Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Table of content 2 3 Executive Summary ................................................................................................................................ 6 4 12.1 Framing ...................................................................................................................................... 9 5 12.2 Methodological approach.......................................................................................................... 11 6 12.3 Climatic impact-drivers for sectors............................................................................................. 14 7 12.3.1 Heat and cold ............................................................................................................................... 16 8 12.3.1.1 Mean air temperature ......................................................................................................... 16 9 12.3.1.2 Extreme heat ....................................................................................................................... 17 10 12.3.1.3 Cold spells ............................................................................................................................ 18 11 12.3.1.4 Frost ..................................................................................................................................... 18 12 12.3.2 Wet and dry ................................................................................................................................. 19 13 12.3.2.1 Mean precipitation .............................................................................................................. 19 14 12.3.2.2 River flood............................................................................................................................ 20 15 12.3.2.3 Heavy precipitation and pluvial flood.................................................................................. 20 16 12.3.2.4 Landslide .............................................................................................................................. 20 17 12.3.2.5 Aridity .................................................................................................................................. 21 18 12.3.2.6 Hydrological drought ........................................................................................................... 21 19 12.3.2.7 Agricultural and ecological drought .................................................................................... 21 20 12.3.2.8 Fire weather......................................................................................................................... 22 21 12.3.3 Wind ............................................................................................................................................ 22 22 12.3.3.1 Mean wind speed ................................................................................................................ 22 23 12.3.3.2 Severe wind storm ............................................................................................................... 22 24 12.3.3.3 Tropical cyclone ................................................................................................................... 22 25 12.3.3.4 Sand and dust storm ............................................................................................................ 23 26 12.3.4 Snow and ice ................................................................................................................................ 23 27 12.3.4.1 Snow, glacier and ice sheet ................................................................................................. 23 28 12.3.4.2 Permafrost ........................................................................................................................... 24 29 12.3.4.3 Lake, river and sea ice ......................................................................................................... 24 30 12.3.4.4 Heavy snowfall and ice storm .............................................................................................. 24 31 12.3.4.5 Hail ....................................................................................................................................... 24 32 12.3.4.6 Snow avalanche ................................................................................................................... 24 33 12.3.5 Coastal ......................................................................................................................................... 25 34 12.3.5.1 Relative sea level ................................................................................................................. 25 35 12.3.5.2 Coastal flood ........................................................................................................................ 25 36 12.3.5.3 Coastal erosion .................................................................................................................... 25 37 12.3.6 Oceanic ........................................................................................................................................ 26 Do Not Cite, Quote or Distribute 12-2 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.3.6.1 Mean ocean temperature ................................................................................................... 26 2 12.3.6.2 Marine heatwave................................................................................................................. 26 3 12.3.6.3 Ocean acidity ....................................................................................................................... 26 4 12.3.6.4 Ocean Salinity ...................................................................................................................... 27 5 12.3.6.5 Dissolved oxygen ................................................................................................................. 27 6 12.3.7 Other climatic impact-drivers ...................................................................................................... 27 7 12.3.7.1 Air pollution weather........................................................................................................... 27 8 12.3.7.2 Atmospheric Carbon Dioxide (CO2) at surface..................................................................... 27 9 12.3.7.3 Radiation at surface ............................................................................................................. 28 10 12.3.7.4 Additional relevant climatic impact-drivers ........................................................................ 28 11 12.4 Regional information on changing climate ................................................................................. 29 12 12.4.1 Africa ............................................................................................................................................ 32 13 12.4.1.1 Heat and cold ....................................................................................................................... 32 14 12.4.1.2 Wet and dry ......................................................................................................................... 33 15 12.4.1.3 Wind .................................................................................................................................... 36 16 12.4.1.4 Snow and Ice........................................................................................................................ 37 17 12.4.1.5 Coastal and Oceanic ............................................................................................................ 37 18 12.4.2 Asia .............................................................................................................................................. 39 19 12.4.2.1 Heat and cold ....................................................................................................................... 40 20 12.4.2.2 Wet and dry ......................................................................................................................... 41 21 12.4.2.3 Wind .................................................................................................................................... 43 22 12.4.2.4 Snow and Ice........................................................................................................................ 45 23 12.4.2.5 Coastal and oceanic ............................................................................................................. 46 24 12.4.3 Australasia ................................................................................................................................... 48 25 12.4.3.1 Heat and Cold ...................................................................................................................... 49 26 12.4.3.2 Wet and Dry ......................................................................................................................... 51 27 12.4.3.3 Wind .................................................................................................................................... 54 28 12.4.3.4 Snow and Ice........................................................................................................................ 55 29 12.4.3.5 Coastal and Oceanic ............................................................................................................ 55 30 12.4.4 Central and South America .......................................................................................................... 58 31 12.4.4.1 Heat and cold ....................................................................................................................... 59 32 12.4.4.2 Wet and dry ......................................................................................................................... 60 33 12.4.4.3 Wind .................................................................................................................................... 62 34 12.4.4.4 Snow and ice ........................................................................................................................ 63 35 12.4.4.5 Coastal and oceanic ............................................................................................................. 64 36 12.4.5 Europe ......................................................................................................................................... 66 Do Not Cite, Quote or Distribute 12-3 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.5.1 Heat and Cold ...................................................................................................................... 67 2 12.4.5.2 Wet and Dry ......................................................................................................................... 68 3 12.4.5.3 Wind .................................................................................................................................... 71 4 12.4.5.4 Snow and ice ........................................................................................................................ 72 5 12.4.5.5 Coastal and Oceanic ............................................................................................................ 73 6 12.4.5.6 Other.................................................................................................................................... 75 7 12.4.6 North America ............................................................................................................................. 76 8 12.4.6.1 Heat and cold ....................................................................................................................... 77 9 12.4.6.2 Wet and dry ......................................................................................................................... 78 10 12.4.6.3 Wind .................................................................................................................................... 81 11 12.4.6.4 Snow and ice ........................................................................................................................ 82 12 12.4.6.5 Coastal and oceanic ............................................................................................................. 84 13 12.4.7 Small islands ................................................................................................................................ 86 14 12.4.7.1 Heat and Cold ...................................................................................................................... 87 15 12.4.7.2 Wet and Dry ......................................................................................................................... 88 16 12.4.7.3 Wind .................................................................................................................................... 90 17 12.4.7.4 Coastal and Oceanic ............................................................................................................ 90 18 12.4.8 Open and deep ocean.................................................................................................................. 93 19 12.4.9 Polar terrestrial regions ............................................................................................................... 96 20 12.4.9.1 Heat and cold ....................................................................................................................... 96 21 12.4.9.2 Wet and dry ......................................................................................................................... 97 22 12.4.9.3 Wind .................................................................................................................................... 97 23 12.4.9.4 Snow and ice ........................................................................................................................ 98 24 12.4.9.5 Coastal and oceanic ............................................................................................................. 99 25 12.4.10 Specific zones and hotspots .................................................................................................. 101 26 12.4.10.1 Hotspots of biodiversity (land, coasts and oceans) ........................................................... 101 27 12.4.10.2 Cities and settlements by the sea...................................................................................... 102 28 12.4.10.3 Deserts and semi-arid areas .............................................................................................. 103 29 12.4.10.4 Mountains.......................................................................................................................... 104 30 12.4.10.5 Tropical forests .................................................................................................................. 105 31 12.5 Global perspective on climatic impact-drivers .......................................................................... 106 32 12.5.1 A global synthesis ...................................................................................................................... 106 33 12.5.2 The emergence of climatic impact-drivers across time and scenarios ..................................... 108 34 Cross-Chapter Box 12.1: Projections by warming levels of hazards relevant to the assessment of 35 Representative Key Risks and Reasons for Concern ....................................... 113 36 12.6 Climate change information in climate services........................................................................ 120 Do Not Cite, Quote or Distribute 12-4 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.6.1 Context of climate services........................................................................................................ 120 2 12.6.2 Assessment of climate services practice and products related to climate change information121 3 12.6.3 Challenges.................................................................................................................................. 123 4 Cross-Chapter Box 12.2: Climate services and climate change information.......................................... 124 5 12.7 Final Remarks ......................................................................................................................... 128 6 Frequently Asked Questions ............................................................................................................... 129 7 References ......................................................................................................................................... 135 8 Figures ............................................................................................................................................... 205 9 Do Not Cite, Quote or Distribute 12-5 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Executive Summary 2 3 Climate change information is increasingly available and robust at regional scale for impacts and risk 4 assessment. Climate services and vulnerability, impacts, and adaptation studies require regional scale multi- 5 decadal climate observations and projections. Since AR5, the increased availability of coordinated ensemble 6 regional climate model projections and improvements in the level of sophistication and resolution of global 7 and regional climate models, completed by attribution and sectoral vulnerability studies, have enabled the 8 investigation of past and future evolution of a range of climatic quantities that are relevant to socio-economic 9 sectors and natural systems. Chapter 12 consolidates core physical knowledge from preceding AR6 WGI 10 chapters and post-AR5 climate impact assessment literature to assess the spatio-temporal evolution of the 11 climatic conditions that may lead to regional scale impacts and risks (following the sectoral classes adopted 12 by AR6 WGII) in the world’s regions (presented in Chapter 1) {12.1} 13 14 The Climatic Impact-Driver (CID) framework adopted in Chapter 12 allows for assessment of 15 changing climate conditions that are relevant for sectoral impacts and risks assessment. CIDs are 16 physical climate system conditions (e.g., means, events, extremes) that affect an element of society or 17 ecosystems and are thus a priority for climate information provision. Depending on system tolerance, CIDs 18 and their changes can be detrimental, beneficial, neutral, or a mixture of each across interacting system 19 elements, regions and society sectors. Each sector is affected by multiple CIDs, and each CID affects 20 multiple sectors. A CID can be measured by indices to represent related tolerance thresholds. {12.1-12.3} 21 22 The current climate in most regions is already different from the climate of the early or mid 20th 23 century with respect to several CIDs. Climate change has already altered CID profiles and resulted in 24 shifts in the magnitude, frequency, duration, seasonality, and spatial extent of associated indices (high 25 confidence). Changes in temperature-related CIDs such as mean temperatures, growing season length, 26 extreme heat, frost, have already occurred and many of these changes have been attributed to human 27 activities (medium confidence). Mean temperatures and heat extremes have emerged above natural 28 variability in all land regions with high confidence. In tropical regions, recent past temperature distributions 29 have already shifted to a range different to that of the early 20th century (high confidence). Ocean 30 acidification and deoxygenation have already emerged over most of the global open ocean, as has reduction 31 in Arctic sea ice (high confidence). Using CID index distributions and event probabilities accurately in both 32 current and future risk assessments requires taking into account the climate change–induced shifts in 33 distributions that have already occurred {12.4, 12.5} 34 35 Several impact-relevant changes have not yet emerged from the natural variability, but will emerge 36 sooner or later in this century depending on the emission scenario (high confidence). Increasing 37 precipitation is projected to emerge before the middle of the century in the high latitudes of the Northern 38 hemisphere (high confidence). Decreasing precipitation will emerge in a very few regions (Mediterranean, 39 South Africa, South Western Australia) (medium confidence) by the mid-century (medium confidence). The 40 anthropogenic forced signal in near-coast relative sea-level rise will emerge by mid-century RCP8.5 in all 41 regions with coasts, except in the West Antarctic region where emergence is projected to occur before 2100 42 (medium confidence). The signal of ocean acidification in the surface ocean is projected to emerge before 43 2050 in every ocean basin (high confidence). However, there is low evidence of emerging drought trends 44 above natural variability in the 21st century {12.5}. 45 46 Every region of the world will experience concurrent changes in multiple CIDs by mid-century (high 47 confidence), challenging the resilience and adaptation capacity of the region. Heat, cold, snow and ice, 48 coastal oceanic, and CO2 at surface CID changes are projected with high confidence in most regions, 49 indicating worldwide challenges, while additional region-specific changes are projected in other CIDs that 50 may lead to more regional challenges. High confidence increases in some of the drought, aridity, and fire 51 weather CIDs will challenge, for example, agriculture, forestry, water systems, health and ecosystems in 52 Southern Africa, the Mediterranean, North Central America, Western North America, the Amazon regions, 53 South-western South America, and Australia. High confidence changes in snow, ice and pluvial or river 54 flooding changes will pose challenges for, for example, energy production, river transportation, ecosystems, 55 infrastructure and winter tourism in North America, Arctic regions, Andes regions, Europe, Siberia, Central, Do Not Cite, Quote or Distribute 12-6 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 South and East Asia, Southern Australia and New Zealand. Only a few CIDs are projected to change with 2 high confidence in the Sahara, Madagascar, Arabian Peninsula, Western Africa, Small Islands; however, the 3 lower confidence levels for CID changes in these regions can originate from knowledge gaps or model 4 uncertainties, and does not necessarily mean that these regions have relatively low risk. {12.5} 5 6 Worldwide changes in heat, cold, snow and ice, coastal, oceanic and CO2-related CIDs will continue 7 over the 21st century, albeit with regionally varying rates of change, regardless of the climate scenario 8 (high confidence). In all regions, there is high confidence that, by 2050, mean temperature and heat extremes 9 will increase, and there is high confidence that sea surface temperature will increase in all oceanic regions, 10 excepting the North Atlantic. With the exception of a few regions with substantial land uplift, relative sea- 11 level rise is very likely to virtually certain (depending on the region) to continue along the 21st century, 12 contributing to increased coastal flooding in most low-lying coastal areas (high confidence) and coastal 13 erosion along most sandy coasts (high confidence), while ocean acidification is virtually certain to increase. 14 It is virtually certain that atmospheric CO2 at the surface will increase in all emissions scenarios until net- 15 zero emissions are achieved. Glaciers will continue to shrink and permafrost to thaw in all regions where 16 they are present (high confidence). These changes will lead to climate states with no recent analogue that are 17 of particular importance for specific regions such as tropical forests or biodiversity hotspots. {12.4} 18 19 A wide range of region-specific CID changes relative to recent past are expected with high or medium 20 confidence, by 2050 and beyond. Most of these changes are concerning CIDs related to the water cycle and 21 storms. Agricultural and ecological drought changes are generally of higher confidence than hydrological 22 drought changes, with increases projected in North and Southern Africa, Madagascar, Southern and Eastern 23 Australia, some regions of Central and South America, Mediterranean Europe, Western North America and 24 North Central America (medium to high confidence). Fire weather conditions will increase by 2050 under 25 RCP4.5 or above in several regions in Africa, Australia, several regions of South America, Mediterranean 26 Europe, and North America (medium to high confidence). Extreme precipitation and pluvial flooding will 27 increase in many regions around the world (high confidence). Increases in river flooding are also expected in 28 Western and Central Europe and in polar regions (high confidence), most of Asia, Australasia and North 29 America, South American Monsoon and Southeastern South America (medium confidence). Mean winds are 30 projected to slightly decrease by 2050 over much of Europe, Asia, and Western North America, and increase 31 in many parts of South America except Patagonia, West and South Africa and Eastern Mediterranean 32 (medium confidence). Storms are expected to have a decreasing frequency but increasing intensity over the 33 Mediterranean, most of North America, and increase over most of Europe. Enhanced convective conditions 34 are expected in North America (medium confidence). Tropical cyclones are expected to increase in intensity 35 despite a decrease in frequency in most tropical regions (medium confidence). Climate change will modify 36 multiple CIDs for small islands in all ocean basins, most notably those related to heat, aridity and droughts, 37 tropical cyclones and coastal impacts. {12.4} 38 39 The level of confidence in the projected direction of change in CIDs and the intensity of the signal 40 depend on mitigation efforts over the 21st century, as reflected by the differences between end-century 41 projections for different climate scenarios. Dangerous humid heat thresholds, such as the NOAA HI of 42 41°C, will be exceeded much more frequently under SSP5-8.5 scenario than under SSP1-2.6 and will affect 43 many regions (high confidence). In many tropical regions, the number of days per year where a HI of 41°C is 44 exceeded will increase by more than 100 days relative to the recent past under SSP5-8.5, while this increase 45 will be limited to less than 50 days under SSP1-2.6 (high confidence). The number of days per year where 46 temperature exceeds 35°C will increase by more than 150 days in many tropical areas, such as the Amazon 47 basin and South-east Asia under SSP5-8.5, while it is expected to increase by less than 2 months in these 48 areas under SSP1-2.6 (except for the Amazon Basin). There is high confidence that the total length of sandy 49 shorelines around the world that are projected to retreat by more than 100 m will be 35% greater under 50 RCP8.5 (~130,000 km) compared to RCP4.5 (~ 95,000 km) by the end of the century. The frequency of the 51 present-day 1-in-100-yr extreme sea level (represented here by extreme total water level) event, in a globally 52 averaged sense, is projected to become an event that occurs multiple times per year under RCP8.5, while 53 under RCP 4.5 it is projected to become a 1-in-5-yr event, representing a 5 fold difference between the two 54 RCPs (high confidence).{12.4, 12.5} 55 Do Not Cite, Quote or Distribute 12-7 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 There is low confidence in past and future changes of several CIDs. In nearly all regions there is low 2 confidence in changes in hail, ice storms, severe storms, dust storms, heavy snowfall, and avalanches, 3 although this does not indicate that these CIDs will not be affected by climate change. For such CIDs, 4 observations are short-term or lack homogeneity, and models often do not have sufficient resolution or 5 accurate parametrizations to adequately simulate them over climate change time scales. {12.4}. 6 7 Many global- and regional-scale CIDs have a direct relation to global warming levels (GWLs) and can 8 thus inform the hazard component of ‘Representative Key Risks’ and ‘Reasons for Concern’ assessed 9 by AR6 WGII. These include heat, cold, wet and dry hazards, both mean and extremes; cryospheric hazards 10 (snow cover, ice extent, permafrost) and oceanic hazards (marine heatwaves) (high confidence). For some of 11 these, a quantitative relation can be drawn (high confidence). For example, with each degree of GSAT 12 warming, the magnitude and intensity of many heat extremes show a linear change, while some changes in 13 frequency of threshold exceedances are exponential; arctic temperatures warm about twice as fast as GSAT; 14 global SSTs increase by ~80% of GSAT change; Northern Hemisphere spring snow cover decreases by ~8% 15 per 1°C. For other hazards (e.g., ice sheet behaviour, glacier mass loss, global mean sea-level rise, coastal 16 floods and coastal erosion) the time and/or scenario dimensions remain critical and a simple relation with 17 GWLs cannot be drawn (high confidence), but still quantitative estimates assuming specific time frames, 18 and/or stabilized GWLs can be derived (medium confidence). Model uncertainty challenges the link between 19 specific GWLs and tipping points and irreversible behavior, but their occurrence cannot be excluded and 20 their chances increase with warming levels (medium confidence) {CCB 12.1}. 21 22 Since AR5, climate change information produced in climate service contexts has increased 23 significantly due to scientific and technological advancements and growing user demand (very high 24 confidence). Climate services involve the provision of climate information in such a way as to assist 25 decision-making. These services include appropriate engagement from users and providers, are based on 26 scientifically credible information and expertise, have an effective access mechanism, and respond to user 27 needs.. Climate services are being developed across regions, sectors, timescales and target users. {12.6} 28 29 Climate services are growing rapidly and are highly diverse in their practices and products (very high 30 confidence). The decision-making context, level of user engagement and co-production between scientists, 31 practitioners and intended users are important determinants of the type of climate service developed and its 32 utility supporting adaptation, mitigation and risk management decisions. User needs and decision-making 33 contexts are very diverse and there is no universal approach to climate services. {12.6} 34 35 Realization of the full potential of climate services is often hindered by limited resources for the co- 36 design and co-production process, including sustained engagement between scientists, service 37 providers and users (high confidence). Further challenges relate to climate services development, provision 38 of climate services, generation of climate service products, communication with users, and evaluation of the 39 quality and socio-economic value of climate services. The development of climate services often uncovers 40 and presents new research challenges to the scientific community. {12.6} 41 42 Do Not Cite, Quote or Distribute 12-8 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.1 Framing 2 3 Climate change is already resulting in significant societal and environmental impacts and will induce major 4 socio-economic damages in the future (AR5 WGII). The society, at large, benefits from information related 5 to climate change risks, which enables the development of options to protect lives, preserve nature, build 6 resilience and prevent avoidable loss and damage. Climate change can also lead to beneficial conditions 7 which can be taken into account in adaptation strategies. 8 9 This chapter assesses climate information relevant to regional impact and risk assessment. It 10 complements other WG1 chapters which focus on the physical processes determining changes in the 11 climate system and on methods for estimating regional changes. 12 13 Impacts of climate change are driven not only by changes in climate conditions, but also by changes in 14 exposure and vulnerability (Cross-Chapter Box 1.3). This chapter concentrates on drivers of impacts that are 15 of climatic origin (see also SR1.5 (IPCC, 2018), and Section 1.3.2), referred to in WGI as “climatic impact- 16 drivers” (CIDs). CIDs are physical climate system conditions (e.g., means, events, extremes) that affect an 17 element of society or ecosystems. Depending on system tolerance, CIDs and their changes can be 18 detrimental, beneficial, neutral, or a mixture of each across interacting system elements and regions. 19 However, this chapter largely focuses on drivers commonly connected to hazards, and adopts the IPCC risk 20 framework (Chapter 1, Cross-Chapter Box 1.3) since the main objective of the UNFCCC convention is to 21 “prevent dangerous anthropogenic interference with the climate system” (Article 2). 22 23 In some cases, risk assessments may require climate information beyond the CIDs identified in this chapter, 24 with further impacts or risk modelling often driven by historical climate forcing datasets (e.g., Ruane et al., 25 2021) and full climate scenario time series (e.g., Lange, 2019) produced using methods described in Chapter 26 10. Chapter 12 focuses on the assessment of a finite number of drivers and how they are projected to evolve 27 with climate change, in order to inform impact and risk assessments. 28 29 This chapter is new IPCC WGI assessment reports, in that it represents a contribution to the “IPCC Risk 30 Framework”. Within this framework, climate-related impacts and risks are determined through an interplay 31 between the occurrence of climate hazards and their consequences depending on the exposure of the affected 32 human or natural system and its vulnerability to the hazardous conditions. In Chapter 12, we are assessing 33 climatic impact-drivers that could lead to hazards or to opportunities, from the literature and model results 34 since AR5. This will particularly support the assessment of key risks related to climate change by WGII 35 (Chapter 16). Despite the fact that impacts may also be induced by climate adaptation and mitigation policies 36 themselves, as well as by socioeconomic trends, changes in vulnerability or exposure, and external 37 geophysical hazards such as volcanoes, the focus here is only on ‘climatic’ impacts and risks induced by 38 shifts in physical climate phenomena that directly influence human and ecological systems. 39 40 This chapter follows the terminology associated with the framing introduced in Chapter 1 (Cross-chapter 41 Box 1.2: Baseline and reference periods in AR6) and as found in the AR6 Glossary. The highlighted terms 42 below are introduced and used extensively in this chapter: 43 44 • Indices for climatic impact-drivers: numerically computable indices using one or a combination of 45 climate variables designed to measure the intensity of the climatic impact-driver, or the probability 46 of exceedance of a threshold. For instance, an index of heat inducing human health stress is the Heat 47 Index (HI) that combines temperature and relative humidity (e.g., Burkart et al., 2011; Lin et al., 48 2012; Kent et al., 2014) and is used by the US National Oceanic and Atmospheric Administration 49 (NOAA) for issuing heat warnings. 50 • Thresholds for climatic impact-drivers: an identified index value beyond which a climatic impact- 51 driver interacts with vulnerability or exposure to create, increase or reduce an impact, risk or 52 opportunity. Thresholds can be used to measure various aspects of the climatic impact-driver 53 (magnitude or intensity, duration, frequency, timing, and spatial extent of threshold exceedance). For 54 instance, a threshold of daily maximum temperature above 35⁰C is considered critical for maize 55 pollination and production (e.g., Schauberger et al., 2017; Tesfaye et al., 2017). Do Not Cite, Quote or Distribute 12-9 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 The approach adopted here is consistent with the United Nations ‘Sendai Framework for Disaster Risk 2 Reduction 2015-2030’, which aims to face disaster consequences (including, but not limited to climate 3 disasters) and reduce risks in natural, managed, and built environments (Aitsi-Selmi et al., 2015; UNISDR, 4 2015). The classification of climatic impact-drivers in this chapter is largely consistent with the classification 5 of hazards used in the Sendai Framework, however, the UNISDR hazard list spans a wider range of hazards 6 inducing damage to society, including hazards that are not directly related to climate (such as volcanoes and 7 earthquakes), which are excluded from the assessment herein. Furthermore, the UNISDR classification of 8 hazards does not include mean climatic conditions which are also discussed as climatic impact-drivers in this 9 chapter. The first priority mentioned in the Sendai Framework is understanding disaster risk as a necessary 10 step for action. Facilitating such an understanding is a clear goal of this chapter. 11 12 The chapter adopts a regional perspective (continental regions as defined in Chapter 1 and used in WGII, see 13 Figure 1.18) on climatic impact-drivers to support decision making across a wide audience of global and 14 regional stakeholders in addition to governments (e.g., civil society organizations, public and private sectors, 15 academia). While the focus here is on future changes, it also describes current levels and observed trends of 16 CIDs as an important point of reference for informing adaptation strategies. 17 18 Figure 12.1 summarizes the rationale behind Chapter 12 as the linkage (also referred to as a handshake) 19 between WGI and WGII, illustrating how the changing profile of risk may be informed by an assessment of 20 climatic impact-drivers, aligning WGI findings on physical climate change with WGII needs. The 21 implementation of mitigation policy shifts may modulate hazard probability changes (i.e., by reducing 22 emissions to limit global warming) as well as regional vulnerability and exposure. The assessment herein is 23 organized around regional climatic impact-drivers, but also relates key indices and thresholds to increasing 24 global drivers (such as mean surface warming) as a contribution to the assessment of ‘Reasons for Concern’ 25 in WGII (O’Neill et al., 2017). 26 27 28 [START FIGURE 12.1 HERE] 29 30 Figure 12.1: Schematic diagram showing the use of climate change information (WGI chapters) for typical 31 impacts or risk assessment (WGII chapters) and the role of Chapter 12, via an illustration of the 32 assessment of property damage or loss in a particular region when extreme sea level exceeds dike 33 height. 34 35 [END FIGURE 12.1 HERE] 36 37 38 The narrative in Chapter 12 is illustrated in Figure 12.2. First, Section 12.2 defines a range of climatic 39 impact-driver categories that are relevant for regional and sectoral impacts and the associated ECVs. Next, 40 Section 12.3 identifies climatic impact-drivers and their relevant indices that are frequently used in the 41 context of climate impacts in the WGII focus sectors (Chapters 2-8). The assessment of changes in regional- 42 scale climatic impact-drivers is then developed within Section 12.4 by continent, following the structure of 43 the WGII assessment report regional chapters (Chapters 9-15), and adding the polar regions, open/deep 44 ocean and other specific zones corresponding to the WGII cross-chapter papers. Section 12.5 then presents a 45 global perspective (both bottom-up and top-down) on the change of regional climatic impact-drivers, 46 including an assessment of the “emergence” of climatic impact-drivers. Section 12.6 discusses how climate 47 information is used in ‘climate services’, which encompasses a range of activities bridging climate science 48 and its use for adaptation and mitigation decision making (see also AR6 WGII Chapter 17). The chapter 49 concludes with final remarks in Section 12.7. 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-10 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 [START FIGURE 12.2 HERE] 2 3 Figure 12.2: Narrative and structure of Chapter 12. 4 5 [END FIGURE 12.2 HERE] 6 7 8 The chapter includes two Cross-Chapter boxes. Cross-chapter Box 12.1 connects climatic impact-drivers 9 them to global climate drivers and levels of warming as an element of the ‘Reasons for Concern’ framework 10 (AR6 WGII Chapter 16). An additional cross-chapter box, including three case studies from Europe, Asia 11 and Africa, describes how climate services draw upon and apply regional climate information to support 12 stakeholder decisions (Cross-Chapter Box 12.2). 13 14 15 12.2 Methodological approach 16 17 This section details the methodological approach followed in Chapter 12 and discusses the underlying 18 rationale for the assessments presented herein. Scientific literature on vulnerability, impacts, and adaptation 19 (as typically asssessed in IPCC WGII) is examined to identify relevant climatic impact-drivers (CIDs) that 20 contribute to sectoral risks and opportunities. Projected changes in corresponding CID indices are then 21 derived from existing literature on changes in the physical climate system, results of other AR6 WGI 22 chapters, and direct calculations based on climate projections from several model ensembles. 23 24 The classification of climatic impact-drivers, the ways that they change (e.g., their magnitude or intensity, 25 duration, frequency, timing, and spatial extent) is described in this section. It is emphasized that this chapter 26 only assesses literature relating to physical climatic impact-drivers, not their impacts on human systems or 27 the environment. Thus, here we do not consider indicators including exposure or vulnerability as assessed by 28 WGII, although the selection of climatic impact-drivers is informed by literature feeding into WGII. 29 30 Chapter 12 assesses climate information relevant for impact and for risk assessment in the seven main 31 sectors corresponding to Chapters 2-8 of the WGII assessment report: 32 • Terrestrial and freshwater ecosystems and their services (WGII Chapter 2); 33 • Ocean and coastal ecosystems and their services (WGII Chapter 3); 34 • Water (WGII Chapter 4); 35 • Food, fibre and other ecosystem products (WGII Chapter 5); 36 • Cities, settlements and key infrastructure (WGII Chapter 6); 37 • Health, wellbeing and the changing structure of communities (WGII Chapter 7); 38 • Poverty, livelihoods and sustainable development (WGII Chapter 8). 39 40 Many of these sectors also include assets affected by climate change that are important for recreation and 41 tourism, including elements of ecosystems services, health and wellbeing, communities, livelihoods, and 42 sustainable development (see also Chapter 1 on the Intergovernmental Science-Policy Platform on 43 Biodiversity and Ecosystem Services (IPBES), and the IPCC Special Report on climate change, 44 desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in 45 terrestrial ecosystems (Hurlbert et al., 2019; IPCC, 2019c). 46 47 CIDs can be captured in seven main types: heat and cold, wet and dry, wind, snow and ice, coastal, oceanic 48 and others. Table 12.1 provides an overview of the seven CID types and the CID categories associated with 49 each type. The type “Other” comprises additional CIDs that are not encompassed with the above six CID 50 types, including air pollution weather (e.g., meteorological conditions that favour high concentrations of 51 surface ozone, particulate matter, or other air pollutants), near-surface atmospheric CO2 concentrations, and 52 mean radiation forcing at the surface, for instance relevant for plant growth. Icebergs, fog and lightning are 53 also noted in this chapter but are not broadly assessed across all sub-sections. In addition, there can be 54 changes in impacts associated with earthquakes that interact with climate variables and climate change, such 55 as liquefaction (e.g., Yasuhara et al., 2012) during earthquakes, or earthquakes caused by snow and water Do Not Cite, Quote or Distribute 12-11 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 changes (Amos et al., 2014; Johnson et al., 2017), which are secondary effects on geophysical hazards that 2 are not further assessed in this chapter. The characteristics and physical description of the climate 3 phenomena or essential climate variables associated with each of these CID categories are assessed and 4 described in previous Chapters 2-11 or Chapter 12 directly as indicated in Table 12.1. The CID categories 5 are further mapped onto different sectors in section 12.3 (Table 12.2). 6 7 8 [START TABLE 12.1 HERE] 9 10 Table 12.1: Overview of the main Climatic impact-driver (CID) types and related CID categories with a short 11 description and their link to other chapters where the underlying climatic phenomenon and its associated 12 essential climate variables are assessed and described. 13 14 Physical CID Description of Type CID Category Brief Description Phenomena Mean air temperature Mean surface air temperature and its diurnal and seasonal cycles. CH2, CH3, Heat and Cold CH4, Atlas Extreme heat Episodic high surface air temperature events potentially exacerbated CH11 by humidity. Cold spell Episodic cold surface air temperature events potentially exacerbated CH11 by wind. Frost Freeze and thaw events near the land surface and their seasonality. CH12 Mean precipitation Mean precipitation, its diurnal and seasonal cycles, and associated soil CH2, CH8, moisture and humidity conditions. Atlas River flood Episodic high water levels in streams and rivers driven by basin runoff CH8, CH11 and the expected seasonal cycle of flooding. Heavy precipitation and High rates of precipitation and resulting episodic, localized flooding CH11 pluvial flood of streams and flat lands. Landslide Ground and atmospheric conditions that lead to geological mass CH12 movements, including landslide, mudslide, and rockfall. Aridity Mean conditions of precipitation and evapotranspiration compared to CH8, CH11, Wet and Dry potential atmospheric and surface water demand, resulting in low Atlas mean surface water, low soil moisture and/or low relative humidity. Hydrological drought Episodic combination of runoff deficit and evaporative demand that CH8, CH11 lead to dry soil. Agricultural and Episodic combination of soil moisture supply deficit and atmospheric CH8, CH11 ecological drought demand requirements that challenge the vegetation’s ability to meet its water needs for transpiration and growth. Note: ‘agricultural’ vs. ‘ecological’ term depends on affected biome. Fire weather Weather conditions conducive to triggering and sustaining wildfires, CH11, CH12 usually based on a set of indicators and combinations of indicators including temperature, soil moisture, humidity, and wind. Fire weather does not include the presence or absence of fuel load. Note: distinct from wildfire occurrence and area burned. Mean wind speed Mean wind speeds and transport patterns and their diurnal and CH2, CH12 seasonal cycles. Severe wind storm Severe storms including thunderstorms, wind gusts, derechos, and CH11, CH12 Wind tornados. Tropical cyclone Strong, rotating storm originating over tropical oceans accompanied CH11 by high winds, rainfall, and storm surge. Sand and dust storm Storms causing the transport of soil and fine dust particles. CH8, CH12 Snow, glacier and ice Snowpack seasonality and characteristics of glaciers and ice sheets CH2, CH9, sheet including calving events and meltwater. Atlas Snow and Ice Permafrost Permanently frozen deep soil layers, their ice characteristics, and the CH2, CH9 characteristics of seasonally frozen soils above. Lake, river and sea ice The seasonality and characteristics of ice formations on the ocean and CH2, CH9 freshwater bodies of water. Heavy snowfall and ice High snowfall and ice storm events including freezing rain and rain- CH11, CH12 storm on-snow conditions. Hail Storms producing solid hailstones. CH11, CH12 Do Not Cite, Quote or Distribute 12-12 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Snow avalanche Cryospheric mass movements and the conditions of collapsing CH12 snowpack. Relative sea level The local mean sea surface height relative to the local solid surface. CH9 Coastal flood Flooding driven by episodic high coastal water levels that result from CH9, CH12 Coastal a combination of relative sea level rise, tides, storm surge, and wave setup. Coastal erosion Long term or episodic change in shoreline position caused by relative CH12 sea level rise, nearshore currents, waves, and storm surges. Mean ocean Mean temperature profile of ocean through the seasons, including heat CH2, CH9 temperature content at different depths and associated stratification. Open Ocean Marine heatwave Episodic extreme ocean temperatures. CH9, CH12 Ocean acidity Profile of ocean water pH and accompanying concentrations of CH5 carbonate and bicarbonate ions. Ocean salinity Profile of ocean salinity and associated seasonal stratification. Note: CH2, CH5 distinct from salinization of freshwater resources. Dissolved oxygen Profile of ocean water dissolved oxygen and episodic low oxygen CH5 events. Air pollution weather Atmospheric conditions that increase the likelihood of high particulate CH6 matter or ozone concentrations or chemical processes generating air pollutants. Note: distinct from aerosol emissions or air pollution Other concentrations themselves. Atmospheric CO2 at Concentration of atmospheric carbon dioxide [CO2] at the surface. CH5 surface Note: distinct from overall radiative effect of CO2 as greenhouse gas. Radiation at surface Balance of net shortwave, longwave and ultraviolet radiation at the CH7 earth’s surface and their diurnal and seasonal patterns. 1 2 [END TABLE 12.1 HERE] 3 4 5 Potential changes in the seasonality of CIDs or the length and characteristics of seasons (e.g. changes in 6 growing season length or pollen season) are also important as they may shift the timing of many CIDs with 7 broad implications for sectors and regional stakeholders (Wanders and Wada, 2015; Cassou and Cattiaux, 8 2016; Hansen and Sato, 2016; Brönnimann et al., 2018; Marelle et al., 2018; Unterberger et al., 2018; Kuriqi 9 et al., 2020). Episodic CIDs characterize impact-relevant conditions persisting from short to long time 10 frames but eventually returning to normal conditions. 11 12 In some situations, phenomena causing severe impacts go well beyond a single extreme event or a single 13 climate variable, and can include interaction of climatic conditions, such as sea level rise and storm surges 14 (Wahl et al., 2015), precipitation in combination with strong winds (Martius et al., 2016) or flooding quickly 15 followed by a heat wave (Wang et al., 2019c) (see also section 10.5.2.4). Such compound events, particularly 16 in context of climate extremes, are assessed in section 11.8. A combination of non-extreme climatic impact- 17 drivers in time or space can also lead to severe impacts (Cutter, 2018). 18 19 Several climatic impact-drivers are reliant on many factors beyond their associated primary climatic 20 phenomenon. For example, river flooding is heavily dependent on river management and engineering and 21 could also be affected by tidal water levels due to sea level rise and/or storm surge. Coastal flooding could be 22 affected by coastal protection structures, port and harbour structures as well as river flows (on inlet- 23 interrupted coasts). Coastal erosion could be influenced by coastal protection measures as well as fluvial 24 sediment supply to the coast. Furthermore, air pollution weather is not the only or dominant driver, for 25 instance, of surface ozone pollution, but precursor emissions from anthropogenic sources can play a 26 significant role (see section 6.5). Chapter 12 focuses only on the influence of the atmospheric, land and 27 oceanic conditions associated with the climatic impact-drivers and the confidence in the direction of CID 28 changes given here does not take into account existing or potential future adaptation measures, unless 29 otherwise stated. 30 31 For each CID category there can be a range of indices that capture the sector- or application-relevant 32 characteristics of a climatic impact-driver as described in sections 12.3 and 12.4. Indices for climatic impact- 33 drivers that are based on absolute or percentile thresholds (e.g. daily maximum temperature above 35⁰C) can 34 be affected by biases in climate model simulations, such as local or regional deviations of a simulated Do Not Cite, Quote or Distribute 12-13 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 climate variable from observed values (Sillmann et al., 2014; Dosio, 2016). Where sensible (i.e., where 2 reliable observational data is available and a climate model that fits for the desired purpose), the output of 3 climate model simulations can be bias-adjusted, potentially involving advanced methods to account for 4 multiple variables and extreme value statistics as assessed in detail in Cross-chapter box 10.2. Yet, there is 5 no general agreement about which bias adjustment methods to apply, as artefacts can arise both from the 6 climate model and from the bias adjustment method, and the number of available methods has considerably 7 grown in recent years (see a detailed discussion of available methods and their performance in Chapter 10 8 sections 10.3.1.3 and 10.3.3.7 and in Cross-chapter box 10.2). The WGI Interactive Atlas illustrates original 9 and bias-adjusted CIDs (see Atlas.1.4.5) and underlying methods are further described in Technical Annex 10 VI. 11 12 A global perspective on climatic impact-drivers related to their evolution for different global warming levels 13 (see section 1.6) is provided in section 12.5.1. Section 12.5.2 focuses on assessing evidence for the 14 emergence (see section 1.4.2.2) of an anthropogenic climate change signal on the change in CIDs beyond 15 natural climate variability, based on the literature assessed in other chapters and additional literature, at both 16 global and regional scales. The process of generating user-relevant regional climate information in context of 17 co-production and climate services is assessed in sections 10.5, 12.6, Box 10.2 and Cross-chapter boxes 10.3 18 and 12.2. 19 20 21 12.3 Climatic impact-drivers for sectors 22 23 Climate change becomes relevant for regional impact management and for risk assessment when changes in 24 mean conditions or episodic events affect natural and societal assets (system components with 25 socioeconomic, cultural, or intrinsic value) positively or negatively (Table 12.2). Decision makers, policy 26 makers, risk managers and engineers therefore benefit from climate information that tracks key trends and 27 exceedance of thresholds that represent crucial challenges for natural and human systems. While useful 28 indices can vary widely for a given sector and precise tolerance threshold values are often unknown, 29 common metrics, categories and progressions of threshold levels allow experts to recognize coherent 30 messages concerning altered regional impacts and risk profiles under climate change. 31 32 This section surveys the links between CIDs and affected sectors; not to perform specific climate change 33 impact or risk assessments (see AR6 WGII), but to describe key indices (among many) that quantify these 34 links as guidance for stakeholders seeking applicable climate information. This survey builds on the work of 35 the World Meteorological Organization Expert Team on Sector-Specific Climate Indices (ET-SCI) and 36 previous IPCC assessments, notably AR5 WGII (Birkmann et al., 2014; IPCC, 2014a) and IPCC Special 37 Reports (IPCC, 2018, 2019b, 2019c) that have assessed climate hazards affecting sectors but is organized 38 from a CID perspective drawing also upon recent summaries of sectoral hazards (Mora et al., 2018; 39 ICOMOS, 2019; Yokohata et al., 2019). Impacts, risks, and opportunities are rarely attributable to a single 40 CID index or threshold, but climate shifts that push conditions outside of expected conditions and beyond 41 tolerance levels are indicative of impact, risk, or benefit given vulnerability and exposure. Focus is on direct 42 sectoral connections of a CID (Hallegatte and Przyluski, 2010) rather than cascading or secondary effects 43 (e.g., water-borne diseases following a flood, mental health challenges following a severe storm, or the 44 effects of drought on poverty), as these are strongly affected by exposure, vulnerability, and response, as 45 discussed in WGII report. 46 47 Table 12.2 presents a summary of Section 12.3 connections between CIDs as defined in Table 12.1 and key 48 sectoral assets, utilizing the WGII organization of sectors (corresponding to WGII chapters 2-8). Colours are 49 shown for connections with at least medium confidence as assessed from sectoral impacts and risk literature, 50 with relevance assessed according to the prominence of that specific CID/asset connection in analyses of 51 current and future impacts and risk. Within each sector there is a multitude of specific sectoral systems that 52 may be affected by CID increases and decreases, with consequences further distinguished by region, 53 background climate, and socioeconomic or ecological context of the affected asset. Our aim is therefore to 54 recognize important drivers and the common attributes of change within each CID that scientists and 55 practitioners monitor to understand current and future challenges for important asset groups, thereby pointing Do Not Cite, Quote or Distribute 12-14 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 to the climate information that needs to be tailored and analysed for impacts and for risk assessment (Section 2 12.6). Additional effects whereby CIDs affect each other (across Table 12.2 columns) are discussed as 3 climatic phenomena within WGI. The ways sectoral assets affect each other (across Table 12.2 rows) are 4 described throughout WGII, for example with information about the suitability of future climate zones and 5 climate velocity challenges for a given asset potentially drawing from multiple CIDs and associated system 6 tolerance thresholds (Hamann et al., 2015). Some broad connections indicated as low confidence may be 7 under-represented in the literature or could be acute under specific circumstances. 8 9 10 [START TABLE 12.2 HERE] 11 12 Table 12.2: Relevance of key climatic impact-drivers (and their respective changes in intensity, frequency, duration, 13 timing, and spatial extent) for major categories of sectoral assets, as assessed with at least medium 14 confidence in Section 12.3 across many studies and applications. ‘High relevance’ indicates climatic 15 impact-drivers that are most prominent and widely studied for their direct connection to assets, while 16 lower relevance indicates weaker linkages and less commonly-studied driving behaviours. Specific levels 17 of risk and opportunity depend on the changing character of regional hazards, vulnerability, and exposure 18 as assessed in WGII. 19 Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal Oceanic Other Agricultural and ecological drought Heavy precipitation and pluvial Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean Ocean temperature Lake, river and sea ice Mean air temperature Hydrological drought Air pollution weather Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Dissolved oxygen Relative sea level Tropical cyclone Snow avalanche Coastal erosion Ocean salinity Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost flood Hail Sector Asset Tropical forests Temperate and boreal forests Terrestrial and Lakes, rivers and wetlands freshwater Grasslands and savanna ecosystems (WGII Chapter 2) Deserts Mountains Polar Coastal land and inertial zones Ocean and Coastal seas coastal Shelf seas and upwelling zones ecosystems (WGII Chapter 3) Polar seas Open ocean and deep sea Cryosphere reservoir Water Aquifers and groundwater (WGII Chapter 4) Streamflow and surface water Water quality Crop systems Food, fibre and Livestock and pasture systems other ecosystems products Forestry systems (WGII Chapter 5) Fisheries and aquaculture systems Cities, Cities settlements, and Land and water transportation key infrastructure Energy infrastructure (WGII Chapter 6) Built environment Labor productivity Health, wellbeing Morbidity and communities (WGII Chapter 7) Mortality Recreations and tourism+ Poverty, Housing stock* livelihoods and Farmland* sustainable development Livestock mortality* (WGII Chapter 8) Indigenous traditions 20 Do Not Cite, Quote or Distribute 12-15 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 *The Recreation and tourism asset category includes outdoor exercise and the tourism industry (including ecosystem services) assessed in many WGII chapters 2 *This asset category is distinguished by the threat of a full loss of key investments and living environments rather than a recoverable damage or loss of productivity or 3 profit 4 5 [END TABLE 12.2 HERE] 6 7 8 12.3.1 Heat and cold 9 10 12.3.1.1 Mean air temperature 11 12 Information about increasing mean annual and seasonal air temperature is relevant in the determination of 13 suitable species range for terrestrial, freshwater, and intertidal species (Thomas et al., 2004; Elith et al., 14 2010; Hincapie and Caicedo, 2013; Cooper, 2014; Krist et al., 2014; Lindner et al., 2014; Saintilan et al., 15 2014; Urban, 2015; Lenoir and Svenning, 2015; Myers-Smith et al., 2015; Thorne et al., 2017). Ocean 16 ecosystems are affected by the ocean temperature CID (described in Section 12.3.6.1). Species redistribution 17 and extinction studies also need information about climate velocity, a comparison of the pace of warming to 18 geographical temperature gradients that indicates the rate at which a species would have to move to maintain 19 its climatological temperature (Thomas et al., 2004; Loarie et al., 2009; Dobrowski et al., 2013; Burrows et 20 al., 2014; Dobrowski and Parks, 2016; Sittaro et al., 2017) with some studies incorporating additional 21 variables beyond temperature (Hamann et al., 2015). Many freshwater ecosystems are strongly constrained 22 by stream and lake temperatures (Scheurer et al., 2009; Comte and Grenouillet, 2013; Contador et al., 2014; 23 Knouft and Ficklin, 2017). Warmer and more stratified lake temperatures are more conducive to 24 cyanobacteria blooms with implications for ecosystem health and water resource quality (Whitehead et al., 25 2009; Moss, 2011; Jones and Brett, 2014; Chapra et al., 2017; Shatwell et al., 2019). Consideration of night- 26 time and daytime temperature trends also elucidates different biophysical effects on vegetation (Peng et al., 27 2013). Changes in the seasonal timing caused by warming trends are critical to species ranges and 28 ecosystem function (Pearce-Higgins et al., 2015; Hughes et al., 2017b), and indices that characterise the 29 onset of spring shed light on plant emergence and development (Ault et al., 2015). 30 31 Mean air temperature dictates many aspects of crop cultivation, livestock production, agroforestry, and 32 output from freshwater aquaculture and fisheries as well as the potential for food contamination. Mean 33 warming alters suitable cultivation zones for crop species (Bragança et al., 2016; Gendron St-Marseille et al., 34 2019; IPCC, 2019c) and tree species (Hanewinkel et al., 2013; Fei et al., 2017). Crop and ecosystem service 35 productivity often responds directly to mean temperatures, although this is dependent on farming system 36 (Bassu et al., 2014; Challinor et al., 2014; Lobell and Tebaldi, 2014; Rosenzweig et al., 2014; Asseng et al., 37 2015; Li et al., 2015; Fleisher et al., 2017; Zhao et al., 2017; Smith and Fazil, 2019). Many studies relate 38 plant development (phenology), insect generation cycles, and pest outbreaks to growing degree days, an 39 aggregation of daily thermal units above a threshold (e.g., Tmean>5℃) that accelerates with warmer 40 conditions (Hof and Svahlin, 2016; Ruosteenoja et al., 2016; Tripathi et al., 2016). Many plants respond to 41 changes in night-time temperatures that affect respiration and transpiration rates (Narayanan et al., 2015; 42 Chen et al., 2019b), and warming of the soil column is also relevant to determine plant sprouting (Grotjahn, 43 2021). A number of indices have been developed to represent the length of the viable local growing season, 44 including a count of days where Tmax > 5℃ (Mueller et al., 2015) or the period between a year’s first and 45 last set of 5 consecutive days with a weighted Tmean ≥ 10℃ (Li et al., 2018b). Warmer conditions and 46 altered seasonality modify the range and metabolism of some pollinators, pests, diseases and weeds (Wolfe 47 et al., 2008; Bebber, 2015; Aljaryian and Kumar, 2016; IPBES, 2016; Ramesh et al., 2017; Deutsch et al., 48 2018; Nyangiwe et al., 2018) and may reduce the effectiveness of winter storage for farmers and caching 49 species (Sutton et al., 2016). 50 51 Warming raises accumulated seasonal heat indices used in livestock production, especially when humidity is 52 high (Key et al., 2014; Lallo et al., 2018), determines aquaculture suitability and is important for wild fish 53 species migration (Tripathi et al., 2016; Brander et al., 2017). Agricultural planners may also calculate how 54 overall warming trends alter the accumulation of vernalization units or chilling hours for agricultural or 55 horticultural crops (often accumulated temperature deficit below a given daily or hourly threshold) (Dennis 56 and Peacock, 2009; Luedeling, 2012; Tripathi et al., 2016; Grotjahn, 2021). Warming in the post-harvest is Do Not Cite, Quote or Distribute 12-16 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 also important for the determination of spoilage and waste (Stathers et al., 2013) as well as food-borne 2 diseases (Kovats et al., 2004; Mbow et al., 2019). 3 4 Warming affects road degradation rates (Chinowsky and Arndt, 2012; Espinet et al., 2016) and warming 5 rates informs designs for long-term energy efficiency of buildings (Kalvelage et al., 2014). Mean 6 temperature drives seasonal energy demand, often expressed using wintertime heating degree days (the 7 accumulated deficit of daily temperatures below a ‘comfortable’ indoor temperature, e.g., 15.5℃) and 8 summertime cooling degree days (the accumulated excess of temperature above a ‘comfortable’ level, e.g., 9 18℃) (Spinoni et al., 2015; Arnell et al., 2019). Energy resources may also need information on warming 10 trends to determine suitable zones and overall productivity for biofuels and solar panels whose efficiency 11 decreases with higher temperatures (Schaeffer et al., 2012; Wild et al., 2015; Solaun and Cerdá, 2019). 12 13 Health impacts and risk studies compare seasonal temperature conditions to limiting thresholds to understand 14 range shifts and incubation rates for pathogens, disease vectors, and zoonotic hosts (e.g., mosquitos, ticks; 15 Caminade et al., 2012, 2014; Eisen and Moore, 2013; Lima et al., 2016; Ogden, 2017; Monaghan et al., 16 2018) and warming of surface ocean and lake waters conducive to bacterial outbreaks (Baker-Austin et al., 17 2013; Jacobs et al., 2015; Vezzulli et al., 2015). Warmer conditions can also affect tourism (Kovács et al., 18 2017) and impact human health by lengthening the allergy season and increasing pollen concentration 19 (Hamaoui-Laguel et al., 2015; Kinney et al., 2015a; Lake et al., 2017; Upperman et al., 2017; Sapkota et al., 20 2019; Ziska et al., 2019). 21 22 23 12.3.1.2 Extreme heat 24 25 Impacts and risk assessments utilize a large variety of indices and approaches tailored to evaluate heat 26 impacts on human health (Sanderson et al., 2017; Gao et al., 2018a; McGregor and Vanos, 2018; Staiger et 27 al., 2019; Zhu et al., 2019; Schwingshackl et al., 2021). A mixture of simple and complex heat stress indices 28 often combine extreme temperatures and high humidity to capture human health challenges (Aström et al., 29 2013; Chow et al., 2016; Dahl et al., 2017a; Im et al., 2017; Coffel et al., 2018; Li et al., 2018c; Vanos et al., 30 2020). Different optimum temperatures and extreme heat thresholds based on local distributions are needed 31 to reflect acclimation of different locations and populations (Cheng et al., 2018; Dosio, 2017; Hajat et al., 32 2014; Kinney et al., 2015b; Lay et al., 2018; Petitti et al., 2016; Russo et al., 2015; Schwingshackl et al., 33 2021; WHO, 2014). Hot and humid heat episodes can be deadly (Mora et al., 2017), are associated with 34 elevated hospital intake (Goldie et al., 2017) and lower safety and productivity of outdoor labourers (Dunne 35 et al., 2013; Graff Zivin and Neidell, 2014; Kjellstrom et al., 2016; Pal and Eltahir, 2016; Zhao et al., 2016b; 36 Mora et al., 2017; Watts et al., 2018; Orlov et al., 2019). Elevated night-time temperatures prevent the 37 human body from experiencing relief from heat stress (Zhang et al., 2012) and can be tracked over extended 38 periods of sequential day and night heat extremes (Murage et al., 2017; Mukherjee and Mishra, 2018). 39 Extreme heat also exacerbates asthma, respiratory difficulties, and response to airborne allergens such as hay 40 fever (Upperman et al., 2017). Extreme heat affects outdoor exercise such as the use of bike-share facilities 41 (Heaney et al., 2019; Vanos et al., 2020). Large-scale recreational and sporting events such as marathons and 42 tennis tournaments monitor heat extremes when determining the viability of host cities (Smith et al., 2016, 43 2018). 44 45 Short-term exposure of crops to temperatures beyond a critical temperature threshold can lead to lower 46 yields and above a limiting temperature threshold, crops may fail altogether (Schlenker and Roberts, 2009; 47 Lobell et al., 2012, 2013; Gourdji et al., 2013; Deryng et al., 2014; Schauberger et al., 2017; Tesfaye et al., 48 2017; Vogel et al., 2019). The exact level of these thresholds depends on species, cultivar, and farm 49 management (Hatfield and Prueger, 2015; Hatfield et al., 2015; Bisbis et al., 2018; Grotjahn, 2021). The 50 timing of heatwaves is particularly important, as extreme heat is more damaging during critical phenological 51 stages (Eyshi Rezaei et al., 2015; Fontana et al., 2015; Mäkinen et al., 2018; Wang et al., 2017). Extreme 52 canopy temperatures, rather than 2 m air temperatures, may be a more robust biophysical indicator of heat 53 impacts on crop production (Siebert et al., 2017). Heat stress indices based upon temperature and humidity 54 determine livestock productivity as well as conception and mortality rates (Dash et al., 2016; Key et al., 55 2014; Pragna et al., 2017; Rojas-Downing et al., 2017). Do Not Cite, Quote or Distribute 12-17 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 2 Heat extremes factor in mortality, morbidity and the range of some thermally sensitive ecosystem species 3 (Smith and Nagy, 2015; Ratnayake et al., 2019; Thomsen et al., 2019). Combined heat and drought stress 4 can reduce forest and grassland primary productivity (Ciais et al., 2005; De Boeck et al., 2018) and even 5 cause tree mortality at higher extremes (Teskey et al., 2015). 6 7 Extreme heat events raise temperatures in buildings and cities already warmed by the urban heat-island 8 effect (Gaffin et al., 2012; Oleson et al., 2018; Zhao et al., 2018; Mauree et al., 2019; Box 10.3) and can 9 induce disruptions in critical infrastructure networks (Chapman et al., 2013). Heat affects transportation 10 infrastructure by warping roads and airport runways (Chinowsky and Arndt, 2012) or buckling railways 11 (Dobney et al., 2010; Dépoues, 2017; Chinowsky et al., 2019), and high temperatures reduce air density 12 leading to aircraft take-off weight restrictions (Coffel et al., 2017; Palko, 2017; Zhou et al., 2018b). Heat 13 extremes increase peak cooling demand and challenge transmission and transformer capacity (Sathaye et al., 14 2013; Russo et al., 2016; Craig et al., 2018; Gao et al., 2018b) and may cause transmission lines to sag or fail 15 (Gupta et al., 2012). Thermal and nuclear electricity plants may be challenged when using warmer river 16 waters for cooling or when mixing waste waters back into waterways without causing ecosystem impacts 17 (Kopytko and Perkins, 2011; van Vliet et al., 2016; Tobin et al., 2018). Extreme temperature can also reduce 18 solar photovoltaic efficiency (Jerez et al., 2015). 19 20 21 12.3.1.3 Cold spells 22 23 The magnitude and timing (relative to developmental stages) of cold extremes (such as the typical coldest 24 day of the year) set limits in the range of species habitat for ecosystems as well as for agricultural and forest 25 pests (Osland et al., 2013; Cavanaugh et al., 2014; Parker and Abatzoglou, 2016; Brunner et al., 2018; 26 Unterberger et al., 2018). Cold air outbreaks can lead to chilling injuries for crops (even above 0℃) and may 27 kill outdoor livestock (particularly young animals) (Mader et al., 2010; Liu et al., 2013; Grotjahn, 2021), but 28 are often necessary for crop chill requirements (Dennis and Peacock, 2009). 29 30 Increases in human mortality can occur on exceptionally cold days (e.g., <1st percentile of temperatures in 31 winter) although thresholds and human-perceived temperatures linked to wind speed (i.e., ‘wind chill’) vary 32 geographically due to acclimatization (Li et al., 2013b, 2018c; Gao et al., 2015; Zhu et al., 2019a). The 33 timing of ‘unseasonal’ cold spells also affect human health (Kinney et al., 2015b). Extreme cold can increase 34 heat and electricity demand (Stuivenvolt-Allen and Wang, 2019), cause water pipes to burst , and 35 mechanically alter roads, railroads and buildings (Underwood et al., 2017). 