Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Table of contents 2 3 Executive Summary ................................................................................................................................. 5 4 5 1.1 Report and chapter overview .......................................................................................................... 8 6 1.1.1 The AR6 WGI Report ................................................................................................................... 8 7 1.1.2 Rationale for the new AR6 WGI structure and its relation to the previous AR5 WGI Report.......... 9 8 1.1.3 Integration of AR6 WGI assessments with other Working Groups ............................................... 12 9 1.1.4 Chapter preview .......................................................................................................................... 13 10 11 1.2 Where we are now .......................................................................................................................... 13 12 1.2.1 The changing state of the physical climate system ....................................................................... 13 13 1.2.1.1 Recent changes in multiple climate indicators............................................................................. 14 14 1.2.1.2 Long-term perspectives on anthropogenic climate change........................................................... 15 15 1.2.2 The policy and governance context ............................................................................................. 18 16 17 Cross-Chapter Box 1.1: The WGI contribution to the AR6 and its potential relevance for the global 18 stocktake ................................................................................................................ 19 19 20 1.2.3 Linking science and society: communication, values, and the IPCC assessment process .............. 29 21 1.2.3.1 Climate change understanding, communication, and uncertainties .............................................. 29 22 23 Box 1.1: Treatment of uncertainty and calibrated uncertainty language in AR6 ............................ 30 24 25 1.2.3.2 Values, science, and climate change communication .................................................................. 32 26 1.2.3.3 Climate information, co-production, and climate services ........................................................... 34 27 1.2.3.4 Media coverage of climate change .............................................................................................. 35 28 29 1.3 How we got here: the scientific context ......................................................................................... 36 30 1.3.1 Lines of evidence: instrumental observations ............................................................................... 36 31 1.3.2 Lines of evidence: paleoclimate .................................................................................................. 39 32 1.3.3 Lines of evidence: identifying natural and human drivers ............................................................ 40 33 1.3.4 Lines of evidence: understanding and attributing climate change ................................................. 43 34 1.3.5 Projections of future climate change ............................................................................................ 45 35 1.3.6 How do previous climate projections compare with subsequent observations? ............................. 48 36 37 Box 1.2: Special Reports in the sixth IPCC assessment cycle: key findings...................................... 50 38 39 1.4 AR6 foundations and concepts ...................................................................................................... 53 40 1.4.1 Baselines, reference periods and anomalies ................................................................................. 53 41 42 Cross-Chapter Box 1.2: Changes in global temperature between 1750 and 1850 ................................. 55 43 44 1.4.2 Variability and emergence of the climate change signal ............................................................... 56 Do Not Cite, Quote or Distribute 1-2 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.4.2.1 Climate variability can influence trends over short periods .......................................................... 57 2 1.4.2.2 The emergence of the climate change signal ............................................................................... 57 3 1.4.3 Sources of uncertainty in climate simulations .............................................................................. 59 4 1.4.3.1 Sources of uncertainty ................................................................................................................ 59 5 1.4.3.2 Uncertainty quantification .......................................................................................................... 60 6 1.4.4 Considering an uncertain future................................................................................................... 61 7 1.4.4.1 Low-likelihood outcomes ........................................................................................................... 62 8 1.4.4.2 Storylines ............................................................................................................................... 62 9 10 Cross-Chapter Box 1.3: Risk framing in IPCC AR6 .............................................................................. 63 11 12 1.4.4.3 Abrupt change, tipping points and surprises ................................................................................ 65 13 14 Cross-Working Group Box: Attribution................................................................................................. 67 15 16 1.4.5 Climate regions used in AR6 ....................................................................................................... 70 17 1.4.5.1 Defining climate regions ............................................................................................................ 70 18 1.4.5.2 Types of regions used in AR6..................................................................................................... 71 19 20 1.5 Major developments and their implications .................................................................................. 72 21 1.5.1 Observational data and observing systems ................................................................................... 72 22 1.5.1.1 Major expansions of observational capacity................................................................................ 73 23 1.5.1.2 Threats to observational capacity or continuity ........................................................................... 77 24 1.5.2 New developments in reanalyses ................................................................................................. 78 25 1.5.3 Climate Models ........................................................................................................................... 82 26 1.5.3.1 Earth System Models .................................................................................................................. 82 27 1.5.3.2 Model tuning and adjustment ..................................................................................................... 84 28 1.5.3.3 From global to regional models .................................................................................................. 85 29 1.5.3.4 Models of lower complexity ....................................................................................................... 86 30 31 Box 1.3: Emission metrics in AR6 WGI............................................................................................. 88 32 33 1.5.4 Modelling techniques, comparisons and performance assessments ............................................... 89 34 1.5.4.1 Model ‘fitness for purpose’ ......................................................................................................... 89 35 1.5.4.2 Ensemble modelling techniques ................................................................................................. 89 36 1.5.4.3 The sixth phase of the Coupled Model Intercomparison Project (CMIP6) ................................... 91 37 1.5.4.4 Coordinated Regional Downscaling Experiment (CORDEX)...................................................... 93 38 1.5.4.5 Model Evaluation Tools ............................................................................................................. 94 39 1.5.4.6 Evaluation of process-based models against observations ........................................................... 94 40 1.5.4.7 Emergent constraints on climate feedbacks, sensitivities and projections .................................... 95 41 1.5.4.8 Weighting techniques for model comparisons ............................................................................. 96 42 Do Not Cite, Quote or Distribute 1-3 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.6 Dimensions of Integration: Scenarios, global warming levels and cumulative carbon emissions 97 2 1.6.1 Scenarios ............................................................................................................................... 98 3 1.6.1.1 Shared Socio-economic Pathways ............................................................................................ 100 4 5 Cross-Chapter Box 1.4: The SSP scenarios as used in Working Group I ............................................ 102 6 7 1.6.1.2 Scenario generation process for CMIP6 .................................................................................... 107 8 1.6.1.3 History of scenarios within the IPCC ........................................................................................ 108 9 1.6.1.4 The likelihood of reference scenarios, scenario uncertainty and storylines ................................ 109 10 1.6.2 Global warming levels .............................................................................................................. 111 11 1.6.3 Cumulative CO2 emissions ........................................................................................................ 112 12 13 Box 1.4: The relationships between ‘net zero’ emissions, temperature outcomes and carbon dioxide 14 removal ............................................................................................................................. 113 15 16 1.7 Final remarks ............................................................................................................................. 114 17 18 Frequently Asked Questions .................................................................................................................. 116 19 FAQ 1.1: Do we understand climate change better now compared to when the IPCC started? 116 20 FAQ 1.2: Where is climate change most apparent?................................................................ 117 21 FAQ 1.3: What can past climate teach us about the future? ................................................... 118 22 23 Acknowledgements ............................................................................................................................. 120 24 25 References ............................................................................................................................. 121 26 27 Appendix 1.A ............................................................................................................................. 165 28 29 Figures ............................................................................................................................. 175 30 31 Do Not Cite, Quote or Distribute 1-4 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Executive Summary 2 3 Working Group I (WGI) of the Intergovernmental Panel on Climate Change (IPCC) assesses the current 4 evidence on the physical science of climate change, evaluating knowledge gained from observations, 5 reanalyses, paleoclimate archives and climate model simulations, as well as physical, chemical and 6 biological climate processes. This chapter sets the scene for the WGI assessment, placing it in the context of 7 ongoing global and regional changes, international policy responses, the history of climate science and the 8 evolution from previous IPCC assessments, including the Special Reports prepared as part of this 9 Assessment Cycle. Key concepts and methods, relevant recent developments, and the modelling and scenario 10 framework used in this assessment are presented. 11 12 Framing and Context of the WGI Report 13 14 The WGI contribution to the IPCC Sixth Assessment Report (AR6) assesses new scientific evidence 15 relevant for a world whose climate system is rapidly changing, overwhelmingly due to human 16 influence. The five IPCC assessment cycles since 1990 have comprehensively and consistently laid out the 17 rapidly accumulating evidence of a changing climate system, with the Fourth Assessment Report (AR4, 18 2007) being the first to conclude that warming of the climate system is unequivocal. Sustained changes have 19 been documented in all major elements of the climate system, including the atmosphere, land, cryosphere, 20 biosphere and ocean. Multiple lines of evidence indicate the unprecedented nature of recent large-scale 21 climatic changes in context of all human history, and that they represent a millennial-scale commitment for 22 the slow-responding elements of the climate system, resulting in continued worldwide loss of ice, increase in 23 ocean heat content, sea level rise and deep ocean acidification. {1.2.1, 1.3, Box 1.2, Appendix 1.A} 24 25 Since the IPCC Fifth Assessment Report (AR5), the international policy context of IPCC reports has 26 changed. The UN Framework Convention on Climate Change (UNFCCC, 1992) has the overarching 27 objective of preventing ‘dangerous anthropogenic interference with the climate system’. Responding to that 28 objective, the Paris Agreement (2015) established the long-term goals of ‘holding the increase in global 29 average temperature to well below 2°C above pre-industrial levels and pursuing efforts to limit the 30 temperature increase to 1.5°C above pre-industrial levels’ and of achieving ‘a balance between 31 anthropogenic emissions by sources and removals by sinks of greenhouse gases in the second half of this 32 century’. Parties to the Agreement have submitted Nationally Determined Contributions (NDCs) indicating 33 their planned mitigation and adaptation strategies. However, the NDCs submitted as of 2020 are insufficient 34 to reduce greenhouse gas emission enough to be consistent with trajectories limiting global warming to well 35 below 2°C above pre-industrial levels (high confidence). {1.1, 1.2} 36 37 This report provides information of potential relevance to the 2023 global stocktake. The 5-yearly 38 stocktakes called for in the Paris Agreement will evaluate alignment among the Agreement’s long-term 39 goals, its means of implementation and support, and evolving global efforts in climate change mitigation 40 (efforts to limit climate change) and adaptation (efforts to adapt to changes that cannot be avoided). In this 41 context, WGI assesses, among other topics, remaining cumulative carbon emission budgets for a range of 42 global warming levels, effects of long-lived and short-lived climate forcers, projected changes in sea level 43 and extreme events, and attribution to anthropogenic climate change. {Cross-Chapter Box 1.1} 44 45 Understanding of the fundamental features of the climate system is robust and well established. 46 Scientists in the 19th-century identified the major natural factors influencing the climate system. They also 47 hypothesized the potential for anthropogenic climate change due to carbon dioxide (CO2) emitted by fossil 48 fuel combustion. The principal natural drivers of climate change, including changes in incoming solar 49 radiation, volcanic activity, orbital cycles, and changes in global biogeochemical cycles, have been studied 50 systematically since the early 20th century. Other major anthropogenic drivers, such as atmospheric aerosols 51 (fine solid particles or liquid droplets), land-use change and non-CO2 greenhouse gases, were identified by 52 the 1970s. Since systematic scientific assessments began in the 1970s, the influence of human activity on the 53 warming of the climate system has evolved from theory to established fact. Past projections of global surface 54 temperature and the pattern of warming are broadly consistent with subsequent observations (limited 55 evidence, high agreement), especially when accounting for the difference in radiative forcing scenarios used Do Not Cite, Quote or Distribute 1-5 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 for making projections and the radiative forcings that actually occurred. {1.3.1 - 1.3.6} 2 3 Global surface temperatures increased by about 0.1°C (likely range –0.1°C to +0.3°C, medium 4 confidence) between the period around 1750 and the 1850–1900 period, with anthropogenic factors 5 responsible for a warming of 0.0°C–0.2°C (likely range, medium confidence). This assessed change in 6 temperature before 1850–1900 is not included in the AR6 assessment of global warming to date, to ensure 7 consistency with previous IPCC assessment reports, and because of the lower confidence in the estimate. 8 There was likely a net anthropogenic forcing of 0.0–0.3 Wm-2 in 1850–1900 relative to 1750 (medium 9 confidence), with radiative forcing from increases in atmospheric greenhouse gas concentrations being 10 partially offset by anthropogenic aerosol emissions and land-use change. Net radiative forcing from solar and 11 volcanic activity is estimated to be smaller than ±0.1 Wm-2 for the same period. {Cross Chapter Box 1.2, 12 1.4.1, Cross Chapter Box 2.3} 13 14 Natural climate variability can temporarily obscure or intensify anthropogenic climate change on 15 decadal time scales, especially in regions with large internal interannual-to-decadal variability. At the 16 current level of global warming, an observed signal of temperature change relative to the 1850–1900 17 baseline has emerged above the levels of background variability over virtually all land regions (high 18 confidence). Both the rate of long-term change and the amplitude of interannual (year-to-year) variability 19 differ from global to regional to local scales, between regions and across climate variables, thus influencing 20 when changes become apparent. Tropical regions have experienced less warming than most others, but also 21 exhibit smaller interannual variations in temperature. Accordingly, the signal of change is more apparent in 22 tropical regions than in regions with greater warming but larger interannual variations (high confidence). 23 {1.4.2, FAQ1.2} 24 25 The AR6 has adopted a unified framework of climate risk, supported by an increased focus in WGI on 26 low-likelihood, high-impact events. Systematic risk framing is intended to aid the formulation of effective 27 responses to the challenges posed by current and future climatic changes and to better inform risk assessment 28 and decision-making. AR6 also makes use of the ‘storylines’ approach, which contributes to building a 29 robust and comprehensive picture of climate information, allows a more flexible consideration and 30 communication of risk, and can explicitly address low-likelihood, high-impact events. {1.1.2, 1.4.4, Cross- 31 Chapter Box 1.3} 32 33 The construction of climate change information and communication of scientific understanding are 34 influenced by the values of the producers, the users and their broader audiences. Scientific knowledge 35 interacts with pre-existing conceptions of weather and climate, including values and beliefs stemming from 36 ethnic or national identity, traditions, religion or lived relationships to land and sea (high confidence). 37 Science has values of its own, including objectivity, openness and evidence-based thinking. Social values 38 may guide certain choices made during the construction, assessment and communication of information 39 (high confidence). {1.2.3, Box 1.1} 40 41 Data, Tools and Methods Used across the WGI Report 42 43 Capabilities for observing the physical climate system have continued to improve and expand overall, 44 but some reductions in observational capacity are also evident (high confidence). Improvements are 45 particularly evident in ocean observing networks and remote-sensing systems, and in paleoclimate 46 reconstructions from proxy archives. However, some climate-relevant observations have been interrupted by 47 the discontinuation of surface stations and radiosonde launches, and delays in the digitisation of records. 48 Further reductions are expected to result from the COVID-19 pandemic. In addition, paleoclimate archives 49 such as mid-latitude and tropical glaciers as well as modern natural archives used for calibration (e.g., corals 50 and trees) are rapidly disappearing owing to a host of pressures, including increasing temperatures (high 51 confidence). {1.5.1} 52 53 Reanalyses have improved since AR5 and are increasingly used as a line of evidence in assessments of 54 the state and evolution of the climate system (high confidence). Reanalyses, where atmosphere or ocean 55 forecast models are constrained by historical observational data to create a climate record of the past, provide Do Not Cite, Quote or Distribute 1-6 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 consistency across multiple physical quantities and information about variables and locations that are not 2 directly observed. Since AR5, new reanalyses have been developed with various combinations of increased 3 resolution, extended records, more consistent data assimilation, estimation of uncertainty arising from the 4 range of initial conditions, and an improved representation of the ocean. While noting their remaining 5 limitations, the WGI report uses the most recent generation of reanalysis products alongside more standard 6 observation-based datasets. {1.5.2, Annex 1} 7 8 Since AR5, new techniques have provided greater confidence in attributing changes in climate 9 extremes to climate change. Attribution is the process of evaluating the relative contributions of multiple 10 causal factors to an observed change or event. This includes the attribution of the causal factors of changes in 11 physical or biogeochemical weather or climate variables (e.g., temperature or atmospheric CO2) as done in 12 WGI, or of the impacts of these changes on natural and human systems (e.g., infrastructure damage or 13 agricultural productivity), as done in WGII. Attributed causes include human activities (such as emissions of 14 greenhouse gases and aerosols, or land-use change), and changes in other aspects of the climate, or natural or 15 human systems. {Cross-WG Box 1.1} 16 17 The latest generation of complex climate models has an improved representation of physical processes, 18 and a wider range of Earth system models now represent biogeochemical cycles. Since the AR5, 19 higher-resolution models that better capture smaller-scale processes and extreme events have become 20 available. Key model intercomparisons supporting this assessment include the Coupled Model 21 Intercomparison Project Phase 6 (CMIP6) and the Coordinated Regional Climate Downscaling Experiment 22 (CORDEX), for global and regional models respectively. Results using CMIP Phase 5 (CMIP5) simulations 23 are also assessed. Since the AR5, large ensemble simulations, where individual models perform multiple 24 simulations with the same climate forcings, are increasingly used to inform understanding of the relative 25 roles of internal variability and forced change in the climate system, especially on regional scales. The 26 broader availability of ensemble model simulations has contributed to better estimations of uncertainty in 27 projections of future change (high confidence). A broad set of simplified climate models is assessed and used 28 as emulators to transfer climate information across research communities, such as for evaluating impacts or 29 mitigation pathways consistent with certain levels of future warming. {1.4.2, 1.5.3, 1.5.4, Cross-chapter Box 30 7.1} 31 32 Assessments of future climate change are integrated within and across the three IPCC Working 33 Groups through the use of three core components: scenarios, global warming levels, and the 34 relationship between cumulative carbon emissions and global warming. Scenarios have a long history in 35 the IPCC as a method for systematically examining possible futures. A new set of scenarios, derived from 36 the Shared Socio-economic Pathways (SSPs), is used to synthesize knowledge across the physical sciences, 37 impact, and adaptation and mitigation research. The core set of SSP scenarios used in the WGI report, SSP1- 38 1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, cover a broad range of emission pathways, including new 39 low-emissions pathways. The feasibility or likelihood of individual scenarios is not part of this assessment, 40 which focuses on the climate response to possible, prescribed emission futures. Levels of global surface 41 temperature change (global warming levels), which are closely related to a range of hazards and regional 42 climate impacts, also serve as reference points within and across IPCC Working Groups. Cumulative carbon 43 emissions, which have a nearly linear relationship to increases in global surface temperature, are also used. 44 {1.6.1-1.6.4, Cross-Chapter Box 1.5, Cross-Chapter Box 11.1} 45 46 47 Do Not Cite, Quote or Distribute 1-7 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.1 Report and chapter overview 2 3 The role of the Intergovernmental Panel on Climate Change (IPCC) is to critically assess the scientific, 4 technical, and socio-economic information relevant to understanding the physical science and impacts of 5 human-induced climate change and natural variations, including the risks, opportunities and options for 6 adaptation and mitigation. This task is performed through a comprehensive assessment of the scientific 7 literature. The robustness of IPCC assessments stems from the systematic consideration and combination of 8 multiple lines of independent evidence. In addition, IPCC reports undergo one of the most comprehensive, 9 open, and transparent review and revision processes ever employed for science assessments. 10 11 Starting with the First Assessment Report (FAR; IPCC, 1990) the IPCC assessments have been structured 12 into three working groups. Working Group I (WGI) assesses the physical science basis of climate change, 13 Working Group II (WGII) assesses associated impacts, vulnerability and adaptation options, and Working 14 Group III (WGIII) assesses mitigation response options. Each report builds on the earlier comprehensive 15 assessments by incorporating new research and updating previous findings. The volume of knowledge 16 assessed and the cross-linkages between the three working groups have substantially increased over time. 17 18 As part of its sixth assessment cycle, from 2015 to 2022, the IPCC is producing three Working Group 19 Reports, three targeted Special Reports, a Refinement to the 2006 IPCC Guidelines for National Greenhouse 20 Gas Inventories, and a Synthesis Report. The AR6 Special Reports covered the topics of ‘Global Warming of 21 1.5°C’ (SR1.5; IPCC, 2018), ‘Climate Change and Land’ (SRCCL; IPCC, 2019a) and ‘The Ocean and 22 Cryosphere in a Changing Climate’ (SROCC; IPCC, 2019b). The SR1.5 and SRCCL are the first IPCC 23 reports jointly produced by all three Working Groups. This evolution towards a more integrated assessment 24 reflects a broader understanding of the interconnectedness of the multiple dimensions of climate change. 25 26 27 1.1.1 The AR6 WGI Report 28 29 The Sixth Assessment Report (AR6) of the IPCC marks more than 30 years of global collaboration to 30 describe and understand, through expert assessments, one of the defining challenges of the 21st century: 31 human-induced climate change. Since the inception of the IPCC in 1988, our understanding of the physical 32 science basis of climate change has advanced markedly. The amount and quality of instrumental 33 observations and information from paleoclimate archives have substantially increased. Understanding of 34 individual physical, chemical and biological processes has improved. Climate model capabilities have been 35 enhanced, through the more realistic treatment of interactions among the components of the climate system, 36 and improved representation of the physical processes, in line with the increased computational capacities of 37 the world's supercomputers. 38 39 This report assesses both observed changes, and the components of these changes that are attributable to 40 anthropogenic influence (or human-induced), distinguishing between anthropogenic and naturally forced 41 changes (see Section 1.2.1.1, Section 1.4.1, Cross Working Group Box: Attribution, and Chapter 3). The 42 core assessment conclusions from previous IPCC reports are confirmed or strengthened in this report, 43 indicating the robustness of our understanding of the primary causes and consequences of anthropogenic 44 climate change. 45 46 The WGI contribution to AR6 is focused on physical and biogeochemical climate science information, with 47 particular emphasis on regional climate changes. These are relevant for mitigation, adaptation and risk 48 assessment in the context of complex and evolving policy settings, including the Paris Agreement, the 49 Global Stocktake, the Sendai Framework and the Sustainable Development Goals (SDGs) Framework. 50 51 The core of this report consists of twelve chapters plus the Atlas (Figure 1.1), which can together be grouped 52 into three categories (excluding this framing chapter): 53 54 Large-Scale Information (Chapters 2, 3 and 4). These chapters assess climate information from global to 55 continental or ocean-basin scales. Chapter 2 presents an assessment of the changing state of the climate Do Not Cite, Quote or Distribute 1-8 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 system, including the atmosphere, biosphere, ocean and cryosphere. Chapter 3 continues with an assessment 2 of the human influence on this changing climate, covering the attribution of observed changes, and 3 introducing the fitness-for-purpose approach for the evaluation of climate models used to conduct the 4 attribution studies. Finally, Chapter 4 assesses climate change projections, from the near to the long term, 5 including climate change beyond 2100, as well as the potential for abrupt and ‘low-likelihood, high-impact’ 6 changes. 7 8 Process Understanding (Chapters 5, 6, 7, 8 and 9). These five chapters provide end-to-end assessments of 9 fundamental Earth system processes and components: the carbon budget and biogeochemical cycles (Chapter 10 5), short-lived climate forcers and their links to air quality (Chapter 6), the Earth’s energy budget and climate 11 sensitivity (Chapter 7), the water cycle (Chapter 8), and the ocean, cryosphere and sea-level changes 12 (Chapter 9). All these chapters provide assessments of observed changes, including relevant paleoclimatic 13 information and understanding of processes and mechanisms as well as projections and model evaluation. 14 15 Regional Information (Chapters 10, 11, 12 and Atlas). New knowledge on climate change at regional 16 scales is reflected in this report with four chapters covering regional information. Chapter 10 provides a 17 framework for assessment of regional climate information, including methods, physical processes, an 18 assessment of observed changes at regional scales, and the performance of regional models. Chapter 11 19 addresses extreme weather and climate events, including temperature, precipitation, flooding, droughts and 20 compound events. Chapter 12 provides a comprehensive, region-specific assessment of changing climatic 21 conditions that may be hazardous or favourable (hence influencing climate risk) for various sectors to be 22 assessed in WGII. Lastly, the Atlas assesses and synthesizes regional climate information from the whole 23 report, focussing on the assessments of mean changes in different regions and on model assessments for the 24 regions. It also introduces the online Interactive Atlas, a novel compendium of global and regional climate 25 change observations and projections. It includes a visualization tool combining various warming levels and 26 scenarios on multiple scales of space and time. 27 28 Embedded in the chapters are Cross-Chapter Boxes that highlight cross-cutting issues. Each chapter also 29 includes an Executive Summary (ES), and several Frequently Asked Questions (FAQs). To enhance 30 traceability and reproducibility of report figures and tables, detailed information on the input data used to 31 create them, as well as links to archived code, are provided in the Input Data Tables in chapter 32 Supplementary Material. Additional metadata on the model input datasets is provided via the report website. 33 34 The AR6 WGI report includes a Summary for Policy Makers (SPM) and a Technical Summary (TS). The 35 integration among the three IPCC Working Groups is strengthened by the implementation of the Cross- 36 Working-Group Glossary. 37 38 39 [START FIGURE 1.1 HERE] 40 41 Figure 1.1: The structure of the AR6 WGI Report. Shown are the three pillars of the AR6 WGI, its relation to the 42 WGII and WGIII contributions, and the cross-working-group AR6 Synthesis Report (SYR). 43 44 [END FIGURE 1.1 HERE] 45 46 47 1.1.2 Rationale for the new AR6 WGI structure and its relation to the previous AR5 WGI Report 48 49 The AR6 WGI report, as a result of its scoping process, is structured around topics such as large-scale 50 information, process understanding and regional information (Figure 1.1). This represents a rearrangement 51 relative to the structure of the WGI contribution to the IPCC Fifth Assessment Report (AR5; IPCC, 2013a), 52 as summarized in Figure 1.2. The AR6 approach aims at a greater visibility of key knowledge developments 53 potentially relevant for policymakers, including climate change mitigation, regional adaptation planning 54 based on a risk management framework, and the Global Stocktake. 55 Do Not Cite, Quote or Distribute 1-9 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 [START FIGURE 1.2 HERE] 3 4 Figure 1.2: Main relations between AR5 WGI and AR6 WGI chapters. The left column shows the AR5 WGI 5 chapter categories. The central column lists the AR5 WGI chapters, with the colour code indicating their 6 relation to the AR6 WGI structure shown in Figure 1.1: Large-Scale Information (red), Process 7 Understanding (gold), Regional Information (light blue), and Whole-Report Information (dark blue). AR5 8 WGI chapters depicted in white have their topics distributed over multiple AR6 WGI chapters and 9 categories. The right column explains where to find related information in the AR6 WGI report. 10 11 [END FIGURE 1.2 HERE] 12 13 14 Two key subjects presented separately in AR5, paleoclimate and model evaluation, are now distributed 15 among multiple AR6 WGI chapters. Various other cross-cutting themes are also distributed throughout this 16 report. A summary of these themes and their integration across chapters is described in Table 1.1. 17 18 19 [START TABLE 1.1 HERE] 20 21 Table 1.1: Cross-cutting themes in AR6 WGI, and the main chapters that deal with them. Bold numbers in the table 22 indicate the chapters that have extensive coverage. 23 Thematic focus Main chapters; additional chapters Aerosols 2, 6, 7, 8, 9, 10, 11; 3, 4, Atlas Atmospheric circulation 3, 4, 8; 2, 5, 10, 11 Biosphere 2, 3, 5, 11, Cross-Chapter Box 5.1; 1, 4, 6, 8 Carbon dioxide removal (CDR) 4, 5; 8 Cities and urban aspects 10, 11, 12; 2, 8, 9, Atlas Climate services 12, Atlas, Cross-Chapter Box 12.2; 1, 10 Climatic impact-drivers 12, Annex VI; 1, 9, 10, 11, Atlas CO2 concentration levels 1, 2, 5, Cross-Chapter Box 1.1; 12, Atlas Coronavirus pandemic (COVID-19) Cross-Chapter Box 6.1; 1 Cryosphere 2, 3, 9; 1, 4, 8, 12, Atlas Deep uncertainty 9; 4, 7, 8, Cross-Chapter Box 11.2, Cross-Chapter Box 12.1 Do Not Cite, Quote or Distribute 1-10 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI Detection and attribution 3, 10, 11, Cross-Working Group Box: Attribution; 5, 6, 8, 9, 12, Atlas Emergence 1, 10, 12; 8, 11 Extremes and abrupt change 11, 12; 1, 5, 7, 8, 9, 10, Atlas, Cross-Chapter Box 12.1 Global warming hiatus Cross-Chapter Box 3.1; 10, 11 Land use 5; 2, 7, 8, 10, 11 Limits of habitability 9, 12; 11 Low-likelihood, high-impact/warming 1, 4, 11; 7, 8, 9, 10, Cross-Chapter Box 1.1, Cross-Chapter Box 1.3, Cross-Chapter Box 4 Model evaluation 1, 3, 9, 10, 11, Atlas; 5, 6, 8 Modes of variability 1, 2, 3, 4, 8, 9, Annex IV; 7, 10, 11, 12, Atlas Monsoons 8; 3, 4, 9, 10, 11, 12, Atlas Natural variability 1, 2, 3, 4, 9, 11; 5, 8, 10 Ocean 3, 5, 9; 1, 2, 4, 7, 12, Atlas Paleoclimate 1, 2; 3, 5, 7, 8, 9, Atlas, Box 11.3 Polar regions 9, 12, Atlas; 2, 3, 7, 8 Radiative Forcing 7; 1, 2, 6, 11 Regional case studies 10, 11, Atlas; 12, Box 8.1, Box 11.4, Cross-Chapter Box 12.2 Risk 1, 11, 12, Cross-Chapter Box 1.3; 4, 5, 9, Cross-Chapter Box 12.1 Sea level 9, 12; 1, 2, 3, 4, 7, 8, 10, 11, Atlas Short-lived climate forcers (SLCF) 6, 7; 1, 2, 4, Atlas Do Not Cite, Quote or Distribute 1-11 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI Solar radiation modification (SRM) 4, 5; 6, 8 Tipping points 5, 8, 9; 4, 11, 12, Cross-Chapter Box 12.1 Values and beliefs 1, 10; 12 Volcanic forcing 2, 4, 7, 8; 1, 3, 5, 9, 10, Annex III Water cycle 8, 11; 2, 3, 10, Box 11.1 1 2 [END TABLE 1.1 HERE] 3 4 5 1.1.3 Integration of AR6 WGI assessments with other Working Groups 6 7 Integration of assessments across the chapters of the WGI Report, and with WGII and WGIII, occurs in a 8 number of ways, including work on a common Glossary, risk framework (see Cross-Chapter Box 1.3), 9 scenarios and projections of future large-scale changes, and the presentation of results at various global 10 warming levels (see Section 1.6). 11 12 Chapters 8 through 12, and the Atlas, cover topics also assessed by WGII in several areas, including regional 13 climate information and climate-related risks. This approach produces a more integrated assessment of 14 impacts of climate change across Working Groups. In particular, Chapter 10 discusses the generation of 15 regional climate information for users, the co-design of research with users, and the translation of 16 information into the user context (in particular directed towards WGII). Chapter 12 provides a direct bridge 17 between physical climate information (climatic impact-drivers) and sectoral impacts and risk, following the 18 chapter organization of the WGII assessment. Notably, Cross-Chapter Box 12.1 draws a connection to 19 representative key risks and Reasons for Concern (RFC). 