36 37 38 12.3.1.4 Frost 39 40 Frost (Tmin < 0 ℃) is a natural and fundamental aspect of many ecosystems, with more extreme conditions 41 defined as ice (or icing) days (Tmax < 0 ℃) (Vincent et al., 2018c). Agricultural systems planning (e.g., 42 planting calendars, seed selection, or the opportunity to double-crop) requires information about the start and 43 end of the frost-free season (Wypych et al., 2017; Wolfe et al., 2018). Crops and wild plants can be directly 44 damaged by frost, but hard or killing frosts (at a threshold several degrees below freezing) can kill crops or 45 lower harvest quality depending on duration (which relates to soil temperature penetration) and plant 46 developmental stage (Crimp et al., 2016; Cradock-Henry, 2017; Li et al., 2018c; Mäkinen et al., 2018; 47 Grotjahn, 2021). Earlier disappearance of snow cover reduces natural insulation that protects plants and 48 burrowing animals from hard frost damages (Trnka et al., 2014; Mäkinen et al., 2018). In some cases an 49 early season warm spell may reduce plant hardiness or induce fruit tree flowering that exposes plants to 50 devastating subsequent frost impacts (Hufkens et al., 2012; Hatfield et al., 2014; Tripathi et al., 2016; 51 Brunner et al., 2018; DeGaetano, 2018; Unterberger et al., 2018; Wolfe et al., 2018). Shifts in the 52 seasonality of frozen soils also affect groundwater recharge and surface streamflow for water resource 53 applications, particularly when peak precipitation is shifted to a season that no longer has frozen soils 54 (Jyrkama and Sykes, 2007). 55 Do Not Cite, Quote or Distribute 12-18 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Regional information about the spring and autumn seasonal periods in which freeze-thaw cycles are common 2 (such as the dates of first spring thaw and last spring frost, or the number of days where Tmax > 0 ℃ and Tmin 3 < 0 ℃) are particularly useful in estimating the rate of potential road and building damages or determining 4 seasonal truck weight restrictions (Kvande and Lisø, 2009; Chinowsky and Arndt, 2012; Palko, 2017; Daniel 5 et al., 2018). The altitude of the freezing level also identifies portions of mountain slopes where freeze/thaw 6 transitions or changes in snowpack condition can influence landslide and snow avalanche hazards (Coe et al., 7 2018). The geographical distribution of frost is also a determining factor in the range of vectors for human 8 diseases such as malaria (Zhao et al., 2016a; Smith et al., 2020). 9 10 Figure 12.3 illustrates how successive heat and cold hazards can potentially affect important natural and 11 human systems, with climatic pressures reaching new sectoral assets or becoming increasingly severe as 12 conditions become more extreme. While the precise value of any CID threshold may depend strongly on 13 local environmental and system characteristics, there are common patterns and interdependencies in the 14 types of thresholds encountered. Changes in the regional profile of CIDs can thus substantially alter 15 threshold exceedance likelihoods. 16 17 18 [START FIGURE 12.3 HERE] 19 20 Figure 12.3: Conceptual illustration of representative climatic impact-driver thresholds showing how 21 graduating thresholds affect successive sectoral assets and lead to potentially more acute hazards as 22 conditions become more extreme (exact values are not shown as these must be tailored to reflect 23 diverse vulnerabilities of regional assets). Representative threshold definitions (T = instantaneous 24 temperature; 𝑇̅ = mean temperature): Cities and Infrastructure: Ttrans = temperature at which energy 25 transmission lines efficiency reduced; Taircraft = temperature at which aircraft become weight-restricted for 26 takeoff; Thotroads = temperature above which roads begin to warp; Tstream = temperature at which streams 27 are not capable of adequately cooling thermal plants; CDDmin = Minimum temperature for calculating 28 cooling degree days; HDDmin = maximum temperature for calculating heating degree days; T ice = 29 temperature at which ice threatens transportation; 𝑇̅permafrost = mean seasonal temperature above which 30 permafrost thaws at critical depths; Tcoldroads = temperature below which road asphalt performance suffers. 31 Health: Tdeadly = temperatures above which prolonged exposure may be deadly (often combined with 32 humidity for heat indices); Tsevere = temperatures above which prolonged exposure may cause elevated 33 morbidity; 𝑇̅blooms = mean temperature for harmful algal or cyanobacteria blooms; Tdanger = level of 34 dangerous cold temperatures (often combined with wind for chill indices); Toverwinter = temperature below 35 which disease vector species cannot survive winter. Ecosystems (CID indices for air and ocean 36 temperature): Thotlim and Tcoldlim = limiting hot and cold temperatures for a given species range; T frost = 37 frost threshold; 𝑇̅max and 𝑇̅min = maximum and minimum suitable annual mean temperatures for a given 38 species; Tcrit = critical temperature above which a given species is stressed. Agriculture: Thotlim= 39 temperature above which a crop or livestock species dies; T hotpest = maximum (or ‘lethal’) temperature 40 above which an agricultural pest/disease/weed cannot survive; T crit = temperature at which productivity 41 for a given crop is depressed; 𝑇̅opt = optimal mean temperature for a given plant’s productivity; GDD min = 42 threshold temperature for growing degree days determining plant development; T chill = temperature below 43 which chilling units are accumulated; Tfrost = temperature below which frost occurs; Thfrost = temperature 44 below which a hard frost threatens crops or livestock; Tcoldpest = minimum wintertime temperature below 45 which a given agricultural pest cannot survive; T coldlim = minimum temperature below which a given crop 46 cannot survive. 47 48 [END FIGURE 12.3 HERE] 49 50 51 12.3.2 Wet and dry 52 53 12.3.2.1 Mean precipitation 54 55 Changes in mean precipitation alter total water resources and long-term surface, snowpack, and groundwater 56 reservoirs (Schewe et al., 2014a). Annual and seasonal wet trends can alter the suitable geographic range of 57 species, with implications for biodiversity and vector borne disaeses (Knouft and Ficklin, 2017; Smith et al., Do Not Cite, Quote or Distribute 12-19 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 2020). The rate at which higher total stream flow increases river erosion and changes sediment loading is 2 relevant for fish breeding (Scheurer et al., 2009), the location of riverine salt fronts that affect coastal 3 agriculture and ecosystems (Chun et al., 2018; Vu et al., 2018), coastal freshwater stratification (Baker- 4 Austin et al., 2013; Bell et al., 2013), and the accretion of sediment in estuaries and beaches (Syvitski and 5 Milliman, 2007). Wetter conditions may shift tourist appeal (Kovács et al., 2017) and alter the pace of 6 degradation for paved and especially unpaved roads (Chinowsky and Arndt, 2012). 7 8 Many agricultural systems require minimum rainfall totals or rely upon irrigation (Mbow et al., 2019). The 9 length of the wet season helps determine the potential for multiple cropping seasons, but inconsistency of 10 wet season arrival times poses challenges for farm management (Waha et al., 2020). Wetter growing season 11 conditions increase the chance of water logging, which can delay planting or damage planted seeds 12 (Rosenzweig et al., 2002; Ben-Ari et al., 2018; Mäkinen et al., 2018; Wolfe et al., 2018; Kolberg et al., 2019; 13 Grotjahn, 2021). Tomasek et al. (2017) calculated ‘workable days’ for agricultural machinery around 14 planting and harvest time set in part by limits in soil moisture saturation below which farmers can utilize 15 critical machinery with less rutting or soil compaction. Wetter conditions may also increase canopy moisture 16 that is conducive to crop pathogens (Garrett et al., 2006; Kilroy, 2015; Grotjahn, 2021). 17 18 19 12.3.2.2 River flood 20 21 A large variety of climate indices and models are utilized to understand how river flooding affects both 22 natural or built environments with highly variable hazard thresholds, given unique local topography and 23 engineered defences such as dams and polders (Arnell and Gosling, 2016; Ekström et al., 2018).. Key 24 transportation routes, built infrastructure, and agricultural lands are threatened when floods exceed design 25 standards commonly based around flood magnitudes of a given historic return period (e.g., 1-in-100 yr flood 26 event), an annual exceedance probability, or precipitation intensity-duration-frequency relationships with key 27 indices (e.g., 10-day cumulative precipitation) related to catchment size and properties (Hirabayashi et al., 28 2013; Arnell and Lloyd-Hughes, 2014; Kundzewicz et al., 2014; Arnell and Gosling, 2016; Dikanski et al., 29 2016; Gosling and Arnell, 2016; Forzieri et al., 2017; Fluixá-Sanmartín et al., 2018; Koks et al., 2019). 30 Floods and high flow events can scour river beds and elevate silt loads, reducing water quality and 31 accelerating deposition in estuaries and reservoirs (Khan et al., 2018; Parasiewicz et al., 2019). Floods can 32 knock down, drown, or wash away crops and livestock, and partially-submerged plants can have yield 33 reduction depending on water turbidity and their development stage (Ruane et al., 2013; Shrestha et al., 34 2019). Basin snowpack properties may also be important during heavy rain events, as rain-on-snow events 35 can lead to rapid acceleration of flood stages that threaten wildlife and society (Hansen et al., 2014). 36 37 38 12.3.2.3 Heavy precipitation and pluvial flood 39 40 Heavy downpours can lead to pluvial flooding in cities, roadways, farmland, subway tunnels and buildings 41 (particularly those with basements) (Grahn and Nyberg, 2017; Palko, 2017; Pregnolato et al., 2017; Orr et 42 al., 2018). Heavy precipitation may overwhelm city transportation and storm water drainage systems, which 43 are typically designed using intensity-duration-frequency information such as the return periods for 1-, 6- or 44 24-hour rainfall totals (Kermanshah et al., 2017; Depietri and McPhearson, 2018; Rosenzweig et al., 2018; 45 Courty et al., 2019). Heavy rain events can directly cause leaf loss and damage or knock over crops, also 46 driving pollutant entrainment and erosion hazards in terrestrial ecosystems and farmland with downstream 47 ramifications for water quality (Hatfield et al., 2014; Segura et al., 2014; Li and Fang, 2016; Chhetri et al., 48 2019). The proportion of total precipitation that falls in heavy events also affects the percentage that is 49 retained in the soil column, altering groundwater recharge and deep soil moisture content for agricultural use 50 (Fishman, 2016; Lesk et al., 2020). 51 52 53 12.3.2.4 Landslide 54 55 Landslides, mudslides, rock falls, and other mass movements can lead to fatalities, destroy infrastructure and Do Not Cite, Quote or Distribute 12-20 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 housing stock, and block critical transportation routes. Climate models cannot resolve these complex slope 2 failure processes (nor triggering mechanisms such as earthquakes), so most studies rely on proxies or 3 conditions conducive to slope failure (Gariano and Guzzetti, 2016; Ho et al., 2017). Common indices include 4 precipitation intensity-duration thresholds (Brunetti et al., 2010; Khan et al., 2012; Melchiorre and Frattini, 5 2012) and thresholds related to antecedent wet periods and extreme rainfall intensities (Alvioli et al., 2018; 6 Monsieurs et al., 2019). Landslides and rockfalls may also be exacerbated by permafrost thaw and receding 7 glaciers in polar and mountain areas (Cook et al., 2016; Haeberli et al., 2017a; Patton et al., 2019). 8 9 10 12.3.2.5 Aridity 11 12 Aridity indices may track long-term changes in precipitation, evapotranspiration demand, surface water, 13 groundwater, or soil moisture (Sherwood and Fu, 2014; Herrera-Pantoja and Hiscock, 2015; Cook et al., 14 2020a). Changes in soil moisture and surface water can shift the rate of carbon uptake by ecosystems 15 (Humphrey et al., 2018) and alter suitable climate zones for wild species and agricultural cultivation (Feng 16 and Fu, 2013; Garcia et al., 2014; Huang et al., 2016b; Schlaepfer et al., 2017; Fatemi et al., 2018; IPCC, 17 2019c) as well as the prevalence of related pests and pathogen-carrying vectors (Paritsis and Veblen, 2011; 18 Smith et al., 2020). Water table depth, in relation to rooting depth, is also important for farms and forests 19 under dry conditions (Feng et al., 2006). A reduction in water availability (via aridity or hydrological 20 drought) challenges water supplies needed for for municipal, industrial, agriculture and hydropower use 21 (Schaeffer et al., 2012; Arnell and Lloyd-Hughes, 2014; Schewe et al., 2014b; Gosling and Arnell, 2016; van 22 Vliet et al., 2016). 23 24 25 12.3.2.6 Hydrological drought 26 27 Water managers often utilize a variety of hydrological drought indices and hydrological models to 28 characterize water resources, low flow conditions and the potential for irrigation (Wanders and Wada, 2015; 29 Mukherjee et al., 2018) (11.9). Low flow volume and intermittency thresholds can indicate reductions in 30 dissolved oxygen, more concentrated pollutants, and higher stream temperatures relevant for ecosystems, 31 water resource quality, and thermal power plant cooling (Feeley et al., 2008; Döll and Schmied, 2012; 32 Schaeffer et al., 2012; Prudhomme et al., 2014; van Vliet et al., 2016). Low water levels may also restrict 33 waterway navigation for commerce and recreation (Forzieri et al., 2018). 34 35 36 12.3.2.7 Agricultural and ecological drought 37 38 Agricultural and ecological drought indices relate to the ability of plants to meet growth and transpiration 39 needs (Zargar et al., 2011; Lobell et al., 2015; Pedro-Monzonís et al., 2015; Bachmair et al., 2016; Wehner et 40 al., 2017; Naumann et al., 2018; Table 11.3) and the timing and duration of droughts can lead to substantially 41 different impacts (Peña-Gallardo et al., 2019). Drought stress for agriculture and ecosystems is difficult to 42 directly observe, and therefore, scientists use a variety of drought indices (Table 11.3), proxy information 43 about changes in precipitation supply and reference evapotranspiration demand, the ratio of actual/potential 44 evapotranspiration or a deficit in available soil water content, particularly at rooting level (Williams et al., 45 2013; Trnka et al., 2014; Allen et al., 2015a; Svoboda and Fuchs, 2017; Mäkinen et al., 2018; Otkin et al., 46 2018). Severe water stress can lead to crop failure, in particular when droughts persist for an extended period 47 or occur during key plant developmental stages (Hatfield et al., 2014; Jolly et al., 2015; Leng and Hall, 48 2019). Projections of high wind speed and low humidity (even for just a portion of the day) can also inform 49 studies examining fruit desiccation and rice cracking (Grotjahn, 2021). Drought also raises disease infection 50 rates for West Nile Virus (Paull et al., 2017), and the alternation of dry and wet spells induces swelling and 51 shrinkage of clay soils that can lead to sinkholes and destabilize buildings (Hadji et al., 2014). 52 53 54 55 Do Not Cite, Quote or Distribute 12-21 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.3.2.8 Fire weather 2 3 Complex fire weather indices shed light on conditions that increase the likelihood of wildfire and shifts in 4 the fire season (Flannigan et al., 2013; Bedia et al., 2015; Jolly et al., 2015; Harvey, 2016; Littell et al., 2016; 5 Westerling, 2016; Abatzoglou et al., 2019) which pose particularly acute challenges for indigenous 6 communities (Christianson and McGee, 2019). Projection of future lightning frequency provides 7 information on an important natural triggering mechanism, particularly when coupled with long-term 8 warming and drying trends (Romps et al., 2014; Jin et al., 2015; Veraverbeke et al., 2017). Fuel aridity 9 metrics also help determine vegetative fuel desiccation and therefore the ignitability, flammability, and 10 spread of fires when they occur (Abatzoglou and Williams, 2016). The presence of snow cover can influence 11 the length of the fire season and the penetration of fire danger into new portions of the Arctic tundra (Young 12 et al., 2017; Abatzoglou et al., 2019). Data on the changing characteristics of local wind circulations like the 13 Santa Ana in California shed light on future intensity and spread patterns for fires (Jin et al., 2015). Fires also 14 produce smoke plumes that reduce air and water quality (via deposition), adversely affecting health, 15 visibility and water resources both near and far downwind (Dennekamp and Abramson, 2011; McKenzie et 16 al., 2014; Dreessen et al., 2016; Liu et al., 2016; Martin, 2016). 17 18 19 12.3.3 Wind 20 21 12.3.3.1 Mean wind speed 22 23 Changes in the speed and direction of prevailing winds can alter the profile of seed dispersal, windblown 24 pest and disease vectors, animal activities, and dust or pollen dispersal affecting ecosystems, agriculture, and 25 human health (Reid and Gamble, 2009; Bullock et al., 2012; Hellberg and Chu, 2016; Nourani et al., 2017). 26 Seasonal winds influence algal blooms, ecosystems and fisheries via lake mixing, ocean currents and coastal 27 upwelling (Bakun et al., 2015; Townhill et al., 2018; Woolway et al., 2020). Changes to wind density also 28 modify a region’s wind and wave renewable energy endowment (Schaeffer et al., 2012; Sierra et al., 2017; 29 Craig et al., 2018; Devis et al., 2018; Tobin et al., 2018; Yalew et al., 2020). Li et al. (2020) and Karnauskas 30 et al. (2018) evaluated wind thresholds at turbine height (~80-100 m above ground) including periods outside 31 of cut-in (2.5-3 m s-1) and cut-out (~25 m s-1) levels beyond which given turbines could not operate. 32 33 34 12.3.3.2 Severe wind storm 35 36 High winds associated with severe storms can level trees and houses, break plant stems and knock fruits, 37 nuts and grains to the ground, with tolerance thresholds depending on crop species and developmental stage 38 (Seidl et al., 2017; Lai, 2018; Elsner et al., 2019; Grotjahn, 2021). Severe storms particularly threaten 39 energy infrastructure, with maximum wind speed associated with treefall and breaking of above-ground 40 electrical transmission lines (Ward, 2013; Nik et al., 2020) . The profile of heavy wind gusts is also required 41 in the design of skyscrapers (Wang et al., 2013a) and bridges (Mondoro et al., 2018). Severe storms are 42 difficult to simulate at the relatively coarse spatial scales of earth system models, thus scientists often project 43 changes by noting areas with increased convective available potential energy (CAPE) and strong low-level 44 wind shear as these are conducive to tornado formation (Diffenbaugh et al., 2013; Tippett et al., 2016; Glazer 45 et al., 2020). 46 47 48 12.3.3.3 Tropical cyclone 49 50 Tropical cyclones and severe coastal storms can deliver wind, water, and coastal hazards with the potential 51 for widespread mortality and damages to cities, housing, transportation and energy infrastructure, 52 ecosystems, and agricultural lands (Burkett, 2011; NASEM, 2012; Bell et al., 2013; Wehof et al., 2014; 53 Ward et al., 2016; Cheal et al., 2017; Godoi et al., 2018; Koks et al., 2019; Pinnegar et al., 2019). Storm 54 planning is often tied to the Saffir-Simpson scale related to peak sustained wind speed (Izaguirre et al., 55 2021), with several indices focusing on storms’ overall power and energy, size and translation speed to Do Not Cite, Quote or Distribute 12-22 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 anticipate destructive potential (Knutson et al., 2015; Wang and Toumi, 2016; Parker et al., 2018; 2 Hassanzadeh et al., 2020). 3 4 5 12.3.3.4 Sand and dust storm 6 7 Sand and dust storms erode topsoils, damage crops and induce problems for health, transportation, 8 mechanical equipment and built infrastructure corresponding to the magnitude and duration of high winds 9 and particulate matter concentrations (Goudie, 2014; O’Loingsigh et al., 2014; Crooks et al., 2016; Barreau 10 et al., 2017; Bhattachan et al., 2018; Al Ameri et al., 2019; Middleton et al., 2019). Dust events may be 11 represented as the number of dust hours per dust storm year and by particulate matter (PM) concentrations 12 (Leys et al., 2011; Spickett et al., 2011; Hand et al., 2016). Photovoltaic panels can lose energy production 13 efficiency with dust accumulation (Patt et al., 2013; Javed et al., 2017). It is also useful to track dust storm 14 deposition of nutrients necessary for coral and tropical forest systems, but they may also feed algal blooms 15 harming lake and coastal ecosystems, health, and recreation (Jickells et al., 2005; Hallegraeff et al., 2014; 16 Gabric et al., 2016). Dust storms also cause air pollution and redistribute the soil-based fungus associated 17 with Valley Fever (Barreau et al., 2017; Coopersmith et al., 2017; Tong et al., 2017; Gorris et al., 2018). 18 19 20 12.3.4 Snow and ice 21 22 Cryospheric changes are a focus of Chapter 9 and were central to the recent IPCC Special Report on the 23 Ocean and Cryosphere in a Changing Climate (IPCC, 2019b). Here we focus on the ways that scientists use 24 snow and ice CIDs to understand current and future societal impacts and risks. 25 26 27 12.3.4.1 Snow, glacier and ice sheet 28 29 A large number of indices have been used in water resource and ecosystem studies to track changes in snow 30 under current and future climate conditions, including measurements of the snow water equivalent at key 31 seasonal dates, the fraction of precipitation falling as snow, the first and last days of snow cover, and cold 32 season temperatures (Mills et al., 2013; Pierce and Cayan, 2013; Berghuijs et al., 2014; Klos et al., 2014; 33 Musselman et al., 2017; Rhoades et al., 2018). Applications also examine shifts in seasonal streamflow for 34 snow-fed river basins (Mote et al., 2005; Pederson et al., 2011; Beniston and Stoffel, 2014; Coppola et al., 35 2014, Fyfe et al., 2017; Coppola et al., 2018; Islam et al., 2017; Knouft and Ficklin, 2017) as well as the 36 geographic extent of snow cover and the depth of frosts when snow cover’s natural insulation is absent 37 (Scheurer et al., 2009; Millar and Stephenson, 2015). Studies examining the impact of snow changes on 38 winter recreation and transportation have used thresholds of ~30 cm snow depth or snow water equivalent > 39 10 cm to determine the length of the season for alpine and cross-country skiing and snowmobiling (Damm et 40 al., 2017; Wobus et al., 2017b; Spandre et al., 2019; Steiger et al., 2019; Abegg et al., 2020). Changes in 41 snow quality also affect recreational activities (Rutty et al., 2017), and artificial snowmaking can augment 42 recreational snowpack depending on the number of suitable snowmaking hours (e.g., where WBGT < -2.2 43 °C; Wobus et al., 2017). Local detail may also be provided by tracking the seasonal rain-snow transition line 44 across space and elevation (Berghuijs et al., 2014). 45 46 Change in ice sheet and glacier spatial extent and surface mass balance is relevant for polar and high 47 mountain ecosystems and downstream assets that rely on glacial water resources (Lee et al., 2017a; Milner et 48 al., 2017; Huss and Hock, 2018; Schaefli et al., 2019). The loss of glaciers reduces the thermal consistency 49 of cold streams suitable for some freshwater species (Giersch et al., 2017), and parks and recreation areas 50 may lose appeal as glaciers and seasonal snow cover retreat (Gonzalez et al., 2018; Wang and Zhou, 2019). 51 Rapid glacial retreat can lead to glacial lakes and outburst floods that endanger downstream communities 52 (Carrivick and Tweed, 2016; Cook et al., 2016; Harrison et al., 2018). 53 54 55 Do Not Cite, Quote or Distribute 12-23 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.3.4.2 Permafrost 2 3 Changes in permafrost temperature, extent, and active layer thickness are metrics that track how permafrost 4 thaw below for e.g. roads, airstrips, rails, and building foundations in high-latitude and mountain regions 5 may destabilize settlements and critical infrastructure (Pendakur, 2016; Derksen et al., 2018; Duvillard et al., 6 2019; Olsson et al., 2019; Streletskiy et al., 2019). Warmer conditions can also affect ecosystems, built 7 infrastructure, and water resources through thawing of especially ice-rich permafrost (≥ 20% ice content) and 8 by thawing of ice wedges (Shiklomanov et al., 2017; Hjort et al., 2018), creation of thermokarst ponds and 9 increased subsurface drainage for polar and high-mountain wetlands (Walvoord and Kurylyk, 2016; 10 Farquharson et al., 2019) and the release of water pollutants such as mercury (Burkett, 2011; Schaeffer et al., 11 2012; Schuster et al., 2018) . 12 13 14 12.3.4.3 Lake, river and sea ice 15 16 Reductions in the duration of thick sea, lake and river ice influence ecosystems as well as ice fishing, 17 hunting, dog sledding, and snowmobiling; which are recreation activities for some but vital aspects of many 18 traditional indigenous communities (Durkalec et al., 2015; AMAP, 2017a; Baztan et al., 2017; Arp et al., 19 2018; Rokaya et al., 2018; Knoll et al., 2019; Meredith et al., 2019; Sharma et al., 2019). The seasonal extent 20 of thin ice and iceberg density also determines the viability of shipping lanes and seasonal roads (Valsson 21 and Ulfarsson, 2011; Pizzolato et al., 2016; AMAP, 2017; Mullan et al., 2017; Sturm et al., 2017), oil and 22 gas exploration timing (Schaeffer et al., 2012) and the seasonality of phytoplankton blooms (Oziel et al., 23 2017). Sea ice is a critical aspect of some ecosystems and fisheries (Massom and Stammerjohn, 2010; 24 Jenouvrier et al., 2014; Bindoff et al., 2019; Meredith et al., 2019). Various definitions of ‘ice free’ Arctic 25 Ocean conditions can be tailored to represent transportation needs, including thresholds of ice coverage (< 26 5% or < 30% or < 1 million km2) in September or over a 4-month period (Laliberté et al., 2016; Jahn, 2018). 27 28 29 12.3.4.4 Heavy snowfall and ice storm 30 31 Heavy snowfall is a substantial concern for cities, settlements, and key transportation and energy 32 infrastructure (Ward, 2013; Palko, 2017; Janoski et al., 2018; Collins et al., 2019). Heavy snowfall can 33 interfere with transportation (Herring et al., 2018) and cause a loss of both work and school days depending 34 on local snow removal infrastructure. Freezing rain and ice storms can be treacherous for road and air travel 35 (Tamerius et al., 2016), and can knock down power and telecommunication lines if ice accumulation is high 36 (Degelia et al., 2016). Rain-on-snow events can create a solid barrier that hinders wildlife and livestock 37 grazing that is important to indigenous communities (Forbes et al., 2016). Shifts in the frequency, seasonal 38 timing and regions susceptible to ice storms shift risks for agriculture and infrastructure (Lambert and 39 Hansen, 2011; Klima and Morgan, 2015; Ning and Bradley, 2015; Groisman et al., 2016). 40 41 42 12.3.4.5 Hail 43 44 Information on the changing frequency and size distribution of hail can help stakeholders build resilience for 45 agriculture, vehicles, transportation infrastructure and buildings, solar panels, and wild species that see 46 critical damage at particular hail size thresholds (Dessens et al., 2007; Webb et al., 2009; Patt et al., 2013; 47 Fiss et al., 2019). Most climate models do not directly resolve hail and therefore studies often examine 48 proxies associated with severe mesoscale storms (Tippett et al., 2015; Prein and Holland, 2018) although 49 some regional studies now utilize hail-resolving models (Mahoney et al., 2012; Brimelow et al., 2017). 50 51 52 12.3.4.6 Snow avalanche 53 54 Information about the changing frequency and seasonal timing of snow avalanches is important to assess 55 threats to transportation routes, infrastructure, recreational skiing, and people living in alpine communities Do Not Cite, Quote or Distribute 12-24 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 (Lazar and Williams, 2008; Mock et al., 2017; Ballesteros-Cánovas et al., 2018; Hock et al., 2019). Like 2 landslides and other mass movements, snow avalanches are not directly resolved by climate models and are 3 thus tracked using proxy climate information describing snow avalanche susceptibility, particularly the snow 4 water equivalent, and triggering mechanisms such as heatwaves, high winds, rain-on-snow and heavy 5 precipitation (Hock et al., 2019). The quality of snow also provides insight into avalanche hazards (Mock et 6 al., 2017), with the seasonal altitude of wet snowpack (>0 .5% liquid water by volume) particularly 7 important in determining characteristics of potential avalanches (Castebrunet et al., 2014). 8 9 10 12.3.5 Coastal 11 12 The recent IPCC Special Report on the Ocean and Cryosphere in a Changing Climate included in-depth 13 discussions of threats facing the world’s coastlines (IPCC, 2019b) and Section 9.6 provides further 14 discussion on coastal processes. Here we note major connections between coastal CIDs and ecosystem and 15 societal assets near coastlines. 16 17 18 12.3.5.1 Relative sea level 19 20 Sea level rise hazards for coastal ecosystems, infrastructure, farmland, cities and settlements in a particular 21 region are often driven by regional changes in relative sea level (RSL) that accounts for land uplift or 22 subsidence and thus represents local asset vulnerability better than global mean sea level (Hallegatte et al., 23 2013; Hinkel et al., 2013; McInnes et al., 2016; Weatherdon et al., 2016; Brown et al., 2018; IPCC, 2019b; 24 Rasoulkhani et al., 2020) (Box 9.1). Vertical land motion (i.e. land subsidence) caused by local fluid (gas or 25 groundwater) extraction can also have a large influence on relative sea levels (Minderhoud et al., 2020). 26 Several indices have been suggested to signify coastal inundation, including a threshold when the local land 27 elevation falls below the local mean higher-high water (MHHW) that is close to the ‘high tide’ level (Kulp 28 and Strauss, 2019) or a threshold when flooding occurs about once every two weeks (Sweet and Park, 2014; 29 Dahl et al., 2017b). RSL rise (or RSLR) can drive increased inland penetration of above-ground and 30 subterranean salt water fronts (i.e., salinity intrusion) affecting coastal ecosystems, agriculture, and water 31 resources (Ferguson and Gleeson, 2012; Kirwan and Megonigal, 2013; Rotzoll and Fletcher, 2013; Chen et 32 al., 2016; Colombani et al., 2016; Holding et al., 2016; Sawyer et al., 2016; Mohammed and Scholz, 2018). 33 The rate of RSLR can determine the survival and net pressure on niche coastal ecosystems such as 34 mangroves, tidal flats, sea grasses and coral reefs (Hubbard et al., 2008; Craft et al., 2009; Bell et al., 2013; 35 Kirwan and Megonigal, 2013; Alongi, 2015; Ellison, 2015; Lovelock et al., 2015; Ward et al., 2016; Lee et 36 al., 2018). 37 38 39 12.3.5.2 Coastal flood 40 41 Episodic coastal flooding of coastal communities, farmland, buildings, transportation routes, industry and 42 other infrastructure is caused by Extreme Total Water Levels (ETWL), which is the combination of RSL, 43 tides, storm surge and high wave setup at the shoreline (Vitousek et al., 2017; Melet et al., 2018; 44 Vousdoukas et al., 2018, 2020a; Koks et al., 2019; Kirezci et al., 2020). Coastal settlement and infrastructure 45 design often uses coastal flooding metrics such as the ETWL frequency distribution or the 100-year average 46 return interval storm tide (storm surge + high tide) level (McInnes et al., 2016; Mills et al., 2016; Walsh et 47 al., 2016b; Zheng et al., 2017). The duration of floods that overtop coastal protection, due to Extreme 48 Coastal Water Levels (ECWL) is important for port and harbour operations and coastal energy infrastructure 49 thresholds (Bilskie et al., 2016; Camus et al., 2017). Frequent inundation by salt water can also have 50 significant impacts on water resources, crops, aquaculture and transportation systems due to corrosion and 51 undercutting of coastal roads, bridges, and rails (Zimmerman and Faris, 2010; Ahmed et al., 2019b; 52 Gopalakrishnan et al., 2019). 53 54 55 12.3.5.3 Coastal erosion Do Not Cite, Quote or Distribute 12-25 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Effective management of coastal ecosystems, cities, settlements, beaches and infrastructure requires 2 information about coastal erosion driven by storm surge, waves, and sea level rise (Dawson et al., 2009; 3 Harley et al., 2017; Mentaschi et al., 2017 Hinkel et al., 2013). Coastal erosion is generally accompanied by 4 shoreline retreat, which can occur as a gradual process (e.g., due to sea level rise) or as an episodic event due 5 to storm surge and/or extreme waves, especially when combined with high tide (Ranasinghe, 2016). The 6 most commonly used shoreline retreat index is the magnitude of shoreline retreat by a pre-determined 7 planning horizon such as 50 or 100 years into the future. Commonly used metrics for episodic coastal 8 erosion include the beach erosion volume due to the 100 yr recurrence storm wave height, the full 9 exceedance probability distribution of coastal erosion volume (Li et al., 2014; Pender et al., 2015; 10 Ranasinghe and Callaghan, 2017) and the cumulative storm energy and storm power index (Godoi et al., 11 2018). The destruction or overtopping of barrier islands may lead to irreversible changes in the physical 12 system as well as in coastal ecosystems (Carrasco et al., 2016; Zinnert et al., 2019). Shoreline position 13 change rates along inlet-interrupted coasts may also be affected by changes in river flows and fluvial 14 sediment supply (Hinkel et al., 2013; Bamunawala et al., 2018; Ranasinghe et al., 2019). Permafrost thaw 15 and Arctic sea ice decline also reduce natural coastal protection from wave erosion for communities and 16 industry (Forbes, 2011; Melvin et al., 2017). 17 18 19 12.3.6 Oceanic 20 21 Oceanic changes and impacts were a substantial focus of the recent IPCC Special Report on the Ocean and 22 Cryosphere in a Changing Climate (IPCC, 2019b). Chapter 9 assesses changes in ocean processes, and here 23 we note major connections used by scientists to understand how oceanic CIDs affect ecosystems and society. 24 25 26 12.3.6.1 Mean ocean temperature 27 28 Shifts in thermal zones affect the suitability of fisheries and marine and coastal species habitat and migration 29 routes (Hoegh-Guldberg and Bruno, 2010; Doney et al., 2012; Burrows et al., 2014; Urban, 2015; Hixson 30 and Arts, 2016; Tripathi et al., 2016; Ahmed et al., 2019b; Bindoff et al., 2019). Intertidal species are 31 particularly dependent on suitable conditions for both air and sea surface temperatures (Monaco and 32 McQuaid, 2019). The structure of ocean warming also affects the intensity of upper-ocean stratification and 33 the timing and strength of coastal upwelling (driven also by mean wind changes), which alters the vertical 34 transport of oxygen- and nutrient-rich waters affecting fishery and marine ecosystem productivity (Wang et 35 al., 2015). 36 37 38 12.3.6.2 Marine heatwave 39 40 Marine heatwaves (MHW) push water temperatures above key thresholds and have been associated with 41 coral bleaching episodes, species shifts, and harmful algal blooms that can disrupt ecosystems, tourism and 42 human health (see Box 9.2; Wernberg et al., 2016; Arias-Ortiz et al., 2018; Oliver et al., 2018; Frölicher, 43 2019; Smale et al., 2019; Sully et al., 2019). The duration and return period of marine heatwaves provide 44 insight into aggregate stresses on marine species, fisheries and ecosystems, with various indices gauging 45 cumulative intensity or the number of days, weeks or months exceeding critical thresholds (Frieler et al., 46 2013; Frölicher et al., 2018; Hughes et al., 2018b; Cheung and Frölicher, 2020). Hobday et al. (2016) defined 47 marine heatwaves as the exceedance of the 30-year 90th percentile of the sea surface temperature (SST) 48 distribution on a given Julian day during 5 or more consecutive days, while Box 9.2, Figure 1 shows MHW 49 as an exceedance of 99th-percentile 11-day de-seasonalised SSTs. The return period of marine heatwaves is 50 also critical in determining a coral system’s ability to recover before the next event (Hughes et al., 2018a). 51 52 53 12.3.6.3 Ocean acidity 54 55 Uptake of atmospheric CO2 and subsequent increases in dissolved CO2 lowers ocean pH and can reduce Do Not Cite, Quote or Distribute 12-26 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 carbonate ion concentrations below critical calcium carbonate saturation thresholds for marine and aquatic 2 organisms growth, reproduction and/or survival with extended implications for marine ecosystems including 3 fisheries (Bell et al., 2013; Kroeker et al., 2013; Barange et al., 2014; Mathis et al., 2015a; Nagelkerken and 4 Connell, 2015; Dutkiewicz et al., 2015; Ekstrom et al., 2015; Gattuso et al., 2015; Nagelkerken and Munday, 5 2016; Tripathi et al., 2016; Behrenfeld et al., 2016b; Weiss et al., 2018; Jiang et al., 2018; Ahmed et al., 6 2019b; Bindoff et al., 2019). Lower pH may provide more favourable conditions for toxic algal blooms 7 (Riebesell et al., 2018) and can interact with hypoxic zones to impact ecosystems (Gobler and Baumann, 8 2016; Cai et al., 2017). 9 10 11 12.3.6.4 Ocean Salinity 12 13 Changes in currents, sea ice brine rejection and net freshwater flux in the ocean can alter salinity with effects 14 on mixed layer structure, density stratification, and the vertical movement of nutrients and marine organisms 15 (Freeland, 2013; Haumann et al., 2016). 16 17 18 12.3.6.5 Dissolved oxygen 19 20 Ocean warming and increased stratification decrease the oxygen content of the ocean (Griffiths et al., 2017; 21 Schmidtko et al., 2017; Bindoff et al., 2019), lead to an expansion of oxygen minimum zones in the open 22 ocean (Stramma et al., 2012; Zhang et al., 2013) and exacerbate the creation of anoxic “dead zones” in the 23 coastal oceans (Breitburg et al., 2018). Such a decline (characterized by successive dissolved oxygen 24 concentration thresholds) could affect a wide range of marine organisms and reduce marine habitats (Chan et 25 al., 2008; Vaquer-Sunyer and Duarte, 2008; Hoegh-Guldberg and Bruno, 2010; Altieri and Gedan, 2015; 26 Breitburg et al., 2018) and can also lead to further local acidification (Zhang and Gao, 2016; Laurent et al., 27 2017). 28 29 30 12.3.7 Other climatic impact-drivers 31 32 12.3.7.1 Air pollution weather 33 34 Although future air pollution will be strongly driven by air quality policies, anthropogenically-driven 35 changes to temperature, humidity, precipitation, and synoptic patterns have the potential to affect the 36 emissions, production, concentration and transport of particulate matter (e.g., from dust, fires, pollen), and 37 gaseous pollutants such as sulphur dioxide, tropospheric ozone, and nitrogen dioxide (see Section 6.5) with 38 resulting impacts on human health, agriculture and ecosystems (Ren et al., 2011; Fiore et al., 2015; Kinney et 39 al., 2015a; Tian et al., 2016; Orru et al., 2017; Emberson et al., 2018; Hayes et al., 2020). Information about 40 conditions leading to poor air quality is also important for visibility in natural parks and tourist locations 41 (Yue et al., 2013; Val Martin et al., 2015), as well as the efficiency of solar photovoltaic panels (Sweerts et 42 al., 2019). Relevant information about conditions favouring air pollution includes tracking warmer 43 conditions that accelerate ozone formation (Peel et al., 2013; Schnell et al., 2016) and the frequency and 44 duration of stagnant air events (Horton et al., 2014; Fann et al., 2015; Lelieveld et al., 2015; Vautard et al., 45 2018), although no regional index has proven sufficient to capture regional changes or acute events (Kerr 46 and Waugh, 2018; Schnell et al., 2018) whereas precipitation and moister air tend to reduce pollution 47 (Section 6.5). 48 49 50 12.3.7.2 Atmospheric Carbon Dioxide (CO2) at surface 51 52 CO2 is a well-mixed greenhouse gas with global repercussions on the Earth’s energy balance; however, 53 atmospheric CO2 concentration changes at the land surface also affect plant functions within ecosystems and 54 agriculture (see also Chapter 5). High CO2 concentration can increase photosynthesis rates and primary 55 production within natural ecosystems (Norby et al., 2010; Ratliff et al., 2015; Zhu et al., 2016) and Do Not Cite, Quote or Distribute 12-27 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 agricultural crops (Hatfield et al., 2011; Leakey et al., 2012; Bell et al., 2013; Glenn et al., 2014; 2 Nagelkerken and Connell, 2015; Behrenfeld et al., 2016a; Deryng et al., 2016; Kimball, 2016). High CO2 3 concentration affects total biomass and plant sugar content important to bioenergy production (Schaeffer et 4 al., 2012), but also helps some pests and weeds flourish (Hamilton et al., 2005; Wolfe et al., 2008; Valerio et 5 al., 2013; Korres et al., 2016; Stinson et al., 2016; Ramesh et al., 2017) while potentially shifting the 6 effectiveness of herbicides (Varanasi et al., 2016; Refatti et al., 2019). Higher CO2 concentration reduces 7 transpiration losses during drought conditions (Cammarano et al., 2016; Deryng et al., 2016; Swann et al., 8 2016; Durand et al., 2018), which also changes the energy balance within the plant canopy (Webber et al., 9 2017). Higher CO2 reduces the nutritional density of crops and forage lands (Loladze, 2014; Müller et al., 10 2014; Myers et al., 2014, 2017; Li et al., 2016c; Lee et al., 2017b; Smith and Myers, 2018; Zhu et al., 2018; 11 Beach et al., 2019) and can increase the production of toxins (Ziska et al., 2007) and allergenic pollen 12 (Schmidt, 2016). 13 14 15 12.3.7.3 Radiation at surface 16 17 Changes in surface solar and longwave radiation fluxes alter photosynthesis rates and potential 18 evapotranspiration for natural ecosystems and food, fibre, and energy crops (Mäkinen et al., 2018) and shift 19 solar energy resources (Schaeffer et al., 2012; Jerez et al., 2015; Wild et al., 2015; Fant et al., 2016; Craig et 20 al., 2018). Plants and aquatic systems particularly respond to changes in photosynthetically-active radiation 21 (PAR) and the fraction of diffuse radiation (Proctor et al., 2018; Ren et al., 2018; Ryu et al., 2018). Increases 22 in ultraviolet radiation can also detrimentally affect ecosystems and human health (Barnes et al., 2019). 23 24 25 12.3.7.4 Additional relevant climatic impact-drivers 26 27 Additional CIDs may be relevant for regional studies but are not the focus of assessment in this report. For 28 example, information about changes in the frequency and seasonal timing of fog helps anticipate airport 29 delays and cool beach days, and is also important for water delivery and retention in coastal ecological and 30 agricultural systems (Torregrosa et al., 2014). 31 32 Threats to many sectoral assets and associated systems may also be compounded when multiple hazards 33 occur simultaneously in the same place, affect multiple regions at the same time, or occur in a sequence that 34 may amplify overall impact (see Section 11.8; Clarke et al., 2018; IPCC, 2012a; Raymond et al., 2020; 35 Zscheischler et al., 2018). There is emerging literature on many connected extremes and their associated 36 hazards (e.g., climatic conditions that could drive multi-breadbasket failures; Kornhuber et al., 2019; Trnka 37 et al., 2019), but a full accounting is not practical here especially considering the many possible CID 38 combinations and the need to assess how exposed systems would be vulnerable to compound CIDs (assessed 39 in Working Group II). Table 12.2 is once again instructive here in considering hazard-related storylines, as 40 the multiple CIDs affecting a given sectoral asset (assessing across a row of Table 12.2) point to potentially 41 dangerous hazard combinations. Similarly, change in a single CID has the potential to affect multiple 42 sectoral assets (assessing down a column of Table 12.2) in a manner with broader systemic implications (see 43 AR6 WGII). 44 45 Recent literature defines CID indices to represent trends and thresholds that influence sectoral assets, 46 albeit with considerable variation owing to the unique characteristics of regional and sectoral assets. 47 Indices include direct information about the CID’s profile (magnitude, frequency, duration, timing, 48 spatial extent) or utilize atmospheric conditions as a proxy for CIDs that are more difficult to directly 49 observe or simulate. Each sector is affected by multiple CIDs, and each CID affects multiple sectors. 50 Assets within the same sector may require different or tailored indices even for the same CID. These 51 indices may be defined to capture graduated thresholds associated with tipping points or inflection 52 points in a particular sectoral vulnerability, with commonalities in the types of processes these 53 thresholds represent even as their precise magnitude may vary by specific sectoral system and asset. 54 55 Do Not Cite, Quote or Distribute 12-28 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4 Regional information on changing climate 2 3 This section describes the historical and projected changes in commonly used indices and thresholds 4 associated with the main climatic impact-drivers (see 12.2 and 12.3) at the scale of AR6 regions described in 5 (Figure 1.18a). The section is organised by continents (12.4.1-12.4.6) with a specific assessment for small 6 islands (12.4.7), open and deep ocean (12.4.8), and Polar regions (12.4.9) as defined in Chapter 1 (Figure 7 1.18c). In addition, CID indices and thresholds relevant to and Specific zones, as defined in WGII AR6 8 “Cross-Chapter Papers”, are assessed in Section 12.4.10 except for the Mediterranean, which is addressed 9 both under Africa and Europe (12.4.1 and 12.4.5) and is a focus in 10.6.4. 10 11 Regional assessment method and tables: In each section herein (12.4.1 – 12.4.10), we assess changes in 12 sector-relevant CIDs following the main CID categories defined in Section 12.2 through commonly used 13 indices and thresholds relevant for sectors described in 12.3. Sections 12.4.1-12.4.9 each include a summary 14 qualitative CID assessment table (Tables 12.3-12.11) showing the confidence levels associated with the 15 direction of projected CID changes (i,e increasing or decreasing) for the mid-century period (2041-2060) 16 relative to the recent past, for scenarios RCP4.5, SSP2-4.5, SRES A1B, or above (RCP6.0, RCP8.5, SSP3- 17 7.0, SSP5-8.5, SRES A2), which approximately encompasses Global Warming Levels of 2.0°C to 2.4°C (as 18 best estimate, see Chapter 4, Table 4.5). For Scenarios RCP2.6, SSP1-2.6 or SSP1-1.9, the signal may have 19 lower confidence levels in some cases due to smaller overall changes, embedded in a similar internal 20 variability, and to the availability of relatively few studies that account for these scenarios. Nevertheless, 21 CID changes under these lower emissions scenarios are included in the text whenever information is 22 available. For each cell in Tables 12.3-12.11, literature is assessed, aided by global Figure 12.4 or regional 23 Figures 12.5-12.10. Confidence in projections is established considering evidence emerging from 24 observations, attribution and projections, as explained in CCBox 10.3 while considering the amount of 25 evidence and agreement across models and studies and model generations. 26 27 The confidence levels associated with the directions of projected CID changes are synthesized assessments 28 based on literature that may utilize different indices and baseline periods or projections by global warming 29 levels. For extreme heat, cold spell, heavy precipitation and drought CIDs that are assessed in Chapter 11, 30 here we draw projections from the 2°C global warming level tables in Section 11.9. In some cases, more 31 details are needed in order to emphasize one aspect of projected CID change. For instance, the change in a 32 CID may be different for intensity, duration, frequency; or there can be strong sub-regional or seasonal 33 signals; or different CID indices may have conflicting signals. A footnote is added in such cases, but a 34 confidence level for a direction of projected change is given based on the 12.3 assessment of aspects of 35 regional CID change that are most relevant for impacts and for risks. As an example, tropical cyclones are 36 increasing in intensity but decreasing in frequency in some regions. Here, in assessing the confidence of the 37 direction of projected change in the Tropical cyclone CID (i.e. the colour of the table cell), we assign more 38 weight to the “intensity” rather than the “frequency”, corresponding to the higher relevance of the intensity 39 major tropical cyclones for risk assessment. Low confidence of changes, arising from lack of evidence, 40 strong spatial or seasonal heterogeneity, or lack of agreement are represented by colour-less cells, and, for 41 the sake of simplicity, only two categories of confidence are given: “medium confidence” and “high 42 confidence” (and higher). In addition, CID assessment tables also indicate observed or projected emergence 43 of the CID change signal from the natural inter-annual variability if assessed with at least medium 44 confidence in Section 12.5.2, using as a basis a criterion of S/N > 1, noise being defined as the interannual 45 variability. The time of emergence (ToE) is given as either: (i) already emerged in the historical period, or 46 (ii) emerging by 2050 at least for RCP8.5/SSP5-8.5, or (iii) emerging after 2050 but before 2100 at least for 47 scenarios RCP8.5 or SSP5-8.5. Table cells that do not include emergence information are indicative of “low 48 confidence of emergence in the 21st century”, which includes situations where assessment indicates 49 emergence will not occur before 2100 or that evidence is not available or insufficient for a confidence 50 assessment of time of emergence. 51 52 Figures: The assessment of changes in CIDs is based on literature, physical understanding (Chapters 2-11), 53 and global and regional climate projections of indices and thresholds presented in the Atlas, as well as in the 54 global and regional figures in 12.4 (Figures 12.4-12.10) showing the future evolution of 9 key CID 55 indices/thresholds used in this assessment (see also Cross-Chapter Box 10.3). The figure indices and Do Not Cite, Quote or Distribute 12-29 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 impacts-relevant thresholds are described in the Technical Annex VI on Climatic impact-drivers and 2 Extreme Indices. 3 4 Figure 12.4 shows changes in 6 CID indices. These global maps are derived from CMIP6 simulations for 5 different time periods and scenarios (except for Extreme Total Water Level where CMIP5 is used). The 6 uncertainty due to climate models, time, scenarios and regional downscaling is illustrated in Supplementary 7 Figures SM.12.1 to SM.12.6 which show the distribution of the spatial average of the index among models 8 over each land region for CMIP5, CMIP6 and CORDEX ensembles for the recent past, mid- and end-21st 9 century, and for GWLs +1.5ºC, +2ºC and +4ºC. The hatching in the figure covers areas where less than 80% 10 of models agree on the sign of change. 11 12 13 [START FIGURE 12.4 HERE] 14 15 Figure 12.4: Median projected changes in selected climatic impact-driver indices based on CMIP6 models. (a-c) 16 mean number of days per year with maximum temperature exceeding 35°C, (d-f) mean number of days 17 per year with the NOAA Heat Index (HI) exceeding 41°C, (g-i) number of negative precipitation anomaly 18 events per decade using the 6-month Standardised Precipitation Index, (j-l) mean soil moisture (%) and 19 (m-o) mean wind speed (%). (p-r) shows change in extreme sea level (1:100 yr return period total water 20 level from Vousdoukas et al. (2018)’s CMIP5 based dataset; meters). Left column is for SSP1-2.6, 2081- 21 2100, middle column is for SSP5-8.5 2041-2060, and right column SSP5-8.5, 2081-2100, all expressed as 22 changes relative to 1995-2014. Exception is extreme total water level which is for (p) RCP4.5 2100, (q) 23 RCP8.5 2050 and (r) RCP8.5 2100, each relative to 1980-2014. Bias correction is applied to daily 24 maximum temperature and HI data (see Technical Annex VI and Atlas.1.4.5). Uncertainty is represented 25 using the simple approach: No overlay indicates regions with high model agreement, where ≥80% of 26 models agree on sign of change; diagonal lines indicate regions with low model agreement, where <80% 27 of models agree on sign of change. For more information on the simple approach, please refer to the 28 Cross-Chapter Box Atlas.1. See Technical Annex VI for details of indices. Figures SM 12.1-12.6 show 29 regionally averaged values of these indices for the WGI reference AR6 regions for various model 30 ensembles, scenarios, time horizons and global warming levels. Further details on data sources and 31 processing are available in the chapter data table (Table 12.SM.1). 32 33 [END FIGURE 12.4 HERE] 34 35 36 Further regional detail is provided for the remaining indices in each continental section in the form of 37 continental maps accompanied by regional box plots displaying changes calculated for AR6 region averages, 38 and the associated regionally averaged uncertainty. 39 40 Climatic impact-drivers changing in a globally coherent way: For the sake of conciseness, assessments 41 pertaining to ocean acidity and the ‘Other’ CID type in 12.2 and 12.3 are not provided per region in Sections 42 12.4.1-9 but are summarised here given the globally coherent way in which they change. 43 44 Ocean acidity: Observations show increasing ocean acidification (robust evidence, high agreement), and it 45 is virtually certain that future ocean acidification will increase given future increases in greenhouse gases 46 (5.4). Areas below calcium carbonate saturation thresholds expanded from the 1990s-2010 and Meredith et 47 al. (2019) indicated that both the Southern and Arctic Oceans will experience year-round under-saturation 48 conditions by 2100 under RCP8.5. The vertical level of the aragonite saturation horizon off the Pacific coast 49 of North America has risen toward the surface by 30-50 m since pre-industrial times (Mathis et al., 2015b; 50 Feely et al., 2016). In a study of US coastlines, Ekstrom et al. (2015) mapped out the projected year when 51 aragonite saturation state drops below 1.5 (a sublethal threshold for bivalve mollusk larvae), finding 52 hazardous conditions before 2030 from northern Oregon to Alaska and before 2100 for the Pacific coast and 53 Atlantic coastline north of New Jersey. Mathis et al. (2015a) found that surface waters in the Beaufort Sea 54 have already dropped below aragonite saturation thresholds, projecting further declines and the Chukchi Sea 55 also dropping below saturation by ~2030. 56 57 Air pollution weather: The effect of climate change on air quality is assessed in Section 6.5 with limitations Do Not Cite, Quote or Distribute 12-30 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 for local planning explained in 6.1.3, and only a brief summary is given here. Section 6.5 notes that climate 2 change will have a small burden on particulate matter (PM) pollution (medium confidence) while the main 3 controlling factor in determining future concentrations will be future emissions policy for PM and their 4 precursors (high confidence). Surface ozone is sensitive to temperature and water vapor changes, but future 5 levels depend on precursor emissions. Although there is low confidence in precise regional changes (Section 6 6.5), climate change will generally introduce a surface O3 penalty (increasing concentrations with increasing 7 warming levels) over regions with high anthropogenic and/or natural ozone precursor emissions, while in 8 less polluted regions higher temperatures and humidity favor destruction of ozone (Schnell et al., 2016). 9 There is low confidence in changes to future stagnation events given the lack of robust projections of related 10 atmospheric conditions, such as future atmospheric blocking events (Sections 3.3.3, 8.4.2). The response of 11 regional air pollution to climate change will also be affected by other CIDs like fire weather, as well as by 12 ecosystem responses such as shifts in emissions by vegetation (Fiore et al., 2015). Section 6.5 assessed 13 medium confidence that climate driven changes to meteorological conditions generally favor extreme air 14 pollution episodes in heavily polluted environments, but noted strong regional and metric dependencies. 15 Given the dominant influence of future air quality policies, uncertainties around stagnation or blocking 16 events, and the potential contrasting regional changes of conditions favoring ozone and PM formation, 17 accumulation and destruction, cells in Tables 12.3-12.11 for air pollution weather are marked as low 18 confidence, and the reader is referred to Section 6.5 for further details. 19 20 Atmospheric CO2 at surface: Observations show rising atmospheric CO2 concentrations at the surface over 21 all earth regions (robust evidence, high agreement) (Sections 2.2, 5.1.1), and it is virtually certain that 22 surface atmospheric CO2 concentrations will continue to increase absent substantial changes to emissions 23 (Section 5.4). 24 25 Radiation at surface: Radiation has undergone decadal variations in past observations which are mostly 26 responding to the so-called dimming and brightening phenomenon driven by the increase and decrease of 27 aerosols. Over the last two decades or so, brightening continues in Europe and Northern America and 28 dimming stabilizes over South and East Asia and increases in some other areas (7.2.2.3). Future regional 29 shortwave radiation projections depend mostly on cloud trends, aerosol and water vapour trends, and 30 stratospheric ozone when considering UV radiation. Over Africa in 2050 and beyond, there is medium 31 confidence that radiation will increase in North and South Africa and decrease over the Sahara, North-East 32 Africa and West Africa (Wild et al., 2015, 2017; Soares et al., 2019; Tang et al., 2019b; Sawadogo et al., 33 2020a, 2020b). Over Asia, the CMIP5 multi-model-mean response shows that solar radiation will decrease in 34 South Asia and increase in East Asia (medium confidence) by the mid-century RCP8.5 (Wild et al., 2015, 35 2017; Ruosteenoja et al., 2019a). Projected solar resources show an increasing trend throughout the 21st 36 century in east Asia under RCP2.6 and RCP8.5 scenarios in CMIP5 simulations (Wild et al., 2015; Zhang et 37 al., 2018a; Shiogama et al., 2020) (medium confidence). More sunshine is projected over Australia in winter 38 and spring by the end of the century (medium confidence) with the increases in southern Australia exceeding 39 10% (CSIRO and BOM, 2015; Wild et al., 2015). In Central and South America, there is medium confidence 40 of increasing solar radiation over the Amazon Basin and the Northern part of South America (Wild et al., 41 2015, 2017; de Jong et al., 2019) (medium confidence). There is low confidence for an increase in surface 42 radiation in Central Europe, owing in particular to disagreement in cloud cover across global and regional 43 models (Jerez et al., 2015; Bartók et al., 2017; Craig et al., 2018), as well as water vapor. The treatment of 44 aerosol appears to be key in explaining these differences (Boé, 2016; Undorf et al., 2018; Boé et al., 2020; 45 Gutiérrez et al., 2020). Regional and global studies however indicate that there is medium confidence in 46 increasing radiation over Southern Europe and decreasing radiation over Northern Europe. Increasing 47 radiation trends are also found over Southern and Eastern U.S.A., and decreasing trends over North Western 48 North America (Wild et al., 2015; Losada Carreño et al., 2020), despite large differences between responses 49 from RCMs and GCMs over South and Eastern U.S.A. (low confidence), where, as for central Europe, the 50 role of aerosols appears important (Chen, 2021). Over polar regions there is medium confidence of a decrease 51 in radiation due to increasing moisture in the atmosphere and clouds (Wild et al., 2015). 52 53 54 55 Do Not Cite, Quote or Distribute 12-31 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.1 Africa 2 3 Previous IPCC assessments results are summarized in the Atlas section 4.1.1 For the purpose of this 4 assessment the Africa region has been divided in 9 sub regions of which 8 Sahara (SAH), Western Africa 5 (WAF), Central Africa (CAF), North East Africa (NEAF), South East Africa (SEAF), West Southern Africa 6 (WSAF), East Southern Africa (ESAF) and Madagascar (MDG) are the official AR6 regions (see Figure 7 Atlas.2) and one North Africa is used in this assessment to indicate the African portion of the Mediterranean 8 region. 9 10 Quite a large body of new literature is now available for the African climate as a result of regionally 11 downscaled CORDEX Africa outputs, in particular, providing projections of both the mean climate (Mariotti 12 et al., 2014; Nikulin et al., 2018; Dosio et al., 2019; Teichmann et al., 2020) and extreme climate phenomena 13 (Giorgi et al., 2014; Nikulin et al., 2018; Dosio et al., 2019; Coppola et al., 2021b). CORDEX Africa 14 simulations are assessed in the Atlas, which finds reasonable skill in mean temperature and precipitation as 15 well as important features of regional climate (e.g., timing of monsoon onset in West Africa) although lower 16 performance in Central Africa. 17 18 19 [START FIGURE 12.5 HERE] 20 21 Figure 12.5: Projected changes in selected climatic impact-driver indices for Africa. (a) Mean change in 1-in-100 22 year river discharge per unit catchment area (Q100, m 3 s-1 km-2) from CORDEX models for 2041-2060 23 relative to 1995-2014 for RCP8.5. (b) Shoreline position change along sandy coasts by the year 2100 24 relative to 2010 (meters; negative values indicate shoreline retreat) from the CMIP5 based data set 25 presented by Vousdoukas et al., (2020). (c) Bar plots for Q100 (m3 s-1 km-2) averaged over land areas for 26 the WGI reference AR6 regions (defined in Chapter 1). The left column within each panel (associated 27 with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute values in grey shades. The other 28 columns (associated with the right y-axis) show the Q100 changes relative to the recent past values for 29 two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for three global warming levels (defined 30 relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C (yellow) and 4°C (brown). The bars 31 show the median (dots) and the 10th-90th percentile range of model ensemble values across each model 32 ensemble. CMIP6 is shown by the darkest colors, CMIP5 by medium, and CORDEX by light. SSP5- 33 8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) Bar plots for shoreline position change 34 show CMIP5 based projections of shoreline position change along sandy coasts for 2050 and 2100 35 relative to 2010 for RCP8.5 (red) and RCP4.5 (blue) from Vousdoukas et al., (2020). Dots indicate 36 regional mean change estimates and bars the 5th-95th percentile range of associated uncertainty. Note that 37 these shoreline position change projections assume that there are no additional sediment sinks/sources or 38 any physical barriers to shoreline retreat. See Technical Annex VI for details of indices. Further details on 39 data sources and processing are available in the chapter data table (Table 12.SM.1). 40 41 [END FIGURE 12.5 HERE] 42 43 44 12.4.1.1 Heat and cold 45 46 Mean air temperature: The African continent has experienced increased warming since the beginning of 47 the 20th century in regions where measurements allow a sufficient homogeneous observation coverage to 48 estimate trends (high confidence) (Figure Atlas.11). This warming is very likely attributable to human 49 influence (Chapter 3, Atlas.4.2) at continental scale. Mean annual temperature have increased in recent 50 decades at a high rate since the mid-20th century, reaching 0.2-0.5°C/decade in some regions such as 51 north, north-eastern, west and south-western Africa (high confidence) (Atlas.4.2, Figure Atlas.11). 52 53 It is very likely that temperatures will increase in all future emission scenarios and all regions of Africa 54 (Atlas.4.4). By the end of century under RCP8.5 or SSP5-8.5, all African regions will very likely experience 55 a warming larger than 3°C except Central Africa where warming is very likely expected above 2.5°C under, 56 while under RCP2.6 or SSP1-2.6, the warming remains very likely limited to below 2°C (Figure Atlas.19). A 57 very likely warming with ranges between 0.5°C and 2.5°C is projected by the mid-century for all scenarios Do Not Cite, Quote or Distribute 12-32 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 depending on the region (high confidence). Mean temperature for all regions are projected to increase with 2 increasing global warming (virtually certain) (Figure Atlas.19). 3 4 Extreme heat: Warm extremes have increased in most of the regions (high confidence), NEAF and MDG 5 (medium confidence) and with low confidence in CAF and SEAF (Chapter 11, Table 11.4). Despite the 6 increasing mean temperature, there is low confidence (limited evidence) that Africa has experienced 7 increased extreme heat stress trend for agriculture or human health in the last two decades of the 20th 8 century in a few regions such as West Africa, South Africa and North Africa considering the period from 9 1973 to 2012 (Knutson and Ploshay, 2016). 10 11 A substantial increase in heatwave magnitude and frequency over most of the Africa domain is projected for 12 even 2℃ global warming (high confidence) (Chapter 11, 11.3,11.9, Table 11.4,), with potential effects on 13 health and agriculture. The number of days with maximum temperature exceeding 35°C is projected to 14 increase (Coppola et al., 2021b) in the range of 50-100 days by 2050 under SSP5-8.5 in WAF, ESAF and 15 WSAF and NEAF (high confidence) (Figure 12.4b). Under SSP1-2.6, the change in the number of 16 exceedance days remains limited to about 40-50 days per year at the end of the Century, at the end of the 17 century in these regions, while it increases by 150 days or more in WAF, CAF, NEAF for SSP5-8.5 18 (Figure 12.4a,c). 19 20 Mortality-related heat stress levels and deadly temperatures are very likely to become more frequent in the 21 future in RCP8.5/SSP5-8.5 and RCP4.5/SSP2-4.5 and for a 2oC global warming (Mora et al., 2017; 22 Nangombe et al., 2018; Sylla et al., 2018; Rohat et al., 2019; Sun et al., 2019b). In particular the equatorial 23 regions where heat is combined with higher humidity levels, but also North Africa, the Sahel and Southern 24 Africa (Figure 12.4d-f) are among the regions with largest increases of heat stress (Zhao et al., 2015; 25 Ahmadalipour and Moradkhani, 2018; Coffel et al., 2018). Mitigation scenario make a large difference in 26 frequency of exceedance of high heat stress indices thresholds (e.g., HI>41°C) by the end of the century 27 (Figure 12.4d-f) (Schwingshackl et al., 2021). In West Africa and Central Africa, under SSP5-8.5, the 28 expected number of days per year with HI>41°C will increase by around 200 days while in SSP1-2.6 such 29 exceedances are expected to increase by less than 50 days per year (Figure 12.4). 30 31 Cold spell and frost: Africa experiences cold events and frost days that can affect agriculture, infrastructure, 32 health and ecosystems, especially in South and North Africa, which have marked cold seasons, and 33 mountainous areas. Cold spells have likely decreased in frequency over subtropical areas. In particular, in 34 North and Southern Africa, the frequency of cold events has likely decreased in the last few decades (Chapter 35 11,11.3,11.9). There is a high confidence that cold spells and low target temperatures will decrease in future 36 climates under all scenarios in West, Central and East Africa. Heating degree days will have a substantial 37 decrease by the end of century for up to about one month under RCP8.5 in North and South Africa(Coppola 38 et al., 2021b) (high confidence). 39 40 There is high confidence that extreme heat has increased in frequency and intensity in most African 41 regions. Heatwaves and deadly heat stress and the frequency of exceedance of hot temperature 42 thresholds (e.g., 35°C) will drastically increase by the end of the century (high confidence) under 43 SSP5-8.5, but limited increases are expected in SSP1-2.6. Dangerous heat stress thresholds (HI>41°C) 44 are projected to be crossed more than 200 days more in West and Central Africa under SSP5-8.5 while 45 this increase remains limited to a few tens of days more for SSP1-2.6. Cold spells and frost days are 46 projected to occur less frequently in all scenarios. 47 48 49 12.4.1.2 Wet and dry 50 51 Mean precipitation: Since the mid-20th century, precipitation trends have varied in Africa but notable 52 drying trends are found in east, central and north eastern part of South Africa, Central Africa, and in the 53 Horn of Africa (Atlas 4.2). 54 55 There is high confidence in projected mean precipitation decreases in North Africa and West Southern Do Not Cite, Quote or Distribute 12-33 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Africa and medium confidence in East Southern Africa by the end of the 21st century (see also Dosio et al., 2 2019; Gebrechorkos et al., 2019; Teichmann et al., 2020) (Atlas.4.5). The West Africa and Northern and 3 Southern East Africa regions each feature a gradient in which precipitation decreases in the west and 4 increases in the east (medium confidence) (Atlas.4.5), with trends in West Africa affecting the boreal summer 5 monsoon (Chen et al., 2020). Increasing precipitation for a 1.5℃ and 2℃ global warming levels are found in 6 central and eastern Sahel with low confidence and the wet signal is getting stronger and more extended for a 7 3℃ and 4℃ warmer world (Atlas 1.1.5). 8 9 A change in monsoon seasonality is also reported in West Africa and Sahel (low confidence) with a forward 10 shift in time (later onset and end) (Mariotti et al., 2011; Seth et al., 2013; Ashfaq et al., 2020) (Chapter 8.2). 11 This shift has been associated with a precipitation decrease during the monsoon season attributed to a 12 decrease of African Easterly Wave activity in the 6-9 day regime (Mariotti et al., 2014) and a soil 13 precipitation feedback reported in Mariotti et al. (2011). 14 15 River flood: Generally in Africa from 1990 through 2014, annual flood frequencies have fluctuated and 16 there is medium confidence in an upward trend in flood events occurrences (Li et al., 2016a). In particular, 17 over West Africa, upward trends in hydrological extremes such as maximum peak discharge have likely 18 occurred during the last few decades (i.e., after 1980) and have caused increased flood events in riparian 19 countries of rivers such as Niger, Senegal and Volta (Nka et al., 2015; Aich et al., 2016; Wilcox et al., 2018; 20 Tramblay et al., 2020) (high confidence). In Southern Africa, trends in flood occurrences are decreasing prior 21 to 1980 and increasing afterwards (medium confidence) (Tramblay et al., 2020). 22 23 Under future climate scenarios, the extreme river discharge as characterized by the 30 year return period of 24 5-day average peak flow, is projected to increase by end of century for the RCP8.5 (more than 10% relative 25 to 1960-1999 period) for most of the tropical African river basins (Dankers et al., 2014) and a consistent 26 increase of flood magnitude across humid tropical Africa by 2050 for the AIB scenario (Arnell et al, 2014) 27 (medium confidence) (Figure 12.5). Specifically, in West Africa there is not univocal pattern of change for 28 future projections (Roudier et al., 2014) however under RCP8.5, there is a medium confidence of projected 29 increase of 20-year flood magnitudes by 2050 in countries within the Niger river basin (Aich et al., 2016) 30 and a low confidence (limited evidence) of an increase in extreme peak flows and their duration in countries 31 of the Volta river basin by 2050 and 2090 (Jin et al., 2018). A significant median change of flood magnitude 32 for the Gambia river (-4.5%) and for the Sessandra (+14.4%) and Niger (+6.1%) are projected under several 33 scenario between mid and end of century (Roudier et al., 2014). In East Africa, extreme flows are projected 34 to increase for region within the Blue Nile with low confidence (limited evidence) (Aich et al., 2014). 35 However, uncertainty due to the climate scenario dominates the projection of extreme flows (Aich et al., 36 2014; Krysanova et al., 2017) for the Blue Nile and Niger river basins. Averaged over the African continent 37 for different levels of global warming, the present-day 100-year return period flood levels will have a return 38 period of 40 years in 1.5℃ and 2℃ (Alfieri et al., 2017) and 21 years for 4℃ warmer climate (Hirabayashi 39 et al., 2013b; Alfieri et al., 2017). 40 41 Heavy precipitation and pluvial flood: Chapter 11 found that heavy precipitation intensity and frequency 42 has likely increased over West Southern Africa but there is no evidence due to a lack of studies that any 43 significant trend is observed in any other region. In addition, East Africa has experienced strong precipitation 44 variability and intense wet spells leading to widespread pluvial flooding events hitting most countries 45 including Ethiopia, Somalia, Kenya and Tanzania (medium confidence). Finally, with respect to Southern 46 Africa, heavy precipitations events have increased in frequency (medium confidence). 47 48 In West Africa and Central Africa, there is high confidence that the intensity of extreme precipitation will 49 increase in future climate under both RCP4.5 and RCP8.5 scenarios and 1.5oC and 2oC global warming 50 levels threatening for widespread flood occurrences before, during and after the mature monsoon season. 51 Extreme precipitation is also increasing in several other regions like SAH, NEAF, SEAF, ESAF and 52 Madagascar (high confidence) for 2oC GWL and higher (Chapter 11). 53 54 Landslide: There is an increase in reported landslides in WAF, CAF, NEAF and SEAF in the past decades 55 but low evidence of significant trends (Gariano and Guzzetti, 2016; Haque et al., 2019). There is low Do Not Cite, Quote or Distribute 12-34 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 confidence (limited evidence) of a future increase in landslides in Central-Eastern Africa, and literature is 2 largely missing to assess this important hazard (Gariano and Guzzetti, 2016). 3 4 Aridity: Section 11.9 assesses medium confidence in observed long-term declines of soil moisture and 5 aridity indices in several African regions (NAF, WSAF, ESAF, MDG). Trends in East Africa are not 6 definitive given uncertain balances between precipitation and potential evaporation (Kew et al., 2021). 7 Projected declines in precipitation and soil moisture trends indicate high confidence in increased aridity over 8 the 21st century in NAF, WSAF, ESAF, and MDG but low confidence elsewhere in Africa (Section 11.9; see 9 also Figure 12.4j-l; Gizaw and Gan, 2017). A growing number of studies provide further regional context on 10 expanding aridity in several places in East and West Africa, respectively (Sylla et al., 2016; Liu et al., 2018b; 11 Haile et al., 2020). 12 13 Hydrological drought: Section 11.9 noted observed decreases in hydrological drought over the 14 Mediterranean (high confidence) and diminished summertime river flows in West Africa (medium 15 confidence). Recent regional modeling studies project substantial increases in hydrological drought affecting 16 major West African river basins under 1.5oC and 2oC global warming levels and RCP4.5 and RCP8.5 17 scenarios; however there remains low confidence in future projections given disagreement with global model 18 runoff projections (e.g., Cook et al., 2020) (low confidence). There is high confidence that a 2 ℃ global 19 warming level would see an increase in hydrological droughts in the Mediterranean region, and medium 20 confidence in increasing hydrological drought conditions in the Southern Africa regions (Section 11.9). 21 22 Agricultural and ecological drought: Farmers and food security experts in East Africa have noted spatial 23 extensions in seasonal agricultural droughts in recent decades (Elagib, 2014), but it is difficult to disentangle 24 these trends from climate variability. In Ethiopia, past severe agricultural drought conditions in the northern 25 regions are moderately common events in recent years (Zeleke et al., 2017). In Southern Africa, the number 26 of “flash” droughts (with rapid onset and durations from a few days to couple of months) have increased by 27 220% between 1961 and 2016 as a result of anthropogenic warming (Yuan et al., 2018). Section 11.9 notes 28 medium confidence increases in agricultural and ecological drought trends in North, West and Central Africa 29 as well as both Southern Africa regions. The most striking drought is the Western Cape drought in 2015- 30 2018, a prolonged drought which resulted in acute water shortages (Wolski, 2018; Burls et al., 2019) 31 (Section 10.6.2). Anthropogenic climate change caused a threefold increase in the probability of such a 32 drought to occur (Botai et al., 2017; Otto et al., 2018) (see also Chapters 10 and 11). Section 11.9 assesses 33 increases in agricultural and ecological drought at 2℃ global warming level for North Africa and West 34 Southern Africa (high confidence) and for East Southern Africa and Madagascar (medium confidence), with 35 confidence generally rising for higher emissions scenarios (see also Sylla et al., 2016; Zhao and Dai, 2017; 36 Diedhiou et al., 2018; Abiodun et al., 2019; Todzo et al., 2020; Coppola et al., 2021). Liu et al. (2018) 37 identified the Southern Africa region as the drought ‘hottest spot’ in Africa in 1.5 and 2 ºC global warming 38 scenarios. 39 40 Fire weather: There is low confidence (low agreement) in recent reductions in fire activity given soil 41 moisture increases in some regions and substantial land use changes (Andela et al., 2017; Forkel et al., 2019; 42 Zubkova et al., 2019). Days prone to fire conditions are going to increase in all extratropical Africa for end 43 of century and fire weather indices are projected to largely increase in North and South Africa, where 44 increasing aridity trends occur (high confidence), with an emerging signal well before the middle of the 45 century where drought and heat increase will combine (Chapter 11) (Engelbrecht et al., 2015; Abatzoglou et 46 al., 2019). There is low confidence (low evidence) of fire weather changes for other African regions. 47 48 Total precipitation is projected to decrease in the northernmost (high confidence) and southernmost 49 regions of Africa (medium confidence), with West and East Africa regions each having a west-to-east 50 pattern of decreasing-to-increasing precipitation (medium confidence). Most African regions will 51 undergo an increase in heavy precipitation that can lead to pluvial floods (high confidence), even as 52 increasing dry climatic impact-drivers (aridity, hydrological, agricultural and ecological droughts, fire 53 weather) are generally projected in the North Africa and Southern African regions (high confidence) 54 and western portions of West Africa (medium confidence). 55 Do Not Cite, Quote or Distribute 12-35 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.1.3 Wind 2 3 Mean wind speed: Decreasing trends in wind speeds have occurred in many parts of Africa (McVicar et al., 4 2012; AR5 WGI) (low confidence due to observations with limited homogeneity). There is high confidence 5 in climate-change induced future decreasing mean wind, wind energy potential and strong winds in North 6 Africa and Mediterranean regions as a consequence of the poleward shift of the Hadley cell (Karnauskas et 7 al., 2018a; Kjellström et al., 2018; Sivakumar and Lucio, 2018; Tobin et al., 2018; Jung and Schindler, 2019) 8 in the RCP4.5 and RCP8.5 scenarios by the middle of the century or beyond, and for a global warming level 9 of 2°C or higher. Over West Africa and South Africa a future significant increase in wind speeds and wind 10 energy potential is expected (medium confidence) (Karnauskas et al., 2018a; Jung and Schindler, 2019) (see 11 also Figure 12.4m-o). 12 13 Severe wind storm: A limited number of studies allow an assessment of past trends in wind storms. In West 14 Africa and specifically in the Sahel band, more intense storms have occurred since the 1980s (low 15 confidence, limited evidence). A persistent and large increase of frequency of Sahelian mesoscale convective 16 storms has been found in several studies (Panthou et al., 2014; Taylor et al., 2017b), with consequences for 17 extreme rainfalls, and potentially on extreme winds (low confidence, limited evidence). There is low 18 confidence of a general increasing trend in extreme winds across West, Central, East and South Africa in a 19 majority of regions by the middle of the century even in high-end scenarios. The frequency of Mediterranean 20 wind storms reaching North Africa, including Medicanes, is projected to decrease, but their intensities are 21 projected to increase, by the mid-century and beyond under SRES A1B, A2 and RCP8.5, (Cavicchia et al.; 22 Walsh et al., 2014; Tous et al., 2016; Romera et al., 2017; Romero and Emanuel, 2017; González‐Alemán et 23 al., 2019) (Chapter 11) (medium confidence). 24 25 Tropical cyclone. In the Southern Indian ocean, an increase of Category 5 cyclones has been observed in 26 recent decades (Fitchett, 2018) as in other basins (Section 11.7). However, there is a projected decrease in 27 the frequency of tropical cyclones making landfall over Madagascar, South Eastern Africa and East South 28 Africa in a 1°C, 2°C and 3ºC warmer world (medium confidence) (Malherbe et al., 2013; Roberts et al., 29 2015, 2020; Muthige et al., 2018; Knutson et al., 2020). There is medium confidence in general increasing 30 intensities for cyclones in such studies for African regions. 31 32 Sand and dust storm: North Africa and the Sahel, and to a lesser extent South Africa, are prone to dust 33 storms, having consequences on health (Querol et al., 2019), transmission of infectious diseases (Agier et al., 34 2013; Wu et al., 2016), and solar power generation and related maintenance costs. There is limited evidence 35 and low agreement of secular 20th century trends in wind speeds or dust emissions (limited length of data 36 records, large variability). Dust variations are controlled by changes in surface winds, precipitation and 37 vegetation, which in turn are modulated at multiple time scales by dominant modes of internal climate 38 variability (see Chapter 10). In North Africa, wind variability explains both the observed high concentrations 39 between the 1970s and 1980s and lower concentrations thereafter (Ridley et al., 2014; Evan et al., 2016). 40 Yet, the effect of vegetation changes may not be negligible (Pu and Ginoux, 2017, 2018). 41 42 Changes to the frequency and intensity of dust storms also remain largely uncertain due to uncertainty in 43 future regional wind and precipitation as the climate warms, CO2 fertilization effects on vegetation (Huang et 44 al., 2017), and anthropogenic land use and land cover change due to land management and invasive species 45 (Ginoux et al., 2012; Webb and Pierre, 2018). Dust loadings and related air pollution hazards (from fine 46 particles that affect health) are projected to generally decrease in many regions of the Sahara and Sahel due 47 to the changing winds (Evan et al., 2016) and slightly increase over the Guinea Coast and West Africa (low 48 confidence) (Ji et al., 2018). 49 50 In summary, there is high confidence of a decrease in mean wind speed and wind energy potential in 51 North Africa and medium confidence of an increase in South and West Africa, by the middle of the 52 century regardless of climate scenario or global warming level equal or superior to 2°C, high 53 confidence of a decrease in frequency of cyclones landing in SEAF, ESAF and MDG, and low 54 confidence of a general increase in wind storms in most of African regions located Southward of Sahel. 55 The evolution of dust storms remains largely uncertain. Do Not Cite, Quote or Distribute 12-36 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.1.4 Snow and Ice 2 3 Snow and glacier: African glaciers are located in East Africa and more specifically on Mount Kenya, the 4 Rwenzori Mountains and Mount Kilimanjaro, with glaciers reducing substantially in each region (high 5 confidence) (Taylor et al., 2006; Cullen et al., 2013; Chen et al., 2018; Prinz et al., 2018; Wang and Zhou, 6 2019). Observation and future projection of African glacier mass changes are assessed in Section 9.5.1 7 within the Low Latitude glacier region, which is one of the regions with the largest mass loss even under 8 low-emission scenarios (assessment of this region is dominated by glaciers in the South American Andes, 9 however) (high confidence). Glaciers in the Low Latitude region will lose 67 ± 42%, 86 ± 24% and 94 ± 10 13% of their mass in 2015 by the end of the century for RCP4.5, RCP6.0 and RCP8.5 scenarios, respectively 11 (Marzeion et al., 2020). Cullen et al., (2013) calculated that even imbalances between the Mount 12 Kilimanjaro glaciers and present-day climate would be enough to eliminate the mountain’s glaciers by 2060. 13 Snow water equivalent and snow cover season duration also decline in the East African mountains, Ethiopian 14 Highlands, and Atlas Mountains with climate change (high confidence) (López-Moreno et al., 2017). 15 16 In conclusion, there is high confidence that African snow and glaciers have very significantly 17 decreased in the last decades and that this trend will continue over the 21st century. 18 19 20 12.4.1.5 Coastal and Oceanic 21 22 Relative sea level: Around Africa, over 1900-2018, a new tide-gauge based reconstruction finds a regional- 23 mean RSL change of 2.07 [1.36–2.77] mm yr-1 in the South Atlantic and 1.33 [0.80-1.86] mm yr-1 in the 24 Indian Ocean (Frederikse et al., 2020), compared to a GMSL change of around 1.7 mm yr-1 (Section 2.3.3.3; 25 Table 9.5). For the period 1993-2018, these RSLR rates, based on satellite altimetry, increased to 3.45 [3.04- 26 3.86] mm yr-1 and 3.65 [3.23-4.08] mm yr-1, respectively (Frederikse et al., 2020), compared to a GMSL 27 change of 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). 28 29 Relative sea-level rise is virtually certain to continue in the oceans around Africa. Regional-mean RSLR 30 projections for the oceans around Africa range from 0.4 m–0.5 m under SSP1-2.6 to 0.8 m–0.9 m under 31 SSP5-8.5 for 2081-2100 relative to 1995-2014 (median values), which is within the range of projected 32 GMSL change {Section 9.6.3.3}. These RSLR projections may however be underestimated due to potential 33 partial representation of land subsidence in their assessment (Section 9.6.3.2). 34 35 Coastal flood: The present day 1:100 yr Extreme Total Water Level is between 0.1 m – 1.2 m around Africa, 36 with values around 1 m or above along the South West, South East and Central East coasts (Vousdoukas et 37 al., 2018). 38 39 Extreme total water level (ETWL) magnitude and occurrence frequency are expected to increase throughout 40 the region (high confidence) (see Figure 12.4p-r and Figure SM 12.6). Across the region, the 5th – 95th 41 percentile range of the 1:100 yr ETWL is projected increase (relative to 1980 – 2014) by 7 cm – 36 cm and 42 by 14 cm – 42 cm by 2050 under RCP4.5 and RCP8.5, respectively. By 2100, this range is projected to be 28 43 cm – 86 cm and 43 cm – 190 cm under RCP4.5 and RCP8.5, respectively (Vousdoukas et al., 2018; Kirezci 44 et al., 2020). In terms of ETWL occurrence frequencies, the present day 1:100 yr ETWL is projected to have 45 median return periods of around 1:10-1:20 yrs by 2050 and 1:1-1:5 yrs by 2100 in Southern and North 46 Africa and occur more than once per year by 2050 and 2100 in most of East and West Africa under RCP4.5 47 (Vousdoukas et al., 2018). The present day 1:50 yr ETWL is projected to occur around 3 times a year by 48 2100 with a SLR of 1 m in Africa (Vitousek et al., 2017). 49 50 Coastal erosion: Shoreline retreat rates upto 1 m yr-1 have been observed around the continent during 1984 51 – 2015, except in ESAF which has experienced a shoreline progration rate of 0.1 m/r over the same period 52 (Luijendijk et al., 2018; Mentaschi et al., 2018). (Mentaschi et al., 2018) report a coastal area losses of 160 53 km2 and 460 km2 over a 30 year period (1984-2015) along the Atlantic and Indian ocean coasts of the 54 continent. At the more regional level, in Ghana along the Gulf of Guinea, about 79% of the shoreline was 55 found to be retreating while 21% was found to be stable or prograding over the period 1974–1996 (Addo and Do Not Cite, Quote or Distribute 12-37 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Addo, 2016). 2 3 Projections indicate that a vast majority of sandy coasts in the region will experience shoreline retreat 4 throughout the 21st century (high confidence), while parts of the ESAF and west MDG coastline are 5 projected to prograde over the 21st century, if present ambient trends continue. Median shoreline change 6 projections (CMIP5), relative to 2010, presented by Vousdoukas et al. (2020) show that, under RCP4.5, 7 shorelines in Africa will retreat by between 30 m (SAH, NEAF, WSAF, ESAF, MDG) and 55 m (WAF, 8 CAF), by mid-century. By the same period but under RCP8.5, the median shoreline retreat is projected to be 9 between 35 m (SAH, NEAF, WSAF, ESAF) and 65 m (WAF, CAF). By 2100, more than 100 m of median 10 retreat is projected in WAF, CAF and SEAF under RCP4.5, while under RCP8.5, more than 100 m of 11 shoreline retreat is projected in all regions except NEAF and WSAF. Under RCP8.5 especially, the 12 projected retreat by 2100 is greater than 150 m in WAF and CAF. The total length of sandy coasts in Africa 13 that is projected to retreat by more than a median of 100 m by 2100 under RCP4.5 and RCP8.5 is about 14 13,000 km and 17,000 km respectively, an increase of approximately 33%. 15 16 Marine heatwave: Over 1982-2016, the coastal oceans of Africa has experienced on average 2.0–3.0 MHW 17 per year, with the coastal oceans around the southern half of the continent experiencing on average 2.5–3 18 MHWs per year. The average duration was between 5 and 15 days (Oliver et al., 2018). Changes over the 19 20th century, derived from MHW proxies, show an increase in frequency between 0.5 and 2.0 MHW per 20 decade over the region, especially off the Horn of Africa; an increase in intensity per event around South 21 Africa; and an increase in MHW duration along the North African coastlines (Oliver et al., 2018). 22 23 There is high confidence that MHWs will increase around Africa. Mean SST, a common proxy for MHWs, is 24 projected to increase by 1 ºC (2ºC) around Africa by 2100, with a hotspot of around 2 ºC (5ºC) along the 25 coastlines of South Africa under RCP4.5 (RCP8.5) (see Interactive Atlas). Under global warming conditions, 26 MHW intensity and duration will increase in the coastal zones of all subregions of Africa (Frölicher et al., 27 2018). Projections for SSP1-2.6 and SSP5-8.5 both show an increase in MHWs around Australasia by 2081 28 – 2100, relative to 1985 – 2014 (Box 9.2, Figure 1). 29 30 In general, there is high confidence that most coastal and ocean related hazards in Africa will increase 31 over the 21st century. Relative sea-level rise is virtually certain to continue around Africa, contributing 32 to increased coastal flooding in low-lying areas (high confidence) and shoreline retreat along most 33 sandy coasts (high confidence). Marine Heatwaves are also expected to increase around the region 34 over the 21st century (high confidence). 35 36 The assessed direction of change in CIDs for Africa and associated confidence levels are illustrated in Table 37 12.3. No relevant literature could be found for permafrost and hail, although these phenomena may be 38 relevant in parts of the continent. 39 40 41 [START TABLE 12.3 HERE] 42 43 Table 12.3: Summary of confidence in direction of projected change in climatic impact-drivers in Africa, representing 44 their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, SRES A1B, or 45 above within each AR6 region (defined in Chapter 1), approximately corresponding (for CIDs that are 46 independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for more details 47 of the assessment method). The table also includes the assessment of observed or projected time-of- 48 emergence of the CID change signal from the natural inter-annual variability if found with at least 49 medium confidence in Section 12.5.2. 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-38 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region North Africa* 3 4 Sahara (SAH) 4 Western Africa (WAF) 1 1 1 1 4 Central Africa (CAF) 4 North Eastern Africa (NEAF) 1,2 1 1 1 4 South Eastern Africa (SEAF) 1 1 1 1 3 4 West Southern Africa (WSAF) 4 East Southern Africa (ESAF) 3 4,5 Madagascar (MDG) 3 4,5 1. Contrasted regional signal: drying in western portions and wettening in eastern portions 2. Likely increase over the Ethiopian Highlands Key 3. Medium confidence of decrease in frequency and increase in intensity High confidence of decrease 4. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. Medium confidence of decrease 5. Substantial parts of the ESAF and MDG coasts are projected to prograde if present-day ambient shoreline change rates continue * North Africa is not an official region of IPCC AR6, but assessment here is based upon the African portions of the Mediterranean Region Low confidence in direction of change Medium confidence of increase High confidence of increase Not broadly relevant 1 2 [END TABLE 12.3 HERE] 3 4 5 12.4.2 Asia 6 7 According to the region definitions given in Chapter 1, Asia is divided into 11 regions: the Arabian 8 Peninsula (ARP), Western Central Asia (WCA), West Siberia (WSB), East Siberia (ESB), the Russian Far 9 East (RFE), East Asia (EAS), East Central Asia (ECA), the Tibetan Plateau (TIB), South Asia (SAS), South- 10 East Asia (SEA) and the Russian Arctic Region (RAR). CID changes in RAR are assessed in the Polar 11 Region section (12.4.9). As assessed in previous IPCC reports, major concerns in Asia are associated 12 particularly with droughts and floods in all regions, heat extremes in SAS and EAS, sand-dust storms in 13 WCA, tropical cyclones in SEA and EAS, snow cover and glacier changes in ECA and the Hindu Kush 14 Himalaya (HKH) region, and sea ice and permafrost thawing in North Asia. 15 16 Since AR5, a large body of new literature is now available relevant to climate change in Asia, which 17 includes projections of both mean climate and extreme climate phenomena from global and regional 18 ensembles of climate simulations such as CMIP6 and CORDEX (see Chapter 10 and the Atlas). Literature 19 has also considerably grown on several climate topics relevant to Asia such as the mountain climate (see in 20 particular Chapter 3 of the SROCC), and the novel regional assessments such as the Hindu Kush Himalaya 21 Assessment (Wester et al., 2019). Figure 12.6 shows the regional changes in indices related to floods, and 22 coastal erosion over Asia, which are assessed on a regional basis along with other climatic impact-driver 23 indices below. 24 25 26 [START FIGURE 12.6 HERE] 27 28 Figure 12.6: Projected changes in selected climatic impact-driver indices for Asia. (a) Mean change in 1-in-100 29 year river discharge per unit catchment area (Q100, m 3 s-1 km-2) from CORDEX models for 2041-2060 30 relative to 1995-2014 for RCP8.5. (b) Shoreline position change along sandy coasts by the year 2100 31 relative to 2010 (meters; negative values indicate shoreline retreat) from the CMIP5 based data set Do Not Cite, Quote or Distribute 12-39 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 presented by Vousdoukas et al., (2020). (c) Bar plots for Q100 (m3 s-1 km-2) averaged over land areas for 2 the WGI reference AR6 regions (defined in Chapter 1). The left column within each panel (associated 3 with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute values in grey shades. The other 4 columns (associated with the right y-axis) show the Q100 changes relative to the recent past values for 5 two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for three global warming levels (defined 6 relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C (yellow) and 4°C (brown). The bars 7 show the median (dots) and the 10th-90th percentile range of model ensemble values across each model 8 ensemble. CMIP6 is shown by the darkest colors, CMIP5 by medium, and CORDEX by light. SSP5- 9 8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) Bar plots for shoreline position change 10 show CMIP5 based projections of shoreline position change along sandy coasts for 2050 and 2100 11 relative to 2010 for RCP8.5 (red) and RCP4.5 (blue) from Vousdoukas et al., (2020). Dots indicate 12 regional mean change estimates and bars the 5th-95th percentiles ranges of associated uncertainty. Note 13 that these shoreline position change projections assume that there are no additional sediment 14 sinks/sources or any physical barriers to shoreline retreat. See Technical Annex VI for details of indices. 15 Further details on data sources and processing are available in the chapter data table (Table 12.SM.1). 16 17 [END FIGURE 12.6 HERE] 18 19 20 12.4.2.1 Heat and cold 21 22 Mean air temperature: A long-term warming trend in annual mean surface temperature has been observed 23 across Asia during 1960-2015, and the warming accelerated after the 1970s (high confidence) (Davi et al., 24 2015; Aich et al., 2017; Cheong et al., 2018; Dong et al., 2018a; IPCC, 2018; Krishnan et al., 2019; Zhang et 25 al., 2019a). Records also indicate a higher rate of warming in minimum temperatures than maximum 26 temperatures in Asia, leading to more frequent warm nights and warm days, and less frequent cold days and 27 cold nights (high confidence) (Supari et al., 2017; Akperov et al., 2018; Cheong et al., 2018; Rahimi et al., 28 2018; Khan et al., 2019a; Li et al., 2019a; Zhang et al., 2019a). 29 30 Projections show continued warming over Asia in the future with contrasted regional patterns across the 31 continent (high confidence) (see Chapter 4 Figure 4.19). For RCP8.5/SSP5-8.5 at the end of the century, the 32 mean estimated warming exceeds 5°C in WSB, ESB and RFE and 7°C in some parts (high confidence). In 33 most areas of ARP and WCA, 5°C is exceded (Ozturk et al., 2017), but EAS, SAS and SEA have a lower 34 projected warming of less than 5°C (Basha et al., 2017; Lu et al., 2019a; Almazroui et al., 2020) (see also 35 Atlas.5). Under SSP1-2.6, the warming remains limited to 2°C in most areas except Arctic regions where it 36 exceeds 2°C (Chapter 4 Figure 4.19). 37 38 Extreme heat: There are increased evidences and high confidence of more frequent heat extremes in the 39 recent decades than in previous ones in most of Asia (Acar Deniz and Gönençgil, 2015; Rohini et al., 2016; 40 Mishra et al., 2017a; You et al., 2017; Imada et al., 2018; Khan et al., 2019b; Krishnan et al., 2019; Rahimi 41 et al., 2019; Yin et al., 2019) (Chapter 11) due to the effects of anthropogenic global warming, El Niño and 42 urbanization (Luo and Lau, 2017; Thirumalai et al., 2017; Imada et al., 2019; Sun et al., 2019c; Zhou et al., 43 2019). But there is medium confidence of heat extremes increasing frequency in many parts of India (Rohini 44 et al., 2016; Mazdiyasni et al., 2017; van Oldenborgh et al., 2018; Roy, 2019; Kumar et al., 2020) partly due 45 to the alleviation of anthropogenic warming by increased air pollution with aerosols and expanding irrigation 46 (van Oldenborgh et al., 2018; Thiery et al., 2020). 47 48 Extreme heat events are very likely to become more intense and/or more frequent in SAS, WCA, ARP, EAS, 49 and SEA by the end of 21st century, especially under RCP6.0 and RCP8.5 (Lelieveld et al., 2016; Pal and 50 Eltahir, 2016; Guo et al., 2017; Mishra et al., 2017a; Dosio et al., 2018; Shin et al., 2018; Lin et al., 2018; 51 Nasim et al., 2018; Su and Dong, 2019; Hong et al., 2019; Khan et al., 2020; Kumar et al., 2020) (see Figure 52 12.4a-c, and Chapter 11). The exceedance of the dangerous heat stress 41°C threshold of the HI is expected 53 to increase by about 250 days in SEA and by 50-150 days in SAS, WCA, ARP and EAS for SSP5-8.5 at the 54 end of century. Under SSP1-2.6, the increase would be restricted to less than 30 days in many of these 55 regions except SEA where the number of exceedance days increases by about 100 days in some areas. Such 56 increases are already present in the middle of the century (see Figure 12.4d-f) (Schwingshackl et al., 2021). Do Not Cite, Quote or Distribute 12-40 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 In these regions, the increase in number of days with exceedance of 35°C of high heat stress is also expected 2 to increase substantially for the mid-century under SSP5-8.5 (typically by 10-50 days except in Arctic and 3 Siberian regions), and by more than 60 days in areas of SEA, and a large difference is found between low 4 and high-end scenarios in the end of the century (high confidence) (Figure 12.4b). Over WSB, ESB and RFE 5 also, an increase of extreme heat durations and frequency is expected in all scenarios (Kattsov et al., 2017; 6 Khlebnikova et al., 2019a) (high confidence). 7 8 Cold spell and frost: Cold spells intensity and frequency, as well as the number of frost days, in most Asian 9 regions have been decreasing since the beginning of the 20th century (high confidence) (Sheikh et al., 2015; 10 Donat et al., 2016; Erlat and Türkeş, 2016; Dong et al., 2018a; Liao et al., 2018, 2020; Lu et al., 2018; van 11 Oldenborgh et al., 2019) (Chapter 11), except for the central Eurasian regions where there was a cooling 12 trend during 1995–2014 which is linked to sea-ice loss in the Barents–Kara Seas (medium confidence) 13 (Wegmann et al., 2018; Blackport et al., 2019; Mori et al., 2019) (Atlas.5.2). 14 15 It is very likely that cold spells will have a decreasing frequency in all future scenarios across Asian regions 16 (Guo et al., 2018b; Sui et al., 2018; Li et al., 2019a), as well as frost days (Wang et al., 2017c; Fallah- 17 Ghalhari et al., 2019) except in tropical Asia (Chapter 11). 18 19 In Asia, temperatures have warmed along the last century (high confidence) and extreme heat episodes 20 have become more frequent in most regions (high confidence), and are very likely projected to increase 21 in all regions of Asia under all warming scenarios along this century. Dangerous heat stress thresholds 22 such as HI>41°C will be crossed much more often (typically 50-150 days per year more than recent 23 past) in many Southern Asia regions at the end of century under SSP5-8.5 while these numbers should 24 remain limited to a few tens under SSP1-2.6 (high confidence). It is very likely that cold spells and frost 25 days will decrease in frequency in all future scenarios across Asian regions along the century. 26 27 28 12.4.2.2 Wet and dry 29 30 Mean precipitation: The most prominent features about changes in precipitation over Asia (1901-2010) are 31 the increasing precipitation trends across higher latitudes, along with some scattered smaller regions of 32 detectable increases and decreases (Knutson and Zeng, 2018); however, spatial variability remains high 33 (Wang et al., 2015b, 2019a; Limsakul and Singhruck, 2016; Supari et al., 2017; Rahimi et al., 2018, 2019; 34 Sein et al., 2018; Kumar et al., 2019) (medium confidence) (see also Atlas.5). 35 36 Mean precipitation is likely to increase in most areas of Northern (WSB, ESB, RFE), Southern (ECA, TIB, 37 SAS) and East Asia (EAS) in different scenarios (Huang et al., 2014; Xu et al., 2017; Kusunoki, 2018; 38 Mandapaka and Lo, 2018; Luo et al., 2019; Wu et al., 2019; Zhu et al., 2019b; Almazroui et al., 2020; Jiang 39 et al., 2020; Rai et al., 2020) (high confidence) (see Atlas.5). Monsoon circulation will also increase seasonal 40 contrasts, with SAS seeing wetter wet seasons and drier dry seasons (Atlas.5.3). Higher uncertainty between 41 CMIP5 and CMIP6 as well as spatial differences lend low confidence to model projections in ARP and WCA 42 (Atlas.5.5), with large seasonal differences (Zhu et al., 2020) and some models projecting decreases in 43 precipitation in Central Asia (Ozturk et al., 2017), Pakistan (Nabeel and Athar, 2020) and SEA (Supari et al., 44 2020). 45 46 River flood: Flood risk has grown in many places in China from 1961 to 2017 (Kundzewicz et al., 2019) (low 47 confidence). In SAS, the numbers of flood events and human fatalities have increased in India during 1978– 48 2006 (Singh and Kumar, 2013), whereas the average country-wide inundation depth has been decreasing 49 during 2002 to 2010 in Bangladesh attributed to improved flood management (Sciance and Nooner, 2018) 50 (low confidence). 51 52 Given the increase of heavy precipitation in most Asian regions the river flood frequency and intensities will 53 change consequently in Asia. Over China floods will increase with different levels under different warming 54 scenarios (Lin et al., 2018; Kundzewicz et al., 2019; Liang et al., 2019; Gu et al., 2020) (medium 55 confidence). Monsoon floods will be more intense in SAS (medium confidence) (Nowreen et al., 2015; Babur Do Not Cite, Quote or Distribute 12-41 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 et al., 2016; Mohammed et al., 2018). The total flood damage will increase greatly in river basins in SEA 2 countries under the conditions of climate change and rapid urbanization in the near future (Dahal et al., 2018; 3 Kefi et al., 2020). A changing snowmelt regime in the mountains may contribute to a shift of spring floods to 4 earlier periods in Central Asia in future (Reyer et al., 2017b) (medium confidence). The annual maximum 5 river discharge can almost double by the mid-21st century in major Siberian rivers, and annual maximum 6 flood area is projected to increase across Siberia mostly by 2–5% relative to the baseline period (1990–1999) 7 under RCP8.5 scenario (Shkolnik et al., 2018) (medium confidence). 8 9 Heavy precipitation and pluvial flood 10 Pluvial floods are driven by extreme precipitation and land use. Observed changes in extreme precipitation 11 vary considerably by region (see also Chapter 11). Heavy precipitation is very likely to become more intense 12 and frequent in all areas of Asia except in ARP (medium confidence) for a 2oC GWL or higher (see Chapter 13 11). 14 15 Landslide: The majority of non-seismic fatal landslide events were triggered by rainfall, and Asia is the 16 dominant geographical area of landslide distribution (Froude and Petley, 2018). Floods and landslides are the 17 most frequently occurring natural hazards in the eastern Himalaya and hilly regions, particularly caused by 18 torrential rain during the monsoon season (Gaire et al., 2015; Syed and Al Amin, 2016). They accounted for 19 nearly half of the events recorded in the countries of the HKH region (Vaidya et al., 2019). Intense monsoon 20 rainfall in northern India and western Nepal in 2013, which led to landslides and one of the worst floods in 21 history, has been linked to increased loading of GHG and aerosols (Cho et al., 2016). Due to an increase of 22 heavy precipitation and permafrost thawing an increase in landslides is expected in some areas of Asia, such 23 as northern Taiwan of China, some South Korea mountains, Himalaya Mountains, and permafrost territories 24 of Siberia, and the increase is expected to be the greatest over areas covered by current glaciers and glacial 25 lakes (Kim et al., 2015; Kharuk et al., 2016; Chen et al., 2019a; Kirschbaum et al., 2020) (medium 26 confidence, medium evidence). 27 28 Aridity: Aridity in West Central Asia and parts of South Asia increased in recent decades (medium 29 confidence), as documented in Afghanistan (Qutbudin et al., 2019), Iran (Zarei et al., 2016; Zolfaghari et al., 30 2016; Pour et al., 2020), most parts of Pakistan (Ahmed et al., 2018, 2019a), and many parts of India (Roxy 31 et al., 2015; Mallya et al., 2016; Matin and Behera, 2017; Ramarao et al., 2019). Some spatial and seasonal 32 differences within these regions remain, with Ambika and Mishra (2020) noting significant aridity declines 33 over the Indo-Gangetic Plain in India during 1979–2018 due in part to the effect of irrigation and Araghi et 34 al. (2018) found that many parts of Iran show no significant trends in aridity. There was slight drying in dry 35 season and significant wetting in wet season in Philippines during 1951–2010 (Villafuerte et al., 2014), and 36 slight wetting in Viet Nam during 1980-2017 (Stojanovic et al., 2020) (low confidence). In EAS there is low 37 confidence of broad aridity changes, as the frequency of droughts have increased (especially in spring) along 38 a strip extending from southwest China to the western part of northeast China; however, there is no evidence 39 of a significant increase in drought severity over China as a whole and many parts in the arid northwest 40 China got wetter during 1961-2012 (Wang et al., 2015b, 2019a; Zhai et al., 2017; Zhang and Shen, 2019). In 41 Siberia, the number of dry days has decreased for much of the region, but increased in its southern parts 42 (Khlebnikova et al., 2019b). 43 44 The counteracting factors of projected increases in precipitation and temperature across most of Asia 45 (Atlas.5; Section 11.9) leads to low confidence (limited evidence, inconsistent trends) for broad, long-term 46 aridity changes with medium confidence only for aridity increases in West Central Asia and East Asia. A 47 growing number of studies highlight the potential for more localized aridity trends, including projection 48 ensembles indicating significant increase in aridity and more frequent and intense droughts in most parts of 49 China (Li et al., 2019b; Yao et al., 2020) and India under RCP4.5 and RCP8.5 for 2020-2100 period (Gupta 50 and Jain, 2018; Bisht et al., 2019; Preethi et al., 2019). 51 52 Hydrological drought: Section 11.9 indicates that limited evidence and inconsistent regional trends gives 53 low confidence to observed and projected changes in hydrological drought in all Asian regions at a 2℃ GWL 54 (approximately mid-century), although West Central Asia hydrological droughts increase at the 4℃ global 55 warming level (approximately end-of-century under higher emissions scenarios) (medium confidence). Do Not Cite, Quote or Distribute 12-42 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Human activities such as reservoir operation and water abstraction have had a profound effect on low river 2 flow characteristics and drought impacts in many Asian regions (Kazemzadeh and Malekian, 2016; Yang et 3 al., 2020b). There was no observed overall long-term change of both meteorological droughts and 4 hydrological droughts over India during 1870-2018 (Mishra, 2020), but there were strong trends toward 5 drying of soil moisture in north-central India (Ganeshi et al., 2020) and intensified droughts in northwest 6 India, parts of Peninsular India, and Myanmar (Malik et al., 2016). The frequency of water scarcity 7 connected with hydrological droughts has increased significantly in southern Russia since the beginning of 8 twenty-first century (Frolova et al., 2017). Higher future temperatures are expected to alter the seasonal 9 profile of hydrologic droughts given reduced summertime snowmelt (medium confidence) downstream of 10 mountains such as the Himalayas and the Tibetan Plateau (Sorg et al., 2014). Several studies project more 11 severe future hydrological drought in the Weihe River basin in northern China (Yuan et al., 2016; Sun and 12 Zhou, 2020). 13 14 Agricultural and ecological drought: Section 11.9 assesses medium confidence in observed increases to 15 agricultural and ecological droughts in West Central Asia, East Central Asia, and East Asia. Persistent 16 droughts were the main factor for grassland degradation and desertification in Central Asia in the early 21st 17 century (Zhang et al., 2018b; Emadodin et al., 2019). Compound meteorological drought and heat events, 18 which lead to water stress conditions for agricultural and ecological systems, have become more frequent, 19 widespread and persistent in China especially since the late 1990s (Yu and Zhai, 2020). There were more 20 agricultural droughts in north China than in south China, and the intensity of agricultural drought increased 21 during 1951-2018 (Zhao et al., 2021). 22 23 Studies examining a 2℃ global warming level give low confidence for projected broad changes to 24 agricultural and ecological drought across all Asia regions, although at the 4℃ global warming level 25 agricultural and ecological drought increases are projected for West Central Asia and East Asia along with a 26 decrease in South Asia (medium confidence) (Section 11.9). Summertime temperature increase will enhance 27 evapotranspiration, facilitating ecological and agricultural drought over Central Asia towards the latter half 28 of this century (Ozturk et al., 2017; Reyer et al., 2017b; Senatore et al., 2019) (see also Figure 12.4 for soil 29 moisture and DF indices) (see also Chapter 11). However, broader changes in droughts could not be 30 determined in Asia due to the mixture of total precipitation signals together with temperature increase 31 patterns (Section 11.9; Atlas.5). 32 33 Fire weather: Under the global warming scenario of 2°C, the magnitude of length and frequency of fire 34 seasons are projected to increase with strong effects in India, China and Russia (Sun et al., 2019b) (medium 35 confidence). (Abatzoglou et al., 2019) found that higher fire weather conditions due to climate change 36 emerge in the first part of the 21st century in South China, WCA as well as in boreal areas of Siberia and 37 RFE. The potential burned areas in five Central Asian countries (Kazakhstan, Kyrgyzstan, Tajikistan, 38 Uzbekistan, and Turkmenistan) will increase by 2%-8% in the 2030s and 3%-13% in the 2080s compared 39 with the baseline (1971-2000) (Zong et al., 2020) (medium confidence). 40 41 In conclusion, there is medium confidence in that extreme precipitation, mean precipitation and river 42 floods will increase across most Asian regions. There is low confidence for projected changes in aridity 43 and drought given overall increases in precipitation and regional inconsistencies, with medium 44 increases for West Central Asia and East Asia especially beyond the middle of the century and global 45 warming levels beyond 2℃. Fire weather seasons are projected to lengthen and intensify particularly 46 in the northern regions (medium confidence). 47 48 49 12.4.2.3 Wind 50 51 Mean wind speed: There is high confidence of the slowdown in terrestrial near-surface wind speed (SWS) 52 in Asia by approximately -0.1 m s-1 per decade since 1950s based on observations and reanalysis data, with 53 the significant decreases in Central Asia among the highest in the world followed by EAS and SAS (Tian et 54 al., 2019; Wu et al., 2018b; Zhang et al., 2019b). But a short-term strengthening in SWS was observed 55 during the winter since 2000 in Eastern China (Zeng et al., 2019; Zha et al., 2019) (medium confidence). Do Not Cite, Quote or Distribute 12-43 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 There is medium confidence of future declining mean SWS in Asia, except in SAS and SEA, as global 2 projections indicate a decreasing trend in all climate scenarios for most of North Asia, TIB and East Asia by 3 the mid-century (Karnauskas et al., 2018a; Fedotova, 2019; Jung and Schindler, 2019; Ohba, 2019; Wu et 4 al., 2020a; Zha et al., 2020) (see also Figure 12.4m-o), with negative effects on wind energy potential. 5 Decreases in North Asia are generally modest, not exceeding 10% for the mid-century and 20% for the end 6 of century for the RCP8.5 and RCP4.5 scenarios (Figure 12.4m-o). 7 8 Severe wind storms: Consistent with the general mean decreasing surface winds, there is medium 9 confidence that strong winds declined faster than weak winds in the past few decades in Asia in general 10 (Vautard et al., 2010; Tian et al., 2019), but evidence is lacking for spatial patterns. There is low confidence 11 that extra-tropical cyclones will decline in number in future climate scenarios over WCA, TIB, WSB and 12 ESB, and intensify over the Arctic regions as a result of the poleward shift of storm-tracks (Basu et al., 2018) 13 (see also Chapter 11). There is limited evidence for projection of changes in severe winds occurring in 14 convective storms in Asia. 15 16 Tropical cyclone: There was an increase in the number and intensification rate of intense tropical cyclones 17 (TC), such as category 4-5 (wind speeds greater than 58 m s-1), in the Western North Pacific (WNP) and Bay 18 of Bengal since the mid-1980s (Kim et al., 2016; Mei and Xie, 2016; Walsh et al., 2016a; Knutson et al., 19 2019) (medium confidence) (see also Section 11.7). There is medium confidence that there has been a 20 significant north-westward shift in TC tracks and a poleward shift in the average latitude where TCs reach 21 their peak intensity in the WNP since the 1980s (Knutson et al., 2019; Sun et al., 2019a; Lee et al., 2020), 22 increasing exposure to TC passage and more destructive landfall over east China, Japan, and Korea in the 23 last few decades (Kossin et al., 2016; Li et al., 2017; Altman et al., 2018; Liu and Chan, 2019), and 24 decreasing exposure in the region of SAS and southern China (Kossin et al., 2016; Cinco et al., 2016) (see 25 also Chapter 11). However, while the analysis shows fewer typhoons, more extreme TCs have affected the 26 Philippines (Cinco et al., 2016; Takagi and Esteban, 2016) (low confidence). The frequency and duration of 27 tropical cyclones has significantly increased over time over the Arabian Sea and insignificantly decreased 28 over the Bay of Bengal during 1977-2018 (Fan et al., 2020) (low confidence). 29 30 There is medium confidence that future TC numbers will decrease but the maximum TC wind intensities will 31 increase in the Western Pacific as elsewhere (Choi et al., 2019; Cha et al., 2020; Knutson et al., 2020) 32 (Chapter 11, see Figure 11.24). The simulations for the late 21st century for the RCP8.5 scenario yield 33 considerably more TCs in the WNP that exceed 49.4 m s−1 (Category 3) intensity (Mclay et al., 2019). There 34 is medium confidence that the average location of the maximum wind will migrate poleward (see Chapter 35 11), and TC translation speeds at the higher latitudes would decrease (Yamaguchi et al., 2020). As a 36 consequence, the intensity of TCs affecting the Japan Islands would increase in the future under the RCP8.5 37 scenario (Yoshida et al., 2017), whereas the frequency of TCs affecting the Philippine region and Vietnam is 38 projected to decrease (Kieu-Thi et al., 2016; Wang et al., 2017b; Gallo et al., 2019) (low confidence). 39 40 Sand and dust storm: The Asia-Pacific region contributes 26.8 per cent to global dust emissions as of 2012 41 (UNESCAP, 2018). In West Asia, the frequency of dust events has increased markedly in some areas (east 42 and northeast of Saudi Arabia, northwest of Iraq and east of Syria) from 1980 to the present (Nabavi et al., 43 2016; Alobaidi et al., 2017). This marked dust increase has been associated to drought conditions in the 44 Fertile Crescent (Notaro et al., 2015; Yu et al., 2015) likely amplified by anthropogenic warming (Kelley et 45 al., 2015) (see Chapter 10). Dust storm frequency in most regions of northern China show a decreasing trend 46 since the 1960s due to the decrease in surface wind speed (Guan et al., 2017) (medium confidence). 47 48 While dust activity has decreased greatly over EAS, current climate models are unable to reproduce the 49 trends (Guan et al., 2015, 2017; Zha et al., 2017; Wu et al., 2018a). Thus, there is limited evidence for future 50 trends of sand and dust storms in Asia. 51 52 In conclusion, surface wind speeds have been decreasing in Asia (high confidence), but there is a large 53 uncertainty in future trends. There is medium confidence that mean wind speeds will decrease in 54 Central and Northern Asia, and that tropical cyclones will have decreasing frequency and increasing 55 intensity overall. Do Not Cite, Quote or Distribute 12-44 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.2.4 Snow and Ice 2 3 Snow: There is no significant interannual trend of total snow cover from 2000 to 2016 over Eurasia (Wang et 4 al., 2017e; Sun et al., 2020). Observations do show significant changes in the seasonal timing of Eurasian 5 snow cover extent (especially for earlier spring snowmelt) since the 1970s, with seasonal changes expected 6 to continue in the future (high confidence) (Yeo et al., 2017; Zhong et al., 2021). By 2100, snowline 7 elevations are projected to rise between 400 and 900 m (4.4 to 10.0 m yr-1) in the Indus, Ganges, and 8 Brahmaputra basins under RCP8.5 scenario (Viste and Sorteberg, 2015). 9 10 Glacier: Observation and future projection of glacier mass changes in Asia are assessed in Section 9.5.1 11 grouped in three main regions: North Asia, High Mountains of Asia, and Caucuses and Middle East. All 12 regions show continuing decline in glacier mass and area in the coming century (high confidence). Under 13 RCP2.6 the pace of glacier loss slows, but glacier losses increase in RCP8.5 and peak in the mid to late 21st 14 century. GlacierMIP projections indicate that glaciers in the High Mountains of Asia lose 42 ± 25%, 15 56 ± 24%, and 71 ± 21% of their 2015 mass by the end of the century for RCP4.5, RCP6.0 and RCP8.5 16 scenarios, respectively. Under the same scenarios, glaciers in North Asia would lose 57 ± 40%, 72 ± 38%, 17 and 85 ± 30% of their mass, and glaciers in the Caucuses and the Middle East would lose 68 ± 32%, 18 83 ± 19%, and 94 ± 13% of their mass (see also Kraaijenbrink et al., 2017; Rounce et al., 2020). 19 20 Although enhanced melt water from snow and glaciers largely offsets hydrological drought-like conditions 21 (Pritchard, 2019), this effect is unsustainable and may reverse as these cryospheric buffers disappear (Gan et 22 al., 2015; Dong et al., 2018b; Huss and Hock, 2018) (medium confidence). In the Himalayas and the TIB 23 region higher temperatures will lead to higher glacier melt rates and significant glacier shrinkage and a 24 summer runoff decrease (Sorg et al., 2014) (medium confidence). Glacier runoff in the Asian high mountains 25 will increase up to mid-century, and after that runoff might decrease due to the loss of glacier storage (Lutz 26 et al., 2014; Huss and Hock, 2018; Rounce et al., 2020) (medium confidence). 27 28 Compared with 1990s, the number of lakes in TIB in the 2010s decreased by 2%, whereas total lake area 29 expanded by 25% (Wang et al., 2020a) due to the joint effect of precipitation increase and glacier retreat. 30 Many new lakes are predicted to form as a consequence of continued glacier retreat in the Himalaya- 31 Karakoram region (Linsbauer et al., 2016). As many of these lakes will develop at the immediate foot of 32 steep icy peaks with degrading permafrost and decreasing slope stability, the risk of glacier lake outburst 33 floods and floods from landslides into moraine-dammed lakes is increasing in Asian high mountains 34 (Haeberli et al., 2017b; Kapitsa et al., 2017; Bajracharya et al., 2018; Narama et al., 2018; Wang et al., 35 2020a) (high confidence). 36 37 Permafrost: Permafrost is thawing in Asia (high confidence). Temperatures in the cold continuous 38 permafrost of northeastern ESB rose from the 1980s up to 2017, and the active layer thicknesses in Siberia 39 and RFE generally increased from late 1990s to 2017 (Romanovsky et al., 2018). The change in mean annual 40 ground temperature for north Siberia is about +0.1 – 0.3 °C per decade since 2000 (Romanovsky et al., 41 2018). Ground temperature in the permafrost regions of TIB (taking 40% of TIB currently) increased 42 (0.02~0.26°C per decade for different boreholes) during 1980 to 2018, and the active layer thickened at a 43 rate of 19.5 cm per decade (Zhao et al., 2020b). There is high confidence that permafrost in Asian high 44 mountains will continue to thaw and the active layer thickness will increase (Bolch et al., 2019). The 45 permafrost area is projected to decline by 13.4–27.7% and 60–90% in TIB (Zhao et al., 2020b), and 46 32% ± 11% and 76% ± 12% in Russia (Guo and Wang, 2016) by the end of the 21st century under the 47 RCP2.6 and RCP8.5 scenario respectively (high confidence). 48 49 Lake and river ice: Lake ice cover duration got shorter in many lakes in TIB (Yao et al., 2016; Cai et al., 50 2019; Guo et al., 2020) and some other areas such as northwest China (Cai et al., 2020) and northeast China 51 (Yang et al., 2019) in last two decades (high confidence). River ice cover extent decreased in TIB as well (Li 52 et al., 2020d; Yang et al., 2020a). Climate warming also leads to a significant reduction in the period with ice 53 phenomena and the decrement of ice regime hazard in Russian lowland rivers (Agafonova et al., 2017), and 54 the Inner Mongolia reach of the Yellow River in north China (Wan et al., 2020) (high confidence). Lake ice 55 and river ice in Asia are expected to decline with projected increases in surface air temperature towards the Do Not Cite, Quote or Distribute 12-45 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 end of this century (Guo et al., 2020; Yang et al., 2020a) (high confidence). 2 3 Heavy snowfall and ice storm: Observed trends in heavy snowfall and ice storms are uncertain. Annual 4 maximum snow depth decreased for the period between 1962 and 2016 on the western side of both eastern 5 and western Japan, at rates of 12.3% and 14.6% per decade, respectively (Ministry of the Environment et al., 6 2018). Observational results generally show a decrease in the frequency and an increase in the mean 7 intensity of snowfalls in most Chinese regions (Zhou et al., 2018) (medium confidence). Because of the 8 decrease in the snow frequency, the occurrence of large-scale snow disasters in TIB decreased (Qiu et al., 9 2018; Wang et al., 2019d) (low confidence). Large parts of northern high-latitude continents (including 10 Siberia and RFE) have experienced cold snaps and heavy snowfalls in the past few winters, and the reduction 11 of Arctic sea ice would increase the chance of heavy snowfall events in those regions in coming decades 12 (Song and Liu, 2017) (medium confidence). Heavy snowfall is projected to occur more frequently in Japan’s 13 Northern Alps, the inland areas of Honshu Island and Hokkaido Island (Kawase et al., 2016, 2020; MOE et 14 al., 2018), and the heavy wet snowfall can be enhanced over the mountainous regions in central Japan and 15 northern part of Japan (Ohba and Sugimoto, 2020) (medium confidence). 16 17 Hail: The hailstorm in the Asian region shows a decreasing trend in several regions (low confidence, limited 18 evidence). In China severe weather days including thunderstorm, hail and/or damaging wind have decreased 19 by 50% from 1961 to 2010 (Li et al., 2016b; Zhang et al., 2017), and the hail size decreased since 1980 (Ni 20 et al., 2017). A rate of decrease of 0.214 hail days per decade has also been reported for Mongolia between 21 1984-2013, where the annual number of hail days averaged is 0.74 (Lkhamjav et al., 2017). 22 23 Snow avalanche: There is as yet limited evidence for the evolution of avalanches in Asia. Tree-ring–based 24 snow avalanche reconstructions in the Indian Himalayas show an increase in avalanche occurrence and 25 runout distances in recent decades (Ballesteros-Cánovas et al., 2018). 26 27 In summary, snowpack and glaciers are projected to continue decreasing and permafrost to continue 28 thawing in Asia (high confidence). There is medium confidence of increasing heavy snowfall in some 29 regions, but limited evidence on future changes in hail and snow avalanches. 30 31 32 12.4.2.5 Coastal and oceanic 33 34 Relative sea level: Around Asia, over 1900-2018, a new tide-gauge based reconstruction finds a regional- 35 mean RSL change of 1.33 (0.80-1.86) mm yr-1 in the Indian Ocean-Southern Pacific and 1.68 [1.27 to 2.09] 36 mm yr-1 in the Northwest Pacific (Frederikse et al., 2020), compared to a GMSL change of around 1.7 mm 37 yr-1 (Section 2.3.3.3; Table 9.5). For the period 1993-2018, the RSLR rates, based on satellite altimetry, 38 increased to 3.65 [3.23 to 4.08] mm yr-1 and 3.53 [2.64 to 4.45] mm yr-1, respectively (Frederikse et al., 39 2020), compared to a GMSL change of 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). The rate of RSL rise along 40 the coastline of China ranges from -2.3 ± 1.9 to 5.7 ± 0.4 mm yr-1 during 1980–2016; after removing the 41 vertical land movement, the average rate of sea level rise is 2.9 ± 0.8 mm yr-1 over 1980-2016 and 3.2 ± 1.1 42 mm yr-1 since 1993 (Qu et al., 2019). However, the rates of land subsidence reported by (Minderhoud et al., 43 2017) are substantially higher than those reported by Qu et al. (2019). RSL change in many coastal areas in 44 Asia, especially in EAS, is affected by land subsidence due to sediment compaction under building mass and 45 groundwater extraction (Erban et al., 2014; Nicholls, 2015; Minderhoud et al., 2019; Qu et al., 2019) (high 46 confidence). During 1991- 2016, the Mekong delta in Vietnam sank on average ~18 cm as a consequence of 47 groundwater withdrawal, and the subsidence related to groundwater extraction has gradually increased with 48 highest sinking rates estimated to be 11 mm yr-1 in 2015 (Minderhoud et al., 2017). 49 50 Relative sea-level rise is very likely to continue in the oceans around Asia. Regional-mean RSLR projections 51 for the oceans around Asia range from 0.3 m–0.5 m under SSP1-RCP2.6 to 0.7 m–0.8 m under SSP5- 52 RCP8.5 for 2081-2100 relative to 1995-2014 (median values), which means local RSL change ranges from 53 just below mean projected GMSL change to above-average values {Section 9.6.3.3}. These RSLR 54 projections may however be underestimated due to potential partial representation of land subsidence in their 55 assessment (Section 9.6.3.2). Do Not Cite, Quote or Distribute 12-46 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Coastal flood: The present day 1:100 yr ETWL is between 0.5 m – 8 m around Asia, with values above 2.5 2 m or above common along the Central and North East coasts of Asia (Vousdoukas et al., 2018; Kirezci et al., 3 2020). Sea level rise and land subsidence will jointly lead to more flooding in delta areas in Asia (Takagi et 4 al., 2016; Wang et al., 2018a) (high confidence). 5 6 Extreme total water level magnitude and occurrence frequency are expected to increase throughout the 7 region (high confidence) (see Figure 12.4p-r and Figure SM 12.6). Across the region, the 5th – 95th percentile 8 range of the 1:100 yr ETWL is projected to increase (relative to 1980 – 2014) by 7 cm – 44 cm and by 10 cm 9 – 42 cm by 2050 under RCP4.5 and RCP8.5, respectively. By 2100, this range is projected to be 11 cm – 91 10 cm and 28 cm – 187 cm under RCP4.5 and RCP8.5, respectively (Vousdoukas et al., 2018; Kirezci et al., 11 2020). Furthermore, the present day 1:100 yr ETWL is projected to have median return periods of around 12 1:50 yrs by 2050 and 1:10 yrs by 2100 under RCP in most of Asia, except SEA and ARP, in which the 13 present day 1:100 yr ETWL is projected occur once per year or more, both by 2050 and 2100 (Vousdoukas 14 et al., 2018). The present day 1:50 yr ETWL is projected to occur around 3 times a year by 2100 with a SLR 15 of 1 m across Asia (Vitousek et al., 2017). Compound impacts of precipitation change, land subsidence, sea- 16 level rise, upstream hydropower development, and local water infrastructure development may lead to larger 17 flood extent and prolonged inundation in the Vietnamese Mekong Delta (Triet et al., 2020). 18 19 Coastal erosion: Over the past 30 years, South, Southeast and East Asia exhibit the most pronounced delta 20 changes globally due to strong human-induced changes to the fluvial sediment flux (Nienhuis et al., 2020). 21 Satellite derived shoreline change estimates over 1984 – 2015 indicate shoreline retreat rates between 0.5 m 22 yr-1 and 1 m yr-1 along the coasts of WCA and ARP, increasing to 3 m yr-1 in SAS. Over the same period, 23 shoreline progradation has been observed along the coasts of RFE (0.2 m yr-1), SEA (0.1 m yr-1) and EAS 24 (0.5 m yr-1) (Luijendijk et al., 2018; Mentaschi et al., 2018 ). Meanwhile, there has been a gross coastal area 25 loss of 3,590 km2 in South Asia, and a loss of 2,350 km2 in Pacific Asia, over a 30 year period (1984-2015) 26 (Mentaschi et al., 2018). 27 28 Projections indicate that a majority of sandy coasts in the Asia region will experience shoreline retreat (Udo 29 and Takeda, 2017; Ritphring et al., 2018; Vousdoukas et al., 2020b) (high confidence), while parts of the 30 RFE, EAS, SEA and WCA coastline are projected to prograde over the 21st century, if present ambient 31 shoreline change trends continue. Median shoreline change projections (CMIP5), relative to 2010, presented 32 by Vousdoukas et al. (2020) show that, by mid-century, sandy shorelines in Asia will retreat by between 10 33 m – 50 m, except in SAS where shoreline retreat is projected to exceed 100 m, under both RCP4.5 and 34 RCP8.5. By 2100, and under RCP4.5, shoreline retreats of around 85 m, 100 m and 300 m are projected 35 along the sandy coastlines of SEA and WCA, ARP and SAS respectively (50 m or less in other Asian 36 regions), while under RCP8.5, over the same period, sandy shorelines along all regions with coastlines, 37 except RFE and EAS, are projected to retreat by more than 100 m, with the retreat in SAS reaching 350 m 38 (2100 RCP8.5 projections for RFE and EAS are ~ 60 m and ~ 85 m respectively) (see Figure 12.6). 39 40 Marine heatwave: There have been frequent marine heatwaves (MHW) in the coastal oceans of Asia, 41 connected to the increase between 0.25°C and 1°C in mean SST of the coastal oceans since 1982-1998 42 (Oliver et al., 2018). There is high confidence that MHWs will increase around most of Asia. Mean SST is 43 projected to increase by 1ºC (2ºC) around Asia by 2100, with a hotspot of around 2 ºC (5 ºC) along the 44 coastlines of the East Sea and the RFE under RCP4.5 (RCP8.5) (see Interactive Atlas). Under global 45 warming conditions, MHW intensity and duration are projected to increase in the coastal zones of all sub- 46 regions of Asia, but most notably in SEA and SAS (Frölicher et al., 2018). Projections for SSP1-2.6 and 47 SSP5-8.5 both show an increase in MHWs around Asia by 2081 – 2100, relative to 1985 – 2014 (Box 9.2, 48 Figure 1). 49 50 In general, there is high confidence that most coastal/ocean related hazards in Asia will increase over 51 the 21st century. Relative sea-level rise is very likely to continue around Asia contributing to increased 52 coastal flooding in low-lying areas (high confidence) and shoreline retreat along most sandy coasts 53 (high confidence). Marine heatwaves are also expected to increase around the region over the 21st 54 century (high confidence). 55 Do Not Cite, Quote or Distribute 12-47 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 The assessed direction of change in climatic impact-drivers for Asia and associated confidence levels are 2 illustrated in Table 12.4. 3 4 5 [START TABLE 12.4 HERE] 6 7 Table 12.4: Summary of confidence in direction of projected change in climatic impact-drivers in Asia, representing 8 their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, SRES A1B, or 9 above within each AR6 region (defined in Chapter 1), approximately corresponding (for CIDs that are 10 independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for more details 11 of the assessment method). The table also includes the assessment of observed or projected time-of- 12 emergence of the CID change signal from the natural inter-annual variability if found with at least 13 medium confidence in Section 12.5.2. 14 Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Arabian Peninsula (ARP) 1 West Central Asia (WCA) 5 1,2 West Siberia (WSB) East Siberia (ESB) Russian Far East (RFE) 1,2 East Asia (EAS) 3 1,2 East Central Asia (ECA) Tibetan Plateau (TIB) South Asia (SAS) 1 South East Asia (SEA) 4 3 1,2 1. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. 2. Substantial parts of the EAS and SEA coasts are projected to prograde if present-day ambient shoreline change rates continue Key 3. Tropical cyclones decrease in number but increase in intensity High confidence of decrease 4. High confidence of decrease in Indonesia (Atlas.5.4.5) Medium confidence of decrease 5. Medium confidence of decreasing in summer and increasing in winter Low confidence in direction of change Medium confidence of increase High confidence of increase Not broadly relevant 15 16 [END TABLE 12.4 HERE] 17 18 19 12.4.3 Australasia 20 21 For the purpose of this assessment, Australasia is sub-divided into five sub-regions as defined in Section 22 1.4.5: Northern Australia (NAU), Central Australia (CAU), Eastern Australia (EAU), Southern Australia 23 (SAU) and New Zealand (NZ). 24 25 Previous IPCC Assessment reports 4 and 5 identify the most damaging historical hazards in this region to be 26 inland flooding, drought, wildfire, and episodic coastal erosion due to storms (Hennessy et al., 2007; 27 Reisinger et al., 2014). The IPCC Special Report on 1.5°C warming (Hoegh-Guldberg et al., 2018) projects 28 very likely increases in the intensity and frequency of warm days and warm nights and decreases in the 29 intensity and frequency of cold days and cold nights in Australasia. Furthermore, a likely increase in the 30 frequency and duration of warm spells is also projected for Australia. The IPCC Special Report on the Ocean Do Not Cite, Quote or Distribute 12-48 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 and Cryosphere (IPCC, 2019) projects a likely global mean sea level rise (RCP8.5) that is up to 0.1 m higher 2 than corresponding AR5 projections. IPCC SROCC also projects an increase of mean significant wave 3 height across the Southern Ocean (high confidence) and an increase in the occurrence of historically rare 4 (1:100 yr) extreme sea levels to 1:1 yr or more frequent events all around the Australasian region by 2100 5 under RCP8.5. 6 7 A detailed national scale climate change assessment of observed and projected climate change, based on over 8 40 CMIP5 models and high resolution downscaling (CSIRO and BOM, 2015) and bi-annual short updates 9 thereafter are available for Australia (CSIRO and BOM, 2016, 2018, 2020). Similar national assessments for 10 New Zealand are also available (Ministry for the Environment & Stats NZ (2017), Ministry for the 11 Environment (2018) and Ministry for the Environment (2020)). The severe extreme events such as heat 12 waves and river floods that have occurred in Australasia, especially over the last decade, have enabled a 13 number of attribution studies, improving the understanding of regional climate change mechanisms that drive 14 such extreme events (see Chapter 11). 15 16 Figure 12.7 illustrates projected changes in two selected hazard indices for Australasia. 17 18 19 [START FIGURE 12.7 HERE] 20 21 Figure 12.7: Projected changes in selected climatic impact-driver indices for Australasia. (a) Mean change in 1- 22 in-100 year river discharge per unit catchment area (Q100, m 3 s-1 km-2) from CORDEX models for 2041- 23 2060 relative to 1995-2014 for RCP8.5. (b) Shoreline position change along sandy coasts by the year 24 2100 relative to 2010 (meters; negative values indicate shoreline retreat) from the CMIP5 based data set 25 presented by Vousdoukas et al. (2020). (c) Bar plots for Q100 (m3 s-1 km-2) averaged over land areas for 26 the WGI reference AR6 regions (defined in Chapter 1). The left column within each panel (associated 27 with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute values in grey shades. The other 28 columns (associated with the right y-axis) show the Q100 changes relative to the recent past values for 29 two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for three global warming levels (defined 30 relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C (yellow) and 4°C (brown). The bars 31 show the median (dots) and the 10th-90th percentile range of model ensemble values across each model 32 ensemble. CMIP6 is shown by the darkest colours, CMIP5 by medium, and CORDEX by light. SSP5- 33 8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) Bar plots for shoreline position change 34 show CMIP5 based projections of shoreline position change along sandy coasts for 2050 and 2100 35 relative to 2010 for RCP8.5 (red) and RCP4.5 (blue) from Vousdoukas et al. (2020). Dots indicate 36 regional mean change estimates and bars the 5th-95th percentiles ranges of associated uncertainty. Note 37 that these shoreline position change projections assume that there are no additional sediment 38 sinks/sources or any physical barriers to shoreline retreat. See Technical Annex VI for details of indices. 39 Further details on data sources and processing are available in the chapter data table (Table 12.SM.1). 40 41 [END FIGURE 12.7 HERE] 42 43 44 12.4.3.1 Heat and Cold 45 46 Mean air temperature: Across Australia mean temperatures have increased by 1.44±0.24 °C during the 47 period 1910-2019, with most of the warming occurring since 1950 (Section Atlas.6.2; CSIRO and BOM, 48 2020; Trewin et al., 2020). In New Zealand, an increase of 1.1°C has been measured from 1909-2016 49 (Section Atlas.6.2; Ministry for the Environment, 2020). In the period 1980 – 2014 a rate of increase of 0.1º 50 – 0.3º per decade has been observed (Figure Atlas.11 and Figure Atlas.23). 51 52 Mean temperature in Australasia is projected to continue to rise through the 21st century (virtually certain) 53 (Section Atlas.6.4). Projections for Australia indicate that the average temperature will increase by +1.1°C 54 (0.84-1.52°C 10th -90th percentile range) by 2041-2060 (mid-century), and by +1.9°C (1.29 to 2.58°C) by 55 2081-2100 (end-century), relative to the baseline period of 1995-2014, under SSP2-4.5 (Interactive Atlas). 56 For SSP5-8.5, the projected changes are up to +1.5°C (1.17 to 1.96°C) and +3.7°C (2.75 to 4.91°C) for mid 57 and end century respectively. For SSP1-2.6, mean temperature is projected to rise by +0.9°C (0.55 to 1.26 Do Not Cite, Quote or Distribute 12-49 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 °C) and +1.0°C (0.55 to 1.54 °C) relative to 1995-2014 by mid and end century, respectively (Interactive 2 Atlas). In New Zealand, an increase of mean temperature of +1.0°C (0.60-1.32°C) relative to 1995-2014 is 3 projected by mid-century, and an increase of +1.6°C (1.03 to 2.26°C) by end century under SSP2-4.5. For 4 SSP5-8.5, the projected increase in mean temperature is +1.3°C (0.91 to 1.66°C) and +3.1°C (2.20 to 4.05°C 5 10-90 percentile range) relative to 1995-2014 by mid and end century, respectively. For SSP1-2.6, the 6 projected increase in mean temperature is +0.75°C (0.39 to 1.06°C) and +0.8°C (0.47 to 1.46°C) relative to 7 1995-2014 by mid and end century, respectively (Interactive Atlas). 8 9 Extreme heat: The region has a very likely trend of increasing frequency and severity of hot extremes since 10 the 1950s (Table 11.6). Extreme minimum temperatures have increased in all seasons over most of Australia 11 and exceeds the increase in extreme maximum temperatures (Wang et al., 2013b; Jakob and Walland, 2016). 12 Heatwave characteristics and hot extremes have increased across many Australian regions since the mid-20th 13 century (Table 11.6; BOM and CSIRO, 2020)). The number of days per year with maximum temperature 14 greater than 35°C has increased over most parts of Australia from 1957 – 2015, with the largest increasing 15 trends of 0.4-1 days/year occurring in north western, northern, north eastern and parts of central Australia 16 (CSIRO and BOM, 2016). Long term changes of hot extremes in Australia have been attributed to 17 anthropogenic influence (Table 11.6). In New Zealand, the number of annual heatwave days increased at 18 18 of 30 sites during the period 1972–2019 (MfE and Stats NZ, 2020). 19 20 More frequent hot extremes and heatwaves are expected over the 21st century in Australia (virtually certain) 21 (Table 11.6). Heat thresholds potentially affecting agriculture and health, such as 35°C or 40°C, are also 22 projected to be exceeded more frequently over the 21st century in Australia under all RCPs (high confidence). 23 By 2090 under RCP4.5, the average number of days per year with maximum temperatures above 35°C is 24 highly spatially variable and is expected to increase by 50%–100%, while the number of days per year with 25 maximum temperatures above 40°C is expected to increase by 200%, relative to 1985 2005 (CSIRO and 26 BOM, 2015). Under RCP8.5 the corresponding projected increases are even greater, with a greater than 27 100% increase in most of Australia, and far greater increases (up to a 20-fold increase in Darwin) in central 28 and northern Australia. Projections for NZ indicate more frequent hot extremes (virtually certain) (Table 29 11.6). Figure 12.4b, c shows CMIP6 projections of mean number of days per year with maximum 30 temperature exceeding 35°C under SSP5-8.5, which are consistent with the above assessed literature and 31 across the two CMIP generations, and indicate a strong difference depending on the mitigation scenario (e.g. 32 over 100 days more per year under SSP5-8.5 in NAU, but, in general, less than 60 days more per year under 33 SSP1-2.6 in NAU) (see also Figure SM 12.1) 34 35 The projected frequency of exceeding dangerous humid heat thresholds is increasing in Australia, with a 36 strong increase in Northern Australia for RCP8.5 (high confidence) (Zhao et al., 2015; Mora et al., 2017; 37 Brouillet and Joussaume, 2019), consistently across CMIP5, CMIP6 and CORDEX simulations (Figure 38 12.4d-f and Figure SM 12.2). Using the HI index, by end-century, the average number of days exceeding 39 41°C is projected to increase in NAU by about 100 days and by about 25 days under SSP5-8.5 and SSP1- 40 2.6, respectively. The projections for New Zealand indicate no appreciable increase in the number of days 41 with HI > 41°C across SSPs, time periods and CMIP generations (Figure 12.4d-f and Figure SM 12.2). 42 43 Cold spell and frost: Excepting parts of southern Australia, the Australasian region has a significant trend of 44 decreasing frequency in cold extremes since the 1950s (high confidence) (Table 11.6) and there is high 45 confidence that such trends are attributable to anthropogenic influence (Table 11.6). The number of frost 46 days per year in Australia has on average declined at a rate of 0.15 days/decade in the past century 47 (Alexander and Arblaster, 2017), except in some regions of southern Australia, where an increase in both 48 number and season length has been reported (Dittus et al., 2014; Crimp et al., 2016b). The number of frost 49 days has decreased at 12 of 30 monitoring sites around New Zealand over the period 1972–2019 (MfE and 50 Stats NZ, 2020). 51 52 Less frequent cold extremes are virtually certain in Australasia (Table 11.6) while a decrease of frost days is 53 projected with high confidence for the region. Projections, relative to 1986–2005, for the number of frost 54 days per year in Australia indicate declines of 0.9 days by mid-century and 1.1 days by end of the century for 55 RCP4.5, while for RCP8.5, the projected declines are 1.0 days and 1.3 days by mid- and end century Do Not Cite, Quote or Distribute 12-50 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 respectively (Alexander and Arblaster, 2017; Herold et al., 2018). Projections for New Zealand indicate that 2 the number of frost days will decrease by 30% (RCP2.6) to 50% (RCP8.5) by 2040, relative to 1986–2005. 3 By 2090, the decrease ranges from 30% (RCP2.6) to 90% (RCP8.5) (MfE and Stats NZ, 2017). 