20 21 The science assessed in Chapters 2 to 7, such as the carbon budget, short-lived climate forcers and emission 22 metrics, are topics in common with WGIII, and relevant for the mitigation of climate change. This includes a 23 consistent presentation of the concepts of carbon budget and net zero emission targets within chapters, in 24 order to support integration in the Synthesis Report. Emission-driven emulators (simple climate models), 25 summarised in Cross-Chapter Box 7.1 in Chapter 7 are used to approximate large-scale climate responses of 26 complex Earth System Models (ESMs) and have been used as tools to explore the expected GSAT response 27 to multiple scenarios consistent with those assessed in WGI for the classification of scenarios in WGIII. 28 Chapter 6 provides information about the impact of climate change on global air pollution, relevant for 29 WGII, including Cross-Chapter Box 6.1 on the implications of the recent coronavirus pandemic (COVID-19) 30 for climate and air quality. Cross-Chapter Box 2.3 in Chapter 2 presents an integrated cross-WG discussion 31 of global temperature definitions, with implications for many aspects of climate change science. 32 33 In addition, Chapter 1 sets out a shared terminology on cross-cutting topics, including climate risk, 34 attribution and storylines, as well as an introduction to emission scenarios, global warming levels and 35 cumulative carbon emissions as an overarching topic for integration across all three Working Groups. 36 37 All these integration efforts are aimed at enhancing the bridges and ‘handshakes’ among Working Groups, 38 enabling the final cross-working group exercise of producing the integrated Synthesis Report. 39 40 Do Not Cite, Quote or Distribute 1-12 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.1.4 Chapter preview 2 3 The main purposes of this chapter are: (1) to set the scene for the WGI assessment and to place it in the 4 context of ongoing global changes, international policy processes, the history of climate science and the 5 evolution from previous IPCC assessments, including the Special Reports prepared as part of the sixth 6 assessment cycle; (2) to describe key concepts and methods, relevant developments since AR5, and the 7 modelling framework used in this assessment; and (3) together with the other chapters of this report, to 8 provide context and support for the WGII and WGIII contributions to AR6, particularly on climate 9 information to support mitigation, adaptation and risk management. 10 11 The chapter comprises seven sections (Figure 1.3). Section 1.2 describes the present state of Earth’s climate, 12 in the context of reconstructed and observed long-term changes and variations caused by natural and 13 anthropogenic factors. It also provides context for the present assessment by describing recent changes in 14 international climate change governance and fundamental scientific values. The evolution of knowledge 15 about climate change and the development of earlier IPCC assessments are presented in Section 1.3. 16 Approaches, methods, and key concepts of this assessment are introduced in Section 1.4. New developments 17 in observing networks, reanalyses, modelling capabilities and techniques since the AR5 are discussed in 18 Section 1.5. The three main ‘dimensions of integration’ across Working Groups in the AR6, i.e. emission 19 scenarios, global warming levels and cumulative carbon emissions, are described in Section 1.6. The Chapter 20 closes with a discussion of opportunities and gaps in knowledge integration in Section 1.7. 21 22 23 [START FIGURE 1.3 HERE] 24 25 Figure 1.3: A roadmap to the contents of Chapter 1. 26 27 [END FIGURE 1.3 HERE] 28 29 30 1.2 Where we are now 31 32 The IPCC sixth assessment cycle occurs in the context of increasingly apparent climatic changes observed 33 across the physical climate system. Many of these changes can be attributed to anthropogenic influences, 34 with impacts on natural and human systems. AR6 also occurs in the context of efforts in international climate 35 governance such as the Paris Agreement, which sets a long-term goal to hold the increase in global average 36 temperature to ‘well below 2°C above pre-industrial levels, and to pursue efforts to limit the temperature 37 increase to 1.5°C above pre-industrial levels, recognizing that this would significantly reduce the risks and 38 impacts of climate change’. This section summarises key elements of the broader context surrounding the 39 assessments made in the present report. 40 41 42 1.2.1 The changing state of the physical climate system 43 44 The WGI contribution to the AR5 (AR5 WGI; IPCC, 2013a) assessed that ‘warming of the climate system is 45 unequivocal’, and that since the 1950s, many of the observed changes are unprecedented over decades to 46 millennia. Changes are evident in all components of the climate system: the atmosphere and the ocean have 47 warmed, amounts of snow and ice have diminished, sea level has risen, the ocean has acidified and its 48 oxygen content has declined, and atmospheric concentrations of greenhouse gases have increased (IPCC, 49 2013b). This Report documents that, since the AR5, changes to the state of the physical and biogeochemical 50 climate system have continued, and these are assessed in full in later chapters. Here, we summarize changes 51 to a set of key large-scale climate indicators over the modern era (1850 to present). We also discuss the 52 changes in relation to the longer-term evolution of the climate. These ongoing changes throughout the 53 climate system form a key part of the context of the present report. 54 55 Do Not Cite, Quote or Distribute 1-13 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.2.1.1 Recent changes in multiple climate indicators 2 3 The physical climate system comprises all processes that combine to form weather and climate. The early 4 chapters of this report broadly organize their assessments according to overarching realms: the atmosphere, 5 the biosphere, the cryosphere (surface areas covered by frozen water, such as glaciers and ice sheets), and the 6 ocean. Elsewhere in the report, and in previous IPCC assessments, the land is also used as an integrating 7 realm that includes parts of the biosphere and the cryosphere. These overarching realms have been studied 8 and measured in increasing detail by scientists, institutions, and the general public since the 18th century, 9 over the era of instrumental observation (see Section 1.3). Today, observations include those taken by 10 numerous land surface stations, ocean surface measurements from ships and buoys, underwater 11 instrumentation, satellite and surface-based remote sensing, and in situ atmospheric measurements from 12 airplanes and balloons. These instrumental observations are combined with paleoclimate reconstructions and 13 historical documentations to produce a highly detailed picture of the past and present state of the whole 14 climate system, and to allow assessments about rates of change across the different realms (see Chapter 2 15 and Section 1.5). 16 17 Figure 1.4 documents that the climate system is undergoing a comprehensive set of changes. It shows a 18 selection of key indicators of change through the instrumental era that are assessed and presented in the 19 subsequent chapters of this report. Annual mean values are shown as stripes, with colours indicating their 20 value. The transitions from one colour to another over time illustrate how conditions are shifting in all 21 components of the climate system. For these particular indicators, the observed changes go beyond the 22 yearly and decadal variability of the climate system. In this Report, this is termed an ‘emergence’ of the 23 climate signal (see Section 1.4.2 and FAQ 1.2). 24 25 Warming of the climate system is most commonly presented through the observed increase in global mean 26 surface temperature (GMST). Taking a baseline of 1850–1900, GMST change until present (2011–2020) is 27 1.09 °C (0.95–1.20 °C) (see Chapter 2, Section 2.3, Cross-Chapter Box 2.3). This evolving change has been 28 documented in previous Assessment Reports, with each reporting a higher total global temperature change 29 (see Section 1.3, Cross-Chapter Box 1.2). The total change in Global Surface Air Temperature (GSAT; see 30 Section 1.4.1 and Cross-Chapter Box 2.3 in Chapter 2) attributable to anthropogenic activities is assessed to 31 be consistent with the observed change in GSAT (see Chapter 3, Section 3.3) 1. 32 33 Similarly, atmospheric concentrations of a range of greenhouse gases are increasing. Carbon dioxide (CO2, 34 shown in Figure 1.4 and Figure 1.5a), found in AR5 and earlier reports to be the current strongest driver of 35 anthropogenic climate change, has increased from 285.5 ± 2.1 ppm in 1850 to 409.9 ± 0.4 ppm in 2019; 36 concentrations of methane (CH4), and nitrous oxide (N2O) have increased as well (see Chapter 2, Sections 37 2.2, Chapter 5, section 5.2, and Annex V). These observed changes are assessed to be in line with known 38 anthropogenic and natural emissions, when accounting for observed and inferred uptake by land, ocean, and 39 biosphere respectively (see Chapter 5, Section 5.2), and are a key source of anthropogenic changes to the 40 global energy balance (or radiative forcing; see Chapter 2, Section 2.2 and Chapter 7, Section 7.3). 41 42 The hydrological (or water) cycle is also changing and is assessed to be intensifying, through a higher 43 exchange of water between the surface and the atmosphere (see Chapter 3, Section 2.3 and Chapter 8, 44 Section 8.3). The resulting regional patterns of changes to precipitation are, however, different from surface 45 temperature change, and interannual variability is larger, as illustrated in Figure 1.4. Annual land area mean 46 precipitation in the Northern Hemisphere temperate regions has increased, while the sub-tropical dry regions 47 have experienced a decrease in precipitation in recent decades (see Chapter 2, Section 2.3). 48 49 The cryosphere is undergoing rapid changes, with increased melting and loss of frozen water mass in most 1 Note that GMST and GSAT are physically distinct but closely related quantities, see Section 1.4.1 and Cross-Chapter Box 2.3 in Chapter 2. Do Not Cite, Quote or Distribute 1-14 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 regions. This includes all frozen parts of the globe, such as terrestrial snow, permafrost, sea ice, glaciers, 2 freshwater ice, solid precipitation, and the ice sheets covering Greenland and Antarctica (see Chapter 9; 3 SROCC, IPCC, 2019b). Figure 1.4 illustrates how, globally, glaciers have been increasingly losing mass for 4 the last fifty years. The total glacier mass in the most recent decade (2010-2019) was the lowest since the 5 beginning of the 20th century. (See Chapter 2, Section 2.3 and Chapter 9, Section 9.5). 6 7 The global ocean has warmed unabatedly since at least 1970 (Sections 1.3, 2.3, 9.2; SROCC, IPCC, 2019b) . 8 Figure 1.4 shows how the averaged ocean heat content is steadily increasing, with a total increase of [0.28– 9 0.55] yottajoule (1024 joule) between 1971 and 2018. (see Chapter 9, Section 9.2). In response to this ocean 10 warming, as well as to the loss of mass from glaciers and ice sheets, the global mean sea level (GMSL) has 11 risen by 0.20 [0.15 to 0.25] metres between 1900 and 2018. GMSL rise has accelerated since the late 1960s. 12 (See Chapter 9, Section 9.6). 13 14 Overall, the changes in these selected climatic indicators have progressed beyond the range of natural year- 15 to-year variability (see Chapters 2, 3, 8, and 9, and further discussion in Sections 1.2.1.2 and 1.4.2). The 16 indicators presented in Figure 1.4 document a broad set of concurrent and emerging changes across the 17 physical climate system. All indicators shown here, along with many others, are further presented in the 18 coming chapters, together with a rigorous assessment of the supporting scientific literature. Later chapters 19 (Chapter 10, 11, 12, and the Atlas) present similar assessments at the regional level, where observed changes 20 do not always align with the global mean picture shown here. 21 22 23 [START FIGURE 1.4 HERE] 24 25 Figure 1.4: Changes are occurring throughout the climate system. Left: Main realms of the climate system: 26 atmosphere, biosphere, cryosphere, and ocean. Right: Six key indicators of ongoing changes since 1850, 27 or the start of the observational or assessed record, through 2018. Each stripe indicates the global (except 28 for precipitation which shows two latitude band means), annual mean anomaly for a single year, relative 29 to a multi-year baseline (except for CO2 concentration and glacier mass loss, which are absolute values). 30 Grey indicates that data are not available. Datasets and baselines used are: (1) CO2: Antarctic ice cores 31 (Lüthi et al., 2008; Bereiter et al., 2015) and direct air measurements (Tans and Keeling, 2020) (see 32 Figure 1.5 for details); (2) precipitation: Global Precipitation Climatology Centre (GPCC) V8 (updated 33 from Becker et al. 2013), baseline 1961-1990 using land areas only with latitude bands 33°N–66°N and 34 15°S–30°S; (3) glacier mass loss: Zemp et al., 2019; (4) global surface air temperature (GMST): 35 HadCRUT5 (Morice et al., 2021), baseline 1961–1990; (5) sea level change: (Dangendorf et al., 2019), 36 baseline 1900–1929; (6) ocean heat content (model-observation hybrid): Zanna et al., (2019), baseline 37 1961–1990. Further details on data sources and processing are available in the chapter data table (Table 38 1.SM.1). 39 40 [END FIGURE 1.4 HERE] 41 42 43 1.2.1.2 Long-term perspectives on anthropogenic climate change 44 45 Paleoclimate archives (e.g, ice cores, corals, marine and lake sediments, speleothems, tree rings, borehole 46 temperatures, soils) permit the reconstruction of climatic conditions before the instrumental era. This 47 establishes an essential long-term context for the climate change of the past 150 years and the projected 48 changes in the 21st century and beyond (IPCC, 2013a; Masson-Delmotte et al., 2013) Chapter 3). Figure 1.5 49 shows reconstructions of three key indicators of climate change over the past 800,000 years 2 – atmospheric 50 CO2 concentrations, global mean surface temperature (GMST) and global mean sea level (GMSL) – 2 as old as the longest continuous climate records based on the ice core from EPICA Dome Concordia (Antarctica). Polar ice cores are the only paleoclimatic archive providing direct information on past greenhouse gas concentrations. Do Not Cite, Quote or Distribute 1-15 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 comprising at least eight complete glacial-interglacial cycles (EPICA Community Members, 2004; Jouzel et 2 al., 2007) that are largely driven by oscillations in the Earth’s orbit and consequent feedbacks on multi- 3 millennial time scales (Berger, 1978; Laskar et al., 1993). The dominant cycles – recurring approximately 4 every 100,000 years – can be found imprinted in the natural variations of these three key indicators. Before 5 industrialisation, atmospheric CO2 concentrations varied between 174 ppm and 300 ppm, as measured 6 directly in air trapped in ice at Dome Concordia, Antarctica (Bereiter et al., 2015; Nehrbass-Ahles et al., 7 2020). Relative to 1850–1900, the reconstructed GMST changed in the range of -6 to +1°C across these 8 glacial-interglacial cycles (see Chapter 2, Section 2.3.1 for an assessment of different paleo reference 9 periods). GMSL varied between about -130 m during the coldest glacial maxima and +5 to +25 m during the 10 warmest interglacial periods (Spratt and Lisiecki, 2016; Chapter 2). They represent the amplitudes of natural, 11 global-scale climate variations over the last 800,000 years prior to the influence of human activity. Further 12 climate information from a variety of paleoclimatic archives are assessed in Chapters 2, 5, 7, 9. 13 14 Paleoclimatic information also provides a long-term perspective on rates of change of these three key 15 indicators. The rate of increase in atmospheric CO2 observed over 1919-2019 CE is one order of magnitude 16 higher than the fastest CO2 fluctuations documented during the last glacial maximum and the last deglacial 17 transition in high-resolution reconstructions from polar ice cores (Marcott et al., 2014, see Chapter 2, Section 18 2.2.3.2.1). Current multi-decadal GMST exhibit a higher rate of increase than over the past two thousand 19 years (PAGES 2k Consortium, 2019; Chapter 2, Section 2.3.1.1.2), and in the 20th century GMSL rise was 20 faster than during any other century over the past three thousand years (Chapter 2, Section 2.3.3.3). 21 22 23 [START FIGURE 1.5 HERE] 24 25 Figure 1.5: Long-term context of anthropogenic climate change based on selected paleoclimatic reconstructions 26 over the past 800,000 years for three key indicators: atmospheric CO2 concentrations, Global Mean 27 Surface Temperature (GMST), and Global Mean Sea Level (GMSL). a) Measurements of CO2 in air 28 enclosed in Antarctic ice cores (Lüthi et al., 2008; Bereiter et al., 2015 [a compilation]; uncertainty 29 ±1.3ppm; see Chapter 2, Section 2.2.3 and Chapter 5, Section 5.1.2 for an assessment) and direct air 30 measurements (Tans and Keeling, 2020; uncertainty ±0.12 ppm). Projected CO2 concentrations for five 31 Shared Socioeconomic Pathways (SSP) scenarios are indicated by dots on the right-hand side panels of 32 the figure (grey background) (Meinshausen et al., 2020; SSPs are described in Section 1.6). b) 33 Reconstruction of GMST from marine paleoclimate proxies (light grey: Snyder (2016); dark grey: 34 Hansen et al. (2013); see Chapter 2, Section 2.3.1 for an assessment). Observed and reconstructed 35 temperature changes since 1850 are the AR6 assessed mean (referenced to 1850–1900; Box TS.3; 36 2.3.1.1); dots/whiskers on the right-hand side panels of the figure (grey background) indicate the 37 projected mean and ranges of warming derived from Coupled Model Intercomparison Project Phase 6 38 (CMIP6) SSP-based (2081–2100) and Model for the Assessment of Greenhouse Gas Induced Climate 39 Change (MAGICC7) (2300) simulations (Chapter 4, Tables 4.5 and 4.9). c) Sea level changes 40 reconstructed from a stack of oxygen isotope measurements on seven ocean sediment cores (Spratt and 41 Lisiecki, 2016; see Chapter 2, Section 2.3.3.3 and Chapter 9, Section 9.6.2 for an assessment). The sea 42 level record from 1850 to 1900 is from Kopp et al. (2016), while the 20th century record is an updated 43 ensemble estimate of GMSL change (Palmer et al., 2021; see also Chapter 2, Section 2.3.3.3 and Chapter 44 9, Section 9.6.1.1). Dots/whiskers on the right-hand side panels of the figure (grey background) indicate 45 the projected median and ranges derived from SSP-based simulations (2081–2100: Chapter 9, Table 9.9; 46 2300: Chapter 9, Section 9.6.3.5). Best estimates (dots) and uncertainties (whiskers) as assessed by 47 Chapter 2 are included in the left and middle panels for each of the three indicators and selected paleo- 48 reference periods used in this report (CO2: Chapter 2, Table 2.1; GMST: Chapter 2, Section 2.3.1.1 and 49 Cross-Chapter Box 2.3, Table 1 in Chapter 2; GMSL: Chapter 2, Section 2.3.3.3 and Chapter 9, Section 50 9.6.2. See also Cross-Chapter Box 2.1 in Chapter 2). Selected paleo-reference periods: LIG – Last 51 Interglacial; LGM – Last Glacial Maximum; MH – mid-Holocene (Cross-Chapter Box 2.1, Table 1 in 52 Chapter 2). The non-labelled best estimate in panel c) corresponds to the sea level high-stand during 53 Marine Isotope Stage 11, about 410,000 years ago (see Chapter 9, Section 9.6.2). Further details on data 54 sources and processing are available in the chapter data table (Table 1.SM.1). 55 56 [END FIGURE 1.5 HERE] 57 58 Do Not Cite, Quote or Distribute 1-16 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Paleoclimate reconstructions also shed light on the causes of these variations, revealing processes that need 2 to be considered when projecting climate change. The paleorecords show that sustained changes in global 3 mean temperature of a few degrees Celsius are associated with increases in sea level of several tens of metres 4 (Figure 1.5). During two extended warm periods (interglacials) of the last 800,000 years, sea level is 5 estimated to have been at least six metres higher than today (Chapter 2; Dutton et al., 2015). During the last 6 interglacial, sustained warmer temperatures in Greenland preceded the peak of sea level rise (Figure 5.15 in 7 Masson-Delmotte et al., 2013). The paleoclimate record therefore provides substantial evidence directly 8 linking warmer GMST to substantially higher GMSL. 9 10 GMST will remain above present-day levels for many centuries even if net CO2 emissions are reduced to 11 zero, as shown in simulations with coupled climate models (Plattner et al., 2008; Section 12.5.3 in Collins et 12 al., 2013; Zickfeld et al., 2013; MacDougall et al., 2020; Chapter 4, Section 4.7.1). Such persistent warm 13 conditions in the atmosphere represent a multi-century commitment to long-term sea level rise, summer sea 14 ice reduction in the Arctic, substantial ice sheet melting, potential ice sheet collapse, and many other 15 consequences in all components of the climate system (Clark et al., 2016; Pfister and Stocker, 2016; Fischer 16 et al., 2018; see also Chapter 9, Section 9.4) (Figure 1.5). 17 18 Paleoclimate records also show centennial- to millennial-scale variations, particularly during the ice ages, 19 which indicate rapid or abrupt changes of the Atlantic Meridional Overturning Circulation (AMOC; see 20 Chapter 9, Section 9.2.3.1) and the occurrence of a ‘bipolar seesaw’ (opposite-phase surface temperature 21 changes in both hemispheres; Stocker and Johnsen, 2003; EPICA Community Members, 2006; Members 22 WAIS Divide Project et al., 2015; Lynch-Stieglitz, 2017; Pedro et al., 2018; Weijer et al., 2019, see Chapter 23 2, Section 2.3.3.4.1). This process suggests that instabilities and irreversible changes could be triggered if 24 critical thresholds are passed (Section 1.4.4.3). Several other processes involving instabilities are identified 25 in climate models (Drijfhout et al., 2015), some of which may now be close to critical thresholds (Joughin et 26 al., 2014; Section 1.4.4.3; see also Chapters 5, 8 and 9 regarding tipping points). 27 28 Based on Figure 1.5, the reconstructed, observed and projected ranges of changes in the three key indicators 29 can be compared. By the first decade of the 20th century, atmospheric CO2 concentrations had already 30 moved outside the reconstructed range of natural variation over the past 800,000 years. On the other hand, 31 global mean surface temperature and sea level were higher than today during several interglacials of that 32 period (Chapter 2, Section 2.3.1, Figure 2.34 and Section 2.3.3). Projections for the end of the 21st century, 33 however, show that GMST will have moved outside of its natural range within the next few decades, except 34 for the strong mitigation scenarios (Section 1.6). There is a risk that GMSL may potentially leave the 35 reconstructed range of natural variations over the next few millennia (Clark et al., 2016; Chapter 9, Section 36 9.6.3.5; SROCC (IPCC, 2019b). In addition, abrupt changes can not be excluded (Section 1.4.4.3). 37 38 An important time period in the assessment of anthropogenic climate change is the last 2000 years. Since 39 AR5, new global datasets have emerged, aggregating local and regional paleorecords (PAGES 2k 40 Consortium, 2013, 2017, 2019; McGregor et al., 2015; Tierney et al., 2015; Abram et al., 2016; Hakim et al., 41 2016; Steiger et al., 2018; Brönnimann et al., 2019b). Before the global warming that began around the mid- 42 19th century (Abram et al., 2016), a slow cooling in the Northern Hemisphere from roughly 1450 to 1850 is 43 consistently recorded in paleoclimate archives (PAGES 2k Consortium, 2013; McGregor et al., 2015). While 44 this cooling, primarily driven by an increased number of volcanic eruptions (PAGES 2k Consortium, 2013; 45 Owens et al., 2017; Brönnimann et al., 2019b; Chapter 3, Section 3.3.1), shows regional differences, the 46 subsequent warming over the past 150 years exhibits a global coherence that is unprecedented in the last 47 2000 years (Neukom et al., 2019). 48 49 The rate, scale, and magnitude of anthropogenic changes in the climate system since the mid-20th century 50 suggested the definition of a new geological epoch, the Anthropocene (Crutzen and Stoermer, 2000; Steffen 51 et al., 2007), referring to an era in which human activity is altering major components of the Earth system 52 and leaving measurable imprints that will remain in the permanent geological record (IPCC, 2018) (Figure 53 1.5). These alterations include not only climate change itself, but also chemical and biological changes in the 54 Earth system such as rapid ocean acidification due to uptake of anthropogenic carbon dioxide, massive 55 destruction of tropical forests, a worldwide loss of biodiversity and the sixth mass extinction of species Do Not Cite, Quote or Distribute 1-17 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 (Hoegh-Guldberg and Bruno, 2010; Ceballos et al., 2017; IPBES, 2019). According to the key messages of 2 the last global assessment of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem 3 Services (IPBES, 2019), climate change is a ‘direct driver that is increasingly exacerbating the impact of 4 other drivers on nature and human well-being’, and ‘the adverse impacts of climate change on biodiversity 5 are projected to increase with increasing warming’. 6 7 8 1.2.2 The policy and governance context 9 10 The contexts of both policymaking and societal understanding about climate change have evolved since the 11 AR5 was published (2013–2014). Increasing recognition of the urgency of the climate change threat, along 12 with still-rising emissions and unresolved issues of mitigation and adaptation, including aspects of 13 sustainable development, poverty eradication and equity, have led to new policy efforts. This section 14 summarizes these contextual developments and how they have shaped, and been used during the preparation 15 of this Report. 16 17 IPCC reports and the UN Framework Convention on Climate Change (UNFCCC). The IPCC First 18 Assessment Report (FAR, IPCC, 1990a) provided the scientific background for the establishment of the 19 United Nations Framework Convention on Climate Change (UNFCCC, 1992), which committed parties to 20 negotiate ways to ‘prevent dangerous anthropogenic interference with the climate system’ (the ultimate 21 objective of the UNFCCC). The Second Assessment Report (SAR, IPCC, 1995a) informed governments in 22 negotiating the Kyoto Protocol (1997), the first major agreement focusing on mitigation under the UNFCCC. 23 The Third Assessment report (TAR, IPCC, 2001a) highlighted the impacts of climate change and need for 24 adaptation and introduced the treatment of new topics such as policy and governance in IPCC reports. The 25 Fourth and Fifth Assessment Reports (AR4, IPCC, 2007a; AR5, IPCC, 2013a) provided the scientific 26 background for the second major agreement under the UNFCCC: the Paris Agreement (2015), which entered 27 into force in 2016. 28 29 The Paris Agreement (PA). Parties to the PA commit to the goal of limiting global average temperature 30 increase to ‘well below 2°C above pre-industrial levels, and to pursue efforts to limit the temperature 31 increase to 1.5°C in order to ‘significantly reduce the risks and impacts of climate change’. In AR6, as in 32 many previous IPCC reports, observations and projections of changes in global temperature are expressed 33 relative to 1850-1900 as an approximation for pre-industrial levels (Cross-Chapter Box 1.2). 34 35 The PA further addresses mitigation (Article 4) and adaptation to climate change (Article 7), as well as loss 36 and damage (Article 8), through the mechanisms of finance (Article 9), technology development and transfer 37 (Article 10), capacity-building (Article 11) and education (Article 12). To reach its long-term temperature 38 goal, the PA recommends ‘achieving a balance between anthropogenic emissions by sources and removals 39 by sinks of greenhouse gases in the second half of this century’, a state commonly described as ‘net zero’ 40 emissions (Article 4) (Section 6, Box 1.4). Each Party to the PA is required to submit a Nationally 41 Determined Contribution (NDC) and pursue, on a voluntary basis, domestic mitigation measures with the 42 aim of achieving the objectives of its NDC (Article 4). 43 44 Numerous studies of the NDCs submitted since adoption of the PA in 2015 (Fawcett et al., 2015; UNFCCC, 45 2015, 2016; Lomborg, 2016; Rogelj et al., 2016, 2017; Benveniste et al., 2018; Gütschow et al., 2018; 46 United Nations Environment Programme (UNEP), 2019) conclude that they are insufficient to meet the Paris 47 temperature goal. In the present IPCC Sixth Assessment cycle, a Special Report on Global Warming of 48 1.5°C (SR1.5, IPCC, 2018) assessed high agreement that current NDCs ‘are not in line with pathways that 49 limit warming to 1.5°C by the end of the century’. The PA includes a ratcheting mechanism designed to 50 increase the ambition of voluntary national pledges over time. Under this mechanism, NDCs will be 51 communicated or updated every five years. Each successive NDC will represent a ‘progression beyond’ the 52 ‘then current’ NDC and reflect the ‘highest possible ambition’ (Article 4). These updates will be informed by 53 a five-yearly periodic review including the ‘Structured Expert Dialogue’ (SED), as well as a ‘global 54 stocktake’, to assess collective progress toward achieving the PA long-term goals. These processes will rely 55 upon the assessments prepared during the IPCC sixth assessment cycle (e.g., Schleussner et al., 2016b; Do Not Cite, Quote or Distribute 1-18 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Cross-Chapter Box 1.1). 2 3 The Structured Expert Dialogue (SED). Since AR5, the formal dialogue between the scientific and policy 4 communities has been strengthened through a new science-policy interface, the Structured Expert Dialogue 5 (SED). The SED was established by UNFCCC to support the work of its two subsidiary bodies, the 6 Subsidiary Body for Scientific and Technological Advice (SBSTA) and the Subsidiary Body for 7 Implementation (SBI). The first SED aimed to ‘ensure the scientific integrity of the first periodic review’ of 8 the UNFCCC, the 2013-2015 review. The Mandate of the periodic review is to ‘assess the adequacy of the 9 long-term (temperature) goal in light of the ultimate objective of the convention’ and the ‘overall progress 10 made towards achieving the long-term global goal, including a consideration of the implementation of the 11 commitments under the Convention'. 12 13 The SED of the first periodic review (2013-2015) provided an important opportunity for face-to-face 14 dialogue between decision makers and experts on review themes, based on ‘the best available scientific 15 knowledge, including the assessment reports of the IPCC’. That SED was instrumental in informing the 16 long-term global goal of the PA and in providing the scientific argument of the consideration of limiting 17 warming to 1.5°C warming (Fischlin et al., 2015; Fischlin, 2017). The SED of the second periodic review, 18 initiated in the second half of 2020, focuses on, inter alia, ‘enhancing Parties’ understanding of the long-term 19 global goal and the scenarios towards achieving it in the light of the ultimate objective of the Convention’. 20 The second SED provides a formal venue for the scientific and the policy communities to discuss the 21 requirements and benchmarks to achieve the ‘long-term temperature goal’ (LTTG) of 1.5°C and well below 22 2°C global warming. The discussions also concern the associated timing of net zero emissions targets and the 23 different interpretations of the PA LTTG, including the possibility of overshooting the 1.5° C warming level 24 before returning to it by means of negative emissions (e.g., Schleussner and Fyson, 2020; Section 1.6). The 25 second periodic review is planned to continue until November 2022 and its focus includes the review of the 26 progress made since the first review, with minimising ‘possible overlaps’ and profiting from ‘synergies with 27 the Global Stocktake’. 28 29 30 [START CROSS-CHAPTER BOX 1.1 HERE] 31 32 Cross-Chapter Box 1.1: The WGI contribution to the AR6 and its potential relevance for the global 33 stocktake 34 35 Contributing Authors: Malte Meinshausen (Australia/Germany), Gian-Kasper Plattner (Switzerland), Aïda 36 Diongue-Niang (Senegal), Francisco Doblas-Reyes (Spain), David Frame (New Zealand), Nathan Gillett 37 (Canada/UK), Helene Hewitt (UK), Richard Jones (UK), Hong Liao (China), Jochem Marotzke (Germany), 38 James Renwick (New Zealand), Joeri Rogelj (Belgium), Maisa Rojas (Chile), Sonia I. Seneviratne 39 (Switzerland), Claudia Tebaldi (USA), Blair Trewin (Australia) 40 41 The global stocktake under the Paris Agreement (PA) evaluates the collective progress of countries’ 42 actions towards attaining the Agreement’s purpose and long-term goals every five years. The first 43 global stocktake is due in 2023, and then every five years thereafter, unless otherwise decided by the 44 Conference of the Parties. The purpose and long-term goals of the PA are captured in Article 2: to 45 ‘strengthen the global response to the threat of climate change, in the context of sustainable development and 46 efforts to eradicate poverty, including by’: mitigation3, specifically, ‘holding the increase in the global 47 average temperature to well below 2°C above pre-industrial levels and to pursue efforts to limit the 48 temperature increase to 1.5°C above pre-industrial levels, recognizing that this would significantly reduce 3 the labels of mitigation, adaptation and means of implementation and support are here provided for reader's guidance only, with no presumption about the actual legal content of the paragraphs and to which extent they encompass mitigation, adaptation and means of implementation in its entirety. Do Not Cite, Quote or Distribute 1-19 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 the risks and impacts of climate change’; adaptation, that is, ‘Increasing the ability to adapt to the adverse 2 impacts of climate change and foster climate resilience and low greenhouse gas emissions development, in a 3 manner that does not threaten food production’; and means of implementation and support, that is, ‘Making 4 finance flows consistent with a pathway towards low greenhouse gas emissions and climate-resilient 5 development’. 6 7 The PA further specifies that the stocktake shall be undertaken in a ‘comprehensive and facilitative manner, 8 considering mitigation, adaptation and the means of implementation and support, and in the light of equity 9 and the best available science’ (Article 14). 10 11 The sources of input envisaged for the global stocktake include the ‘latest reports of the Intergovernmental 12 Panel on Climate Change’ as a central source of information 4. The global stocktake is one of the key formal 13 avenues for scientific inputs into the UNFCCC and PA negotiation process alongside, for example, the 14 Structured Expert Dialogues under the UNFCCC5 (Section 1.2.2). 15 16 The WGI assessment provides a wide range of information potentially relevant for the global 17 stocktake, complementing the IPCC AR6 Special Reports, the contributions from WGII and WGIII 18 and the Synthesis Report. This includes the state of greenhouse gas emissions and concentrations, the 19 current state of the climate, projected long-term warming levels under different scenarios, near-term 20 projections, the attribution of extreme events, and remaining carbon budgets. Cross-Chapter Box 1.1, Table 1 21 provide pointers to the in-depth material that WGI has assessed and may be relevant for the global stocktake. 22 23 The following tabularised overview of potentially relevant information from the WGI contribution for 24 the global stocktake is structured into three sections: the current state of the climate, the long-term 25 future, and the near-term. These sections and their order align with the three questions of the Talanoa 26 dialogue, launched during COP23 based on the Pacific concept of talanoa 6: ‘Where are we’, ‘Where do we 27 want to go’ and ‘How do we get there?’. 28 29 30 [START CROSS-CHAPTER BOX 1.1, TABLE 1 HERE] 31 32 Cross-Chapter Box 1.1, Table 1: WGI assessment findings and their potential relevance for the global stocktake. The 33 table combines information assessed in this report that could potentially be relevant 34 for the global stocktake process. Section 1 focuses on the current state of the climate 35 and its recent past. Section 2 focuses on long-term projections in the context of the 36 PA’s 1.5°C and 2.0°C goals and on progress towards net zero greenhouse gas 37 emissions. Section 3 considers challenges and key insights for mitigation and 38 adaptation in the near term from a WGI perspective. Further Information on 39 potential relevance of the aspects listed here in terms of, for example, impacts and 40 socio-economic aspects can be found in the WGII and WGIII reports 41 Section 1: State of the Climate – ‘Where are we?’ WGI assessment to inform about past changes in the climate system, current climate and committed changes. Question Chapter Potential Relevance and Explanatory Remarks 4 paragraph 37b in 19/CMA.1 in FCCC/PA/CMA/2018/3/Add.2, pursuant decision 1/CP.21, paragraph 99 of the adoption of the PA in FCCC/CP/2015/10/Add.1, available at: https://unfccc.int/documents/193408 5 Decision 5/CP.25, available at: https://unfccc.int/sites/default/files/resource/cp2019_13a01E.pdf 6 Decision 1/CP.23, in FCCC/CP/2017/L.13, available at https://unfccc.int/resource/docs/2017/cop23/eng/l13.pdf Do Not Cite, Quote or Distribute 1-20 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI How much warming have Cross- Knowledge about the current warming relative to pre- we observed in global Chapter industrial levels allows us to quantify the remaining mean surface air Box 1.2; distance to the PA goal of keeping global mean temperatures? Cross- temperatures well below 2°C above pre-industrial level or Chapter pursue best efforts to limit warming to 1.5°C above pre- Box 2.3; industrial level. Many of the report’s findings are provided 2.3.1.1, against a proxy for pre-industrial temperature levels with especially Cross-Chapter Box 1.2 examining the difference between 2.3.1.1.3 pre-industrial levels and the 1850–1900 period. How much has the ocean 2.3.3.1, A warming ocean can affect marine life (e.g., coral warmed? 9.2.1.1; Box bleaching) and also are among the main contributors to 9.1; 7.2; long-term sea level rise (thermal expansion). Marine Box 7.2 heatwaves can accentuate the impacts of ocean warming on marine ecosystems. Also, knowing the heat uptake of the ocean helps to better understand the response of the climate system and hence helps to project future warming. How much have the land 2.3.4; 5.4.3; A stronger than global-average warming over land, areas warmed and how has 5.4.8; 8.2.1; combined with changing precipitation patterns, and/or precipitation changed? 8.2.3; 8.5.1; increased aridity in some regions (like the Mediterranean) can severely affect land ecosystems and species distributions, the terrestrial carbon cycle and food production systems. Amplified warming in the Arctic can enhance permafrost thawing, which in turn can result in overall stronger anthropogenic warming (a positive feedback loop). Intensification of heavy precipitation events can cause more severe impacts related to flooding. How did the sea ice area 2.3.2.1.1; Sea ice area influences mass and energy (ice-albedo, heat change in recent decades 2.3.2.1.2; and momentum) exchange between the atmosphere and the in both the Arctic and 9.3; Cross- ocean, and its changes in turn impact polar life, adjacent Antarctic? Chapter land and ice masses and complex dynamical flows in the Box 10.1; atmosphere. The loss of a year-round sea-ice cover in the 12.4.9 Arctic can severely impact Arctic ecosystems, affect the livelihood of First Nations in the Arctic, and amplify Arctic warming with potential consequences for the warming of the surrounding permafrost regions and ice sheets. How much have 2.2.3; 2.2.4; The main human influence on the climate is via atmospheric CO2 and other 5.1.1; 5.2.2; combustion of fossil fuels and land use-change-related CO2 GHG concentrations 5.2.3; 5.2.4 emissions, the principal causes of increased CO2 increased? concentrations since the pre-industrial period. Historical observations indicate that current atmospheric concentrations are unprecedented within at least the last 800,000 years. An understanding of historical fossil fuel emissions and the carbon cycle interactions, as well as CH4 and N2O sinks and sources are crucial for better estimates of future GHG emissions compatible with the PA’s long- term goals. Do Not Cite, Quote or Distribute 1-21 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI How much did sea level 2.3.3.3; Sea level rise is a comparatively slow consequence of a rise in past centuries and 9.6.1; 9.6.2; warming world. Historical warming committed the world how large is the long-term FAQ 9.1; already to long-term sea level rise that is not reversed in commitment? Box 9.1; even the lowest emissions scenarios (such as 1.5°C), which 9.6.3; 9.6.4 come with a multimeter sea level commitment. Regional sea level change near the coastlines differs from global- mean sea level change due to vertical land movement, ice mass changes, and ocean dynamical changes. How much has the ocean 2.3.4.3; Ocean acidification is affecting marine life, especially acidified and how much 2.3.4.2; 5.3 organisms that build calciferous shells and structures (e.g., oxygen have they lost? coral reefs). Together with less oxygen in upper ocean waters and increasingly widespread oxygen minimum zones and in addition to ocean warming, this poses adaptation challenges for coastal and marine ecosystems and their services, including seafood supply. How much of the observed 3.3.1 To monitor progress toward the PA’s long-term goals it is warming was due to important to know how much of the observed warming is anthropogenic influences? due to human activities. Chapter 3 assesses human-induced warming in global mean near-surface air temperature for the decade 2010–2019, relative to 1850–1900 with associated uncertainties, based on detection and attribution studies. This estimate can be compared with observed estimates of warming for the same decade reported in Chapter 2, and is typically used to calculate carbon budgets consistent with remaining below a particular temperature threshold. How much has 3.3.2; 3.3.3; Climate change impacts are driven by changes in many anthropogenic influence 3.4; 3.5; aspects of the climate system, including changes in the changed other aspects of 3.6; 3.7; 8; water cycle, atmospheric circulation, ocean, cryosphere, the climate system? 12; 10.4 biosphere and modes of variability, and to better plan climate change adaptation it is relevant to know which observed changes have been driven by human influence. How much are 9.6.4; 11.3- Adaptation challenges are often accentuated in the face of anthropogenic emissions 11.8; 12.3; extreme events, including floods, droughts, bushfires, and contributing to changes in Cross- tropical cyclones. For agricultural management, the severity and frequency Chapter infrastructure planning, and designing for climate of extreme events? Box 3.2; resilience it is relevant to know whether extreme events 1.5; Cross- will become more frequent in the near future. In that Chapter respect it is important to understand whether observed Box 1.3 extreme events are part of a natural background variability or caused by past anthropogenic emissions. This attribution of extreme events is therefore key to understanding current events, as well as to better project the future evolution of these events, such as temperature extremes; heavy precipitation; floods; droughts; extreme storms and compound events; extreme sea level. Also, loss and damage events are often related to extreme events, which means that future disasters can be fractionally attributed to Do Not Cite, Quote or Distribute 1-22 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI past human emissions Section 2: Long-term climate futures. – ‘Where do we want to go?’ WGI assessment to inform how long-term climate change could unfold depending on chosen emission futures. Question Chapter Potential Relevance and Explanatory Remarks How are climate model Box 4.1; The scientific literature provides new insights in a projections used to 3.8.2; developing field of scientific research regarding evaluating project the range of future Cross- model performance and weighting. This can lead to more global and regional Chapter constrained projection ranges for a given scenario and some climate changes? Box 3.1; variables, which take into account the performance of 10.3; 10.4; climate models and interdependencies among them. These 12.4 techniques have a strong relevance to quantifying future uncertainties, for example regarding the likelihood of the various scenarios exceeding the PA’s long-term temperature goals of 1.5°C or 2°C. If emission scenarios are 1.2.2; 4.6, Understanding of the response to a change of anthropogenic pursued that achieve FAQ 4.2, emissions is important to estimate the scale and timing of mitigation goals by 2050, 12.4, 9, 11; mitigation compatible with the PA’s long-term goals. The what are the differences Atlas; new generation of scenarios spans the response space from in climate over the 21st Interactive very low emission scenarios (SSP1-1.9) under the century compared to Atlas; assumption of accelerated and effective climate policy emission scenarios where implementation, to very high emission scenarios in the no additional climate absence of additional climate policies (SSP3-7.0 or SSP5- policies are implemented? 8.5). It can be informative to place current NDCs and their emission mitigation pledges within this low and high-end scenario range, that is, in the context of medium-high emission scenarios (RCP4.5, RCP6.0 or SSP4-6.0). Climate response differences between those future medium or high emission scenarios and those compatible with the PA’s long-term temperature goals can help inform policymakers about the corresponding adaptation challenges. Do Not Cite, Quote or Distribute 1-23 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI What is the climatic Box 1.4; Understanding the long-term climate effect of global effect of net zero GHG 4.7.2; emission levels, including the effect of net zero emission emissions and a balance 5.2.2; 5.2.3 targets adopted by countries as part of their long-term between anthropogenic and 5.2.4; climate strategies, can be important when assessing whether sources and 7.6 the collective level of mitigation action is consistent with anthropogenic sinks? long-term goals of the PA. Understanding the dynamics of natural sources of CO2, CH4 and N2O is a fundamental prerequisite to derive climate projections. Net zero GHG emissions, that is, the balance between anthropogenic sources and anthropogenic sinks of CO2 and other greenhouse gases, will halt human-induced global warming and/or lead to slight reversal below peak warming levels. Net zero CO2 emissions will approximately lead to a stabilisation of CO2-induced global warming. What is the remaining 5.5 The remaining carbon budget provides an estimate of how carbon budget that is much CO2 can still be emitted into the atmosphere by consistent with the PA’s human activities while keeping global mean surface long-term temperature temperature to a specific warming level. It thus provides key goals? geophysical information about emissions limits consistent with limiting global warming to well below 2°C above pre- industrial levels and to pursue efforts to limit the temperature increase to 1.5°C. Remaining carbon budgets can be seen in the context of historical CO2 emissions to date. The concept of the transient climate response to cumulative emissions of CO2 (TCRE) indicates that one tonne of CO2 has the same effect on global warming irrespective of whether it is emitted in the past, today, or in the future. In contrast, the global warming from short-lived climate forcers is dependent on their rate of emission rather than their cumulative emission. What is our current Cross- Synthesis information on projected changes in indices of knowledge on the Chapter climatic impact-drivers feeds into different ‘Reasons for ‘Reasons for Concern’ Box 12.1; Concern’. Where possible, an explicit transfer function related to the PA’s long- with between different warming levels and indices quantifying term temperature goals individual characteristics of these hazards is provided, or the and higher warming domains difficulties in doing so documented. Those indices include levels? discussed Arctic sea ice area in September; global average change in in 2.3.3, ocean acidification; volume of glaciers or snow cover; ice 3.5.4, volume change for the West Antarctic Ice Sheet (WAIS) 4.3.2; 5.3; and Greenland Ice Sheet (GIS); Atlantic Meridional 8.4.1; Overturning Circulation (AMOC) strength ; amplitude and 9.4.2, 9.5; variance of El Niño Southern Oscillation (ENSO) mode 11; 12 (Nino3.4 index); and weather and climate extremes. What are the climate 6.6.3 ; Understanding to what degree rapid decarbonisation effects and air pollution 6.7.3 ; Box strategies bring about reduced air pollution due to co-benefits of rapid 6.2 reductions in co-emitted short-lived climate forcers can be decarbonisation due to useful to consider integrated and/or complementary policies, the reduction of co- with synergies for pursuing the PA goals, the World Health emitted short-lived Organization (WHO) air quality guidelines and the climate forcers (SLCF)? Sustainable Development Goals (SDGs). Do Not Cite, Quote or Distribute 1-24 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI What are the Equilibrium Box 4.1; Equilibrium Climate Sensitivity (ECS) measures the long- Climate Sensitivity, the 5.4, 5.5.1, term global-mean warming in response to doubling CO2 Transient Climate 7.5 concentrations from pre-industrial levels, while Transient Response, and Transient Climate Response (TCR) also takes into account the inertia Climate Response to of the climate system and is an indicator for the near- and Emissions and what do medium-term warming. TCRE is similar to TCR, but asks these indicators tell us the question of what is the implied warming is in response about expected warming to cumulative CO2 emissions (rather than CO2 concentration over the 21st century changes). The higher the ECS, TCR or TCRE, the lower are under various scenarios? the greenhouse gas emissions that are consistent with the PA’s long-term temperature goals. What is the Earth's 7.2.2 The current global energy imbalance implies that one can energy imbalance and expect additional warming before the Earth’s climate why does it matter? system attains equilibrium with the current level of concentrations and radiative forcing. Note though, that future warming commitments can be different depending on how future concentrations and radiative forcing change. What are the regional and 8.4.1, 8.5; Changes in regional precipitation – in terms of both long-term changes in 8.6; 10.4; extremes and long-term averages – are important for precipitation, evaporation 10.6; 12.4; estimating adaptation challenges. Projected changes of and runoff? 11.4; 11.9; precipitation minus evaporation (P-E) are closely related to 11.6; 11.7; surface water availability and drought probability. Atlas; Understanding water cycle changes over land, including Interactive seasonality, variability and extremes, and their uncertainties, Atlas is important to estimate a broad range of climate impacts and adaptation, including food production, water supply and ecosystem functioning. Are we committed to 4.7.2; Unlike many regional climate responses, global-mean sea irreversible sea level rise 9.6.3; level keeps rising even in the lowest scenarios and is not and what is the expected 9.6.4; 12.4; halted when warming is halted. This is due to the long sea level rise by the end Interactive timescales on which ocean heat uptake, glacier melt, ice of the century if we Atlas sheets react to temperature changes. Tipping points and pursue strong mitigation thresholds in polar ice sheets need to be considered. Thus, or high emission sea level rise commitments and centennial-scale scenarios? irreversibility of ocean warming and sea level rise are important for future impacts under even the lowest of the emission scenarios. Can we project future 11, 12.4; Projections of future weather and climate extreme events climate extremes under Interactive and their regional occurrence, including at different global various global warming Atlas warming levels, are important for adaptation and disaster levels in the long term? risk reduction. The attribution of these extreme events to natural variability and human-induced changes can be of relevance for both assessing adaptation challenges and issues of loss and damage. Do Not Cite, Quote or Distribute 1-25 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI What is the current 1.4.4; From a risk perspective, it is useful to have information of knowledge of potential 4.7.2; 4.8; lower probability events and system changes, if they could surprises, abrupt changes, 5.4.8; Box potentially result in high impacts, given the dynamic tipping points and low- 5.1; interactions between climate-related hazards and socio- likelihood, high impact 8.5.3.2; economic drivers (exposure, vulnerability of the affected events related to different 8.6.2; Box human or ecological systems). Examples include permafrost levels of future emissions 9.4; 11.2.4; thaw, CH4 clathrate feedbacks, ice sheet mass loss, ocean or warming? Cross- turnover circulation changes, either accelerating warming Chapter globally or yielding particular regional responses and Box 4.1; impacts. Cross- Chapter Box 12.1 Section 3: The near term. – ‘How do we get there?’ WGI assessment to inform near-term adaptation and mitigation options Questions Chapter Potential Relevance and Explanatory Remarks What are projected key 4.3; 4.4; Much of the near-term information and comparison to climate indices under FAQ 4.1, historical observations allows us to quantify the climate low, medium and high 10.6; 12.3; adaptation challenges for the next decades as well as the emission scenarios in the Atlas; opportunities to reduce climate change by pursuing lower near term, that is, the Interactive emissions. For this timescale both the forced changes and the next 20 years? Atlas internal variability are important. How can the climate 7.6 For mitigation challenges, it is important to compare efforts to benefit of mitigating reduce emissions of CO2 versus emissions of other climate emissions of different forcers, such as, short-lived CH4 or long-lived N2O. Global greenhouse gases be Warming Potentials (GWPs), which are used in the UNFCCC compared? and in emission inventories, are updated and various other metrics are also investigated. While the NDCs of Parties to the PA, emission inventories under the UNFCCC, and various emission trading schemes work on the basis of GWP- weighted emissions, some recent discussion in the scientific literature also considers projecting temperatures induced by short-lived climate forcers on the basis of emission changes, not emissions per se. Do mountain glaciers 2.3.2.3; Mountain glaciers and seasonal snow cover often feed shrink currently and in 9.5; downstream river systems during the melting period, and can the near-future in regions Cross- be an important source of freshwater. Changing river that are currently chapter discharge can pose adaptation challenges. Melting mountain dependent on this Box 10.4; glaciers are among the main contributors to observed global seasonal freshwater 12.4: mean sea level rise. supply? 8.4.1; Atlas.5.2. 2; Atlas.5.3. 2; Atlas.6.2; Atlas.9.2 Do Not Cite, Quote or Distribute 1-26 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI What are the capacities 10.5;10.6; Challenges for adaptation and risk management are and limitations in the Box 10.2; predominantly local, even if globally interlinked. There are a provision of regional Cross- number of approaches used in the production of regional climate information for Chapter climate information for adaptation purposes focusing on adaptation and risk Box 10.4; regional scales. All of them consider a range of sources of management? 11.9; data and knowledge that are distilled into, at times contextual, Cross- climate information. A wealth of examples can be found in Chapter this Report, including assessments of extremes and climatic Box 1.3; impact-drivers, and attribution at regional scales. Specific 12.6; regions and case studies for regional projections are Cross- considered, like the Sahel and West African monsoon drought Chapter and recovery, the Southern Australian rainfall decline, the Box 12.1 Caribbean small island summer drought, and regional projections are discussed for Cape Town, the Mediterranean region and Hindu Kush Himalaya. How important are 6.1; 6.6; While most of the radiative forcing which causes climate reductions in short-lived 6.7; 7.6 change comes from CO2 emissions, short-lived climate climate forcers compared forcers also play an important role in the anthropogenic effect to the reduction of CO2 on climate change. Many aerosol species, especially SO4, tend and other long-lived to cool the climate and their reduction leads to a masking of greenhouse gases? greenhouse gas induced warming. On the other hand, many short-lived species themselves exert a warming effect, including black carbon and CH4, the second most important anthropogenic greenhouse gas (in terms of current radiative forcing). Notably, the climate response to aerosol emissions has a strong regional pattern and is different from that of greenhouse gas driven warming. What are potential co- 5.6.2; 6.1; The reduction of fossil-fuel-related emissions often goes benefits and side-effects 6.7.5 hand-in-hand with a reduction of air pollutants, such as of climate change aerosols and ozone. Reductions will improve air quality and mitigation? result in broader environmental benefits (reduced acidification, eutrophication, and often tropospheric ozone recovery). More broadly, various co-benefits are discussed in WGII and WGIII, as well as co-benefits and side-effects related to certain mitigation actions, like increased biomass use and associated challenges to food security and biodiversity conservation. What large near-term 1.4; 4.4.4; Surprises can come from a range of sources: from incomplete surprises could result in Cross- understanding of the climate system, from surprises in particular adaptation Chapter emissions of natural (e.g., volcanic) sources, or from challenges? Box 4.1; disruptions to the carbon cycle associated with a warming 8.5.2; climate (e.g., methane release from permafrost thawing, 11.2.4; tropical forest dieback). There could be large natural Cross- variability in the near-term; or also accelerated climate Do Not Cite, Quote or Distribute 1-27 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI Chapter change due to a markedly more sensitive climate than Box 12.1 previously thought. When the next large explosive volcanic eruption will happen is unknown. The largest volcanic eruptions over the last few hundred years led to substantial but temporary cooling, including precipitation changes. 1 2 [END CROSS-CHAPTER BOX 1.1, TABLE 1 HERE] 3 4 5 [END CROSS-CHAPTER BOX 1.1 HERE] 6 7 8 Sustainable Development Goals (SDGs). Many interactions among environmental problems and 9 development are addressed in the United Nations 2030 Agenda for Sustainable Development and its 10 Sustainable Development Goals. The 2030 Agenda, supported by the finance-oriented Addis Ababa Action 11 Agenda (UN DESA, 2015), calls on nations to ‘take the bold and transformative steps which are urgently 12 needed to shift the world onto a sustainable and resilient path.’ The 2030 Agenda recognizes that ‘climate 13 change is one of the greatest challenges of our time and its adverse impacts undermine the ability of all 14 countries to achieve sustainable development.’ SDG 13 deals explicitly with climate change, establishing 15 several targets for adaptation, awareness-raising and finance. Climate and climate change are also highly 16 relevant to most other SDGs, while acknowledging UNFCCC as the main forum to negotiate the global 17 response to climate change. For example, both long-lived greenhouse gases (LLGHGs), through mitigation 18 decisions, and SLCFs, through air quality, are relevant to SDG 11 (sustainable cities and communities). 19 Chapter 6 assesses SLCF effects on climate and the implications of changing climate for air quality, 20 including opportunities for mitigation relevant to the SDGs (Chapter 6, Box 6.2). Also, the UN Conference 21 on Housing and Sustainable Development established a New Urban Agenda (United Nations, 2017) 22 envisaging cities as part of the solutions for sustainable development, climate change adaptation and 23 mitigation. 24 25 The Sendai Framework for Disaster Risk Reduction (SFDRR). The Sendai Framework for Disaster Risk 26 Reduction is a non-binding agreement to reduce risks associated with disasters of all scales, frequencies and 27 onset rates caused by natural or human-made hazards, including climate change. The SFDRR outlines targets 28 and priorities for action including ‘Understanding disaster risk’, along the dimensions of vulnerability, 29 exposure of persons and assets and hazard characteristics. Chapter 12 assesses climate information relevant 30 to regional impact and risk assessment with a focus on climate hazards and other aspects of climate that 31 influence society and ecosystem and makes the link with Working Group II. AR6 adopts a consistent risk 32 and solution-oriented framing (Cross-Chapter Box 1.3) that calls for a multidisciplinary approach and cross- 33 working group coordination in order to ensure integrative discussions of major scientific issues associated 34 with integrative risk management and sustainable solutions (IPCC, 2017). 35 36 The Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES). 37 Efforts to address climate change take place alongside and in the context of other major environmental 38 problems, such as biodiversity loss. The Intergovernmental Science-Policy Platform on Biodiversity and 39 Ecosystem Services (IPBES), established in 2012, builds on the IPCC model of a science-policy interface 40 and assessment. The Platform's objective is to ‘strengthen the science-policy interface for biodiversity and 41 ecosystem services for the conservation and sustainable use of biodiversity, long-term human well-being and 42 sustainable development’ (UNEP, 2012). SROCC (IPCC, 2019b) and SRCCL (IPCC, 2019a) assessed the 43 relations between changes in biodiversity and in the climate system. The rolling work programme of IPBES 44 up to 2030 will address interlinkages among biodiversity, water, food and health. This assessment will use a 45 nexus approach to examine interlinkages between biodiversity and the above-mentioned issues, including 46 climate change mitigation and adaptation. Furthermore, IPBES and IPCC will directly collaborate on 47 biodiversity and climate change under the rolling work programme. 48 49 Addressing climate change alongside other environmental problems, while simultaneously supporting Do Not Cite, Quote or Distribute 1-28 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 sustainable socioeconomic development, requires a holistic approach. Since AR5, there is increasing 2 attention on the need for coordination among previously independent international agendas, recognizing that 3 climate change, disaster risk, economic development, biodiversity conservation and human well-being are 4 tightly interconnected. The current COVID-19 pandemic provides an example of the need for such 5 interconnection, with its widespread impacts on economy, society and environment (e.g., Shan et al., 2020). 6 Cross Chapter Box 6.1 in Chapter 6 assesses the consequences of the COVID-19 lockdowns on emissions of 7 GHGs and SLCFs and related implications for the climate. Another example is the close link between SLCF 8 emissions, climate evolution and air quality concerns (see Chapter 6). Emissions of halocarbons have 9 previously been successfully regulated under the Montreal Protocol and its Kigali Amendment. This has 10 been achieved in an effort to reduce ozone depletion that has also modulated anthropogenic climate influence 11 (Estrada et al., 2013; Wu et al., 2013). In the process, emissions of some SLCFs are jointly regulated to 12 reduce environmental and health impacts from air pollution (e.g., Gothenburg Protocol; Reis et al., 2012). 13 Considering the recognized importance of SLCFs for climate, the IPCC decided in May 2019 to approve that 14 the IPCC Task Force on National Greenhouse Gas Inventories produces an IPCC Methodology Report on 15 SLCFs to develop guidance for national SLCFs inventories. 16 17 The evolving governance context since AR5 challenges the IPCC to provide policymakers and other actors 18 with information relevant for both adaptation to and mitigation of climate change and for the loss and 19 damage induced. 20 21 22 1.2.3 Linking science and society: communication, values, and the IPCC assessment process 23 24 This section assesses how the process of communicating climate information has evolved since AR5. It 25 summarizes key issues regarding scientific uncertainty addressed in previous IPCC assessments and 26 introduces the IPCC calibrated uncertainty language. Next it discusses the role of values in problem-driven, 27 multidisciplinary science assessments such as this one. The section introduces climate services and how 28 climate information can be tailored for greatest utility in specific contexts, such as the global stocktake. 29 Finally, we briefly evaluate changes in media coverage of climate information since AR5, including the 30 increasing role of internet sources and social media. 31 32 33 1.2.3.1 Climate change understanding, communication, and uncertainties 34 35 The response to climate change is facilitated when leaders, policymakers, resource managers, and their 36 constituencies share basic understanding of the causes, effects, and possible future course of climate change 37 (SR1.5, IPCC, 2018; SRCCL, IPCC, 2019a). Achieving shared understanding is complicated, since scientific 38 knowledge interacts with pre-existing conceptions of weather and climate built up in diverse world cultures 39 over centuries and often embedded in strongly held values and beliefs stemming from ethnic or national 40 identities, traditions, religion, and lived relationships to weather, land and sea (Van Asselt and Rotmans, 41 1996; Rayner and Malone, 1998; Hulme, 2009, 2018; Green et al., 2010; Jasanoff, 2010; Orlove et al., 2010; 42 Nakashima et al., 2012; Shepherd and Sobel, 2020). These diverse, more local understandings can both 43 contrast with and enrich the planetary-scale analyses of global climate science (high confidence). 44 45 Political cultures also give rise to variation in how climate science knowledge is interpreted, used, and 46 challenged (Leiserowitz, 2006; Oreskes and Conway, 2010; Brulle et al., 2012; Dunlap and Jacques, 2013; 47 Mahony, 2014, 2015; Brulle, 2019). A meta-analysis of 87 studies carried out between 1998 and 2016 (62 48 USA national, 16 non-USA national, 9 cross-national) found that political orientation and political party 49 identification were the second-most important predictors of views on climate change after environmental 50 values (the strongest predictor) (McCright et al. 2016). Ruiz et al. (2020) systematically reviewed 34 studies 51 of non-US nations or clusters of nations and 30 studies of the USA alone. They found that in the non-US 52 studies, ‘changed weather’ and ‘socio-altruistic values’ were the most important drivers of public attitudes. 53 For the USA case, by contrast, political affiliation and the influence of corporations were most important. 54 Widely varying media treatment of climate issues also affects public responses (see Section 1.2.3.4). In 55 summary, environmental and socio-altruistic values are the most significant influences on public opinion Do Not Cite, Quote or Distribute 1-29 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 about climate change globally, while political views, political party affiliation, and corporate influence also 2 had strong effects, especially in the USA (high confidence). 3 4 Furthermore, climate change itself is not uniform. Some regions face steady, readily observable change, 5 while others experience high variability that masks underlying trends (Section 1.4.1); most regions are 6 subject to hazards, but some may also experience benefits, at least temporarily (see Chapters 11, 12, and 7 Atlas). This non-uniformity may lead to wide variation in public climate change awareness and risk 8 perceptions at multiple scales (Howe et al., 2015; Lee et al., 2015). For example, short-term temperature 9 trends, such as cold spells or warm days, have been shown to influence public concern (Hamilton and 10 Stampone, 2013; Zaval et al., 2014; Bohr, 2017). 11 12 Given these manifold influences and the highly varied contexts of climate change communication, special 13 care is required when expressing findings and uncertainties, including IPCC assessments that inform 14 decision making. Throughout the IPCC’s history, all three Working Groups (WGs) have sought to explicitly 15 assess and communicate scientific uncertainty (Le Treut et al., 2007; Cubasch et al., 2013). Over time, the 16 IPCC has developed and revised a framework to treat uncertainties consistently across assessment cycles, 17 reports, and Working Groups through the use of calibrated language (Moss and Schneider, 2000; IPCC, 18 2005). Since its First Assessment Report (IPCC, 1990a), the IPCC specified terms and methods for 19 communicating authors’ expert judgments (Mastrandrea and Mach, 2011). During the AR5 cycle, this 20 calibrated uncertainty language was updated and unified across all WGs (Mastrandrea et al., 2010, 2011). 21 Box 1.1 summarizes this framework as used in AR6. 22 23 24 [START BOX 1.1 HERE] 25 26 Box 1.1: Treatment of uncertainty and calibrated uncertainty language in AR6 27 28 The AR6 follows the approach developed for AR5 (Box 1.1, Figure 1), as described in the ‘Guidance Notes 29 for Lead Authors of the IPCC Fifth Assessment Report on Consistent Treatment of Uncertainties’ 30 (Mastrandrea et al., 2010). The uncertainty Guidance Note used in AR6 clarifies the relationship between the 31 qualitative description of confidence and the quantitative representation of uncertainty expressed by the 32 likelihood scale. The calibrated uncertainty language emphasizes traceability of the assessment throughout 33 the process. Key chapter findings presented in the chapter Executive Summary are supported in the chapter 34 text by a summary of the underlying literature that is assessed in terms of evidence and agreement, 35 confidence, and also likelihood if applicable. 36 37 In all three WGs, author teams evaluate underlying scientific understanding and use two metrics to 38 communicate the degree of certainty in key findings. These metrics are: 39 40 1. Confidence: a qualitative measure of the validity of a finding, based on the type, amount, quality and 41 consistency of evidence (e.g., data, mechanistic understanding, theory, models, expert judgment) and 42 the degree of agreement. 43 2. Likelihood: a quantitative measure of uncertainty in a finding, expressed probabilistically (e.g., 44 based on statistical analysis of observations or model results, or both, and expert judgement by the 45 author team or from a formal quantitative survey of expert views, or both). 46 47 Throughout IPCC reports, the calibrated language indicating a formal confidence assessment is clearly 48 identified by italics (e.g., medium confidence). Where appropriate, findings can also be formulated as 49 statements of fact without uncertainty qualifiers. 50 51 52 [START BOX 1.1, FIGURE 1 HERE] 53 54 Box 1.1, Figure 1: The IPCC AR6 approach for characterizing understanding and uncertainty in assessment findings. 55 This diagram illustrates the step-by-step process authors use to evaluate and communicate the state Do Not Cite, Quote or Distribute 1-30 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 of knowledge in their assessment (Mastrandrea et al., 2010). Authors present evidence/agreement, 2 confidence, or likelihood terms with assessment conclusions, communicating their expert judgments 3 accordingly. Example conclusions drawn from this report are presented in the box at the bottom of 4 the figure. [adapted from Mach et al. (2017)]. 5 6 [END BOX 1.1, FIGURE 1 HERE] 7 8 9 Box.1.1, Figure 1 (adapted from Mach et al., 2017) shows the idealized step-by-step process by which IPCC 10 authors assess scientific understanding and uncertainties. It starts with the evaluation of the available 11 evidence and agreement (Steps 1–2). The following summary terms are used to describe the available 12 evidence: limited, medium, or robust; and the degree of agreement: low, medium, or high. Generally, 13 evidence is most robust when there are multiple, consistent, independent lines of high-quality evidence. 14 15 If the author team concludes that there is sufficient evidence and agreement, the level of confidence can be 16 evaluated. In this step, assessments of evidence and agreement are combined into a single metric (Steps 3–5). 17 The assessed level of confidence is expressed using five qualifiers: very low, low, medium, high, and very 18 high. Step 4 depicts how summary statements for evidence and agreement relate to confidence levels. For a 19 given evidence and agreement statement, different confidence levels can be assigned depending on the 20 context, but increasing levels of evidence and degrees of agreement correlate with increasing confidence. 21 When confidence in a finding is assessed to be low, this does not necessarily mean that confidence in its 22 opposite is high, and vice versa. Similarly, low confidence does not imply distrust in the finding; instead, it 23 means that the statement is the best conclusion based on currently available knowledge. Further research and 24 methodological progress may change the level of confidence in any finding in future assessments. 25 26 If the expert judgement of the author team concludes that there is sufficient confidence and 27 quantitative/probabilistic evidence, assessment conclusions can be expressed with likelihood statements 28 (Box.1.1, Figure 1, Steps 5–6). Unless otherwise indicated, likelihood statements are related to findings for 29 which the authors’ assessment of confidence is ‘high’ or ‘very high’. Terms used to indicate the assessed 30 likelihood of an outcome include: virtually certain: 99–100% probability, very likely: 90–100%, likely: 66– 31 100%, about as likely as not: 33–66%, unlikely: 0–33%, very unlikely: 0–10%, exceptionally unlikely: 0–1%. 32 Additional terms (extremely likely: 95–100%, more likely than not >50–100%, and extremely unlikely 0–5%) 33 may also be used when appropriate. 34 35 Likelihood can indicate probabilities for single events or broader outcomes. The probabilistic information 36 may build from statistical or modelling analyses, other quantitative analyses, or expert elicitation. The 37 framework encourages authors, where appropriate, to present probability more precisely than can be done 38 with the likelihood scale, for example with complete probability distributions or percentile ranges, including 39 quantification of tails of distributions important for risk management (Mach et al., 2017; see also Sections 40 1.2.2 and 1.4.4). In some instances, multiple combinations of confidence and likelihood are possible to 41 characterize key findings. For example, a very likely statement might be made with high confidence, whereas 42 a likely statement might be made with very high confidence. In these instances, the author teams consider 43 which statement will convey the most balanced information to the reader. 44 45 Throughout this WGI report and unless stated otherwise, uncertainty is quantified using 90% uncertainty 46 intervals. The 90% uncertainty interval, reported in square brackets [x to y], is estimated to have a 90% 47 likelihood of covering the value that is being estimated. The range encompasses the median value and there 48 is an estimated 10% combined likelihood of the value being below the lower end of the range (x) and above 49 its upper end (y). Often the distribution will be considered symmetric about the corresponding best estimate 50 (as in the illustrative example in the figure), but this is not always the case. In this report, an assessed 90% 51 uncertainty interval is referred to as a ‘very likely range’. Similarly, an assessed 66% uncertainty interval is 52 referred to as a ‘likely range’. 