4 5 In general, there is high confidence that most heat hazards in Australasia will increase and that cold 6 hazards will decrease over the 21st century. The mean temperature in Australasia is virtually certain to 7 continue to rise through the 21st century, accompanied by less frequent cold extremes (virtually 8 certain) and frost days (high confidence), and more frequent hot extremes (virtually certain). Heat 9 stress is projected to increase in Australia (high confidence). 10 11 12 12.4.3.2 Wet and Dry 13 14 Mean precipitation: Here, only increases in precipitation (under Wet) are addressed, with decreases (under 15 Dry) are addressed in Aridity below. 16 17 In terms of wet climatic impact-drivers, detectable anthropogenic increases in precipitation in Australia have 18 been reported particularly for north central Australia for the period 1901-2010 (Knutson and Zeng, 2018). 19 Figure Atlas.11 indicates no significant trend in precipitation over the region during the baseline period 1960 20 – 2015, except for the GPCP dataset which shows an increasing trend in north central Australia. In New 21 Zealand, increases in annual rainfall have been observed between 1960-2019 in the south and west of the 22 South Island and east of the North Island. Note however, for the most part, the above reported trends in New 23 Zealand have been classified as statistically not significant (Figure Atlas.23). 24 25 Annual mean precipitation is projected to increase in central and north east Australia (low confidence) and in 26 the south and west of New Zealand (medium confidence) (Section Atlas.6.4). Liu et al. (2018a) show that 27 under 1.5°C warming, central and northeast Australia will become wetter. In New Zealand, projected 28 patterns in annual precipitation exhibit increases in the west and south of New Zealand (Section 29 Atlas.6.4).(Liu et al., 2018a) project that the South Island will be wetter under both 1.5°C and 2°C warming. 30 However, there is limited model agreement for projected rainfall changes in Australasia as shown in the 31 Atlas. 32 33 River flood: Streamflow observations in Australia have shown that negative trends dominate in annual 34 maximum flow and that stations with significant negative trends were mostly located in the southeast and 35 southwest (Gu et al., 2020). The observed peak flow trend in southern Australia is attributed to the decrease 36 of soil moisture, although an increase of flood magnitude is possible for very rare events. For the more 37 frequent flood events, the increase of extreme precipitation is balanced by the decrease of soil moisture. 38 (Wasko and Nathan, 2019). 39 40 While median annual runoff is projected to decrease in most of Australia (Chiew et al., 2017), consistent 41 with projected decreases in average rainfall (CSIRO and BOM, 2015; Alexander and Arblaster, 2017), river 42 floods are projected to increase due to more intense extreme rainfall events and associated increase in runoff 43 (medium confidence). Asadieh and Krakauer (2017) found a decrease in the value of the 95% percentile of 44 mean streamflow with RCP8.5 by the end of the century in all of Australia, except in a small part in centre of 45 the country. In terms of relative increases, flooding is expected to increase more in the northern Australia 46 (driven by convective rainfall systems) than in southern Australia (where more intense extreme rainfall may 47 be compensated by drier antecedent moisture conditions) (Alexander and Arblaster, 2017; Dey et al., 2019) 48 with flood frequency increasing in northern Australia and along parts of the east coast and decreasing in 49 south-western Western Australia (Hirabayashi et al. 2013). Gu et al. (2020) project larger flood magnitude 50 and volumes under both RCP2.6 and RCP8.5 in northern Australia, and smaller flood magnitudes and 51 volumes in southern Australia under the same RCPs. These findings are in general agreement with the 52 patterns in peak flow, corresponding to the 1:100 yr return period streamflow, shown in Figure 12.7a,c for 53 mid-21st century under RCP8.5. 54 55 There is medium confidence that river flooding will increase in New Zealand. Projections for New Zealand Do Not Cite, Quote or Distribute 12-51 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 indicate that the 1:50 yr and 1:100 yr flood peaks for rivers in many parts of the country may increase by 5 to 2 10% by 2050 and more by 2100 (with large variation between models and emissions scenarios), with a 3 corresponding decrease in return periods for specific flood levels (Gray et al., 2005; Carey-Smith et al., 4 2010; McMillan et al., 2010, 2012; Ballinger et al., 2011). 5 6 Heavy precipitation and pluvial flood: Rainfall extremes have been detected to increase in Australasia, 7 with low confidence (Table 11.6). There is high confidence that Rx1day and Rx5day precipitation extremes 8 will increase for 2ºC or lower warming for the region as a whole, but on a sub-regional basis there is only 9 medium confidence of increases in NAU and CAU and low confidence of increases on EAU, SAU and NZ. 10 For warming levels exceeding 2ºC, these extremes are very likely to increase in NAU and CAU and they are 11 likely to increase elsewhere in the region (Section 11.9). 12 13 Landslide: Based on local slope characteristics, lithology and seismic activity, the South Island and the 14 eastern half of the North Island of New Zealand are vulnerable to landslide occurrence (Broeckx et al., 15 2020). The potential for land and rockslides increases with, amongst other factors, total precipitation rates, 16 precipitation intensity, mountain permafrost thaw rates, glacier retreat and air temperature (Allen and 17 Huggel, 2013; Crozier, 2010; Gariano and Guzzetti, 2016; IPCC, 2019). Given the increase of the magnitude 18 of these physical variables in areas that are already highly susceptible to mass movements (MfE, 2018), there 19 is low confidence that the occurrence of landslides will increase under future climate conditions. 20 21 Aridity: In terms of dry climatic impact-drivers, a substantial decrease in precipitation has been observed 22 across southern Australia during the cool season (April–October) (medium confidence). The drying trend has 23 been particularly strong over southwest Western Australia between May and July, with rainfall since 1970 24 being around 20% less than the 1900-1969 average (CSIRO and BOM, 2020). Detectable decreases in mean 25 precipitation, attributable at least in part to anthropogenic forcing, have been reported for parts of southwest 26 Australia (Delworth and Zeng, 2014; Knutson and Zeng, 2018), southeast Australia, and Tasmania (Knutson 27 and Zeng, 2018; see also case study in Section 10.4.1.2.3). In New Zealand, the northeast of the South Island 28 and western and the northern parts of the North Island show decreasing precipitation trends during 1960- 29 2019 (MfE and Stats NZ, 2020). 30 31 Aridity is projected to increase, especially during winter and spring, with medium confidence in SAU but 32 with high confidence in southwest Western Australia (Section Atlas.6.4, Table 11.6). In EAU and in the 33 north and east of NZ, aridity is projected to increase with medium confidence, while a decrease is projected 34 with medium confidence in the south and west of NZ (Section Atlas.6.4). Although there is only low 35 confidence in the projected decrease of mean annual precipitation in south western and eastern Australia and 36 the north and east of New Zealand, there is high confidence of reduced winter and spring precipitation in 37 Australia in future, mostly in southwestern and eastern Australia (Section Atlas.6.4). Liu et al. (2018b) show 38 that under 2°C warming, most of Australia is projected to become drier based on PDSI, with the exception of 39 the tropical northeast. Ferguson et al. (2018) project that between 1976-2005 and 2070-2099, winters will 40 become drier (mainly in southern Australia) under RCP8.5. (Liu et al., 2018c) project that the north island of 41 New Zealand will be drier under both 1.5°C and 2°C warming. 42 43 Hydrological Drought: There is low confidence of observed changes in hydrological droughts in 44 Australasia, except in SAU where there is medium comfidence of an observed increase in the south-east and 45 south-west. Future projections indicate medium confidence in further hydrological drought increases for 46 Southern Australia for warming levels of 2°C or higher (Section 11.9). Mean annual runoff in far south-east 47 and far south-west Australia are projected to decline by median values of 20% and 50%, respectively, by 48 mid-century under RCP8.5 (Chiew et al., 2017). Prudhomme et al. (2014) assess changes in the Drought 49 Index (DI), defined as areal runoff less than the 10th percentile over the reference period 1976 – 2005, and 50 project DI increases for both Australia and New Zealand by 10%–20% by 2070-2099 under RCP8.5 with the 51 greatest effects being in the southern parts of the Australian continent. Those projections are consistent with 52 the trends shown in Figure 12.4g-i (see also Figure SM 12.3). The SPI drought frequency is projected to 53 increase in SAU and particularly in southwest Western Australia by mid- century, while by the end of the 54 century SPI drought frequency is projected to increase all over Australia, and particularly strongly in 55 southwest Western Australia as well as southern Victoria (see Figure 12.4g-i). For the Murray-Darling basin, Do Not Cite, Quote or Distribute 12-52 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Ferguson et al. (2018) project effectively no change (-1%) in mean precipitation, a 27% decrease in P-E, and 2 30% increase in runoff in 2070-2099 relative to 1976-2005 with RCP8.5. 3 4 Agricultural and ecological Drought: There is medium confidence in observations of Agricultural and 5 ecological droughts increasing in SAU and decreasing in NAU, while there is low confidence of changes 6 elsewhere in the region (Section 11.9). More regional studies have observed an increase in agricultural and 7 ecological drought intensity in southwest Australia and an increase in drought intensity in parts of southeast 8 Australia, while the length of droughts therein has increased (Section 11.9). In New Zealand, since 1972/73, 9 soils at 7 of 30 monitored sites became drier, while the 2012–13 drought was one of the most extreme in the 10 previous 41 years (MfE and Stats NZ, 2017). Future evaporative demand is projected to lead to medium 11 confidence increases in agricultural and ecological droughts for a 2° global warming level in SAU and EAU 12 and low confidence for changes in CAU, NAU and NZ although there is medium confidence of CAU 13 increases at the 4℃ global warming level (Section 11.9). There is medium confidence for more time in 14 agricultural and ecological drought in SAU by mid-21st century (Coppola et al., 2021b) as well as by the end 15 of the 21st century (Herold et al., 2018). The SPEI shows a springtime intensification in SAU with moderate 16 and severe droughts in the southwest and moderate droughts in the southeast (Herold et al., 2018). There is 17 consensus among the different model ensembles (CORDEX-CORE, CMIP5 and CMIP6) that the drought 18 frequency (DF), one of several proxies for agricultural and ecological drought, will increase in all four 19 Australian regions for both mid-century (NAU 0.2 to 2 DF increase, CEU 0.5 to 2 DF increase, SAU 1 to 3 20 DF increase and EAU 0.8 to 3 DF increase) and end-century (0.8–2.7 DF increase for NAU, 1.2–2 DF 21 increase for CAU, 2.2–3.8 for SAU, and 0.2–3 for EAU) for both RCP8.5 and SSP5-85, with CMIP6 22 showing the lowest increase (see Figure 12.4g-l and Figure SM 12.4) (Coppola et al., 2021b). 23 24 Fire weather: Dowdy and Pepler (2018) examined atmospheric conditions conducive to pyroconvection in 25 the period 1979-2016, and found an increased risk in southeast Australia during spring and summer, due to 26 changes in vertical atmospheric stability and humidity, in combination with adverse near-surface fire weather 27 conditions. CSIRO and BOM (2018) and Dowdy (2018) found that the annual 90th percentile daily Forest 28 Fire Danger Index (FFDI) has increased from 1950-2016 in parts of Australia, especially in southern 29 Australia (1 to 2.5 decade-1) and in spring and summer. These studies indicate an increase in the frequency 30 and magnitude of FFDI extreme quantiles, as well as a shift of the fire season start toward spring, 31 lengthening the fire season. The unprecedented large fires of austral spring and summer of 2019 in SE 32 Australia were a result of extreme hot and dry weather in significantly drier than average conditions that had 33 persisted since 2017, in combination with consistently stronger than average winds, resulting in above 34 average to highest on record FFDI values in much of the country (Abram et al., 2021). These fires have been 35 attributed to climate change through the temperature component of fire weather indices (van Oldenborgh et 36 al., 2021). In New Zealand, days with very high and extreme fire weather increased in 12 out of 28 37 monitored sites, and decreased in 8, in the period 1997 to 2019 (MfE and Stats NZ, 2020). Attribution 38 studies indicate that there is medium confidence of an anthropogenically-driven past increase in fire weather 39 conditions, essentially due to increase in frequency of extreme heat waves. (Hope et al., 2019; Lewis et al., 40 2020; van Oldenborgh et al., 2021). 41 42 Fire weather indices are projected to increase in most of Australia (high confidence) and many parts of New 43 Zealand (medium confidence), in particular with respect to extreme fire and induced pyroconvection (Dowdy 44 et al., 2019b). Increasing mean temperature, cool season rainfall decline, and changes in tropical climate 45 variability all contribute to a future increase in extreme fire risk in Australia (Abram et al., 2021). Projections 46 indicate that the annual cumulative FFDI will increase by 31-33% in southern and eastern Australia, and by 47 17-25% in northern Australia and the Rangelands by 2090 (relative to 1995) under RCP8.5 (CSIRO and 48 BOM, 2015). Using a CMIP5 ensemble of 17 models, Abatzoglou et al. (2019) found a statistically 49 significant positive trend for fire weather intensity and fire season length for future mid-century conditions 50 under RCP8.5, including a detectable anthropogenic influence on fire risk magnitude and fire season length 51 by 2040 in Western Australia and along the Queensland coastline. Using the C-Haines and FFDI indices 52 with A2 and RCP8.5 respectively, Di Virgilio et al. (2019) and Clarke et al. (2019) have shown that extreme 53 fire weather frequency will increase in south eastern Australia by the end of the 21st century. Most of these 54 projections indicate that the biggest increases in fire weather conditions will be in late spring, effectively 55 resulting in longer (stronger) fire seasons in areas where spring is the shoulder (peak) season. Do Not Cite, Quote or Distribute 12-53 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 In New Zealand, Watt et al. (2019) projected that the number of days with very high to extreme fire risk will 2 increase by 71 per cent by 2040, and by a further 12 per cent by 2090, for the A1B scenario, with fire risk increase 3 all along the east coast. The most marked relative changes by 2090 were projected for Wellington and Dunedin 4 where very high to extreme fire risk is projected to increase by, respectively, 89% to 32 days and 207% to 18 5 days, compared to the baseline period 1970-1999. 6 7 Annual mean precipitation is projected to increase in central and north east Australia (low confidence) 8 and in the south and west of New Zealand (medium confidence), while it is projected to decrease in 9 southern Australia (medium confidence), albeit with high confidence in southwest western Australia, in 10 eastern Australia (medium confidence), and in the north and east of New Zealand (medium confidence). 11 Heavy precipitation and pluvial flooding are projected to increase with medium confidence in northern 12 Australia and central Australia. There is medium confidence that river flooding will increase in New 13 Zealand and Australia, with higher increases in northern Australia. Aridity is projected to increase 14 with medium confidence in southern Australia (high confidence in southwest Western Australia), 15 eastern Australia (medium confidence), and in the north and east of New Zealand (medium confidence). 16 Hydrological droughts are projected to increase in Southern Australia (medium confidence), while 17 Agricultural and ecological droughts are projected to increase with medium confidence in Southern 18 Australia and Eastern Australia. Fire weather is projected to increase throughout Australia (high 19 confidence) and New Zealand (medium confidence). 20 21 22 12.4.3.3 Wind 23 24 Mean wind speed: There is low confidence of a mean wind-speed trend in the last decades (low agreement) 25 (McVicar et al., 2012; Troccoli et al., 2012; Azorin-Molina et al., 2018; Wu et al., 2018b), as long-term 26 measurements are not homogeneous. 27 28 In future climate scenarios wind speed trends exhibit generally weak amplitudes with low agreement among 29 models (Figure 12.4m-o and Figure SM 12.5) in Australia with uncertain consequences on wind power 30 potential (CSIRO and BOM, 2015; Karnauskas et al., 2018a; Jung and Schindler, 2019). However, there is 31 medium confidence that, by the end of the century, annual mean wind power will significantly increase in 32 northeastern Australia under RCP8.5, but there is low confidence of an increase by end-century under 33 RCP4.5, and for any scenario by mid-century (Karnauskas et al., 2018). In New Zealand, mean wind patterns 34 are projected to become more north-easterly in summer, and westerlies to become more intense in winter 35 (low confidence), in agreement with the strengthening of the southern hemisphere storm tracks (Section 36 4.5.1). 37 38 Severe wind storm: There is generally low confidence in observed changes in extreme winds and 39 extratropical storms in Australasia (Section 11.7.2). CMIP5 projections of severe winds indicate a general 40 increase in north eastern Australia, and decrease in some parts in southern and central Australia (medium 41 confidence) by the end of the century under RCP8.5 (CSIRO and BOM, 2015; Kumar et al., 2015; Jung and 42 Schindler, 2019). Elsewhere trends are diverse and vary across simulations with low agreement. Projections 43 of changes in the 1:25 yr return period winds (based on annual maxima) for 2074–2100 relative to 1979– 44 2005 for RCP8.5 show an increase in tropical areas of northern Australia (Kumar et al., 2015). 45 46 In New Zealand, the frequency and magnitude of extreme wind have decreased (from 1980 – 2019) at 12 of 47 14 monitored sites and increased at two monitored sites (MfE and Stats NZ, 2020). Due to the intensification 48 and the shift of the Austral storm track by the end of the century (Yin, 2005), a continuous increase in 49 extreme wind speed in New Zealand is projected over the South Island and the southern part of the North 50 Island by mid- and end of century for all RCP’s (low confidence) (MfE, 2018). 51 52 Tropical cyclone: In Australia, the number of TCs has generally declined since 1982, and the frequency of 53 intense TCs that make landfall in north eastern Australia has declined significantly since the 19th century 54 (medium confidence) (Kuleshov et al., 2010; Callaghan and Power, 2011; Holland and Bruyère, 2014; 55 Knutson et al., 2019; CSIRO and BOM, 2020). There is high confidence that cyclones making landfall along Do Not Cite, Quote or Distribute 12-54 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 north eastern and north Australian coastlines will decrease in number and low confidence of an increase in 2 their intensities for a 2°C global warming level as well as for the mid-century period with scenarios RCP4.5 3 and above (Roberts et al., 2015, 2020; Bacmeister et al., 2018; Knutson et al., 2020) , with the amplitude of 4 changes increasing from RCP4.5 to RCP8.5 (Bacmeister et al., 2018). Decreases in frequency are projected 5 for “east coast lows” (Walsh et al., 2016b; Dowdy et al., 2019a). 6 7 Sand and dust storm: Australia is recognized to be the largest dust source in the Southern Hemisphere 8 (Zheng et al., 2016). Land use and land cover change have increase dust emissions in Australia in the past 9 200 years (Marx et al., 2014). While projections suggest a decrease in severe winds in central and southern 10 Australia, changes in vegetation due to increased aridity and hydrological drought could be expected to result 11 in increased wind erosion and dust emission across the country (medium confidence) (Webb et al., 2020). 12 13 In Australasia, there is low confidence in projected mean wind speeds and wind power potential, with 14 a medium confidence increase projected only in northeastern Australia under high emission scenarios 15 and by the end of the 21st century. Tropical cyclones in north eastern and north Australia are 16 projected to decrease in number (high confidence) while their intensity is projected to increase (low 17 confidence). 18 19 20 12.4.3.4 Snow and Ice 21 22 Snow: The snow season length in Australia has decreased by 5% during 2000-2013 relative to 1954-1999, 23 especially in spring (Pepler et al., 2015). A shift in the date of peak snowfall has also been observed with an 24 11 day advance over the same period (Pepler et al., 2015). A decreasing trend in maximum snow depth has 25 been observed for Australian alpine regions since the late 1950s, with the largest declines during spring and 26 at lower altitudes. Maximum snow depth is highly variable and is strongly influenced by rare heavy snowfall 27 days, which have no observed trends in frequency (CSIRO and BOM, 2020). 28 29 Projections for Southern Australia and New Zealand show a continuing reduction in snowfall during the 21st 30 century (high confidence). The magnitude of decrease varies with the altitude of the region and the emission 31 scenario. At elevations lower than 1500 m, years without snowfall are projected from 2030 in some models. 32 By 2090, and under RCP8.5, such years are projected to become common (CSIRO and BOM, 2015). The 33 number of annual snow days in New Zealand is projected to decrease under all RCP’s, by up to 30 days or 34 more by 2090 under RCP8.5, relative to 1986 – 2005 (MfE, 2018). 35 36 Glacier: Glacier mass and areal extent in New Zealand is projected to continue to decease over the 21st 37 century (high confidence) (Section 9.5.1.3). Glacier ice volume from 1977 to 2018 in New Zealand has 38 decreased from 26.6 km3 to 17.9 km3 (a loss of 33%) (Salinger et al., 2019). Relative to 2015, glaciers in 39 New Zealand are projected to lose 36 ± 44%, 53 ± 33%, and 77 ± 27% of their mass by the end of the century 40 under RCP2.6, RCP4.5, and RCP8.5, respectively, with the loss rates decreasing over time under RCP2.6 41 and increasing under RCP8.5 (Marzeion et al., 2020). 42 43 In summary, snowfall is expected to decrease throughout the region at high altitudes in both Australia 44 (high confidence) and New Zealand (medium confidence). In New Zealand, glacier ice mass and extent 45 are expected to decrease over the 21st century for all scenarios (high confidence). 46 47 48 12.4.3.5 Coastal and Oceanic 49 50 Relative sea level: Around Australasia, over 1900-2018, a new tide-gauge based reconstruction finds a 51 regional-mean RSL change of 1.33 [0.80-1.86] mm yr-1 in the Indian ocean - South Pacific region (Frederikse 52 et al., 2020), compared to a GMSL change of around 1.7 mm yr-1 (Section 2.3.3.3; Table 9.5). For the period 53 1993-2018, the RSLR rates, based on satellite altimetry, increased to 3.65 [3.23-4.08] mm yr-1 (Frederikse et 54 al., 2020), compared to a GMSL change of 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). 55 Do Not Cite, Quote or Distribute 12-55 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Relative sea level is virtually certain to increase throughout the region over the 21st century (Section 9.6.3, 2 Figure 9.28). Regional-mean RSLR projections for the oceans around Australasia range from 0.4 m–0.5 m 3 under SSP1-RCP2.6 to 0.7 m–0.9 m under SSP5-RCP8.5 for 2081-2100 relative to 1995-2014 (median 4 values), which means local RSL change falls within the range of mean projected GMSL change (Section 5 9.6.3.3). However these RSLR projections may be underestimated due to potential partial representation of 6 land subsidence (Section 9.6.3.2). 7 8 Coastal flood: The most commonly used index for episodic coastal inundation in Australia is the summation 9 of a high end SLR and the 1:100 yr storm tide level (the combined sea level due to storm surge and tide) 10 (CSIRO and BOM, 2016; McInnes et al., 2016). However, episodic coastal flooding is caused by Extreme 11 Total Water Levels (ETWL), which is the combination of SLR, tides, surge and wave setup (12.3.5.2). The 12 present day 1:100 yr ETWL is between 0.5 m – 2.5 m around most of Australia, except the northwestern 13 coast where 1:100 yr ETWL can be as large as 6 m–7 m (Vousdoukas et al., 2018; O’Grady et al., 2019; 14 Kirezci et al., 2020). 15 16 Extreme total water level magnitude and occurrence frequency are expected to increase throughout the 17 region (high confidence) (see Figure 12.4p-r and Figure SM 12.6). Across the region, the 5th – 95th percentile 18 range of the 1:100 yr ETWL is projected increase (relative to 1980 – 2014) by 5 cm – 35 cm and by 10 cm – 19 40 cm by 2050 under RCP4.5 and RCP8.5, respectively (Figure 12.4q). By 2100 (Figure 12.4p,r), this range 20 is projected to be 25 cm – 80 cm and 50 cm – 190 cm under RCP4.5 and RCP8.5, respectively (Vousdoukas 21 et al., 2018; Kirezci et al., 2020). Furthermore, the present day 1:100 yr ETWL is projected to have median 22 return periods of around 1:20 yrs by 2050 and 1:1 yrs by 2100 in SAU and NZ and return periods of around 23 1:50 yrs by 2050 and 1:20 yrs by 2100 in NAU under RCP4.5 (Vousdoukas et al., 2018), while the present 24 day 1:50 yr ETWL is projected to occur around 3 times a year by 2100 with a SLR of 1 m around Australasia 25 (Vitousek et al., 2017). 26 27 Coastal erosion: Satellite derived shoreline retreat rates for the period between 1984 – 2015 show retreat 28 rates between 0.5 m yr-1 and 1 m yr-1 around the region, except in SAU where a shoreline progradation rate 29 of 0.1 m yr-1 has been observed (Luijendijk et al., 2018; Mentaschi et al., 2018). Mentaschi et al. (2018) 30 report a coastal area loss of 350 km2 over the same period in West Australia from satellite observations. 31 32 Projections indicate that a majority of sandy coasts in the region will experience shoreline retreat, throughout 33 the 21st century (high confidence) (Figure 12.7b,d). Median shoreline change projections (CMIP5) under 34 both RCP4.5 and RCP8.5 presented by Vousdoukas et al. (2020) show that, by mid-century, sandy shorelines 35 will retreat (relative to 2010) by between 50 m and 80 m all around Australasia, except in SAU and NZ 36 where the projected retreat (relative to 2010) is between 35 m and 50 m. By 2100, median shoreline retreats 37 exceeding 100 m (relative to 2010) are projected along the sandy coasts of NAU (~ 150 m), CAU ( ~ 160 m), 38 and EAU ( ~ 110 m) under RCP4.5m, while projections for SAU and NZ are around 80 m – 90 m. Under 39 RCP8.5, shoreline retreat exceeding 100 m is projected all around the region by 2100 (relative to 2010) with 40 retreats as high as 220 m in NAU and CAU (~ 170 m in EAU and around 130 m in SAU and NZ) (Figure 41 12.7b, d). The total length of sandy coasts in Australasia that is projected to retreat by more than a median of 42 100 m by 2100 under RCP4.5 and RCP8.5 is about 12,500 km and 16,000 km respectively, an increase of 43 approximately 30%. 44 45 Distinct from long term coastline recession, storms and storm surges also result in episodic coastal erosion. 46 In general, the historically measured maximum episodic coastal erosion (either eroded volume or coastline 47 retreat distance) or that due to a 1:100 yr return period storm wave height is used as a design criterion for 48 coastal zone management and planning in Australia (Wainwright et al., 2014; Mortlock et al., 2017). 49 50 While there is wide recognition in Australia that the combined effect of SLR, changing storm surge and 51 wave climates will directly affect future episodic coastal erosion (Harley et al., 2017; McInnes et al., 2016; 52 Ranasinghe, 2016) only a few projections of how this hazard may evolve are available for Australia. In one 53 such study, Jongejan et al. (2016) provide projections of how the full exceedance probability curve of the 54 maximum erosion per year may evolve over the 21st century (due to the combined action of SLR, storm 55 surge and storm waves). Their results show that, for example, the 0.01 exceedance probability maximum Do Not Cite, Quote or Distribute 12-56 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 coastline retreat in 2025 will have an exceedance probability of 0.015 by 2050 and 0.07 by 2100. 2 3 Marine heatwave: The mean SST of the ocean around Australia and east of New Zealand has warmed at a 4 rate of about 0.22ºC per decade between 1992 and 2016 (Wijffels et al., 2018), which is higher than the 5 global average SST increase of 0.16ºC per decade (Oliver et al., 2018). This mean ocean surface warming is 6 connected to longer and more frequent marine heatwaves in the region (Oliver et al., 2018). Over the period 7 1982-2016, the coastal ocean of Australia experienced on average more than 1.5 MHW per year, with the 8 north coast of Western Australia and the Tasman Sea experiencing on average 2.5-3 MHWs per year. The 9 average duration was between 10 and 15 days, with somewhat longer and hotter MHWs in the Tasman Sea. 10 In New Zealand, the SE coast of the South Island experiences the most MHWs (2.5-3.0 per year). The 11 duration of MHW in New Zealand is on average 10-15 days (Oliver et al., 2018). Changes over the 20th 12 century, derived from MHW proxies, show an increase in frequency between 0.3 and 1.5 MHW per decade, 13 except along the south east coast of New Zealand (Box 9.1); an increase in duration per event; and the total 14 number of MHW days per decade, with the change being stronger in the Tasman Sea than elsewhere (Oliver 15 et al., 2018). 16 17 There is high confidence that MHWs will increase around most of Australasia. Under RCP4.5 and RCP8.5 18 respectively, mean SST is projected to increase by 1ºC and 2 ºC around Australia by 2100, with a hotspot of 19 around 2 ºC for RCP4.5 and of 4ºC for RCP8.5 along the southeast coast between Sydney and Tasmania (see 20 Interactive Atlas). Under all RCPs, the mean SST around Australia is expected to increase in the future, with 21 median values of around 0.4-1.0 ºC by 2030 under RCP4.5, and 2-4 ºC by 2090 under RCP8.5 (CSIRO and 22 BOM, 2015). Warming is expected to be largest along the north-west coast of Australia, southern Western 23 Australia, and along the east coast of Tasmania (CSIRO and BOM, 2018). More frequent, extensive, intense 24 and longer-lasting MHWs are projected around Australia and New Zealand for GWLs of 1.5 ºC, 2 ºC and 3.5 25 ºC relative to the modelled reference value for 1861-1880 (Frölicher et al., 2018) Projections for SSP1-2.6 26 and SSP5-8.5 both show an increase in MHWs around Australasia by 2081 – 2100, relative to 1985 – 2014 27 (Box 9.2, Figure 1). 28 29 In general, there is high confidence that most coastal/ocean related hazards in Australasia will increase 30 over the 21st century. Relative sea-level rise is virtually certain to continue in the oceans around 31 Australasia, contributing to increased coastal flooding in low-lying areas (high confidence) and 32 shoreline retreat along most sandy coasts (high confidence). Marine heatwaves are also expected to 33 increase around the region over the 21st century (high confidence). 34 35 The assessed direction of change in climatic impact-drivers for Australasia and associated confidence levels 36 are illustrated in Table 12.5, together with emergence time information (see Section 12.5.2). No assessable 37 literature could be found for Hail and Snow avalanches, although these phenomena may be relevant in parts 38 of the region. 39 40 41 [START TABLE 12.5 HERE] 42 43 Table 12.5: Summary of confidence in direction of projected change in climatic impact-drivers in Australasia, 44 representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, SRES 45 A1B, or above within each AR6 region (defined in Chapter 1), approximately corresponding (for CIDs that 46 are independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for more 47 details of the assessment method). The table also includes the assessment of observed or projected time-of- 48 emergence of the CID change signal from the natural inter-annual variability if found with at least medium 49 confidence in Section 12.5.2. 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-57 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Northern Australia (NAU) 5 7 Central Australia (CAU) 7 Eastern Australia (EAU) 7 Southern Australia (SAU) 1 3 7 New Zealand (NZ) 2 4 6 7 1. High confidence of decrease in southwest Western Australia 2. Medium confidence of decrease in north and east and increase in south and west 3. High confidence of increase in southwest Western Australia Key 4. Medium confidence of increase in the north and east and decrease in south and west High confidence of decrease 5. Low confidence of increasing intensity, and high confidence of decreasing occurrence Medium confidence of decrease 6. High confidence of decrease in Glacier volume, medium confidence of decrease in snow Low confidence in direction of change 7. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. Medium confidence of increase High confidence of increase Not broadly relevant 1 2 [END TABLE 12.5 HERE] 3 4 5 12.4.4 Central and South America 6 7 For the purpose of this assessment, Central and South America is sub-divided into eight sub-regions, as 8 defined in Chapter 1: southern Central America (SCA), north western South America (NWS), northern South 9 America (NSA), South American Monsoon (SAM), north eastern South America (NES), south western 10 South America (SWS), south eastern South America (SES), and southern South America (SSA). The 11 Caribbean is placed under the Small Islands section (12.4.7) of this chapter. 12 13 Previous assessments have documented ongoing and projected changes in several CIDs. IPCC AR5 14 projections (IPCC, 2014b) pointed to increases in mean temperature between 2-6oC by end of the century 15 (high confidence) and increases in the occurrence of warm days and nights under various future climate 16 scenarios (medium confidence). AR5 also pointed to patterns of changes in precipitation (medium 17 confidence), changes in the duration of dry spells (medium confidence), and decreases in water supply (high 18 confidence). IPCC SR1.5 projections indicated expected increases in river flooding and extreme runoff at 19 2°C warming in parts of South America, and decreases in runoff in Central America, central and southern 20 South America, and the Amazon basin. The IPCC SROCC reported an increased number of landslides, a 21 decreased volume of lahars from ice and snow-clad volcanoes, and increased frequency of Glacier Lake 22 Outburst Floods. SROCC also indicated that regional and local-scale projections point to decreasing trends 23 in glacier runoff. 24 25 New literature is now available for the regional climate as a result of observational research and coordinated 26 modelling outputs of CORDEX South America (Solman, 2013; Sánchez et al., 2015) and CORDEX-CORE 27 (Giorgi et al., 2018; Teichmann et al., 2019; Coppola et al., 2021b). Of particular interest are the new 28 projections of both mean climate and extremes. This new regional climate information is key to main sectors 29 sensitive to climate change in Central and South America such as water resources, infrastructure, agriculture, 30 livestock, forestry, silviculture, and fisheries (Magrin, 2015; López Feldman and Hernández Cortés, 2016), 31 human health (changes in morbidity and mortality, and emergence of diseases in previously non-endemic 32 areas; (Núñez et al., 2016)) and biodiversity (Uribe Botero, 2015), urban planning, navigation and tourism. 33 Do Not Cite, Quote or Distribute 12-58 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 [START FIGURE 12.8 HERE] 2 3 Figure 12.8: Projected changes in selected climatic impact-driver indices for Central and South America. (a) 4 Mean change in 1-in-100 year river discharge per unit catchment area (Q100, m 3 s-1 km-2) from CORDEX 5 models for 2041-2060 relative to 1995-2014 for RCP8.5. (b) Shoreline position change along sandy 6 coasts by the year 2100 relative to 2010 (meters; negative values indicate shoreline retreat) from the 7 CMIP5 based data set presented by Vousdoukas et al. (2020). (c) Bar plots for Q100 (m3 s-1 km-2) 8 averaged over land areas for the WGI reference AR6 regions (defined in Chapter 1). The left column 9 within each panel (associated with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute 10 values in grey shades. The other columns (associated with the right y-axis) show the Q100 changes 11 relative to the recent past values for two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for 12 three global warming levels (defined relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C 13 (yellow) and 4°C (brown). The bars show the median (dots) and the 10th-90th percentile range of model 14 ensemble values across each model ensemble. CMIP6 is shown by the darkest colours, CMIP5 by 15 medium, and CORDEX by light. SSP5-8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) Bar 16 plots for shoreline position change show CMIP5 based projections of shoreline position change along 17 sandy coasts for 2050 and 2100 relative to 2010 for RCP8.5 (red) and RCP4.5 (blue) from Vousdoukas et 18 al. (2020). Dots indicate regional mean change estimates and bars the 5 th-95th percentiles ranges of 19 associated uncertainty. Note that these shoreline position change projections assume that there are no 20 additional sediment sinks/sources or any physical barriers to shoreline retreat. See Technical Annex VI 21 for details of indices. Further details on data sources and processing are available in the chapter data table 22 (Table 12.SM.1). 23 24 [END FIGURE 12.8 HERE] 25 26 27 12.4.4.1 Heat and cold 28 29 Mean air temperature: New literature confirms a continuous warming since the beginning of the 20th 30 century in the majority of the eight sub-regions (Atlas.7). However, observational datasets in several areas 31 are still short and trend estimation is hindered by year-to-year and interannual variability. Atlas projections 32 point to a virtually certain warming across all sub-regions, with the largest increases taking place in the 33 Amazon Basin (NSA and SAM) (Atlas.7.2.4). A consistent increase in temperature-related indices linked to 34 several climate-sensitive sectors (GDD, CDD) is found across CMIP5, CMIP6 and CORDEX-CORE 35 projections, with lesser increase for CDD in mid-latitude regions than in SCA and the Amazon (Coppola et 36 al., 2021b). Daily mean temperature exceedances of a typical 21.5°C threshold for a successful incubation of 37 disease pathogens inside many mosquito vectors (Lambrechts et al., 2011; Blanford et al., 2013; Mordecai et 38 al., 2013, 2017) will be crossed much more frequently, potentially driving increases in the incidence of 39 vector-borne diseases (Laporta et al., 2015; Messina et al., 2019). 40 41 Extreme heat: Chapter 11 found high confidence of increased heat waves in all regions but SSA over the 42 past decades. There is evidence of increasing heat stress over summertime in much of SES and SWS using 43 the WBGT index for the period 1973-2012, and this has been attributed to human influence on the climate 44 system (Knutson and Ploshay, 2016). Climate change projections point to major increases in several heat 45 indices across the region for all scenarios (high confidence). Largest increases in the frequency of hot days 46 (maximum temperatures, Tx > 35°C) are projected for the Amazon basin under SSP5-8.5 with more than 47 200 days per year at the end of the century under SSP5-8.5, while such increases remain moderate (50-100 48 days) in SSP1-2.6. For the dangerous heat threshold of HI>41°C, increases in frequency are similar to that in 49 Tx>35°C (Figure 12.4 and Figure SM 1.1 and 1.2) (Coppola et al., 2021b; Schwingshackl et al., 2021). 50 51 Cold spell and frost: A decreasing frequency of cold days and nights has been observed in many sub- 52 regions (Chapter 11). There is medium confidence (limited agreement) of a decrease in frost days in SWS, 53 SES and SSA. Projections consistently suggest a general decrease in the frequency of cold spells and frost 54 days in all sub-regions as indicated by several indices based on minimum temperature (Chou et al., 2014; 55 López-Franca et al., 2016; Li et al., 2020b). Heating Degree Days are consistently projected to decrease by 5 56 degree days per year in the Amazon region, and up to 20-30 degree days in NSA, SWS and SES, under 57 future RCP8.5/SSP5-8.5 (Coppola et al., 2021b). Do Not Cite, Quote or Distribute 12-59 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 In conclusion, it is virtually certain that warming will continue everywhere in Central and South 2 America and there is high confidence that by the end of century most regions will undergo extreme 3 heat stress conditions much more often than in recent past (e.g., increase of dangerous heat HI>41°C, 4 or Tx>35°C) with more than 200 additional days per year under SSP5-8.5, while such conditions will 5 be met up typically 50-100 more days per year under SSP1-2.6 over the same regions. Cold spells and 6 frost days will have a decreasing trend (high confidence). 7 8 9 12.4.4.2 Wet and dry 10 11 Mean precipitation: The Atlas documents diverse historical precipitation trends in the region, including: a 12 small but not significant increasing trend in SCA, a decreasing trend in south-eastern and north-eastern 13 Brazil, and an increasing trend in SSA. Projections indicate a drying signal for SCA (medium confidence) 14 (Coppola et al., 2014; Nakaegawa et al., 2014), NES and SWS (high confidence) (Atlas 7.2.5) and the well- 15 known dipole for South America, meaning increasing precipitation over subtropical regions like the Río de 16 La Plata basin (SES) (high confidence) and decreased precipitation in the Amazon (NSA) (medium 17 confidence) (Chou et al., 2014; Llopart et al., 2014; Reboita et al., 2014; Sánchez et al., 2015b; Teichmann et 18 al., 2019). These features are consistent among observations (Sena et al., 2018) and are consistently evident 19 in regional and global model projections by mid- and end-of-century for both RCP4.5 and RCP8.5 (Jones 20 and Carvalho, 2013) . 21 22 River flood: Emerging literature in the region documents ongoing changes in river floods: (Mernild et al., 23 2018) report decreases and increases in annual runoff on the west of the Andes Cordillera's continental 24 divide, with the greatest decreases in the number of low (<10th percentile) runoff and the greatest increases in 25 high (>90th percentile) runoff conditions. In coastal northeast Peru, extreme precipitation events recently 26 caused devastating river floods and landslides (Son et al., 2020). In Brazil, floods are becoming more (less) 27 frequent and intense in wet (drier) regions (Bartiko et al., 2019; Borges de Amorim and Chaffe, 2019), with 28 higher propagation of hydrological changes through anthropogenically-modified agricultural basins (Chagas 29 and Chaffe, 2018). Record, catastrophic, unprecedented, and once-in-a-century flooding events have also 30 been reported in recent decades in the tributaries of the Amazon river or along its mainstream (Sena et al., 31 2012; Espinoza et al., 2013; Marengo et al., 2013; Filizola et al., 2014), in Argentinean rural and urban areas 32 (Barros et al., 2015), in the lower reaches of the Atrato, Cauca and Magdalena rivers in Colombia (Hoyos et 33 al., 2013; Ávila et al., 2019), in basins whose mainstreams flow through important metropolitan areas such 34 as Concepción, Chile (Rojas et al., 2017), and even in one of Earth's driest regions, the Atacama Desert 35 (Wilcox et al., 2016). In the Amazon basin, the significant increase in extreme flow was associated with the 36 strengthening of the Walker circulation (Barichivich et al., 2018). 37 38 Emerging literature in the region documents projected increases for river floods in SES and SAM (medium 39 confidence). Projections indicate that SES and the coasts of Ecuador and Peru will experience a tendency 40 toward wetter conditions that can be a proxy for longer periods of flooding and enhanced river discharges 41 (Zaninelli et al., 2019). CORDEX models project the strongest changes for the peak flow with a return 42 period of 100 years in SES by mid-century and under RCP8.5. (Figure 12.8). At the continental scale, on the 43 contrary, (Alfieri et al., 2017) suggest that 100-yr river floods are projected to decrease under RCP8.5. 44 Regional projections of river floods have high uncertainty, however; owing to differences in hydrological 45 models (Reyer et al., 2017a) (low confidence). Fábrega et al. (2013) projected increases in surface runoff for 46 Panamá, while Zulkafli et al. (2016) identified increases in 100-yr floods of 7.5% and 12.0% in projections 47 for the Peruvian Amazon wet season under RCPs 4.5 and 8.5, respectively. Wetter conditions and ±20% 48 variations in annual mean streamflow are also projected for the Río de La Plata under the warming levels of 49 1.5°C, 2°C and 3°C above pre‐industrial conditions (Montroull et al., 2018). In central Chile, 50-yr peak 50 flows are expected to be greater by mid-century than 100-yr peak flows observed over the reference period 51 (Bozkurt et al., 2018). 52 53 Heavy precipitation and pluvial flood: Chapter 11 indicated that there is low confidence due to limited 54 evidence of extreme precipitation trends in almost all Central and South America, except in SES where 55 increases in the magnitude and frequency of heavy precipitation have been observed (high confidence). In Do Not Cite, Quote or Distribute 12-60 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 general, data scarcity persists for a representative continental assessment. Chapter 11 projections indicate 2 low confidence of increase, compared to the modern period, in the intensity and frequency of heavy 3 precipitation in SCA and SWS for all GWLs, a medium confidence of increase in NSA, NES, SSA, SAM and 4 SES for GWL of 4°C. In NWS, a wide range of changes is projected (low confidence). 5 6 Landslide: Several regions in Central America, as well as Colombia and southeastern Brazil, are considered 7 areas of high incidence of observed fatal landslides. In these areas, ENSO-driven fluctuations in rainfall 8 amounts (Sepúlveda and Petley, 2015) and climate change (Nehren et al., 2019) seem to be key factors. Rock 9 falls, ice- and rock-ice avalanches, lahars, and landslides have been reported in frequent number in the 10 southern, extratropical Andes in the last decades (Gariano and Guzzetti, 2016). A large number of ice- and 11 moraine-dammed lakes have consequently failed, causing floods that rank amongst the largest events ever 12 recorded (Iribarren Anacona et al., 2015). However, due to research deficit, evidence is largely missing for a 13 reliable assessment of past and future trends of such hazards. 14 15 Aridity: Several new regional studies suggest positive trends in the frequency and length of droughts in the 16 region, such as: over NWS (Domínguez-Castro et al., 2018), NSA (Marengo and Espinoza, 2016; Cunha et 17 al., 2019) and NES (Marengo and Bernasconi, 2015), over southern Amazonia (Fu et al., 2013; Boisier et al., 18 2015), in the São Francisco River Basin and the capital city Distrito Federal in Brazil (Borges et al., 2018; 19 Bezerra et al., 2019), in the southern Andes (Vera and Díaz, 2015), in central southern Chile (Boisier et al., 20 2018), in SES (Rivera and Penalba, 2014), and during recent years in SSA (Rivera and Penalba, 2014). 21 Chapter 8 indicated medium confidence of anthropogenic forcing on observed drying trends in central Chile. 22 Additional discussion on droughts and aridity trends in South America is presented in Chapter 8 (Sections 23 8.3.1.6, 8.4.1.6, 8.6.2.1). 24 25 Chapter 8 projections point to two important drying hotspots in South America with long-term soil moisture 26 decline and precipitation declines, namely the Amazon basin (SAM and NSA) and SWS (medium 27 confidence) (see also Figure 12.4). End-of-century RCP8.5 projections show a longer dry season in the 28 central part of South America and decreased precipitation over the Amazon and central Brazil (Coppola et 29 al., 2014; Giorgi et al., 2014b; Llopart et al., 2014; Teichmann et al. 2013) (Atlas.7). Seasonal changes are 30 also apparent, with decreases in June-July-August rainfall projected for NSA, the coastal region of SES, 31 SAM and the southern portion of the SWS (Marengo et al., 2016). Decreases in December-January-February 32 rainfall are projected for the central part of South America in the near term (Kitoh et al., 2011; Chou et al., 33 2014; Cabré et al., 2016). Regional projections for Central and South America also indicate an increase in 34 dryness in SCA and NES by mid to end of the century (medium confidence) (Chou et al., 2014; Marengo and 35 Bernasconi, 2015). 36 37 Hydrological drought: Chapter 11 assessed mostly low confidence in observed changes in hydrological 38 droughts given a lack of studies and clear evidence, with medium confidence only for a streamflow decrease 39 in subregions of SWS. Some trends are becoming more clear, such as the ones reported for Colombia (NWS) 40 by Carmona and Poveda (2014), who indicated that 62% of the 25- to 50-year long monthly average 41 streamflow time series exhibited significant decreasing trends. However, studies of discharge changes 42 indicate that uncertainty is still large, as argued by Pabón-Caicedo et al. (2020) for the full extent of the 43 Andes. 44 45 A number of studies project decreases in runoff and river discharge for SCA, Colombia, Brazil and the 46 southern part of South America by the end of this century (Nakaegawa and Vergara, 2010; Arnell and 47 Gosling, 2013; van Vliet et al., 2013). Section 11.9 assessed high confidence in projections of increases in 48 hydrological droughts in NSA, SAM, SWS, and SSA under a +4°C GWL, medium confidence in SCA, and 49 low confidence in the rest of the sub-regions given insufficient evidence, lack of signal or mixed signals 50 among the available studies. Signals are much more uncertain for the middle of the century (or for a 2°C 51 GWL), with only SAM projected to have increased hydrological droughts with medium confidence. 52 53 Agricultural and ecological drought: Section 11.9 assessed low confidence in observed changes to 54 agricultural and ecological drought across Central and South America due to regional heterogeneity and 55 differences depending on the drought metrics used, except in NES which has seen a dominant increase in Do Not Cite, Quote or Distribute 12-61 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 drought severity (medium confidence). 2 3 NSA and SAM are the two regions where the strongest signal of increasing number of dry days (NDD) and 4 drought frequency (DF) is projected compared to other regions of the world (Coppola et al., 2021b). By the 5 end of this century and under RCP8.5, the NSA area average value for NDD reaches 43, 32 and 27, within 6 the CORDEX-CORE, CMIP5 and CMIP6 ensembles, respectively. For the frequency of droughts, the NSA 7 area average value is of about 4.6, 3.4 and 3.8. The SAM region shows NDD and DF values of 29, 20 and 8 21, and of 4, 3 and 3.5, respectively (see also Figure 12.4j-l). In Central America, a significantly drier 9 northern region and a wetter southern region were projected for mid-century by (Hidalgo et al., 2017), whilst 10 Fuentes-Franco et al. (2015) pointed to more pronounced dry periods during the rainy season in SCA by the 11 end of this century under RCP8.5. Increases in the frequency of meteorological droughts that may initiate 12 other drought types are projected for the eastern part of the Amazon and the opposite for the west under 13 RCP8.5 (Duffy et al., 2015). In central Chile, the occurrence of extended droughts, such as the recently- 14 experienced 2010-2015 mega-drought (which is still driving impacts), is projected to increase from one to up 15 to five events per 100 years under RCP8.5 (Bozkurt et al., 2018). Section 11.9 highlights change in 16 confidence in increases in drought severity in SCA, NSA, NES, SAM, SWS, and SSA from low to high 17 under the three GWLs of +1.5°C, +2°C and +4°C. NES and SES change from low confidence to medium 18 confidence increases in agricultural and ecological drought severity by +4°C GWL with different metrics and 19 high agreement between studies. Only SAM and SSA have projections of agricultural and ecological drought 20 increasing with high confidence for the middle of the century, or for a +2°C GWL, and NSA, NES and SCA 21 are increasing with medium confidence. 22 23 Fire weather: There is evidence of increases in forest fire activity (number of fires, burned area and fire 24 duration) such as in central and south-central Chile, where more conducive fire weather conditions have been 25 proposed as the main driver (González et al., 2018; Urrutia-Jalabert et al., 2018). Projections indicate that the 26 Amazon is one of the regions in the world with the highest increase in fire weather indices over the 21st 27 century and under all RCPs (high confidence) (Betts et al., 2015; Abatzoglou et al., 2019; Sun et al., 2019b). 28 This is consistent with the large increase in the frequency of joint occurrence of extreme hot and dry days for 29 a 2°C warming level or more (Vogel et al., 2020). Projections of fire weather indices also show an increased 30 risk in SWS (high confidence), SSA and SCA (medium confidence). However, wildfires highly depend on 31 land-use and appropriate management may help mitigate future increases in fire risk (Fonseca et al., 2019). 32 33 Mean precipitation is projected to change in a dipole pattern with increases in NWS and SES and 34 decreases in NES and SWS (high confidence) with further decreases in NSA and SCA (medium 35 confidence). There is medium confidence of an increase in river floods in SAM and SES. There is high 36 confidence of a dominant increase in drought duration in NES, an in the number of dry days and 37 drought frequency in NSA and SAM. Dry climatic impact-drivers have more prevalent regional 38 increases with higher global warming levels, with strongest projections of agricultural and ecological 39 drought (high confidence) and aridity and fire changes (medium confidence) over the Amazon and with 40 both drought types increasing over the SWS (medium confidence). 41 42 43 12.4.4.3 Wind 44 45 Mean wind speed: Due to the lack of long-term homogeneous records or limited observations in the region, 46 past wind speed trends are difficult to establish. Global climate models project an increase in wind speeds, 47 under all future scenarios, augmenting wind power potential in most parts of Central and South America, 48 especially in NES where changes lie in the range 0-20% by 2050 under RCP8.5 and 0-40% under RCP8.5 49 (medium confidence) (Karnauskas et al., 2018a; Reboita et al., 2018; Jung and Schindler, 2019). In 50 Patagonia, wind speeds are projected to decrease. For RCP4.5 changes remain marginal and have low 51 confidence (low agreement). Further information in (Figure 12.4m-o). 52 53 Severe wind storm: Similar observational limitations inhibit an assessment of long-term extreme wind 54 trends. However, (Pes et al., 2017) found extreme wind increases in most of Brazil over the past decades. 55 Future projections indicate a slight decrease in the number of extra-tropical cyclones in mid latitudes (limited Do Not Cite, Quote or Distribute 12-62 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 evidence, low confidence) (Reboita et al., 2018), and an increase of extreme winds in tropical areas (limited 2 evidence, low confidence) (Kumar et al., 2015). Climate models project a shift and an intensification of 3 southern storm tracks, with most effects offshore over the Southern ocean (Chapter 4), with low confidence 4 (low agreement) of significant extreme wind changes over land and coastal areas across the 21st century 5 (Chang, 2017; Augusto Sanabria and Carril, 2018; Reboita et al., 2020). 6 7 Tropical cyclone: CMIP5 and CMIP6 simulations, including the new HighResMIP, project a decrease in the 8 frequency of tropical cyclones in the Atlantic and Pacific coasts of Central America for the mid-century or 9 under a 2°C global warming level, accompanied with an increased frequency of intense cyclones (medium 10 confidence) (Diro et al., 2014; Knutson et al., 2020; Roberts et al., 2020) (Chapter 11). 11 12 In summary, there is limited evidence of current trends in observed wind speed and wind storms in 13 South America. Climate projections indicate a decrease in frequency of tropical cyclones in Central 14 America accompanied with an increased frequency of intense cyclones, and an increase in mean wind 15 speed and wind power potential in NES, NSA and SAM (medium confidence). 16 17 18 12.4.4.4 Snow and ice 19 20 Snow: Studies on seasonal snow cover are limited and restricted to the Andes Cordillera. Mernild et al. 21 (2017) indicated that much of the area north of 23°S experienced a decrease in the number of snow cover 22 days, while the southern half experienced the opposite. A reduction in snow cover of about 15% was 23 simulated for areas in the range [3000-5000 m], whereas in regions with altitude below 1000 m (Patagonia) 24 snow cover extent increased. The reduced snowfall over the Chilean and Argentinean Andes mountains, 25 which has resulted in unprecedented reductions in river flow, reservoir volumes and groundwater levels, led 26 to the most severe and long-lasting hydrological drought (2010-2015) reported in the adjacent semi-arid 27 foothills of the central Andes (Garreaud et al., 2017; Rivera et al., 2017; Masiokas et al., 2020). Projections 28 (based on process understanding) in Chapter 9 point to decreases in seasonal snow cover extent and duration 29 across South America as global climate continues to warm (high confidence). 30 31 Glacier: Observation and future projection of Central and South America glacier mass changes are assessed 32 in Section 9.5.1, grouped in two main regions: Low Latitude region (98% of which is glaciers in the Andes) 33 and the Southern Andes region. An increase in the number and areal extent of glacial lakes in the Southern 34 Andes was reported for the period 1986-2016 (Wilson et al., 2018). Similar changes are being observed in 35 the central Andes (Colonia et al., 2017). Since 1800 at least 15 ice‐dammed lakes and 16 moraine‐dammed 36 lakes have failed in the extratropical Andes, causing high-magnitude Glacial Lake Outburst Floods (Rojas et 37 al., 2014; Cook et al., 2016; Wilson et al., 2018; Drenkhan et al., 2019). Partially due to glaciers shrinkage 38 and lake growth, the frequency of outburst floods has increased in the last 30-40 years (Carey et al., 2012; 39 Iribarren Anacona et al., 2015). 40 41 Glaciers across South America are expected to continue to lose mass and glacier area in the coming century 42 (high confidence) (Section 9.5). In term of their mass, glaciers in the Low Latitude region are projected by 43 GlacierMIP to lose 67 ± 42%, 86 ± 24% and 94 ± 13%, of their 2015 baseline by the end of the century 44 under RCP2.6, RCP4.5 and RCP8.5 scenarios, respectively (Marzeion et al., 2020). Glaciers in the Southern 45 Andes show decreasing mass loss rates for RCP2.6, and increasing rates for RCP8.5, which peak in the mid 46 to late 21st century. Glaciers in the Southern Andes are projected to lose 26 ± 27%, 33 ± 26%, and 47 ± 26% 47 of their mass in 2015 by the end of the century under RCP2.6, RCP4.5 and RCP8.5 scenarios, respectively. 48 Details in (Section 9.5.1.3). 49 50 Permafrost: There is limited information on the ongoing changes and projections of permafrost conditions 51 in the region. Based on model projections under the IPCC A1B scenario, permafrost areas in the Bolivian 52 Andes will eventually be lost, but this could take years to decades or longer depending on permafrost 53 thickness (Rangecroft et al., 2016). The central Andes will experience the highest disturbance to the thermal 54 regime of the twenty-first century. As a consequence, in the Argentinian Andes up to 95% of rock glaciers in 55 the southern Desert Andes and in the central Andes will rest in areas above 0 °C under the worst case Do Not Cite, Quote or Distribute 12-63 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 scenario of warming (the freezing level might move up more than twice as much as during the entire 2 Holocene) (Drewes et al., 2018). 3 4 In conclusion, glacier volume loss and permafrost thawing will likely continue in the Andes Cordillera 5 under all climate scenarios, causing important reductions in river flow and potentially high-magnitude 6 Glacial Lake Outburst Floods. 7 8 9 12.4.4.5 Coastal and oceanic 10 11 Relative sea level: Around Central and South America, over 1900-2018, a new tide-gauge based 12 reconstruction finds a regional-mean relative sea level (RSL) change of 2.07 [1.36 to 2.77] mm yr-1 in the 13 South Atlantic, 2.49 [1.89 to 3.06] mm yr-1 in the subtropical North Atlantic and 1.20 [0.76 to1.62] mm yr-1 14 in the East Pacific (Frederikse et al., 2020), compared to a global mean sea level (GMSL) change of around 15 1.7 mm yr-1 (Section 2.3.3.3; Table 9.5). For the period 1993-2018, these RSLR rates, based on satellite 16 altimetry, increased to 3.45 [3.04to 3.86)] mm yr-1, 4.04 [2.77 to 5.24] mm yr-1 and 2.35 [0.70 to 4.06] mm 17 yr-1, respectively (Frederikse et al., 2020), compared to a GMSL change of 3.25 mm yr-1 (Section 2.3.3.3; 18 Table 9.5). 19 20 Relative sea-level rise is extremely likely to continue in the oceans around CSA. Regional-mean RSLR 21 projections for the oceans around Central and South America range from 0.3 m–0.5 m under SSP1-RCP2.6 22 to 0.5 m–0.9 m under SSP5-RCP8.5 for 2081-2100 relative to 1995-2014 (median values), which is around 23 the projected GMSL change {Section 9.6.3.3}. These RSLR projections may however be underestimated due 24 to potential partial representation of land subsidence in their assessment (Section 9.6.3.2). 25 26 Coastal flood: Present day 1:100 yr Extreme Total Water Level (ETWL) ranges from 0.5 to 2.5 m around 27 most of Central and South America, except in SSA and SWS subregions where 1:100 yr ETWLs can be as 28 large as 5 to 6 m (Vousdoukas et al., 2018). ETWL magnitude and occurrence frequency are expected to 29 increase throughout the region (high confidence) (see Figure 12.4p-r and Figure SM 12.6). Across the region, 30 the 5th – 95th percentile range of the 1:100 yr ETWL is projected increase (relative to 1980 – 2014) by 8 cm – 31 34 cm and by 10 cm – 43 cm by 2050 under RCP4.5 and RCP8.5, respectively (Vousdoukas et al., 2018; 32 Kirezci et al., 2020). By 2100, this range is projected to be 21 cm – 93 cm and 34 cm – 190 cm under 33 RCP4.5 and RCP8.5, respectively (Vousdoukas et al., 2018; Kirezci et al., 2020). Furthermore, under 34 RCP4.5, the present day 1:100 yr ETWL is projected to have median return periods of 1:10 yrs – 1:50 yrs by 35 2050 and 1:1 yr by 2100 in SES, SSA and SWS. In other regions of Central and South America, the present 36 day 1:100 yr ETWL is projected to occur once per year or more by both 2050 and 2100 under RCP4.5. The 37 present day 1:50 yr ETWL is projected to occur around 3 times a year by 2100 with a SLR of 1 m (Vitousek 38 et al., 2017). 39 40 Coastal erosion: According to satellite data, shoreline retreat rates of around 1 m yr-1 have been observed 41 along the sandy coasts of SCA, SES and SSA over the period 1984-2015, while shoreline progradation rates 42 of around 0.25 m /yr has been observed in NWS and NSA. The sandy shorelines in NES and SWS have 43 remained more or less stable over the same period (Luijendijk et al., 2018; Mentaschi et al., 2018). Using 44 satellite observations, Mentaschi et al. (2018) report a coastal area loss of 250 km2 over a 30-year period 45 (1984-2015) along the Pacific coast of South America, and of 780 km2 along the Atlantic coastlines. 46 47 Projections indicate that a majority of sandy coasts in the region will experience shoreline retreat throughout 48 the 21st century (high confidence). Median shoreline change projections (CMIP5) for the mid-century period 49 show that, relative to 2010, presented by Vousdoukas et al. (2020) show that sandy shorelines will retreat by 50 between 30 m and 75 m in SCA, NES, SES and SSA under both RCP4.5 and RCP8.5, while the projected 51 mid-century retreats are less than 30 m in NSA, NWS and SWS for both RCPs (Vousdoukas et al., 2020). 52 Parts of the coastline in these latter three regions projected to prograde over the 21st century, if present 53 ambient shoreline change trends continue (Vousdoukas et al., 2020). By 2100, median retreats of more than 54 100 m are projected in SCA, NES, SES and SSA under both RCPs, while retreats between 50 m – 100 m are 55 projected for NSA, NWS, and SWS under both RCPs (Figure 12.8). Notably, the projected shoreline retreats Do Not Cite, Quote or Distribute 12-64 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 in SCA and SES approach 150 m by 2100 under RCP4.5 and 200 m under RCP8.5. The total length of sandy 2 coasts in Central and South America that is projected to retreat by more than a median of 100 m by 2100 3 under RCP4.5 and RCP8.5 is about 15,000 km and 12,000 km respectively, an increase of approximately 4 30%. 5 6 Marine heatwave: The mean sea surface temperature (SST) of the Atlantic Ocean and the Caribbean around 7 Central and South America increased from 0.25°C to 1°C over the period 1982-1998 (Oliver et al., 2018). 8 This mean ocean surface warming is connected to longer and more frequent marine heatwaves (MHW) in the 9 region (Oliver et al., 2018). Over the period 1982-2016, the coastlines experienced on average more than 1.0 10 MHW per year, with the Pacific coast of North Central America and the coast of SES (Atlantic) experiencing 11 on average 2.5-3.0 MHWs per year. The average duration was between 10 and 15 days, with the notable 12 exception of the Equatorial Pacific coastline, which experiences MHWs with >30 days average duration 13 related to ENSO conditions. In the Southwestern Atlantic shelf (32-38°S), more than half of the days with 14 MHWs have occurred since 2014, and the most intense event (1.7°C above previous maximum) took place 15 in the austral summer of 2017 (Manta et al., 2018). Changes over the 20th century, derived from MHW 16 proxies, show an increase in frequency between 0.5 and 2.0 MHW per decade in the South Atlantic, the 17 Caribbean and the Pacific coast of North Central America, an increase in intensity per event in the South 18 Atlantic, and a decrease along the Equatorial Pacific coastline. The total number of MHW days per year 19 increases around Central and South America, with the exception of the Equatorial Pacific coastline (Oliver et 20 al., 2018). 21 22 There is high confidence that MHWs will increase around Central and South America. Mean SST is 23 projected to increase by 1ºC (2ºC) by 2100, with a hotspot of about 2ºC (4ºC) along the coast of South East 24 South America and North West South America under RCP4.5 (RCP8.5). (See Interactive Atlas). More 25 frequent, MHWs are projected around the region for GWLs of 1.5 ºC, 2 ºC and 3.5 ºC relative to the 26 modelled reference value for 1861-1880 (Frölicher et al., 2018). Projections for SSP1-2.6 and SSP5-8.5 both 27 show an increase in MHWs around Central and South America by 2081 – 2100, relative to 1985 – 2014 (Box 28 9.2, Figure 1). 29 30 In summary, relative sea-level rise is extremely likely to continue in the oceans around Central and 31 South America, contributing to increased coastal flooding in low-lying areas (high confidence) and 32 shoreline retreat along most sandy coasts (high confidence). Marine heatwaves are also expected to 33 increase around the region over the 21st century (high confidence). 34 35 The assessed direction of change in climatic impact-drivers for Central and South America and associated 36 confidence levels are illustrated in Table 12.6. No assessable literature could be found for sand and dust 37 storm, lake and sea ice, heavy snowfall and ice storms, hail, and snow avalanches, although these 38 phenomena may be relevant in parts of the region. 39 40 41 [START TABLE 12.6 HERE] 42 43 Table 12.6: Summary of confidence in direction of projected change in climatic impact-drivers in Central and South 44 America, representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3- 45 4.5, SRES A1B, or above within each AR6 region (defined in Chapter 1), approximately corresponding 46 (for CIDs that are independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 47 for more details of the assessment method). The table also includes the assessment of observed or projected 48 time-of-emergence of the CID change signal from the natural inter-annual variability if found with at least 49 medium confidence in Section 12.5.2. 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-65 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Southern Central America (SCA) 2 3 Northwestern South America (NWS) 3,4 Northern South America (NSA) 2 3,4 South American Monsoon (SAM) 1 Northeastern South America (NES) 3 Southwestern South America (SWS) 3,4 Southeastern South America (SES) 3 Southern South America (SSA) 3 1. Increase in extreme flow in the Amazon basin. Key 2. Tropical cyclones decrease in number but increase in intensity High confidence of decrease 3. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. Medium confidence of decrease 4. Substantial parts of the NWS, NSA and NES coasts are projected to prograde if present-day ambient shoreline change rates continue. Low confidence in direction of change Medium confidence of increase High confidence of increase Not broadly relevant 1 2 [END TABLE 12.6 HERE] 3 4 5 12.4.5 Europe 6 7 The regional European climate and main hazards have been previously assessed in SREX, AR5 (WGII), 8 SR1.5, SROCC and SRCCL and a summary of key findings can be found in the Europe section of the Atlas 9 8.1. For the purpose of these assessment Europe is divided in four climatic regions North Europe (NEU), 10 Western and Central Europe (WCE), Eastern Europe (EE) and Mediterranean (MED) (see Figure Atlas.27). 11 12 Since AR5 and SR1.5, a large body of literature that uses the EURO-CORDEX and MED-CORDEX 13 ensembles of high-resolution simulations (Jacob et al., 2014; Ruti et al., 2016; Kjellström et al., 2018; 14 Vautard et al., 2020; Coppola et al., 2021a) to assess signals of climate change in Europe has emerged. 15 These scenario simulations have been the basis of a number of impact studies (see e.g., Jacob et al., 2014b, 16 2018; Somot et al., 2018; Faggian and Decimi, 2019) ) highlighting the use of the climatic impact-drivers. 17 The development of the science of attribution of weather events (Stott et al., 2016) has provided evidence of 18 links between climate change and hazard changes such as the 2017 Mediterranean heat wave (Kew et al., 19 2019) and many others (see Chapter 11). 20 21 The ability of global and regional models to reproduce the observed changes in mean and extreme 22 temperature and precipitation in Europe is assessed in the literature (see Atlas 8.3). In summary, both GCMs 23 and RCMs have their limitations but, in general, the increased resolution of RCMs is shown to clearly add 24 valued in terms of resolving spatial patterns and seasonal cycles of precipitation and precipitation extremes 25 in many European regions, especially in regions of complex topography such as the Alps and for quantities 26 such as snowmelt driven runoff, regional winds and Mediterranean hurricanes (Medicanes). 27 28 Examples of projected hazard thresholds are illustrated in Figures 12.4 and 12.9 based on the most recently 29 updated Euro-CORDEX RCM projections, CMIP5 and CMIP6 GCMs for comparison. For a more 30 comprehensive representation of other hazard index trends assessed in this section the reader is referred to 31 the interactive Atlas. 32 Do Not Cite, Quote or Distribute 12-66 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 [START FIGURE 12.9 HERE] 2 3 Figure 12.9: Projected changes in selected climatic impact-driver indices for Europe. (a) Mean change in 1-in-100 4 year river discharge per unit catchment area (Q100, m3 s-1 km-2), and (b) median change in the number of 5 days with snow water equivalent (SWE) over 100 mm (from November to March), from EURO- 6 CORDEX models for 2041-2060 relative to 1995-2014 and RCP8.5. Diagonal lines indicate where less 7 than 80% of models agree on the sign of change. (c) Bar plots for Q100 (m3 s-1 km-2) averaged over land 8 areas for the WGI reference AR6 regions (defined in Chapter 1). The left column within each panel 9 (associated with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute values in grey shades. 10 The other columns (associated with the right y-axis) show the Q100 changes relative to the recent past 11 values for two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for three global warming levels 12 (defined relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C (yellow) and 4°C (brown). 13 The bars show the median (dots) and the 10th-90th percentile range of model ensemble values across each 14 model ensemble. CMIP6 is shown by the darkest colours, CMIP5 by medium, and CORDEX by light. 