53 54 [END BOX 1.1 HERE] 55 Do Not Cite, Quote or Distribute 1-31 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Considerable critical attention has focused on whether applying the IPCC framework effectively achieves 3 consistent treatment of uncertainties and clear communication of findings to users (Shapiro et al., 2010; 4 Adler and Hirsch Hadorn, 2014). Specific concerns include, for example, the transparency and traceability of 5 expert judgements underlying the assessment conclusions (Oppenheimer et al., 2016) and the context- 6 dependent representations and interpretations of probability terms (Budescu et al., 2009, 2012; Janzwood, 7 2020). Budescu et al. (2014) surveyed 25 samples in 24 countries (a total of 10,792 individual responses), 8 finding that even when shown IPCC uncertainty guidance, lay readers systematically misunderstood IPCC 9 likelihood statements. When presented with a ‘high likelihood’ statement, they understood it as indicating a 10 lower likelihood than intended by the IPCC authors. Conversely, they interpreted ‘low likelihood’ statements 11 as indicating a higher likelihood than intended. In another study, British lay readers interpreted uncertainty 12 language somewhat differently from IPCC guidance, but Chinese lay people reading the same uncertainty 13 language translated into Chinese differed much more in their interpretations (Harris et al., 2013). Further, 14 even though it is objectively more probable that wide uncertainty intervals will encompass true values, wide 15 intervals were interpreted by lay people as implying subjective uncertainty or lack of knowledge on the part 16 of scientists (Løhre et al., 2019). Mach et al. (2017) investigated the advances and challenges in approaches 17 to expert judgment in the AR5. Their analysis showed that the shared framework increased the overall 18 comparability of assessment conclusions across all WGs and topics related to climate change, from the 19 physical science basis to resulting impacts, risks, and options for response. Nevertheless, many challenges in 20 developing and communicating assessment conclusions persist, especially for findings drawn from multiple 21 disciplines and Working Groups, for subjective aspects of judgments, and for findings with substantial 22 uncertainties (Adler and Hirsch Hadorn, 2014). In summary, the calibrated language cannot entirely prevent 23 misunderstandings, including a tendency to systematically underestimate the probability of the IPCC’s 24 higher-likelihood conclusions and overestimate the probability of the lower-likelihood ones (high 25 confidence), however a consistent and systematic approach across Working Groups to communicate the 26 assessment outcomes is an important characteristic of the IPCC. 27 28 Some suggested alternatives are impractical, such as always including numerical values along with calibrated 29 language (Budescu et al., 2014). Others, such as using positive instead of negative expressions of low to 30 medium probabilities, show promise but were not proposed in time for adoption in AR6 (Juanchich et al., 31 2020). This report therefore retains the same calibrated language used in AR5 (Box 1.1). Like previous 32 reports, AR6 also includes FAQs that express its chief conclusions in plain language designed for lay 33 readers. 34 35 The framework for communicating uncertainties does not address when "deep uncertainty" is identified in 36 the assessment (Adler and Hirsch Hadorn, 2014). The definition of deep uncertainty in IPCC assessments 37 has been described in the context of the SROCC (IPCC, 2019b; Box 5 in Abram et al. (2019)). A situation of 38 deep uncertainty exists when experts or stakeholders do not know or cannot agree on: (1) appropriate 39 conceptual models that describe relationships among key driving forces in a system; (2) the probability 40 distributions used to represent uncertainty about key variables and parameters; and/or (3) how to weigh and 41 value desirable alternative outcomes (Abram et al., 2019). (See also Cross-Chapter Box 1.2, Annex VII 42 Glossary) Since AR5, the ‘storylines’ or ‘narratives’ approach has been used to address issues related to deep 43 uncertainty, for example low-likelihood events that would have high impact if they occurred, to better inform 44 risk assessment and decision making (see Section 1.4.4). Chapter 9 (Section 9.2.3) notes deep uncertainty in 45 long term projections for sea level rise, and in processes related to Marine Ice Sheet Instability and Marine 46 Ice Cliff Instability. 47 48 49 1.2.3.2 Values, science, and climate change communication 50 51 As noted above, values — fundamental attitudes about what is important, good, and right — play critical 52 roles in all human endeavours, including climate science. In AR5, Chapters 3 and 4 of the WGIII assessment 53 addressed the role of cultural, social, and ethical values in climate change mitigation and sustainable 54 development (Fleurbaey et al., 2014; Kolstad et al., 2014). These values include widely accepted concepts of 55 human rights, enshrined in international law, that are relevant to climate impacts and policy objectives (Hall Do Not Cite, Quote or Distribute 1-32 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 and Weiss, 2012; Peel and Osofsky, 2018; Setzer and Vanhala, 2019). Specific values – human life, 2 subsistence, stability, and equitable distribution of the costs and benefits of climate impacts and policies – 3 are explicit in the texts of the UNFCCC and the PA (Breakey et al., 2016; Dooley and Parihar, 2016). Here 4 we address the role of values in how scientific knowledge is created, verified, and communicated. Chapters 5 10, 12, and Cross-Chapter Box 12.2 address how the specific values and contexts of users can be addressed 6 in the co-production of climate information. 7 8 The epistemic (knowledge-related) values of science include explanatory power, predictive accuracy, 9 falsifiability, replicability, and justification of claims by explicit reasoning (Popper, 1959; Kuhn, 1977). 10 These are supported by key institutional values, including openness, ‘organized scepticism,’ and objectivity 11 or ‘disinterestedness’ (Merton, 1973), operationalized as well-defined methods, documented evidence, 12 publication, peer review, and systems for institutional review of research ethics (COSEPUP, 2009). In recent 13 decades, open data, open code, and scientific cyberinfrastructure (notably the Earth System Grid Federation, 14 a partnership of climate modeling centers dedicated to supporting climate research by providing secure, web- 15 based, distributed access to climate model data) have facilitated scrutiny from a larger range of participants, 16 and FAIR data stewardship principles – making data Findable, Accessible, Interoperable and Reusable 17 (FAIR) – are being mainstreamed in many fields (Wilkinson et al., 2016). Climate science norms and 18 practices embodying these scientific values and principles include the publication of data and model code, 19 multiple groups independently analysing the same problems and data, model intercomparison projects 20 (MIPs), explicit evaluations of uncertainty, and comprehensive assessments by national academies of science 21 and the IPCC. 22 23 The formal Principles Governing IPCC Work (1998, amended 2003, 2006, 2012, 2013) specify that 24 assessments should be ‘comprehensive, objective, open and transparent.’ The IPCC assessment process 25 seeks to achieve these goals in several ways: by evaluating evidence and agreement across all relevant peer- 26 reviewed literature, especially that published or accepted since the previous assessment; by maintaining a 27 traceable, transparent process that documents the reasoning, data, and tools used in the assessment; and by 28 maximizing the diversity of participants, authors, experts, reviewers, institutions, and communities 29 represented, across scientific discipline, geographical location, gender, ethnicity, nationality, and other 30 characteristics. The multi-stage review process is critical to ensure an objective, comprehensive and robust 31 assessment, with hundreds of scientists, other experts, and governments providing comments to a series of 32 drafts before the report is finalised. 33 34 Social values are implicit in many choices made during the construction, assessment, and communication of 35 climate science information (Heymann et al., 2017; Skelton et al., 2017). Some climate science questions are 36 prioritised for investigation, or given a specific framing or context, because of their relevance to climate 37 policy and governance. One example is the question of how the effects of a 1.5°C global warming would 38 differ from those of a 2°C warming, an assessment specifically requested by Parties to the PA. SR1.5 (2018) 39 explicitly addressed this issue ‘within the context of sustainable development; considerations of ethics, 40 equity and human rights; and the problem of poverty’ (Chapters 1 and 5; see also Hoegh-Guldberg et al., 41 2019) following the outcome of the approval of the outline of the Special Report by the IPCC during its 44th 42 Session (Bangkok, Thailand, 17-20 October 2016). Likewise, particular metrics are sometimes prioritized in 43 climate model improvement efforts because of their practical relevance for specific economic sectors or 44 stakeholders. Examples include reliable simulation of precipitation in a specific region, or attribution of 45 particular extreme weather events to inform rebuilding and future policy (see Chapters 8 and 11; Intemann, 46 2015; Otto et al., 2018; James et al., 2019). Sectors or groups whose interests do not influence research and 47 modelling priorities may thus receive less information in support of their climate-related decisions (Parker 48 and Winsberg, 2018). 49 50 Recent work also recognizes that choices made throughout the research process can affect the relative 51 likelihood of false alarms (overestimating the probability and/or magnitude of hazards) or missed warnings 52 (underestimating the probability and/or magnitude of hazards), known respectively as Type I and Type II 53 errors. Researchers may choose different methods depending on which type of error they view as most 54 important to avoid, a choice that may reflect social values (Douglas, 2009; Knutti, 2018; Lloyd and Oreskes, 55 2018). This reflects a fundamental trade-off between the values of reliability and informativeness. When Do Not Cite, Quote or Distribute 1-33 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 uncertainty is large, researchers may choose to report a wide range as ‘very likely’, even though it is less 2 informative about potential consequences. By contrast, high-likelihood statements about a narrower range 3 may be more informative, yet also prove less reliable if new evidence later emerges that widens the range. 4 Furthermore, the difference between narrower and wider uncertainty intervals has been shown to be 5 confusing to lay readers, who often interpret wider intervals as less certain (Løhre et al., 2019). 6 7 8 1.2.3.3 Climate information, co-production, and climate services 9 10 In AR6, ‘climate information’ refers to specific information about the past, current, or future state of the 11 climate system that is relevant for mitigation, adaptation and risk management. Cross-Chapter Box 1.1 is an 12 example of climate information at the global scale. It provides climate change information potentially 13 relevant to the global stocktake, and indicates where in AR6 this information may be found. 14 15 Responding to national and regional policymakers' needs for tailored information relevant to risk assessment 16 and adaptation, AR6 emphasizes assessment of regional information more than earlier reports. Here the 17 phrase ‘regional climate information’ refers to predefined reference sets of land and ocean regions; various 18 typological domains (such as mountains or monsoons); temporal frames including baseline periods as well as 19 near-term (2021–2040), medium-term (2041–2060), and long-term (2081–2100); and global warming levels 20 (Sections 1.4.1 and 1.4.5; Chapters 10, 12, and Atlas). Regional climate change information is constructed 21 from multiple lines of evidence including observations, paleoclimate proxies, reanalyses, attribution of 22 changes and climate model projections from both global and regional climate models (Section 1.5.3, Chapter 23 10, Section 10.2 to 10.4). The constructed regional information needs to take account of user context and 24 values for risk assessment, adaptation and policy decisions (Section 1.2.3, Chapter 10, Section 10.5). 25 26 As detailed in Chapter 10, scientific climate information often requires ‘tailoring’ to meet the requirements 27 of specific decision-making contexts. In a study of the UK Climate Projections 2009 project, researchers 28 concluded that climate scientists struggled to grasp and respond to users’ information needs because they 29 lacked experience interacting with users, institutions, and scientific idioms outside the climate science 30 domain (Porter and Dessai, 2017). Economic theory predicts the value of ‘polycentric’ approaches to climate 31 change informed by specific global, regional, and local knowledge and experience (Ostrom, 1996, 2012). 32 This is confirmed by numerous case studies of extended, iterative dialogue among scientists, policymakers, 33 resource managers, and other stakeholders to produce mutually understandable, usable, task-related 34 information and knowledge, policymaking and resource management around the world (Lemos and 35 Morehouse, 2005; Lemos et al., 2012, 2014, 2018; see Vaughan and Dessai, 2014 for a critical view). SR1.5 36 (2018) assessed that ‘education, information, and community approaches, including those that are informed 37 by indigenous knowledge and local knowledge, can accelerate the wide-scale behaviour changes consistent 38 with adapting to and limiting global warming to 1.5°C. These approaches are more effective when combined 39 with other policies and tailored to the motivations, capabilities and resources of specific actors and contexts 40 (high confidence).’ These extended dialogic ‘co-production’ and education processes have thus been 41 demonstrated to improve the quality of both scientific information and governance (high confidence) 42 (Chapter 10, Section 10.5; Cross Chapter Box 12.2 in Chapter 12). 43 44 Since AR5, ‘climate services’ have increased at multiple levels (local, national, regional, and global) to aid 45 decision-making of individuals and organizations and to enable preparedness and early climate change 46 action. These services include appropriate engagement from users and providers, are based on scientifically 47 credible information and producer and user expertise, have an effective access mechanism, and respond to 48 the users’ needs (Hewitt et al., 2012; Annex VII Glossary). A Global Framework for Climate Services 49 (GFCS) was established in 2009 by the World Meteorological Organization (WMO) in support of these 50 efforts (Hewitt et al., 2012; Lúcio and Grasso, 2016). Climate services are provided across sectors and 51 timescales, from sub-seasonal to multi-decadal and support co-design and co-production processes that 52 involve climate information providers, resource managers, planners, practitioners and decision makers 53 (Brasseur and Gallardo, 2016; Trenberth et al., 2016; Hewitt et al., 2017). For example, they may provide 54 high-quality data on temperature, rainfall, wind, soil moisture and ocean conditions, as well as maps, risk 55 and vulnerability analyses, assessments, and future projections and scenarios. These data and information Do Not Cite, Quote or Distribute 1-34 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 products may be combined with non-meteorological data, such as agricultural production, health trends, 2 population distributions in high-risk areas, road and infrastructure maps for the delivery of goods, and other 3 socio-economic variables, depending on users’ needs (WMO, 2020a). Cross-chapter Box 12.2 in Chapter 12 4 illustrates the diversity of climate services with three examples from very different contexts. 5 6 The current landscape of climate services is assessed in detail in Chapter 12 (Section 12.6), with a focus on 7 multi-decadal timescales relevant for climate change risk assessment. Other information relevant to 8 improving climate services for decision making includes the assessment of methods to construct regional 9 information (Chapter 10), as well as projections at the regional level (Atlas) relevant for impact and risk 10 assessment in different sectors (Chapter 12). 11 12 13 1.2.3.4 Media coverage of climate change 14 15 Climate services focus on users with specific needs for climate information, but most people learn about 16 climate science findings from media coverage. Since AR5, research has expanded on how mass media report 17 climate change and how their audiences respond (Dewulf, 2013; Jaspal and Nerlich, 2014; Jaspal et al., 18 2014). For example, in five European Union (EU) countries, television coverage of the AR5 used ‘disaster’ 19 and ‘opportunity’ as its principal themes, but virtually ignored the ‘risk’ framing introduced by AR5 WGII 20 (Painter, 2015) and now extended by the AR6 (see Cross-Chapter Box 1.3). Other studies show that people 21 react differently to climate change news when it is framed as a catastrophe (Hine et al., 2015), as associated 22 with local identities (Sapiains et al., 2016), or as a social justice issue (Howell, 2013). Similarly, audience 23 segmentation studies show that responses to climate change vary between groups of people with different, 24 although not necessarily opposed, views on this phenomenon (e.g., Maibach et al., 2011; Sherley et al., 2014; 25 Detenber et al., 2016). In Brazil, two studies have shown the influence of mass media on the high level of 26 public climate change concern in that country (Rodas and DiGiulio, 2017; Dayrell, 2019). In the USA, 27 analyses of television network news show that climate change receives minimal attention, is most often 28 framed in a political context, and largely fails to link extreme weather events to climate change using 29 appropriate probability framing (Hassol et al., 2016). However, recent evidence suggests that Climate 30 Matters (an Internet resource for US TV weathercasters to link weather to climate change trends) may have 31 had a positive effect on public understanding of climate change (Myers et al., 2020). Also, some media 32 outlets have recently adopted and promoted terms and phrases stronger than the more neutral ‘climate 33 change’ and ‘global warming’, including ‘climate crisis’, ‘global heating’, and ‘climate emergency’ (Zeldin- 34 O’Neill, 2019). Google searches on those terms, and on ‘climate action,’ increased 20-fold in 2019, when 35 large social movements such as the School Strikes for Climate gained worldwide attention (Thackeray et al., 36 2020). We thus assess that specific characteristics of media coverage play a major role in climate 37 understanding and perception (high confidence), including how IPCC assessments are received by the 38 general public. 39 40 Since AR5, social media platforms have dramatically altered the mass-media landscape, bringing about a 41 shift from uni-directional transfer of information and ideas to more fluid, multi-directional flows (Pearce et 42 al., 2019). A survey covering 18 Latin American countries (StatKnows-CR2, 2019) found that the main 43 sources of information about climate change mentioned were the Internet (52% of mentions), followed by 44 social media (18%). There are well-known challenges with social media, such as misleading or false 45 presentations of scientific findings, incivility that diminishes the quality of discussion around climate change 46 topics, and ‘filter bubbles’ that restrict interactions to those with broadly similar views (Anderson and 47 Huntington, 2017). However, at certain moments (such as at the release of the AR5 WGI report), Twitter 48 studies have found that more mixed, highly-connected groups existed, within which members were less 49 polarized (Pearce et al., 2014; Williams et al., 2015). Thus, social media platforms may in some 50 circumstances support dialogic or co-production approaches to climate communication. Because the contents 51 of IPCC reports speak not only to policymakers, but also to the broader public, the character and effects of 52 media coverage are important considerations across Working Groups. 53 54 Do Not Cite, Quote or Distribute 1-35 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.3 How we got here: the scientific context 2 3 Scientific understanding of the climate system’s fundamental features is robust and well established. This 4 section briefly presents the major lines of evidence in climate science (Figure 1.6). It illustrates their long 5 history and summarises key findings from the WGI contribution to AR5, where relevant referencing 6 previous IPCC assessments for comparison. Box 1.2 summarises major findings from three Special Reports 7 already released during the sixth IPCC assessment cycle. This chapter’s Appendix 1A summarises the 8 principal findings of all six IPCC WGI Assessment Reports, including the present one, in a single table for 9 ready reference. 10 11 12 [START FIGURE 1.6 HERE] 13 14 Figure 1.6: Climate science milestones, between 1817-2021. Milestones in observations (top); Curves of global 15 surface air temperature (GMST) using HadCRUT5 (Morice et al., 2021) and atmospheric CO2 16 concentrations from Antarctic ice cores (Lüthi et al., 2008; Bereiter et al., 2015) and direct air 17 measurements from 1957 onwards (Tans and Keeling, 2020) (see Figure 1.4 for details) (middle). 18 Milestone in scientific understanding of the CO2 enhanced greenhouse effect (bottom). Further details on 19 each milestone are available in Chapter 1, Section 1.3, and Chapter 1 of AR4. 20 21 [END FIGURE 1.6 HERE] 22 23 24 1.3.1 Lines of evidence: instrumental observations 25 26 Instrumental observations of the atmosphere, ocean, land, biosphere, and cryosphere underpin all 27 understanding of the climate system. This section describes the evolution of instrumental data for major 28 climate variables at Earth’s land and ocean surfaces, at altitude in the atmosphere, and at depth in the ocean. 29 Many data records exist, of varying length, continuity, and spatial distribution; Figure 1.7 gives a schematic 30 overview of temporal coverage. 31 32 Instrumental weather observation at the Earth’s surface dates to the invention of thermometers and 33 barometers in the 1600s. National and colonial weather services built networks of surface stations in the 34 1800s. By the mid-19th century, semi-standardized naval weather logs recorded winds, currents, 35 precipitation, air pressure, and temperature at sea, initiating the longest continuous quasi-global instrumental 36 record (Maury, 1849, 1855, 1860). Because the ocean covers over 70% of global surface area and constantly 37 exchange energy with the atmosphere, both air and sea surface temperatures (SST) recorded in these naval 38 logs are crucial variables in climate studies. Dove (1853) mapped seasonal isotherms over most of the globe. 39 By 1900, a patchy weather data-sharing system reached all continents except Antarctica. Regular 40 compilation of climatological data for the world began in 1905 with the Réseau Mondial (Meteorological 41 Office and Shaw, 1920), and the similar compilations World Weather Records (Clayton, 1927) and Monthly 42 Climatic Data for the World (est. 1948) have been published continuously since their founding. 43 44 Land and ocean surface temperature data have been repeatedly evaluated, refined, and extended (Section 45 1.5.1). As computer power increased and older data were recovered from handwritten records, the number of 46 surface station records used in published global land temperature time series grew. A pioneering study for 47 1880–1935 used fewer than 150 stations (Callendar, 1938). A benchmark study of 1880–2005 incorporated 48 4300 stations (Brohan et al., 2006). A study of the 1753–2011 period included previously unused station 49 data, for a total of 36,000 stations (Rohde et al., 2013); recent versions of this dataset comprise over 40,000 50 land stations (Rohde and Hausfather, 2020). Several centres, including NOAA, Hadley, and Japan 51 Meteorological Agency (JMA), each produce SST datasets independently calculated from instrumental 52 records. In the 2000s, adjustments for bias due to different measurement methods (buckets, engine intake 53 thermometers, moored and drifting buoys) resulted in major improvements of SST data (Thompson et al., 54 2008), and these improvements continue (Huang et al., 2017; Kennedy et al., 2019). SST and land-based data 55 are incorporated into global surface temperature datasets calculated independently by multiple research Do Not Cite, Quote or Distribute 1-36 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 groups, including NOAA, NASA, Berkeley Earth, Hadley-CRU, JMA, and China Meteorological 2 Administration (CMA). Each group aggregates the raw measurement data, applies various adjustments for 3 non-climatic biases such as urban heat-island effects, and addresses unevenness in geospatial and temporal 4 sampling with various techniques (see Chapter 2, Section 2.3.1.1.3 and Table 2.4 for references). Other 5 research groups provide alternative interpolations of these datasets using different methods (e.g., Cowtan and 6 Way, 2014; Kadow et al., 2020). Using the then available global surface temperature datasets, WGI AR5 7 assessed that the global mean surface temperature (GMST) increased by 0.85°C from 1880 to 2012 and 8 found that each of the three decades following 1980 was successively warmer at the Earth’s surface than any 9 preceding decade since 1850 (IPCC, 2013b). Marine air temperatures, especially those measured during 10 night-time, are increasingly also used to examine variability and long-term trends (e.g., Rayner et al., 2006; 11 Kent et al., 2013; Cornes et al., 2020; Junod and Christy, 2020). Cross-Chapter Box 2.3 in Chapter 2 12 discusses updates to the global temperature datasets, provides revised estimates for the observed changes and 13 considers whether marine air temperatures are changing at the same rate as SSTs. 14 15 Data at altitude came initially from scattered mountain summits, balloons, and kites, but the upper 16 troposphere and stratosphere were not systematically observed until radiosonde (weather balloon) networks 17 emerged in the 1940s and 1950s. These provide the longest continuous quasi-global record of the 18 atmosphere’s vertical dimension (Stickler et al., 2010). New methods for spatial and temporal 19 homogenisation (intercalibration and quality control) of radiosonde records were introduced in the 2000s 20 (Sherwood et al., 2008, 2015; Haimberger et al., 2012). Since 1978, Microwave Sounding Units (MSU) 21 mounted on Earth-orbiting satellites have provided a second high-altitude data source, measuring 22 temperature, humidity, ozone, and liquid water throughout the atmosphere. Over time, these satellite data 23 have required numerous adjustments to account for such factors as orbital precession and decay (Edwards, 24 2010). Despite repeated adjustments, however, marked differences remain in the temperature trends from 25 surface, radiosonde, and satellite observations; between the results from three research groups that analyse 26 satellite data (UAH, RSS, and NOAA); and between modelled and satellite-derived tropospheric warming 27 trends (Thorne et al., 2011; Santer et al., 2017). These differences are the subject of ongoing research 28 (Maycock et al., 2018). In the 2000s, Atmospheric Infrared Sounder (AIRS) and radio occultation (GNSS- 29 RO) measurements provided new ways to measure temperature at altitude, complementing data from the 30 MSU. GNSS-RO is a new independent, absolutely calibrated source, using the refraction of radio-frequency 31 signals from the Global Navigation Satellite System (GNSS) to measure temperature, pressure, and water 32 vapour (Chapter 2, Section 2.3.1.2.1; Foelsche et al., 2008; Anthes, 2011). 33 34 Heat-retaining properties of the atmosphere’s constituent gases were closely investigated in the 19th century. 35 Foote, (1856) measured solar heating of CO2 experimentally and argued that higher concentrations in the 36 atmosphere would increase Earth’s temperature. Water vapour, ozone, carbon dioxide, and certain 37 hydrocarbons were found to absorb longwave (infrared) radiation, the principal mechanism of the 38 greenhouse effect (Tyndall, 1861). 19th-century investigators also established the existence of a natural 39 biogeochemical carbon cycle. CO2 emitted by volcanoes is removed from the atmosphere through a 40 combination of silicate rock weathering, deep-sea sedimentation, oceanic absorption, and biological storage 41 in plants, shellfish, and other organisms. On multi-million-year timescales, the compression of fossil organic 42 matter stores carbon as coal, oil, and natural gas (Chamberlin, 1897, 1898; Ekholm, 1901). 43 44 Arrhenius (1896) calculated that a doubling of atmospheric carbon dioxide would produce a 5–6°C warming, 45 but in 1900 new measurements seemed to rule out CO2 as a greenhouse gas due to overlap with the 46 absorption bands of water vapour (Ångström, 1900; Very and Abbe, 1901). Further investigation and more 47 sensitive instruments later overturned Ångström’s conclusion (Fowle, 1917; Callendar, 1938). Nonetheless, 48 the major role of CO2 in the energy balance of the atmosphere was not widely accepted until the 1950s 49 (Callendar, 1949; Plass, 1956, 1961; Manabe and Möller, 1961; Weart, 2008; Edwards, 2010). Revelle and 50 Keeling established carbon dioxide monitoring stations in Antarctica and Hawaii during the 1957–1958 51 International Geophysical Year (Revelle and Suess, 1957; Keeling, 1960). These stations have tracked rising 52 atmospheric CO2 concentrations from 315 ppm in 1958 to 414 ppm in 2020. Ground-based monitoring of 53 other greenhouse gases followed. The Greenhouse Gases Observing Satellite (GOSat) was launched in 2009, 54 and two Orbiting Carbon Observatory satellite instruments have been in orbit since 2014. 55 Do Not Cite, Quote or Distribute 1-37 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 WGI AR5 highlighted ‘the other CO2 problem’ (Doney et al., 2009), that is, ocean acidification caused by 2 the absorption of some 20–30% of anthropogenic carbon dioxide from the atmosphere and its conversion to 3 carbonic acid in seawater. WGI AR5 assessed that the pH of ocean surface water has decreased by 0.1 since 4 the beginning of the industrial era (high confidence), indicating approximately a 30% increase in acidity 5 (IPCC, 2013b). 6 7 With a heat capacity about 1000 times greater than that of the atmosphere, Earth’s ocean stores the vast 8 majority of energy retained by the planet. Ocean currents transport the stored heat around the globe and, over 9 decades to centuries, from the surface to its greatest depths. The ocean’s thermal inertia moderates faster 10 changes in radiative forcing on land and in the atmosphere, reaching full equilibrium with the atmosphere 11 only after hundreds to thousands of years (Yang and Zhu, 2011). The earliest subsurface measurements in 12 the open ocean date to the 1770s (Abraham et al., 2013). From 1872–76, the research ship HMS Challenger 13 measured global ocean temperature profiles at depths up to 1700 m along its cruise track. By 1900, research 14 ships were deploying instruments such as Nansen bottles and Mechanical BathyThermographs (MBTs) to 15 develop profiles of the upper 150 m in areas of interest to navies and commercial shipping (Abraham et al., 16 2013). Starting in 1967, eXpendable BathyThermographs (XBTs) were deployed by scientific and 17 commercial ships along repeated transects to measure temperature to 700 m (Goni et al., 2019). Ocean data 18 collection expanded in the 1980s with the Tropical Ocean Global Experiment (TOGA; Gould, 2003). Marine 19 surface observations for the globe, assembled in the mid-1980s in the International Comprehensive Ocean- 20 Atmosphere Data Set (ICOADS; Woodruff et al., 1987, 2005), were extended to 1662–2014 using newly 21 recovered marine records and metadata (Woodruff et al., 1998; Freeman et al., 2017). The Argo submersible 22 float network developed in the early 2000s provided the first systematic global measurements of the 700– 23 2000 m layer. Comparing the HMS Challenger data to data from Argo submersible floats revealed global 24 subsurface ocean warming on the centennial scale (Roemmich et al., 2012). WGI AR5 assessed with high 25 confidence that ocean warming accounted for more than 90% of the additional energy accumulated by the 26 climate system between 1971 and 2010 (IPCC, 2013b). In comparison, warming of the atmosphere 27 corresponds to only about 1% of the additional energy accumulated over that period (IPCC, 2013a). Chapter 28 2 summarises the ocean heat content datasets used in AR6 (Chapter 2, Section 2.3.3.1; Table 2.7). 29 30 Water expands as it warms. This thermal expansion, along with glacier mass loss, were the dominant 31 contributors to global mean sea level rise during the 20th century (high confidence) according to AR5 (IPCC, 32 2013b). Sea level can be measured by averaging across tide gauges, some of which date to the 18th century. 33 However, translating tide gauge readings into global mean sea levevl (GMSL) is challenging, since their 34 spatial distribution is limited to continental coasts and islands, and their readings are relative to local coastal 35 conditions that may shift vertically over time. Satellite radar altimetry, introduced operationally in the 1990s, 36 complements the tide gauge record with geocentric measurements of GMSL at much greater spatial coverage 37 (Katsaros and Brown, 1991; Fu et al., 1994). WGI AR5 assessed that global mean sea level rose by 0.19 38 [0.17 to 0.21] m over the period 1901–2010, and that the rate of sea level rise increased from 2.0 [1.7 to 2.3] 39 mm yr–1 in 1971–2010 to 3.2 [2.8 to 3.6] mm yr–1 from 1993–2010. Warming of the ocean very likely 40 contributed 0.8 [0.5 to 1.1] mm yr–1 of sea level change during 1971–2010, with the majority of that 41 contribution coming from the upper 700 m (IPCC, 2013b). Chapter 2, Section 2.3.3.3 assesses current 42 understanding of the extent and rate of sea level rise, past and present. 43 44 Satellite remote sensing also revolutionised studies of the cryosphere (Chapter 2, Section 2.3.2 and Chapter 45 9, Sections 9.3 to 9.5), particularly near the poles where conditions make surface observations very difficult. 46 Satellite mapping and measurement of snow cover began in 1966, with land and sea ice observations 47 following in the mid-1970s. Yet prior to the Third Assessment Repor, researchers lacked sufficient data to 48 tell whether the Greenland and Antarctic Ice Sheets were shrinking or growing. Through a combination of 49 satellite and airborne altimetry and gravity measurements, and improved knowledge of surface mass balance 50 and perimeter fluxes, a consistent signal of ice loss for both ice sheets was established by the time of AR5 51 (Shepherd et al., 2012). After 2000, satellite radar interferometry revealed rapid changes in surface velocity 52 at ice-sheet margins, often linked to reduction or loss of ice shelves (Scambos et al., 2004; Rignot and 53 Kanagaratnam, 2006). Whereas sea ice area and concentration were continuously monitored since 1979 from 54 microwave imagery, datasets for ice thickness emerged later from upward sonar profiling by submarines 55 (Rothrock et al., 1999) and radar altimetry of sea-ice freeboards (Laxon et al., 2003). A recent reconstruction Do Not Cite, Quote or Distribute 1-38 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 of Arctic sea ice extent back to 1850 found no historical precedent for the Arctic sea ice minima of the 21st 2 century (Walsh et al., 2017). Glacier length has been monitored for decades to centuries; internationally 3 coordinated activities now compile worldwide glacier length and mass balance observations (World Glacier 4 Monitoring Service, Zemp et al., 2015), global glacier outlines (Randolph Glacier Inventory, Pfeffer et al., 5 2014), and ice thickness of about 1100 glaciers (GlaThiDa, Gärtner-Roer et al., 2014). In summary, these 6 data allowed WGI AR5 to assess that over the last two decades, the Greenland and Antarctic Ice Sheets have 7 been losing mass, glaciers have continued to shrink almost worldwide, and Arctic sea ice and Northern 8 Hemisphere spring snow cover have continued to decrease in extent (high confidence) (IPCC, 2013b). 9 10 11 [START FIGURE 1.7 HERE] 12 13 Figure 1.7: Schematic of temporal coverage of selected instrumental climate observations (top) and selected 14 paleoclimate archives (bottom). The satellite era began in 1979 CE (Common Era). The width of the taper 15 gives an indication of the amount of available records. 16 17 18 [END FIGURE 1.7 HERE] 19 20 21 1.3.2 Lines of evidence: paleoclimate 22 23 With the gradual acceptance of geological ‘deep time’ in the 19th century came investigation of fossils, 24 geological strata, and other evidence pointing to large shifts in the Earth’s climate, from ice ages to much 25 warmer periods, across thousands to billions of years. This awareness set off a search for the causes of 26 climatic changes. The long-term perspective provided by paleoclimate studies is essential to understanding 27 the causes and consequences of natural variations in climate, as well as crucial context for recent 28 anthropogenic climatic change. The reconstruction of climate variability and change over recent millennia 29 began in the 1800s (Brückner et al., 2000; Brückner, 2018 [1890]; Coen, 2018, 2020). In brief, 30 paleoclimatology reveals the key role of carbon dioxide and other greenhouse gases in past climatic 31 variability and change, the magnitude of recent climate change in comparison to past glacial-interglacial 32 cycles, and the unusualness recent climate change (Section 1.2.1.2; Cross Chapter Box 2.1 in Chapter 2; 33 Tierney et al., 2020). FAQ 1.3 provides a plain-language summary of its importance. 34 35 Paleoclimate studies reconstruct the evolution of Earth’s climate over hundreds to billions of years using pre- 36 instrumental historical archives, indigenous knowledge and natural archives left behind by geological, 37 chemical, and biological processes (Figure 1.7). Paleoclimatology covers a wide range of temporal scales, 38 ranging from the human historical past (decades to millennia) to geological deep time (millions to billions of 39 years). Paleoclimate reference periods are presented in Cross Chapter Box 2.1 in Chapter 2. 40 41 Historical climatology aids near-term paleoclimate reconstructions using media such as diaries, almanacs, 42 and merchant accounts that describe climate-related events such as frosts, thaws, flowering dates, harvests, 43 crop prices, and droughts (Lamb, 1965, 1995; Le Roy Ladurie, 1967; Brázdil et al., 2005). Meticulous 44 records by Chinese scholars and government workers, for example, have permitted detailed reconstructions 45 of China’s climate back to 1000 CE, and even beyond (Louie and Liu, 2003; Ge et al., 2008). Climatic 46 phenomena such as large-scale, regionally and temporally distributed warmer and cooler periods of the past 47 2000 years were originally reconstructed from European historical records (Lamb, 1965, 1995; Le Roy 48 Ladurie, 1967; Neukom et al., 2019). 49 50 Indigenous and local knowledge have played an increasing role in historical climatology, especially in areas 51 where instrumental observations are sparse. Peruvian fishermen named the periodic El Niño warm current in 52 the Pacific, linked by later researchers to the Southern Oscillation (Cushman, 2004). Inuit communities have 53 contributed to climatic history and community based monitoring across the Arctic (Riedlinger and Berkes, 54 2001; Gearheard et al., 2010). Indigenous Australian knowledge of climatic patterns has been offered as a 55 complement to sparse observational records (Green et al., 2010; Head et al., 2014), such as those of sea-level Do Not Cite, Quote or Distribute 1-39 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 rise (Nunn and Reid, 2016). Ongoing research seeks to conduct further dialogue, utilise Indigenous and local 2 knowledge as an independent line of evidence complementing scientific understanding, and analyse their 3 utility for multiple purposes, especially adaptation (Laidler, 2006; Alexander et al., 2011; IPCC, 2019c). 4 Indigenous and local knowledge are used most extensively by IPCC Working Group II. 5 6 Certain geological and biological materials preserve evidence of past climate changes. These ‘natural 7 archives’ include corals, trees, glacier ice, speleothems (stalactites and stalagmites), loess deposits (dust 8 sediments), fossil pollen, peat, lake sediment, and marine sediment (Stuiver, 1965; Eddy, 1976; Haug et al., 9 2001; Wang et al., 2001; Jones et al., 2009; Bradley, 2015). By the early 20th century, laboratory research 10 had begun using tree rings to reconstruct precipitation and the possible influence of sunspots on climatic 11 change (Douglass, 1914, 1919, 1922). Radiocarbon dating, developed in the 1940s (Arnold and Libby, 12 1949), allows accurate determination of the age of carbon-containing materials from the past 50,000 years; 13 this dating technique ushered in an era of rapid progress in paleoclimate studies. 14 15 On longer timescales, tiny air bubbles trapped in polar ice sheets provide direct evidence of past atmospheric 16 composition, including CO2 levels (Petit et al., 1999), and the 18O isotope in frozen precipitation serves as a 17 proxy marker for temperature (Dansgaard, 1954). Sulphate deposits in glacier ice and as ash layers within 18 sediment record major volcanic eruptions, providing another mechanism for dating. The first paleoclimate 19 reconstructions used an almost 100,000-year ice core taken at Camp Century, Greenland (Dansgaard et al., 20 1969; Langway Jr, 2008). Subsequent cores from Antarctica extended this climatic record to 800,000 years 21 (EPICA Community Members, 2004; Jouzel, 2013). Comparisons of air contained in these ice samples 22 against measurements from the recent past enabled WGI AR5 to assess that atmospheric concentrations of 23 CO2, methane (CH4), and nitrous oxide (N2O) had all increased to levels unprecedented in at least the last 24 800,000 years (IPCC, 2013b) (see Section 1.2.1.2, Figure 1.5). 25 26 Global reconstructions of sea surface temperature were developed from material contained in deep-sea 27 sediment cores (CLIMAP Project Members et al., 1976), providing the first quantitative constraints for 28 model simulations of ice age climates (e.g., Rind and Peteet, 1985). Paleoclimate data and modelling showed 29 that the Atlantic Ocean circulation has not been stable over glacial-interglacial time periods, and that many 30 changes in ocean circulation are associated with abrupt transitions in climate in the North Atlantic region 31 (Ruddiman and McIntyre, 1981; Broecker et al., 1985; Boyle and Keigwin, 1987; Manabe and Stouffer, 32 1988). 33 34 By the early 20th century, cyclical changes in insolation due to the interacting periodicities of orbital 35 eccentricity, axial tilt, and axial precession had been hypothesised as a chief pacemaker of ice age- 36 interglacial cycles on multi-millennial timescales (Milankovich, 1920). Paleoclimate information derived 37 from marine sediment provides quantitative estimates of past temperature, ice volume, and sea level over 38 millions of years (Section 1.2.1.2, Figure 1.5) (Emiliani, 1955; Shackleton and Opdyke, 1973; Siddall et al., 39 2003; Lisiecki and Raymo, 2005; Past Interglacials Working Group of PAGES, 2016). These estimates have 40 bolstered the orbital cycles hypothesis (Hays et al., 1976; Berger, 1977, 1978). However, paleoclimatology 41 of multi-million to billion-year periods reveals that methane, carbon dioxide, continental drift, silicate rock 42 weathering, and other factors played a greater role than orbital cycles in climate changes during ice-free 43 ‘hothouse’ periods of Earth’s distant past (Frakes et al., 1992; Bowen et al., 2015; Zeebe et al., 2016). 44 45 The WGI AR5 (IPCC, 2013b) used paleoclimatic evidence to put recent warming and sea level rise in a 46 multi-century perspective and assessed that 1983–2012 was likely the warmest 30-year period of the last 47 1400 years in the Northern Hemisphere (medium confidence). AR5 also assessed that the rate of sea level 48 rise since the mid-19th century has been larger than the mean rate during the previous two millennia (high 49 confidence). 50 51 52 1.3.3 Lines of evidence: identifying natural and human drivers 53 54 The climate is a globally interconnected system driven by solar energy. Scientists in the 19th-century 55 established the main physical principles governing Earth’s temperature. By 1822, the principle of radiative Do Not Cite, Quote or Distribute 1-40 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 equilibrium (the balance between absorbed solar radiation and the energy Earth re-radiates into space) had 2 been articulated, and the atmosphere’s role in retaining heat had been likened to a greenhouse (Fourier, 3 1822). The primary explanations for natural climate change — greenhouse gases, orbital factors, solar 4 irradiance, continental position, volcanic outgassing, silicate rock weathering, and the formation of coal and 5 carbonate rock — were all identified by the late 1800s (Fleming, 1998; Weart, 2008). 6 7 The natural and anthropogenic factors responsible for climate change are known today as radiative ‘drivers’ 8 or ‘forcers’. The net change in the energy budget at the top of the atmosphere, resulting from a change in one 9 or more such drivers, is termed radiative forcing (RF; see Annex VII: Glossary) and measured in Watts per 10 square metre (W m-2). The total radiative forcing over a given time interval (often since 1750) represents the 11 sum of positive drivers (inducing warming) and negative ones (inducing cooling). Past IPCC reports have 12 assessed scientific knowledge of these drivers, quantified their range for the period since 1750, and presented 13 the current understanding of how they interact in the climate system. Like all previous IPCC reports, AR5 14 assessed that total radiative forcing has been positive at least since 1850–1900, leading to an uptake of 15 energy by the climate system, and that the largest single contribution to total radiative forcing is the rising 16 atmospheric concentration of CO2 since 1750 (IPCC, 2013a; see Cross-Chapter Box 1.2 and Chapter 7). 17 18 Natural drivers include changes in solar irradiance, ocean currents, naturally occurring aerosols, and natural 19 sources and sinks of radiatively active gases such as water vapour, carbon dioxide, methane, and sulphur 20 dioxide. Detailed global measurements of surface-level solar irradiance were first conducted during the 21 1957–1958 International Geophysical Year (Landsberg, 1961), while top-of-atmosphere irradiance has been 22 measured by satellites since 1959 (House et al., 1986). Measured changes in solar irradiance have been small 23 and slightly negative since about 1980 (Matthes et al., 2017). Water vapour is the most abundant radiatively 24 active gas, accounting for about 75% of the terrestrial greenhouse effect, but because its residence time in the 25 atmosphere averages just 8–10 days, its atmospheric concentration is largely governed by temperature (van 26 der Ent and Tuinenburg, 2017; Nieto and Gimeno, 2019). As a result, non-condensing greenhouse gases with 27 much longer residence times serve as ‘control knobs’, regulating planetary temperature, with water vapour 28 concentrations as a feedback effect (Lacis et al., 2010, 2013). The most important of these non-condensing 29 gases is carbon dioxide (a positive driver), released naturally by volcanism at about 637 MtCO2 yr-1 in recent 30 decades, or roughly 1.6% of the 37 GtCO2 emitted by human activities in 2018 (Burton et al., 2013; Le 31 Quéré et al., 2018). Absorption by the ocean and uptake by plants and soils are the primary natural CO2 sinks 32 on decadal to centennial time scales (see Chapter 5, Section 5.1.2 and Figure 5.3). 33 34 Aerosols (tiny airborne particles) interact with climate in numerous ways, some direct (e.g. reflecting solar 35 radiation back into space) and others indirect (e.g., cloud droplet nucleation); specific effects may cause 36 either positive or negative radiative forcing. Major volcanic eruptions inject sulphur dioxide (SO2, a negative 37 driver) into the stratosphere, creating aerosols that can cool the planet for years at a time by reflecting some 38 incoming solar radiation. The history and climatic effects of volcanic activity have been traced through 39 historical records, geological traces, and observations of major eruptions by aircraft, satellites, and other 40 instruments (Dörries, 2006). The negative RF of major volcanic eruptions was considered in the First 41 Assessment Report (FAR; IPCC, 1990a). In subsequent assessments, the negative RF of smaller eruptions 42 has also been considered (e.g., Chapter 2, section 2.4.3 in IPCC, 1995; Cross-Chapter Box 4.1 in Chapter 4 43 of this report). Dust and other natural aerosols have been studied since the 1880s (e.g., Aitken, 1889; 44 Ångström, 1929, 1964; Twomey, 1959), particularly in relation to their role in cloud nucleation, an aerosol 45 indirect effect whose RF may be either positive or negative depending on such factors as cloud altitude, 46 depth, and albedo (Stevens and Feingold, 2009; Boucher et al., 2013). 47 48 Anthropogenic (human) drivers of climatic change were hypothesised as early as the 17th century, with a 49 primary focus on forest clearing and agriculture (Grove, 1995; Fleming, 1998). In the 1890s, Arrhenius was 50 first to calculate the effects of increased or decreased CO2 concentrations on planetary temperature, and 51 Högbom estimated that worldwide coal combustion of about 500 Mt yr-1 had already completely offset the 52 natural absorption of CO2 by silicate rock weathering (Högbom, 1894; Arrhenius, 1896; Berner, 1995; 53 Crawford, 1997). As coal consumption reached 900 Mt yr-1 only a decade later, Arrhenius wrote that 54 anthropogenic carbon dioxide from fossil fuel combustion might eventually warm the planet (Arrhenius, 55 1908). In 1938, analysing records from 147 stations around the globe, Callendar calculated atmospheric Do Not Cite, Quote or Distribute 1-41 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 warming over land at 0.3-0.4°C from 1880-1935 and attributed about half of this warming to anthropogenic 2 CO2 (Callendar, 1938; Fleming, 2007; Hawkins and Jones, 2013; Figure 1.8). 3 4 5 [START FIGURE 1.8 HERE] 6 7 Figure 1.8: G.S. Callendar’s estimates of global land temperature variations and their possible causes. (a) The 8 original figure from Callendar (1938), using measurements from 147 surface stations for 1880–1935, 9 showing: (top) ten-year moving departures from the mean of 1901-1930 (°C), with the dashed line 10 representing his estimate of the ‘CO2 effect’ on temperature rise, and (bottom) annual departures from the 11 1901–1930 mean (°C). (b) Comparing the estimates of global land (60°S–60°N) temperatures tabulated 12 in Callendar (1938, 1961) with a modern reconstruction (Osborn et al., 2021) for the same period, after 13 (Hawkins and Jones (2013). Further details on data sources and processing are available in the chapter 14 data table (Table 1.SM.1). 15 16 [END FIGURE 1.8 HERE] 17 18 19 Studies of radiocarbon (14C) in the 1950s established that increasing atmospheric CO2 concentrations were 20 due to fossil fuel combustion. Since all the 14C once contained in fossil fuels long ago decayed into non- 21 radioactive 12C, the CO2 produced by their combustion reduces the overall concentration of atmospheric 14C 22 (Suess, 1955). Related work demonstrated that while the ocean was absorbing around 30% of anthropogenic 23 CO2, these emissions were also accumulating in the atmosphere and biosphere (see Section 1.3.1 and 24 Chapter 5, Section 5.2.1.5). Further work later established that atmospheric oxygen levels were decreasing in 25 inverse relation to the anthropogenic CO2 increase, because combustion of carbon consumes oxygen to 26 produce CO2 (Keeling and Shertz, 1992; IPCC, 2013a, Chapters 2 and 6). Revelle and Suess (1957) 27 famously described fossil fuel emissions as a ‘large scale geophysical experiment’, in which ‘within a few 28 centuries we are returning to the atmosphere and ocean the concentrated organic carbon stored in 29 sedimentary rocks over hundreds of millions of years’. The 1960s saw increasing attention to other 30 radiatively active gases, especially ozone (Manabe and Möller, 1961; Plass, 1961). Methane and nitrous 31 oxide were not considered systematically until the 1970s, when anthropogenic increases in those gases were 32 first noted (Wang et al., 1976). In the 1970s and 1980s, scientists established that synthetic halocarbons (see 33 Annex VII: Glossary), including widely used refrigerants and propellants, were extremely potent greenhouse 34 gases (Ramanathan, 1975; Chapter 2, Section 2.2.4.3; Chapter 6, section 6.2.2.9). When these chemicals 35 were also found to be depleting the stratospheric ozone layer, they were stringently and successfully 36 regulated on a global basis by the 1987 Montreal Protocol on the Ozone Layer and successor agreements 37 (Parson, 2003). 38 39 Radioactive fallout from atmospheric nuclear weapons testing (1940s–1950s) and urban smog (1950s– 40 1960s) first provoked widespread attention to anthropogenic aerosols and ozone in the troposphere 41 (Edwards, 2012). Theory, measurement, and modelling of these substances developed steadily from the 42 1950s (Hidy, 2019). However, the radiative effects of anthropogenic aerosols did not receive sustained study 43 until around 1970 (Bryson and Wendland, 1970; Rasool and Schneider, 1971), when their potential as 44 cooling agents was recognised (Peterson et al., 2008). The US Climatic Impact Assessment Program (CIAP) 45 found that proposed fleets of supersonic aircraft, flying in the stratosphere, might cause substantial aerosol 46 cooling and depletion of the ozone layer, stimulating efforts to understand and model stratospheric 47 circulation, atmospheric chemistry, and aerosol radiative effects (Mormino et al., 1975; Toon and Pollack, 48 1976). Since the 1980s, aerosols have increasingly been integrated into comprehensive modelling studies of 49 transient climate evolution and anthropogenic influences, through treatment of volcanic forcing, links to 50 global dimming and cloud brightening, and their influence on cloud nucleation and other properties (e.g., 51 thickness, lifetime, and extent) and precipitation (e.g., Hansen et al., 1981; Charlson et al., 1987, 1992; 52 Albrecht, 1989; Twomey, 1991). 53 54 The FAR (1990) focused attention on human emissions of carbon dioxide, methane, tropospheric ozone, 55 chlorofluorocarbons (CFCs), and nitrous oxide. Of these, at that time only the emissions of CO2 and CFCs 56 were well measured, with methane sources known only ‘semi-quantitatively’ (IPCC, 1990a). The FAR Do Not Cite, Quote or Distribute 1-42 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 assessed that some other trace gases, especially CFCs, have global warming potentials hundreds to thousands 2 of times greater than CO2 and methane, but are emitted in much smaller amounts. As a result, CO2 remains 3 by far the most important positive anthropogenic driver, with methane next most significant (Section 1.6.3); 4 anthropogenic methane stems from such sources as fossil fuel extraction, natural gas pipeline leakage, 5 agriculture, and landfills. In 2001, increased greenhouse forcing attributable to CO2, methane, ozone, CFC- 6 11, and CFC-12 was detected by comparing satellite measurements of outgoing longwave radiation 7 measurements taken in 1970 and in 1997 (Harries et al., 2001). AR5 assessed that the 40% increase in 8 atmospheric CO2 contributed most to positive RF since 1750. Together, changes in atmospheric 9 concentrations of CO2, methane, nitrous oxide, and halocarbons from 1750–2011 were assessed to contribute 10 a positive RF of 2.83 [2.26 to 3.40] W m–2 (IPCC, 2013b). 11 12 All IPCC reports have assessed the total RF as positive when considering all sources. However, due to the 13 considerable variability of both natural and anthropogenic aerosol loads, the FAR characterised total aerosol 14 RF as ‘highly uncertain’ and was unable even to determine its sign (positive or negative). Major advances in 15 quantification of aerosol loads and their effects have taken place since then, and IPCC reports since 1992 16 have consistently assessed total forcing by anthropogenic aerosols as negative (IPCC, 1992, 1995a, 1996). 17 However, due to their complexity and the difficulty of obtaining precise measurements, aerosol effects have 18 been consistently assessed as the largest single source of uncertainty in estimating total RF (Stevens and 19 Feingold, 2009; IPCC, 2013a). Overall, AR5 assessed that total aerosol effects, including cloud adjustments, 20 resulted in a negative RF of –0.9 [–1.9 to −0.1] W m−2 (medium confidence), offsetting a substantial portion 21 of the positive RF resulting from the increase in greenhouse gases (high confidence) (IPCC, 2013b). Chapter 22 7 provides an updated assessment of the total and per-component RF for the WGI contribution to AR6. 23 24 25 1.3.4 Lines of evidence: understanding and attributing climate change 26 27 Understanding the global climate system requires both theoretical understanding and empirical measurement 28 of the major forces and factors that govern the transport of energy and mass (air, water and water vapour) 29 around the globe; the chemical and physical properties of the atmosphere, ocean, cryosphere, and land 30 surfaces; and the biological and physical dynamics of natural ecosystems, as well as the numerous feedbacks 31 (both positive and negative) among these processes. Attributing climatic changes or extreme weather events 32 to human activity (see Cross Working Group Box: Attribution) requires, additionally, understanding of the 33 many ways that human activities may affect the climate, along with statistical and other techniques for 34 separating the ‘signal’ of anthropogenic climate change from the ‘noise’ of natural climate variability (see 35 Section 1.4.2). This inter- and trans-disciplinary effort requires contributions from many sciences. 36 37 Due to the complexity of many interacting processes ranging in scale from the molecular to the global, and 38 occurring on timescales from seconds to millennia, attribution makes extensive use of conceptual, 39 mathematical, and computer simulation models. Modelling allows scientists to combine a vast range of 40 theoretical and empirical understanding from physics, chemistry, and other natural sciences, producing 41 estimates of their joint consequences as simulations of past, present, or future states and trends (Nebeker, 42 1995; Edwards, 2010, 2011). 43 44 In addition to radiative transfer (discussed above in Section 1.3.3), forces and factors such as 45 thermodynamics (energy conversions), gravity, surface friction, and the Earth's rotation govern the 46 planetary-scale movements or ‘circulation’ of air and water in the climate system. The scientific theory of 47 climate began with Halley (1686), who hypothesized vertical atmospheric circulatory cells driven by solar 48 heating, and Hadley (1735), who showed how the Earth’s rotation affects that circulation. Ferrel (1856) 49 added the Coriolis force to existing theory, explaining the major structures of the global atmospheric 50 circulation. In aggregate, prevailing winds and ocean currents move energy poleward from the equatorial 51 regions where the majority of incoming solar radiation is received. 52 53 Climate models provide the ability to simulate these complex circulatory processes, and to improve the 54 physical theory of climate by testing different mathematical formulations of those processes. Since 55 controlled experiments at planetary scale are impossible, climate simulations provide one important way to Do Not Cite, Quote or Distribute 1-43 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 explore the differential effects and interactions of variables such as solar irradiance, aerosols, and greenhouse 2 gases. To assess their quality, models or compontents of models may be compared with observations. For 3 this reason, they can be used to attribute observed climatic effects to different natural and human drivers 4 (Hegerl et al., 2011). As early as Arrhenius (1896), simple mathematical models were used to calculate the 5 effects of doubling atmospheric carbon dioxide over pre-industrial concentrations (~550 ppm vs ~275 ppm). 6 In the early 1900s Bjerknes formulated the Navier-Stokes equations of fluid dynamics for motion of the 7 atmosphere (Bjerknes, 1906; Bjerknes et al., 1910), and Richardson (1922) developed a system for numerical 8 weather prediction based on these equations. When electronic computers became available in the late 1940s, 9 the methods of Bjerknes and Richardson were successfully applied to weather forecasting (Charney et al., 10 1950; Nebeker, 1995; Harper, 2008). 11 12 In the 1960s similar approaches to modelling the weather were used to model the climate, but with much 13 longer runs than daily forecasting (Smagorinsky et al., 1965; Manabe and Wetherald, 1967). Simpler 14 statistical and one- and two-dimensional modelling approaches continued in tandem with the more complex 15 General Circulation Models (GCMs) (Manabe and Wetherald, 1967; Budyko, 1969; Sellers, 1969). The first 16 coupled atmosphere-ocean model (AOGCM) with realistic topography appeared in 1975 (Bryan et al., 1975; 17 Manabe et al., 1975). Rapid increases in computer power enabled higher resolutions, longer model 18 simulations, and the inclusion of additional physical processes in GCMs, such as aerosols, atmospheric 19 chemistry, sea ice, and snow. 20 21 In the 1990s, AOGCMs were state of the art. By the 2010s, Earth system models (ESMs, also known as 22 coupled carbon-cycle climate models) incorporated land surface, vegetation, the carbon cycle, and other 23 elements of the climate system. Since the 1990s, some major modelling centres have deployed ‘unified’ 24 models for both weather prediction and climate modelling, with the goal of a seamless modelling approach 25 that uses the same dynamics, physics, and parameterisations at multiple scales of time and space (Cullen, 26 1993; Brown et al., 2012; NRC Committee on a National Strategy for Advancing Climate Modeling, 2012; 27 Brunet et al., 2015; Chapter 10, Section 10.1.2). Because weather forecast models make short-term 28 predictions that can be frequently verified, and improved models are introduced and tested iteratively on 29 cycles as short as 18 months, this approach allows major portions of the climate model to be evaluated as a 30 weather model and more frequently improved. However, all climate models exhibit biases of different 31 degrees and types, and the practice of ‘tuning’ parameter values in models to make their outputs match 32 variables such as historical warming trajectories has generated concern throughout their history (Randall and 33 Wielicki, 1997; Edwards, 2010; Hourdin et al., 2017; see also 1.5.3.2). Overall, the WGI AR5 assessed that 34 climate models had improved since previous reports (IPCC, 2013b) . 35 36 Since climate models vary along many dimensions, such as grid type, resolution, and parameterizations, 37 comparing their results requires special techniques. To address this problem, the climate modelling 38 community developed increasingly sophisticated Model Intercomparison Projects (MIPs) (Gates et al., 1999; 39 Covey et al., 2003). MIPs prescribe standardised experiment designs, time periods, output variables, or 40 observational reference data, to facilitate direct comparison of model results. This aids in diagnosing the 41 reasons for biases and other differences among models, and furthers process understanding (Section 1.5). 42 Both the CMIP3 and CMIP5 model intercomparison projects included experiments testing the ability of 43 models to reproduce 20th century global surface temperature trends both with and without anthropogenic 44 forcings. Although some individual model runs failed to achieve this (Hourdin et al., 2017), the mean trends 45 of multi-model ensembles did so successfully (Meehl et al., 2007a; Taylor et al., 2012). When only natural 46 forcings were included (creating the equivalent of a ‘control Earth’ without human influences), similar multi- 47 model ensembles could not reproduce the observed post-1970 warming at either global or regional scales 48 (Edwards, 2010; Jones et al., 2013). The GCMs and ESMs compared in CMIP6 (used in this report) offer 49 more explicit documentation and evaluation of tuning procedures (Schmidt et al., 2017; Burrows et al., 2018; 50 Mauritsen and Roeckner, 2020); see Section 1.5). 51 52 The FAR (IPCC, 1990a) concluded that while both theory and models suggested that anthropogenic 53 warming was already well underway, its signal could not yet be detected in observational data against the 54 ‘noise’ of natural variability (also see Barnett and Schlesinger (1987) and Section 1.4.2). Since then, 55 increased warming and progressively more conclusive attribution studies have identified human activities as Do Not Cite, Quote or Distribute 1-44 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 the ‘dominant cause of the observed warming since the mid-20th century’ (IPCC, 2013b). ‘Fingerprint’ 2 studies seek to detect specific observed changes – expected from theoretical understanding and model results 3 – that could not be explained by natural drivers alone, and to attribute statistically the proportion of such 4 changes that is due to human influence. These include global-scale surface warming, nights warming faster 5 than days, tropospheric warming and stratospheric cooling, a rising tropopause, increasing ocean heat 6 content, changed global patterns of precipitation and sea-level air pressure, increasing downward longwave 7 radiation, and decreasing upward longwave radiation (Hasselmann, 1979; Schneider, 1994; Karoly et al., 8 1994; Santer et al., 1995, 2013, Hegerl et al., 1996, 1997; Gillett et al., 2003; Santer, 2003; Zhang et al., 9 2007; Stott et al., 2010; Davy et al., 2017; Mann et al., 2017). Cross Working Group Box 1.1 outlines 10 attribution methods and uses from across the AR6, now including event attribution (specifying the influence 11 of climate change on individual extreme events such as floods, or on the frequency of classes of events such 12 as tropical cyclones). Overall, the evidence for human influence has grown substantially over time and from 13 each IPCC report to the subsequent one. 14 15 A key indicator of climate understanding is whether theoretical climate system budgets or ‘inventories’, such 16 as the balance of incoming and outgoing energy at the surface and at the top of the atmosphere, can be 17 quantified and closed observationally. The global energy budget, for example, includes energy retained in 18 the atmosphere, upper ocean, deep ocean, ice, and land surface. Church et al. (2013) assessed in AR5 with 19 high confidence that independent estimates of effective radiative forcing (ERF), observed heat storage, and 20 surface warming combined to give an energy budget for the Earth that is consistent with the WGI AR5 21 assessed likely range of equilibrium climate sensitivity (ECS) [1.5°C to 4.5°C] to within estimated 22 uncertainties (IPCC, 2013a; on ECS, see Section 1.3.5 below). Similarly, over the period 1993 to 2010, when 23 observations of all sea level components were available, WGI AR5 assessed the observed global mean sea 24 level rise to be consistent with the sum of the observed contributions from ocean thermal expansion (due to 25 warming) combined with changes in glaciers, the Antarctic and Greenland Ice Sheets, and land water storage 26 (high confidence). Verification that the terms of these budgets balance over recent decades provides strong 27 evidence for our understanding of anthropogenic climate change (Cross-Chapter Box 9.1 in Chapter 9). 28 29 The Appendix to Chapter 1 (Appendix 1A) lists the key detection and attribution statements in the 30 Summaries for Policymakers of WGI reports since 1990. The evolution of these statements over time reflects 31 the improvement of scientific understanding and the corresponding decrease in uncertainties regarding 32 human influences. The SAR stated that ‘the balance of evidence suggests a discernible human influence on 33 global climate’ (IPCC, 1995b). Five years later, the TAR concluded that ‘there is new and stronger evidence 34 that most of the warming observed over the last 50 years is attributable to human activities’ (IPCC, 2001b). 35 AR4 further strengthened previous statements, concluding that ‘most of the observed increase in global 36 average temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic 37 greenhouse gas concentrations’ (IPCC, 2007b). AR5 assessed that a human contribution had been detected to 38 changes in warming of the atmosphere and ocean; changes in the global water cycle; reductions in snow and 39 ice; global mean sea level rise; and changes in some climate extremes. AR5 concluded that ‘it is extremely 40 likely that human influence has been the dominant cause of the observed warming since the mid-20th 41 century’ (IPCC, 2013b). 42 43 44 1.3.5 Projections of future climate change 45 46 It was recognised in IPCC AR5 that information about the near term was increasingly relevant for adaptation 47 decisions. In response, WGI AR5 made a specific assessment for how global surface temperature was 48 projected to evolve over the next two decades, concluding that the change for the period 2016–2035 relative 49 to 1986–2005 will likely be in the range of 0.3°C to 0.7°C (medium confidence), assuming no major volcanic 50 eruptions or secular changes in total solar irradiance (IPCC, 2013b). AR5 was also the first IPCC assessment 51 report to assess ‘decadal predictions’ of the climate, where the observed state of the climate system was used 52 to start forecasts for a few years ahead. AR6 examines updates to these decadal predictions (Chapter 4, 53 Section 4.4.1). 54 55 The assessments and predictions for the near-term evolution of global climate features are largely Do Not Cite, Quote or Distribute 1-45 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 independent of future carbon emissions pathways. However, WGI AR5 assessed that limiting climate change 2 in the long-term future will require substantial and sustained reductions of greenhouse gas emissions (IPCC, 3 2013b). This assessment results from decades of research on understanding the climate system and its 4 perturbations, and projecting climate change into the future. Each IPCC report has considered a range of 5 emission scenarios, typically including a scenario in which societies choose to continue on their present 6 course as well as several others reflecting socioeconomic and policy responses that may limit emissions 7 and/or increase the rate of carbon dioxide removal from the atmosphere. Climate models are used to project 8 the outcomes of each scenario. However, future human climate influence cannot be precisely predicted 9 because greenhouse gas and aerosol emissions, land use, energy use, and other human activities may change 10 in numerous ways. Common emission scenarios used in the WGI contribution to AR6 are detailed in Section 11 1.6. 12 13 Based on model results and steadily increasing CO2 concentrations (Bolin and Bischof, 1970; SMIC, 1971; 14 Meadows et al., 1972), concerns about future ‘risk of effects on climate’ were addressed in Recommendation 15 70 of the Stockholm Action Plan, resulting from the 1972 United Nations Conference on the Human 16 Environment. Numerous other scientific studies soon amplified these concerns (summarised in Schneider 17 (1975), and Williams (1978); see also Nordhaus (1975, 1977). In 1979, a US National Research Council 18 (NRC) group led by Jule Charney reported on the ‘best present understanding of the carbon dioxide/climate 19 issue for the benefit of policymakers’, initiating an era of regular and repeated large-scale assessments of 20 climate science findings. 21 22 The 1979 Charney NRC report estimated equilibrium climate sensitivity (ECS) at 3°C, stating the range as 23 2°C–4.5°C, based on ‘consistent and mutually supporting’ model results and expert judgment (NRC, 1979). 24 ECS is defined in IPCC assessments as the global surface air temperature (GSAT) response to CO2 doubling 25 (from pre-industrial levels) after the climate has reached equilibrium (stable energy balance between the 26 atmosphere and ocean). Another quantity, transient climate response (TCR), was later introduced as the 27 global surface air temperature change, averaged over a 20-year period, at the time of CO2 doubling in a 28 scenario of concentration increasing at 1% per year). Calculating ECS from historical or paleoclimate 29 temperature records in combination with energy budget models has produced estimates both lower and 30 higher than those calculated using GCMs and ESMs; in AR6, these are assessed in Chapter 7, Section 7.5.2. 31 32 ECS is typically characterised as most relevant on centennial timescales, while TCR was long seen as a more 33 appropriate measure of the 50-100 year response to gradually increasing CO2; however, recent studies have 34 raised new questions about how accurately both quantities are estimated by GCMs and ESMs (Grose et al., 35 2018; Meehl et al., 2020; Sherwood et al., 2020). Further, as climate models evolved to include a full-depth 36 ocean, the time scale for reaching full equilibrium became longer and new methods to estimate ECS had to 37 be developed (Gregory et al., 2004; Meehl et al., 2020; Meinshausen et al., 2020). Because of these 38 considerations as well as new estimates from observation-based, paleoclimate, and emergent-constraints 39 studies (Sherwood et al., 2020), the AR6 definition of ECS has changed from previous reports; it now 40 includes all feedbacks except those associated with ice sheets. Accordingly, unlike previous reports, the AR6 41 assessments of ECS and TCR are not based primarily on GCM and ESM model results (see Chapter 7, Box. 42 7.1 and Section 7.5.5 for a full discussion). 43 44 Today, other sensitivity terms are sometimes used, such as transient climate response to emissions (TCRE, 45 defined as the ratio of warming to cumulative CO2 emissions in a CO2-only simulation) and Earth system 46 sensitivity (ESS), which includes multi-century Earth system feedbacks such as changes in ice sheets. Table 47 1.2 shows estimates of ECS and TCR for major climate science assessments since 1979. The table shows 48 that despite some variation in the range of GCM and (for the later assessments) ESM results, expert 49 assessment of ECS changed little between 1979 and the present report. Based on multiple lines of evidence, 50 AR6 has narrowed the likely range of ECS to 2.5-4.0 °C (Chapter 7, Section 7.5.5). 51 52 53 [START TABLE 1.2 HERE] 54 Do Not Cite, Quote or Distribute 1-46 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Table 1.2: Estimates of equilibrium climate sensitivity (ECS) and transient climate response (TCR) from successive 2 major scientific assessments since 1979. No likelihood statements are available for reports prior to 2001 because those 3 reports did not use the IPCC calibrated uncertainty language. The assessed range of ECS differs from the range derived 4 from General Circulation Model (GCM) and Earth System Model (ESM) results because assessments take into account 5 other evidence, other types of models, and expert judgment. The AR6 definition of ECS differs from previous reports, 6 now including all long-term feedbacks except those associated with ice sheets. AR6 estimates of ECS are derived 7 primarily from process understanding, historical observations, and emergent constraints, informed by (but not based on) 8 GCM and ESM model results. CMIP6 is the 6th phase of the Coupled Model Intercomparison Project. See Chapter 7, 9 Box 7.1 and Section 7.5.5. 10 Assessment ECS range Assessed Assessed Assessed derived from range of central range of GCM and ECS (°C) estimate of TCR (°C) ESM results ECS (°C) (°C) NAS 1979 (NRC, 1979) 2.0–3.5 1.5–4.5 3.0 NAS 1983 (National Research Council 2.0–3.5 1.5–4.5 3.0 and Carbon Dioxide Assessment Committee, 1983) Villach 1985 (WMO/UNEP/ICSU, 1986) 1.5–5.5 1.5–4.5 3.0 IPCC FAR 1990 (IPCC, 1990a) 1.9–5.2 1.5–4.5 2.5 IPCC 1992 Supplementary Report (IPCC, 1.7–5.4 1.5–4.5 2.5 discussed 1992) but not assessed IPCC 1994 Radiative Forcing report not given 1.5–4.5 2.5 (IPCC, 1995a) IPCC SAR (IPCC, 1996) 1.9–5.2 1.5–4.5 2.5 discussed but not assessed IPCC TAR (IPCC, 2001a) 2.0–5.1 1.5–4.5 2.5 1.1–3.1 (likely) IPCC AR4 (IPCC, 2007a) 2.1–4.4 2.0–4.5 3.0 1.0–3.0 (likely) IPCC AR5 (IPCC, 2013a) 2.1–4.7 1.5–4.5 not given 1.0–2.5 (likely) World Climate Research Programme Models not 2.6–3.9 not given not given (Sherwood et al., 2020) used in (66% estimate uncertainty interval, likely) 2.3–4.7 (90% uncertainty interval, very likely) IPCC AR6 2021 1.8–5.6 2.5–4.0 3.0 1.4–2.2 (CMIP6). Not (likely) (likely) used directly in assessing 2.0-5.0 ECS range (Ch (very likely) 7). 11 12 [END TABLE 1.2 HERE] 13 14 15 WGI AR5 assessed that there is a close relationship of cumulative total emissions of CO2 and global mean Do Not Cite, Quote or Distribute 1-47 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 surface temperature response that is approximately linear (IPCC, 2013b). This finding implies that continued 2 emissions of carbon dioxide will cause further warming and changes in all components of the climate 3 system, independent of any specific scenario or pathway. Scenario-based climate projections using the 4 Representative Concentration Pathways (RCPs) assessed in WGI AR5 result in continued warming over the 5 21st century in all scenarios except a strong climate change mitigation scenario (RCP2.6). Similarly, under 6 all RCP scenarios, AR5 assessed that the rate of sea level rise over the 21st century will very likely exceed 7 that observed during 1971–2010 due to increased ocean warming and increased loss of mass from glaciers 8 and ice sheets. Further increases in atmospheric CO2 will also lead to further uptake of carbon by the ocean, 9 which will increase ocean acidification. By the mid-21st century the magnitudes of the projected changes are 10 substantially affected by the choice of scenario. The set of scenarios used in climate change projections 11 assessed as part of the AR6 are discussed in Section 1.6. 12 13 From the close link between cumulative emissions and warming it follows that any given level of global 14 warming is associated with a total budget of GHG emissions, especially CO2 as it is the largest long-lasting 15 contributor to radiative forcing (Allen et al., 2009; Collins et al., 2013; Rogelj et al., 2019). Higher emissions 16 in earlier decades imply lower emissions later on to stay within the Earth's carbon budget. Stabilising the 17 anthropogenic influence on global surface temperature thus requires that CO2 emissions and removals reach 18 net zero once the remaining carbon budget is exhausted (see Cross-Chapter Box 1.4). 