15 SSP5-8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) As for (c) but showing absolute 16 values for number of days with SWE > 100mm, masked to grid cells with at least 14 such days in the 17 recent-past. See Technical Annex VI for details of indices. Further details on data sources and processing 18 are available in the chapter data table (Table 12.SM.1). 19 20 [END FIGURE 12.9 HERE] 21 22 23 12.4.5.1 Heat and Cold 24 25 Mean air temperature: Since AR5, studies have confirmed that the mean warming trend in Europe is 26 increasing (Atlas.8.2). The observed warming trend patterns are largely consistent with those simulated by 27 global and regional climate models and it is very likely that such trends are, in large part, due to human 28 influence on climate (Chapter 3). 29 30 All temperature trends are very likely to continue for a global warming of 1.5°C or 2°C and 3oC (Atlas 8.4). 31 Future warming leads to the exceedance of different temperature thresholds relevant for vector-borne 32 diseases (medium confidence) (Caminade et al., 2012; Medlock et al., 2013), invasive allergens (medium 33 confidence) (Storkey et al., 2014; Hamaoui-Laguel et al., 2015), SST thresholds in the Mediterranean (likely 34 to exceed 20°C), or relevant for the vibrio bacteria development (Vezzulli et al., 2015). Future warming is 35 also projected to lead to the exceedance of cooling degree-day index (>22°C) thresholds, characterizing a 36 potential increase in energy demand for cooling in Southern Europe with increases likely exceeding 40% in 37 some areas (Spinoni et al., 2015) by 2050 under RCP8.5 (high confidence) (Coppola et al., 2021a) (see also 38 Atlas and Section 12.3). 39 40 Extreme heat: The frequency of heat waves observed in Europe has very likely increased in recent decades 41 due to human-induced change in atmospheric composition (see Chapter 11) and a detectable anthropogenic 42 increase in a summertime heat stress index over all regions of Europe has been identified based on wet bulb 43 globe temperature (WBGT) index trends for 1973-2012 (medium confidence, limited evidence) (Knutson and 44 Ploshay, 2016). 45 46 It is very likely that the frequency of heat waves will increase during the 21st century regardless of the 47 emission scenario in each European region, and for 1.5°C and 2°C global warming levels (GWLs) (Chapter 48 11). Heat stress due to both high temperature and humidity, affecting morbidity, mortality and labor capacity 49 (see Section 12.3) is projected to increase under all emission scenarios and GWLs by the middle of the 50 century (Figure 12.4a-f). Under RCP8.5, the expected number of days with WBGT larger than 31°C of about 51 25, 30 and 40 days per year, projected by EURO-CORDEX, CMIP5 and CMIP6 respectively on average 52 over the Mediterranean region, and specifically of 30, 40 and 60 days per year in low coastal plain areas such 53 as the Po Valley, the Italian, Greek and Spanish coasts and the Mediterranean islands in (Coppola et al., 54 2021a). An average increase of a few days per year of maximum daily temperature exceeding 35°C, a typical 55 critical threshold for crop productivity, is expected by the mid-century in Central Europe, and a further 56 increase of 10-20 days is expected for the Mediterranean areas (see Figure 12.4b) (Coppola et al., 2021a). By Do Not Cite, Quote or Distribute 12-67 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 contrast, under SSP1-2.6, the increase in this number of days remain limited to less than about 10 days, and 2 confined to the Mediterranean regions. Mitigation is expected to have a strong effect: by the end of century, 3 and under SSP5-8.5, dangerous heat threshold of HI>41°C is projected to be crossed 5-10 days more per 4 year in the Mediterranean regions and a few days per year more in WCE and EEU while these increases 5 would virtually be absent under SSP1-2.6 (Figure 12.4d-f). 6 7 Cold spell and frost: Temperature observations for winter cold spells in Europe show a long-term 8 decreasing frequency (Brunner et al., 2018), with their probability of occurrence projected to decrease in the 9 future (high confidence) and virtually disappear by the end of the century (Chapter 11). The frequency of 10 frost days will very likely decrease for all scenario and all time horizons (Lindner et al., 2014; Coppola et al, 11 2021) with consequences on agriculture and forests. A simple heating degree-day index, characterizing 12 heating demand, shows a large observed decreasing trend for winter heating energy demand in Europe 13 (Spinoni et al., 2015),. This trend is very likely to continue through the 21st century, with decreases in the 14 range of 20-30% for Northern Europe, about 20% for Central Europe and 35% for Southern Europe, by mid- 15 century under RCP8.5 (Spinoni et al., 2018b; Coppola et al., 2021a) (See also the interactive Atlas). 16 17 In summary, irrespective of the scenario, it is virtually certain that warming will continue in Europe, 18 and there is high confidence that the observed increase in heat extremes is due to human activities. It is 19 very likely that the frequency of heat extremes will increase over the 21st century with an increasing 20 gradient toward the Southern regions. Extreme heat will exceed critical thresholds for health, 21 agriculture and other sectors more frequently (high confidence), with strong differences between 22 mitigation scenarios. It is very likely that the frequency of cold spells and frost days will keep 23 decreasing over the course of this century and it is likely that cold spells will virtually disappear 24 towards the end of the century. 25 26 27 12.4.5.2 Wet and Dry 28 29 Mean precipitation: Precipitation has generally increased in Northern Europe and decreased in Southern 30 Europe, especially in winter (Fischer and Knutti, 2016, Knutson and Zeng, 2018) but in the latter 31 precipitation trends are strongly dependent on the examined period (Chapter 11). The trend in precipitation 32 increase in the north and decreasing in the south is also represented by global and regional climate 33 simulations (Jacob et al., 2014; Rajczak and Schär, 2017; Lionello and Scarascia, 2018; Coppola et al., 34 2021a) (see Atlas.8.2) and has been attributed to climate change (Chapter 3, Chapter 8). 35 36 Studies since AR5, together with EURO-CORDEX and MED-CORDEX experiments and the latest CMIP6 37 ensemble, have increased confidence in regional projections of mean and extreme precipitation (Prein et al., 38 2016) despite their wet bias, and show that it is very likely that precipitation will increase in Northern Europe 39 in DJF and decrease in the Mediterranean in JJA under all climate scenarios except RCP2.6/SSP1-2.6 and 40 for both mid and end century periods (Coppola et al., 2021a). (Atlas 8.5). 41 42 River flood: There is high confidence of an observed increasing trend of river floods in Central and Western 43 Europe (WCE) and medium confidence of a decrease in northern (NEU) and Southern Europe (MED). 44 45 SR1.5 shows evidence of an increase in reported floods in the UK over the period 1884-2013, and increasing 46 trends in annual maximum daily streamflow data over 1966–2005 in parts of Europe. Although, high flow 47 does not show uniform trends for the entire region (Hall et al., 2014; Mediero et al., 2015) or specific region 48 (Mudersbach et al., 2017; Tramblay et al., 2019; Vicente-Serrano et al., 2017), regional patterns of 49 significant flood trends do exist. Based on the most extended river flow database spanning the period 1960- 50 2010, an increase in floods frequency in northwestern Europe, decreasing in medium and large catchments 51 in southern Europe and decreasing floods in eastern Europe has been detected (Blöschl et al., 2019) in line 52 with (Mediero et al., 2014; Arheimer and Lindström, 2015; Gudmundsson et al., 2017; Krysanova et al., 53 2017; Kundzewicz et al., 2018; Mangini et al., 2018) 54 55 There is high confidence of river floods increasing in Central and Western Europe (WCE) and medium Do Not Cite, Quote or Distribute 12-68 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 confidence of a decrease in northern (NEU) Eastern (EE) and Southern Europe (MED) for mid and end of 2 century under RCP8.5 and low confidence under RCP2.6. The projected increase in WCE is roughly 10% 3 (18% by end of century) and the projected decrease in NEU is of 5% (11% by end of century) for the peak 4 flow with a return period of 100 years for mid-century, under RCP8.5 (Di Sante et al., 2021) (See Figure 5 12.9a for mid of century Q(100) projections of flood discharges per unit catchment area (Blöschl et al., 2019) 6 based on EURO-CORDEX models). 7 8 Using frequency analysis of extreme peak flow events above a 100-year return period as a threshold, which 9 is the average protection level of the European river network (Rojas et al., 2013), Alfieri et al. (2017) and 10 Alfieri et al. (2015) show that Europe is likely to be one of the regions where the largest increases in flood 11 risk may occur, with only few countries in Eastern Europe showing a decrease (Poland Lithuania, Belarus) 12 (Osuch et al., 2017). They find a significant increase of events with peak discharge above 100-year return 13 period (Q100) in most of Europe in line with (Rojas et al., 2012; Hirabayashi et al., 2013; Dankers et al., 14 2014; Forzieri et al., 2016; Roudier et al., 2016; Thober et al., 2018) and an increase in the magnitude of 15 floods in southern Europe, although (Giuntoli et al., 2015) projects no change. A modest but significant 16 decrease in the 100-year return period river flood is projected for Southern (due to reduction of precipitation) 17 and north Eastern European regions the latter because of the strong reduction in snowmelt induced river 18 floods (Thober et al., 2018; Di Sante et al., 2021). 19 20 Heavy precipitation and pluvial flood: Heavy precipitation frequency trends have been detected in Europe 21 with high confidence for the NEU and Alpine regions and with medium confidence in WCE, and also 22 attributed to climate change with high confidence in NEU (Section 11.9). 23 24 Projections based on multiple lines of evidence from global to convective permitting model scales show high 25 confidence in extreme precipitation increase in the northern, central and eastern European regions (NEU, 26 WCE, EEU and in the Alpine area, Increases with medium confidence are projected for the Mediterranean 27 basin (with a negative gradient toward the south) for mid and end of century under RCP8.5, RCP4.5, and 28 SSP5-8.5 (Section 11.9) (MedECC, 2020). Guerreiro et al. (2017), based on observations, showed that 20% 29 of city areas in WCE and MED regions is affected by pluvial flooding and less than 10% of city areas in the 30 Northern and Western coastal cities. 31 32 Landslide: Increase of rainfall periods connected to landslides are projected to increase in central Europe by 33 up to 1 more period per year in flat areas in low altitudes and by up to 14 more periods per year at higher 34 altitudes by mid-century, becoming even more evident by end of the century (Schlögl and Matulla, 2018). 35 An increase of landslides by up to + 45.7% and + 21.2% is also projected for Southern Italy (Calabria 36 region) by mid of century under both RCP4.5 and RCP8.5 (Gariano and Guzzetti, 2016) and by up to 40% in 37 Central Italy (Umbria) during the winter season (Ciabatta et al., 2016). A decrease of landslides is projected 38 in the Peloritani Mountains in Southern Italy (RCP4.5 and 8.5) by mid of the century (Peres and Cancelliere, 39 (2018)). A slight increase (10 year return period) in landslides is projected in the Eastern Carpathians, the 40 Moldavian Subcarpathian and the northern part of the Moldavian Tableland and higher increase in the 41 western hilly and plateau areas of Romania (100 year return period) (Jurchescu et al., 2017). 42 43 Aridity: The Mediterranean region shows evidence of large-scale decreasing precipitation trends over 1901- 44 2010, which are at least partly attributable to anthropogenic forcing according to CMIP5 models (Knutson 45 and Zeng (2018). Nevertheless, there is low agreement among studies on observed precipitation trend in the 46 Mediterranean region. (Chapter 11; Atlas 8.2). 47 48 The precipitation is projected to decrease by mid and end of century for the RCP8.5 and SSP5-85 with 49 strong agreement among CMIP5, CMIP6 and CORDEX regional climate ensemble models on the direction 50 of change. With both temperature increase and precipitation decrease there is high confidence on increase 51 aridity in the MED region. (Coppola et al., 2021 Atlas.8.2; Chapter 4; Chapter 8). 52 53 Hydrological drought: There is high confidence that hydrological droughts have increased in the 54 Mediterranean basin with medium confidence in anthropogenic attribution of the signal, and high confidence 55 that they will continue to increase through the 21st century for 2oC GWL and higher and all scenarios. Do Not Cite, Quote or Distribute 12-69 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 (Section 11.9; Chapter 8). There is medium confidence in hydrological drought increase in WCE and low 2 confidence in direction of change for EEU and NEU from mid-century onward and for 2°C GWL and higher 3 and all scenarios (Section 11.9) (see Figure 12.4g-i) 4 5 Streamflow droughts are projected become more severe and persistent in Mediterranean and Western Europe 6 (current 100-year events could occur approximately every 2 to 5 years by 2080) (Forzieri et al., 2016). 7 The opposite tendency is projected in Northern, Eastern and Central Europe where higher precipitation that 8 outweigh the effects of increased evapotranspiration is expected to result in a decrease in streamflow drought 9 frequency (Forzieri et al. 2014). For a 2oC GWL droughts will become more intense in the MED and in 10 France and longer mainly due to less rainfall and higher evapotranspiration. A reduction of drought length 11 and magnitude is projected for NEU and EEU (Roudier et al., 2016). In the Southern Alps, both winter and 12 summer low flows are projected to be more severe, with a 25% decrease in the 2050s (Vidal et al., 2016). 13 14 Agricultural and ecological drought: There is medium confidence that agricultural and ecological droughts 15 have increased in Western and Central Europe and in the Mediterranean region, and medium confidence that 16 anthropogenic drivers contributed to the Mediterranean increase (Section 11.9, Chapter 8). 17 18 Chapter 11 assesses that agricultural and ecological droughts will increase in the Mediterranean regions 19 (high confidence) and Western and Central Europe (medium confidence) by mid-century with high 20 confidence by the end of century for the MED for 2oC GWL and higher and all scenarios. Low confidence in 21 direction of change is assessed for EEU and NEU (see Figure 12.4k). 22 23 Recent local studies provide additional risk-relevant context to changes in European drought. Agricultural 24 and ecological drought conditions are expected to intensify in Southern Europe by end-of-century based on 25 the 12-month rainfall Drought Severity Index (a soil moisture indicator), precipitation deficit SPI and SPEI 26 indices. There will be regions in Southern Europe where this type of drought could be up to 14 times worse 27 than the worst drought in the historical period (Guerreiro et al., 2018). One-in-10 year drought events are 28 projected to happen every second year (Mora et al., 2018; Ruosteenoja et al., 2018). The Mediterranean 29 region will have 100 additional stress years (years with three consecutive months of precipitation deficits 30 greater than 25%) (Giorgi et al., 2018); an increase of both drought frequency (up to 2 events per decade) 31 and severity (Spinoni et al., 2014, 2019a) and an increase of consecutive dry day (CDD) in the southern part 32 of the MED region (Lionello and Scarascia, 2020). In contrast, droughts are expected to decrease in winter in 33 Northern Europe (Spinoni et al., 2018a; Section 11.9). These findings are confirmed by the EURO- 34 CORDEX, CMIP5 and CMIP6 ensemble that show a change of frequency of drought events in the MED 35 between 2 and 3 events per decade for mid of the century RCP8.5 scenario (Coppola et al., 2021a) (see also 36 Figure SM 1.3 ) 37 38 Fire weather: Fire weather conditions have been increasing since about 1980 over a few regions in Europe 39 including Mediterranean areas (low confidence) (Venäläinen et al., 2014; Urbieta et al., 2019; Barbero et al., 40 2020; Giannaros et al., 2021). However beyond a few studies, evidence is largely missing about attribution 41 of these trends to anthropogenic climate change (Forzieri et al., 2016). An increase in fire weather is 42 projected for most of Europe, especially Western, Eastern and Central regions, by 2080 (current 100-yr. 43 events will occur every 5 to 50 years), with a progressive increase in confidence and model agreement along 44 the 21st century (medium confidence) (Forzieri et al., 2016; Abatzoglou et al., 2019). With increased drying 45 and heat combined, in Mediterranean areas, an increase in fire weather indices is projected under RCP4.5 46 and RCP8.5, or SRES A1B, as early as by the mid century (Bedia et al., 2014; Abatzoglou et al., 2019; 47 Dupuy et al., 2020; Fargeon et al., 2020; Ruffault et al., 2020) (high confidence) and an increase in burned 48 area of 40% and 100% for a 2oC and 3oC respectively (Turco et al., 2018). 49 50 In summary, there is high confidence that river floods will increase in Central and Western Europe 51 and medium confidence that they will decrease in Northern, Eastern and Southern Europe, for mid 52 and end of century under RCP8.5 and with low confidence under RCP2.6. There is high confidence 53 that aridity will increase by mid and end of century for the RCP8.5 and SSP5-85, and high confidence 54 that agricultural, ecological and hydrological droughts will increase in the Mediterranean region by 55 mid and far end of century under all RCPs except RCP2.6/SSP1-2.6 and also for 2oC and higher Do Not Cite, Quote or Distribute 12-70 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 GWLs. There is high confidence in fire weather increase in the Mediterranean region. 2 3 4 12.4.5.3 Wind 5 6 Mean wind speed: Mean surface wind speeds have decreased in Europe as in many other areas of the 7 Northern Hemisphere over the past four decades (AR5, WGI) (medium confidence), with a reversal to an 8 increasing trend in the last decade (low confidence) that is however not fully consistent across studies (Tian 9 et al., 2019; Zeng et al., 2019; Zhang et al., 2019d; Deng et al., 2021) (Chapter 2). Reanalyses also show 10 declining winds in Europe (Deng et al., 2021) with large interdecadal variability (Laurila et al., 2020). The 11 declining trend has induced a corresponding decline in wind power potential indices across Europe (low 12 confidence) (Tian et al., 2019). However, there is low agreement and low evidence that climate model 13 historical trends are consistent with observed trends (Tian et al., 2019; Deng et al., 2021) . Several factors 14 have been attributed to these trends, including forest growth, urbanization, local changes in wind 15 measurement exposure and aerosols (Bichet et al., 2012), as well as natural variability (Zeng et al., 2019). 16 17 Due to changes in mean surface wind speed patterns (Li et al., 2018a) and the poleward shift of the North 18 Atlantic jet stream exit, mean surface wind speeds are projected to decrease in the Mediterranean areas under 19 RCP4.5 and RCP8.5 by the middle of the century and beyond, or for global warming levels of 2 degrees and 20 higher (high confidence), with a subsequent decrease in wind power potential (medium confidence) (Hueging 21 et al., 2013; Tobin et al., 2015, 2018; Davy et al., 2018; Karnauskas et al., 2018a; Kjellström et al., 2018; 22 Moemken et al., 2018) (see also Fig. 12.4). However, subregional patterns of change are present from 23 regional climate models such as an increase in wind speeds in the Aegean Sea and in the northern Adriatic 24 Sea where a reduction of Bora events and an increase of Scirocco events are projected for mid-century and 25 beyond under RCP4.5 and RCP8.5 (medium confidence) (Tobin et al., 2016; Davy et al., 2018; Belušić 26 Vozila et al., 2019). Projections (as cited above) also indicate a decrease in mean wind speed in Northern 27 Europe (medium confidence, medium agreement) (Karnauskas et al., 2018a; Tobin et al., 2018; Jung and 28 Schindler, 2019). Daily and interannual wind variability is projected to increase under RCP8.5 (only) 29 Northern Europe (Moemken et al., 2018) (low confidence), which can influence electrical grid management 30 and wind energy production (low confidence). Wind speeds are projected to shift towards more frequent 31 occurrences below thresholds inhibiting wind power production (Weber et al., 2018). Wind stagnation events 32 may become more frequent in future climate scenarios in some areas of Europe in the second half of the 21st 33 century (Horton et al., 2014; Vautard et al., 2018), with potential consequences on air quality (low 34 confidence). 35 36 Severe wind storm: There are large uncertainties in past evolutions of windstorms and extreme winds in 37 Europe. Extreme near-surface winds have been decreasing in the past decades (Smits et al., 2005; Tian et al., 38 2019; Vautard et al., 2019) according to near-surface observations. Significant negative trends of cyclone 39 frequency in spring and positive trends in summer have been found in the Mediterranean basin for the period 40 1979-2008 (Lionello et al., 2016). By contrast increasing trands have been found in Arctic ocean areas 41 (Wickström et al., 2020). These trends are not associated with significant trends in extratropical cyclones 42 (Chapter 2), and are consistent with the mean wind decreasing trends. 43 44 There is medium confidence that serial clustering of storms, inducing cumulated economic losses, in future 45 climate will increase in many areas in Europe under climate projections over Europe (Karremann et al., 46 2014; Economou et al., 2015). Strong winds and extra-tropical storms are projected to have a slightly 47 increasing frequency and amplitude in the future in Northern, Western and central Europe (Outten and Esau, 48 2013; Feser et al., 2015; Forzieri et al., 2016; Mölter et al., 2016; Ruosteenoja et al., 2019b; Vautard et al., 49 2019) in RCP8.5 and SRES A1B by the end of the century (medium confidence), as well as off the European 50 coasts (Martínez-Alvarado et al., 2018) due to the increase of intensity of extratropical storms at a 2°C global 51 warming level or above (Zappa et al., 2013) in these areas. The frequency of storms, including Medicanes, is 52 projected to decrease in Mediterranean regions, and their intensities are projected to increase, by the middle 53 of the century and beyond for SRES A1B, A2 and RCP8.5 (medium confidence) (Nissen et al., 2014; Feser et 54 al., 2015; Forzieri et al., 2016; Mölter et al., 2016; Tous et al., 2016; Romera et al., 2017; González‐Alemán 55 et al., 2019) (MedECC, 2020)(Chapter 11). Do Not Cite, Quote or Distribute 12-71 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Projections of smaller-scale hazard phenomena such as tornadoes, wind gusts, hail storms and lightning are 2 currently not directly available partly due to the inability of climate models to simulate such phenomena. 3 Observational networks for such phenomena usually lack homogeneity over long climate periods, hindering 4 clear trends to be detected. For instance, while no robust trends have been identified (Hermida et al., 2015; 5 Mohr et al., 2015a; Burcea et al., 2016; Ćurić and Janc, 2016), hail storm environments (favourable 6 atmospheric configurations) have increased in frequency (Sanchez et al., 2017) (low confidence, low 7 evidence). In future climate periods it is more likely than not that severe convection environments will 8 become more frequent by the end of the century under scenario RCP8.5 (Mohr et al., 2015b; Púčik et al., 9 2017), and there is medium confidence that such environments become more frequent by the 2050s in 10 scenario RCP4.5. There is no evidence for changes in tornado frequencies in Europe in the observations 11 (Groenemeijer and Kühne, 2014) as well as in future climate projections. Insufficient observational record 12 length for lightning numbers does not allow an assessment of trends. 13 14 There is high confidence that mean wind speeds will decrease in Mediterranean areas and medium 15 confidence of such decreases in Northern Europe for global warming levels of 2°C or more and beyond 16 the middle of the century. A slightly increased frequency and amplitude of extratropical cyclones, 17 strong winds and extra-tropical storms is projected for Northern, Central and Western Europe by the 18 middle of the century and beyond and for global warming levels of 2°C or more (medium confidence). 19 The frequency of Medicanes is projected to decrease (medium confidence), but their intensity is 20 projected to increase. Proxies of intense convection indicate that the large-scale conditions conducive 21 to severe convection will tend to increase in the future climate (low confidence). 22 23 24 12.4.5.4 Snow and ice 25 26 Snow: Widespread and accelerated declines of snow depth are currently observed in Europe (Fontrodona 27 Bach et al., 2018) and snow water equivalent (Marty et al., 2017b) (see also Figures 12.9b) for a typical 28 range of altitudes of higher settlement elevation and ski resorts. In the Pyrenees a slow snow cover decline 29 has been observed starting from the industrial period with a sharp increase since 1955 (López-Moreno et al., 30 2020). Under the RCP2.6, RCP4.5 and RCP8.5 scenario the reliability elevation using snowmaking will rise 31 of 200-300 m in the Alps and 400-600 in the Pyrenees by mid century. By end of century projections are 32 highly dependent from the scenario being stationary for the RCP2.6 and continuously decreasing under the 33 RCP8.5 up to not have anymore natural snow conditions for any of the locations in the French Alps and 34 Pyrenees (Spandre et al., 2019). Similarly also Norway and Austria will see a rising of the natural snow 35 elevation with consequences for the ski season (Scott et al., 2020; Steiger and Scott, 2020). In the Alps, 36 recent simulations project a reduction in Snow Water Equivalent (SWE) at 1500 m a.s.l. of 80–90% by 2100 37 under the A1B scenario and a snow season that would start 2-4 weeks later and end 5-10 weeks earlier than 38 in the 1992-2012 average (Schmucki et al., 2015), which is equivalent to a shift in elevation of about 700 m 39 (Marty et al., 2017a). For elevations above 3000 m a.s.l., a decline in SWE of at least 10% is expected by the 40 end of the century even when assuming the largest projected precipitation increase. Similar trends are 41 observed for the Pyrenees and Scandinavia (López-Moreno et al., 2009; Räisänen and Eklund, 2012). For the 42 Northern French Alps above 1500 m and the Ötztal locations in the Austrian alps SWE has a similar 43 decreasing trend altitudinal dependent for all 3 scenarios (RCP2.6, RPC4.5 and RCP8.5) until mid of century 44 and with significant differentiation among them in the second half of the century up to free snow condition 45 for the RCP8.5 scenario (Hanzer et al., 2018; Verfaillie et al., 2018). 46 47 Glacier: Observation and future projections of European glacier mass changes are assessed in Section 9.5.1 48 grouped in two main regions: Scandinavia and Central Europe regions. It is virtually certain that glaciers 49 will shrink in the future and there is medium confidence in the timing and mass change rates (Chapter 9). 50 Central Europe is one of the regions where glaciers are projected to lose substantial mass even under low- 51 emission scenarios (Section 9.5.1.3) (MedECC, 2020). GlacierMIP projections indicate that glaciers in the 52 Central Europe region lose 63 ± 31%, 80 ± 22% and 93 ± 13% of their 2015 mass by the end of the century 53 for RCP2.6, RCP4.5 and RCP8.5, respectively (Marzeion et al., 2020). In those same scenarios, glaciers in 54 Scandinavia are projected to lose 55 ± 33%, 66 ± 34% and 82 ± 24% of their 2015 mass. The virtually 55 certain shrink in glaciers is bolstered by RCM simulations from the EURO-CORDEX ensemble, with the Do Not Cite, Quote or Distribute 12-72 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Global Glacier Evolution Model (GloGEM) indicating a substantial reduction of glacier ice volumes in the 2 European Alps by 2050 (47–52% with respect to 2017 for RCP2.6, 4.5 and 8.5). Under RCP2.6, about two- 3 thirds (63 ± 11%) of the present-day (2017) ice volume is projected to be lost by 2100. In contrast, under the 4 strong warming of RCP8.5, glaciers in the European Alps are projected to largely disappear by 2100 (94 ± 5 4% volume loss compared to 2017; Zekollari et al., 2019). 6 7 Permafrost: In Europe, permafrost is found in high mountains and in Scandinavia, as well as in Arctic 8 Islands (e.g., Iceland, Novaya Zemlia or Svalbard). In recent decade permafrost has been lost (Chapter 9) 9 and accelerated warming at high altitudes and latitudes has favored an increase of permafrost temperatures in 10 the order of 0.2 ± 0.1°C between 2007 and 2016 (Romanovsky et al., 2018; Noetzli et al., 2019). Over the 11 21st century, permafrost is very likely to undergo increasing thaw and degradation (Hock et al., 2019) and it is 12 virtually certain that permafrost extent and volume with decrease with increase of global warming (Chapter 13 9). 14 15 Permafrost thawing is projected to affect the frequency and magnitude of high-mountain mass wasting 16 processes (Stoffel and Huggel, 2012). The temporal frequency of periglacial debris flows in the Alps is 17 unlikely to change significantly by the mid-21st century but is likely to decrease during the second part of the 18 century, especially in summer (Stoffel et al., 2011, 2014; Lane et al., 2017). There is medium confidence that 19 most of the Northern Europe periglacial processes will disappear by the end of the century, even in the 20 RCP2.6 scenario (Aalto et al., 2017). The magnitude of debris flow events might increase (Lugon and 21 Stoffel, 2010) (low confidence) and the debris-flow season may last longer in a warming climate (Stoffel and 22 Corona, 2018) (medium confidence). Quantitative data for the European Alps is highly site dependent 23 (Haeberli, 2013). 24 25 Heavy Snowfall, ice storms and hail: There is low confidence that climate change will affect ice and snow 26 episodic hazards (limited evidence). The change in snowpack in the Alps is expected to lead to a possible 27 reduction in overall avalanche activity (low confidence), except possibly in winter and at high altitudes 28 (Castebrunet et al., 2014). 29 30 For ice storms, or freezing rainstorms, there is also limited evidence due to a limited number of studies. 31 Heavy snowfalls have decreased in frequency in the past decades and this is expected to continue in the 32 future climate (Beniston et al., 2018) (low confidence). Freezing rain is projected to increase in western, 33 central and southern Europe (Kämäräinen et al., 2018) (low confidence). Rain-on-snow events, are 34 decreasing in northern regions (Pall et al., 2019) and by 48% on average in southern Scandinavia (Poschlod 35 et al., 2020) due to decreases in snowfall. 36 37 In summary, in the future snow cover extent and seasonal duration will reduce (high confidence) and 38 it is virtually certain that glaciers will continue to shrink. A reduction of glacier ice volume is projected 39 in the European Alps and Scandinavia (high confidence). There is high confidence that permafrost will 40 undergo increasing thaw and degradation over the 21st century. Most of the Northern Europe 41 periglacial will disappear by the end of the century even for a lower emissions scenario (medium 42 confidence) and the debris-flow season may last longer in a warming climate (medium confidence). 43 44 45 12.4.5.5 Coastal and Oceanic 46 47 Relative sea level: Around Europe, over 1900-2018, a new tide-gauge based reconstruction finds a regional- 48 mean RSL change of 1.08 [0.79 to 1.38] mm yr-1 in the subpolar North Atlantic (Frederikse et al., 2020), 49 compared to a GMSL change of around 1.7 mm yr-1 (Section 2.3.3.3; Table 9.5). For the period 1993-2018, 50 the RSLR rates, based on satellite altimetry, increased to 2.17 [1.66 to 2.66] mm yr-1 (Frederikse et al., 51 2020), compared to a GMSL change of 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). 52 53 Relative sea level rise is extremely likely to continue in the oceans around Europe. Regional-mean RSLR 54 projections for the oceans around Europe range from 0.4 m–0.5 m under SSP1-RCP2.6 to 0.7 m–0.8 m under 55 SSP5-RCP8.5 for 2081-2100 relative to 1995-2014 (median values), which means that there are locally large Do Not Cite, Quote or Distribute 12-73 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 deviations from the projected GMSL change {Section 9.6.3.3}. These RSLR projections may however be 2 underestimated due to potential partial representation of land subsidence in their assessment (Section 3 9.6.3.2). The signal is strongest for the North Sea and Atlantic coasts, followed by the Black Sea. The Baltic 4 Sea on the contrary shows the lowest increase due to land uplift (Vousdoukas et al., 2017). The model 5 agreement is higher for the Mediterranean and in line with the previous findings by Gualdi et al. (2013). 6 7 Coastal flood: The present day 1:100 yr ETWL is between 0.5-1.5m in the MED basin and 2.5-5.0 m in the 8 western Atlantic European coasts, around the UK and along the North Sea coast, and lower at 1.5-2.5m 9 along the Baltic Sea coast. (Kirezci et al 2020). Similar values are reported by Vousdoukas et al., (2018). 10 11 There is high confidence that extreme total water level (ETWL) magnitude and occurrence frequency will 12 increase throughout Europe (see Figure 12.4p-r), except in the northern Baltic Sea. Across the region, the 5th 13 – 95th percentile range of the 1:100 yr ETWL is projected to increase (relative to 1980 – 2014) by 4 cm – 40 14 cm and by 6 cm – 47 cm by 2050 under RCP4.5 and RCP8.5, respectively. By 2100, this range is projected 15 to be 6 cm – 88 cm and 25 cm – 186 cm under RCP4.5 and RCP8.5, respectively (Vousdoukas et al., 2018; 16 Kirezci et al., 2020) (see Figure SM 12.6). Mass addition across the Gibraltar Strait may play a role even if 17 this is quite unclear to which extent (Lionello et al., 2017). Furthermore, under RCP4.5, the present day 18 1:100 yr ETWL is projected to have median return periods of between 1:5 and 1:20 yrs by 2050 and occur 19 once per year or more by 2100 in the Mediterranean and Black Sea, while in the rest of Europe it is mostly 20 projected to have median return periods of between 1:20 yrs and 1:50 yrs by 2050 and between 1:5 yrs and 21 1:20 yrs by 2100 (Vousdoukas et al., 2018). Under RCP8.5, occurrence of the present day 1:100 yr ETWL is 22 projected to increase further to median return periods of 1:1 yr to 1:5 yrs by 2050 and occur more than once 23 per year by 2100 in the Mediterranean and Black Sea, while in the rest of Europe it is mostly projected to 24 have median return periods between 1:5 yrs and more than once per year by 2100. 25 26 Coastal erosion: Satellite derived shoreline change estimates over 1984 – 2015 indicate shoreline retreat 27 rates of around 0.5 m yr-1 along the sandy coasts of CEU and MED, around 4 m yr-1 in EEU (Caspian Sea 28 region) and more or less stable shorelines in NEU (Luijendijk et al., 2018; Mentaschi et al., 2018). Mentaschi 29 et al. (2018) report a coastal area loss of 270 km2 over a 30 year period (1984-2015) along the Atlantic 30 coastlines of Europe. 31 32 Projections indicate that sandy coasts throughout the continent (except those bordering the northern Baltic 33 Sea) will experience shoreline retreat through the 21st century (high confidence). Median shoreline change 34 projections (CMIP5) relative to 2010, show that, by mid-century, shorelines will retreat by between 25 m 35 and 60 m along sandy coasts in CEU and MED under both RCP4.5 and RCP8.5 (Vousdoukas et al., 2020; 36 Athanasiou et al., 2020). Mid-century median projections for NEU indicate virtually no shoreline retreat 37 under RCP4.5, but a retreat of around 40 m under RCP8.5. By 2100, median shoreline retreats of around 50 38 m are projected in NEU and MED under RCP4.5, increasing to around 80 m under RCP8.5. End century 39 median projections for CEU are far higher at 100 m (RCP4.5) and 160 m (RCP8.5). The total length of sandy 40 coasts in Europe that is projected to retreat by more than a median of 100 m by 2100 under RCP4.5 and 41 RCP8.5 is about 12,000 km and 18,000 km respectively, an increase of approximately 54% (Vousdoukas et 42 al., 2020) 43 44 Local assessments of both long term shoreline retreat and episodic coastal erosion are given by (Li et al., 45 2013a; Toimil et al., 2017; Bon de Sousa et al., 2018; Le Cozannet et al., 2019). In terms of episodic coastal 46 erosion, 31%–88% of all Aegean beaches are projected to experience complete erosion, with a RCP4.5 sea 47 level rise of 0.5 m and a surge of 0.6 m, but with substantial uncertainty (Monioudi et al., 2017). 48 49 Marine heatwave: The mean SST of the Atlantic Ocean and the Mediterranean has increased between 50 0.25°C and 1°C since 1982-1998. This mean ocean surface warming is correlated to longer and more 51 frequent marine heatwaves in the region (Oliver et al., 2018). Over the period 1982-2016, the coastlines of 52 Europe experienced on average more than 2.0 MHW per year, with the eastern Mediterranean and 53 Scandinavia experiencing 2.5–3 MHWs per year. The average duration was between 10 and 15 days. 54 Changes over the 20th century, derived from MHW proxies, show an increase in frequency of between 1.0 55 and 2.0 MHWs per decade in Europe, although the trend is not statistically significant; with an increase in Do Not Cite, Quote or Distribute 12-74 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 intensity per event in the North Atlantic and the Mediterranean, and a decrease in the Atlantic off the British 2 Isles. The total number of MHW days per decade has increased in the Mediterranean (Oliver et al., 2018). 3 4 Mean SST is projected to increase by 1 ºC to 3ºC around Europe by 2100, with a hotspot of around 4 ºC to 5 5 ºC along the Arctic coastline of Europe under RCP4.5 and RCP8.5 scenarios (see Interactive Atlas), leading 6 to a continued increase in MHW frequency, magnitude and duration (Oliver et al., 2018; MedECC, 2020). 7 Projections for SSP1-2.6 and SSP5-8.5 both show an increase in MHWs around Europe by 2081 – 2100, 8 relative to 1985 – 2014 (Box 9.2, Figure 1). Darmaraki et al. (2019) project that, by the end of the 21st 9 century and under RCP8.5, there will be one MHW occurring every year in the Northern Mediterranean sea, 10 and that these MHWs would be 3 months longer, 4 times more intense, and 42 times more severe than 11 present day MHWs in the region. Frölicher et al., (2018) show that, in Europe, the change in the probability 12 for the number of days of MHWs exceeding the 99th percentile of the pre-industrial level is 4%, 15% and 13 30% for global warming levels of 1oC, 2oC and 3.5oC, respectively. MHW increase in the Mediterranean will 14 impact on many species that live in shallow waters and have reduced motility, with consequences for the 15 related economic activities (Galli et al., 2017). 16 17 In general, there is high confidence that most coastal/ocean related climatic impact-drivers in Europe 18 will increase over the 21st century for all scenarios and time horizons. Relative sea level rise is 19 extremely likely to continue around Europe (except in the northern Baltic Sea), contributing to 20 increased coastal flooding in low-lying areas and shoreline retreat along most sandy coasts (high 21 confidence). Marine Heatwaves are also expected to increase around the region over the 21st century 22 (high confidence). 23 24 25 12.4.5.6 Other 26 27 Compound events. One typical compound event that is observed in the European area is compound flooding 28 due to the combination of extreme sea level events and extreme precipitation events associated with high 29 level of runoff. In the present climate, the Mediterranean coasts are exposed to a higher probability of this 30 type of compound flooding events (Bevacqua et al., 2019). Under RCP8.5, the probability of these events are 31 projected to increase along northern European coasts (west coast of UK, northern France, the east and south 32 coast of the North Sea, and the eastern half of the Black Sea) with the percentage of coastline currently 33 experiencing such events with a return period lower than 6 years will increasing from 3% to 11% by the end 34 of the 21st century (Bevacqua et al., 2019). 35 36 Under RCP8.5, regions in Russia, France and Germany are projected to experience an increase in the 37 frequency and the length of wet and cold compound events, while Spain and Bulgaria are projected to stay 38 longer in the hot and dry state by mid-century (Sedlmeier et al., 2016). 39 40 Compound events of dry and hot summers have increased in Europe. (Manning et al., 2019) found that the 41 probability of such compound events has increased across much of Europe between 1950–1979 and 1984– 42 2013, notably in southern, eastern and western Europe. Compound hot and dry extremes are projected to 43 increase in Europe by mid-century for the SRES A1B and RCP8.5 but a particularly strong signal is 44 projected in southern and eastern Germany and the Czech Republic (Sedlmeier et al., 2016). 45 46 47 The assessed direction of change in climatic impact-drivers for Europe and associated confidence levels are 48 illustrated in Table 12.7, together with emergence time information (see Section 12.5.2). No assessable 49 literature could be found for Sand and dust storms, although these phenomena may be relevant in parts of the 50 region. 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-75 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 [START TABLE 12.7 HERE] 2 3 Table 12.7: Summary of confidence in direction of projected change in climatic impact-drivers in Europe, representing 4 their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, SRES A1B, or 5 above within each AR6 region (defined in Chapter 1), approximately corresponding (for CIDs that are 6 independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for more details 7 of the assessment method). The table also includes the assessment of observed or projected time-of- 8 emergence of the CID change signal from the natural inter-annual variability if found with at least medium 9 confidence in Section 12.5.2. 10 11 Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Mediterranean (MED) 5 6 7 2 Western and Central Europe (WCE) 4 2 Eastern Europe (EEU) Northern Europe (NEU) 1 8 2,3 1. Excluding southern UK. 2. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. 3. The Baltic sea shoreline is projected prograde if present-day ambient shoreline change rates continue. Key 4. For the Alps landslide risk is likely to increase. High confidence of decrease 5. Low confidence of decrease in the southernmost part of the region. Medium confidence of decrease 6. General decrease except in Aegean Sea exhibiting increase. 7. Medium confidence of decrease in frequency and increase in intensities. Low confidence in direction of change 8. Except in the Northern Baltic Sea region. Medium confidence of increase High confidence of increase Not broadly relevant 12 13 [END TABLE 12.7 HERE] 14 15 16 12.4.6 North America 17 18 Major changes in North American CIDs were assessed in WGII AR5 Chapter 26 (Romero-Lankao et al., 19 2014), with additional detail on connections to warming levels provided by SR1.5 (IPCC, 2018), and climate 20 information related to land degradation and land use suitability in SRCCL (IPCC, 2019c), and ocean and 21 coastal hazards in the SROCC (IPCC, 2019b). Recent national assessments in the United States (USGCRP, 22 2017, 2018) and Canada (Bush and Lemmen, 2019) enhance the local perspective and assessments across a 23 number of CIDs and their sectoral connections. For the purpose of this assessment, North America is sub- 24 divided into six sub-regions as defined in Chapter 1: Central North America (CNA), Western North America 25 (WNA), Central North America (CAN), Eastern North America (ENA), Northeast North America (NEN), 26 and Northwest North America (NWN). Greenland and Arctic regions of Northeast and Northwest North 27 America are further assessed in 12.4.9, and the Caribbean and Hawaiian Islands are assessed in 12.4.7. 28 29 30 [START FIGURE 12.10 HERE] 31 32 Figure 12.10: Projected changes in selected climatic impact-driver indices for North America. (a) Mean change in 33 1-in-100 year river discharge per unit catchment area (Q100, m 3 s-1 km-2), and (b) median change in the Do Not Cite, Quote or Distribute 12-76 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 number of days with snow water equivalent (SWE) over 100 mm (from November to March), from 2 CORDEX-North America models for 2041-2060 relative to 1995-2014 and RCP8.5. Diagonal lines 3 indicate where less than 80% of models agree on the sign of change. (c) Bar plots for Q100 (m3 s-1 km-2) 4 averaged over land areas for the WGI reference AR6 regions (defined in Chapter 1). The left column 5 within each panel (associated with the left y-axis) shows the ‘recent past’ (1995-2014) Q100 absolute 6 values in grey shades. The other columns (associated with the right y-axis) show the Q100 changes 7 relative to the recent past values for two time periods (‘mid’ 2041-2060 and ‘long’ 2081-2100) and for 8 three global warming levels (defined relative to the preindustrial period 1850-1900): 1.5°C (purple), 2°C 9 (yellow) and 4°C (brown). The bars show the median (dots) and the 10th-90th percentile range of model 10 ensemble values across each model ensemble. CMIP6 is shown by the darkest colours, CMIP5 by 11 medium, and CORDEX by light. SSP5-8.5/RCP8.5 is shown in red and SSP1-2.6/RCP2.6 in blue. (d) As 12 for (c) but showing absolute values for number of days with SWE > 100mm, masked to grid cells with at 13 least 14 such days in the recent-past. See Technical Annex VI for details of indices. A Caribbean Q100 14 bar plot is included here but assessed in the Small islands section (12.4.7). Further details on data sources 15 and processing are available in the chapter data table (Table 12.SM.1). 16 17 [END FIGURE 12.10 HERE] 18 19 20 12.4.6.1 Heat and cold 21 22 Mean air temperature: Section Atlas.9.2 assessed very likely mean warming in observations across North 23 America with highest increases at higher latitudes and in the winter season. Section Atlas.9.4 assessed very 24 likely mean warming in future decades in all North American regions, with CMIP and CORDEX models 25 showing median increases exceeding 2℃ in much of the continental interior under RCP8.5 (2041-2060 26 compared to 1995-2014) and higher increases toward the north. Mean temperatures at the end of century 27 show strong scenario dependence, rising between 1°C and 2.5℃ in RCP2.6 and about 4 to 8℃ in RCP8.5 28 (Figures Atlas.15, Atlas.32, Atlas.33). Warming also raises stream temperatures across the continent (DOE, 29 2015; Trtanj et al., 2016; van Vliet et al., 2016; Chapra et al., 2017), and Hill et al. (2014) projected US 30 stream warming by 0.6°C (±0.3°C) per 1°C increase in local air temperature. Mean warming drives shifts in 31 the seasonal timing of temperature thresholds, including increasing growing degree-days (Mu et al., 2017), 32 longer growing seasons (Gowda et al., 2018; Li et al., 2018b; Vincent et al., 2018c), reduced chill hours 33 (Luedeling, 2012; Lee and Sumner, 2015; Xie et al., 2015; Parker and Abatzoglou, 2019), and longer pollen 34 and allergy seasons (Fann et al., 2016; Anenberg et al., 2017; Sapkota et al., 2019). Warmer temperatures 35 will very likely reduce heating degree days and increase cooling degree days (Bartos et al., 2016; US EPA, 36 2016; Craig et al., 2018; Zhang et al., 2019c; Coppola et al., 2021b). 37 38 Extreme heat: Section 11.9 assessed that extreme temperatures in North America have increased in recent 39 decades (medium evidence, medium agreement) other than Central and Eastern North America (low 40 confidence), and extreme heat in all regions is projected to increase with climate change (high confidence). 41 Observed trends in extreme heat are more positive for heat extreme indices that include temperature and 42 humidity given historical expansion of irrigation and intensification of agriculture (Mueller et al., 2017; 43 Grotjahn and Huynh, 2018; Thiery et al., 2020). Several studies noted statistically significant increases in 44 intensity and particularly the frequency, duration, and seasonal length of the physiologically-hazardous 45 extreme heat conditions across North American (Grineski et al., 2015; Habeeb et al., 2015; Martínez-Austria 46 et al., 2016; Petitti et al., 2016; Vincent et al., 2018c; García-Cueto et al., 2019). 47 48 Figure 12.4b shows over a month of additional days at CMIP6 SSP5-8.5 mid-century where temperatures 49 exceed 35°C across much of southern Mexico and regions near the US-Mexico border and these extreme 50 temperatures occur at least once per year up to southern Canada. (Coppola et al., 2021b) found similar 51 patterns in CMIP5 and CORDEX-Core. Using locally-tailored heat thresholds, Maxwell et al. (2018) found 52 that “very hot” days in 5 US cities will occur a median of 3 to 5 times more often by RCP8.5 2036-2065 (2 53 to 3.5x in RCP4.5), Oleson et al. (2018) projected that annual heatwave duration will exceed one month in 54 Houston in RCP8.5 2061-2080, and Anderson et al. (2018) projected 7 to 12 times more exceedances of 55 thresholds associated with high-mortality in RCP8.5 2061-2080 (6 to 7x in RCP4.5). (Schwingshackl et al., 56 2021) found that Central and Eastern North America are among the regions with the strongest trend in heat Do Not Cite, Quote or Distribute 12-77 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 stress indicators. Studies also project increasingly surpassed heat extreme thresholds for North American 2 crops (Gourdji et al., 2013), airplane weight restrictions (Coffel et al., 2017), and peak load energy systems 3 (Auffhammer et al., 2017). 4 5 The number of days crossing dangerous heat thresholds such as HI>41°C will be very sensitive to the 6 mitigation scenario at the end of century (Wuebbles et al., 2014; Zhao et al., 2015; Dahl et al., 2019; 7 Schwingshackl et al., 2021). At the end of the century under SSP5-8.5, a CMIP6 median increase of 8 exceedances of 75-150 days per year is projected over much of North Central America, Central North 9 America and the U.S. Southwest while this increase is projected to remain limited below 60 days under 10 SSP1-2.6 (Figure 12.4d,f and Figure SM 1.2). Steinberg et al. (2018) also projected more frequent and longer 11 ‘heat-health’ events in California extending into October. 12 13 Cold spell: Chapter 11 assessed high confidence in decreasing frequency and intensity of cold spells over 14 North America (Section 11.9). The number of days with extreme wind chill hours (humidex<-30) decreased 15 at 76% of examined Canadian stations from 1953-2012 (Mekis et al., 2015) and cold days and coldest nights 16 decreased in Mexico from 1980-2010 (García-Cueto et al., 2019). 17 18 Cold spells are projected to decrease over North America under climate change, with the largest decreases 19 most common in the winter season (high confidence) (Section 11.9). Minimum winter temperatures are 20 projected to rise faster than the mean wintertime temperature (Underwood et al., 2017) and alter cold 21 hardiness zones used to determine agricultural suitability (Parker and Abatzoglou, 2016). Wuebbles et al. 22 (2014) projections for RCP8.5 end-of-century found that the 4-day cold spell that happens on average once 23 every 5 years is projected to warm by more than 10 ºC and CMIP5 models do not project current 1-in-20 24 year annual minimum temperature extremes to recur over much of the continent. Multiple studies have 25 shown that Arctic warming can alter large-scale variability and change the frequency and duration of mid- 26 latitude cold air outbreaks, potentially leading to increasing cold hazards in some regions (Barcikowska et 27 al., 2019; Cohen et al., 2019; Zhou et al., 2021) (low agreement). 28 29 Frost: An expansion of the frost-free season is underway and projections for North America indicate a 30 continuation of this trend in the future (high confidence). Significant decreases in frost days, consecutive 31 frost days, and ice days were identified in 1948-2016 station observations across Canada, along with a 32 resulting lengthening of the frost-free season by more than a month in many regions (Vincent et al., 2018c). 33 Frost days also declined in nearly all Mexican cities from 1980-2010 (García-Cueto et al., 2019), and a 1917- 34 2016 decline of about 3 weeks in frigid winter conditions challenges ecosystems in the Northeast US and 35 Southeast Canada (Contosta et al., 2020). Studies connect projections of a longer frost-free season in North 36 America to a longer outdoor construction season, orchard production, and weight restrictions on runways 37 (Daniel et al., 2018; DeGaetano, 2018; Jacobs et al., 2018). Frosts are projected to persist as an episodic 38 hazard in many regions given natural variability and cold air outbreaks even as mean temperature rises (high 39 confidence). 40 41 Climate change is virtually certain to shift the balance of temperature toward warming trends and 42 away from cold extremes, with increases in the magnitude, frequency, duration and seasonal and 43 spatial extent of heat extremes driving impacts across North America. There is a particular sensitivity 44 to scenario pathway in resulting changes to the frequency of exceeding dangerous heat thresholds such 45 as HI>41°C in NCA, CNA and the US Southwest, with 75-150 days more under SSP5-8.5 but less than 46 60 days more under SSP1-2.6 in the end of the century. 47 48 49 12.4.6.2 Wet and dry 50 51 Mean precipitation: Atlas.9.2 found that trends in annual precipitation over 1960–2015 are generally non- 52 significant though there are consistent positive trends over parts of ENA and CNA, together with significant 53 decreases in precipitation in parts of south-western US and north-western Mexico. 54 55 Atlas.9.4 assessed very high confidence in increases in precipitation over most of Northern and Eastern Do Not Cite, Quote or Distribute 12-78 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 North America, with medium confidence of decrease over Northern Central America and low confidence 2 elsewhere (see Figure Atlas.33, Cross-Chapter Box Atlas.1 Figure 1). Changes are most dramatic in the 3 Spring and Winter, when wet conditions are projected to extend from the northern portions of the continent 4 as far south as the central Great Plains while Mexico becomes drier; in contrast, summertime changes are 5 uncertain across most of the continent other than wetter conditions in Northern Canada (Easterling et al., 6 2017; Bukovsky and Mearns, 2020; Teichmann et al., 2020; Almazroui et al., 2021) . 7 8 River flood: There is limited evidence and low agreement on observed climate change influences for river 9 floods in North America (see also Section 11.5). Trends in streamflow indices are mixed and difficult to 10 separate from river engineering influences, with large changes but little spatial coherence across the US, 11 making it difficult to identify trends with confidence (Villarini and Slater; Peterson et al., 2013; Mallakpour 12 and Villarini, 2015; Archfield et al., 2016; Wehner et al., 2017; Hodgkins et al., 2019; Neri et al., 2019). 13 There is high confidence in historical shifts in the timing of peak streamflow toward higher winter and earlier 14 spring flows in snowmelt-driven basins in Canada (Burn and Whitfield, 2016; Bonsal et al., 2019) and the 15 United States (Dudley et al., 2017; Wehner et al., 2017). Some rivers show ice-jam floods occurring a week 16 earlier, but changes are mixed given localized positive and negative changes across the continent (Rokaya et 17 al., 2018). There is medium confidence that climate change will increase river floods over the United States 18 and Canada but low confidence for changes in Mexico. Wobus et al. (2017a) used a regional hydrologic 19 model for 57000 streams to project more than a doubling in the frequency of current 100 yr flow events in 20 many portions of the United States for RCP8.5 2050 with additional contributions from earlier snowmelt. 21 CMIP6 projections for SSP5-8.5 2065-2099 show strongest peak United States runoff increases in the East 22 (Villarini and Zhang, 2020); however, several studies applying global hydrological models disagree with 23 regional streamflow projections, indicating a decrease in the magnitude or frequency of floods over a large 24 portion of North America (e.g., Hirabayashi et al., 2013; Arnell and Gosling, 2016) (Figure 12.10a,c). 25 26 Heavy precipitation and pluvial flood: Section 11.4 assessed high confidence in observed increases in 27 extreme precipitation events (including hourly totals) in Central and Eastern North America with low 28 confidence in broad trends elsewhere in the continent despite observational increases in some portions of 29 each region (Vincent et al., 2018c; García-Cueto et al., 2019; Zhang et al., 2019c). 30 31 Section 11.4 found that high precipitation is projected to increase across North America (high confidence) 32 except for portions of Western North America where projections are mixed (medium confidence of increase). 33 Maxwell et al. (2018) identified regional “heavy precipitation day” thresholds for 5 cities across the US and 34 projected that a tripling (or more) of these events is possible by RCP8.5 mid-century. Projections indicate 35 changes to intensity-duration-frequency (IDF) curves typically used for construction design and automobile 36 hazards, as well as increases in the 10 year recurrence level of 24 hour rainfall intensities that challenge 37 storm water drainage systems (Hambly et al., 2013; Cheng and AghaKouchak, 2015; Neumann et al., 2015b; 38 Prein et al., 2017b; Hettiarachchi et al., 2018; Ragno et al., 2018). Precise levels of regional IDF 39 characteristics may still depend substantially on the method and resolution of downscaling applied 40 (DeGaetano and Castellano, 2017; Cook et al., 2020b). 41 42 Landslide: There is growing but yet limited evidence for unique climate-driven changes in landslide and 43 rockfall hazards in North America, even as theory suggests decreases in slope and rockface stability due to 44 more intense rainfall, rain-on-snow events, mean warming, permafrost thaw, glacier retreat, and coastal 45 erosion (Cloutier et al., 2017; Coe et al., 2018; Handwerger et al., 2019; Hock et al., 2019; Patton et al., 46 2019) although dry trends can decelerate mass movements (Bennett et al., 2016). Landslide frequency has 47 likely increased in British Columbia (Geertsema et al., 2006) and is expected to increase in Northwest North 48 America given the combination of these factors (medium confidence) (Gariano and Guzzetti, 2016). Cloutier 49 et al. (2017) projected an increase in landslides in Western Canada due to wetter overall conditions and 50 reduced return period for extreme rainfall. Robinson et al. (2017) used scenarios based upon projection of 50 51 yr recurrence of 7 d precipitation periods to highlight the potential for increased landslide hazards near 52 Seattle. Broad US projections are more uncertain given increases in evapotranspiration that will counteract 53 precipitation changes over much of the country (Coe, 2016). 54 55 Aridity: Chapter 8 showed that aridity in North America generally moves opposite to mean precipitation Do Not Cite, Quote or Distribute 12-79 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 change with an added evaporative demand from warmer temperatures (high confidence in aridity increase for 2 North Central America, medium confidence increase in Central North America; high confidence decrease in 3 Northeast North America, medium confidence decrease in Eastern and Northwest North America; see also 4 Section 11.9). Projected soil moisture declines (Figure 12.4j-l) are most widespread across North America 5 during the summertime, with largest declines in Mexico and the southern Great Plains but also extending 6 into Canada (Section 8.4.1.6; Swain and Hayhoe, 2015; Easterling et al., 2017; Bonsal et al., 2019; Lu et al., 7 2019). Yoon et al. (2018) found net reduction in southern Great Plains groundwater storage in RCP8.5 mid- 8 century projections despite increases in mean precipitation and both wet and dry extremes. Soil moisture 9 drying could reach unprecedented levels by the CMIP6 RCP8.5 end of century, even when evaluating deeper 10 soil columns relevant for crop rooting depth (Cook et al., 2020a). Projected changes in the aridity index 11 portray a shift the geographic range of temperate drylands northward and eastward in Central and Western 12 North America (Schlaepfer et al., 2017; Seager et al., 2018) which also diminishes aquifer recharge rates in 13 the southern Great Plains and in some Western regions where snowpack is reduced (Meixner et al., 2016). 14 15 Hydrological drought: Section 11.9 asssessed low confidence of significant observational trends and 16 projected future changes in the characteristics of episodic hydrological drought in North America given 17 limited evidence and low agreement in modeled changes. Zhao et al. (2020) found that increases in 18 hydrological drought frequency (particularly the 100 yr drought) were far more prevalent than for 19 meteorological drought across 5797 watersheds in the US and Canada, indicating a strong influence of 20 evaporative demand. Reductions in the overall supply of meltwater from a declining snowpack also increase 21 the potential for intermittent hydrological droughts in the Western US (Mote et al., 2018; Livneh and 22 Badger, 2020). 23 24 Agricultural and ecological drought: Section 11.9 assessed medium confidence for an increase in 25 agricultural and ecological drought in Western North America but otherwise found limited evidence for 26 broadly observed changes in North American agricultural and ecological drought even as increasing 27 evaporative demand intensified vegetation stress and soil moisture deficits in recent events (Sections 11.6, 28 11.9). Section 11.9 asssessed medium confidence for more intense agricultural and ecological drought 29 conditions over North Central America, Western North America and Central North America in a 2℃ global 30 warming level (about mid-century), with medium confidence extending to Eastern North America and high 31 confidence for North Central America and Central North America under a 4℃ global warming level 32 associated with higher emissions scenarios past 2050. Figure 12.4g-i shows that the frequency of 33 meteorological droughts (which often initiate hydrological, agricultural and ecological drought) is largely 34 projected to increase in North American areas where total precipitation decreases (and vice versa; see 35 Section 11.9 and Coppola et al., 2021b), and higher evaporative demand will extend the regions where more 36 intense ecological and agricultural droughts develop when meteorological droughts occur (Wehner et al., 37 2017; Cook et al., 2019). Studies utilizing a variety of drought indices and soil moisture projections 38 consistently project increased drought extending from Mexico into the southern Canadian Plains during the 39 summer (Swain and Hayhoe, 2015; Ahmadalipour et al., 2017; Feng et al., 2017; Bonsal et al., 2019; Tam et 40 al., 2019; Cook et al., 2020a). 41 42 Fire weather: Climatic conditions conducive to wildfire have increased in Mexico, Western and Northwest 43 North America, essentially due to warming (high confidence). Abatzoglou and Williams (2016) found 44 climate changes led to higher values for 8 fuel aridity indices over the Western US in recent decades, with 45 2000-2015 changes exposing 75% more forested area to high fuel aridity and adding 9 more high fire 46 potential days each year, similar to 1979-2013 Western US and Mexico fire season expansion in Jolly et al., 47 2015). Increases in lightning-initiated fires have been distinguished from trends in man-made fire in 48 Western Canada and the United States (Balch et al., 2017; Hanes et al., 2019). Jain et al. (2017) identified a 49 1979-2015 expansion in fire weather season in Eastern Canada and the southwestern United States (with a 50 smaller reduction in the northern mountain West) along with regional shifts in the 99th percentile Canadian 51 Fire Weather Index (FWI) and potential fire spread days. Girardin and Wotton (2009) noted that 1951-2002 52 trends in the Monthly Drought Code fire index in Eastern Canada could hardly be distinguished from decadal 53 variability. 54 55 Climate change drives future increases in North American fire weather, particularly in the Southwest (high Do Not Cite, Quote or Distribute 12-80 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 confidence), although shifts in exposure and vulnerability are needed to understand overall fire risks (see 2 WGII Chapter 14). A significant increase of FWI is apparent before RCP8.5 2050 in much of North 3 America, including the frequency of 95th percentile FWI days, peak seasonal FWI average, fire weather 4 season length, and maximum fire weather index (Abatzoglou et al., 2019), and fire season across North 5 America expands dramatically beyond 2℃ global warming levels (Sun et al., 2019b; Jain et al., 2020). Wang 6 et al. (2017) simulated fire spread days across Canada and found median increases by 2071-2100 of -20 to 7 140% (RCP2.6), -20 to 250% (RCP4.5) and 40 to 360% (RCP8.5) compared to 1976-2015. Prestemon et al. 8 (2016) found more conducive conditions for lightning-ignited fires in the Southeastern US by mid-century, 9 while warming conditions in Alaska increasingly push July temperatures above 13.4 ºC, a threshold for fire 10 danger across Alaska’s tundra and boreal forest (Partain et al., 2016; Young et al., 2017). Longer and more 11 intense fire seasons would also raise particulate matter and black carbon concentrations in the Western 12 United States, reducing visibility at many National Parks (Yue et al., 2013; Val Martin et al., 2015). 13 14 Changes in North American wet and dry climatic impact-drivers are largely organized by the 15 northeast (more wet) to southwest (more dry) pattern of mean precipitation change, although heavy 16 precipitation increases are widespread and increasing evaporative demand expands aridity, 17 agricultural and ecological drought, and fire weather (particularly in summertime) (high confidence). 18 19 20 12.4.6.3 Wind 21 22 Mean wind speed: Mean wind speeds have declined in past decades in North America as in other Northern 23 Hemisphere areas over the past four decades (AR5, WGI) (medium confidence) with a reversal in the last 24 decade (low confidence) not fully consistent across studies (Tian et al., 2019; Zeng et al., 2019; Zhang et al., 25 2019d) (Chapter 2). Tian et al. (2019) found a corresponding reduction in the wind power potential across 26 the eastern parts of North America. 27 28 Mean wind speeds are expected to decline over much of North America (Figure 12.4m-o), but the only broad 29 signal of consistent change across model types is a reduction in wind speed in Western North America (high 30 confidence). These declines reduce wind power endowment by 2050 and as early as the 2020-2040 near-term 31 period in the US Mountain West, while there is disagreement between global and regional model change 32 projections in the upper and lower Great Plains, Ohio River Valley, Mexico and Eastern Canada (Karnauskas 33 et al., 2018a; Jung and Schindler, 2019; Chen, 2020). 34 35 Severe wind storm: There is limited evidence and low agreement in observed changes in North American 36 CID indices associated with extratropical cyclones (Chapter 11), severe thunderstorms, severe wind bursts 37 (derechos), tornadoes, or lightning strikes (Vose et al., 2014; Easterling et al., 2017; Kossin et al., 2017). 38 Observational studies have indicated a reduction in the number of tornado days in the US, but increases in 39 outbreaks with 30 or more tornados in one day (Brooks et al., 2014), the density of tornado clusters (Elsner 40 et al., 2015), and overall tornado power (Elsner et al., 2019). 41 42 There is medium confidence of a general decrease in the number of extratropical cyclones producing high 43 wind speeds in North America, except over northernmost parts, for a global warming level of 2°C or by the 44 end of the century in under RCP4.5 and RCP8.5 (Kumar et al., 2015; Jeong and Sushama, 2018; Li et al., 45 2018a). GCMs cannot directly resolve tornadoes and severe thunderstorms, however projections of 46 favorable environments for severe storms (based on convective available potential energy and wind shear) 47 indicate medium confidence for more severe storms and a longer convective storm season in the United 48 States weaker increases extending north and east (Seeley and Romps, 2015; Glazer et al., 2020) and a 49 corresponding increase in fall and winter tornadic storms (Brooks, 2013a; Diffenbaugh et al., 2013; Brooks 50 et al., 2014) (see also Section 11.7.1). Prein et al. (2017a) used a convection-permitting model to project a 51 tripling of mesoscale convective systems over the United States for end-of-century RCP8.5. 52 53 Tropical cyclone: Section 11.7.1 identified recent reductions in tropical cyclone translation speed and higher 54 tropical cyclone rainfall totals over the North Atlantic as well as substantial natural variability. Projections 55 indicate low confidence in change in North Atlantic tropical cyclone numbers, but medium confidence in Do Not Cite, Quote or Distribute 12-81 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Mexico and the US Gulf and Atlantic coasts for more intense storms with higher wind, precipitation, and 2 storm surge totals when they do occur (Diro et al., 2014; DOE, 2015; Walsh et al., 2016a; Kossin et al., 3 2017; Marsooli et al., 2019; Ting et al., 2019; Knutson et al., 2020) (see Section 11.7.1). A more rapid 4 intensification of tropical cyclone winds and destructive power also heightens the tropical cyclone hazard 5 (Bhatia et al., 2019). Greenhouse gas forcing is projected to shift tropical cyclones poleward (Kossin et al., 6 2016), while also holding the potential for higher precipitation totals (Risser and Wehner, 2017; Knutson et 7 al., 2020) particularly given evidence that storms increasingly stall near North American coastlines (Hall and 8 Kossin, 2019). 9 10 Sand and dust storm: Land use change has increased dust emission in the western U.S. in the past 200 11 years (Neff et al., 2008). However, there is medium confidence for observed increases in Western North 12 American sand and dust storm activity since 1980. In their study of Valley Fever spread, Tong et al. (2017) 13 identified a rapid intensification of dust storm activity using PM10 and PM2.5 observations from 1980-2011 14 across 29 monitoring sites in the Southwestern US, similar to contiguous US observations by Brahney et al. 15 (2013). Hand et al. (2016) attributed the earlier onset of spring dusts in the Southwest in large part to the 16 Pacific Decadal Oscillation, however. The increasing trend in dust since the 1990s in the southwest US can 17 be explained by precipitation deficit and surface bareness (Pu and Ginoux, 2018). Projections of future sand 18 and dust storms over North America are based on aridity as a primary proxy for conducive conditions which 19 lends medium confidence of an increase over Mexico and the Southwest US. Pu and Ginoux (2017) project 20 about 5 more dusty days in spring and summer in the southern Great Plains under RCP8.5 at the end of the 21 century, while dusty days decrease in northern regions where mean precipitation tends toward wetter 22 conditions. 23 24 Tropical cyclones, severe wind and dust storms in North America are shifting toward more extreme 25 characteristics, with a stronger signal toward heightened intensity than increased frequency, although 26 specific regional patterns are more uncertain (medium confidence). Mean wind speed and wind power 27 potential are projected to decrease in Western North America (medium confidence) with differences 28 between global and regional models lending low confidence elsewhere. 29 30 31 12.4.6.4 Snow and ice 32 33 Snow: The seasonal extent of snow cover has reduced over North America in recent decades (robust 34 evidence, high agreement) (see also Section 2.3.2.2, Section 9.5.3 and Atlas). The average North America 35 area covered by snow decreased at a rate of about 8500 km2 per year over the 1972-2015 period, reducing 36 the average snow cover season by two weeks primarily due to earlier spring melt (US EPA, 2016). 37 Observations indicate earlier spring snowpack melting (Dudley et al., 2017) and a reduction in end-of-season 38 snowpack metrics important to water resources over the Rockies (particularly since 1980) and Pacific 39 Northwest (Pederson et al., 2013; Kormos et al., 2016; Kunkel et al., 2016; Fyfe et al., 2017; Mote et al., 40 2018). In situ measurements in Canada show more heterogenous trends in snow amount and density (Brown 41 et al., 2019). 42 43 Climate change is expected to reduce the total snow amount and the length of the snow cover season over 44 most of North America, with a corresponding decrease in the proportion of total precipitation falling as snow 45 and a reduction in end-of-season snowpack (high confidence) (see Atlas.9). Changes include a reduction in 46 days with snowfall in all but Northern Canada (Danco et al., 2016; McCrary and Mearns, 2019), a delay of 47 about a week in first snowfall in the US West in RCP8.5 2050 (Pierce and Cayan, 2013), and more 48 prominent reductions in Canadian snow cover in the October-December period (Mudryk et al., 2018). 