19 20 Past, present and future emissions of CO2 therefore commit the world to substantial multi-century climate 21 change, and many aspects of climate change would persist for centuries even if emissions of CO2 were 22 stopped immediately (IPCC, 2013b). According to AR5, a large fraction of this change is essentially 23 irreversible on a multi-century to millennial time scale, barring large net removal (‘negative emissions’) of 24 CO2 from the atmosphere over a sustained period through as yet unavailable technological means (IPCC, 25 2013a, 2018; see Chapters 4 and 5). However, significant reductions of warming due to SLCFs could reduce 26 the level at which temperature stabilises once CO2 emissions reach net zero, and also reduce the long-term 27 global warming commitment by reducing radiative forcing from SLCFs (Chapter 5). 28 29 In summary, major lines of evidence – observations, paleoclimate, theoretical understanding, and natural and 30 human drivers — have been studied and developed for over 150 years. Methods for projecting climate 31 futures have matured since the 1950s and attribution studies since the 1980s. We conclude that 32 understanding of the principal features of the climate system is robust and well established. 33 34 35 1.3.6 How do previous climate projections compare with subsequent observations? 36 37 Many different sets of climate projections have been produced over the past several decades, so it is valuable 38 to assess how well those projections have compared against subsequent observations. Consisent findings 39 build confidence in the process of making projections for the future. For example, Stouffer and Manabe 40 (2017) compared projections made in the early 1990s with subsequent observations. They found that the 41 projected surface pattern of warming, and the vertical structure of temperature change in both the atmosphere 42 and ocean, were realistic. Rahmstorf et al. (2007, 2012) examined projections of global surface 43 temperatureand global mean sea level assessed by the TAR and AR4 and found that the global surface 44 temperature projections were in good agreement with the subsequent observations, but that sea level 45 projections were underestimates compared to subsequent observations. WGI AR5 also examined earlier 46 IPCC Assessment Reports to evaluate their projections of how global surface temperature and global mean 47 sea level would change (Cubasch et al., 2013) with similar conclusions. 48 49 Although these studies generally showed good agreement between the past projections and subsequent 50 observations, this type of analysis is complicated because the scenarios of future radiative forcing used in 51 earlier projections do not precisely match the actual radiative forcings that subsequently occurred. 52 Mismatches between the projections and subsequent observations could be due to incorrectly projected 53 radiative forcings (e.g., aerosol emissions, greenhouse gas concentrations or volcanic eruptions that were not 54 included), an incorrect modelled response to those forcings, or both. Alternatively, agreement between 55 projections and observations may be fortuitous due to a compensating balance of errors, for example, too low Do Not Cite, Quote or Distribute 1-48 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 climate sensitivity but too strong radiative forcings. 2 3 One approach to partially correct for mismatches between the forcings used in the projections and the 4 forcings that actually occurred is described by Hausfather et al. (2020). Model projections of global surface 5 temperatureand estimated radiative forcings were taken from several historical studies, along with the 6 baseline no-policy scenarios from the first four IPCC assessment reports. These model projections of 7 temperature and radiative forcing are then compared to (a) the observed change in temperature through time 8 over the projection period, and (b) the observed change in temperature relative to the observationally- 9 estimated radiative forcing over the projection period (Figure 1.9; data from Hausfather et al. (2020)). 10 11 Although this approach has limitations when the modelled forcings differ greatly from the forcings 12 subsequently experienced, they were generally able to project actual future global warming when the 13 mismatches between forecast and observed radiative forcings are accounted for. For example, the Scenario B 14 presented in Hansen et al. (1988) projected around 50% more warming than has been observed during the 15 1988–2017 period, but this is largely because it overestimated subsequent radiative forcings. Similarly, while 16 the FAR (IPCC, 1990a) projected a higher rate of global surface temperature warming than has been 17 observed, this is largely because it overestimated future greenhouse gas concentrations: the FAR’s projected 18 increase in total anthropogenic forcing between 1990 and 2017 was 1.6 W m-2, while the observational 19 estimate of actual forcing during that period is 1.1 W m-2 (Dessler and Forster, 2018). Under these actual 20 forcings, the change in temperature in the FAR aligns with observations (Hausfather et al., 2020). 21 22 23 [START FIGURE 1.9 HERE] 24 25 Figure 1.9: Assessing past projections of global temperature change. Projected temperature change post-publication 26 on a temperature vs time (1970–2020, top panel) and temperature vs radiative forcing (1970–2017, 27 bottom panel) basis for a selection of prominent climate model projections (taken from Hausfather et al., 28 2020). Model projections (using global surface air temperature, GSAT) are compared to temperature 29 observations (using global mean surface temperature, GMST) from HadCRUT5 (black) and 30 anthropogenic forcings (through 2017) from Dessler and Forster (2018), and have a baseline generated 31 from the first five years of the projection period. Projections shown are: Manabe (1970), Rasool and 32 Schneider (1971), Broecker (1975), Nordhaus (1977), Hansen et al. (1981, H81), Hansen et al. (1988, 33 H88), Manabe and Stouffer (1993), along with the Energy Balance Model (EBM) projections from the 34 FAR, SAR and TAR, and the multi-model mean projection using CMIP3 simulations of the Special 35 Reports on Emission Scenarios (SRES) A1B scenario from AR4. H81 and H88 show most excpected 36 scenarios 1 and B, respectively. See Hausfather et al. (2020) for more details of the projections. Further 37 details on data sources and processing are available in the chapter data table (Table 1.SM.1). 38 39 [END FIGURE 1.9 HERE] 40 41 42 In addition to global surface temperature, past regional projections can be evaluated. For example, the FAR 43 presented a series of temperature projections for 1990 to 2030 for several regions around the world. Regional 44 projections were given for a best global warming estimate of 1.8°C since 1850-1900 by 2030, and were 45 assigned low confidence. The FAR also suggested that regional temperature changes should be scaled by - 46 30% to +50% to account for the uncertainty in projected global warming. 47 48 The regional projections presented in the FAR are compared to the observed temperature change in the 49 period since 1990 (Figure 1.10), following Grose et al. (2017). Subsequent observed temperature change has 50 tracked within the FAR projected range for the best estimate of regional warming in the Sahel, South Asia 51 and Southern Europe. Temperature change has tracked at or below this range for the Central North America 52 and Australia, yet remains within the range reduced by 30% to generate the FAR’s lower global warming 53 estimate, consistent with the smaller observed estimate of radiative forcing compared to the FAR central 54 estimate. Note that the projections assessed in Chapter 4 of AR6 WGI suggest that global temperatures will 55 be around 1.2°C–1.8°C above 1850–1900 by 2030, also lower than the FAR central estimate. 56 Do Not Cite, Quote or Distribute 1-49 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Overall, there is medium confidence that past projections of global temperature are consistent with 2 subsequent observations, especially when accounting for the difference in radiative forcings used and those 3 which actually occurred (limited evidence, high agreement). FAR regional projections are broadly consistent 4 with subsequent observations, allowing for regional-scale climate variability and differences in projected and 5 actual forcings. There is medium confidence that the spatial warming pattern has been reliably projected in 6 past IPCC reports (limited evidence, high agreement). 7 8 9 [START FIGURE 1.10 HERE] 10 11 Figure 1.10: Range of projected temperature change for 1990–2030 for various regions defined in IPCC First 12 Assessment Report (FAR).The left panel shows the FAR projections (IPCC, 1990a) for Southern 13 Europe, with darker red bands representing the range of projected change given for the best estimate of 14 1.8°C global warming since pre-industrial to 2030, and the fainter red bands show the range scaled by – 15 30% to +50% for lower and higher estimates of global warming. Blue lines show the regionally averaged 16 observations from several global temperature gridded datasets, and blue dashed lines show the linear 17 trends in those datasets for 1990–2020 extrapolated to 2030. Observed datasets are: HadCRUT5, Cowtan 18 and Way, GISTEMP, Berkeley Earth and NOAA GlobalTemp. The inset map shows the definition of the 19 FAR regions used. The right panel shows projected temperature changes by 2030 for the various FAR 20 regions, compared to the extrapolated observational trends, following Grose et al. (2017). Further details 21 on data sources and processing are available in the chapter data table (Table 1.SM.1). 22 23 [END FIGURE 1.10 HERE] 24 25 26 [START BOX 1.2 HERE] 27 28 Box 1.2: Special Reports in the sixth IPCC assessment cycle: key findings 29 30 The Sixth Assessment Cycle started with three Special Reports. The Special Report on Global Warming of 31 1.5°C (SR1.5, (IPCC, 2018), invited by the Parties to the UNFCCC in the context of the Paris Agreement, 32 assessed current knowledge on the impacts of global warming of 1.5°C above pre-industrial levels and 33 related global greenhouse gas (GHG) emission pathways. The Special Report on Climate Change and Land 34 (SRCCL, IPCC, 2019a) addressed GHG fluxes in land-based ecosystems, land use and sustainable land 35 management in relation to climate change adaptation and mitigation, desertification, land degradation and 36 food security. The Special Report on the Ocean and Cryosphere in a Changing Climate (SROCC, IPCC, 37 2019b) assessed new literature on observed and projected changes of the ocean and the cryosphere, and their 38 associated impacts, risks, and responses. 39 40 The SR1.5 and SRCCL were produced through a collaboration between the three IPCC Working Groups, the 41 SROCC by only WGs I and II. Here we focus on key findings relevant to the physical science basis covered 42 by WGI. 43 44 1) Observations of climate change 45 46 The SR1.5 estimated with high confidence that human activities caused a global warming of approximately 47 1°C between the 1850-1900 and 2017. For the period 2006–2015, observed global mean surface temperature 48 (GMST 7) was 0.87±0.12°C higher than the average over the 1850–1900 period (very high confidence). 49 Anthropogenic global warming was estimated to be increasing at 0.2±0.1°C per decade (high confidence) 50 and likely matches the level of observed warming to within ±20%. The SRCCL found with high confidence 51 that over land, mean surface air temperature increased by 1.53±0.15°C from 1850–1900 to 2006–2015, or 52 nearly twice as much as the global average. This observed warming has already led to increases in the 7 Box 1.2 reproduces the temperature metrics as they appeared in the respective SPMs of the SRs. In AR6 long-term changes of GMST (Global Mean Surface Temperature) and GSAT (Global Surface Air Temperature) are considered to be equivalent, differing in uncertainty estimates only (see Cross-Chapter Box 2.3 in Chapter 2). Do Not Cite, Quote or Distribute 1-50 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 frequency and intensity of climate and weather extremes in many regions and seasons, including heat waves 2 in most land regions (high confidence), increased droughts in some regions (medium confidence), and 3 increases in the intensity of heavy precipitation events at the global scale (medium confidence). These 4 climate changes have contributed to desertification and land degradation in many regions (high confidence). 5 Increased urbanisation can enhance warming in cities and their surroundings (heat island effect), especially 6 during heat waves (high confidence), and intensify extreme rainfall (medium confidence). 7 8 With respect to the ocean, the SROCC assessed that it is virtually certain that the ocean has warmed 9 unabated since 1970 and has taken up more than 90% of the excess heat contributed by global warming. The 10 rate of ocean warming has likely more than doubled since 1993. Over the period 1982–2016, marine 11 heatwaves have very likely doubled in frequency and are increasing in intensity (very high confidence). In 12 addition, the surface ocean acidified further (virtually certain) and loss of oxygen occurred from the surface 13 to a depth of 1000 m (medium confidence). The report expressed medium confidence that the Atlantic 14 Meridional Overturning Circulation (AMOC) weakened in 2004–2017 relative to 1850–1900. 15 16 Concerning the cryosphere, the SROCC reported widespread continued shrinking of nearly all components. 17 Mass loss from the Antarctic Ice Sheet tripled over the period 2007–2016 relative to 1997–2006, while mass 18 loss doubled for the Greenland Ice Sheet (likely, medium confidence). The report concludes with very high 19 confidence that due to the combined increased loss from the ice sheets, global mean sea level (GMSL) rise 20 has accelerated (extremely likely). The rate of recent GMSL rise (3.6±0.5 mm yr-1 for 2006–2015) is about 21 2.5 times larger than for 1901–1990. The report also found that Arctic sea ice extent has very likely 22 decreased for all months of the year since 1979 and that September sea ice reductions of 12.8±2.3% per 23 decade are likely unprecedented for at least 1000 years. Feedbacks from the loss of summer sea ice and 24 spring snow cover on land have contributed to amplified warming in the Arctic (high confidence), where 25 surface air temperature likely increased by more than double the global average over the last two decades. By 26 contrast, Antarctic sea ice extent overall saw no statistically significant trend for the period 1979 to 2018 27 (high confidence). 28 29 The SROCC assessed that anthropogenic climate change has increased observed precipitation (medium 30 confidence), winds (low confidence), and extreme sea level events (high confidence) associated with some 31 tropical cyclones. It also found evidence for an increase in annual global proportion of Category 4 or 5 32 tropical cyclones in recent decades (low confidence). 33 34 2) Drivers of climate change 35 36 The SRCCL stated that the land is simultaneously a source and sink of CO2 due to both anthropogenic and 37 natural drivers. It estimates with medium confidence that Agriculture, Forestry and Other Land Use 38 (AFOLU) activities accounted for around 13% of CO2, 44% of methane, and 82% of nitrous oxide emissions 39 from human activities during 2007–2016, representing 23% (12.0±3.0 GtCO2 equivalent yr-1) of the total net 40 anthropogenic emissions of GHGs. The natural response of land to human-induced environmental change 41 such as increasing atmospheric CO2 concentration, nitrogen deposition, and climate change, caused a net 42 CO2 sink equivalent of around 29% of total CO2 emissions (medium confidence); however, the persistence of 43 the sink is uncertain due to climate change (high confidence). 44 45 The SRCCL also assessed how changes in land conditions affect global and regional climate. It found that 46 changes in land cover have led to both a net release of CO2, contributing to global warming, and an increase 47 in global land albedo, causing surface cooling. However, the report estimated that the resulting net effect on 48 globally averaged surface temperature was small over the historical period (medium confidence). 49 50 The SROCC found that the carbon content of Arctic and boreal permafrost is almost twice that of the 51 atmosphere (medium confidence), and assessed medium evidence with low agreement that thawing northern 52 permafrost regions are currently releasing additional net methane and CO2. 53 54 3) Projections of climate change 55 Do Not Cite, Quote or Distribute 1-51 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 The SR1.5 concluded that global warming is likely to reach 1.5°C between 2030 and 2052 if it continues to 2 increase at the current rate (high confidence). However, even though warming from anthropogenic emissions 3 will persist for centuries to millennia and will cause ongoing long-term changes, past emissions alone are 4 unlikely to raise global surface temperatur to 1.5°C above 1850-1900 levels. 5 6 The SR1.5 also found that reaching and sustaining net zero anthropogenic CO2 emissions and reducing net 7 non-CO2 radiative forcing would halt anthropogenic global warming on multi-decadal time scales (high 8 confidence). The maximum temperature reached is then determined by cumulative net global anthropogenic 9 CO2 emissions up to the time of net zero CO2 emissions (high confidence) and the level of non-CO2 radiative 10 forcing in the decades prior to the time that maximum temperatures are reached (medium confidence). 11 12 Furthermore, climate models project robust differences in regional climate characteristics between the 13 present day and a global warming of 1.5°C, and between 1.5°C and 2°C, including mean temperature in most 14 land and ocean regions and hot extremes in most inhabited regions (high confidence). There is medium 15 confidence in robust differences in heavy precipitation events in several regions and the probability of 16 droughts in some regions. 17 18 The SROCC projected that global-scale glacier mass loss, permafrost thaw, and decline in snow cover and 19 Arctic sea ice extent will continue in the near term (2031–2050) due to surface air temperature increases 20 (high confidence). The Greenland and Antarctic Ice Sheets are projected to lose mass at an increasing rate 21 throughout the 21st century and beyond (high confidence). Sea level rise will also continue at an increasing 22 rate. For the period 2081–2100 with respect to 1986–2005, the likely ranges of global mean sea level 23 (GMSL) rise are projected at 0.26–0.53 m for RCP2.6 and 0.51–0.92 m for RCP8.5. For the RCP8.5 24 scenario, projections of GMSL rise by 2100 are higher by 0.1 m than in AR5 due to a larger contribution 25 from the Antarctic Ice Sheet (medium confidence). Extreme sea level events that occurred once per hundred 26 years in the recent past are projected to occur at least once per year at many locations by 2050, especially in 27 tropical regions, under all RCP scenarios (high confidence). According to SR1.5, by 2100, GMSL rise would 28 be around 0.1 m lower with 1.5°C global warming compared to 2°C (medium confidence). If warming is held 29 to 1.5°, GMSLwill still continue to rise well beyond 2100, but at a slower rate and a lower magnitude. 30 However, instability and/or irreversible loss of the Greenland and Antarctic Ice Sheets, resulting in multi- 31 metre rise in sea level over hundreds to thousands of years, could be triggered at 1.5°C to 2°C of global 32 warming (medium confidence). According to the SROCC, sea level rise in an extended RCP2.6 scenario 33 would be limited to around 1 m in 2300 (low confidence) while multi-metre sea-level rise is projected under 34 RCP8.5 by then (medium confidence). 35 36 The SROCC projected that over the 21st century, the ocean will transition to unprecedented conditions with 37 increased temperatures (virtually certain), further acidification (virtually certain), and oxygen decline 38 (medium confidence). Marine heatwaves are projected to become more frequent (very high confidence) as are 39 extreme El Niño and La Niña events (medium confidence). The AMOC is projected to weaken during the 40 21st century (very likely), but a collapse is deemed very unlikely (albeit with medium confidence due to 41 known biases in the climate models used for the assessment). 42 43 4) Emission pathways to limit global warming 44 45 The SR1.5 focused on emission pathways and system transitions consistent with 1.5°C global warming over 46 the 21st century. Building upon the understanding from WGI AR5 of the quasi-linear relationship between 47 cumulative net anthropogenic CO2 emissions since 1850–1900 and maximum global mean temperature, the 48 report assessed the remaining carbon budgets compatible with the 1.5°C or 2°C warming goals of the Paris 49 Agreement. Starting from year 2018, the remaining carbon budget for a one-in-two chance of limiting global 50 warming to 1.5°C is about 580 GtCO2, and about 420 GtCO2 for a two-in-three chance (medium confidence). 51 At constant 2017 emissions, these budgets would be depleted by about the years 2032 and 2028, 52 respectively. Using GMST instead of GSAT gives estimates of 770 and 570 GtCO2, respectively (medium 53 confidence). Each budget is further reduced by approximately 100 GtCO2 over the course of this century 54 when permafrost and other less well represented Earth-system feedbacks are taken into account. 55 Do Not Cite, Quote or Distribute 1-52 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 It is concluded that all emission pathways with no or limited overshoot of 1.5°C imply that global net 2 anthropogenic CO2 emissions would need to decline by about 45% from 2010 levels by 2030, reaching net 3 zero around 2050, together with deep reductions in other anthropogenic emissions, such as methane and 4 black carbon. To limit global warming to below 2°C, CO2 emissions would have to decline by about 25% by 5 2030 and reach net zero around 2070. 6 7 [END BOX 1.2 HERE] 8 9 10 1.4 AR6 foundations and concepts 11 12 AR6 WGI builds on previous assessments using well established foundations and concepts. This section 13 highlights some of the cross-cutting methods applied in the climate change literature and topics discussed 14 repeatedly throughout this report. The choices related to baseline, or reference periods, are first highlighted 15 (Section 1.4.1), including a specific discussion on the pre-industrial baseline used in AR6 WGI (Cross- 16 Chapter Box 1.2). The relationships between long-term trends, climate variability and the concept of 17 emergence of changes (Section 1.4.2) and the sources of uncertainty in climate simulations (Section 1.4.3) 18 are discussed next. The topic of low-likelihood outcomes, storylines, abrupt changes and surprises follows 19 (Section 1.4.4), including a description of the AR6 WGI risk framing (Cross-Chapter Box 1.3). The Cross- 20 Working Group Box: Attribution describes attribution methods, including those for extreme events. Various 21 sets of geographical regions used in later Chapters are also defined and introduced (Section 1.4.5). 22 23 24 1.4.1 Baselines, reference periods and anomalies 25 26 Several ‘baselines’ or ‘reference periods’ are used consistently throughout AR6 WGI. Baseline refers to a 27 period against which differences are calculated whereas reference period is used more generally to indicate a 28 time period of interest, or a period over which some relevant statistics are calculated (see Annex VII: 29 Glossary). Variations in observed and simulated climate variables over time are often presented as 30 ‘anomalies’, i.e., the differences relative to a baseline, rather than using the absolute values. This is done for 31 several reasons. 32 33 First, anomalies are often used when combining data from multiple locations, because the absolute values 34 can vary over small spatial scales which are not densely observed or simulated, whereas anomalies are 35 representative for much larger scales (e.g., for temperature, Hansen and Lebedeff 1987). Since their baseline 36 value is zero by definition, anomalies are also less susceptible to biases arising from changes in the 37 observational network. Second, the seasonality in different climate indicators can be removed using 38 anomalies to more clearly distinguish variability from long-term trends. 39 40 Third, different datasets can have different absolute values for the same climate variable that should be 41 removed for effective comparisons of variations with time. This is often required when comparing climate 42 simulations with each other, or when comparing simulations with observations, as simulated climate 43 variables are also affected by model bias that can be removed when they are presented as anomalies. It can 44 also be required when comparing observational datasets or reanalyses (see Section 1.5.2) with each other, 45 due to systematic differences in the underlying measurement system (see Figure 1.11). Understanding the 46 reasons for any absolute difference is important, but whether the simulated absolute value matters when 47 projecting future change will depend on the variable of interest. For example, there is not a strong 48 relationship between climate sensitivity of a model (which is an indicator of the degree of future warming) 49 and the simulated absolute global surface temperature (Mauritsen et al. 2012; Hawkins and Sutton 2016). 50 51 For some variables, such as precipitation, anomalies are often expressed as percentages in order to more 52 easily compare changes in regions with very different climatological means. However, for situations where 53 there are important thresholds (e.g., phase transitions around 0°C) or for variables which can only take a 54 particular sign or be in a fixed range (e.g., sea ice extent or relative humidity), absolute values are normally 55 used. Do Not Cite, Quote or Distribute 1-53 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 The choice of a baseline period has important consequences for evaluating both observations and simulations 3 of the climate, for comparing observations with simulations, and for presenting climate projections. There is 4 usually no perfect choice of baseline as many factors have to be considered and compromises may be 5 required (Hawkins and Sutton 2016). It is important to evaluate the sensitivity of an analysis or assessment to 6 the choice of the baseline. 7 8 For example, the collocation of observations and reanalyses within the model ensemble spread depends on 9 the choice of the baseline, and uncertainty in future projections of climate is reduced if using a more recent 10 baseline, especially for the near-term (Figure 1.11). The length of an appropriate baseline or reference period 11 depends on the variable being considered, the rates of change of the variable and the purpose of the period, 12 but is usually 20 to 50 years long. The World Meteorological Organization (WMO) uses 30-year periods to 13 define ‘climate normals’, which indicate conditions expected to be experienced in a given location. 14 15 16 [START FIGURE 1.11 HERE] 17 18 Figure 1.11: Choice of baseline matters when comparing observations and model simulations. Global surface air 19 temperature (GSAT, grey) from a range of CMIP6 historical simulations (1850–2014, 25 models) and 20 SSP1-2.6 (2015–2100) using absolute values (top) and anomalies relative to two different baselines: 21 1850–1900 (middle) and 1995–2014 (bottom). An estimate of GSAT from a reanalysis (ERA-5, orange, 22 1979–2020) and an observation-based estimate of global mean surface air temperature (GMST) (Berkeley 23 Earth, black, 1850–2020) are shown, along with the mean GSAT for 1961–1990 estimated by Jones et al. 24 (1999), light blue shading, 14.0±0.5°C). Using the more recent baseline (bottom) allows the inclusion of 25 datasets which do not include the periods of older baselines. The middle and bottom panels have scales 26 which are the same size but offset. Further details on data sources and processing are available in the 27 chapter data table (Table 1.SM.1). 28 29 [END FIGURE 1.11 HERE] 30 31 32 For AR6 WGI, the period 1995–2014 is used as a baseline to calculate the changes in future climate using 33 model projections and also as a ‘modern’ or ‘recent past’ reference period when estimating past observed 34 warming. The equivalent period in AR5 was 1986–2005, and in SR1.5, SROCC and SRCCL it was 2006– 35 2015. The primary reason for the different choice in AR6 is that 2014 is the final year of the historical 36 CMIP6 simulations. These simulations subsequently assume different emission scenarios and so choosing 37 any later baseline end date would require selecting a particular emissions scenario. For certain assessments, 38 the most recent decade possible (e.g. 2010–2019 or 2011–2020, depending on the availability of 39 observations) is also used as a reference period (see Cross Chapter Box 2.3 in Chapter 2). 40 41 Figure 1.12 shows changes in observed global mean surface temperature (GMST) relative to 1850–1900 and 42 illustrates observed global warming levels for a range of reference periods that are either used in AR6 or 43 were used in previous IPCC Reports. This allows changes to be calculated between different periods and 44 compared to previous assessments. For example, AR5 assessed the change in GMST from the 1850–1900 45 baseline to 1986–2005 reference period as 0.61 (0.55–0.67) °C, whereas it is now assessed to be 0.69 (0.52– 46 0.82) °C using improved GMST datasets (also see Cross-Chapter Box 2.3 in Chapter 2). 47 48 The commonly used metric for global surface warming tends to be global mean surface temperature (GMST) 49 but, as shown in Figure 1.11, climate model simulations tend to use global surface air temperature (GSAT). 50 Although GMST and GSAT are closely related, the two measures are physically distinct. GMST is a 51 combination of land surface air temperatures (LSAT) and sea surface temperatures (SSTs), whereas GSAT is 52 surface air temperatures over land, ocean and ice. A key development in AR6 is the assessment that long- 53 term changes in GMST and GSAT differ by at most 10% in either direction, with low confidence in the sign 54 of any differences (see Cross Chapter Box 2.3 for details). 55 56 Three future reference periods are used in AR6 WGI for presenting projections: near-term (2021–2040), Do Not Cite, Quote or Distribute 1-54 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 mid-term (2041–2060) and long-term (2081–2100) (see Figure 1.11). In AR6, 20-year reference periods are 2 considered long enough to show future changes in many variables when averaging over ensemble members 3 of multiple models, and short enough to enable the time dependence of changes to be shown throughout the 4 21st century. Projections with alternative recent baselines (such as 1986–2005 or the current WMO climate 5 normal period of 1981–2010) and a wider range of future reference periods are presented in the Interactive 6 Atlas. Note that ‘long-term’ is also sometimes used to refer to durations of centuries to millennia when 7 examining past climate, as well as future climate change beyond the year 2100. Cross-Chapter Box 2.1 in 8 Chapter 2 discusses the paleo reference periods used in AR6. 9 10 11 [START FIGURE 1.12 HERE] 12 13 Figure 1.12: Global warming over the instrumental period. Observed global mean surface temperature (GMST) from 14 four datasets, relative to the average temperature of 1850–1900 in each dataset (see Cross-Chapter Box 15 2.3 and Section 2.3.1.1 for more details). The shaded grey band indicates the assessed likely range for the 16 period around 1750 (see Cross-Chapter Box 1.2). Different reference periods are indicated by the 17 coloured horizontal lines, and an estimate of total GMST change up to that period is given, enabling a 18 translation of the level of warming between different reference periods. The reference periods are all 19 chosen because they have been used in the AR6 or previous IPCC assessment reports. The value for the 20 1981–2010 reference period, used as a ‘climate normal’ period by the World Meteorological 21 Organization, is the same as the 1986–2005 reference period shown. Further details on data sources and 22 processing are available in the chapter data table (Table 1.SM.1). 23 24 [END FIGURE 1.12 HERE] 25 26 27 [START CROSS-CHAPTER BOX 1.2 HERE] 28 29 Cross-Chapter Box 1.2: Changes in global temperature between 1750 and 1850 30 31 Contributing Authors: Ed Hawkins (UK), Paul Edwards (USA), Piers Forster (UK), Darrell Kaufman 32 (USA), Jochem Marotzke (Germany), Malte Meinshausen (Australia/Germany), Maisa Rojas (Chile), Bjørn 33 H. Samset (Norway), Peter Thorne (Ireland/UK). 34 35 36 The Paris Agreement aims to limit global temperatures to specific thresholds ‘above pre-industrial levels’. In 37 AR6 WGI, as in previous IPCC reports, observations and projections of changes in global temperature are 38 generally expressed relative to 1850–1900 as an approximate pre-industrial state (SR1.5, IPCC, 2018). This 39 is a pragmatic choice based upon data availability considerations, though both anthropogenic and natural 40 changes to the climate occurred before 1850. The remaining carbon budgets, the chance of crossing global 41 temperature thresholds, and projections of extremes and sea level rise at a particular level of global warming 42 can all be sensitive to the chosen definition of the approximate pre-industrial baseline (Millar et al., 2017a; 43 Schurer et al., 2017; Pfleiderer et al., 2018; Rogelj et al., 2019; Tokarska et al., 2019). This Cross-Chapter 44 Box assesses the evidence on change in radiative forcing and global temperature from the period around 45 1750 to 1850–1900; variations in the climate before 1750 are discussed in Chapter 2. 46 47 Although there is some evidence for human influence on climate before 1750 (e.g., Ruddiman and Thomson, 48 2001; Koch et al., 2019), the magnitude of the effect is still disputed (e.g., Joos et al., 2004; Beck et al., 49 2018b; see Chapter 5, Section 5.1.2.3), and most studies analyse the human influence on climate over the 50 industrial period. Historically, the widespread use of coal-powered machinery started the Industrial 51 Revolution in Britain in the late 18th century (Ashton, 1997), but the global effects were small for several 52 decades. In line with this, previous IPCC assessment reports considered changes in radiative forcing relative 53 to 1750, and temperature changes were often reported relative to the ‘late 19th century’. AR5 and SR1.5 54 made the specific pragmatic choice to approximate pre-industrial global temperatures by the average of the 55 1850–1900 period, when permanent surface observing networks emerged that provide sufficiently accurate Do Not Cite, Quote or Distribute 1-55 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 and continuous measurements on a near-global scale (see Sections 1.3.1 and Chapter 2, Section 2.3.1.1), and 2 because the model simulations of the historical period used 1850 as their start date. For the same reasons, to 3 ensure continuity with previous assessments, and because of larger uncertainties and lower confidence in 4 climatic changes before 1850 than after, AR6 makes the same choice to approximate pre-industrial global 5 temperatures by the average of the 1850-1900 period. 6 7 Here we assess improvements in our understanding of climatic changes in the period 1750-1850. 8 Anthropogenic influences on climate between 1750 and 1900 were primarily increased anthropogenic GHG 9 and aerosol emissions, and changes in land use. Between 1750 and 1850 atmospheric CO2 levels increased 10 by from about 278 ppm to about 285 ppm (Chapter 2, Section 2.2.3, equivalent to around 3 years of current 11 rates of increase), corresponding to about 55 GtCO2 in the atmosphere. Estimates of emissions from fossil 12 fuel burning (about 4 GtCO2, Boden et al., 2017) cannot explain the pre-1850 increase, so CO2 emissions 13 from land use changes are implicated as the dominant source. The atmospheric concentration of other GHGs 14 also increased over the same period, and there was a cooling influence from other anthropogenic radiative 15 forcings (such as aerosols and land use changes), but with a larger uncertainty than for GHGs (e.g., Carslaw 16 et al., 2017; Owens et al., 2017; Hamilton et al., 2018; Chapter 2, Section 2.2.6; Chapter 7, Section 7.3.5.2; 17 Cross-Chapter Box 1.2, Figure 1). It is likely that there was a net anthropogenic forcing of 0.0–0.3 Wm-2 in 18 1850–1900 relative to 1750 (medium confidence). The net radiative forcing from changes in solar activity 19 and volcanic activity in 1850–1900, compared to the period around 1750, is estimated to be smaller than 20 ± 0.1 W m-2, but note there were several large volcanic eruptions between 1750 and 1850 (Cross-Chapter 21 Box 1.2, Figure 1). 22 23 Several studies since AR5 have estimated changes in global temperatures following industrialisation and 24 before 1850. Hawkins et al. (2017) used observations, radiative forcing estimates and model simulations to 25 estimate the warming from 1720–1800 until 1986–2005 and assessed a likely range of 0.55°C–0.80°C, 26 slightly broader than the equivalent range starting from 1850–1900 (0.6°C–0.7°C). From proxy evidence, 27 PAGES 2k Consortium (2019) found that GMST for 1850–1900 was 0.02°C [-0.22 to 0.16°C] warmer than 28 the 30-year period centred on 1750. Schurer et al. (2017) used climate model simulations of the last 29 millennium to estimate that the increase in GHG concentrations before 1850 caused an additional likely 30 range of 0.0–0.2°C global warming when considering multiple reference periods. Haustein et al. (2017) 31 implies an additional warming of around 0.05°C attributable to human activity from 1750 to 1850–1900, and 32 the AR6 emulator (Chapter 7, Section 7.3.5.3) estimates the likely range of this warming to be 0.04°C– 33 0.14°C. 34 35 Combining these different sources of evidence, we assess that from the period around 1750 to 1850–1900 36 there was a change in global temperature of around 0.1°C [-0.1 to +0.3°C](medium confidence), with an 37 anthropogenic component of a likely range of 0.0°C–0.2°C (medium confidence). 38 39 40 [START CROSS-CHAPTER BOX1.2, FIGURE 1 HERE] 41 42 Cross-Chapter Box 1.2, Figure 1: Changes in radiative forcing from 1750 to 2019. The radiative forcing estimates 43 from the AR6 emulator (see Cross-Chapter Box 7.1 in Chapter 7) are split into GHG, other anthropogenic (mainly 44 aerosols and land use) and natural forcings, with the average over the 1850–1900 baseline shown for each. Further 45 details on data sources and processing are available in the chapter data table (Table 1.SM.1). 