49 Reduced total snowpack and earlier snowmelt lower dry season streamflow (Kormos et al., 2016; Rhoades et 50 al., 2018). Figure 12.10b shows a reduction in days suitable for skiing (SWE>10 cm; Wobus et al., 2017b) 51 across the US and Southern Canada, although some portions of Northern Central Canada see an increase. 52 53 Glacier: Section 9.5.1 assessed that glaciers in Alaska, Western Canada and the Western United States are 54 expected to continue to lose mass and areal extent (high confidence). Compared to their 2015 state, glaciers 55 in the Western Canada and the United States region will lose 62 ± 30%, 75 ± 29% and 85 ± 23%, of their Do Not Cite, Quote or Distribute 12-82 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 mass by the end of the century for RCP2.6, RCP4.5 and RCP8.5 scenarios, respectively (Marzeion et al., 2 2020). Meanwhile glaciers in Alaska will lose 26 ± 21%, 31 ± 24% and 44 ± 27%, of their mass in 2015 3 under the same scenarios. The overall loss of glacial mass can act as a meltwater supply for freshwater 4 resources, although this is expected to peak in the middle of the century and then fade as glaciers disappear 5 (Fyfe et al., 2017; Derksen et al., 2018). Continued shrinkage of glaciers is projected to create further glacial 6 lakes (medium confidence) similar to those that have led to outburst floods in Alaska and Canada (Carrivick 7 and Tweed, 2016; Harrison et al., 2018). 8 9 Permafrost: Warmer ground temperatures are expected to extend the geographical extent and depth of 10 permafrost thaw across northern North America (very high confidence) (see also Section 9.5.2). Observations 11 across Canada show that permafrost temperature is increasing and the active layer is getting thicker (Chapter 12 2.3.2.5; Biskaborn et al., 2019; Derksen et al., 2019; Romanovsky et al., 2020). Slater and Lawrence (2013) 13 note that end-of-century RCP8.5 in North America only has shallow permafrost as the most probable 14 condition in the Canadian Archipelago. Melvin et al. (2017) noted the loss of shallow permafrost in 5 15 RCP8.5 CMIP5 models across a wide swath of southern Alaska by 2050 along with increases of active layer 16 thickness. There is high confidence in continued reductions in mountain near-surface permafrost area with 17 high spatial variability given local snow and temperature changes (Chapter 9.5.2; Peng et al., 2018; Hock et 18 al., 2019). 19 20 Lake, river and sea ice: Anthropogenic warming reduces the seasonal extent of lake and river ice over 21 many North American freshwater systems, with ice-free winter conditions pushing further north with rising 22 temperatures (high confidence). Observations in Central and Eastern North America show reduced average 23 seasonal lake ice cover duration (Benson et al., 2012; Mason et al., 2016; US EPA, 2016). Satellite 24 observations show declines in lake ice (Du et al., 2017) and loss of more than 20% of winter river ice length 25 in much of Alaska (2008-2018 compared to 1984-1994; Yang et al., 2020). Spring lake and river ice in 26 Canada is projected to break up 10-25 days earlier while fall freeze-up occurs 5-15 days later by mid- 27 century, with larger declines in lake ice season by the coasts (Dibike et al., 2012) and for rivers in the 28 Rockies and Northeastern United States (Yang et al., 2020a) although global models have difficulty with 29 frozen freshwater system dynamics (Derksen et al., 2018). Substantial ice loss is projected over the 30 Laurentian Great Lakes (Hewer and Gough, 2019; Matsumoto et al., 2019). The southern extent of lakes 31 experiencing intermittent winter ice cover moves northward with rising temperature, pushing nearly out of 32 the continental United States at low elevations when GMT increases by 4.5℃ (Sharma et al., 2019). Higher 33 spring flows and the potential for winter thaws are also projected to heighten the threat of ice jams (Rokaya 34 et al., 2018; Bonsal et al., 2019) while reducing the seasonal viability of ice roads and recreational use 35 (Pendakur, 2016; Mullan et al., 2017; Knoll et al., 2019). 36 37 Seasonal sea ice coverage along the majority of Canadian and Alaskan coastlines is declining (robust 38 evidence, high agreement) and there is high confidence that hazards associated with a loss of sea ice increase 39 under climate change, as further assessed in Section 12.4.9. 40 41 Heavy snowfall: There is low agreement (limited evidence) for observed changes in heavy snowfall in 42 North America. Kluver and Leathers (2015) noted a 1930-2008 frequency increase for all snow intensities in 43 the Northern Great Plains but declines in heavier snow events in the Pacific Northwest and declines in the 44 Southeastern US. Changnon (2018) found that most extreme 30-day high snowfall periods in the 1900-2016 45 record over the Eastern US occurred in the 1959-1987 period that lies between the Dust Bowl and recent 46 warming. There is low agreement and medium evidence for broad projected changes to heavy snowfall over 47 North America given increased heavy precipitation and warmer winter temperatures. Several recent regional 48 studies have projected that low-intensity events decrease more rapidly than heavy snowfall events, resulting 49 in an increase in the snowfall proportion from heavy snowfall events even as the number of such events 50 decreases (O’Gorman, 2014; Lute et al., 2015; Zarzycki, 2016; Janoski et al., 2018; Ashley et al., 2020) 51 52 Ice storm: There is limited evidence in the literature of unique changes in ice storms observed or projected 53 over North America. Groisman et al. (2016) examined 40 years of observations and found weak decreases in 54 freezing rain events over the Southeastern US in the most recent decade. Ning and Bradley (2015) project 55 that the average snow-rain transition line, which is associated with mixed precipitation, moves 2˚ latitude Do Not Cite, Quote or Distribute 12-83 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 northward over Eastern North America by the end of the 21st century under RCP4.5 (4˚under RCP8.5) (see 2 also Klima and Morgan, 2015).. 3 4 Hail: There is limited evidence and low agreement for observed changes in the frequency or intensity of 5 North American hail storms. Allen et al., (2015) and Allen (2018) found that temporal inconsistencies in the 6 US and Canadian hail records made long-term climate analysis difficult, although Tang et al. (2019) 7 identified increasing environments conducive for hail ≥ 5 cm over the Central and Eastern United States. 8 There is limited evidence and medium agreement in projections of increased hail damage potential over 9 North America. Some regional and convective-permitting model projections indicate a longer hail season 10 with fewer events and larger hail sizes that result in higher hail damage potential (Brimelow et al., 2017; 11 Trapp et al., 2019). 12 13 Snow avalanche: There is limited evidence of directional changes in snow avalanches over North America. 14 Mock and Birkeland (2000) identified a 1969-1995 decrease in snow avalanches over the Western United 15 States, although they note the heavy influence of natural variability. A similar decline was observed over 16 Western Canada (Bellaire et al., 2016; Sinickas et al., 2016), but clear trends are difficult to discern given 17 sparse observations and shifts in avalanche management. We concur with the SROCC assessment of 18 medium confidence and high agreement that snow avalanche hazards generally decrease at low elevations 19 given lower snowpack even as high elevations are increasingly susceptible to wet snow avalanches (Hock et 20 al., 2019; see also Lazar and Williams, 2008). 21 22 Observations and projections agree that snow and ice CIDs over North America are characterized by 23 reduction in glaciers and the seasonality of snow and ice formation, loss of shallow permafrost, and 24 shifts in the rain/snow transition line that alters the seasonal and geographic range of snow and ice 25 conditions in the coming decades (very high confidence). 26 27 28 12.4.6.5 Coastal and oceanic 29 30 Relative sea level: Chapter 9 found that observations indicate increasing sea levels along most North 31 American coasts (robust evidence, high agreement), although there is substantial regional variation in 32 relative sea level rise (robust evidence, high agreement). Around North America, over 1900-2018, a new 33 tide-gauge based reconstruction finds a regional-mean RSL change of 1.08 [0.79-1.38] mm yr-1 in the 34 subpolar North Atlantic, 2.49 [1.89-3.06] mm yr-1 in the subtropical North Atlantic, and 1.20 [0.76-1.62] in 35 the East Pacific (Frederikse et al., 2020), compared to a GMSL change of around 1.7 mm yr-1 (Section 36 2.3.3.3; Table 9.5). For the period 1993-2018, these RSLR rates, based on satellite altimetry, increased to 37 2.17 [1.66-2.66] mm yr-1, 4.04 [2.77-5.24] mm yr-1 and 2.35 [0.70-4.06] mm yr-1, respectively (Frederikse et 38 al., 2020), compared to a GMSL change of 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). Relative sea level 39 (RSL) is falling in portions of southern Alaska (Sweet et al., 2018) and much of northern Northeastern 40 Canada and around Hudson Bay (where land is rising by >10 mm/year; Greenan et al., 2019). 41 42 Relative sea-level rise is virtually certain to continue in the oceans around North America, except in northern 43 Northeastern Canada and potions of Southern Alaska. Regional-mean RSLR projections for the oceans 44 around North America range from 0.4 m -1.0 m under SSP1-RCP2.6 to 0.7 m -1.4 m under SSP5-RCP8.5 for 45 2081-2100 relative to 1995-2014 (median values), which means that there are locally large deviations from 46 the projected GMSL change {Section 9.6.3.3}, including decreases in RSL in northern Northeastern Canada 47 from land uplift (see also Sweet et al., 2017; Greenan et al., 2019; Oppenheimer et al., 2019). The RSLR 48 projections here may however be underestimated due to potential partial representation of land subsidence in 49 their assessment (Section 9.6.3.2). 50 51 Coastal flood: Observations indicate that episodic coastal flooding is increasing along many coastlines in 52 North America (robust evidence, high agreement), and this episodic coastal flooding will increase in many 53 North American regions under future climate change (high confidence) although land uplift from glacial 54 isostatic adjustment in northern and Hudson Bay portions of Northeast North America leads to only medium 55 confidence of coastal flood increases in that region. Sweet et al. (2018) found 2000-2015 observed increases Do Not Cite, Quote or Distribute 12-84 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 of about 125% in high tide flooding frequencies along the southern Atlantic US coastline, with 75% 2 increases along the US Gulf Coast and US northern Atlantic coastlines. That same study noted that a GMSL 3 of 0.5m in 2100 would increase high tide (‘nuisance’) flooding from current rates of about once a month for 4 most coastal regions to about once every other day along the US Atlantic and Gulf Coasts and smaller 5 frequency increases along the Pacific coast, and Dahl et al. (2017) found similar trends on the US East Coast 6 prior to mid-century (see also . The present day 1:50 yr ETWL is projected to occur around 3 times a year by 7 2100 with a SLR of 1 m all around the region, except in most of Eastern North America where it is expected 8 to have return periods of 1:1 yr – 1:2 yrs (Vitousek et al., 2017). Ghanbari et al. (2019) projected 9 corresponding shifts toward higher frequencies of major flooding events for 20 US cities. Figure 12.4r and 10 Figure SM.12.6 show increases of 70 cm or more in the 100-yr return period extreme total water level 11 (ETWL) over much of the US East Coast, British Columbia, Alaska, and the Hudson Bay under RCP8.5 by 12 2100 (relative to 1980–2014), with lower increases in Northern Mexico, Northern Canada, Labrador, and the 13 Pacific and Gulf Coasts of the US (Vousdoukas et al., 2018). Projected coastal flooding increases generally 14 follow patterns of RSL change although sea ice loss in the north also increases open water storm surge 15 (Greenan et al., 2019). 16 17 Coastal erosion: There is limited evidence of changes in North American episodic storm erosion caused by 18 waves and storm surges. Observations show increased extreme wave energy on the Pacific coast, but no clear 19 trend on other US coasts given substantial natural variability (Bromirski et al., 2013; Vose et al., 2014). In 20 terms of long term coastal erosion, shoreline retreat rates of around 1 m yr-1 have been observed during 21 1984-2015 along the sandy coasts of NWN and NCA while portions of the US Gulf Coast have seen a retreat 22 rate approaching 2.5 m yr-1 (Luijendijk et al., 2018; Mentaschi et al., 2018).Sandy shorelines along ENA and 23 WNA have remained more or less stable during 1984 – 2014, but a shoreline progradation rate of around 0.5 24 m yr-1 has been observed in NEN Mentaschi et al. (2018) report 1984-2015 coastal area land losses of 630 25 km2 and 1,260 km2 along the Pacific and Atlantic coasts of USA, respectively. 26 27 Projections indicate that sandy coasts in most of the region will experience shoreline retreat through the 21st 28 century (high confidence). Median shoreline change projections presented by Vousdoukas et al. (2020) 29 compated CMIP5 projections show that sandy shorelines in NWN, ENA, and NCA will retreat by between 30 40 m – 80 m by mid-century (relative to 2010) under both RCP4.5 and RCP8.5. Projections for NEN and 31 WNA are lower at 20 m – 30 m under the same RCPs. The highest median mid-century projection in the 32 region is for CNA at around 125 m under both RCPs. RCP4.5 projections for 2100 show shoreline retreats of 33 100 m or more along the sandy coasts of NWN, CNA, and NCA, while retreats of between 40 m – 80 m are 34 projected in other regions. Under RCP8.5, retreats exceeding 100 m are projected in all regions except NEN 35 and WNA (approximately 80 m) by 2100, with particularly high retreats in NWN (160 m), CNA (330 m) and 36 SCA (200 m). The total length of sandy coasts in North America that are projected to retreat by more than a 37 median of 100 m by 2100 under RCP4.5 and RCP8.5 is about 15,000 km and 25,000 km respectively, an 38 increase of approximately 70%. 39 40 Marine heatwave: There is high confidence in observed increases in marine heatwave frequency and future 41 increases in marine heatwaves are very likely around North America (Box 9.2). The total number of MHW 42 days per decade increased in the North American coastal zone, albeit somewhat more in the Pacific (Oliver 43 et al., 2018; Smale et al., 2019). Projected increases in degree heating weeks (Heron et al., 2016) and degree 44 heating months (Frieler et al., 2013) indicate increasing bleaching-level and mortality-level heating stress 45 threshold events for reefs in Florida and Mexico. 46 47 Mean SST is projected to increase by 1 ºC (3 ºC) around North America by 2100, with a hotspot of around 4 48 ºC (5 ºC) off the North America Atlantic coastline under RCP4.5 (RCP8.5) conditions (see Interactive Atlas). 49 Frölicher et al. (2018) projected increasing MHW frequency and spatial extent at a 2℃ global warming level 50 with the largest increases in the Gulf of Mexico and off the southern US East Coast (>20x) as well as off the 51 coast of the Pacific Northwest (>15x). Projections for SSP1-2.6 and SSP5-8.5 both show an increase in 52 MHWs around North America by 2081 – 2100, relative to 1985 – 2014 (Box 9.2, Figure 1). 53 54 There is high confidence that most coastal CIDs in North America will continue to increase in the 55 future with climate change. An observed increase in relative sea level rise is virtually certain to Do Not Cite, Quote or Distribute 12-85 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 continue in North America (other than around the Hudson Bay and Southern Alaska) contributing to 2 more frequent and severe coastal flooding in low-lying areas (high confidence) and shoreline retreat 3 along most sandy coasts (high confidence). Marine heatwaves are also expected to increase all around 4 the region over the 21st century (high confidence). 5 6 The assessed direction of change in climatic impact-drivers for North America and associated confidence 7 levels are illustrated in Table 12.8. 8 9 10 [START TABLE 12.8 HERE] 11 12 Table 12.8: Summary of confidence in direction of projected change in climatic impact-drivers in North America, 13 representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, 14 SRES A1B, or above within each AR6 region (defined in Chapter 1), approximately corresponding (for 15 CIDs that are independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for 16 more details of the assessment method). The table also includes the assessment of observed or projected 17 time-of-emergence of the CID change signal from the natural inter-annual variability if found with at 18 least medium confidence in Section 12.5.2. 19 Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region North Central America (NCA) 2 Western North America (WNA) 3 5 5 4,7 6,7 6,7 8 6 1 1 5 2 Central North America (CNA) 7 7 7 8 4 2 Eastern North America (ENA) 5 7 8 1 1 1 2 Northeast North America (NEN) 5 5 6,7 6,7 8 1,6 1 4 4,6 2,6 Northwest North America (NWN) 5 6 5 6,7 6,7 8 1 1,6 9 2 1. Snow may increase in some high elevations and during the cold season and decrease in other seasons and at lower elevations. 2. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. 3. Increasing in northern regions and decreasing toward south. 4. Decreasing in northern regions and increasing toward south. Key 5. Higher confidence in northern regions and lower toward south. High confidence of decrease 6. Higher confidence in southern regions and lower toward north. Medium confidence of decrease 7. Higher confidence in increase for some climatic impact driver indices during summertime. Low confidence in direction of change 8. Increase in convective conditions but decrease in winter extratropical cyclones. Medium confidence of increase 9. Relative sea level rise reduced given land uplift in Southern Alaska. High confidence of increase Not broadly relevant 20 21 [END TABLE 12.8 HERE 22 23 24 12.4.7 Small islands 25 26 This section covers the climatic impact-drivers affecting small islands around the world (see definition of 27 SIDS in the Glossary; Cross-Chapter Box Atlas.2) with a particular focus on small islands in the Caribbean 28 (CAR) Sea and the Pacific Ocean. Caribbean and Pacific small islands have mostly tropical climates and 29 local conditions are also influenced by diverse topography ranging from low-lying islands and atolls to 30 volcanic and mountainous terrain. Climate variability in these islands is influenced by the trade winds, the Do Not Cite, Quote or Distribute 12-86 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 easterly waves, tropical cyclones (TC), and the migrations of the Inter-Tropical Convergence Zone (ITCZ), 2 the North Atlantic Subtropical High, and the South Pacific Convergence Zone (SPCZ), and other modes of 3 climate variability as discussed in Cross-Chapter Box Atlas.2. Furthermore, changes in the ocean 4 temperature and chemistry, and relative sea level have strong impacts on these small islands given their 5 geographical location and dependence on coastal and marine ecosystem services. 6 7 IPCC AR5 recognized the heterogeneity in these small islands in terms of physical geography, socio- 8 economic and cultural backgrounds, as well as their vulnerability to the impacts of climate change. Similar to 9 previous reports, these regions have been assessed together in this section given similarities in the challenges 10 they face in addressing climate change impacts and risk, which was thought to be dominated by sea level rise 11 until IPCC AR4 (Nurse et al., 2014; Betzold, 2015). Since then there has been a substantial increase in the 12 number and complexity of the literature on the drivers and impacts of climate change on small islands (BOM 13 and CSIRO, 2011, 2014; Nurse et al., 2014; Gould et al., 2018; Keener et al., 2018). There are also 14 increasing efforts in producing higher resolution climate projections for the small islands through 15 downscaling methods (Elison Timm et al., 2015; McLean et al., 2015; Khalyani et al., 2016; Zhang et al., 16 2016a; Stennett-Brown et al., 2017; Bhardwaj et al., 2018; Bowden et al., 2021). 17 18 IPCC AR5 identified the key climate and ocean-related hazards affecting the small islands, which occur at 19 different timescales and have diverse impacts on multiple sectors (Christensen et al., 2013; Nurse et al., 20 2014). Recent findings from IPCC SR1.5 and IPCC SROCC emphasize that the multiple interrelated climate 21 hazards currently faced by low-lying islands and coastal areas will be amplified in the future, especially at 22 higher global warming levels (Hoegh-Guldberg et al., 2018; IPCC, 2019b). 23 24 25 12.4.7.1 Heat and Cold 26 27 Mean air temperature: Significant warming trends are clearly evident in the small islands, such as those in 28 the Pacific, CAR, and Western Indian Ocean, particularly over the latter half of the 20th century (see Figure 29 Atlas.11; Section Atlas.10.2; Cross-Chapter Box Atlas.2, Table 1). This observed warming signal in the 30 tropical western Pacific has been attributed to anthropogenic forcing (Wang et al., 2016). There is high 31 confidence of warming over small islands even at 1.5°C global warming level (GWL) (Section Atlas.10.4; 32 Figure Atlas.31) (Hoegh-Guldberg et al., 2018). Mean temperature is very likely to increase by 1°C–2°C 33 (2°C–4°C) by 2041–2060 (2081–2100) under RCP8.5 (BOM and CSIRO, 2014) and SSP3-7.0 (Figure 4.19 34 and Figure Atlas.13) (Section Atlas.10.4) (Almazroui et al., 2021). 35 36 Extreme heat: Observation records indicate warming trends in the temperature extremes since the 1950s in 37 CAR and the Pacific small islands (high confidence) (Section 11.3.2; Section 11.9, Table 11.7). A detectable 38 anthropogenic increase in summertime heat stress index has been identified over a number of island regions 39 in CAR, western tropical Pacific, and tropical Indian Ocean, based on wet bulb globe temperature (WBGT) 40 index trends for 1973–2012 (medium confidence) (Knutson and Ploshay, 2016). An increasing trend in the 41 maximum daytime heat index is also noted in CAR during the 1980–2014 period, as well as more extreme 42 heat events since 1991 (Ramirez-Beltran et al., 2017). 43 44 Compared with the recent past, it is likely that the intensity and frequency of hot (cold) temperature extremes 45 will increase (decrease) in the small islands (Section 11.9, Table 11.7) (BOM and CSIRO, 2014). Warm 46 spell conditions will occur up to half the year in CAR at 1.5°C GWL with an additional 70 days at 2°C 47 (Taylor et al., 2018), with livestock temperature-humidity tolerance thresholds increasingly surpassed (Lallo 48 et al., 2018). In CAR, a median increase of almost a month where temperatures exceed 35°C is projected by 49 end of the 21st century under SSP5-8.5 (Figure 12.4a.c and Figure SM 12.1). Heat waves are projected to 50 increase in CAR by the mid- and end of the 21st century under RCP8.5 (Section 11.3.5; Section 11.9, Table 51 11.7). Figure 12.4d-f and Figure SM 12.2 also show an increase of about 30-60 days with HI exceeding 41°C 52 by 2041-2060 under SSP5-8.5 relative to 1995-2014 in CAR, with an additional increase of about days by 53 end of the 21st century for RCP8.5/SSP5-8.5, but this increase remains below 50 days for RCP2.6/SSP1-2.6. 54 The Pacific Islands is also among regions projected to have an increase in number of days with mean HI 55 exceeding 41°C by 2091-2100 under RCP8.5/SSP5-8.5, increasing the risk of heat stress in the region Do Not Cite, Quote or Distribute 12-87 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 (Newth and Gunasekera, 2018). 2 3 It is very likely that the significant recent warming trends observed in the small islands will continue in 4 the 21st century, which will likely further increase heat stress in these regions. 5 6 7 12.4.7.2 Wet and Dry 8 9 Mean precipitation: Observational datasets have generally revealed no significant long-term trends in 10 rainfall in the Caribbean over the 20th century when analysed at seasonal and inter-decadal timescales, 11 except for some areas where there is evidence for decreasing trends for the period 1901-2010 but not for the 12 period 1951-2010 (Knutson and Zeng, 2018; Section Atlas.10.2; Cross-Chapter Box Atlas.2, Table 1). 13 Although there are spatial variations, annual rainfall trends in the Western Indian Ocean are mostly 14 decreasing, with generally non-significant trends in the Western tropical Pacific since the 1950s (low 15 confidence). Significant drying trends are noted in the southern Pacific subtropics and southwestern French 16 Polynesia during the 1951-2015 period (McGree et al., 2019), and in some areas of Hawaii during the 1920- 17 2012 period (medium confidence) (Section Atlas.10.2; Cross-Chapter Box Atlas.2, Table 1). 18 19 Atlas.10.4 projects precipitation reduction over the Caribbean (high confidence) and parts of the 20 Atlantic and Indian Oceans, particularly in June to August by end of 21st century. Precipitation is 21 generally projected to increase in the small islands in parts of the western and equatorial Pacific, but there is 22 low confidence in broad changes givend drier conditionsprojected for the southern sub-tropical and eastern 23 Pacific Ocean (limited agreement given spatial and seasonal variability; Section Atlas.10.4; Figure Atlas.31) 24 (Almazroui et al., 2021). 25 26 River flood: There is limited evidence on observed changes in river flooding in small islands. Long-term 27 records in Hawaii indicate no clear trends in peak flow, except for the significant decrease in peak 28 streamflow in Hawaii Island over the period 1967-2016 (Bassiouni and Oki, 2013; Clilverd et al., 2019). 29 Similarly, there is no significant trend in the frequency and height (after adjusting for average sea level rise) 30 of river flood in Fiji over the period 1892-2013 (McAneney et al., 2017). There is low confidence on the 31 direction of future change of river flooding in the small islands due to the limited currently available 32 literature. In Oahu, Hawaii, extreme peak flow events with high return periods are projected to increase by 33 end of 21st century under RCP8.5, but there is also high uncertainty in these projections (Leta et al., 2018). 34 35 Heavy precipitation and pluvial flood: Heavy precipitation days in CAR have increased in magnitude, and 36 have been more frequent in the northern part during the latter part of the 20th century (low confidence) 37 (Section 11.4.2; Table 11.7). The direction of change in extreme precipitation varies across the Pacific and 38 depends on the season (low confidence) (Section 11.4.2; Cross-Chapter Box Atlas.2, Table 1). Although 39 pluvial flooding events have been observed in some islands, there is limited evidence for an assessment on 40 past changes in pluvial flooding unlike in other regions. There is low confidence in the projected increase in 41 magnitude of very heavy precipitation days in CAR across different GWLs (Table 11.7). On the other hand, 42 there is high confidence in the increase in frequency and intensity of extreme rainfall events (i.e. 1-in-20 year 43 rainfall events) in the western tropical Pacific in the 21st century even for RCP2.6 scenario based on model 44 agreement and mechanistic understanding but low confidence in the magnitude of change in extreme rainfall 45 due to model bias (BOM and CSIRO, 2014). 46 47 Landslide: Heavy rainfall, such as from tropical cyclones, can trigger landslides over steep terrain in the 48 small islands (Bessette-Kirton et al., 2019). There is limited evidence to determine long-term trends in 49 rainfall-induced landslides in the small islands (Kirschbaum et al., 2015; Sepúlveda and Petley, 2015; Froude 50 and Petley, 2018; Bessette-Kirton et al., 2019). There is low confidence in the future change in landslides in 51 the small islands. The direction of change may depend on future changes in precipitation, tropical cyclones, 52 climate modes (e.g., El Niño-Southern Oscillation (ENSO)), as well as human disturbance, but more data 53 and understanding of the complexity of these relationships are needed, especially in these vulnerable areas 54 (Sepúlveda and Petley, 2015; Gariano and Guzzetti, 2016; Froude and Petley, 2018). 55 Do Not Cite, Quote or Distribute 12-88 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Aridity: Current estimates identify many small islands as being under water stress and thus particularly 2 sensitive to variations in rainfall and groundwater, population growth and demand, land use change, among 3 others (Holding et al., 2016) (Cross-Chapter Box Atlas.2). From 1950 to 2016, a heterogeneous but prevalent 4 drying trend is found in CAR (medium confidence), where drought variability is modulated by the tropical 5 Pacific and North Atlantic oceans (Herrera and Ault, 2017) (Table 11.7, Cross-Chapter Box Atlas.2, Table 6 1). In the future, increased aridity and decreased freshwater availability are projected in many small islands 7 due to higher evapotranspiration in a warmer climate that partially offsets increases or exacerbates reductions 8 in precipitation (Karnauskas et al., 2016, 2018b; Hoegh-Guldberg et al., 2018). Increased aridity is projected 9 for the majority of the small islands, such as in CAR, southern Pacific and Western Indian Ocean, by 2041- 10 2059 relative to 1981-1999, and at 1.5°C and 2.0°C GWLs, under RCP8.5, which will further intensify by 11 2081-2099 (Karnauskas et al., 2016, 2018b) (medium confidence). Groundwater recharge is projected to 12 increase in Maui, Hawaii except on the leeward sides of the island, which underscores the importance of 13 topography and elevation on freshwater availability in different island microclimates (Brewington et al., 14 2019; Mair et al., 2019). 15 16 Hydrological drought: There is low confidence of widespread changes to hydrological drought in the 17 Caribbean or Pacific small islands in recent decades, although an increasing number of studies document 18 local changes. Records in Hawaii indicate downward trends in low streamflow and base flow from 1913 to 19 2008 (Bassiouni and Oki, 2013). Decadal variability of Hawaiian streamflow coincides with rainfall 20 fluctuations associated with the Pacific Decadal Variability although significant average declines in surface 21 and baseflow runoff of about 8% and 11% per decade, respectively, have been noted during the 1987-2016 22 period (Clilverd et al., 2019). 23 24 There is low confidence in hydrological drought change projections given low signal to noise ratios and the 25 challenge in representing island scales in global analyses. Prudhomme et al. (2014) recognized the 26 Caribbean as one of the regions with highest increase in regional deficit index (RDI; a measure of the 27 fraction of area in hydrological drought conditions) by end of the 21st century under RCP8.5. Daily 28 streamflow and extreme low flows in two watersheds in Oahu, Hawaii are projected to decline by mid- and 29 end of the 21st century under RCP4.5 and RCP8.5, which would result in more frequent hydrological 30 droughts in this area (Leta et al., 2018). 31 32 Agricultural and ecological drought: Recent trends toward more frequent and severe droughts have been 33 noted in the small islands but only generate low confidence in broad trend patterns given high spatial 34 variability including heightened drought on the leeward side of islands (e.g., Frazier and Giambelluca, 2017; 35 Herrera and Ault, 2017; McGree et al., 2019; Table 11.7, Cross-Chapter Box Atlas.2, Table 1). There is 36 medium confidence that agricultural and ecological droughts will increase in frequency, duration, magnitude, 37 and extent in the small islands, such as in CAR and parts of the Pacific, particularly where there are future 38 declines in precipitation compounded by higher evapotranspiration, under increasing levels of warming 39 (Naumann et al., 2018; Taylor et al., 2018; Vichot‐Llano et al., 2021). Relative to the period 1985-2014, 40 decreases in annual surface and total column soil moisture becomes more robust in more areas in CAR by 41 2071-2100 under SSP3-7.0 and SSP5-8.5 scenarios (Cook et al., 2020a), but global simulations are 42 challenged to represent drought features in small island domains (see also Section 11.9). 43 44 Fire weather: There is limited evidence on trends in wildfire in CAR and the Pacific. Records of wildfire in 45 Hawaii from 2005-2011 indicate a peak in area burned during the hot and dry summer months, but 46 Trauernicht et al. (2015) note the difficulty in establishing the link between past climate and wildfire trends 47 due to human activities and vegetation changes. Availability of literature limits assessment on future fire 48 weather in the small islands (low confidence). Drying and warming trends tend to increase fire probability 49 aside from the climate impact on fuel loading, e.g., grassland fires in Hawaii (Trauernicht, 2019), and 50 wildfires in Puerto Rico (Van Beusekom et al., 2018). 51 52 Observed and projected rainfall trends vary spatially across the small islands. Higher 53 evapotranspiration under a warming climate can partially offset future increases or amplify future 54 reductions in rainfall resulting in drier conditions and increase water stress in the small islands 55 (medium confidence). Do Not Cite, Quote or Distribute 12-89 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 12.4.7.3 Wind 2 3 Mean wind speed: Scarcity of observations limits assessment of long-term changes in winds over the small 4 islands in the Pacific and CAR. Records indicate that average daily wind speeds have slowly declined in 5 Hawaii, but trends have remained constant across western and south Pacific sites since the mid-20th century 6 (Marra and Kruk, 2017). Recent studies of reanalyses and hindcast simulations indicate an intensification of 7 the Pacific trade winds during the 1992-2011 period, which contributed to the ocean cooling in the tropical 8 central and eastern Pacific (England et al., 2014; Takahashi and Watanabe, 2016). Projections estimate up to 9 0.4 m s-1 (8%) increase in annual winds in CAR under RCP8.5, which is associated with changes in the 10 extension of the North Atlantic Subtropical High that enhances the Caribbean low-level jet during the wet 11 season, and stronger local easterlies due to enhanced land-ocean temperature differences in the dry season 12 (Costoya et al., 2019) (low confidence). 13 14 Tropical cyclone: Tropical cyclones have devastating impacts on the small islands due to intense winds, 15 storm surge, and rainfall, although the associated rainfall can also be beneficial for freshwater resources. It is 16 likely that tropical cyclone intensity and intensification rates at a global scale have increased in the past 40 17 years but it is not clear if regional-scale changes are basin-wide or due to shifts in tracks (Section 11.7.1.2). 18 Other features are less sensitive to these issues, such as the poleward migration where tropical cyclones 19 reach peak intensity in the western North Pacific and the slowdown in tropical cyclone translational speed in 20 most basins in the latter half of the 20th century (low confidence), which can enhance rainfall and flooding 21 events, including over small islands in CAR and the Pacific (Section 11.7.1.2). 22 23 Future global changes in tropical cyclones include more frequent Category 4-5 storms (high confidence) and 24 increased rain rates (very high confidence) (Knutson et al., 2020), with relative sea level rise exacerbating 25 storm surge potential but large differences per region depending also on changes to future storm tracks (see 26 Section 11.7.1.5). By the late 21st century, tropical cyclones are projected to be less frequent in the basins of 27 the western and eastern North Pacific, Bay of Bengal, Caribbean Sea and in the Southern Hemisphere, but 28 will be more frequent in the subtropical central Pacific (Murakami et al., 2014; Yoshida et al., 2017; Bell et 29 al., 2019; Knutson et al., 2020). Over CAR, tropical cyclone intensity is expected to increase due to higher 30 sea surface temperatures but can be inhibited by increases in vertical wind shear in the region (medium 31 confidence) (Kossin, 2017; Ting et al., 2019). The poleward movement of the area where tropical cyclones 32 reach peak intensity in the western North Pacific is likely to continue, which affects the tropical cyclone 33 frequency over the small islands in the area (Kossin et al., 2016) (Section 11.7.1.5). Projections also indicate 34 an increase (decrease) in the tropical cyclone frequency during El Niño (La Niña) events in the Pacific at the 35 end of the 21st century (Chand et al., 2017). RCP8.5 2080-2099 projections indicate an increase in tropical 36 cyclone number by 2% in the North Central Pacific relative to 1980-1999, with tracks shifting northward 37 towards Hawaii (Li et al., 2018d). Given projected reductions to the overall number of storms but increases 38 in storm intensity, total rainfall and storm surge potential, we assess medium confidence of overall changes to 39 tropical cyclones affecting the Caribbean and Pacific small islands. 40 41 Global changes indicate that small islands will generally face fewer but more intense tropical cyclones 42 (medium confidence) although there is substantial variability across small island regions given 43 projected regional shifts in storm tracks. 44 45 46 12.4.7.4 Coastal and Oceanic 47 48 Relative sea level: Relative sea level rise (SLR) continues to be a major threat to small islands and atolls, 49 since it can exacerbate the impacts of other climate hazards on low-lying coastal communities and 50 infrastructures, ecosystems, and freshwater resources (Nurse et al., 2014; Hoegh-Guldberg et al., 2018). In 51 the Indian Ocean- South Pacific region, a new tide-gauge based reconstruction finds a regional-mean RSL 52 change of 1.33 [0.80 to 1.86] mm yr-1 over 1900-2018 (Frederikse et al., 2020) compared to a GMSL change 53 of around 1.7 mm yr-1 (Section 2.3.3.3; Table 9.5). RSLR rates based on satellite altimetry for the period 54 1993-2018 increased to 3.65 [3.23 to 4.08] mm yr-1 (Frederikse et al., 2020), compared to a GMSL change of 55 3.25 mm yr-1 (Section 2.3.3.3; Table 9.5). Do Not Cite, Quote or Distribute 12-90 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Relative sea-level rise is very likely to continue in the oceans in the Small Island States. Around the small 2 islands, regional-mean RSLR projections vary widely, from 0.4 m–0.6 m under SSP1-RCP2.6 to 0.7 m–1.6 3 m under SSP5-RCP8.5 for 2081–2100 relative to 1995–2014 (median values), but in general they are 4 situated in areas with RSL changes ranging from the mean projected GMSL change to above-average values 5 {Section 9.6.3.3}. These RSLR projections may however be underestimated due to potential partial 6 representation of land subsidence in their assessment (Section 9.6.3.2). 7 8 Coastal flood: Relative sea-level rise, storm surges, and swells contribute to coastal inundation in the small 9 islands, where studies on historical trends in coastal flooding are currently limited. For example, a swell 10 event due to distant extra-tropical cyclones in December 2008 raised extreme water levels leading to 11 flooding affecting five Pacific island nations: Marshall Islands, Micronesia, Papua New Guinea, Kiribati and 12 Solomon Islands (Hoeke et al., 2013; Merrifield et al., 2014). Over low-lying atoll islands in the northwest 13 tropical Pacific, potential increases in the areal extent in coastal flooding with reduced return periods, 14 especially at higher SLR scenarios, are expected to have negative consequences on freshwater resources and 15 island habitability (Storlazzi et al., 2015, 2018). Select tide gauges across the Pacific also indicate increasing 16 trends in the frequency of minor flooding since the 1960s (Marra and Kruk, 2017). 17 18 As relative sea levels increase, the potential for coastal flooding increases in the small islands (high 19 confidence). Across the Pacific and CAR small islands, the 5th – 95th percentile range of the 1:100 yr ETWL 20 is projected to increase (relative to 1980 – 2014) by 10 cm – 35 cm and by 14 cm – 41 cm by 2050 under 21 RCP4.5 and RCP8.5, respectively (Figure 12.4q). By 2100, this range is projected to be 27 cm – 81 cm and 22 44 cm – 188 cm under RCP4.5 and RCP8.5, respectively (Figure 12.4p,r) (Vousdoukas et al., 2018; Kirezci 23 et al., 2020). Furthermore, by 2050, the present day 1:100 yr ETWL is projected to have median return 24 periods of between 1:1 yr and 1:50 yr in both the Pacific and CAR small islands, with some Pacific islands 25 projected to experience the present day 1:100 yr ETWL more than once a year (Vousdoukas et al., 2018; 26 Oppenheimer et al., 2019). By 2100, the present day 1:50 yr ETWL is projected to occur around 3 times a 27 year by 2100 with a SLR of 1 m at Pacific and CAR small islands (Vitousek et al., 2017). In the Western 28 Tropical Pacific, the magnitude and frequency of coastal flooding due to SLR can be modulated by changes 29 in the wave climate (Shope et al., 2016). 30 31 Coastal erosion: Recent studies have indicated variable and dynamic changes in shorelines of reef islands 32 (medium confidence), including both erosion and accretion, which suggest factors other than SLR affecting 33 shoreline changes, such as in the central and western Pacific within the past 50 to 60 year timeframe (Webb 34 and Kench, 2010; Le Cozannet et al., 2014; Ford and Kench, 2015; Duvat and Pillet, 2017). For example, 35 islands on atolls in the central and western Pacific have not substantially eroded or reduced in size in the past 36 decades when sea level has increased but rather have changed their position and morphology due to 37 anthropogenic factors (e.g., seawalls, reclamation) and climate-ocean processes (Biribo and Woodroffe, 38 2013; McLean and Kench, 2015). Analysis of aerial and satellite imagery revealed severe shoreline retreat in 39 six islands and the disappearance of five vegetated reef islands in Solomon Islands in the western Pacific 40 between 1947 to 2014, which may be due to the interaction between SLR and waves (Albert et al., 2016). In 41 French Polynesia, changes in shoreline and island area have been observed since the 1960s, partly due to the 42 effect of TCs on sediment changes and human activities (Duvat and Pillet, 2017; Duvat et al., 2017). Coastal 43 erosion have also been noted over the small, low-lying, sandy islands, such as in French Polynesia and 44 Solomon Islands, among others, due to high relative sea level rise and storms (Luijendijk et al., 2018a; 45 Mentaschi et al., 2018). Average shoreline retreat rates between 1 m yr-1 and 2 m yr-1 are estimated for the 46 islands in the Equatorial Pacific and in CAR, while a retreat rate of 0.5 m is estimated for islands in the south 47 Pacific, based on satellite observations from 1984-2016 (Luijendijk et al., 2018a; Mentaschi et al., 2018). 48 There was also a loss of 610 km2 vs. a gain of 520 km2 in coastal area in Oceania during the 1984-2015 49 period (Mentaschi et al., 2018). 50 51 Projections indicate that shoreline retreat will occur over most of the small islands in the Pacific and CAR 52 throughout the 21st century with spatial variability (high confidence). Median shoreline change projections 53 (CMIP5), relative to 2010, presented by Vousdoukas et al. (2020) show that, by mid-century, shorelines in 54 the islands in the equatorial Pacific and south Pacific will retreat by around 40 m, under both RCP4.5 and 55 RCP8.5. In CAR islands, sandy shorelines are projected to retreat by about 80 m by mid-century under both Do Not Cite, Quote or Distribute 12-91 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 RCPs. By 2100, more than 100 m of median shoreline retreat is projected for all small islands under both 2 RCPs; notably in CAR where retreats approaching 200 m (relative to 2010) are projected under both RCPs. 3 The total length of sandy coasts in CAR and Pacific small islands that is projected to retreat by more than a 4 median of 100 m by 2100 under RCP4.5 and RCP8.5 is about 1,100 km and 1,200 km respectively, an 5 increase of approximately 14%. 6 7 Marine heatwave: Ocean temperatures from satellite observations noted a moderate increase of 1–4 annual 8 marine heat wave (MHW) events between 1982–1988 and 2000–2016 over some areas in the Indian Ocean, 9 subtropical parts of the North and South Atlantic, and central and western parts of the North and South 10 Pacific, but a decrease in frequency (2 annual events) over the eastern Pacific Ocean (Oliver et al., 2018) 11 (see Box 9.2). The intensity of MHWs has also increased between 0.2°C and 0.5°C over the equatorial 12 portions of the North Atlantic, and the South Pacific. Over the eastern tropical Pacific, the decrease in 13 intensity and duration of MHW is between 0.5–1.0 °C and between 30–75 days, respectively (Oliver et al., 14 2018) (see Box 9.2). There is high confidence that MHWs will increase around all small island nations. 15 Marine heatwaves are projected to be more intense and prolonged where the largest changes are noted in the 16 equatorial region with maximum annual intensities up to 1.2°C (1.8°C) and annual mean duration reaching 17 100 days (200 days) at 1.5°C (2.0 °C) warming levels (Frölicher et al., 2018). Projections for SSP1-2.6 and 18 SSP5-8.5 both show an increase in MHWs around all small island nations by 2081–2100, relative to 1985– 19 2014 (Box 9.2, Figure 1). 20 21 In summary, relative sea level rise is very likely in the oceans around small islands, and along with 22 storm surges and waves will exacerbate coastal inundation in small islands. Shoreline retreat is 23 projected along sandy coasts of most small islands (high confidence). There is high confidence that 24 MHWs will increase around all small island nations. 25 26 The assessed direction of change in climatic impact-drivers for CAR and Pacific small islands and associated 27 confidence levels are illustrated in Table 12.9. Cold, Snow, Ice related climatic impact-drivers, and sand and 28 dust storms are not broadly relevant in small islands. 29 30 31 [START TABLE 12.9 HERE] 32 33 Table 12.9: Summary of confidence in direction of projected change in climatic impact-drivers in the small islands, 34 representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, 35 SRES A1B, or above within each AR6 region (defined in Chapter 1), approximately corresponding (for 36 CIDs that are independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for 37 more details of the assessment method). The table also includes the assessment of observed or projected 38 time-of-emergence of the CID change signal from the natural inter-annual variability if found with at 39 least medium confidence in Section 12.5.2. 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-92 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Caribbean (CAR) 6 7 Pacific Islands 1 2 3 4 5 6 7 1. Very high confidence in the direction of change, but low to medium confidence in the magnitude of change due to model uncertainty. 2. Decrease in eastern Pacific and southern Pacific subtropics, but Increase in parts of western and equatorial Pacific; with seasonal variation in future changes. 3. High confidence in increase in extreme rain frequency and intensity in western tropical Pacific; low confidence in magnitude of change due to Key model bias. High confidence of decrease 4. Increase in southern Pacific. Medium confidence of decrease 5. Particularly in parts of the Pacific with projected rainfall declines. Low confidence in direction of change 6. Increase in intensity; decrease in frequency except over central north Pacific. Medium confidence of increase 7. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. High confidence of increase Not broadly relevant 1 2 [END TABLE 12.9 HERE] 3 4 5 12.4.8 Open and deep ocean 6 7 Oceans face challenges from anthropogenic perturbations to the global Earth system, which cause increasing 8 ocean warming, carbon dioxide induced acidification and oxygen loss (Bindoff et al., 2019). Climate change 9 will affect the major oceanic CIDs described in Section 12.2; mean ocean temperature, marine heatwaves, 10 ocean acidity, ocean salinity, dissolved oxygen (O2), severe wind storms and sea ice melting. These changes 11 result in a shifting profile of hazards relevant to impact and risk assessments (Section 12.3). New evidence, 12 the IPCC (2019b) and advances in the new CMIP6 climate simulations reinforce confidence in projected 13 changes in climatic impact-drivers in the global oceans. As the ocean has taken up about 90% of the global 14 warming for the period 1971-2018 (Section 7.2.2.2), the emergence of the sea surface temperature increase 15 signal has already been observed in global oceans over the last century (Hawkins et al., 2020). The 16 emergence signal in in sea ice extent decrease has already emerged in the Arctic Ocean (Landrum and 17 Holland, 2020), while ocean acidification and low oxygen have also already emerged in many ocenic regions 18 and will emerge in all global oceans by 2050 under RCP8.5 (see Section 12.5.2; Table 12.10). This section 19 assesses key hazards that can be linked with sectoral and regional vulnerability and exposure in open and 20 deep oceans, drawing from previous Chapters (2,3,4, 5, and 9). 21 22 Mean ocean temperature: It is very likely that global mean sea surface temperature global mean SST has 23 increased by 0.88°C [0.68 to 1.01°C] from 1850-1900 to 2011-2020, and 0.60°C [0.44 to 0.74°C] from 24 1980-2020 (Section 2.3.1.1.6; Table 2.4). There is very high confidence that the Indian Ocean, western 25 equatorial the Pacific Ocean, and western boundary currents have warmed faster than the global average, 26 while the Southern Ocean, the eastern equatorial Pacific, and the North Atlantic Ocean have warmed more 27 slowly or slightly cooled (Section 9.2.1.1). It is virtually certain that SST will continue to increase in the 21st 28 century at a rate depending on future emission scenario, while the global mean ocean surface temperature 29 will continue to increase throughout the 21st century (very high confidence), with CMIP6 projections 30 indicating an increase of 0.86°C (likely range: 0.43-1.47°C) under SSP1-2.6 and 2.89°C (2.01-4.07°C) under 31 SSP5-8.5, by 2081-2100, relative to 1995-2014 (Section 9.2.1.1). Global warming of 2°C above pre- 32 industrial levels is projected to increase sea surface temperature, resulting in the exceedance of numerous 33 hazard thresholds for pathogens, seagrasses, mangroves, kelp forests, rocky shores, coral reefs and other 34 marine ecosystems (Poloczanska et al., 2013b, 2016, 2013a; Pörtner et al., 2014; Liu et al., 2014; Graham et Do Not Cite, Quote or Distribute 12-93 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 al., 2015; Schoepf et al., 2015; Gobler et al., 2017; Henson et al., 2017; Hoegh-Guldberg et al., 2017; 2 Krueger et al., 2017; Hughes et al., 2018b, 2018a; Perry et al., 2018) (medium confidence). It is virtually 3 certain that upper ocean stratification has increased at a rate of 4.9±1.5% during 1970-2018 and that this will 4 continue to increase in the 21st century (Section 9.2.1.3), potentially leading to reduced nutrient supply and 5 total productivity (Moore et al., 2018) (low confidence). 6 7 Marine heatwave: Marine heaywaves have increased in frequency over the 20th century, with an 8 approximate doubling since the 1980s (high confidence), and their intensity and duration have also increased 9 (medium confidence) (Box 9.2). Projections show that this increasing trend likely continues with 2-9 times 10 more frequent marine heatwaves (at global scale) projected by 2081-2100, relative to 1995-2014 under 11 SSP1-2.6, and 4-18 times more frequent under SSP5-8.5. The largest changes in MHW frequency likely to 12 occur in the tropical ocean and the Arctic, while there is medium confidence of moderate increases in the 13 mid-latitudes, and of small increases in the Southern Ocean (Box 9.2). Permanent MHWs (more than 360 14 days per year, relative to the historical climate conditions) are projected to occur in the 21st century in parts 15 of the tropical ocean, in the Arctic Ocean and around latitude 45°S, under SSP5-8.5 (Box 9.2). The 16 occurrence of such permanent MHWs can be largely avoided under SSP1-2.6 scenario (Box 9.2). MHW can 17 have devastating and long-term impacts on ecosystems (Oliver et al., 2018), making them an emerging 18 hazard for marine ecosystems (Frölicher and Laufkötter, 2018; Smale et al., 2019). A series of marine 19 heatwaves that occurred in 2010-2011 had consequences for seagrass in Western Australia (Wernberg et al., 20 2013; Arias-Ortiz et al., 2018), and for loster fishery in the Gulf of Maine (Pershing et al., 2018). The 21 MHWs that occured Western Australia in 2015/2016 lead to the third-highest mass bleaching globally (Le 22 Nohaïc et al., 2017). 23 24 Ocean acidity: With the increasing CO2 concentration, the global mean ocean surface pH is decreasing and 25 is now the lowest it has been for at least a thousand years (very high confidence) (Section 2.3.3.5). It is very 26 likely that, since the 1980s, ocean surface pH has changed at a rate of –0.016 to –0.019 per decade in the 27 subtropical open oceans, at –0.010 to –0.026 per decade in the tropical Pacific, and at –0.003 to –0.026 per 28 decade in open subpolar and polar zones (Section 2.3.3.5; Section 5.3.3.2). Over the period 1870–1899 to 29 2080–2099, ocean surface pH is projected to decline by -0.16 ± 0.002 under SSP1–2.6, and by -0.44 ± 0.005 30 under SSP5–8.5 (Section 4.3.2.5; Section 5.3.4.1). Declining ocean pH will exacerbate negative impacts on 31 marine species (Albright et al., 2016; Kwiatkowski et al., 2016; Watson et al., 2017) (medium confidence). 32 33 Ocean salinity: Salinity contrasts have increased since the 1950s, near the ocean surface (virtually certain) 34 and in the sub-surface (very likely), with high salinity regions becoming more saline and low salinity regions 35 becoming fresher (Section 2.3.3.2). At the basin scale, it is very likely that the Pacific and the Southern 36 Oceans have freshened, and the Atlantic has become more saline (Section 2.3.3.2). The IPCC (2019) 37 assessment that fresh ocean regions get fresher and salty ocean regions get saltier will continue in the 21st 38 century is confirmed in Section 9.2.2.2. 39 40 At the regional scale, by 2100, the average Arctic surface salinity is projected to decrease by 1.5 ± 1.1 psu, 41 and the liquid freshwater column in the Arctic Ocean is projected increase by 5.4 ± 3.8 m under RCP8.5, 42 (Shu et al., 2018). In the Indian Ocean, sea surface salinity is projected to decrease by 0.49 psu and 0.75 psu 43 by 2080, compared to 2015, under RCP2.6 and RCP2.6, respectively (Akhiljith et al., 2019). Projections for 44 the North and South Atlantic Oceans indicate increasing salinity in the upper layer (0 – 500 m) under both 45 RCP4.5 and RCP8.5, due to the decreasing freshwater input from the Equator and increasing net evaporation 46 (Skliris et al., 2020). There is medium confidence that fresh ocean regions (Pacific, Southern and Indian 47 oceans) will get fresher and salty ocean regions (Atlantic ocean) will get saltier over the 21st century (IPCC, 48 2019; Section 9.2.2.2). Ocean warming and high-latitude surface freshening is projected to continue to 49 increase upper ocean stratification in the 21st century (Section 9.2.1.3). 50 51 Dissolved oxygen: Since the middle of the last century, oxygen concentrations of open and coastal waters 52 have been declining, and such deoxygenation affects biological and biogeochemical processes in the ocean 53 (Schmidtko et al., 2017). In recent decades, low oxygen zones in ocean ecosystems have expanded, and 54 projections indicate an acceleration with global warming (Diaz and Rosenberg, 2008; Gilly et al., 2013; 55 Gobler et al., 2014) (medium confidence). A 2% loss (4.8 ± 2.1 Pmoles O2) in total dissolved oxygen in the Do Not Cite, Quote or Distribute 12-94 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 upper ocean layer (100–600 m) has been observed during 1970–2010 (Section 2.3.4.2), with the highest 2 oxygen loss of up to 30 mol m-2 per decade in the equatorial and North Pacific, the Southern Ocean and the 3 South Atlantic Ocean (Section 5.3.3.2). Global mean ocean oxygen concentration is projected to decrease by 4 6.36 ± 2.92 mmol m-3 under SSP1-2.6 and by 13.27 ± 5.28 mmol m-3 under SSP5-8.5 in the subsurface (100- 5 600 m) by 2080–2099, compared to 1870–1899, which is respectively 71% and 40% greater than previous 6 estimates based on CMIP5 models (Section 5.3.3.2). In the benthic ocean, projected future losses of 7 dissolved oxygen concentration by 2080–2099, compared to 1870– 1899, are −5.14 ± 2.04 mmol m−3 under 8 SSP1-2.6 and −6.04 ± 2.19 mmol m−3 under SSP5-8.5 (Kwiatkowski et al., 2020). Section 5.3.3.2 assessed 9 very likely global decreases in ocean oxygen concentrations although there is medium confidence in specific 10 regional declines that are expected to expand both anoxic and hypoxic zones, with such reductions of oxygen 11 expected to persist for thousands of years (Yamamoto et al., 2015; Frölicher et al., 2020). 12 13 Sea ice: The Arctic sea ice area for September has decreased from 6.23 to 3.76 million km2 and for March 14 from 14.52 to 13.42 million km2 between 1979-1988 and 2010-19 (Section 2.3.2.1.1). There is high 15 confidence that sea ice in the Arctic will further decrease in the future under all emission scenarios (Section 16 9.3.1.1). In contrast, there is no clear observed trend in the Antarctic sea ice area over the past few decades 17 and there is low confidence of future changes therein (Section 9.3.1.1). The duration of the summer season in 18 the Arctic has increased by 5 to 20 weeks between 1979 and 2013, with a significant trend ranging from 5 to 19 17 days/decade for earlier spring retreat and from 5 to 25 days/decade for later fall advance, with 20 consequences for Arctic marine mammals (AMMs) due to sea ice habitat loss (Laidre et al., 2015). The 21 Arctic is projected to be ice-free more often during summer under 2°C global warming compared to 1.5°C 22 global warming (Section 9.3.1.1; see also Section 12.4.9 and Section 4.4.2.1), opening new shipping lanes 23 for international commerce (Valsson and Ulfarsson, 2011) and lengthening the season for offshore resource 24 extraction (Schaeffer et al., 2012). Iceberg numbers are expected to increase as a result of global warming, 25 forming an elevated hazard to shipping and offshore facilities (Bigg et al., 2018). 26 27 28 It is virtually certain that global mean SST will continue to increase throughout the 21st century, 29 resulting in the exceedance of numerous hazard thresholds relevant to marine ecosystems (medium 30 confidence). Marine heatwaves day are projected to increase in global oceans with a larger increase in 31 the tropical ocean and Arctic Ocean (high confidence). It is virtually certain that upper ocean 32 stratification will continue to increase in the 21st century. Future Ocean warming will very likely assist 33 the development of both anoxic and hypoxic zones, with such reductions of oxygen expected to persist 34 for thousands of years. Future projections also indicate freshening of the Pacific, Southern and Indian 35 Oceans and a saltier Atlantic Ocean (medium confidence). 36 37 The assessed direction of change in climatic impact-drivers for open and deep ocean regions and associated 38 confidence levels are illustrated in Table 12.10, following the AR6 WGI ocean reference regions as proposed 39 by the Atlas (Figure Atlas.2b). 40 41 42 [START TABLE 12.10 HERE] 43 44 Table 12.10: Summary of confidence in direction of projected change in climatic impact-drivers in open and deep 45 ocean regions, representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, 46 SSP3-4.5, SRES A1B, or above within each AR6 region (defined in Chapter 1), approximately 47 corresponding (for CIDs that are independent of sea-level rise) to global warming levels between 2°C and 48 2.4°C (see 12.4 for more details of the assessment method). The table also includes the assessment of 49 observed or projected time-of-emergence of the CID change signal from the natural inter-annual 50 variability if found with at least medium confidence in Section 12.5.2. 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-95 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Mean Ocean Temperature Marine Heatwave Dissolved Oxygen Ocean acidity Salinity Sea Ice Region Arctic Ocean (ARO) South Pacific Ocean (SPO) Equatorial Pacific Ocean (EPO) North Pacific Ocean (NPO) South Atlantic Ocean (SAO) Equatorial Atlantic Ocean (EAO) North Atlantic Ocean (NAO) Equatorial Indian Ocean (EIO) South Indian Ocean (SIO) Arabian Sea (ARS) Bay of Bengal (BOB) Southern Ocean (SOO) 1 Key High confidence of decrease Medium confidence of decrease Low confidence in direction of change Medium confidence of increase High confidence of increase 2 Not broadly relevant 3 4 [END TABLE 12.10 HERE] 5 6 7 12.4.9 Polar terrestrial regions 8 9 Several recent climate assessments on polar regions describe robust patterns of recent and future climatic 10 changes driving impacts and risk for polar environmental, societal, and economic assets, including the IPCC 11 SROCC (Meredith et al., 2019), the Report on Snow, Water, Ice and Permafrost in the Arctic (AMAP, 12 2017b), and national assessments for the United States (Markon et al., 2018) and Canada (Derksen et al., 13 2018). This section examines Greenland and Iceland, the Russian Arctic, Antarctica, and the arctic portions 14 of Northern Europe and North America (Figure 1.18c). 15 16 17 12.4.9.1 Heat and cold 18 19 Mean air temperature: Atlas.11.2 shows high confidence in warming of the Arctic in observations and 20 projections, measuring among the fastest-warming places at more than twice the global mean, with 21 substantially higher temperature increases in the cold season (see also AMAP, 2017; Meredith et al., 2019). 22 Atlas.11.1 assessed very likely warming in observations of West Antarctica from 1957-2016, but limited 23 evidence of mean air temperature change across East Antarctica even as there is high confidence in future 24 warming across the continent (Meredith et al., 2019) (Figures Atlas.32, Atlas.33). 25 26 Extreme heat, cold spell and frost: Ecosystem and societal temperature thresholds in polar regions often 27 reflect lower tolerance to heat and higher tolerance to cold. Extreme heat events have increased around the 28 Arctic and Iceland since 1979, including increases in cold season warm days and nights, melt days, and 29 Arctic winter warm events (T>-10℃) as well as decreases in cold days and nights (Mernild et al., 2014; 30 Matthes et al., 2015; Vikhamar-Schuler et al., 2016; Graham et al., 2017; Sui et al., 2017; Dobricic et al., 31 2020; Peña-Angulo et al., 2020). Heatwaves causing high temperature records have been recently 32 documented in West and East Antarctica (Wille et al., 2019; Robinson et al., 2020). There is high confidence 33 that polar amplification will drive increases in Arctic heat extremes as well as continuing declines in the Do Not Cite, Quote or Distribute 12-96 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 magnitude and frequency of cold extremes (Matthes et al., 2015; Kharin et al., 2018), although dynamical 2 effects will still bring substantial cold air anomalies over the Arctic (Wu and Francis, 2019). There is 3 medium confidence for equivalent changes in extreme heat in Antarctica based primarily on higher mean 4 temperatures, with Lee et al. (2017) projecting more than 50 additional degree days above freezing (2098 5 RCP8.5 vs. 2014) over parts of the Antarctic Peninsula but smaller changes over mainland Antarctica. 6 7 8 12.4.9.2 Wet and dry 9 10 Mean precipitation: Atlas.11.2 indicated medium confidence in observed increases in Arctic precipitation 11 with the largest rises in the cold season. Antarctic precipitation showed no significant overall trend since the 12 1970s, with a positive trend over the 20th century (Atlas.11.1, 9.4.2.1). Increases in Arctic and Antarctic 13 precipitation during the 21st century are very likely with projected percentage increases that are much higher 14 than most sub-polar regions of the world (Figure Atlas.32). 15 16 Floods and heavy precipitation: Observations and model projections indicate high confidence in 17 increasing Arctic river runoff in response to increasing total precipitation (Box et al., 2019; Durocher et al., 18 2019; Meredith et al., 2019) with a shift toward earlier meltwater flooding (AMAP, 2017b). Higher Arctic 19 precipitable water totals are also connected with observed increases in heavy precipitation and convective 20 activity (high confidence) (Ye et al., 2015; Kharin et al., 2018; Chernokulsky et al., 2019). Higher flood 21 magnitudes are also driven by future increases in rain-on-snow event days, amounts, and runoff, which are 22 more significant in the Arctic than in mid-latitudes (where seasonal snow cover is often further reduced) 23 (AMAP, 2017b; Il Jeong and Sushama, 2018). 24 25 Landslide and snow avalanche: There is a growing number of studies on mass movements in polar 26 regions. Although there is low confidence in widespread observational trends for landslides or snow 27 avalanches, a rise in the number of future landslides is supported by strong links to increases in heavy 28 precipitation, glacier retreat, and thawing of ice-rich permafrost that can lead to retrogressive thaw slumps in 29 Arctic regions (Kokelj et al., 2015; Derksen et al., 2018; Lewkowicz and Way, 2019; Patton et al., 2019; 30 Ward Jones et al., 2019) (Section 2.3.2.5). 31 32 Aridity and drought: Recent decades have seen a general decrease in Arctic aridity with projections 33 indicating a continuing trend toward reduced aridity (high confidence) as increased moisture transport leads 34 to higher precipitation, humidity and streamflow (Meredith et al., 2019) and a corresponding decrease in dry 35 days (Khlebnikova et al., 2019b). There is low confidence overall of recent or projected drought changes in 36 polar regions (Section 11.9) even as increasing evidence shows that drainage from permafrost thaw, higher 37 potential evapotranspiration, and changing seasonal patterns of melt have caused lake reduction and soil 38 moisture deficits in several areas that match with projections of future drought increase despite overall 39 precipitation increases (Andresen and Lougheed, 2015; Bring et al., 2016; Spinoni et al., 2018a; Feng et al., 40 2019; Finger Higgens et al., 2019). 41 42 Fire weather: Fire season lengthened from 1979-2015 over Arctic portions of North America (Jain et al., 43 2017), corresponding also to a 1975-2015 increase in lightning-ignited fires in Northwestern North America 44 (Girardin et al., 2013; Veraverbeke et al., 2017). Abatzoglou et al. (2019) climate model simulations project 45 significant fire weather index increases in boreal forests of Arctic Europe, Arctic Russia and Northeast North 46 America (medium confidence). Trends toward more frequent fires in tundra regions are expected to continue, 47 driven in particular by increasing potential evapotranspiration and changes in vegetation (high confidence) 48 (Hu et al., 2015; AMAP, 2017b; Young et al., 2017). 49 50 51 12.4.9.3 Wind 52 53 Mean wind speed and severe storm: There is medium confidence of mean wind decrease over the Russian 54 Arctic, Greenland and Iceland and Arctic Northeast North America (Karnauskas et al., 2018a; Jung and 55 Schindler, 2019), but low confidence of changes in the other Arctic regions and Antarctica. Bintanja et al. Do Not Cite, Quote or Distribute 12-97 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 (2014) projected that a strengthening of the Southern Annular Mode would decrease easterlies along 2 Antarctica’s coasts with only small changes in katabatic winds (although this effect may diminish with 3 stratospheric ozone recovery). In contrast, Gorter et al. (2014) regional climate model projections indicated a 4 reduction in mean winds over the interior of Greenland by RCP4.5 2100 while coastal winds increase. 5 Reanalysis data and climate models indicate few coherent regional trends of polar cyclone frequency or 6 relationships with cyclone depth and size (Akperov et al., 2018, 2019; Day and Hodges, 2018; Zahn et al., 7 2018). 8 9 10 12.4.9.4 Snow and ice 11 12 Snow: Atlas.11.1 identified likely increases in surface mass balance (driven by snowfall) across Antarctica in 13 the 20th century (medium confidence). In the Arctic, overall snow extent and seasonal duration are projected 14 to continue recent declines (high confidence), although mid-winter snowpack increases in some of the 15 coldest and high-elevation locations given higher precipitation totals (medium confidence) (Bring et al., 16 2016; Danco et al., 2016; AMAP, 2017; Meredith et al., 2019) (9.5.3; Atlas.11.2; Atlas.9). Higher 17 temperatures result in a higher percentage of Arctic precipitation falling as rain (particularly in fall and 18 spring) (high confidence), with most land regions (outside of Greenland and Antarctica) becoming 19 dominated by rainfall (more than 50% of total precipitation) by RCP8.5 2100 (Bintanja and Andry, 2017; 20 Irannezhad et al., 2017). 21 22 Glacier and ice sheet: Section 9.5.1 and Section 2.3.2.3 found that glaciers have lost mass in all polar 23 regions since 2000 (high confidence), and Section 9.4 assessed high confidence in Greenland ice sheet mass 24 losses since 1980 and Antarctic Ice Sheet losses since 1992 (dominated by West Antarctica with losses in 25 parts of East Antarctica in the past two decades). New simulations from GlacierMIP (Marzeion et al., 2020) 26 indicate glaciers in Iceland will lose 31 ± 35%, 41 ± 46% and 53 ± 45% of their mass in 2015 by the end of 27 the century for RCP2.6, RCP4.5 and RCP8.5 scenarios, respectively. Marzeion et al. (2020) projected mass 28 losses (high confidence) for those same scenarios in the Greenland Periphery: 22 ± 23%, 29 ± 26%, and 42 ± 29 28%; Svalbard: 35 ± 34%, 50 ± 36%, and 66 ± 35%; Russian Arctic: 26 ± 26%, 38 ± 28%, and 52 ± 30%; 30 Northern Arctic Canada: 12 ± 13%, 18 ± 12%, and 27 ± 18%; Southern Arctic Canada: 23 ± 27%, 33 ± 29%, 31 and 48 ± 32%; and Antarctic Periphery: 7 ± 12%, 13 ± 10%, and 16 ± 19%. Areas with receding glaciers are 32 also potentially vulnerable to glacial lake outburst floods (Harrison et al., 2018). 33 34 Permafrost: Observations from recent decades (assessed in Section 9.5.2 and Section 2.3.2.5) show 35 increases in permafrost temperature (very high confidence) and active layer thickness (medium confidence) 36 across the Arctic (AMAP, 2017; Markon et al., 2018; Biskaborn et al., 2019; Derksen et al., 2019; 37 Farquharson et al., 2019; Meredith et al., 2019; Romanovsky et al., 2020). Section 9.5.2 noted that 38 observations of active layer thickness in Antarctica are too limited to assess long-term trends (see also 39 Biskaborn et al., 2019; Hrbáček et al., 2018). Future projections indicate continuing increases in permafrost 40 temperature and active layer thickness with loss of permafrost across the Arctic (Section 9.5.2). Streletskiy et 41 al. (2019) noted that changes to Russian permafrost temperature and active layer thickness are most 42 pronounced in areas where permafrost is continuous (underlying >90% of landmass). CMIP5 analyses by 43 Slater and Lawrence (2013) projected that, by RCP8.5 2100, shallow (<3 m) permafrost would be most 44 probable only in portions of the Canadian Arctic Archipelago and the Russian Arctic coastal and eastern 45 upland regions. 46 47 Sea ice: Consistent with the SROCC (Meredith et al., 2019), Section 9.3.1 and Section 2.3.2.1.1 assess very 48 high confidence that Arctic sea ice thickness, extent, and average age have significantly decreased over the 49 past four decades with largest declines in September (when sea ice is at an annual minimum). Declines in 50 landfast ice are most rapid in the Laptev Sea (Selyuzhenok et al., 2015), are breaking perennial landfast ice 51 blocking ocean channels (‘ice plugs’) in the Canadian Archipelago (Pope et al., 2017), and declining in the 52 cold season by 7% decade-1 across the Arctic (1976-2007) (Yu et al., 2014). Observed trends and projections 53 suggest that perennial sea ice is being replaced by thin, seasonal ice, although multi-year ice will persist 54 above the Canadian Archipelago and drift into sea transportation lanes (Howell et al., 2016; Derksen et al., 55 2018). Trends from 1979-2013 show slightly earlier spring melt for Arctic sea ice, but substantially-delayed Do Not Cite, Quote or Distribute 12-98 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 fall freeze up and a melt season lengthened by more than 3 days decade-1 off northern Alaska and Canada 2 with the exception of portions of the Bering Sea (Parkinson, 2014; Stroeve et al., 2014). Section 9.3.2 3 assessed low confidence in long-term trends in sea ice extent or thickness near Antarctica. 4 5 Future declines in Arctic sea ice are virtually certain, although there is low confidence in declines of 6 Antarctic sea ice given dynamical processes in the Southern Ocean and the recovery of stratospheric ozone 7 (Section 9.3; Meredith et al., 2019). Projections of an ‘ice-free’ Arctic vary depending on definitions 8 representing transportation needs, however Laliberté et al. (2016) noted that the median of 42 CMIP5 models 9 projected <5% sea ice for the month of September by 2050, with equivalent conditions for the entirety of the 10 August-October period by 2090. Section 9.3.1 assessed high confidence that practically ice-free conditions 11 (<1 million km2 in the September mean) would likely first appear before 2050 even under strong mitigation 12 scenarios (Sigmond et al., 2018; Stroeve and Notz, 2018; Notz and SIMIP, 2020). 13 14 Lake and river ice: There is high confidence in observations of significant declines in seasonal ice cover 15 thickness and duration over most Arctic lakes, with many lakes projected to lose more than month of ice 16 cover by mid-century (medium confidence) (Meredith et al., 2019; Sharma et al., 2019). Some lakes that 17 previously froze to the bottom (‘bedfast’) now maintain liquid bottom water year round, and others shift 18 from perennial to seasonal ice cover (Surdu et al., 2016; Engram et al., 2018). Yang et al. (2020) identified a 19 decline in Arctic cold-season river ice extent in satellite observations (particularly in Alaska) and projected 20 reductions in average Northern Hemisphere seasonal river ice duration of 6.10 ± 0.08 days per degree global 21 surface air temperature. 22 23 Heavy snowfall and ice storm: There is limited evidence of changes in heavy snowfall due to competing 24 influences of shortened snowfall seasonality with more intense (and larger overall) precipitation in the 25 Arctic. Episodic heavy snowfall trends in Antarctica are difficult to separate from large interannual 26 variability (limited evidence) (Gorodetskaya et al., 2014, Turner et al., 2020). Limited evidence also hinders 27 clear signals in ice storms, although warming shifts the freezing line (around which ice storms occur) 28 poleward and upslope (Bintanja and Andry, 2017). Groisman et al. (2016) used 40 years of observations to 29 identify an increase of freezing rain events in Norway, North America, and Eastern and Western Russia. 30 Increases in wintertime rainfall have led to more frequent development of difficult wildlife and livestock 31 grazing conditions as basal ice conditions coat the ground below snowpack (Peeters et al., 2019). 32 33 34 12.4.9.5 Coastal and oceanic 35 36 Relative sea level: Satellite altimetry and tide data show that relative sea levels (with glacial isostatic 37 adjustment) are rising in Arctic Europe and Northwest North America, declining in portions of Southern 38 Alaska and Arctic Northeast North America and no clear trend in Greenland and Arctic Russia (Sweet et al., 39 2018; Rose et al., 2019), which is broadly consistent with findings in Oppenheimer et al. (2019). Areas with 40 low or negative change have substantial land uplift counteracting the global mean sea level trend (Greenan et 41 al., 2018; Sweet et al., 2018; Madsen et al., 2019). SROCC projections indicate high confidence in future 42 rises in relative sea level for all Arctic regions other than areas of substantial land uplift in Northeastern 43 Canada, the west coast of Greenland, and narrow portions of West Antarctica (Oppenheimer et al., 2019). 44 45 Coastal flooding and erosion: Higher sea levels and reduced coastal sea ice protection will increase future 46 extreme sea levels in the Arctic (high confidence for Arcitc NEU, RAR, and Arctic NWN; medium 47 confidence for GIC and Arctic NEN given glacial isostatic adjustment). Vousdoukas et al. (2018) project 48 that the current 1:100 yr extreme total water level would have median return periods of 1:20 yrs – 1:50 yrs 49 by 2050, increasing to 1:5 yrs – 1:20 yrs by 2100 under RCP4.5 along nearly the entire Arctic coastline by 50 2100 (excluding GIC for which projections are not available). Projections for RCP8.5 indicate that the 51 present day 1:100 yr ETWL would have median return periods of 1:10 yrs – 1:50 yrs 2050 and would occur 52 once every 5 years (or more frequently). Arctic coastal erosion is also expected to increase with climate 53 change (medium confidence; high agreement but limited evidence of projections), accelerated in some 54 regions by subsurface permafrost thaw and increased wave energy (Gibbs and Richmond, 2015; Fritz et al., 55 2017; Oppenheimer et al., 2019; Casas-Prat and Wang, 2020). A longer ice-free season for the RCP8.5 Do Not Cite, Quote or Distribute 12-99 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 2080s is projected to help drive more than 100 m of shoreline retreat in Northwest North America Arctic 2 coastal communities (Melvin et al., 2017; Greenan et al., 2019; Magnan et al., 2019). Assessment of coastal 3 flooding and erosion changes in Antarctica are limited by a lack of studies. 4 5 Marine heatwave: Recent years have seen marine heatwaves and increasing extreme coastal SSTs in Arctic 6 systems (Lima and Wethey, 2012; Collins et al., 2019; Frölicher, 2019). Projections show increases in 7 marine heatwave intensity, frequency, and duration will be larger over the Arctic Ocean than mid-latitude 8 oceans due in part to low interannual variability under current sea ice (high confidence). Frölicher et al. 9 (2018) used 12 CMIP5 models to project median MHW days increasing about 25-fold and 50-fold at the 2℃ 10 and 3.5℃ GWLs in response to mean ocean warming and sea ice loss, and the smallest global changes still 11 being increases in the Southern Ocean around Antarctica (see also Cross-chapter Box 9.1). 12 13 Climate change has caused and will continue to induce enhanced warming trend, increasing heat- 14 related extremes and decreasing cold spells and frosts in the Arctic (high confidence), with similar 15 changes in Antarctica but medium confidence for extreme heat increases and West Antarctic frost 16 change decreases and low confidence for cold spell changes and East Antarctica frost. The water cycle 17 is projected to intensify in polar regions, leading to more rainfall, higher river flood potential and 18 more intense precipitation (high confidence). Projections indicate reductions in glaciers and sea ice at 19 both poles, with enhanced permafrost warming, decreasing permafrost extent, and decreasing 20 seasonal duration and extent of snow cover in the Arctic (high confidence) even as some of the coldest 21 regions will see higher total snowfall given increased precipitation (medium confidence). Projections 22 indicate relative sea level rises in Polar regions (high confidence), with the exception of regions with 23 substantial land uplift including Northeast North America (high confidence), Western Greenland, the 24 Northern Baltic Sea, and portions of West Antarctica. Higher sea levels also contribute to high 25 confidence for projected increases of Arctic coastal flooding and higher coastal erosion (aided by sea 26 ice loss) (medium confidence) with lower confidence for those CIDs in regions with substantial land 27 uplift (Arctic Northeast North America and Greenland). 28 29 30 [START TABLE 12.11 HERE] 31 32 Table 12.11: Summary of confidence in direction of projected change in climatic impact-drivers in the polar regions, 33 representing their aggregate characteristic changes for mid-century for scenarios RCP4.5, SSP3-4.5, 34 SRES A1B, or above within each AR6 region (defined in Chapter 1), approximately corresponding (for 35 CIDs that are independent of sea-level rise) to global warming levels between 2°C and 2.4°C (see 12.4 for 36 more details of the assessment method). The table also includes the assessment of observed or projected 37 time-of-emergence of the CID change signal from the natural inter-annual variability if found with at 38 least medium confidence in Section 12.5.2. Note that the Arctic portions of the NEU, NEN, and NWN 39 differ from the full AR6 regions assessed in the Europe and North America sections above (see also 40 Figure 1.18c). 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 Do Not Cite, Quote or Distribute 12-100 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Climatic Impact-Driver Heat and Cold Wet and Dry Wind Snow and Ice Coastal & Oceanic Other Heavy precipitation and pluvial flood Agricultural and ecological drought Heavy snowfall and ice storm Atmospheric CO2 at surface Snow, glacier and ice sheet Mean air temperature Lake, river and sea ice Air pollution weather Hydrological drought Sand and dust storm Radiation at surface Mean precipitation Severe wind storm Mean wind speed Marine heatwave Relative sea level Snow avalanche Tropical cyclone Coastal erosion Ocean acidity Extreme heat Coastal flood Fire weather Permafrost River flood Cold spell Landslide Aridity Frost Hail Region Greenland and Iceland (GIC) 2,3 1 5 Arctic Northern Europe (aNEU) 2,3 1 6 7 Russian Arctic (RAR) 2,3 1,4 7 Arctic Northwest North America (aNWN) 2,3 1 7 Arctic Northeast North America (aNEN) 2,3 1,4 West Antarctica (WAN) 1,4 East Antarctica (EAN) 1. Snow may increase in some high elevations and during the cold season and decrease in other seasons and at lower elevations 2. Higher confidence in southern regions and lower toward north 3. Higher confidence in increase for some climatic impact-driver indices during summertime Key 4. Glaciers decline even as some regional snow climatic impact-driver indices increase High confidence of decrease 5. Decreasing in west and increasing in east Medium confidence of decrease 6. Except for Northern Baltic Sea coasts where relative sea levels fall Low confidence in direction of change 7. Along sandy coasts and in the absence of additional sediment sinks/sources or any physical barriers to shoreline retreat. Medium confidence of increase High confidence of increase Not broadly relevant 1 2 [END TABLE 12.11 HERE] 3 4 5 12.4.10 Specific zones and hotspots 6 7 This section focuses on CIDs affecting specific zones heightened vulnerability and coherent characteristics 8 that cut across traditional continental regions (see also Section 12.3). It is designed to match the structure of 9 the cross-chapter papers in the WGII report, although polar regions were addressed in more extensive detail 10 in 12.4.8 and 12.4.9 and the Mediterranean Region will not be handled separately given that its climatic 11 impact-drivers are discussed in Sections 12.4.1 and 12.4.5 as well as in Cross-Chapter Box 10.3. 12 13 14 12.4.10.1 Hotspots of biodiversity (land, coasts and oceans) 15 16 Hotspots of biodiversity are defined by the AR6 WGII as “geographic areas with exceptionally high richness 17 of species, including rare (endemic) species” (WGII Cross-Chapter Paper 1). The AR6 assessment is based 18 on 238 distinctive regions often called the “Global 200 ecoregions” (Olson and Dinerstein, 2002). 19 20 Mean temperature increase is a major climatic impact-driver for biodiversity hotspots, and it is very likely 21 that it will affect all hotspot areas identified in the literature, at various rates in all climate scenarios, except 22 those located in the North Atlantic where warming is uncertain (see Chapter 4). Terrestrial ecosystems will 23 experience an enhanced warming compared to ocean ecosystems, because land temperatures are warming 24 faster than ocean temperatures (Chapter 4). Marine ecoregions will experience ocean acidification and 25 temperatures that increase faster in high latitudes (high confidence), but critical temperature and oxygen 26 thresholds are projected to be crossed earlier (by mid-century RCP8.5) in tropical areas (Hughes et al., 27 2017a; Bruno et al., 2018). A warming trend is also expected for freshwater ecosystems, with different local 28 magnitudes due to combined effects of groundwater system inertia as well as hydrology changes (Knouft and 29 Ficklin, 2017). In tropical land areas, because interannual temperature variability is weak compared to 30 changes, the temperature distribution range is likely to be shifted to a very different range in all projection 31 scenarios, with unprecedented values relative to preindustrial. High climate velocities are particularly Do Not Cite, Quote or Distribute 12-101 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 noteworthy for biodiversity hotspots given complex ecosystem dynamics and niche climates not easily 2 replicated under shifted geographies (Burrows et al., 2014; Halpern et al., 2015; Dobrowski and Parks, 3 2016). In some regions (Central Africa, Amazon, South East Asia) the mean temperature change is already 4 beyond the normal range of variations as it has reached levels higher than 3 (and up to 6) times larger than 5 the standard deviation of the interannual variations (Hawkins et al., 2020). Together with global warming, 6 land and marine heat waves are very likely to increase in the future climate in biodiversity hotspots (Sections 7 12.4.1-12.4.7). 8 9 There is low confidence in broad patterns of future drying or wet trends across the land and freshwater 10 biodiversity hotspots from the humid tropics, although drying trends have been detected and predicted in 11 parts of the Amazon (Fu et al., 2013; Boisier et al., 2015). There is medium confidence (limited evidence, 12 high agreement) that in several regions the length of the dry season has already increased and is projected to 13 further increase in some parts of the Mediterranean, Amazonia and sub-Saharan Africa (S. Debortoli et al., 14 2015; Dunning et al., 2018; Hochman et al., 2018; Saeed et al., 2018). Longer dry seasons also extend the 15 seasonal length and geographical extent of fire weather in all future scenarios (medium confidence) (Jolly et 16 al., 2015; Abatzoglou et al., 2019). 17 18 In conclusion, biodiversity hotspots located around the world will each face unique challenges in 19 climatic impact-drivers changes. However, heat, drought and length of dry season, wildfire weather, 20 sea surface temperature and deoxygenation are relevant drivers to terrestrial and freshwater 21 ecosystems, and have marked increasing trends. 22 23 24 12.4.10.2 Cities and settlements by the sea 25 26 Cities and settlements (C&S) by the sea are exposed to specific climate and climate change patterns and to 27 compound coastal hazard risks (AR6 WGII Cross-chapter paper 2). AR5 WGII found that, in general, “urban 28 climate change-related risks are increasing (including rising sea levels and storm surges, heat stress, extreme 29 precipitation, inland and coastal flooding, landslides, drought, increased aridity, water scarcity, and air 30 pollution)”. Since AR5 a number of studies have been carried out to understand urban climate and its 31 change. Box 10.3 identified a continuing strong role of the urban heat island in amplifying heat extremes in 32 cities, although changes in the urban heat island are an order of magnitude smaller than projected localized 33 warming trends (very high confidence). 34 35 Coastal cities’ proximity to the sea somewhat mitigates the effect of urban heat islands (high confidence) 36 (Salvati et al., 2017; Santamouris et al., 2017; Wang et al., 2018b; Martinelli et al., 2020). Cities and 37 settlements by the sea typically experience higher humidity levels than inland regions, combining with heat 38 to enhance heat stress and induce exceedance of critical heat stress thresholds for outdoor activities, with 39 potential enhanced exposure to heat for informal settlements (Wang et al., 2019b). Such threshold 40 exceedances are projected to increase for many coastal areas (high confidence), including the Persian Gulf 41 where heat stress is projected to be extreme (Pal and Eltahir, 2016; Ahmadalipour and Moradkhani, 2018), 42 and some low-lying areas in Europe such as the Po Valley and coastal Mediterranean areas (Coppola et al., 43 2021a) (Schwingshackl et al., 2021) (see also the heat stress index shown in Figure 12.4d-f). 44 45 Climate change related variations in oceanic drivers (e.g., relative sea level, storm surge, ocean waves), 46 combined with tropical cyclones, extreme precipitation and river flooding, are expected to lead to more 47 frequent and more intense coastal flooding and erosion (very high confidence) impacting C&S located 48 especially in low elevation coastal zones and mega-deltas (Chan et al., 2012, 2018; Karymbalis et al., 2012; 49 Hemer et al., 2013; Aerts et al., 2014; Neumann et al., 2015a; Ranasinghe, 2016; Hauer et al., 2016; Hinkel 50 et al., 2018; Mavromatidi et al., 2018; Marcos et al., 2019) (See also Sections 12.3, 12.4.1-12.4.7 and 51 12.4.9). Coastal erosion and flooding also pose challenges to critical infrastructure such as roads, subway 52 tunnels, electricity and phone networks, wastewater management plants and buildings (Grahn and Nyberg, 53 2017; Pregnolato et al., 2017). Compound flooding due to simultaneous storm surges and high river flows 54 have been found to be increasingly frequent in several cities and/or low-lying areas in Europe and the U.S.A. 55 (Wahl et al., 2015; Paprotny et al., 2018; Bevacqua et al., 2019; Ganguli and Merz, 2019) (high confidence). Do Not Cite, Quote or Distribute 12-102 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Chapter 11 found that the frequency of such compound flood events is projected to increase (high 2 confidence). In addition to sea level rise induced changes, many C&S by the sea are in regions where tropical 3 cyclones are projected to become more intense and frequent (high confidence) (Section 11.7). 4 5 The SROCC report highlighted coastal settlements in the Arctic as being particularly exposed to several CID 6 changes (Magnan et al., 2019). Enhanced waves due to extended season of sea ice retreat are projected to 7 foster coastal flooding and erosion (Gudmestad, 2018; Casas-Prat and Wang, 2020) (Section 12.4.9). Climate 8 change is also affecting sea ice quality and season length along coasts of the Arctic ocean where populations 9 depend on sea ice for hunting or transportation (Pearce et al., 2015) (Section 12.4.9). 10 11 In summary, coastal cities and settlements are particularly affected by a number of climatic impact- 12 drivers that have already changed and will continue to change whatever the emission scenario. These 13 include increases in extreme heat, pluvial floods, coastal erosion and coastal flood (high confidence). 14 Increasing relative sea level compounding with increasing tropical cyclone storm surge and rainfall 15 intensity will increase the probability of coastal city flooding (high confidence). Arctic coastal 16 settlements are particularly exposed to climate change due to sea ice retreat (high confidence). 17 18 19 12.4.10.3 Deserts and semi-arid areas 20 21 Drylands, which include hyper arid, arid, semi-arid and dry sub-humid areas (IPCC, 2019c), lie on all 22 continents and cover 46% of the global land area and host over a third of the current population (Olsson et 23 al., 2019) . Huang et al. (2016) found that aridity changes have helped expand dryland area by ~4% from 24 1948 to 2004, with the largest expansion of drylands occurring in semi-arid regions since the early 1960s. 25 Section 4.5.1 assessed high confidence of a future poleward expansion of the Hadley cell, leading to a 26 poleward shift of dryland areas in all scenarios considered. There is no evidence of a future global trend in 27 aridification of drylands (IPCC, 2019a), but high confidence of aridification in some areas (Mediterranean, 28 Central America, South Africa) (IPCC, 2019a) (see also Figure 12.4j-l). However, drivers of desertification 29 largely include land cover changes and land use management, along with climate change (IPCC, 2019a). 30 31 Warming temperatures and extreme heat are major climatic impact-drivers with multiple potential impacts 32 on societies, health, and habitability in semi-arid and arid regions that are already near physiological limits 33 for outdoor activities. Semi-arid regions will very likely undergo a warming in all future scenarios (Chapter 34 4; Atlas) and likely undergo an increase in duration, magnitude and frequency of heatwaves (Chapter 11) 35 (see also Figure 12.4a-c). It is likely that heat stress will be much more intense by the end of the century in 36 many areas under all scenarios, such as deserts and semi-arid zones in Asia (Murari et al., 2015; Mishra et 37 al., 2017b), Australia and Africa (Dosio et al., 2018; Guo et al., 2017; Schwingshackl et al., 2021; Xia et al., 38 2016; Zhao et al., 2015a), with consequences on labour productivity with respect to very heat-humidity 39 conditions (see also Figure 12.4d-f). 40 41 Drought is another major climatic impact-driver for semi-arid areas, imposing major challenges on 42 agriculture given existing water availability constraints (Kusunose and Lybbert, 2014; Barlow et al., 2016; 43 Wolski et al., 2018). Over the period 1961-2013, the annual area of drylands in drought has increased, on 44 average by slightly more than 1% per year, with large inter-annual variability (Olsson et al., 2019). In 45 general, droughts have increased in several arid and semi-arid areas over the last decades (medium 46 confidence), and are likely to increase in the future as indicated by a number of indices calculated from 47 climate (Liu et al., 2018b; Zkhiri et al., 2019; Coppola et al., 2021b; Driouech et al., 2021)(see also Figure 48 12.4j-l). 49 50 Deserts and semi-arid areas are prone to dust storms, which can drive impacts on health and several other 51 sectors (Zhang et al., 2016b; Tong et al., 2017). SRCCL indicated that the evolution of dust under climate 52 change is uncertain (Mirzabaev et al., 2019), and there is a lack of evidence and agreement of a change in 53 their frequency or intensity so far in most regions (Sections 12.4.1-12.4.9). Model projections of future 54 changes in dust are hindered by the uncertainties in future regional wind and precipitation as the climate 55 warms (Evan et al., 2016), in the effect of CO2 fertilization on source extent (Huang et al., 2017), and in the Do Not Cite, Quote or Distribute 12-103 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 impact of human activities upon the land surface (Ginoux et al., 2012) (see Chapter 10). Projected trends on 2 dust storms and dust loads in deserts and semi-arid areas vary from region to region. Dust loadings are 3 expected to decrease over most of the Sahara and Sahel (low confidence) (Section 12.4.1), increase over 4 Mexico and Southwestern U.S. (medium confidence) (Section 12.4.6), and there is low confidence of a future 5 trend due to climate change in other continents (Sections 12.4.2-12.4.5). 6 7 In conclusion, desert and semi-arid areas are strongly affected by climatic impact-drivers such as 8 extreme heat, drought and dust storms. Heat hazards are very likely increasing in all future climate 9 scenarios, but broadly consistent desert and semi-arid region changes for other climatic impact- 10 drivers’ future evolution remains uncertain. 11 12 13 12.4.10.4 Mountains 14 15 Mountains cover about 30% of the land areas on Earth (not counting Antarctica) and deliver a number of 16 vital services to humanity (IPCC, 2019b) (WGII Cross-chapter paper 5). Climate change in high mountains 17 was addressed in the SROCC, which emphasized changes in several climatic impact-drivers, such as: an 18 observed general decline in low-elevation snow cover, glaciers and permafrost (high confidence), which 19 induced changes in natural hazards such as decrease in slope stability (high confidence), changes to the 20 frequency of glacial lake outbursts (limited evidence), and climate effects on other hazards (avalanche, rain- 21 on-snow floods) with various degrees of confidence (Hock et al., 2019). 22 23 There is a growing body of literature indicating elevation-dependent warming (EDW; different rates of 24 warming by altitude although not necessarily increasing with altitude) in several mountain regions but not 25 globally (Hock et al., 2019; Pepin et al., 2019; Ahmed et al., 2020; Li et al., 2020a; Williamson et al., 2020; 26 You et al., 2020; Micu et al., 2021). Statistically significant elevational enhancement to long-term trends in 27 maximum near-surface air temperatures and diurnal temperature range were observed in southern central 28 Himalaya and in the Swiss Alps (Rottler et al., 2019; Thakuri et al., 2019). Aguilar-Lome et al. (2019) 29 reported that winter daytime land surface temperatures in the Andean region between 7°S and 20°S show the 30 strongest trends at higher elevations: +1.7°C per decade above 5000 masl. Palazzi et al. (2019) identified 31 changes in albedo and downward thermal radiation as key drivers of EDW according to the simulation 32 outputs of a high-spatial resolution model in three important mountainous areas: the Colorado Rocky 33 Mountains, the Greater Alpine Region and the Himalayas-Tibetan Plateau, but mechanisms for EDW remain 34 complex (Hock et al., 2019). Warming is also affecting mountain lake surface temperatures, increasing 35 probabilities of ice-free winters and the frequency and duration of “lake heatwaves” (O’Reilly et al., 2015; 36 Woolway et al., 2020, 2021) (high confidence) with a high variability from lake to lake. 37 38 EDW could speed up the observed, rapid upward shifts of the Freezing Level Height (FLH) in several 39 mountainous regions of the world and lead to faster changes in the snowline, the glacier equilibrium-line 40 altitude and the snow/rain transition height (high confidence). In the Indus, Ganges and Brahmaputra basins 41 in Asia, the FLH is projected to rise at a rate of 4.4 to 10.0 m yr-1 under RCP8.5 (Viste and Sorteberg, 2015). 42 In the Argentinian Andes, FLH is projected under RCP8.5 to move up more than twice as much by 2070 as 43 during the entire Holocene under the worst case scenario (Drewes et al., 2018). On the western slope of the 44 subtropical Andes (30°-38°S) in central Chile, the mean value of the free tropospheric height of the 0°C 45 isotherm under wet conditions is projected to be close to or higher than the upper quartile of the distribution 46 in the current climate, towards the end of the century and under RCP8.5 (Mardones and Garreaud, 2020). In 47 the Alps and the Pyrenees, Spandre et al. (2019) projected a rise in the natural snow elevation of 200 to 300 48 m and 400 to 600 m by mid-century under RCP2.6 and RCP8.5, respectively. In the same region, the 49 environmental equilibrium-line altitude is projected to exceed the maximum elevation of 69%, 81% and 92% 50 of the glaciers by the end of the century under RCPs 2.6, 4.5 and 8.5, respectively (Žebre et al., 2021). 51 52 Orographic effects enhance convection and stratiform heavy precipitation (due to uplift) and make 53 mountains prone to extreme precipitation events. These events are projected to increase in major 54 mountainous regions (Alps, parts of the Andes, British Columbia, Northwest North America, Calabria, 55 Carpathian, Hindu-Kush-Himalaya, Rockies, Umbria; medium to high confidence depending on location), Do Not Cite, Quote or Distribute 12-104 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 with potential cascading consequences of floods, landslides and lake outbursts in mountainous areas in all 2 scenarios (medium confidence) (Geertsema et al., 2006; Kim et al., 2015; Gaire et al., 2015; Kharuk et al., 3 2016; Syed and Al Amin, 2016; Ciabatta et al., 2016; Gariano and Guzzetti, 2016; Jurchescu et al., 2017; 4 Rajczak and Schär, 2017; Cloutier et al., 2017; Gądek et al., 2017; Alvioli et al., 2018; Schlögl and Matulla, 5 2018; Coe et al., 2018; Chen et al., 2019a; Handwerger et al., 2019; Hock et al., 2019; Patton et al., 2019; 6 Vaidya et al., 2019; Kirschbaum et al., 2020; Coppola et al., 2021b) (Sections 12.4.1-12.4.9, Chapter 11). 7 8 Declines in low-elevation snow depth and seasonal extent are projected for all SSP-RCPs (see Sections 9 12.4.1-6), along with reductions in mountain glacier surface area, increases in permafrost temperature, 10 decreases in permafrost thickness, changes in lake and river ice, changes in the amount and seasonality of 11 streamflows and hydrologic droughts in snow-dominated and glacier-fed river basins (e.g., in Central Asia; 12 Sorg et al., 2014; Reyer et al., 2017) (medium confidence), and decreases in the stability of mountain slopes 13 and snowfields. Glacier recession could lead to the creation of new glacial lakes in places like the Himalaya- 14 Karakoram region; (Linsbauer et al., 2016) and in Alaska and Canada (Carrivick and Tweed, 2016; Harrison 15 et al., 2018) (medium confidence). With increasing temperature and precipitation these can increase the 16 occurrence of glacier lake outburst floods and landslides over moraine-dammed lakes (Carey et al., 2012; 17 Rojas et al., 2014; Iribarren Anacona et al., 2015; Cook et al., 2016; Haeberli et al., 2017a; Kapitsa et al., 18 2017; Narama et al., 2018; Wilson et al., 2018; Drenkhan et al., 2019; Wang et al., 2020a) (high confidence). 19 20 In conclusion, mountains face complex challenges from specific climatic impact-drivers drastically 21 influenced by climate change: regional elevation-dependent warming (high confidence), low-to-mid- 22 altitude snow cover and snow-season decrease even as some high elevations see more snow (high 23 confidence), glacier mass reduction and permafrost thawing (high confidence), and increases in 24 extreme precipitation and floods in most parts of major mountain ranges (medium confidence). 25 26 27 12.4.10.5 Tropical forests 28 29 Tropical forests, which are among the world’s most biologically diverse ecosystems, are essentially located 30 in Central and South America, Africa and South-East Asia. AR5 and SR1.5 indicated several specific 31 climatic impact-driver changes that are particularly important to tropical forests: mean temperature increase, 32 long-term drying trends (including shifts in the length of the dry season), prolonged drought, wildfires and 33 surface CO2 increase for inland forests (IPCC, 2013, 2018). SRCCL assessed an enhanced risk and severity 34 of wildfires in tropical rainforests (high confidence), but fires are not only natural and also due to 35 deforestation and other human influences (IPCC, 2019a). 36 37 Temperature is rising in all tropical regions covered with forests and will very likely continue to rise, 38 reaching levels unprecendented in recent decades as the temperature trends rapidly emerge from weak 39 historical interannual variability (12.4.1-4; 12.5.2) (see Chapter 4, Atlas). 40 41 Regional patterns of increasing drought or unusual wet and dry periods are predicted with agreement over 42 many climate models such as over the Amazon basin (Boisier et al., 2015; Duffy et al., 2015; Zulkafli et al., 43 2016; Coppola et al., 2021b). There is medium confidence (limited evidence, high agreement) that in several 44 tropical-forest regions the dry season length has increased (Amazonia, West Africa) (Fu et al., 2013; S. 45 Debortoli et al., 2015; Saeed et al., 2017; Dunning et al., 2018; Wadsworth et al., 2019), and low confidence 46 (limited evidence) that deforestation influences the shift in the onset of the wet season in South Amazonia 47 (Leite-Filho et al., 2019). In contrast, the wet season is increasing in Northern Australia Tropical Forests 48 (Catto et al., 2012). 49 50 Tropical forests typically reach peak fire weather conditions in the dry season (Taufik et al., 2017), in 51 particular during long-lived droughts (Brando et al., 2014; Marengo et al., 2018), with consequences on tree 52 mortality, forest and carbon sink loss (Brando et al., 2019), and on the hydrological cycle in South America 53 (Martinez and Dominguez, 2014; Espinoza et al., 2020). Observations and reanalyses over the past 3-4 54 decades, combined into fire risk indices, show that the fire weather season length has been increasing by 55 about 20% globally (Jolly et al., 2015), and this index exhibits particularly high trend values over tropical Do Not Cite, Quote or Distribute 12-105 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 forest areas of South and Central America and Africa. There is generally low confidence in future projections 2 of general fire weather risk evolution in tropical forests and evolutions depend on the region (Abatzoglou et 3 al., 2019). Over the Amazon basin the fire risk increase is emerging well before 2050 while for other 4 equatorial forests no significant evolution is found. In Savanna areas the risk increase is found to be more 5 general. 6 7 In conclusion, most tropical forests are challenged by a mix of emerging warming trends that are 8 particularly large in comparison to historical variability (medium confidence). Water cycle changes 9 bring prolonged drought, longer dry seasons, and increased fire weather to many tropical forests, with 10 plants also responding to CO2 increases (medium confidence). 11 12 13 12.5 Global perspective on climatic impact-drivers 14 15 12.5.1 A global synthesis 16 17 Section 12.4 assessed changes in climatic impact-drivers by region, primarily based on a large number of 18 local, regional-scale studies (even though global studies are also used). This section presents an assessment 19 of changes in CID at the global scale. It is based on both a bottom-up synthesis of the results in 12.4, and a 20 top-down assessment from global-scale studies undertaken here. Box 12.1 summarizes global-scale CIDs 21 with levels of warming. 22 23 Global-scale studies use similar indices of climatic impact-drivers across space, although these indices may 24 not be always those used at the local or regional scale. Most published global-scale studies concentrate on 25 single sectors or climatic impact-drivers, but some take a multi-sectoral perspective (e.g., Warszawski et al., 26 2014; Arnell et al., 2016; Mitchell et al., 2017; O’Neill et al., 2018; Arnell et al., 2019a; Schleussner et al., 27 2016; Betts et al., 2018; Byers et al., 2018; Mora et al., 2018; O’Neill et al., 2018; Zscheischler et al., 2018). 28 Only a few published global-scale studies (e.g., Coppola et al., 2021; Schwingshackl et al., 2021) have used 29 CMIP6 scenarios to date. Box 12.1 summarizes global-scale CIDs with levels of warming. 30 31 All regions will experience, before 2050, increased warming, an increase of extreme heat and a decrease in 32 cold spells, regardless of the emissions trajectory (high confidence). Tropical regions, but also mid latitude 33 regions to a lesser extent, will experience an increasing number of days with heat indices crossing dangerous 34 thresholds used to characterize heat stress, such as HI>41°C (Figure 12.4). The increase, by the end of 35 century, exceeds 100 days per year in most tropical areas under SSP5-8.5 but remains much more limited to 36 almost half under SSP1-2.6. Several global-scale studies have shown that high temperature extremes will 37 increase everywhere (high confidence) ) (Gourdji et al., 2013; Perkins-Kirkpatrick and Gibson, 2017; 38 Harrington et al., 2018; Jones et al., 2018; Lehner et al., 2018; Shi et al., 2018; Tebaldi and Wehner, 2018; 39 Arnell et al., 2019; Russo et al., 2019; Schwingshackl et al., 2021), although the change depends on the 40 indicator (see also Chapter 11). For example, by 2080, at least 80% of the land surface is expected to 41 experience average summer temperatures greater than the historical (1920-2014) maximum with high 42 RCP8.5 emissions (Lehner et al., 2018). The areas of rice and maize cropland with damaging extreme 43 temperatures during the reproductive season will increase by a factor of three under RCP8.5 (Gourdji et al., 44 2013). Heatwaves that are currently considered rare become the norm almost everywhere with high 45 emissions by the end of the century (Russo et al., 2014). Heat stress as a combined function of temperature 46 and humidity also increases at the global scale, especially with high emissions (e.g. Matthews et al., 2017). 47 Growing degree-days and cooling degree-days also increase everywhere (Arnell et al., 2019) with the 48 absolute and proportional changes depending on temperature threshold. Increases in temperatures will result 49 in reductions in heating degree-days (Arnell et al., 2019; Coppola et al., 2021b) and a widespread reduced 50 frequency of cold extremes (high confidence). 51 52 Integrating the results of the regional assessments in 12.4 shows that changes in CIDs linked with the water 53 cycle or atmospheric dynamics (e.g. storms) vary more among regions, largely due to the spatial pattern of 54 changes in atmospheric circulation and changes in precipitation and evaporation (Chapters 8 and 11). There 55 is high confidence that heavy precipitation and pluvial floods will be increasing in a majority of land regions, Do Not Cite, Quote or Distribute 12-106 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 primarily due to the well-understood Clausius-Clapeyron relationship describing the increase in moisture 2 content with air temperature (Chapters 8 and 11), but there is a large spatial variability in fluvial flood 3 hazards. Top-down global-scale studies show that although fluvial flood hazards are projected to decrease in 4 regions where there are large reductions in seasonal rainfall totals or where warmer temperatures mean less 5 accumulated snow, at the global scale, fluvial flood hazard (characterized as the area affected, size of peak or 6 likelihood of an event) is projected to increase substantially through the century (Giuntoli et al., 2015; Arnell 7 and Gosling, 2016; Winsemius et al., 2016; Alfieri et al., 2017; Dottori et al., 2018; Arnell et al., 2019). 8 Projected changes in agricultural and hydrological drought characteristics are dependent on the indicator 9 used to define drought (Section 12.3; Chapter 11), but there is at least medium confidence of an increase in 10 the drought hazard in many parts of the world. This is also reflected in global scale studies, with for example 11 Naumann et al. (2018) showing that the global mean average drought duration (based on the SPEI index 12 which is calculated from the difference between precipitation and potential evaporation) increased from 7 13 months with the current climate to 18.5 months for a global warming level of 3oC. The apparent global 14 increase in drought occurrence is greater when evaporation is captured in the drought indicator (e.g. SPEI) 15 than when the indicator is based on precipitation alone (as in SPI) (Carrão et al., 2018). There is evidence 16 that the likelihood of simultaneous events in several locations will increase: (Trnka et al., 2019) found that 17 the proportion of wheat-growing areas experiencing simultaneous severe water stress events (based on SPEI) 18 in a year increased from 15% under current conditions to up to 60% at the end of the 21st century under high 19 emissions. 20 21 The regional assessment in 12.4 shows that fire weather is projected in increase with medium or high 22 confidence in every continent of the world, including Arctic polar regions. Globally, fire weather is projected 23 to increase in future, primarily due to higher temperatures and exacerbated where precipitation reduces. By 24 2050, 60% of the global land area would see a significant increase in fire weather under RCP8.5 (Abatzoglou 25 et al., 2019). There is less confidence in the projected distribution of change in fire weather across regions in 26 global scale studies. For example, (Moritz et al., 2012) projected an increase in fire weather in mid and high 27 latitudes but a reduction in the tropics, whilst Yu et al. (2019) and Bedia et al. (2015) projected an increase in 28 the tropics. These differences reflect differences in methodologies and fire weather indices adopted in 29 different studies. 30 31 Integration of the results of 12.4 shows that the total number of tropical cyclones is projected to decrease 32 through the 21st century, particularly with high emissions, but the number of very intense tropical cyclones is 33 projected to increase in most areas (at least medium confidence) (e.g. Bacmeister et al., 2018) (Chapter 11). 34 Furthermore, regions with glaciers will lose glacier mass and regions concerned with snow cover will see a 35 reduction in snow depth, the duration, or extent of cover (medium confidence in Polar regions, high 36 confidence elsewhere). 37 38 Relative sea-level rise is projected in all regions (excepting a few Arctic polar regions) with likelihoods 39 varying from very likely to virtually certain depending on the region. This will increase the frequency of 40 extreme sea levels and, depending on the level of coastal flood protection, coastal flooding (Vousdoukas et 41 al., 2018; Kirezci et al., 2020).. In terms of globally averaged extreme total water level (ETWL) frequency 42 changes, the present-day 1:100 yr event is projected to become 1:30 yr and 1:20 yr events by 2050 under 43 RCP4.5 and RCP8.5 respectively. By 2100 the present day 1:100 yr ETWL is projected to become a 1:5 yr 44 event under RCP4.5, while under RCP8.5, such events are projected to occur more than once a year 45 (Vousdoukas et al., 2018). 46 47 There is high confidence that most of the world’s sandy coasts will experience shoreline retreat, in the 48 absence of terrestrial or offshore sediment sources. Median projections presented by (Vousdoukas et al., 49 2020b) indicate that 13.6% (36,097 km) and 15.2% (40,511 km) of the world’s sandy beaches could retreat 50 by more than 100 m by 2050 (relative to 2010) under RCP4.5 and RCP8.5 respectively, implying a 12% 51 increase in severely threatened shoreline length under RCP8.5, relative to RCP4.5. These median projections 52 increase to 35.7% (RCP4.5)–49.5% (RCP8.5) (or 95,061 km–131,745 km) by the end of the century, 53 implying a 38% increase in severely threatened shoreline length under RCP8.5, relative to RCP4.5. 54 55 Figure 12.11 highlights that each region will, with high confidence, experience changes in multiple CIDs, Do Not Cite, Quote or Distribute 12-107 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 challenging the vulnerability of the region and its adaptation and mitigation capacity. All non-polar regions 2 with a coastline will see an increase in relative sea level, extreme sea level and coastal erosion, and will also 3 see an increase in hot extremes, a decrease in cold extremes, and many will experience an increase in heavy 4 precipitation. One cluster of regions – South eastern and western Africa regions, the Mediterranean, North 5 Central America, West North America, the Amazon regions, Western South America, and Australia – will 6 experience, in addition to the aforementioned globally changing CIDs, increases in either drought/aridity or 7 fire weather with high confidence. This will impact upon agricultural resources, infrastructure and health and 8 ecosystems. A second cluster of regions including mountainous areas or regions with seasonal snow cover 9 will experience (in addition to increases in heat extremes, more intense short-duration rainfall, and increases 10 in coastal hazards where coasts exist) reductions in snow and ice cover and/or increases in river flooding in 11 many cases (Western, Northwestern, Central and Eastern North America, Arctic regions, Andes regions, 12 Europe, Siberia, central and East Asia, Southern Australia and New Zealand) (high confidence). These are 13 places where energy production, ski tourism, river transportation, infrastructure could in particular face 14 increased risks. 15 16 In a few other regions, only a few CIDs are projected to change with high confidence (e.g., Sahara, Central 17 Africa, Western Africa, Madagascar, Arabian Peninsula, South eastern South America, New Zealand, Small 18 Islands). The lower confidence levels associated with changes in CIDs in these regions can be due either to 19 weaker change signals compared to natural variability, or due to limited evidence and model uncertainties 20 leading to low agreement, and does not mean that climate change may affect these regions any less than in 21 other regions. 22 23 In summary, there is high confidence that all regions of the world will experience changes in several 24 climatic impact-drivers by mid-century, albeit at region specific rates of change and confidence levels 25 per CID. Consequently, changing CIDs have the potential to affect climate-related risks in all regions 26 of the world. 27 28 29 [START FIGURE 12.11 HERE] 30 31 Figure 12.11: Synthesis of the CID changes projected by 2050 (2041-2060) with high confidence, relative to 32 reference period (1995-2014), together with the sign of change. Information is taken from the CID 33 tables in Section 12.4. Some CIDs are grouped in order to streamline the information in order to fit in all 34 information in the figure. Mean temperature, extreme heat, cold spells and frost are grouped under a 35 single icon “heat”, as they are projected to change simultaneously, albeit heat and cold are changing in 36 opposite directions. Coastal CIDs (relative sea level, coastal flooding and coastal erosion at sandy 37 beaches) are also grouped. In the figure, the “coastal” icon indicates regions where at least two of the 38 three individual coastal CIDs are projected to change with high confidence. Cases where only two of the 39 three CIDs increase with high confidence are in Arctic North Europe, Russian Arctic and Arctic North 40 West North America. A single icon is used for aridity, hydrological drought, and agricultural and 41 ecological drought, and only the number of drought types that change is indicated. For the “Snow, ice” 42 icon, information is taken from the evolution of the “Snow, Glacier and ice sheet” CID, and in most 43 regions also have similar changes for “permafrost” and “lake, river and sea ice”. Exceptions are for NEN, 44 RAR and Arctic NWN where snow is decreasing with medium confidence (thus not appearing in the 45 figure), while permafrost and Lake, river and sea ice is decreasing with high confidence. The location of 46 the icon within the regions is arbitrary. Further details on data sources and processing are available in the 47 chapter data table (Table 12.SM.1). Icon sources: https://www.flaticon.com/authors/freepik 48 49 [END FIGURE 12.11 HERE] 50 51 52 12.5.2 The emergence of climatic impact-drivers across time and scenarios 53 54 The emergence of a climate change signal occurs when that signal exceeds some critical threshold (usually 55 taken to be a measure of natural variability, see e.g. Hawkins and Sutton, 2012) or when the probability 56 distribution of an indicator becomes significantly different to that over a reference period (e.g. Chadwick et 57 al., 2019) (see also Section 1.4.2, Chapter 10), in which case external anthropogenic forcings can be detected Do Not Cite, Quote or Distribute 12-108 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 as causal factors. The “time of emergence” (ToE) or “temperature of emergence” is the time or global 2 warming level thresholds associated with this exceedance. Emergence is particularly relevant to impacts, 3 risks assessment and adaptation because human and natural systems are largely adapted to natural variability 4 but may be vulnerable if exposed to changes that go beyond this variability range; this is not to say that 5 changes within natural variability have no impact, as occurrence of damaging extremes proves. Emergence 6 also informs the timing of adaptation measures. The emergence of a change is always relative to a reference 7 period (e.g. the pre-industrial period or a recent past), depending on the framing question. In the former case, 8 the goal is to estimate the amplitude of an anthropogenically driven change while in the latter, it is to 9 estimate the amplitude of change relative to a baseline that is familiar to stakeholders. Both questions are 10 important for risk assessment, but the former may be more directly interpretable in a mitigation context. The 11 variability also refers to a time scale, generally interannual to interdecadal. The reader is referred to Section 12 1.4.2 and Chapter 10 for more details about how emergence is defined and used in the literature. 13 14 Changes in climatic impact-drivers may remain within the range of natural variability or have a time of 15 emergence that varies by region and scenario. This section assesses the evidence for the effects of 16 anthropogenic climate change on the emergence of changes in CID index, past, resent and future, as 17 evidenced by the literature assessed in other chapters, as well as additional literature assessed here, at both 18 global and regional scales. In many cases, however, sufficient literature for a robust region by region 19 assessment of times of emergence is lacking. The assessment herein is made by CID. Regional emergence 20 assessment is reported in Tables 12.3-11 but is undertaken in this section. 21 22 ToE estimation must be done with caution given the many sources of inherent uncertainties, such as 23 observations representing only a single realization of climate history, internal variability (whose frequency -- 24 e.g., annual or decadal -- needs to be precisely defined), model biases, and potential low-frequency changes 25 in variability (Chapter 10) (Lehner et al., 2017). In addition, a homogeneous interpretation of multiple 26 studies is hampered by heterogeneous methodologies used to calculate emergence. In this section, we assess 27 emergence and its confidence level based on such multiple methods as provided by the literature, and unless 28 specified otherwise, emergence here refers to S/N>1 (S/N= signal to noise ratio) relative to a preindustrial 29 baseline and interannual variability (the “noise”). Furthermore, observed trends and attribution are taken into 30 account in combination with climate simulations (historical or projections) for assessing whether a trend has 31 already emerged in the historical period. 32 33 Mean air temperature: Warming of mean annual temperatures has already emerged in all land regions as 34 obtained from past observations and confirmed by historical simulations (high confidence) (King et al., 2015; 35 Hawkins et al., 2020) (Figure 1.13), with S/N ratios larger than 2. In the current climate, the highest S/N 36 ratios exceed 5 over Central Africa, Amazonia, East and South East Asia. Seasonal warming emergence 37 depends on the season. Because the temperature variability in the mid-latitudes is higher in winter than in 38 summer, the emergence of seasonal warming occurs for summer but not for winter in most of this part of the 39 world. In Europe, summer warming has emerged in all regions (medium confidence, medium agreement), 40 and in North America, it has emerged only over Eastern and Western regions while in winter there is low 41 confidence of an emergence in warming in all regions for both Europe and North America (Lehner et al., 42 2017; Hawkins et al., 2020). When considering the climate of the end of the 20th century (i.e. recent past) as 43 a baseline, the emergence of mean temperature is projected at very different times depending on the scenario. 44 For instance, emergence is reached by 2050 under RCP8.5 in most areas of Europe, Australia or East Asia, 45 but it does not occur within the 21st century under RCP2.6 (medium confidence) (Sui et al., 2014; Im et al., 46 2020). This means that RCP2.6 is efficient to keep mean temperatures within the recent climate range in the 47 mid-latitudes. However, even under RCP2.6, mean temperatures in tropical regions that have not already 48 emerged are projected to emerge before 2050 (medium confidence). 49 50 Extreme heat and cold: Increase in heat extremes have emerged or will emerge in the coming 3 decades in 51 most land regions (high confidence) (King et al., 2015; Seneviratne and Hauser, 2020) (Chapter 11), relative 52 to the pre-inudtrial period, as found by testing significance of differences in distributions of yearly 53 temperature maxima in simulated 20-year periods. In tropical regions, wherever observed changes can be 54 established with statistical significance, and in most mid-latitude regions, there is high confidence that hot 55 and cold extremes have emerged in the historical period, but only medium confidence elsewhere. In other Do Not Cite, Quote or Distribute 12-109 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 regions emergence is projected at the latest in the first half of the 21st century (high confidence) under 2 RCP8.5 at least (King et al., 2015; Seneviratne and Hauser, 2020). Relative to the end of 20th-century 3 conditions, changes in humid heat stress as characterized by wet bulb temperature, indicates a ToE as early 4 as in the first two decades of the 21st century in RCP8.5 at least in many tropical regions (most of Africa in 5 the band [20°S-20°N], South Asia and South-East Asia) (Im et al., 2020) (medium confidence) . By 2050 and 6 under RCP8.5, wet bulb temperature is projected to emerge in many other areas such as South Africa, North 7 Africa, Europe, and most of Central, Southern, Eastern Asia and Northern and Eastern Australia, while under 8 RCP2.6, emergence is either reached later in the century (Europe, Central Asia, Northern Australia), or never 9 reached in the century (Im et al., 2020). Decrease of cold spells has already emerged above the interannual 10 variability in Australasia, Africa and most of northern South America, and they are projected to emerge 11 before 2050 in the Northern mid-latitudes and in southern South America (King et al., 2015) under RCP8.5 12 (medium confidence, low evidence and high agreement). 13 14 Mean precipitation: Over only a few regions mean precipitation changes have emerged in the historical 15 period (increase in Northern and Eastern Europe and decrease in West Africa, Amazonia) from observations 16 with an S/N ratio larger than 1 (Hawkins et al., 2020) (low confidence). The emergence of increasing 17 precipitation before the middle of the 21st century is found across scenarios in Siberian regions, Russian far 18 east, northern Europe, Arctic regions and the northernmost parts of North America (high confidence) and 19 later in other Northern mid-latitude areas, depending on the scenario, albeit with different methods and 20 emergence definitions used in climate projections (Chapter 8) (Giorgi and Bi, 2009; Maraun, 2013; King et 21 al., 2015; Akhter et al., 2018; Kumar and Ganguly, 2018; Nguyen et al., 2018; Barrow and Sauchyn, 2019; 22 Rojas et al., 2019; Kusunoki et al., 2020; Pohl et al., 2020; Li et al., 2021). Decreases in mean precipitation 23 are projected to emerge in parts of Africa by the middle of trhe century, and later in the Mediterranean and 24 southern Australia, but the emergence depends on the scenario, and specific seasons for crop growth 25 (Nguyen et al., 2018; Rojas et al., 2019). Mean precipitation does not emerge in any of these regions at 26 anytime in the 21st century under RCP2.6, but emerges in all under RCP8.5. ToE under RCP4.5 is projected 27 to be around 25 years later relative to RCP8.5 in many of the early emergence regions, highlighting the 28 importance of mitigation to gain more time for adaptation. 29 30 Heavy precipitation and floods: There is low confidence in the emergence of heavy precipitation and 31 pluvial and fluvial flood frequency in observations, despite trends that have been found in a few regions 32 (Chapter 8, Chapter 11, and across Section 12.4). In climate projections, the emergence of increase in heavy 33 precipitation strongly depends on the scale of aggregation (Kirchmeier-Young et al., 2019), with, in general, 34 no emergence before a 1.5°C or 2°C warming level, and before the middle of the century (medium 35 confidence), but results depend on the method used for the calculation of the ToE (Maraun, 2013; King et al., 36 2015; Kusunoki et al., 2020). Emergent increases in heavy precipitation are found in several regions when 37 aggregated at a regional scale in Northern Europe, Northern Asia and East Asia, at latest by the end of the 38 century in SRES A1B or RCP8.5 scenarios or when considering the decadal variability as a reference 39 (medium confidence) (Maraun, 2013; Li et al., 2018e, 2021; Kusunoki et al., 2020). There have been few 40 emergence studies for streamflow and flooding, although one study showed emergence of different 41 hydrological regimes at different times during the 21st century across the United States (Leng et al., 2016). 42 Variability in extreme streamflows from year to year can be high relative to a trend (Zhuan et al., 2018). 43 Given the heterogeneity of methods and results, there is only low confidence in the emergence of heavy 44 precipitation and flood signals in any region when considering the S/N ratio. 45 46 Droughts, aridity and fire weather: There is low confidence in the emergence of drought frequency in 47 observations, for any type of drought, in all regions. Even though significant drought trends are observed in 48 several regions with at least medium confidence (Section 12.4), agricultural and ecological drought indices 49 have interannual variability that dominates trends, as can be seen from their time series (Guo et al., 2018a; 50 Spinoni et al., 2019b; Haile et al., 2020; Wu et al., 2020b) (medium confidence). Studies of the emergence of 51 drought with systematic comparisons between trends and variability of indices are lacking, precluding a 52 comprehensive assessment of future drought emergence. Historical climate simulations indicate that fire 53 weather indices have already emerged in several regions (the Amazon basin, Mediterranean, Central 54 America, West and South Africa) (low confidence, low evidence) (Abatzoglou et al., 2019), and emergence is 55 projected with low confidence by the middle of the century in several other regions (Southern Australia, Do Not Cite, Quote or Distribute 12-110 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 Siberia, most of Northern America and Europe) when considering several indices together. 2 3 Wind: Observed mean surface wind speed trends are present in many areas (12.4), but the emergence of 4 these trends from the interannual natural variability and their attribution to human-induced climate change 5 remains of low confidence due to various factors such as changes in the type and exposure of recording 6 instruments, and their relation to climate change is not established. For future conditions, there is limited 7 evidence of the emergence of trends in mean wind speeds due to the lack of studies quantifying wind speed 8 changes and their interannual variability. The same limitation also holds for wind extremes (severe storms, 9 cyclones, dust and sandstorms). 10 11 Snow and ice: The decrease in the Northern Hemisphere snow cover extent in spring has already emerged 12 from natural variability (Section 3.4.2). The snow cover duration period is projected to emerge over large 13 parts of Eastern and Western North America and Europe by the mid-century both in spring and autumn, and 14 emergence is expected in the second half of the 21st century in the Arctic regions in the high RCP8.5 15 scenario (Chapter 9, SROCC) (medium confidence). For snow depth or snow water equivalent, there is low 16 confidence (limited evidence) of the emergence of a decrease before 2050 because climate change also 17 increases the variability of the snow depth signal, in Europe (Willibald et al., 2020) (Section 3.4.2). 18 Terrestrial permafrost is warming worldwide due to climate change (Sections 2.3.2.5, 9.5.2). Due to weak 19 interannual variability of permafrost temperatures, terrestrial permafrost warming has emerged above natural 20 variability in almost all observed time series of the Northern Hemisphere (Biskaborn et al., 2019) (medium 21 confidence, limited evidence, high agreement), but the active layer thickness exhibits considerable 22 interannual variability inhibiting evidence for emergence (Chapter 9). 23 24 Sea ice: Sea ice area decrease in the Arctic in all seasons has already emerged from the interannual 25 variability (high confidence) (Chapter 9). By contrast, the Antarctic sea ice area shows no significant trend, 26 and therefore no emergence. 27 28 For other snow and ice CIDs (heavy snowfall and ice storm, hail, snow avalanche), there is limited evidence 29 of emerging signals. 30 31 Relative sea level, coastal flood and coastal erosion: Near-coast RSLR will emerge before 2050 for 32 RCP4.5 along the coasts of all AR6 regions (with coasts) except EAS, RFE, MDG, the southern part of ENA 33 and the Antarctic regions (9.6.1.4) (Bilbao et al., 2015) (medium confidence). Under RCP8.5, emergence of 34 near-coast RSLR is projected by mid-century along the coasts of all AR6 regions (with coasts), except WAN 35 where emergence is projected to occur before 2100 (Section 9.6.1.4) (Lyu et al., 2014) (medium 36 confidencce). Emergence studies for ETWL and coastal erosion are lacking and hence it is not currently 37 possible to robustly assess emergence in these CIDs. 38 39 Mean ocean temperature and marine heatwave: The emergence of the sea surface temperature increase 40 signal has been observed in global oceans over the last century, and the largest S/N values are found in the 41 tropical Atlantic and tropical Indian Oceans (Hawkins et al., 2020). There is high confidence in the 42 widespread occurrence of marine heatwaves in all basins and marginal seas over the last decades (Chapter 43 9), but the emergence of this signal above the natural variability has not yet been addressed in detail. 44 45 Ocean acidity, ocean salinity and dissolved oxygen: The global ocean pH has very likely emerged from 46 natural variability for more than 95% of the global open ocean (SROCC, Chapter 2). The regional signals are 47 more variable, but in all ocean basins, the signal of ocean acidification in the surface ocean is projected to 48 emerge in the early 21st century (Chapter 5). The mean time of emergence for acidity in the coastal 49 subtropical to temperate northeast Pacific and northwest Atlantic is above two decades (5.3.5.2) (high 50 agreement, medium evidence). Salinity change signals have already emerged with 20–45% of the zonally 51 averaged basin in the Atlantic, 20–55% in the Pacific and 25–50% in the Indian Oceans and will be reaching 52 35–55% in the Atlantic in 2050 to 55–65% in 2080; 45–65% to 60–75% in the Pacific; 45–65% to 60–80% 53 in the Indian Oceans (Silvy et al., 2020) (Chapter 9). Deoxygenization has already emerged in many open 54 oceans. The signal is most evident in the Pacific and Southern Oceans but not evident in the North Atlantic 55 Ocean (Andrews et al., 2013; Levin, 2018). However, there is medium confidence in the emergence of the Do Not Cite, Quote or Distribute 12-111 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 anthropogenic signal in many other oceanic regions by 2050 (Henson et al., 2017; Levin, 2018). 2 3 There is high confidence that several CID changes have already emerged above natural variability in 4 the historical period in many regions (e.g. mean temperature in most regions, heat extremes in tropical 5 areas, sea ice, salinity). Heat and cold CIDs (excluding frost) that have not already emerged will 6 emerge by 2050 whatever the scenario (medium confidence) in almost all land regions. The emergence 7 of increasing precipitation before the middle of the century is also projected in in Siberian regions, 8 Russian far east, northern Europe and the northernmost parts of North America and Arctic regions 9 across scenarios with the various methods and emergence definitions used (high confidence). Studies 10 are missing to properly assess S/N emergence for droughts and for wind CIDs. Arctic sea ice extent 11 declines have mostly emerged above noise level (medium to high confidence), and the emergence of 12 declining snow cover is expected by the end of the century under RCP8.5. There is medium confidence 13 that, under RCP8.5, the anthropogenic forced signal in near coast relative sea-level change will emerge 14 by mid-century in all regions with coasts, except in the West Antarctic region where emergence is 15 projected to occur before 2100. In all ocean basins, the signal of ocean acidification in the surface 16 ocean is projected to emerge before 2050 (high confidence). 17 18 19 [START TABLE 12.12 HERE] 20 21 Table 12.12: Emergence of CIDs in different time periods, as assessed in this section; the colour corresponds to the 22 confidence of the region with the highest confidence; White cells indicate where evidence is lacking or 23 the signal is not present, leading to overall low confidence of an emerging signal. 24 High Medium Low Medium High Confidence Confidence confidence Confidence Confidence (decreasing) (decreasing) (increasing) (increasing) 25 Emerging between 2050 least for RCP8.5/SSP5- Emerging by 2050 at and 2100 for at least Already emerged in historical period RC8.5/SSP5-8.5 Climatic Impact- Climatic Impact-Driver Category Driver Type 8.5 Mean air temperature 1 1 High confidence except over a few regions (CNA and NWS) where there Heat and Extreme heat 2 3 is low agreement across observation Cold Cold spell 4 5 datasets Frost 2 High confidence in tropical regions Mean precipitation 6 7 where observations allow trend estimation and in most regions in the River flood mid-latitudes, medium confidence Heavy precipitation and pluvial flood 8 elsewhere Wet and Landslide 3 High confidence in all land regions 4 Emergence in Australia, Africa and Dry Aridity most of northern South America where Hydrological drought observations allow trend estimation Agricultural and ecological drought 5 Emergence in other regions 6 Increase (Most Northern mid- Fire weather latitudes, Siberia, Arctic regions by Mean wind speed mid-century, other later in the century) Severe wind storm 7 decrease in the Mediterranean area, Wind South Africa, South West Australia Tropical cyclone 8 Northern Europe, Northern Asia and Sand and dust storm East Asia under RCP8.5 and not in Snow, glacier and ice sheet 9 10 low-end scenarios Snow and 9 Europe, Eastern and Western North Permafrost Ice America (snow) Lake, river and sea ice 11 Do Not Cite, Quote or Distribute 12-112 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Heavy snowfall and ice storm 10 Arctic (snow) 11 Arctic sea ice only Hail 12 Everywhere except WAN under Snow avalanche RCP8.5 Relative sea level 12 13 With varying area fraction Coastal flood depending on basin Coastal 14 Pacific and Southern Ocean then Coastal erosion many other regions by 2050 Mean ocean temperature Marine heatwave Oceanic Ocean acidity Ocean salinity 13 Dissolved oxygen 14 Air pollution weather Other Atmospheric CO2 at surface Radiation at surface 1 2 [END TABLE 12.12 HERE] 3 4 5 [START CROSS-CHAPTER BOX 12.1 HERE] 6 7 Cross-Chapter Box 12.1: Projections by warming levels of hazards relevant to the assessment of 8 Representative Key Risks and Reasons for Concern 9 10 Contributors: 11 Claudia Tebaldi (USA), Guofinna Aoalgeirsdottir (Iceland), Sybren Drijfhout (UK), John Dunne (USA), 12 Tamsin Edwards (UK), Erich Fischer (Switzerland), John Fyfe (Canada), Richard Jones (UK), Robert Kopp 13 (USA), Charles Koven (USA), Gerhard Krinner (France), Friederike Otto (UK/Germany), Alex C. Ruane 14 (USA), Sonia I. Seneviratne (Switzerland), Jana Sillmann (Norway/Germany), Sophie Szopa (France), 15 Prodromos Zanis (Greece). 16 17 A consistent risk framework (Reisinger et al., 2020) has been adopted across the three Working Groups 18 (WG) in IPCC AR6 while recognizing the diversity of risk concepts across disciplines. WGI is assessing 19 changes in climatic impact-drivers (CIDs), which are physical climate system conditions (e.g., means, events 20 and extremes) that affect an element of society or ecosystems. Depending on system tolerance, CIDs and 21 their changes can be detrimental, beneficial, neutral, or a mixture of each across interacting system elements 22 and regions (Sections 12.1-12.3). In the assessment of Representative Key Risk (RKR) categories and 23 Reasons for Concerns (RFCs) in WGII Chapter 16, the focus lies on the adverse consequences of climate 24 change, for which many types of CIDs (i.e., ‘hazards’ in the context of identified risks) play a key role. This 25 box synthesizes the assessment of such hazards according to global warming levels (GWLs) from various 26 chapters of WGI to inform understanding of their potential changes and associated risks with temperature 27 levels in general, and in particular to facilitate WGII integrated assessments of RKRs and RFCs. Another 28 cross-chapter box, CCB 11.1, connects the organization of regional information according to GWLs to the 29 other common dimension along which future projections are organized, i.e., scenarios. Section 1.6 describes 30 all dimensions of integration adopted in this report, adding cumulative carbon emissions to GWLs and 31 scenarios. 32 33 Eight RKRs are identified within WGII Chapter 16: 34 RKR-A: risk to the integrity of low-lying coastal socio-ecological systems; 35 RKR-B: risk to terrestrial and ocean ecosystems; 36 RKR-C: risk to critical infrastructure and networks; 37 RKR-D: risk to living standards; 38 RKR-E: risk to human health; 39 RKR-F: risk to food security; 40 RKR-G: risk to water security; and 41 RKR-H: risk to peace and migration. Do Not Cite, Quote or Distribute 12-113 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 1 2 RFCs further synthesize the landscape of risks from climatic changes into five categories (from IPCC TAR 3 onward; Smith et al., 2001): 4 RFC1: Risks to unique and threatened systems; 5 RFC2: Risks associated with extreme weather events; 6 RFC3: Risks associated with the distribution of impacts; 7 RFC4: Risks associated with global aggregate impacts; and 8 RFC5: Risks associated with large-scale singular events. 9 Importantly, the assessment of risk in WGII considers hazards as only one component of an integrated 10 assessment that involves their complex interaction with exposure and vulnerability of the systems at risk 11 (Reisinger et al., 2020). 12 13 Hazards relevant to RKRs and RFCs are identified among aspects of the climate system that have an 14 episodic, short term nature, like extreme events (particularly relevant to RFC2 but contributing to many other 15 risk categories). Increasing GWLs translate into changing characteristics of frequency, duration, intensity, 16 seasonality and spatial extent for many of these hazards also apparent in scenario-based results (Chapter 11, 17 Chapter 12, Sections 12.4 and 12.5.1). Also, increasing GWLs increase the likelihood of compound temporal 18 or spatial occurrence of similar or different hazards (Chapter 11, Section 11.8). Other relevant hazards 19 coincide with long-term trends embodying a gradual change that may result in unfavorable environmental 20 conditions. Furthermore, RFC5’s focus on singular events includes concern surrounding potential tipping 21 points and irreversible behavior in the physical climate system. 22 23 Cross-Chapter Box12.2 Table 1 organizes information by hazard and presents current state and future change 24 assessments with increasing GWLs (defined by increasing GSAT, see CCB2.3). We draw on individual 25 chapters across the WG1 report for the assessment of how these hazards vary with GWL. Hazards for which 26 a relation to GWLs has not been assessed are not reported in the table. 27 28 29 [START CROSS-CHAPTER BOX 12.1, TABLE 1 HERE] 30 31 Cross-chapter Box 12.1 Table 1: Summary of CIDs/hazards that are identified as driving RKRs and RFCs. 32 The behavior of each (in most cases considered at the global scale, but 33 for some types in terms of spatially resolved patterns) as a function of 34 GWLs is described and when possible quantified (in which case the cell 35 is colored), together with the level of confidence of the assessment, to be 36 found in more detail in the chapter/sections indicated in the 37 corresponding column. For the relation with GSAT levels, two columns 38 detail current state, which can be associated to ~1C of global warming, 39 and future behavior. Tipping Points and Irreversibility is 40 comprehensively assessed with CMIP6 models up to GWL = 3C with 41 fewer studies and lower confidence at higher GWLs up to 5C. 42 43 44 Do Not Cite, Quote or Distribute 12-114 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Hazard Category RKR/RFC WGI Chapter Behavior at ~1C (present) Behavior as a function of GWL (future) Sub-category relevance references Extreme events Frequency and intensity of hot extremes increased and cold extremes Strong linear relation between magnitude and intensity of heat and cold extremes decreased at the global scale and in most regions since 1950 (GSAT change and GSAT, detectable from warming as low as 1.5C; changes in the extreme Sections 11.3, ~0.6C) (virtually certain). Number of warm days and nights increased; metrics twice as large (in mid-latitude regions) or more (in high latitude regions) 11.9, 12.4; Hot and Cold All RKRs; intensity and duration of heat waves increased; number of cold days and than GSAT warming (very likely); metrics related to frequency of exceedance may Figures 11.4, extremes RFC2, RFC3 nights decreased (virtually certain). Regional-to-continental scale trends show stronger than linear relationships (exponential) (very likely).Compared to 11.5, 11.9; generally consistent global-scale trends (high confidence). Limited data in a today, changes in extremes at +2°C at least two times larger than at +1.5°C, and Table 11.2. few regions (esp. of Africa) hampers trend assessment. four times larger at +3°C. Frequency and intensity of heavy precipitation events increased at the Precipitation events - including those associated with Tropical Cyclones (TCs) - global scale over a majority of land regions with good observational increase with GSAT. For GWLs >2°C very rare (e.g., 1 in 10 or more years) heavy coverage (high confidence) and at the continental scale in North America, precipitation events more frequent and more intense over all continents (virtually Europe, and Asia. Larger percentage increases in heavy precipitation certain) and nearly all AR6 regions (likely). Likelihood lower at lower GWLs and for Sections 11.4, observed in the northern high latitudes in all seasons, and in the mid Extreme less rare events. At the global scale, intensification of heavy precipitation 11.7, 12.4; All RKRs; latitudes in the cold season (high confidence). Regional increases in the precipitation generally follows Clausius-Clapeyron (~6-7% per degree C of GSAT warming; high Figure 11.5, RFC2, RFC3 frequency and/or the intensity of heavy rainfall also observed in most parts events confidence). Increase in frequency of heavy precipitation events accelerates with 11.13, 11.14; of Asia, northwest Australia, northern Europe, southeastern South warming, higher for rarer events (high confidence), with approximately a doubling Table 11.2. America, and most of the United States (high confidence), and west and and tripling frequency of 10-year and 50-year events, respectively, at 4°C of global southern Africa, central Europe, the eastern Mediterranean region, warming. Mexico, and northwestern South America (medium confidence). GHGs likely the main cause. Increased atmospheric evaporative demand in dry seasons over a majority Sections 11.6, of land areas due to human induced climate change (medium confidence). All RKRs; Upward trend with GSAT (high confidence). 