46 47 [END FIGURE CROSS-CHAPTER 1.2, FIGURE 1 HERE] 48 49 50 [END CROSS-CHAPTER BOX 1.2 HERE] 51 52 53 1.4.2 Variability and emergence of the climate change signal 54 55 Climatic changes since the pre-industrial era are a combination of long-term anthropogenic changes and Do Not Cite, Quote or Distribute 1-56 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 natural variations on time scales from days to decades. The relative importance of these two factors depends 2 on the climate variable or region of interest. Natural variations consist of both natural radiatively forced 3 trends (e.g. due to volcanic eruptions or solar variations) and ‘internal’ fluctuations of the climate system 4 which occur even in the absence of any radiative forcings. The internal ‘modes of variability’, such as ENSO 5 and the NAO, are discussed further in Annex IV. 6 7 8 1.4.2.1 Climate variability can influence trends over short periods 9 10 Natural variations in both weather and longer timescale phenomena can temporarily obscure or intensify any 11 anthropogenic trends (e.g., Deser et al., 2012; Kay et al., 2015). These effects are more important on small 12 spatial and temporal scales but can also occur on the global scale as well (see Cross-Chapter Box 3.1in 13 Chapter 3). 14 15 Since AR5, many studies have examined the role of internal variability through the use of ‘large ensembles’. 16 Each such ensemble consists of many different simulations by a single climate model for the same time 17 period and using the same radiative forcings. These simulations differ only in their phasing of the internal 18 climate variations (also see Section 1.5.4.2). A set of illustrative examples using one such large ensemble 19 (Maher et al., 2019) demonstrates how variability can influence trends on decadal timescales (Figure 1.13). 20 The long-term anthropogenic trends in this set of climate indicators are clearly apparent when considering 21 the ensemble as a whole (grey shading), and all the individual ensemble members have very similar trends 22 for ocean heat content (OHC), which is a robust estimate of the total energy stored in the climate system 23 (e.g., Palmer and McNeall, 2014). However, the individual ensemble members can exhibit very different 24 decadal trends in global surface air temperature (GSAT), UK summer temperatures, and Arctic sea-ice 25 variations. More specifically, for a representative 11-year period, both positive and negative trends can be 26 found in all these surface indicators, even though the long-term trend is for increasing temperatures and 27 decreasing sea ice. Periods in which the long-term trend is substantially obscured or intensified for more than 28 20 years are also visible in these regional examples, highlighting that observations are expected to exhibit 29 short-term trends which are larger or smaller than the long-term trend or differ from the average projected 30 trend from climate models, especially on continental spatial scales or smaller (see Cross Chapter Box 3.1 in 31 Chapter 3). The actual observed trajectory can be considered as one realisation of many possible alternative 32 worlds which experienced different weather, as also demonstrated by the construction of ‘observation-based 33 large ensembles’ that are alternate possible realisations of historical observations, which retain the statistical 34 properties of observed regional weather (e.g., McKinnon and Deser, 2018). 35 36 37 [START FIGURE 1.13 HERE] 38 39 Figure 1.13: Simulated changes in various climate indicators under historical and RCP4.5 scenarios using the 40 MPI ESM Grand Ensemble. The grey shading shows the 5–95% range from the 100-member ensemble. 41 The coloured lines represent individual example ensemble members, with linear trends for the 2011–2021 42 period indicated by the thin dashed lines. Changes in Ocean Heat Content (OHC) over the top 2000m 43 represents the integrated signal of global warming (left). The top row shows surface air temperature- 44 related indicators (annual GSAT change and UK summer temperatures) and the bottom row shows Arctic 45 sea-ice related indicators (annual ice volume and September sea ice extent). For smaller regions and for 46 shorter time period averages the variability increases and simulated short-term trends can temporarily 47 obscure or intensify anthropogenic changes in climate. Data from Maher et al., (2019). Further details on 48 data sources and processing are available in the chapter data table (Table 1.SM.1). 49 50 [END FIGURE 1.13 HERE] 51 52 53 1.4.2.2 The emergence of the climate change signal 54 55 In the 1930s it was noted that temperatures were increasing at both local and global scales (Kincer, 1933; 56 Callendar, 1938; Figure 1.8). At the time it was unclear whether the observed changes were part of a longer- Do Not Cite, Quote or Distribute 1-57 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 term trend or a natural fluctuation; the ‘signal’ had not yet clearly emerged from the ‘noise’ of natural 2 variability. Numerous studies have since focused on the emergence of changes in temperature using 3 instrumental observations (e.g., Madden and Ramanathan, 1980; Wigley and Jones, 1981; Mahlstein et al., 4 2011, 2012; Lehner and Stocker, 2015; Lehner et al., 2017) and paleo-temperature data (e.g., Abram et al., 5 2016). 6 7 Since the IPCC Third’s Assessment report in 2001, the observed signal of climate change has been 8 unequivocally detected at the global scale (see Section 1.3), and this signal is increasingly emerging from the 9 noise of natural variability on smaller spatial scales and in a range of climate variables (see also FAQ1.2). In 10 this Report emergence of a climate change signal or trend refers to when a change in climate (the ‘signal’) 11 becomes larger than the amplitude of natural or internal variations (defining the ‘noise’). This concept is 12 often expressed as a ‘signal to-noise’ ratio (S/N) and emergence occurs at a defined threshold of this ratio 13 (e.g. S/N > 1 or 2). Emergence can be estimated using observations and/or model simulations and can refer 14 to changes relative to a historical or modern baseline (see Chapter 12, Section 12.5.2, Annex VII: Glossary). 15 The concept can also be expressed in terms of time (the ‘time of emergence’; Annex VII: Glossary) or in 16 terms of a global warming level (Kirchmeier‐ Young et al., 2019; see Chapter 11, Section 11.2.5) and is also 17 used to refer to a time when we can expect to see a response of mitigation activities that reduce emissions of 18 greenhouse gases or enhance their sinks (emergence with respect to mitigation, see Chapter 4, Section 19 4.6.3.1). Whenever possible, emergence should be discussed in the context of a clearly defined level of S/N 20 or other quantification, such as ‘the signal has emerged at the level of S/N > 2’, rather than as a simple 21 binary statement. For an extended discussion, see Chapter 10 (Section 10.4.3). 22 23 Related to the concept of emergence is the detection of change (see Chapter 3). Detection of change is 24 defined as the process of demonstrating that some aspect of the climate or a system affected by climate has 25 changed in some defined statistical sense, often using spatially aggregating methods that try to maximise 26 S/N, such as ‘fingerprints’ (e.g., Hegerl et al., 1996), without providing a reason for that change. An 27 identified change is detected in observations if its likelihood of occurrence by chance due to internal 28 variability alone is determined to be small, for example, <10% (Annex VII: Glossary). 29 30 An example of observed emergence in surface air temperatures is shown in Figure 1.14. Both the largest 31 changes in temperature and the largest amplitude of year-to-year variations are observed in the Arctic, with 32 lower latitudes showing less warming and smaller year-to-year variations. For the six example regions 33 shown (Figure 1.14), the emergence of changes in temperature is more apparent in northern South America, 34 East Asia and central Africa, than for northern North America or northern Europe. This pattern was predicted 35 by Hansen et al. (1988) and noted in subsequent observations by Mahlstein et al. (2011) (see Chapter 10, 36 Section 10.3.4.3, Chapter 12, Section 12.5.2). Overall, tropical regions show earlier emergence of 37 temperature changes than at higher latitudes (high confidence). 38 39 Since AR5, the emergence of projected future changes has also been extensively examined, in variables 40 including surface air temperature (Hawkins and Sutton, 2012; Kirtman et al., 2013; Tebaldi and 41 Friedlingstein, 2013), ocean temperatures and salinity (Banks and Wood, 2002), mean precipitation (Giorgi 42 and Bi, 2009; Maraun, 2013), drought (Orlowsky and Seneviratne, 2013), extremes (Diffenbaugh and 43 Scherer, 2011; Fischer et al., 2014; King et al., 2015; Schleussner and Fyson, 2020), and regional sea level 44 change (Lyu et al., 2014). The concept has also been applied to climate change impacts such as effects on 45 crop growing regions (Rojas et al., 2019). In AR6, the emergence of oceanic signals such as regional sea 46 level change and changes in water mass properties is assessed in Chapter 9 (Section 9.6.1.4), emergence of 47 future regional changes is assed in Chapter 10 (Section 10.4.3), the emergence of extremes as a function of 48 global warming levels is assessed in Chapter 11 (Section 11.2.5) and the emergence of climatic impact- 49 drivers for AR6 regions and many climate variables is assessed in Chapter 12 (Section 12.5.2). 50 51 Although the magnitude of any change is important, regions which have a larger signal of change relative to 52 the background variations will potentially face greater risks than other regions, as they will see unusual or 53 novel climate conditions more quickly (Frame et al., 2017). As in Figure 1.14, the signal of temperature 54 change is often smaller in tropical countries, but their lower amplitude of variability means they may 55 experience the effects of climate change earlier than the mid-latitudes. In addition, these tropical countries Do Not Cite, Quote or Distribute 1-58 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 are often amongst the most exposed, due to large populations (Lehner and Stocker, 2015), and often more 2 vulnerable (Harrington et al., 2016; Harrington and Otto, 2018; Russo et al., 2019); both of these factors 3 increase the risk from climate-related impacts (Cross Chapter Box 1.3). The rate of change is also important 4 for many hazards (e.g., Loarie et al., 2009). Providing more information about changes and variations on 5 regional scales, and the associated attribution to particular causes (see Cross-Working Group Box: 6 Attribution), is therefore important for adaptation planning. 7 8 9 [START FIGURE 1.14 HERE] 10 11 Figure 1.14: The observed emergence of changes in temperature. Top left: the total change in temperature 12 estimated for 2020 relative to 1850–1900 (following Hawkins et al. 2020), showing the largest warming 13 in the Arctic. Top right: the amplitude of estimated year-to-year variations in temperature. Middle left: 14 the ratio of the observed total change in temperature and the amplitude of temperature variability (the 15 ‘signal-to-noise (S/N) ratio’), showing that the warming is most apparent in the tropical regions (also see 16 FAQ1.2). Middle right: the global warming level at which the change in local temperature becomes larger 17 than the local year-to-year variability. The bottom panels show time series of observed annual mean 18 surface air temperatures over land in various example regions, as indicated as boxes in the top left panel. 19 The 1 and 2 standard deviations of estimated year-to-year variations for that region are shown by the pink 20 shaded bands. Observed temperature data from Berkeley Earth (Rohde and Hausfather, 2020). Further 21 details on data sources and processing are available in the chapter data table (Table 1.SM.1). 22 23 [END FIGURE 1.14 HERE] 24 25 26 1.4.3 Sources of uncertainty in climate simulations 27 28 When evaluating and analysing simulations of the physical climate system, several different sources of 29 uncertainty need to be considered (e.g., Hawkins and Sutton, 2009; Lehner et al., 2020). Broadly, these 30 sources are: uncertainties in radiative forcings (both those observed in the past and those projected for the 31 future); uncertainty in the climate response to particular radiative forcings; internal and natural variations of 32 the climate system (which may be somewhat predictable) and interactions among these sources of 33 uncertainty. 34 35 Ensembles of climate simulations (see Section 1.5.4.2), such as those produced as part of the sixth phase of 36 the Coupled Model Intercomparison Project (CMIP6), can be used to explore these different sources of 37 uncertainty and estimate their magnitude. Relevant experiments with climate models include both historical 38 simulations constrained by past radiative forcings and projections of future climate which are constrained by 39 specified drivers, such as GHG concentrations, emissions, or radiative forcings. (The term ‘prediction’ is 40 usually reserved for estimates of the future climate state which are also constrained by the observed initial 41 conditions of the climate system, analogous to a weather forecast.) 42 43 44 1.4.3.1 Sources of uncertainty 45 46 Radiative forcing uncertainty 47 Future radiative forcing is uncertain due to as-yet-unknown societal choices that will determine future 48 anthropogenic emissions; this is considered ‘scenario uncertainty’. The RCP and SSP scenarios, which form 49 the basis for climate projections assessed in this report, are designed to span a plausible range of future 50 pathways (see Section 1.6) and can be used to estimate the magnitude of scenario uncertainty, but the real 51 world may also differ from any one of these example pathways. 52 53 Uncertainties also exist regarding past emissions and radiative forcings. These are especially important for 54 simulations of paleoclimate time periods, such as the Pliocene, Last Glacial Maximum or the last 55 millennium, but are also relevant for the CMIP historical simulations of the instrumental period since 1850. 56 In particular, historical radiative forcings due to anthropogenic and natural aerosols are less well constrained Do Not Cite, Quote or Distribute 1-59 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 by observations than the greenhouse gas radiative forcings. There is also uncertainty in the size of large 2 volcanic eruptions (and in the location for some that occurred before around 1850), and the amplitude of 3 changes in solar activity, before satellite observations. The role of historical radiative forcing uncertainty was 4 considered previously (Knutti et al., 2002; Forster et al., 2013) but, since AR5, specific simulations have 5 been performed to examine this issue, particularly for the effects of uncertainty in anthropogenic aerosol 6 radiative forcing (e.g., Jiménez-de-la-Cuesta and Mauritsen, 2019; Dittus et al., 2020). 7 8 Climate response uncertainty 9 Under any particular scenario (see Section 1.6.1), there is uncertainty in how the climate will respond to the 10 specified emissions or radiative forcing combinations. A range of climate models is often used to estimate 11 the range of uncertainty in our understanding of the key physical processes and to define the ‘model response 12 uncertainty’ (see Section 1.5.4 and Chapter 4, Section 4.2.5). However, this range does not necessarily 13 represent the full ‘climate response uncertainty’ in how the climate may respond to a particular radiative 14 forcing or emissions scenario. This is because, for example, the climate models used in CMIP experiments 15 have structural uncertainties not explored in a typical multi-model exercise (e.g., Murphy et al., 2004) and 16 are not entirely independent of each other (Masson and Knutti, 2011; Abramowitz et al., 2019; see Section 17 1.5.4.8); there are small spatial-scale features which cannot be resolved; and long time-scale processes or 18 tipping points are not fully represented. Section 1.4.4 discusses how some of these issues can still be 19 considered in a risk assessment context. For some metrics, such as Equilibrium Climate Sensitivity (ECS), 20 the CMIP6 model range is found to be broader than the very likely range assessed by combining multiple 21 lines of evidence (see Chapter 4, Section 4.3.4 and Chapter 7, Section 7.5.6). 22 23 Natural and internal climate variations 24 Even without any anthropogenic radiative forcing, there would still be uncertainty in projecting future 25 climate because of unpredictable natural factors such as variations in solar activity and volcanic eruptions. 26 For projections of future climate, such as those presented in Chapter 4, the uncertainty in these factors is not 27 normally considered. However, the potential effects on the climate of large volcanic eruptions (Cross- 28 Chapter Box 4.1in Chapter 4, Zanchettin et al., 2016; Bethke et al., 2017) and large solar variations (Feulner 29 and Rahmstorf, 2010; Maycock et al., 2015) are studied. On longer timescales, orbital effects and plate 30 tectonics also play a role. 31 32 Further, even in the absence of any anthropogenic or natural changes in radiative forcing, Earth’s climate 33 fluctuates on timescales from days to decades or longer. These ‘internal’ variations, such as those associated 34 with modes of variability (e.g., ENSO, Pacific Decadal Variability (PDV), or Atlantic Multi-decadal 35 Variability (AMV) – see Annex IV) are unpredictable on timescales longer than a few years ahead and are a 36 source of uncertainty for understanding how the climate might become in a particular decade, especially 37 regionally. The increased use of ‘large ensembles’ of complex climate model simulations to sample this 38 component of uncertainty is discussed above in Section 1.4.2.1 and further in Chapter 4. 39 40 Interactions between variability and radiative forcings 41 It is plausible that there are interactions between radiative forcings and climate variations, such as influences 42 on the phasing or amplitude of internal or natural climate variability (Zanchettin, 2017). For example, the 43 timing of volcanic eruptions may influence Atlantic multi-decadal variability (e.g., Otterå et al., 2010; Birkel 44 et al., 2018) or ENSO (e.g., Maher et al., 2015; Khodri et al., 2017; Zuo et al., 2018), and anthropogenic 45 aerosols may influence decadal modes of variability in the Pacific (e.g., Smith et al., 2016). In addition, 46 melting of glaciers and ice caps due to anthropogenic influences has been speculated to increase volcanic 47 activity (e.g., a specific example for Iceland is discussed in Swindles et al., 2018). 48 49 50 1.4.3.2 Uncertainty quantification 51 52 Not all of these listed sources of uncertainty are of the same type. For example, internal climate variations 53 are an intrinsic uncertainty that can be estimated probabilistically, and could be more precisely quantified, 54 but cannot usually be reduced. However, advances in decadal prediction offer the prospect of narrowing 55 uncertainties in the trajectory of the climate for a few years ahead (e.g., Meehl et al., 2014; Yeager and Do Not Cite, Quote or Distribute 1-60 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Robson, 2017; Chapter 4, Section 4.2.3). 2 3 Other sources of uncertainty, such as model response uncertainty, can in principle be reduced, but are not 4 amenable to a frequency-based interpretation of probability, and Bayesian methods to quantify the 5 uncertainty have been considered instead (e.g., Tebaldi, 2004; Rougier, 2007; Sexton et al., 2012). The 6 scenario uncertainty component is distinct from other uncertainties, given that future anthropogenic 7 emissions can be considered as the outcome of a set of societal choices (see Section 1.6.1). 8 9 For climate model projections it is possible to approximately quantify the relative amplitude of various 10 sources of uncertainty (e.g., Hawkins and Sutton, 2009; Lehner et al., 2020). A range of different climate 11 models are used to estimate the model response uncertainty to a particular emissions pathway, and multiple 12 pathways are used to estimate the scenario uncertainty. The unforced component of internal variability can 13 be estimated from individual ensemble members of the same climate model (e.g., Deser et al., 2012; Maher 14 et al., 2019; Section 1.5.4.8). 15 16 Figure 1.15 illustrates the relative size of these different uncertainty components using a ‘cascade of 17 uncertainty’ (Wilby and Dessai, 2010), with examples shown for global mean temperature, northern South 18 American annual temperatures and East Asian summer precipitation changes. For global mean temperature, 19 the role of internal variability is small, and the total uncertainty is dominated by emissions scenario and 20 model response uncertainties. Note that there is considerable overlap between individual simulations for 21 different emissions scenarios even for the mid-term (2041–2060). For example, the slowest-warming 22 simulation for SSP5-8.5 produces less mid-term warming than the fastest-warming simulation for SSP1-1.9. 23 For the long-term, emissions scenario uncertainty becomes dominant. 24 25 The relative uncertainty due to internal variability and model uncertainty increases for smaller spatial scales. 26 In the regional example shown for changes in temperature, the same scenario and model combination has 27 produced two simulations which differ by 1°C in their projected 2081–2100 averages due solely to internal 28 climate variability. For regional precipitation changes, emissions scenario uncertainty is often small relative 29 to model response uncertainty. In the example shown, the SSPs overlap considerably, but SSP1-1.9 shows 30 the largest precipitation change in the near-term even though global mean temperature warms the least; this 31 is due to differences between regional aerosol emissions projected in this and other scenarios (Wilcox et al., 32 2020). These cascades of uncertainty would branch out further if applying the projections to derive estimates 33 of changes in hazard (e.g., Wilby and Dessai, 2010; Halsnæs and Kaspersen, 2018; Hattermann et al., 2018). 34 35 36 [START FIGURE 1.15 HERE] 37 38 Figure 1.15: The ‘cascade of uncertainties’ in CMIP6 projections. Changes in GSAT (left), northern South 39 America (region NSA) temperature change (middle), and East Asia (region EAS) summer (JJA) 40 precipitation change (right) are shown for two time periods (2041–2060, top, and 2081–2100, bottom). 41 The SSP-radiative forcing combination is indicated at the top of each cascade at the value of the multi- 42 model mean for each scenario. This branches downwards to show the ensemble mean for each model, and 43 further branches into the individual ensemble members, although often only a single member is available. 44 These diagrams highlight the relative importance of different sources of uncertainty in climate 45 projections, which varies for different time periods, regions and climate variables. See Section 1.4.5 for 46 the definition of the regions used. Further details on data sources and processing are available in the 47 chapter data table (Table 1.SM.1). 48 49 [END FIGURE 1.15 HERE] 50 51 52 1.4.4 Considering an uncertain future 53 54 Since AR5 there have been developments in how to consider and describe future climate outcomes which are 55 considered possible but very unlikely, highly uncertain, or potentially surprising. To examine such futures 56 there is a need to move beyond the usual ‘likely’ or ‘very likely’ assessed ranges and consider low-likelihood Do Not Cite, Quote or Distribute 1-61 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 outcomes, especially those that would result in significant impacts if they occurred (e.g., Sutton, 2018; 2 Sillmann et al., 2021). This section briefly outlines some of the different approaches used in the AR6 WGI. 3 4 5 1.4.4.1 Low-likelihood outcomes 6 7 In the AR6, certain low-likelihood outcomes are described and assessed because they may be associated with 8 high levels of risk and the greatest risks may not be associated with the most expected outcome. The aim of 9 assessing these possible futures is to better inform risk assessment and decision making. Two types are 10 considered: (1) low-likelihood high warming (LLHW) scenarios, which describe the climate in a world with 11 very high climate sensitivity, and (2) low-likelihood high impact (LLHI) events that have a low likelihood of 12 occurring, but would cause large potential impacts on society or ecosystems. 13 14 An illustrative example of how low-likelihood outcomes can produce significant additional risks is shown in 15 Figure 1.16. The Reasons for Concern (RFCs) produced by the IPCC AR5 WGII define the additional risks 16 due to climate change at different global warming levels. These have been combined with Chapter 4 17 assessments of projected global temperature for different emissions scenarios (SSPs; see Section 1.6), and 18 Chapter 7 assessments about ECS. For example, even following a medium emissions scenario could result in 19 high levels of additional risk if ECS is at the upper end of the very likely range. However, not all possible 20 low-likelihood outcomes relate to ECS, and AR6 considers these issues in more detail than previous IPCC 21 assessment reports (see Table 1.1 and below for some examples). 22 23 24 [START FIGURE 1.16 HERE] 25 26 Figure 1.16: Illustrating concepts of low-likelihood scenarios. Left: schematic likelihood distribution consistent with 27 the IPCC AR6 assessments that equilibrium climate sensitivity (ECS) is likely in the range 2.5 to 4.0°C, 28 and very likely between 2.0 and 5.0°C (Chapter 7). ECS values outside the assessed very likely range are 29 designated low-likelihood scenarios in this example (light grey). Middle and right columns: additional 30 risks due to climate change for 2020–2090 using the Reasons For Concern (RFCs, see IPCC, 2014), 31 specifically RFC1 describing the risks to unique and threatened systems and RFC3 describing risks from 32 the distribution of impacts (O’Neill et al., 2017b; Zommers et al., 2020). The projected changes of GSAT 33 used are the 95%, median and 5% assessed ranges from Chapter 4 for each SSP (top, middle and bottom); 34 these are designated High ECS, Mid-range ECS and Low ECS respectively. The burning-ember risk 35 spectrum is usually associated with levels of committed GSAT change; instead, this illustration associates 36 the risk spectrum with the GSAT reached in each year from 2020 to 2090. Note that this illustration does 37 not include the vulnerability aspect of each SSP scenario. Further details on data sources and processing 38 are available in the chapter data table (Table 1.SM.1). 39 40 [END FIGURE 1.16 HERE] 41 42 43 1.4.4.2 Storylines 44 45 As societies are increasingly experiencing the impacts of climate change related events, the climate science 46 community is developing climate information tailored for particular regions and sectors. There is a growing 47 focus on explaining and exploring complex physical chains of events or on predicting climate under various 48 future socio-economic developments. Since AR5, ‘storylines’ or ‘narratives’ approaches have been used to 49 better inform risk assessment and decision making, to assist understanding of regional processes, and 50 represent and communicate climate projection uncertainties more clearly. The aim is to help build a cohesive 51 overall picture of potential climate change pathways that moves beyond the presentation of data and figures 52 (Annex VII: Glossary; Fløttum and Gjerstad, 2017; Moezzi et al., 2017; Dessai et al., 2018; Shepherd et al., 53 2018b). 54 55 In the broader IPCC context, the term ‘scenario storyline’ refers to a narrative description of one or more 56 scenarios, highlighting their main characteristics, relationships between key driving forces and the dynamics Do Not Cite, Quote or Distribute 1-62 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 of their evolution (for example, short-lived climate forcers emissions assessed in Chapter 6 are driven by 2 ‘scenario storylines’, see Section 1.6). WGI is mainly concerned with ‘physical climate storylines’. These are 3 self-consistent and possible unfolding of a physical trajectory of the climate system, or a weather or climate 4 event, on timescales from hours to multiple decades (Shepherd et al., 2018b). This approach can be used to 5 constrain projected changes or specific events on specified explanatory elements such as projected changes 6 of large-scale indicators (Chapter 10, Box 10.2). For example, Hazeleger et al. (2015) suggested using ‘tales 7 of future weather’, blending numerical weather prediction with a climate projection to illustrate the potential 8 behaviour of future high-impact events (also see Hegdahl et al. 2020). Several studies describe how possible 9 large changes in atmospheric circulation would affect regional precipitation and other climate variables, and 10 discuss the various climate drivers which could cause such a circulation response (James et al., 2015; Zappa 11 and Shepherd, 2017; Mindlin et al., 2020). Physical climate storylines can also help frame the causal factors 12 of extreme weather events (Shepherd, 2016) and then be linked to event attribution (Chapter 11, Section 13 11.2.2; Cross Working Group Box: Attribution). 14 15 Storyline approaches can be used to communicate and contextualise climate change information in the 16 context of risk for policymakers and practitioners (e.g., de Bruijn et al., 2016; Dessai et al., 2018; Scott et al., 17 2018; Jack et al., 2020; Chapter 10, Box 10.2). They can also help in assessing risks associated with LLHI 18 events (Weitzman, 2011; Sutton, 2018), because they consider the ‘physically self-consistent unfolding of 19 past events, or of plausible future events or pathways’ (Shepherd et al., 2018b), which would be masked in a 20 probabilistic approach. These aspects are important as the greatest risk need not be associated with the 21 highest-likelihood outcome, and in fact will often be associated with low-likelihood outcomes. The storyline 22 approach can also acknowledge that climate-relevant decisions in a risk-oriented framing will rarely be taken 23 on the basis of physical climate change alone; instead, such decisions will normally take into account socio- 24 economic factors as well (Shepherd, 2019). 25 26 In the AR6 WGI Assessment Report, these different storyline approaches are used in several places (see 27 Table 1.1). Chapter 4 uses a storyline approach to assess the upper tail of the distribution of global warming 28 levels (the storylines of high global warming levels) and their manifestation in global patterns of temperature 29 and precipitation changes. Chapter 9 uses a storyline approach to examine the potential for, and early 30 warning signals of, a high-end sea-level scenario, in the context of deep uncertainty related to our current 31 understanding the physical processes that contribute to long-term sea-level rise. Chapter 10 assesses the use 32 of physical climate storylines and narratives as a way to explore uncertainties in regional climate projections, 33 and to link to the specific risk and decision context relevant to a user, for developing integrated and context- 34 relevant regional climate change information. Chapter 11 uses the term storyline in the framework of 35 extreme event attribution. Chapter 12 assesses the use of a storylines approach with narrative elements for 36 communicating climate (change) information in the context of climate services (Cross-Chapter Box 12.2 in 37 Chapter 12). 38 39 40 [START CROSS-CHAPTER BOX 1.3 HERE] 41 42 Cross-Chapter Box 1.3: Risk framing in IPCC AR6 43 44 Contributing Authors: Andy Reisinger (New Zealand), Maisa Rojas (Chile), Maarten van Aalst 45 (Netherlands), Aïda Diongue-Niang (Senegal), Mathias Garschagen (Germany), Mark Howden (Australia), 46 Margot Hurlbert (Canada), Katie Mach (USA), Sawsan Mustafa (Sudan), Brian O’Neill (USA), Roque 47 Pedace (Argentina), Jana Sillmann (Norway), Carolina Vera (Argentina), David Viner (UK). 48 49 The IPCC SREX presented a framework for assessing risks from climate change, linking hazards (due to 50 changes in climate) with exposure and vulnerability (Cardona et al., 2012). This framework was further 51 developed by AR5 WGII (IPCC, 2014b), while AR5 WGI focussed only on the hazard component of risk. 52 As part of AR6, a cross-Working Group process expanded and refined the concept of risk to allow for a 53 consistent risk framing to be used across the three IPCC working groups (IPCC, 2019b; Box 2 in Abram et 54 al., 2019; Reisinger et al., 2020). 55 Do Not Cite, Quote or Distribute 1-63 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 In this revised definition, risk is the ‘potential for adverse consequences for human or ecological systems, 2 recognising the diversity of values and objectives associated with such systems. In the context of climate 3 change, risks can arise not only from impacts of climate change, but also from potential human responses 4 to climate change. Relevant adverse consequences include those on lives, livelihoods, health and wellbeing, 5 economic, social and cultural assets and investments, infrastructure, services (including ecosystem services), 6 ecosystems and species. 7 8 In the context of climate change impacts, risks result from dynamic interactions between climate-related 9 hazards with the exposure and vulnerability of the affected human or ecological system to hazards. Hazards, 10 exposure and vulnerability may each be subject to uncertainty in terms of magnitude and likelihood of 11 occurrence, and each may change over time and space due to socio-economic changes and human decision- 12 making. 13 14 In the context of climate change responses, risks result from the potential for such responses not achieving 15 the intended objective(s), or from potential trade-offs with, or negative side-effects on, other societal 16 objectives, such as the Sustainable Development Goals. Risks can arise for example from uncertainty in 17 implementation, effectiveness or outcomes of climate policy, climate-related investments, technology 18 development or adoption, and system transitions’. 19 20 The following concepts are also relevant for the definition of risk (see Annex VII: Glossary): 21 22 Exposure: The presence of people, livelihoods, species or ecosystems, environmental functions, services, 23 and resources, infrastructure, or economic, social, or cultural assets in places and settings that could be 24 adversely affected. 25 26 Vulnerability: The propensity or predisposition to be adversely affected. Vulnerability encompasses a 27 variety of concepts and elements including sensitivity or susceptibility to harm and lack of capacity to cope 28 and adapt. 29 30 Hazard: The potential occurrence of a natural or human-induced physical event or trend that may cause loss 31 of life, injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods, 32 service provision, ecosystems and environmental resources. 33 34 Impacts: The consequences of realised risks on natural and human systems, where risks result from the 35 interactions of climate-related hazards (including extreme weather and climate events), exposure, and 36 vulnerability. Impacts generally refer to effects on lives, livelihoods, health and wellbeing, ecosystems and 37 species, economic, social and cultural assets, services (including ecosystem services), and infrastructure. 38 Impacts may be referred to as consequences or outcomes, and can be adverse or beneficial. 39 40 Risk in AR6 WGI 41 42 The revised risk framing clarifies the role and contribution of WGI to risk assessment. Risk in IPCC 43 terminology applies only to human or ecological systems, not to physical systems on their own. 44 45 Climatic impact-drivers: CIDs are physical climate system conditions (e.g., means, extremes, events) that 46 affect an element of society or ecosystems. Depending on system tolerance, CIDs and their changes can be 47 detrimental, beneficial, neutral, or a mixture of each across interacting system elements and regions. 48 49 In AR6, WGI uses the term ‘climatic impact-drivers’ (CIDs) to describe changes in physical systems rather 50 than ‘hazards’, because the term hazard already assumes an adverse consequence. The terminology of 51 ‘climatic impact-driver’ therefore allows WGI to provide a more value-neutral characterisation of climatic 52 changes that may be relevant for understanding potential impacts, without pre-judging whether specific 53 climatic changes necessarily lead to adverse consequences, as some could also result in beneficial outcomes 54 depending on the specific system and associated values. Chapter 12 and the Atlas assess and provide 55 information on climatic impact-drivers for different regions and sectors to support and link to WGII Do Not Cite, Quote or Distribute 1-64 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 assessment of the impacts and risks (or opportunities) related to the changes in the climatic impact-drivers. 2 Although CIDs can lead to adverse or beneficial outcomes, focus is given to CIDs connected to hazards, and 3 hence inform risk. 4 5 ‘Extremes’ are a category of CID, corresponding to unusual events with respect to the range of observed 6 values of the variable. Chapter 11 assesses changes in weather and climate extremes, their attribution and 7 future projections. 8 9 As examples of the use of this terminology, the term ‘flood risk’ should not be used if it only describes 10 changes in the frequency and intensity of flood events (a hazard); the risk from flooding to human and 11 ecological systems is caused by the flood hazard, the exposure of the system affected (e.g., topography, 12 human settlements or infrastructure in the area potentially affected by flooding) and the vulnerability of the 13 system (e.g., design and maintenance of infrastructure, existence of early warning systems). As another 14 example, climate-related risk to food security can arise from both potential climate change impacts and 15 responses to climate change and can be exacerbated by other stressors. Drivers for risks related to climate 16 change impacts include climate hazards (e.g., drought, temperature extremes, humidity), mediated by other 17 climatic impact-drivers (e.g., increased CO2 fertilisation of certain types of crops may help increase yields), 18 the potential for indirect climate-related impacts (e.g., pest outbreaks triggered by ecosystem responses to 19 weather patterns), exposure of people (e.g., how many people depend on a particular crop) and vulnerability 20 or adaptability (how able are affected people to substitute other sources of food, which may be related to 21 financial access and markets). 