12.4; Figure Drought Esp. observed in dry summer climates in Europe, North America and Africa RFC2, RFC3 11.18; Table (high confidence). 11.2. Upward trend with GSAT for flooded area extent, starting from 2C vs. 1.5C and higher levels. Increase in the frequency and magnitude of pluvial floods (high Sections 11.5, All RKRs; Inland floods confidence). Increasing flood potential in urban areas where heavy precipitation 12.4; Table RFC2, RFC3 projected to increase, especially at high GWLs (high confidence). 11.2. Increase in precipitation from TC with GSAT; Average peak TC wind speeds, Human contribution to extreme rainfall amount from specific TC events proportion of intense TCs, and peak wind speeds of most intense TCs increase Sections 11.7.1, All RKRs; Tropical Cyclones (high confidence). Global proportion of major TC intensities likely increased globally with GSAT (high confidence). Decrease or lack of change in global 12.4; Table RFC2, RFC3 over the past four decades. frequency of TCs (all categories) with GSAT (medium confidence). 11.2. High confidence that MHWs have increased in frequency over the 20th MHWs very likely become 2-9 times more frequent in 2081-2100 compare to Marine RKR A,B,F; century, with an approximate doubling from 1982-2016, and medium Section 12.4; 1985-2014 under SSP2.1.6 corresponding to a GWL of 2.1°C (1.3°C -2.8°C 95%CI), Heatwaves RFC1,2,3; confidence that they have become more intense and longer since the Box 9.2. or 3-15 times more frequent under SSP5-8.5 corresponding to a GWL of 5°C (3.6- 1980s. Do Not Cite, Quote or Distribute 12-115 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI 6.9 95%CI). Spatial heterogeneity with larger changes in the tropical oceans and Arctic Ocean (medium confidence) Higher frequency with increasing GSAT. Increasing trend in more frequent Higher frequency already detected: More frequent concurrent heat waves Section 11.8; Concurrent concurrent heat waves and droughts with GSAT (high confidence). More frequent All RKRs; and droughts. Increased compound flooding risk (storm surge, extreme Table 11.2; Events in Time concurrent (in time) extreme events at different locations with increasing GSAT, RFC2, RFC3 rainfall and/or river flow) in some locations; the probability of concurring Boxes 11.3, and Space for GWLs > 2°C (high confidence). Compound flooding risk (storm surge, extreme events likely increased. 11.4. rainfall and/or river flow) increasing with GSAT (high confidence). Trends Weather conditions that promote wildfire (compound hot, dry and windy Fire Weather RKR-B, C; Weather conditions promoting wildfire (compound hot, dry and windy events) Section 12.4; events) more probable in some regions over the last century (medium trends RFC1,2,3 likely more frequent with GSAT. Table 11.2. confidence). Behavior to first order controlled by emissions and policies, not by meteorology. Air Pollution RKR-E; Ozone decreases with GSAT in low polluted regions (-0.2 to -2 ppbv per degree C). Sections 6.5, Not discernible. weather RFC3 Ozone increases with GSAT in regions close to sources of precursors (0.2 to 2 ppbv 12.4. per degree C). Sections 4.6.1.1, 12.4, Temperatures scale approximately linearly with GSAT, largely independently of Atlas; Figure scenario (high confidence). High latitudes of Northern Hemisphere warm faster 4.36. Spatial patterns of temperature changes associated with the 0.5°C (virtually certain). Antarctic polar amplification smaller than Arctic (high Figure.Atlas.15 Patterns of mean RKR-B, D, F, difference in GMST warming between 1991–2010 and 1960–1970 confidence). Arctic annual mean temperatures warm between 2 and 2.4 times and all Atlas warming RFC1,3,4 consistent with projected changes under 1.5°C and 2°C of global warming. faster for GWLs between 1.5°C-4°C. In the Southern Hemisphere relatively high sections’ rates of warming in subtropical continental areas of South America, southern figures for Africa, Australia (high confidence). mean temperature changes. Sections 4.6.1, 7.4.4.1, 12.4.9; RKR- Arctic Warming Very likely more pronounced (2–2.4 times faster) than the global average over the Atlas.11; A,C,G,H; Emerged already from internal variability. trends 21st century (high confidence). Figures 4.22, RFC1, RFC3 Atlas.32; Table 4.2. Changes in large-scale atmospheric circulation and precipitation with each half degree of warming (high confidence). Stable pattern of change over time and Sections Patterns of Regional patterns of recent trends, over at least the past three decades, scenarios. Some departures from linearity possible at regional scale (medium 2.3.1.3.4, 4.5.1, RKR-B, D, F, Precipitation consistent with documented increase in precipitation over tropical wet confidence). Precipitation increase on land higher at 3°C and 4°C compared to 4.6.1, 12.4; RFC1, RFC3 Change regions and decrease over dry areas. 1.5°C and 2°C. Precipitation increases in large parts of the monsoon regions, Figures 2.14, tropics and high latitudes, decreases in the Mediterranean and large parts of the 4.37, Atlas.15. subtropics (high confidence). Model and observations show globally averaged SSTs warming at a lower rate of ~80% than GSAT. It is virtually certain that SST will continue to increase at a rate RKR-A, B, depending on future emission scenario ranging from 0.4 – 1.5°C in 2081-2100 Sections Increased 0.81 (0.65-0.94) per degree C of GSAT (1850-1900 average vs. SST warming D; RFC1, relative to 1995-2014 under SSP1-2.6 correspomding to a GWL of 2.1°C (1.3°C - 2.3.1.1.6, 9.2.1, 2009-2018 average). RFC4 2.8°C 95%CI), to 2–4°C under SSP5-8.5 correspomding to a GWL of 5°C (3.6°C - 12.4 6.9°C 95%CI). Do Not Cite, Quote or Distribute 12-116 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Sections Virtually certain decline of surface pH globally over the last 40 years at a Increase of net ocean carbon flux throughout the century irrespectively of the 2.3.3.5, 4.3.2.4, Ocean RKR-A,B; rate of 0.017–0.027 pH units per decade, decline also in the subsurface emission scenario considered (high confidence). Decrease of ocean surface pH 5.3.4, 5.4.2, Acidification/pH RFC1, RFC4 over the past 2-3 decades (medium confidence). Surface pH now the lowest through the 21st century, except for SSP1-1.9 and SSP1-2.6 where values increase 12.4; Figure of at least the last 26 thousand years (very high confidence). slightly starting from 2070 to 2100 (high confidence). 4.6. RKR- 9.4% chance of at least 3 months of drought in a year at current levels Increase at the global scale in the chance of at least 3 months of drought in a year Sections 12.4, SPEI index Global B,F,G,H; (~1C) to about 20% (15-30%) at 1.5°C, 35%(20-45%) at 2°C to 60% (45-75%) at 4°C. 12.5.1. RFC3 Medium confidence that both ENSO amplitude and frequency of high- No change in the amplitude of ENSO variability (medium confidence); enhanced RKR- magnitude events since 1950 higher than over the pre-industrial period Sections 2.4.2, ENSO-related variability of precipitation under SSP2-4.5 and higher (high ENSO Variability B,D,F,G; (before 1850) but low confidence of this being outside the range of internal 4.3.3.1, 4.5.3.2; confidence). Likely shift eastward of the pattern of teleconnection over N.Pacific RFC2,3,5 variability. No clear evidence shifts in ENSO or associated features or its Figure 4.10. and N. America. teleconnections. Arctic Sea Ice Area decreased for all months since 1970s; strongest Sections 2.3.2, decrease in summer (very high confidence). Arctic sea-ice younger, thinner The Arctic Ocean will likely become sea ice–free in September before 2050 in all 4.3.2.1, 9.3.1, RKR-A, B, and faster moving (very high confidence). Current pan-Arctic sea ice levels considered SSP scenarios; such disappearance consistently occurring in most- Sea-Ice Loss 9.3.2, 12.4.9; H; RFC1,3,5 unprecedented since 1850 (high confidence). Low confidence in all aspects years at 2-3°C (medium confidence) and including several months in most years at Figures 4.2c, of Antarctic sea-ice prior to the satellite era. Antarctic SIA experienced little 3-5°C (high confidence). 4.5. net change since 1979 (high confidence). Global permafrost volume in the top 3 m decreasing by about 25±5% per °C for Sections RKR-A,C; Increases in permafrost temperatures in the upper 30 m over the past GWLs < 4°C. Relative to 1995-2014: at 1.5 and 2 °C decreasing by less than 40% Permafrost Thaw 2.3.2.5, 9.5.2, RFC3,5 three to four decades throughout the permafrost regions (high confidence). (medium confidence), at 2 and 3 °C by less than 75% (medium confidence), at 3 12.4.9. and 5 °C by more than at least 60% loss (medium confidence). Up to 2050, limited scenario/GWL dependency (likely SLR=~0.15-0.30 m). By 2100, GMSL is rising with accelerated rate since the 19th century (high likely GMSL rise wrt 1995-2014 of 0.51 (0.40-0.69) m, 0.62 (0.50-0.81) m and 0.70 Sea-Level RKR- Sections confidence), almost doubled during past 2 decades (about 0.1 mm yr-2). (0.58-0.91) m for, respectively, GWLs of 2.0°C, 3.0°C, and 4.0°C (medium Change over the A,C,D,E,F,G, 9.6.1.2, 9.6.3.3, GMSL increase over the 20th century faster than over any preceding confidence). Deep uncertainty in projections for GWLs> 3°C because of Ice Sheet 21st century H; RFC1,3,4 9.6.3.4, 12.4. century in at least the last three millennia (high confidence). behavior. For example, incorporation of low confidence ice-sheet processes under SSP5-8.5 (approximately 5°C) leads to a rise of 0.6-1.6 m rather than 0.7-1.1 m. Sea-Level Global Mean Sea Level commitment (over the 2000-year-long period following Change peak warming) of 2-6 m for 2°C peak warming, 4-10 m for 3°C peak warming, 12- Commitment RFC5 Section 9.6.3.5. 16 m for 4°C peak warming, and 19-22 m for 5°C peak warming (medium (2,000 years agreement, limited evidence). after peak GWL) Northern Linear change of NH snow cover in Spring of about 8% (area) per °C of (for GWLs Substantial reductions in spring snow cover extent in the NH since 1978 Sections Hemisphere RKR-G, <4°C). Relative to 1995-2014: at 1.5-2 °C NH spring snow cover extent likely (very high confidence). Since 1981, general decline in NH spring snow water 2.3.2.2, Spring Snow RFC1,3 decreases by less than 20% (medium confidence); at 2-3 °C likely decreases by less equivalent (high confidence). 9.5.3, 12.4. Cover than 30%; At 3-5 °C, likely decreases by more than 25%. For 1.5- 2 °C about 50-60% (low confidence) of glacier mass outside the two ice Very high confidence global glaciers continuing retreat since ~1850. Current sheets and excluding peripheral glaciers in Antarctica remaining, predominantly in global glacier mass loss highly unusual over at least the last 2000 years the polar regions. At 2- 3 °C about 40-50% (low confidence) of current glacier mass Sections Mass Loss of RKR-B,G; (medium confidence). Increased rate of glacier mass loss over the last 3-4 outside Antarctica remaining. At sustained 3–5 °C 25-40% (low confidence).of 2.3.2.3, 9.5.1, Glaciers RFC1, RFC3 decades (high confidence). ). Glaciers not in balance with respect to current glacier mass outside Antarctica remaining. likely nearly all glacier mass 12.4. current climate conditions and will continue to lose mass for at least lost in low latitudes, Central Europe, Caucasus, Western Canada and USA, North several decades. Asia, Scandinavia and New Zealand. Do Not Cite, Quote or Distribute 12-117 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Tipping Points / Irreversibility Sections 5.4.3, Amazon drying and deforestation expected to cause a rapid change in the regional Amazon Forest 5.4.8, 8.6.2.1, RFC1, RFC5 Highly dependent on human disturbance water cycle, possibly linked to the crossing of a climate threshold. Low confidence Dieback 12.4.10; Table change will occur by 2100. 4.11. Boreal Forest Possible if climate threshold is exceeded, but counteracted by poleward RFC1, RFC5 Highly dependent on human disturbance Section 5.4.8. Dieback expansion. At sustained warming levels between 1.5°C and 2°C, the ice sheets will continue to lose mass (high confidence); on time scales of multiple centuries, the Greenland and West Antarctic Ice Sheets will partially be lost (medium confidence); there is limited evidence that the Greenland and West Antarctic Ice Sheets will be lost almost completely and irreversibly over multiple millennia; At sustained warming Sections Greenland Ice Sheet mass loss rate increased substantially since the turn of levels between 2°C and 3°C, there is limited evidence that the Greenland and West 2.3.2.4, the 21st century (high confidence). The Antarctic Ice Sheet has lost mass Ice Sheets RFC5 Antarctic Ice Sheets will be lost almost completely and irreversibly over multiple 9.4.1, 9.4.2; between 1992 and 2017 (very high confidence), with an increasing mass- millennia, and high confidence in increasing risk of complete loss and increasing Table 4.11. loss rate over this period (medium confidence) rate of mass loss for higher warming; At sustained warming levels between 3°C and 5°C, near-complete loss of the Greenland Ice Sheet and complete loss of the West Antarctic Ice Sheet will occur irreversibly over multiple millennia (medium confidence); substantial parts or all of Wilkes Subglacial Basin in East Antarctica will be lost over multiple millennia (low confidence). Sections 9.5.1, 12.4.9; Table Glaciers RFC5 See Trends section of this table Continuing substantial global mass loss. 4.11 Sections 4.7.2, Global Ocean 9.6.3; Table RFC5 See Trends section of this table Centennial-scale irreversibility of ocean warming. Temperature 4.11. Sections 4.7.2, Centennial-scale irreversibility of sea level rise. Tipping point linked to ice-sheet SLR RFC5 See Trends section of this table 9.6.3; Table behavior. Deep uncertainty on SLR above 3°C warming 4.11. Atlantic Sections Meridional Low agreement on 20th century trend between models and most There is medium confidence an abrupt collapse will not occur before 2100; for 1.5- 2.3.3.4.1, Overturning RFC5 reconstructions. Observed decline since the mid-2000s cannot be 2, 2-3, 3-5°C warming in 2100, AMOC decline is 29, 32 and 39%, respectively, of 9.2.3.1; Table Circulation distinguished from internal variability (high confidence). its pre-industrial strength. 4.11. (AMOC) Will contribute as a feedback with warming, of approximately 18 +/- 12 Pg C per Section 5.4.8; Permafrost RFC5 See Trends section of this table °C. Possibly nonlinear but low confidence in the value of any threshold for such Table 4.11; Carbon behavior. Likely irreversible at centennial time scales. Box 5.1. Sections 4.3.2, Reversible within years to decades; No tipping point or threshold beyond which Arctic Sea Ice RFC5 Abrupt change already observed 9.3.1; Table loss of ice becomes irreversible (high confidence) 4.11. Snow Cover of Sections Northern RFC5 See Trends section of the table Not anticipated to present tipping point/irreversible behavior. 8.6.2.3, 9.5.3; Hemisphere Table 4.11. Do Not Cite, Quote or Distribute 12-118 Total pages: 227 Final Government Distribution Chapter 12 IPCC AR6 WGI Sections Has likely increased over the last 40 years (medium confidence) and can be Not anticipated to present tipping point/irreversible behaviour, unless AMOC 4.4.1.4, 4.5.1.5, Global Monsoon RFC5 explained by a phase change in Atlantic Multidecadal Variability. collapse occurs. 8.6.1; Table 4.11. Section 4.5.3.2; ENSO RFC5 See Trends section of the table Not anticipated to present tipping point/irreversible behavior. Table 4.11. -1 Methane Methane release from shelf clathrates is <10 TgCH4 yr Section 5.4.8; RFC5 Not anticipated to present tipping point/irreversible behavior. Clathrates Table 4.11. 1 2 [END CROSS-CHAPTER BOX 12.1, TABLE 1 HERE] 3 4 [END CROSS-CHAPTER BOX 12.1] Do Not Cite, Quote or Distribute 12-119 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 12.6 Climate change information in climate services 2 3 Climate services are a significantly evolving source for climate change information to support adaptation, 4 mitigation and risk management decisions. As an evolving field, there are multiple definitions of climate 5 services (cf. Brasseur and Gallardo, 2016). The Global Framework for Climate Services defines a climate 6 service as the provision of climate information to assist decision-making. The service includes appropriate 7 engagement from users and providers, is based on scientifically credible information and expertise, has an 8 effective access mechanism, and responds to user needs (Hewitt et al., 2012) 9 10 IPCC AR5 WGII introduced climate services as bridging the generation and application of climate 11 knowledge, also describing their history and concepts (Jones et al., 2014). Since then, this transdisciplinary 12 field has been growing rapidly (Brasseur and Gallardo, 2016; Hewitt et al., 2020a) with the social sciences in 13 particular pointing out knowledge requirements for co-design and co-development of climate services 14 (Larosa and Mysiak, 2019; Daniels et al., 2020; Steynor et al., 2020). Climate services differ from more 15 research-driven vulnerability, impacts, and adaptation research in their orientation toward decision support 16 (Stone and Meinke, 2005; Ruane et al., 2016; Golding et al., 2019), but overlaps exist (Bruno Soares and 17 Buontempo, 2019). Climate services are often targeted at building resilience to climate-related hazards from 18 near real-time to seasonal and multi-decadal time horizons, to inform adaptation to climate variability and 19 change (Hewitt et al., 2012), widely recognized as an important challenge for sustainable development and 20 risk management (Moss et al., 2010; Jones et al., 2014; Vaughan et al., 2018). This section focuses largely 21 on climate change timescales (past, present and future), which are the focus of AR6 WGI. 22 23 This section introduces the current climate services landscape, assesses climate services practices and 24 products related to climate change information and associated challenges. Cross-chapter box 12.2 provides 25 concrete examples of climate services. The section builds on the introduction to climate services in Section 26 1.6.1.4 and the assessment of regional climate information construction including storylines discussed in 27 sections 10.3.4.2, 10.5.3 and Box 10.2. The Atlas supports the provision of climate information across WGs 28 by providing interactive maps and further details to the material made publicly accessible for use in climate 29 services. WGII (Chapter 17) further elaborates on climate services as enablers for climate risk management. 30 31 32 12.6.1 Context of climate services 33 34 The idea of climate services is not new and has its roots in meteorology and climatology (Larosa and 35 Mysiak, 2019). It can be traced back to the late 1970s and the US National Climate Program Act of 1978 36 (Henderson et al., 2016). The development of the Global Framework for Climate Services (GFCS) after the 37 World Climate Conference-3 in Geneva brought international attention and renewed impetus to the climate 38 services field (Hewitt et al., 2012). As a result, large investments have been made globally and regionally in 39 the development of user-driven climate services. WMO has created Regional Climate Centres (RCCs) to 40 facilitate climate service development by regional and national providers (Hewitt et al., 2020a). The 41 European Union declared its ambition to stimulate “the creation of a community of climate services 42 application developers and users that matches supply and demand for climate information and prediction”, 43 giving primacy to climate services that are user-driven and science-informed (Lourenço et al., 2016), thus 44 embracing concepts of co-design, co-development and co-evaluation of climate services (Street, 2016). 45 Diverse and action-driven international initiatives allowed climate services to progressively shift from 46 mitigation towards adaptation (Larosa and Mysiak, 2019). Opportunities for the development of climate 47 services have emerged through the 2015 Agendas (Paris Agreement, Sustainable Development Goals and 48 Sendai Framework), Nationally Determined Contributions, National Adaptation Plans, Multilateral 49 Development Banks and Task Force on Climate-related Financial Disclosure (see Chapter 1.2.2). 50 51 Scientific advancements in climate services related to meteorology and climatology are still closely linked to 52 essential climate variables (Larosa and Mysiak, 2019) and benefit from consistently growing computational 53 power, infrastructure and storage capacity to meet the demands of higher spatially and temporally resolved 54 climate information (Buontempo et al., 2020). Climate services also focus on impact chains providing 55 decision-makers with information on climate change with cross-sectoral impact assessments for adaptation Do Not Cite, Quote or Distribute 12-120 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 (Jacob and Solman, 2017). Today there is a diversity of climate services that involve interpretation, analysis, 2 and communication of different sources of climate data, ideally combining different types of knowledge 3 (scientific/technical, experiential, indigenous, etc.), to a targeted group of decision-makers (Parris et al., 4 2016; Olazabal et al., 2018; Pezij et al., 2019). (Jacobs and Street, 2020) argue that climate services should 5 be expanded to also address societal challenges, such as system transformation that includes climate in the 6 context of other risks and development challenges. 7 8 Climate services are undertaken in public and private sectors at global, regional, national, and local scales 9 (Hewitt et al., 2012, 2020b; Cortekar et al., 2020). Intermediaries such as private sector consulting 10 companies, national climate service providers as well as research organizations, government agencies or 11 academic institutions provide climate services that translate aspects of climate research to the specific 12 context of decision-makers (see also Chapter 10.5). The EU Roadmap for Climate Services ( 13 14 an Commission, 2015; Street, 2016) focuses on developing a market for climate services comprising of both 15 public and private domains. The GFCS, under the leadership of several United Nations Agencies, 16 emphasizes the public domain by supporting national and regional capacity building and development of 17 climate services mainly through National Meteorological and Hydrological Services (Hewitt et al., 2012; 18 Domingos et al., 2016; Sivakumar and Lucio, 2018; WMO, 2018). There are on-going debates about the 19 commercialization of climate services (Brooks, 2013b; WMO, 2015; Webber and Donner, 2017; Hoa, 2018; 20 Troccoli et al., 2018; Bruno Soares and Buontempo, 2019; Hewitt et al., 2020a). Some argue that the 21 commercialization of climate services is needed to meet the diverse needs of specific clients and to drive 22 innovation in the field (Brooks, 2013b; Troccoli, 2018a). Others argue that if climate services shift incentives 23 for climate science away from the public interest towards profit-seeking, this will result in less publicly 24 accessible and transparent climate information and more private knowledge (Keele, 2019; Tart et al., 2020). 25 26 Some climate adaptation planning already use climate information as provided by the IPCC, however, 27 depending on the decision context, this information may be too coarse, too broad or too disciplinary to 28 directly inform decision-making at the scale where adaptation measures are taken (Howarth and Painter, 29 2016; Nissan et al., 2019). Thus, while the role of IPCC is clearly felt as a reference, authoritative, starting 30 point, there is a need for complementary information to translate the assessments at the national, local or 31 sectoral level (Howarth and Painter, 2016; Kjellström et al., 2016; van den Hurk et al., 2018; Vaughan et al., 32 2018). The AR6 Interactive Atlas (see section Atlas.2) does provide a collection of observational data and 33 global and regional climate projections. It is designed as a climate service towards the WGI needs and 34 beyond to assess the state of the climate by offering data, maps and a level of expert analysis by aggregation 35 of results to regions, scenarios and warming levels. 36 37 38 12.6.2 Assessment of climate services practice and products related to climate change information 39 40 The climate services landscape is fast growing and very broad as reflected in the vast diversity of practices 41 and products that can be found in the peer-reviewed literature (very high confidence). However, a large part 42 of climate services practices and products is published in ‘grey’ literature (i.e. non-peer reviewed or non- 43 academic) by private consultancy and non-scientific civil organizations of which many are not in the public 44 domain. In addition, the respective climate service context of a specific stakeholder in a sector dictates what 45 climate information is required and on what scales and in what format it is most usefully provided. The 46 extent and type of engagement between scientists and users is another critical aspect of climate services (see 47 Cross-chapter box 12.2 Fig 1, and2section 10.5). The assessment here can thus only provide a partial and 48 rather general representation of available practices and products of the evolving climate services field. 49 50 User needs and decision making contexts are very diverse and there is no “one size fits all” solution to 51 climate services (very high confidence) (Hewitt et al., 2017a; Vincent et al., 2018b). In many cases this 52 requires recognizing that stakeholders make decisions through a combination of scientific information and 53 additional values (Vanderlinden et al., 2017; Parker and Lusk, 2019) (see also Section 1.2.3 and Section 54 10.5.4). The emerging climate service literature may clarify some features of climate information requested 55 by users, for instance climatic impact-driver identification and prioritization through stakeholder Do Not Cite, Quote or Distribute 12-121 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 engagement; the specification of thresholds for various regions/sectors; the types of metrics 2 (magnitude/intensity, frequency, duration, timing, spatial extent) that are of primary interest; and decision 3 support systems where informatics allow stakeholders to custom-make impact-relevant thresholds and then 4 query databases to understand current and future characteristics (Bachmair et al., 2016; Buontempo et al., 5 2020). However, users also ask for capacity building activities related to basic knowledge in climate change 6 sciences and climate-related risks (De Bruin et al., 2020; Sultan et al., 2020). 7 8 Since AR5 and SROCC (Chapter 2) there has been considerable progress in understanding climate 9 information user needs (Baztan et al., 2017; Golding et al., 2017b, 2017a, 2019; Bruno Soares et al., 2018a; 10 Hewitt and Golding, 2018; Singh et al., 2018; Sivakumar and Lucio, 2018; Bessembinder et al., 2019; Sultan 11 et al., 2020; Wang et al., 2020b; Hewitt et al., 2020b), better facilitation of user engagement (Buontempo et 12 al., 2014, 2018; Buontempo and Hewitt, 2018) and an appreciation of climate scientists to involve 13 communication specialists and social scientists to support the co-design and co-development process that is 14 fundamental to a successful climate service (Buontempo et al., 2014; Gregow et al., 2016; Damm et al., 15 2019). 16 17 Climate services require user engagement and can take various forms in which climate information and data 18 is delivered or communicated to the users (very high confidence). Different levels of user engagement exist, 19 which can range from passive engagement, interactive group activities, to focused relationships between 20 climate service provider and users, and which result in different types of climate service products including 21 websites, capacity building, and co-design of tailored climate indices (Hewitt et al., 2017b) (see Cross- 22 Chapter Box 12.2, Figure 1). The fundamental basis for climate service development is the co-production 23 process between climate services provider and user (Valiela, 2006; Briley et al., 2015; Golding et al., 2017a; 24 Vincent et al., 2018a; Bruno Soares and Buontempo, 2019; Schipper et al., 2019), which can be very 25 resource intensive (Buontempo et al., 2018; Falloon et al., 2018; Kolstad et al., 2019) and varies strongly 26 from case to case (Reinecke, 2015; Bremer et al., 2019; Goodess et al., 2019; Jung and Schindler, 2019). 27 Climate services scholars and practitioners can better facilitate and embrace the knowledge co-production 28 process if it is recognized as a multi-faceted phenomenon with several dimensions (e.g., constitutive, 29 interactional, institutional, pedagogical, empowerment) (Kruk et al., 2017; Knaggård et al., 2019; 30 Weichselgartner and Arheimer, 2019). 31 32 Information moves from useful to usable only when users effectively incorporate this information into a 33 decision process (Lemos et al., 2012; Bruno Soares and Dessai, 2016; Prokopy et al., 2017) (see also WGII, 34 Chapter 17). Climate services include a range of knowledge brokerage activities such as: identify knowledge 35 needs; dissemination of knowledge; coordinate and network; compile and translate; build capacity through 36 informed decision-making; analyse, evaluate and develop policy; and personal consultation (e.g., (De Bruin 37 et al., 2020). When analysing four European climate services, (Reinecke, 2015) found that different climate 38 services emphasized different knowledge brokerage activities. 39 40 There are various types of providers and products of climate services related to the key sectors and regions 41 such as those described in previous sections (Hewitt et al., 2017a). For instance, studies have described 42 sectoral climate services in support of agriculture (Falloon et al., 2018; Hansen et al., 2019), health (Jancloes 43 et al., 2014; Lowe et al., 2017), tourism (Morin et al., 2018; Damm et al., 2019; Matthews et al., 2020), 44 energy (Troccoli, 2018b; Goodess et al., 2019; Soret et al., 2019), disaster risk reduction (Golding et al., 45 2019; Street et al., 2019), water (van den Hurk et al., 2016; Vano et al., 2018), ocean and coastal ecosystems 46 (Weisse et al., 2015; Le Cozannet et al., 2017), cities (Rosenzweig and Solecki, 2014; Rosenzweig et al., 47 2015; Gidhagen et al., 2020), and cultural heritage (ICOMOS, 2019). Many countries (including almost 48 every country in Europe – see Atlas.5.6.3.4) have established a climate service centre, which follow different 49 practices of user engagement and provide different products (e.g., Kjellström et al., 2016; Kolstad, 2019; 50 Skelton et al., 2017). Climate services in other countries may be distributed across agencies and programs, 51 although these are often not centrally coordinated (Parris et al., 2016). One of the key pillars of the GFCS is 52 the Climate Services Information System (CSIS), which is the principal mechanism through which 53 information about past, present and future climate is archived, analysed, modelled, exchanged and processed 54 for users (Hewitt et al., 2020a). Some national governments also have organized national climate projections 55 to be used for official planning (e.g.,(EEA, 2018). A list of available national products (e.g., observational Do Not Cite, Quote or Distribute 12-122 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 datasets) and projections can be found in the Atlas (e.g., Section Atlas.1.1.3 and 2.3). 2 3 Figure 12.12 maps a general categorization of practices and products that have emerged from reviewing the 4 climate service literature and user interviews (Visscher et al., 2020). The categories range from very generic 5 products or expert analysis focused particularly on climate information (climate-centric approaches) to more 6 integrated products that include shared open-source products and capacity building as well as tailored 7 products that treat climate information as part of a larger decision-making context (climate-inclusive 8 approaches). Three specific examples that elaborate in more detail on specific practices and products related 9 to those general categories are provided in the Cross-chapter box 12.2. 10 11 12 [START FIGURE 12.12 HERE] 13 14 Figure 12.12:Illustration of different types of climate services. Products, for instance, can focus only on climate- 15 related information or can be designed to integrate climate information with other decision-relevant 16 context (vertical axis) and they can be very generic in terms of relevance to a wide range of sectors or 17 stakeholders or customized to fit the needs of a specific sector or stakeholder (horizontal axis). Source: 18 Adapted from (Visscher et al., 2020). 19 20 [END FIGURE 12.12 HERE] 21 22 23 12.6.3 Challenges 24 25 Climate services set new scientific challenges to physical climate research (high confidence). Over at least 26 the last decade, for instance, many questions have appeared in terms of optimal estimation of changes and 27 uncertainties from projections of model ensembles, ensemble optimization, or adjustment of model biases 28 while preserving essential information on trends and cross-variable, time and space consistencies, 29 downscaling information at the local scale (Benestad et al., 2017; Hewitt et al., 2017a; Marotzke et al., 2017; 30 Hewitt and Lowe, 2018; Knutti, 2019) (see also Section 10.5). Other challenges related to climate services 31 are the inter-operability of data (Giuliani et al., 2017), access to data (open/FAIR Guiding principles 32 (Wilkinson et al., 2016; Georgeson et al., 2017)), format of data (including moving away from percentile- 33 based probabilistic forecasts (e.g., Haines, 2019) and funding mechanisms (Bruno Soares and Buontempo, 34 2019). 35 36 Understanding and modeling of weather and climate extremes is of great relevance for climate services and 37 is continuing to set challenges for research, such as modeling changes in impact-relevant threshold 38 exceedance and return periods for a variety of extremes (Maraun et al., 2015; Sillmann et al., 2017; Hewitt et 39 al., 2020c; Schwingshackl et al., 2021) (see also Chapter 11). Extreme event attribution has also been used in 40 context of climate services (Philip et al., 2020) as it is of interest to some stakeholder groups (Sippel et al., 41 2015; Marjanac and Patton, 2018; Jézéquel et al., 2019, 2020). The usefulness or applicability of available 42 extreme event attribution methods (see Cross-chapter box 1.4 and section 11.2.4) for assessing climate- 43 related risks remains subject to debate (Shepherd, 2016; Mann et al., 2017; Lloyd and Oreskes, 2018). 44 45 The design of climate services involves key challenges, such as a domain challenge where users, tasks and 46 data may be unknown; and an informational challenge related to the use and adoption of novel and complex 47 scientific data (Christel et al., 2018). This includes challenges in the uptake of climate information in terms 48 of coordinated delivery of data, information, expertise and training by public research institutes, the 49 inclusion of climate change adaptation in public and private regulation, and uncertainties and confidence in 50 climate projections (Cavelier et al., 2017). Quality control and quality assurance are still weak elements in 51 the development of climate service products (Jacob, 2020). Quality criteria or standards (that go beyond 52 good practice) will have to be developed and agreed (cf. (Pacchetti et al., 2021). These challenges reflect the 53 dilemma that exists at the interface between the climate modelling community and climate services 54 regarding: 1) the purposes of the models for climate research versus service development; 2) the gap between 55 the spatial and temporal scales of the models versus the scales needed in applications; and 3) tailoring Do Not Cite, Quote or Distribute 12-123 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 climate model results to real-world applications (Benestad et al., 2017; Hackenbruch et al., 2017; van den 2 Hurk et al., 2018). 3 4 Climate services require a sustained engagement between scientists, service providers and users that is often 5 hindered by limited resources for the co-design and co-production process (high confidence). There are 6 recurring challenges related to successful climate service applications: 1) climate services are not visible and 7 poorly understood by ‘end users’ (Weichselgartner and Arheimer, 2019); 2) data can be of unknown or poor 8 quality, data formats can be hard to access or process, and it can be difficult to choose from large databases 9 (e.g. Section 1.5.4) without appropriate user guidance; 3) users are unsure how to choose from available 10 climate services to meet their needs (Rössler et al., 2019); 4) building trust between climate service users and 11 providers (Baztan et al., 2020); 5) the lack of understanding of users and their contexts by the climate science 12 and service community (Porter and Dessai, 2017); 6) the difficulty in scaling up services (Tall et al., 2014; 13 van Huysen et al., 2018); 7) the lack of trained scientists skilled at conducting societally-relevant research 14 (Rozance et al., 2020). 15 16 Challenges also arise in determining the effectiveness and added value of climate services, particularly in 17 terms of providing quantitative estimates of economic benefits and making a business case for climate 18 services (Bruno Soares, 2017). The market for climate services is still in its infancy (Cavelier et al., 2017; 19 Bruno Soares et al., 2018b; Tall et al., 2018; Damm et al., 2019). One form of value may be determined by a 20 particular user community’s willingness to pay (Acquah and Onumah, 2011; Ouédraogo et al., 2018; Antwi- 21 Agyei et al., 2021), which however cannot reflect the value of climate services as a public good and for 22 society as a whole (Hewitt et al., 2012). Literature is only recently emerging on the socio-economic benefits 23 of weather and climate services (Vaughan et al., 2019). Early studies and guidelines from the WMO focus on 24 cost-benefit ratios (Perrels et al., 2013; WMO, 2015). Issues related to demand-driven versus supply-driven 25 climate services (Lourenço et al., 2016; Street, 2016; Daniels et al., 2020), public versus private climate 26 services (Hewitt et al., 2020a) and business models for climate services (Hoa, 2018) have been raised. A 27 large share of climate services documented in peer-reviewed literature is currently provided in non-market 28 frameworks (e.g., public service obligations and R&D grants) (Hoa, 2018; Kolstad et al., 2019; Cortekar et 29 al., 2020). 30 31 Other challenges related to governance and dealing with complex systems are sometimes acknowledged but 32 less well described in the climate services domain (Hewitt et al., 2020a). Importantly, decision contexts are 33 strongly rooted in past practice (which often does not even make optimal use of past climate information), 34 stakeholder experience and history. Even important emerging concepts of co-production, entry points, and 35 champions do not always fall naturally into these realities without significant effort. The social sciences have 36 an important role in helping understand and tackle these challenges (Bruno Soares and Buontempo, 2019). 37 38 39 [START CROSS-CHAPTER BOX 12.2 HERE] 40 41 Cross-Chapter Box 12.2: Climate services and climate change information 42 43 Contributing Authors: 44 Suraje Dessai (UK/Portugal), Jana Sillmann (Norway/Germany), Carlo Buontempo (UK/Italy), Cecilia 45 Conde (Mexico), Aida Diongue-Niang (Senegal), Francisco J. Doblas-Reyes (Spain), Christopher Jack 46 (South Africa), Richard Jones (UK), Benjamin Lamptey (Niger/Ghana), Xianfu Lu (UK/China), Douglas 47 Maraun (Austria/Germany), Ben Orlove (USA), Roshanka Ranasinghe (Netherlands/Sri Lanka/Australia), 48 Alex C. Ruane (USA), Anna Steynor (South Africa), Bart van den Hurk (Netherlands), Robert Vautard 49 (France) 50 51 Climate services involve the provision of climate information in such a way as to assist decision-making. 52 The service needs to have appropriate engagement from users and providers, be based on scientifically 53 credible information and expertise, have an effective access mechanism, and meet the users’ needs (Hewitt et 54 al., 2012). Predominantly, climate services are targeted at informing and enabling risk management in 55 adaptation to climate variability and change (Jones et al., 2014; Vaughan et al., 2018). Chapter 1 introduces Do Not Cite, Quote or Distribute 12-124 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 climate services in a broader context of interaction between science and society, including how climate 2 information can be tailored and co-produced for greatest utility in specific contexts. Chapter 10 assesses the 3 key foundations for the generation of climate information about regional climate change. Chapters 11, 12 4 and Atlas comprehensively assess regional climate change information. The Interactive Atlas gives access to 5 various repositories of quantitative climate information. In WG2, Chapter 17 assesses climate services in the 6 context of climate risk management. 7 8 Climate services contexts are diverse and complex. They can be characterised using different factors such as 9 sectors, regions, purposes, time horizons, data sources, level of processing of climate data, background 10 knowledge, type of climate services providers, as well as the nature of the interactions between providers, 11 users and other stakeholders (Bessembinder et al., 2019). To illustrate the wide diversity of climate change 12 information in climate services, a useful categorisation is by user-provider engagement of climate services 13 (Cross-Chapter Box 12.2, Figure 1). One broad category includes Websites and web tools which generally 14 focuses on data and information provision (Hewitson et al., 2017). Websites are generally able to reach many 15 users, but engagement is passive through one-way transfer of information. The second broad category 16 involves Interactive group activities, such as workshops, meetings and interactive forums, which create a 17 stronger dialogue between climate service providers and decision-makers. Multi-way communication and 18 regular interaction enable building of trust, co-learning and co-production of products and services. The third 19 broad category involves Focused relationships which are tailored, targeted and address very specific needs 20 of the user. Effective engagement arises from an iterative process between the provider and user to ensure 21 the user’s needs are being addressed appropriately (Hewitt et al., 2017b). 22 23 The diversity of climate services practices and products is illustrated here using three case studies that 24 represent each of the broad categories of provider-user engagement (Cross-chapter box Figure 1). 25 26 27 [START CROSS-CHAPTER BOX 12.2, FIGURE 1 HERE] 28 29 Cross-Chapter Box 12.2, Figure 1: Schematic of three broad categories of engagement between users and providers of 30 climate services (adapted from Hewitt et al., (2017b). 31 32 [END CROSS-CHAPTER BOX 12.2, FIGURE 1 HERE] 33 34 35 Case study 1 – Websites and web tools. The Copernicus Climate Change Service (C3S) provides free and 36 open access to climate data, tools and information through a website. It also includes demonstration projects 37 that show how C3S data can be used in practice through case studies, training sessions and workshops 38 (Thepaut et al., 2018). A large audience of the service is composed of intermediate users, loosely defined as 39 the community of operators in one of the intermediate steps between the primary producers of climate data 40 and the ultimate beneficiaries. 41 42 To address this audience, the strategy of C3S is to provide free and open access climate data and tools such 43 as historical observations (both satellite and ground-based), climate data records relevant for a number of 44 Essential Climate Variables (Bojinski et al., 2014), global and regional reanalyses, climate monitoring 45 bulletins, seasonal predictions, as well as both global (a selection of simulations from the Coupled Model 46 Intercomparison Project (CMIP (Taylor et al., 2012; Eyring et al., 2016)) and regional climate projections 47 from the Coordinated Regional Downscaling Experiment (Euro- and Med-CORDEX (Jacob et al., 2014; 48 Ruti et al., 2016)). A number of indices for various sectors can be calculated through cloud-based tools. For 49 instance, in order to address the specific needs of key sectoral users, climate impact indicators for common 50 variables such as “heating degree-days” can be calculated by the users and made available to others 51 (Buontempo et al., 2020). All this material is quality controlled following a standardised, transparent and 52 traceable framework. 53 54 C3S also facilitates the tailoring process, by providing a series of working open-source on-the-cloud 55 demonstrators which show how climate data can be transformed into actionable information to meet specific Do Not Cite, Quote or Distribute 12-125 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 user requirements. This tailoring process covers the chain between the definition of key indicators all the 2 way to the user interface. The C3S products definition and production involve scientists that produced and 3 assessed the data. A variety of potential users are involved in the definition of indicators and other products. 4 5 Through its quality assurance process and demonstrators, C3S provides a basic evaluation of all climate data 6 it provides access, and it encourages the users to develop their own case-specific analysis within the C3S 7 infrastructure. Trustworthiness and relevance of such an analysis are substantially strengthened through a 8 distillation process, co-designed by the user and data provider, and drawing upon multiple lines of evidence 9 and process-based evaluation of model fitness (Section 10.5). 10 11 Case study 2 – Interactive group activities. Science-application engagement is extremely challenging 12 especially in critically important but complex contexts such as rapidly growing cities in developing nations 13 (Culwick and Patel, 2017). The publicly funded Future Resilience for African CiTies And Lands 14 (FRACTAL) project was conceived and designed in response to extensive and strong evidence and 15 experience that useful and useable climate services require strong mutual relationships across the science- 16 application interface that can be built using supportive processes and structures (Taylor et al., 2017). 17 18 Informed by this understanding, FRACTAL was grounded in a very reflexive and context guided approach 19 with city decision making at its core (Taylor et al., 2017a). Representatives from selected southern African 20 cities were included in the proposal design and, throughout the project, a core principle was to allow the city 21 partners to lead and guide the process. 22 23 Two important elements were deployed in FRACTAL: “embedded researchers” and “learning labs”. 24 Embedded Researchers were seconded into the municipality and served as the essential connection for the 25 learning process within each city (Steynor et al., 2020). Learning Labs (Arrighi et al., 2016) were interactive 26 structures in which participants from academia, local city government and councils, state-owned enterprises, 27 communities and community development institutions, etc could interact. Embedded Researchers and 28 Learning Labs were the backbone of ongoing learning processes within each city and resulted in more 29 focused small-group dialogues, capacity development and training processes, and within-city research and 30 engagement activities. Each Learning Lab focused initially on identifying “burning issues” without a 31 requirement that they involve strong climate linkages. However, with the over-arching focus on resilience, 32 discussions evolved in that direction and the burning issues identified often centered around water in peri- 33 urban areas e.g., in Windhoek (Scott et al., 2018). 34 35 The learning labs also introduced and developed the concept of Climate Risk Narratives as a process and 36 product (CRN; see Cross-Chapter Box 12.2, Figure 2 and Chapter 10 Box 10.2 on storylines) to generate and 37 integrate climate and socio-economic information relevant to adaptation and resilience (Jack et al., 2020). 38 The first CRNs were informed primarily by climate evidence, but also included some tentative socio- 39 economic impact elements gleaned from literature and other studies. Their content was intentionally 40 provocative and designed to promote debate and discussion, and subsequent iteration. Many participants 41 noted that this was the first time that various important conversations across governance structures and 42 disciplinary areas had occurred around what climate change may actually mean. This demonstrates the 43 engagement value of CRNs as a key element in an iterative co-production process to ensure important details 44 are included correctly, such as the local context, terms and names as well as providing reality checks on the 45 impacts and societal responses (Jack et al., 2020). 46 47 This case study emphasises the positive contributions of the fit, tailoring and contextualization of climate 48 information with respect to the specific decision-making needs of particular users (10.5 and WG2 49 17.4.4.2.2), the importance of participatory planning for risk management in urban areas (WG2 6.1.5 and 50 6.4.2) and the importance of networks and organizations which link researchers, policy-makers and end- 51 users to promote adaptation in African cities (WG2 9.8.5.1). Upscaling this type of interactive activity to 52 cater for the large number of user demands remains a challenge. 53 54 55 Do Not Cite, Quote or Distribute 12-126 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 [START CROSS-CHAPTER BOX 12.2, FIGURE 2 HERE] 2 3 Cross-Chapter Box 12.2, Figure 2: Climate Risk Narrative infographic developed through the FRACTAL Windhoek 4 Learning Lab process (adapted from Jack et al., (2020)). 5 6 [END CROSS-CHAPTER BOX 12.2, FIGURE 2 HERE] 7 8 9 Case study 3 – Focused relationships. This broad category involves one-to-one engagement between a 10 provider and a user with very specific needs. One such user is the Asian Development Bank (ADB) who has 11 committed to making all its investments climate resilient by implementing a climate risk management 12 framework (ADB, 2014, 2018; Lu, 2019). The climate risk management framework mandates all climate- 13 sensitive investment projects undertake a climate risk and adaptation assessment, to identify material risks of 14 a changing climate to the proposed project and potential adaptive measures to be incorporated into project 15 design, implementation, maintenance and/or monitoring. Typically, loan project processing teams procure 16 consulting services for a bespoke climate risk and adaptation assessment (CRA) for a specific project. The 17 user-provider engagement is highly targeted and goal-oriented. An example of such a focused user-provider 18 engagement is the CRA carried out as part of an investment project in Viet Nam, the Water Efficiency 19 Improvement in Drought-Affected Provinces (WEIDAP) project. 20 21 In the wake of the El Niño-induced 2015-2016 severe drought causing major damage to agricultural land in 22 the Central Highlands of Viet Nam, the WEIDAP project was initiated to improve water productivity of 23 irrigated agriculture. Proposed project interventions include a package of both “soft” (e.g., policy, 24 institutional and capacity building, on-farm water efficiency practices) and “hard” (modernized irrigation 25 schemes) activities. To ensure that the project delivers expected benefits under a changing climate, 26 consultants were recruited to carry out a detailed CRA, working as part of the overall project processing 27 team. Through extensive consultations with the rest of the project team and review of literature including 28 relevant climate projections, the CRA consultants chose to construct three broad climate scenarios for the 29 2050s (a time frame appropriate for the lifetime of the irrigation schemes being proposed under the project): 30 a warm-and-wet, a hot-and-wet, and a hotter future. Outputs from a selection of CMIP5 models were 31 analysed under these three scenarios, to derive changes in temperature, rainfall and potential 32 evapotranspiration, which in turn were used as inputs to hydrological, crop and agro-economics models to 33 assess the impacts of climate change on the overall project performance. Table 1 presents the summary of the 34 key parameters under the three scenarios. Recommendations from the CRA included (largely minor) 35 refinements and additional activities for drought planning, detailed engineering design of the relevant project 36 components (such as access roads, river crossings and foundations), and support for poorer farmers who may 37 not be able to afford access to water and climate-resilient technologies. 38 39 This case study illustrates that climate information distillation including a sustained iterative engagement 40 between climate information users, producers and translators can improve the quality of the information and 41 the decision-making (Section 10.5, WG2 17.4.4.2.2). 42 43 44 [START CROSS-CHAPTER BOX 12.2, TABLE 1 HERE] 45 46 Cross-Chapter Box 12.2, Table 1: Summary of Annual Province-Level Changes in Temperature, Precipitation 47 and Evapotranspiration under the Three Broad Scenarios in Southern Viet 48 Nam (Scenario 1: warm-and-wet; Scenario 2: hot-and-wet; Scenario 3: hotter) 49 (Source: Table 3 in (ADB, 2020)) 50 51 52 53 54 55 56 Do Not Cite, Quote or Distribute 12-127 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI Province and Scenario Number Binh Thuan Dak Lak Dak Nong Khanh Hoa Ninh Thuan Item 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 ΔT (°C) 1.1 1.8 2.6 1.1 1.5 2.0 1.2 2.1 2.7 1.1 1.8 2.6 1.1 1.5 2.6 ΔP (%) 28 -12 4 8 17 -8 8 -8 7 3 -10 7 27 1 5 ΔPET (%) 3 6 8 4 5 7 4 7 9 3 6 8 3 5 8 1 ΔP = change in precipitation, ΔPET = change in potential evapotranspiration, ΔT = change in temperature. 2 Note: Color scale indicates significance of changes for the water balance. Gray = no significant change, green = 3 medium positive impact, blue = high positive impact, yellow = medium negative impact, orange = high negative impact 4 5 [END CROSS-CHAPTER BOX 12.2, TABLE 1 HERE] 6 7 [END CROSS-CHAPTER BOX 12.2 HERE] 8 9 10 12.7 Final Remarks 11 12 The assessment in this chapter is based on a rapidly growing body of new evidence from the peer-reviewed 13 literature, direct calculations of climate projections from several new model ensembles, and results from 14 other AR6 WGI chapters. Although a large amount a new information on CID changes and their uptake in 15 climate services has become available since AR5, some challenges still remain. This section summarizes 16 some of these main challenges, with a view to facilitating improved assessments in future. The section is 17 organized following the order of chapter sections and consolidated according to key assessment components. 18 19 • The adoption of the climatic impact-driver (CID) framework could benefit more from stronger 20 connections across physical climate and impact scientists, and between the science community and 21 practitioners/stakeholders on the ground. Co-development of CID index definitions with impact 22 scientists or stakeholders helps ensure their salience and utility {12.1, 12.2, 12.3, 12.6}. 23 24 • The ability to project all aspects of shifting CID profiles and their effects at fine, local scales is often 25 reliant on dynamical downscaling and additional impact modelling steps, making a robust and full 26 quantification of the uncertainties involved more challenging. Availability of multiple models and 27 ease of connecting physical climate models at different scales can facilitate assessment {12.2, 12.3, 28 12.4, 12.5}. 29 30 • Regional and sub-regional differences in coverage and access of homogeneous historical records, in 31 the deployment of regional model ensembles and the exploration of scenarios, and ultimately in 32 peer-reviewed studies addressing the full range of past and current behavior, detection and 33 attribution, and future projections challenge a uniformly robust assessment across all CIDs and 34 regions of the world {12.4, 12.5}. 35 36 • Efforts to assess a consistent global, large scale view of CID changes across regions and sectors 37 would benefit from additional coordinated studies adopting common CID indices, model protocols, 38 time horizons and scenarios or global warming levels {12.3, 12.5}. 39 40 • Even though the body of peer-reviewed literature regarding climate services practices and products 41 is growing, a large part is still documented only in grey literature arising from commercial 42 consultancy, and thus not publicly and freely accessible {12.6}. Do Not Cite, Quote or Distribute 12-128 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 Frequently Asked Questions 2 3 FAQ 12.1: What is a climatic impact-driver (CID)? 4 5 A climatic impact-driver is a physical climate condition that directly affects society or ecosystems. Climatic 6 impact-drivers may represent a long-term average condition (such as the average winter temperatures that 7 affect indoor heating requirements), a common event (such as a frost that kills off warm-season plants), or 8 an extreme event (such as a coastal flood that destroys homes). A single climatic impact-driver may lead to 9 detrimental effects for one part of society while benefiting another, while others are not affected at all. A 10 climatic impact-driver (or its change caused by climate change) is therefore not universally hazardous or 11 beneficial, but we refer to it as a ‘hazard’ when experts determine it is detrimental to a specific system. 12 13 Climate change can alter many aspects of the climate system, but efforts to identify impacts and risks usually 14 focus on a smaller set of changes known to affect, or potentially affect, things that society cares about. 15 These climatic impact-drivers (CIDs) are formally defined in this Report as ‘physical climate system 16 conditions (e.g., means, events, extremes) that affect an element of society or ecosystems. Depending on 17 system tolerance, CIDs and their changes can be detrimental, beneficial, neutral, or a mixture of each across 18 interacting system elements and regions’. Because people, infrastructure and ecosystems interact directly 19 with their immediate environment, climate experts assess CIDs locally and regionally. CIDs may relate to 20 temperature, the water cycle, wind and storms, snow and ice, oceanic and coastal processes or the chemistry 21 and energy balance of the climate system. Future impacts and risk may also be directly affected by factors 22 unrelated to the climate (such as socio-economic development, population growth, or a viral outbreak) that 23 may also alter the vulnerability or exposure of systems. 24 25 CIDs capture important characteristics of the average climate and both common and extreme events that 26 shape society and nature (see FAQ 12.2). Some CIDs focus on aspects of the average climate (such as the 27 seasonal progression of temperature and precipitation, average winds and the chemistry of the ocean) that 28 determine, for example, species distribution, farming systems, the location of tourist resorts, the availability 29 of water resources and the expected heating and cooling needs for buildings in an average year. CIDs also 30 include common episodic events that are particularly important to systems, such as thaw events that can 31 trigger springtime plant development, cold spells that are important for fruit crop chill requirements, or frost 32 events that eliminate summer vegetation as winter sets in. Finally, CIDs include many extreme events 33 connected to impacts such as hailstorms that damage vehicles, coastal floods that destroy shoreline property, 34 tornadoes that damage infrastructure, droughts that increase competition for water resources, and heatwaves 35 that can strain the health of outdoor laborers. 36 37 Many aspects of our daily lives, businesses and natural systems depend on weather and climate, and there is 38 great interest in anticipating the impacts of climate change on the things we care about. To meet these needs, 39 scientists engage with companies and authorities to provide climate services – meaningful and possibly 40 actionable climate information designed to assist decision-making. Climate science and services can focus on 41 CIDs that substantially disrupt systems to support broader risk management approaches. A single CID 42 change can have dramatically different implications for different sectors or even elements of the same sector, 43 so engagement between climate scientists and stakeholders is important to contextualize the climate changes 44 that will come. Climate services responding to planning and optimization of an activity can focus on more 45 gradual changes in climate operating conditions. 46 47 FAQ 12.1, Figure 1 tracks example outcomes of seasonal snow cover changes that connect climate science to 48 the need for mitigation, adaptation and regional risk management. The length of the season with snow on the 49 ground is just one of many regional climate conditions that may change in the future, and it becomes a CID 50 because there are many elements of society and ecosystems that rely on an expected seasonality of snow 51 cover. Climate scientists and climate service providers examining human-driven climate change may 52 identify different regions where the length of the season with snow cover could increase, decrease, or stay 53 relatively unaffected. In each region, change in seasonal snow cover in turn may affect different systems in 54 beneficial or detrimental ways (in the latter case, changing seasonal snow cover would be a ‘hazard’), 55 although systems such as coastal aquaculture remain relatively unaffected. The changing profile of benefits Do Not Cite, Quote or Distribute 12-129 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 and hazards connected to these changes in the seasonal snow cover CID in turn affects the profile of impacts, 2 risks and benefits that stakeholder in the region manage in response to climate change. 3 4 5 [START FAQ12.1, FIGURE 1 HERE] 6 7 FAQ 12.1, Figure 1: A single climatic impact-driver can affect ecosystems and society in different ways. A variety 8 of impacts from the same climatic impact-driver change, illustrated with the example of regional 9 seasonal snow cover. 10 11 [END FAQ12.1, FIGURE 1 HERE] 12 13 14 Do Not Cite, Quote or Distribute 12-130 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 FAQ 12.2: What are climatic thresholds and why are they important? 2 3 Climatic thresholds tell us about the tolerance of society and ecosystems so that we can better scrutinize the 4 types of climate changes that are expected to impact things we care about. Many systems have natural or 5 structural thresholds. If conditions exceed those thresholds, the result can be sudden changes or even 6 collapses in health, productivity, utility or behavior. Adaptation and risk management efforts can change 7 these thresholds, altering the profile of climate conditions that would be problematic and increasing overall 8 system resilience. 9 10 Decision makers have long observed that certain weather and climate conditions can be problematic, or 11 hazardous, for things they care about (i.e., things with socio-economic, cultural or intrinsic value). Many 12 elements of society and ecosystems operate in a suitable climate zone selected naturally or by stakeholders 13 considering the expected climate conditions. However, as climate change moves conditions beyond expected 14 ranges, they may cross a climatic ‘threshold’ – a level beyond which there are either gradual changes in 15 system behaviour or abrupt, non-linear and potentially irreversible impacts. 16 17 Climatic thresholds can be associated with either natural or structural tolerance levels. Natural thresholds, for 18 instance, include heat and humidity conditions above which humans cannot regulate their internal 19 temperatures through sweat, drought durations that heighten competition between species, and winter 20 temperatures that are lethal for pests or disease-carrying vector species. Structural thresholds include 21 engineered limits of drainage systems, extreme wind speeds that limit wind turbine operation, the height of 22 coastal protection infrastructure, and the locations of irrigation infrastructure or tropical cyclone sheltering 23 facilities. 24 25 Thresholds may be defined according to raw values (such as maximum temperature exceeding 35℃) or 26 percentiles (such as the local 99th percentile daily rainfall total). They also often have strong seasonal 27 dependence (see FAQ 12.3). For example, the amount of snowfall that a deciduous tree can withstand 28 depends on whether the snowfall occurs before or after the tree sheds its leaves. Most systems respond to 29 changes in complex ways, and those responses are not determined solely or precisely by specific thresholds 30 of a single climate variable. Nonetheless, thresholds can be useful indicators of system behaviours, and an 31 understanding of these thresholds can help inform risk management decisions. 32 33 34 FAQ 12.2 Figure 1 illustrates how threshold conditions can help us understand climate conditions that are 35 suitable for normal system operation and the thresholds beyond which impacts occur. Crops tend to grow 36 most optimally within a suitable range of daily temperatures that is influenced by the varieties being 37 cultivated and the way the farm is managed. As daily temperatures rise above a ‘critical’ temperature 38 threshold, plants begin to experience heat stress that reduces growth and may lower resulting yields. If 39 temperatures reach a higher ‘limiting’ temperature threshold, crops may suffer leaf loss, pollen sterility, or 40 tissue damage that can lead to crop failure. Farmers typically select a cropping system with some 41 consideration to the probability of extreme temperature events that may occur within a typical season, and so 42 identifying hot temperature thresholds helps farmers select their seed and field management strategies as part 43 of their overall risk management. Climate experts may therefore aim to assist farm planning by providing 44 information about the climate change-induced shifts to the expected frequency of daily heat extremes that 45 exceed crop tolerance thresholds. 46 47 Adaptation and other changes in societies and environment can shift climatic thresholds by modifying 48 vulnerability and exposure. For example, adaptation efforts may include breeding new crops with higher heat 49 tolerance levels so that corresponding dangerous thresholds occur less frequently. Likewise, increasing the 50 height of a flood embankment protecting a given community can increase the level of river flow that may be 51 tolerated without flooding, reducing the frequency of damaging floods. Stakeholders therefore benefit from 52 climate services that are based on a co-development process, with scientists identifying system-relevant 53 thresholds and developing tailored climatic impact-driver indices that represent these thresholds (FAQ 12.1). 54 These thresholds help focus the provision of action-relevant climate information for adaptation and risk 55 management. Do Not Cite, Quote or Distribute 12-131 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 [START FAQ12.2, FIGURE 1 HERE] 2 3 FAQ12.2, Figure 1: Crop response to maximum temperature thresholds. Crop growth rate responds to daily 4 maximum temperature increases, leading to reduced growth and crop failure as temperatures 5 exceed critical and limiting temperature thresholds, respectively. Note that changes in other 6 environmental factors (such as carbon dioxide and water) may increase the tolerance of plants to 7 increasing temperatures. 8 9 [END FAQ12.2, FIGURE 1 HERE] 10 11 Do Not Cite, Quote or Distribute 12-132 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 FAQ 12.3: How will climate change affect the regional characteristics of a climate hazard? 2 3 Human-driven climate change can alter the regional characteristics of climate hazard by changing the 4 magnitude or intensity of a climate hazard, the frequency with which it occurs, the duration that hazardous 5 conditions persist, the timing when a hazard occurs, or the spatial extent threatened by a hazard. By 6 examining each of these aspects of a hazard’s profile change, climate services may provide climate risk 7 information that allows decision makers to better tailor adaptation, mitigation and risk management 8 strategies. 9 10 A climate hazard is a climate condition with the potential to harm natural systems or society. Examples 11 include heatwaves, droughts, heavy snowfall events and sea level rise. Climate scientists look for patterns in 12 climatic impact-drivers to detect the signature of changing hazards that may influence stakeholder planning 13 (FAQ 12.1). Climate service providers work with stakeholders and impacts experts to identify key system 14 responses and tolerance thresholds (FAQ 12.2) and then examine historical observations and future climate 15 projections to identify associated changes to the characteristics of a regional hazard’s profile. Climate change 16 can alter at least five different characteristics of the hazard profile of a region (FAQ 12.3, Figure 1): 17 18 Magnitude or intensity is the raw value of a climate hazard, such as an increase in the maximum yearly 19 temperature or in the depth of flooding that results from a coastal storm with a 1% change of occurring each 20 year. 21 22 Frequency is the number of times that a climate hazard reaches or surpasses a threshold over a given period. 23 For example, increases to the number of heavy snowfall events, tornadoes, or floods experienced in a year or 24 in a decade. 25 26 Duration is the length of time over which hazardous conditions persist beyond a threshold, such as an 27 increase in the number of consecutive days where maximum air temperature exceeds 35°C, the number of 28 consecutive months of drought conditions, or the number of days that a tropical cyclone affects a location. 29 30 Timing captures the occurrence of a hazardous event in relation to the course of a day, season, year, or other 31 period in which sectoral elements are evolving or co-dependent (such as the time of year when migrating 32 animals expect to find a seasonal food supply). Examples include a shift toward an earlier day of the year 33 when the last spring frost occurs or a delay in the typical arrival date for the first seasonal rains, the length of 34 the winter period when the ground is typically covered by snow, or a reduction in the typical time needed for 35 soil moisture to move from normal to drought conditions. 36 37 Spatial extent is the region in which a hazardous condition is expected, such as the area currently threatened 38 by tropical cyclones, geographical areas where the coldest day of the year restricts a particular pest or 39 pathogen, terrain where permafrost is present, the area that would flood following a common storm, zones 40 where climate conditions are conducive to outdoor labour, or the size of a marine heatwave. 41 42 Hazard profile changes are often intertwined or stem from related physical changes to the climate system. 43 For example, changes in the frequency and magnitude of extreme events are often directly related to each 44 other as a result of atmospheric dynamics and chemical processes. In many cases, one aspect of hazard 45 change is more apparent than others, which may provide a first emergent signal indicating a larger set of 46 changes to come (FAQ1.2). 47 48 Information about how a hazard has changed or will change helps stakeholders prioritize more robust 49 adaptation, mitigation and risk management strategies. For example, allocation of limited disaster relief 50 resources may be designed to recognize that tropical cyclones are projected to become more intense even as 51 the frequency of those storms may not change. Planning may also factor in the fact that even heatwaves that 52 are not record-breaking in their intensity can still be problematic for vulnerable populations when they 53 persist over a long period. Likewise, firefighters recognize new logistical challenges in the lengthening of the 54 fire weather season and an expansion of fire conditions into parts of the world where fires were not 55 previously a great concern. Strong engagement between climate scientists and stakeholders therefore helps Do Not Cite, Quote or Distribute 12-133 Total pages: 227 Final Government Draft Chapter 12 IPCC AR6 WGI 1 climate services tailor and communicate clear information about the types of changing climate hazards to be 2 addressed in resilience efforts. 3 4 5 [START FAQ12.3, FIGURE 1 HERE] 6 7 FAQ 12.3, Figure 1: Types of changes to a region’s hazard profile. The first five panels illustrate how climate 8 changes can alter a hazard’s intensity (or magnitude), frequency, duration, and timing (by 9 seasonality and speed of onset) in relation to a hazard threshold (horizontal grey line). The 10 difference between the historical climate (blue) and future climate (red) shows the changing 11 aspects of climate change that stakeholders will have to manage. The bottom-right panel shows 12 how a given climate hazard (such as a once-in-100-year river flood) may reach new geographical 13 areas under a future climate change. 14 15 [END FAQ12.3, FIGURE 1 HERE] 16 Do Not Cite, Quote or Distribute 12-134 Total pages: 227