22 23 Information provided by WGI may or may not be relevant to understand risks related to climate change 24 responses. For example, the risk to a company arising from emissions pricing, or the societal risk from 25 reliance on an unproven mitigation technology, are not directly dependent on actual or projected changes in 26 climate but arise largely from human choices. However, WGI climate information may be relevant to 27 understand the potential for maladaptation, such as the potential for specific adaptation responses not 28 achieving the desired outcome or having negative side-effects. For example, WGI information about the 29 range of sea level rise can help inform understanding of whether coastal protection, accommodation, or 30 retreat would be the most effective risk management strategy in a particular context. 31 32 From a WGI perspective also relevant for risk assessment are low-likelihood high impact events and the 33 concept of deep uncertainty. 34 35 Low-likelihood, high-impact events: (LLHI) ‘These are events whose probability of occurrence is low but 36 whose potential impacts on society and ecosystems are high. To better inform risk assessment and decision 37 making, such low likelihood outcomes are described as they may be associated with very high levels of risk 38 and because the greatest risks might not be associated with the most expected outcome. 39 40 The AR6 WGI report provides more detailed information about these types of events compared to the AR5 41 (see Table 1.1, Section 1.4.4). 42 43 Recognising the need for assessing and managing risk in situations of high uncertainty, the SROCC 44 advanced the treatment of situations with deep uncertainty (IPCC, 2019b; Box 5 in Abram et al., 2019); 45 Section 1.2.3). A situation of deep uncertainty exists when experts or stakeholders do not know or cannot 46 agree on: (1) appropriate conceptual models that describe relationships among key driving forces in a 47 system; (2) the probability distributions used to represent uncertainty about key variables and parameters; 48 and/or (3) how to weigh and value desirable alternative outcomes (Abram et al., 2019). The concept of deep 49 uncertainty can complement the IPCC calibrated language and thereby broaden the communication of risk. 50 51 [END CROSS-CHAPTER BOX 1.3 HERE] 52 53 54 1.4.4.3 Abrupt change, tipping points and surprises 55 Do Not Cite, Quote or Distribute 1-65 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 An abrupt change is defined in this report as a change that takes place substantially faster than the rate of 2 change in the recent history of the affected component of a system (see Annex VII: Glossary). In some cases, 3 abrupt change occurs because the system state actually becomes unstable, such that the subsequent rate of 4 change is independent of the forcing. We refer to this class of abrupt change as a tipping point, defined as a 5 critical threshold beyond which a system reorganizes, often abruptly and/or irreversibly (Lenton et al., 2008); 6 Annex VII: Glossary). Some of the abrupt climate changes and climate tipping points discussed in this report 7 could have severe local climate responses, such as extreme temperature, droughts, forest fires, ice sheet loss 8 and collapse of the thermohaline circulation (see Chapter 4, Section 4.7.2, Chapter 5, Section 5.4.9, Chapter 9 8, Section 8.6 and Chapter 9, Section 9.2.3). 10 11 There is evidence of abrupt change in Earth’s history, and some of these events have been interpreted as 12 tipping points (Dakos et al., 2008) . Some of these are associated with significant changes in the global 13 climate, such as deglaciations in the Quaternary (past 2.5 million years) and rapid warming at the 14 Palaeocene-Eocene Thermal Maximum (around 55.5 million years ago) (Bowen et al., 2015; Hollis et al., 15 2019). Such events changed the planetary climate for tens to hundreds of thousands of years, but at a rate 16 that is actually much slower than projected anthropogenic climate change over this century, even in the 17 absence of tipping points. 18 19 Such paleoclimate evidence has even fuelled concerns that anthropogenic GHGs could tip the global climate 20 into a permanent hot state (Steffen et al., 2018). However, there is no evidence of such non-linear responses 21 at the global scale in climate projections for the next century, which indicate a near-linear dependence of 22 global temperature on cumulative GHG emissions (Section 1.3.5, Chapter 5, Section 5.5 and Chapter 7, 23 Section 7.4.3.1). At the regional scale, abrupt changes and tipping points, such as Amazon forest dieback and 24 permafrost collapse, have occurred in projections with Earth System Models (Drijfhout et al., 2015; Bathiany 25 et al., 2020; Chapter 4, Section 4.7.3). In such simulations, tipping points occur in narrow regions of 26 parameter space (e.g., CO2 concentration or temperature increase), and for specific climate background 27 states. This makes them difficult to predict using ESMs relying on parmeterizations of known processes. In 28 some cases, it is possible to detect forthcoming tipping points through time-series analysis that identifies 29 increased sensitivity to perturbations as the tipping point is approached (e.g., ‘critical slowing-down’, 30 Scheffer et al., 2012). 31 32 Some suggested climate tipping points prompt transitions from one steady state to another (see Figure 1.17). 33 Transitions can be prompted by perturbations such as climate extremes which force the system outside of its 34 current well of attraction in the stability landscape; this is called noise-induced tipping (Ashwin et al., 2012; 35 Figure 1.17, panels a/b). For example, the tropical forest dieback seen in some ESM projections is 36 accelerated by longer and more frequent droughts over tropical land (Good et al., 2013). 37 38 Alternatively, transitions from one state to another can occur if a critical threshold is exceeded; this is called 39 bifurcation tipping (Ashwin et al., 2012; Figure 1.17, panels c/d). The new state is defined as irreversible on 40 a given timescale if the recovery from this state takes substantially longer than the timescale of interest, 41 which is decades to centuries for the projections presented in this report. A well-known example is the 42 modelled irreversibility of the ocean’s thermohaline circulation in response to North Atlantic changes such 43 as freshwater input from rainfall and ice-sheet melt (Rahmstorf et al., 2005; Alkhayuon et al., 2019), which 44 is assessed in detail in Chapter 9, Section 9.2.3. 45 46 The tipping point concept is most commonly framed for systems in which the forcing changes relatively 47 slowly. However, this is not the case for most scenarios of anthropogenic forcing projected for the 21st 48 century. Systems with inertia lag behind rapidly-increasing forcing, which can lead to the failure of early 49 warning signals or even the possibility of temporarily overshooting a bifurcation point without provoking 50 tipping (Ritchie et al., 2019). 51 52 53 [START FIGURE 1.17 HERE] 54 55 Figure 1.17: Illustration of two types of tipping points: noise-induced (panels a, b) and bifurcation (panels c, d). Do Not Cite, Quote or Distribute 1-66 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 (a), (c) example time-series (coloured lines) through the tipping point with black solid lines indicating 2 stable climate states (e.g., low or high rainfall) and dashed lines represent the boundary between stable 3 states. (b), (d) stability landscapes provide an intuitive understanding for the different types of tipping 4 point. The valleys represent different climate states the system can occupy, with hill tops separating the 5 stable states. The resilience of a climate state is implied by the depth of the valley. The current state of the 6 system is represented by a ball. Both scenarios assume that the ball starts in the left-hand valley (black 7 dashed lines) and then through different mechanisms dependent on the type of tipping transitions to the 8 right valley (coloured lines). Noise-induced tipping events, for instance drought events causing sudden 9 dieback of the Amazonian rainforest, develop from fluctuations within the system. The stability landscape 10 in this scenario remains fixed and stationary. A series of perturbations in the same direction or one large 11 perturbation are required to force the system over the hill top and into the alternative stable state. 12 Bifurcation tipping events, such as a collapse of the thermohaline circulation in the Atlantic Ocean under 13 climate change, occur when a critical level in the forcing is reached. Here the stability landscape is 14 subjected to a change in shape. Under gradual anthropogenic forcing the left valley begins to shallow and 15 eventually vanishes at the tipping point, forcing the system to transition to the right-hand valley. 16 17 [END FIGURE 1.17 HERE] 18 19 20 Surprises are a class of risk that can be defined as low-likelihood but well-understood events, and events that 21 cannot be predicted with current understanding. The risk from such surprises can be accounted for in risk 22 assessments (Parker and Risbey, 2015). Examples relevant to climate science include: a series of major 23 volcanic eruptions or a nuclear war, either of which would cause substantial planetary cooling (Robock et al., 24 2007; Mills et al., 2014); significant 21st century sea level rise due to marine ice sheet instability (MISI, 25 Chapter 9, Box 9.4); the potential for collapse of the stratocumulus cloud decks (Schneider et al., 2019) or 26 other substantial changes in climate feedbacks (see Chapter 7, Section 7.4); and unexpected biological 27 epidemics among humans or other species, such as the COVID-19 pandemic (Forster et al., 2020; Le Quéré 28 et al., 2020; see Cross-Chapter Box 6.1 in Chapter 6). The discovery of the ozone hole was also a surprise 29 even though some of the relevant atmospheric chemistry was known at the time. The term ‘unknown 30 unknowns’ (Parker and Risbey, 2015) is also sometimes used in this context to refer to events that cannot be 31 anticipated with present knowledge or were of an unanticipated nature before they occurred. 32 33 34 [START CROSS-WORKING GROUP BOX: ATTRIBUTION HERE] 35 36 Cross-Working Group Box: Attribution 37 38 Contributing Authors: Wolfgang Cramer (France/Germany), Pandora Hope (Australia), Maarten van Aalst 39 (Netherlands), Greg Flato (Canada), Katja Frieler (Germany), Nathan Gillett (Canada/UK), Christian Huggel 40 (Switzerland), Jan Minx (Germany), Friederike Otto (UK/Germany), Camille Parmesan (France/UK/USA), 41 Joeri Rogelj (UK/Belgium), Maisa Rojas (Chile), Sonia I. Seneviratne (Switzerland), Aimee Slangen 42 (Netherlands), Daithi Stone (New Zealand), Laurent Terray (France), Robert Vautard (France), Xuebin 43 Zhang (Canada) 44 45 Introduction 46 47 Changes in the climate system are becoming increasingly apparent, as are the climate-related impacts on 48 natural and human systems. Attribution is the process of evaluating the contribution of one or more causal 49 factors to such observed changes or events. Typical questions addressed by the IPCC are for example: ‘To 50 what degree is an observed change in global temperature induced by anthropogenic greenhouse gas and 51 aerosol concentration changes or influenced by natural variability?’ or ‘What is the contribution of climate 52 change to observed changes in crop yields that are also influenced by changes in agricultural management?’ 53 Changes in the occurrence and intensity of extreme events can also be attributed, addressing questions such 54 as: ‘Have human greenhouse gas emissions increased the likelihood or intensity of an observed heat wave?’ 55 56 This Cross-Working Group Box briefly describes why attribution studies are important. It also describes Do Not Cite, Quote or Distribute 1-67 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 some new developments in the methods used and provides recommendations for interpretation. 2 3 Attribution studies serve to evaluate and communicate linkages associated with climate change, for example: 4 between the human-induced increase in greenhouse gas concentrations and the observed increase in air 5 temperature or extreme weather events (WGI Chapter 3, 10, 11); or between observed changes in climate 6 and changing species distributions and food production (WGII Chapters 2 and others, summarised in Chapter 7 16) (e.g., Verschuur et al., 2021); or between climate change mitigation policies and atmospheric greenhouse 8 gas concentrations (WGI Chapter 5; WGIII Chapter 14). As such, they support numerous statements made 9 by the IPCC (IPCC, 2013b, 2014b; WGI Chapter 1, Section 1.3, Appendix 1A). 10 11 Attribution assessments can also serve to monitor mitigation and assess the efficacy of applied climate 12 protection policies (e.g., Nauels et al., 2019; Banerjee et al., 2020; WGI Chapter 4, Section 4.6.3), inform 13 and constrain projections (Gillett et al., 2021; Ribes et al., 2021; WGI Chapter 4, Section 4.2.3) or inform the 14 loss and damages estimates and potential climate litigation cases by estimating the costs of climate change 15 (Huggel et al., 2015; Marjanac et al., 2017; Frame et al., 2020). These findings can thus inform mitigation 16 decisions as well as risk management and adaptation planning (e.g., CDKN, 2017). 17 18 Steps towards an attribution assessment 19 20 The unambiguous framing of what is being attributed to what is a crucial first step for an assessment 21 (Easterling et al., 2016; Hansen et al., 2016; Stone et al., 2021), followed by the identification of the possible 22 and plausible drivers of change and the development of a hypothesis or theory for the linkage (see Cross- 23 Working Group Box: Attribution, Figure 1). The next step is to clearly define the indicators of the observed 24 change or event and note the quality of the observations. There has been significant progress in the 25 compilation of fragmented and distributed observational data, broadening and deepening the data basis for 26 attribution research (e.g., Poloczanska et al., 2013; Ray et al., 2015; Cohen et al., 2018; WGI Chapter 1, 27 Section 1.5). The quality of the observational record of drivers should also be considered (e.g., volcanic 28 eruptions: WGI Chapter 2, section 2.2.2). Impacted systems also change in the absence of climate change; 29 this baseline and its associated modifiers such as agricultural developments or population growth need to be 30 considered, alongside the exposure and vulnerability of people depending on these systems. 31 32 There are many attribution approaches, and several methods are detailed below. In physical and biological 33 systems, attribution often builds on the understanding of the mechanisms behind the observed changes and 34 numerical models are used, while in human systems other methods of evidence-building are employed. 35 Confidence in the attribution can be increased if more than one approach is used and the model is evaluated 36 as fit-for-purpose (Hegerl et al., 2010; Vautard et al., 2019; Otto et al., 2020; Philip et al., 2020) (WGI 37 Chapter 1, Section 1.5). Finally, appropriate communication of the attribution assessment and the 38 accompanying confidence in the result (e.g., Lewis et al., 2019). 39 40 Attribution methods 41 42 Attribution of changes in atmospheric greenhouse gas concentrations to anthropogenic activity 43 44 AR6 WGI Chapter 5 presents multiple lines of evidence that unequivocally establish the dominant role of 45 human activities in the growth of atmospheric CO2, including through analysing changes in atmospheric 46 carbon isotope ratios and the atmospheric O2-N2 ratio (WGI Chapter 5, Section 5.2.1.1). Decomposition 47 approaches can be used to attribute emissions underlying those changes to various drivers such as 48 population, energy efficiency, consumption or carbon intensity (Hoekstra and van den Bergh, 2003; Raupach 49 et al., 2007; Rosa and Dietz, 2012). Combined with attribution of their climate outcomes, the attribution of 50 the sources of greenhouse gas emissions can inform the attribution of anthropogenic climate change to 51 specific countries or actors (Matthews, 2016; Otto et al., 2017; Skeie et al., 2017; Nauels et al., 2019), and in 52 turn inform discussions on fairness and burden sharing (WGIII Chapter 14). 53 54 55 Attribution of observed climate change to anthropogenic forcing Do Not Cite, Quote or Distribute 1-68 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Changes in large-scale climate variables (e.g., global mean temperature) have been reliably attributed to 3 anthropogenic and natural forcings (e.g., Hegerl et al., 2010; Bindoff et al., 2013; WGI Chapter 1, Section 4 1.3.4). The most established method is to identify the ‘fingerprint’ of the expected space-time response to a 5 particular climate forcing agent such as the concentration of anthropogenically induced greenhouse gases or 6 aerosols, or natural variation of solar radiation. This technique disentangles the contribution of individual 7 forcing agents to an observed change (e.g., Gillett et al., 2021). New statistical approaches have been applied 8 to better account for internal climate variability and the uncertainties in models and observations (e.g., 9 Naveau et al., 2018; Santer et al., 2019) (WGI, Chapter 3 Section 3.2). There are many other approaches, for 10 example, global mean sea-level change has been attributed to anthropogenic climate forcing by attributing 11 the individual contributions from, for example, glacier melt or thermal expansion, while also examining 12 which aspects of the observed change are inconsistent with internal variability (WGI Chapter 3, Section 3.5.2 13 and WGI Chapter 9, Section 9.6.1.4). 14 15 Specific regional conditions and responses may simplify or complicate attribution on those scales. For 16 example, some human forcings, such as regional land use change or aerosols, may enhance or reduce 17 regional signals of change (Lejeune et al., 2018; Undorf et al., 2018; Boé et al., 2020; Thiery et al., 2020; see 18 also WGI Chapter 10, Section 10.4.2; WGI Chapter 11, Sections 11.1.6 and 11.2.2. In general, regional 19 climate variations are larger than the global mean climate, adding additional uncertainty to attribution (e.g., 20 in regional sea-level change, WGI Chapter 9, Section 9.6.1). These statistical limitations may be reduced by 21 ‘process-based attribution’, focusing on the physical processes known to influence the response to external 22 forcing and internal variability (WGI Chapter 10, Section 10.4.2). 23 24 Attribution of weather and climate events to anthropogenic forcing 25 26 New methods have emerged since AR5 to attribute the change in likelihood or characteristics of weather or 27 climate events or classes of events to underlying drivers (National Academies of Sciences Engineering and 28 Medicine, 2016; Stott et al., 2016; Jézéquel et al., 2018; Wehner et al., 2018; Wang et al., 2020; WGI 29 Chapter 10, Section 10.4.1; WG1 Chapter 11, Section 11.2.2). Typically, historical changes, simulated under 30 observed forcings, are compared to a counterfactual climate simulated in the absence of anthropogenic 31 forcing. Another approach examines facets of the weather and thermodynamic status of an event through 32 process-based attribution (Hauser et al., 2016; Shepherd et al., 2018b; Grose et al., 2019; WGI Chapter 10 33 Section 10.4.1 and Chapter 11). Events where attributable human influences have been found include hot and 34 cold temperature extremes (including some with wide-spread impacts), heavy precipitation, and certain types 35 of droughts and tropical cyclones (e.g., Vogel et al., 2019; Herring et al., 2021; AR6 WGI Chapter 11, 36 Section 11.9). Event attribution techniques have sometimes been extended to ‘end-to-end’ assessments from 37 climate forcing to the impacts of events on natural or human systems (Otto, 2017, examples in WGII Table 38 16.1, SI of WGII Chapter 16, Section 16.2). 39 40 Attribution of observed changes in natural or human systems to climate-related drivers 41 42 The attribution of observed changes to climate-related drivers across a diverse set of sectors, regions and 43 systems is part of each chapter in the WGII contribution to the AR6 and is synthesised in WGII Chapter 16 44 (Section 16.2). The number of attribution studies on climate change impacts has grown substantially since 45 AR5, generally leading to higher confidence levels in attributing the causes of specific impacts. New studies 46 include the attribution of changes in socio-economic indicators such as economic damages due to river 47 floods (e.g., Schaller et al., 2016; Sauer et al., 2021), the occurrence of heat related human mortality (e.g., 48 Sera et al., 2020, Vicedo-Cabrera et al., 2018;) or economic inequality (e.g., Diffenbaugh and Burke, 2019). 49 50 Impact attribution covers a diverse set of qualitative and quantitative approaches, building on experimental 51 approaches, observations from remote sensing, long-term in situ observations, and monitoring efforts, 52 teamed with local knowledge, process understanding and empirical or dynamical modelling (WGII Chapter 53 16, Section 16.2; Stone et al., 2013; Cramer et al., 2014). The attribution of a change in a natural or human 54 system (e.g., wild species, natural ecosystems, crop yields, economic development, infrastructure or human 55 health) to changes in climate-related systems (i.e., climate, and ocean acidification, permafrost thawing or Do Not Cite, Quote or Distribute 1-69 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 sea-level rise) requires accounting for other potential drivers of change, such as technological and economic 2 changes in agriculture affecting crop production (Hochman et al., 2017; Butler et al., 2018), changes in 3 human population patterns and vulnerability affecting flood or wildfire induced damages (Huggel et al., 4 2015; Sauer et al., 2021), or habitat loss driving declines in wild species (IPBES, 2019). These drivers are 5 accounted for by estimating a baseline condition that would exist in the absence of climate change. The 6 baseline might be stationary and be approximated by observations from the past, or it may change over time 7 and be simulated by statistical or process-based impact models (Cramer et al. 2014, WGII Chapter 16, 8 Section 16.2). Assessment of multiple independent lines of evidence, taken together, can provide rigorous 9 attribution when more quantitative approaches are not available (Parmesan et al., 2013). These include 10 paleodata, physiological and ecological experiments, natural ‘experiments’ from very long-term datasets 11 indicating consistent responses to the same climate trend/event, and ‘fingerprints’ in species' responses that 12 are uniquely expected from climate change (e.g. poleward range boundaries expanding and equatorial range 13 boundaries contracting in a coherent pattern world-wide, Parmesan and Yohe, 2003). Meta-analyses of 14 species/ecosystem responses, when conducted with wide geographic coverage, also provide a globally 15 coherent signal of climate change at an appropriate scale for attribution to anthropogenic climate change 16 (Parmesan and Yohe, 2003; Parmesan et al., 2013). 17 18 Impact attribution does not always involve attribution to anthropogenic climate forcing. However, a growing 19 number of studies include this aspect (e.g., Frame et al., 2020 for the attribution of damages induced by 20 hurricane Harvey; or Diffenbaugh and Burke, 2019 for the attribution of economic inequality between 21 countries; or Schaller et al., 2016 for flood damages). 22 23 24 [START CROSS-WORKING GROUP BOX: ATTRIBUTION, FIGURE 1 HERE] 25 26 Cross-Working Group Box: Attribution, Figure 1: Schematic of the steps to develop an attribution assessment, 27 and the purposes of such assessments. Methods and systems 28 used to test the attribution hypothesis or theory include model- 29 based fingerprinting, other model-based methods, evidence- 30 based fingerprinting, process-based approaches, empirical or 31 decomposition methods and the use of multiple lines of 32 evidence. Many of the methods are based on the comparison of 33 the observed state of a system to a hypothetical counterfactual 34 world that does not include the driver of interest to help estimate 35 the causes of the observed response. 36 37 [END CROSS-WORKING GROUP BOX: ATTRIBUTION, FIGURE 1 HERE] 38 39 40 [END CROSS-WORKING GROUP BOX: ATTRIBUTION HERE] 41 42 43 1.4.5 Climate regions used in AR6 44 45 1.4.5.1 Defining climate regions 46 47 AR5 assessed regional scale detection and attribution and assessed key regional climate phenomena and their 48 relevance for future regional climate projections. This report shows that past and future climate changes and 49 extreme weather events can be substantial on local and regional scales (Chapters 8–12, Atlas), where they 50 may differ considerably from global trends, not only in intensity but even in sign (e.g., Fischer et al., 2013). 51 52 Although the evolution of global climate trends emerges as the net result of regional phenomena, average or 53 aggregate estimates often do not reflect the intensity, variability, and complexity of regional climate changes 54 (Stammer et al., 2018; Shepherd, 2019). A fundamental aspect of the study of regional climate changes is the 55 definition of characteristic climate zones, clusters or regions, across which the emergent climate change 56 signal can be properly analysed and projected (see Atlas). Suitable sizes and shapes of such zones strongly Do Not Cite, Quote or Distribute 1-70 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 depend not only on the climate variable and process of interest, but also on relevant multiscale feedbacks. 2 3 There are several approaches to the classification of climate regions. When climate observation data was 4 sparse and limited, the aggregation of climate variables was implicitly achieved through the consideration of 5 biomes, giving rise to the traditional vegetation-based classification of Köppen (1936). In the last decades, 6 the substantial increases in climate observations, climate modelling, and data processing capabilities have 7 allowed new approaches to climate classification, e.g., through interpolation of aggregated global data from 8 thousands of stations (Peel et al., 2007; Belda et al., 2014; Beck et al., 2018a) or through data-driven 9 approaches applied to delineate ecoregions that behave in a coherent manner in response to climate 10 variability (Papagiannopoulou et al., 2018). Experience shows that each method has strengths and 11 weaknesses through trade-offs between detail and convenience. For instance, a very detailed classification, 12 with numerous complexly shaped regions derived from a large set of variables, may be most useful for the 13 evaluation of climate models (Rubel and Kottek, 2010; Belda et al., 2015; Beck et al., 2018a) and climate 14 projections (Feng et al., 2014; Belda et al., 2016). In contrast, geometrically simple regions are often best 15 suited for regional climate modelling and downscaling (e.g., the Coordinated Regional Climate Downscaling 16 Experiment (CORDEX) domains; see Giorgi and Gutowski, 2015, and Section 1.5.3). 17 18 19 1.4.5.2 Types of regions used in AR6 20 21 IPCC’s recognition of the importance of regional climates can be traced back to its First Assessment Report 22 (IPCC, 1990a), where climate projections for 2030 were presented for five subcontinental regions (see 23 Section 1.3.6 for an assessment of those projections). In subsequent reports, there has been a growing 24 emphasis on the analysis of regional climate, including two special reports: one on regional impacts (IPCC, 25 1998) and another on extreme events (SREX, IPCC, 2012). A general feature of previous IPCC reports is 26 that the number and coverage of climate regions vary according to the subject and across WGs. Such varied 27 definitions have the advantage of optimizing the results for a particular application (e.g., national boundaries 28 are crucial for decision making, but they rarely delimit distinctive climate regions), whereas variable region 29 definitions may have the disadvantage of hindering multidisciplinary assessments and comparisons between 30 studies or WGs. 31 32 In this Report, regional climate change is primarily addressed through the introduction of four classes of 33 regions (unless otherwise explicitly mentioned and justified). The first two are the unified WGI Reference 34 Sets of (1) Land and (2) Ocean Regions, which are used in the entire Report. These are supplemented by 35 additional sets of (3) Typological Regions — used in Chapters 5, 8–12 and Atlas — and (4) Continental 36 Regions, which are mainly used for linking Chapters 11, 12 and Atlas with WGII (Figure 1.18). All four 37 classes of regions are defined and described in detail in the Atlas. Here we summarize their basic features. 38 39 40 [START FIGURE 1.18 HERE] 41 42 Figure 1.18: Main types of regions used in this report. (a) AR6 WGI Reference Set of Land and Ocean Regions 43 (Iturbide et al., 2020), consisting of 46 land regions and 15 ocean regions, including 3 hybrid regions 44 (CAR, MED, SEA) that are both land and ocean regions. Acronyms are explained on the right of the map. 45 Notice that RAR, SPO, NPO and EPO extend beyond the 180º meridian, therefore appearing at both sides 46 of the map (indicated by dashed lines). A comparison with the previous reference regions of AR5 WGI 47 (IPCC, 2013a) is presented in the Atlas. (b) Example of typological regions: monsoon domains adopted in 48 Chapter 8. Acronyms are explained on the right of the map. The black contour lines represent the global 49 monsoon zones, while the coloured regions denote the regional monsoon domains. The two stippled 50 regions (EqAmer and SAfri) do receive seasonal rainfall, but their classification as monsoon regions is 51 still under discussion. (c) Continental Regions used mainly in Chapter 12 and the Atlas. Stippled zones 52 define areas that are assessed in both regions (e.g., the Caribbean is assessed as Small Islands and also as 53 part of Central America). Small Islands are ocean regions containing small islands with consistent climate 54 signals and/or climatological coherence. 55 56 [END FIGURE 1.18 HERE] Do Not Cite, Quote or Distribute 1-71 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 3 Reference Land and Ocean Regions are polygonal, sub-continental domains defined through a combination 4 of environmental, climatic and non-climatic (e.g., pragmatic, technical, historical) factors, in accordance 5 with the literature and climatological reasoning based on observed and projected future climate. Merging the 6 diverse functions and purposes of the regions assessed in the literature into a common reference set implies a 7 certain degree of compromise between simplicity, practicality, and climate consistency. For instance, Spain 8 is fully included in the Mediterranean (MED) Reference Region, but is one of the most climatically diverse 9 countries in the world. Likewise, a careful comparison of panels (a) and (b) of Figure 1.18) reveals that the 10 simplified southern boundary of the Sahara (SAH) Reference Region slightly overlaps the northern boundary 11 of the West African Monsoon Typological Region. As such, the resulting Reference Regions are not 12 intended to precisely represent climates, but rather to provide simple domains suitable for regional synthesis 13 of observed and modelled climate and climate change information (Iturbide et al., 2020). In particular, 14 CMIP6 model results averaged over Reference Regions are presented in the Atlas. 15 16 The starting point for defining the AR6 Reference Sets of Land Regions was the collection of 26 regions 17 introduced in SREX (IPCC, 2012). The SREX collection was then revised, reshaped, complemented and 18 optimized to reflect the recent scientific literature and observed climate-change trends, giving rise to the 19 novel AR6 reference set of 46 land regions. Additionally, AR6 introduces a new reference set of 15 ocean 20 regions (including 3 hybrid regions that are treated as both, land and ocean), which complete the coverage of 21 the whole Earth (Iturbide et al., 2020). 22 23 Particular aspects of regional climate change are described by specialized domains called Typological 24 Regions (Figure 1.18b). These regions cover a wide range of spatial scales and are defined by specific 25 features, called typologies. Examples of typologies include: tropical forests, deserts, mountains, monsoon 26 regions, and megacities, among others. Typological Regions are powerful tools to summarize complex 27 aspects of climate defined by a combination of multiple variables. For this reason, they are used in many 28 chapters of AR6 WGI and WGII (e.g., Chapters 8–12 and the Atlas). 29 30 Finally, consistency with WGII is also pursued in Chapter 11, 12 and Atlas through the use of a set of 31 Continental Regions (Figure 1.18c), based on the nine continental domains defined in AR5 WGII Part B 32 (Hewitson et al., 2014). These are classical geopolitical divisions of Africa, Asia, Australasia, Europe, North 33 America, Central and South America, plus Small Islands, Polar Regions, and the Ocean. In AR6 WGI, five 34 hybrid zones (Caribbean–Small islands, East Europe–Asia, European Arctic, North American Arctic, and 35 North Central America) are also identified, which are assessed in more than one continental region. 36 Additional consistency with WGIII is pursued by Chapter 6 through the use of sub-continental domains 37 which essentially form a subset of the Continental Set of Regions (Figure 1.18c and Chapter 6, Section 6.1). 38 39 40 1.5 Major developments and their implications 41 42 This section presents a selection of key developments since the AR5 of the capabilities underlying the lines 43 of evidence used in the present report: observational data and observing systems (Section 1.5.1), new 44 developments in reanalyses (Section 1.5.2), climate models (Section 1.5.3), and modelling techniques, 45 comparisons and performance assessments (Section 1.5.4). For brevity, we focus on the developments that 46 are of particular importance to the conclusions drawn in later chapters, though we also provide an assessment 47 of potential losses of climate observational capacity. 48 49 50 1.5.1 Observational data and observing systems 51 52 Progress in climate science relies on the quality and quantity of observations from a range of platforms: 53 surface-based instrumental measurements, aircraft, radiosondes and other upper-atmospheric observations, 54 satellite-based retrievals, ocean observations, and paleoclimatic records. An historical perspective to these 55 types of observations is presented in Section 1.3.1. Do Not Cite, Quote or Distribute 1-72 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Observed large-scale climatic changes assessed in Chapter 2, attribution of these changes in Chapter 3, and 3 regional observations of specific physical or biogeochemical processes presented in other Chapters, are 4 supported by improvements in observational capacity since the AR5. Attribution assessments can be made at 5 a higher likelihood level than AR5, due in part to the availability of longer observational datasets (Chapter 6 3). Updated assessments are made based on new and improved datasets, for example of global temperature 7 change (Cross-Chapter Box 2.3 in Chapter 2) or regional climate information (Chapter 10, Section 10.2). Of 8 particular relevance to the AR6 assessment are the ‘Essential Climate Variables’ (ECVs, Hollmann et al., 9 2013; Bojinski et al., 2014), and ‘Essential Ocean Variables’ (EOVs; Lindstrom et al., 2012), compiled by 10 the Global Climate Observing System (GCOS; WMO, 2016), and the Global Ocean Observing System 11 (GOOS), respectively. These variables include physical, chemical, and biological variables or groups of 12 linked variables and underpin ‘headline indicators’ for climate monitoring (Trewin et al., 2021). 13 14 We highlight below the key advances in observational capacity since the AR5, including major expansions 15 of existing observational platforms as well as new and/or emerging observational platforms that play a key 16 role in AR6. We then discuss potential near-term losses in key observational networks due to climate change 17 or other adverse human-caused influence. 18 19 20 1.5.1.1 Major expansions of observational capacity 21 22 Atmosphere, land and hydrological cycle 23 24 Satellites provide observations of a large number of key atmospheric and land surface variables, ensuring 25 sustained observations over wide areas. Since AR5, such observations have expanded to include satellite 26 retrievals of atmospheric CO2 via the NASA Orbiting Carbon Observatory satellites (OCO-2 and OCO-3, 27 Eldering et al., 2017), following on from similar efforts employing the Greenhouse Gases Observing 28 Satellite (GOSAT, Yokota et al., 2009; Inoue et al., 2016). Improved knowledge of fluxes between the 29 atmosphere and land surface results from combining remote sensing and in situ measurements (Rebmann et 30 al., 2018). FLUXNET (https://fluxnet.org/) has been providing eddy covariance measurements of carbon, 31 water, and energy fluxes between the land and the atmosphere, with some of the stations operating for over 32 20 years (Pastorello et al., 2017), while the Baseline Surface Radiation Network (BSRN) has been 33 maintaining high-quality radiation observations since the 1990s (Ohmura et al., 1998; Driemel et al., 2018). 34 35 Observations of the composition of the atmosphere have been further improved through expansions of 36 existing surface observation networks (Bodeker et al., 2016; De Mazière et al., 2018) and through in situ 37 measurements such as aircraft campaigns (Chapter 2, Section 2.2; Chapter 5, Section 5.2; Chapter 6, Section 38 6.2). Examples of expanded networks include Aerosols, Clouds, and Trace Gases Research InfraStructure 39 (ACTRIS) (Pandolfi et al., 2018), which focuses on short-lived climate forcers, and the Integrated Carbon 40 Observation System (ICOS), which allows scientists to study and monitor the global carbon cycle and 41 greenhouse gas emissions (Colomb et al., 2018). Examples of recent aircraft observations include the 42 Atmospheric Tomography Mission (ATom), which has flown repeatedly along the north-south axis of both 43 the Pacific and Atlantic oceans, and the continuation of the In-service Aircraft for a Global Observing 44 System (IAGOS) effort, which measures atmospheric composition from commercial aircraft (Petzold et al., 45 2015). 46 47 Two distinctly different but important remote sensing systems can provide information about temperature 48 and humidity since the early 2000s. Global Navigation Satellite Systems (e.g., GPS) radio occultation and 49 limb soundings provide information, although only data for the upper troposphere and lower stratosphere are 50 suitable to support climate change assessments (Angerer et al., 2017; Scherllin-Pirscher et al., 2017; Steiner 51 et al., 2019; Gleisner et al., 2020). These measurements complement those from the Atmospheric Infrared 52 Sounder (AIRS; Chahine et al., 2006). AIRS has limitations in cloudy conditions, although these limitations 53 have been partly solved using new methods of analysis (Blackwell and Milstein, 2014; Susskind et al., 54 2014). These new data sources now have a sufficient length of the record to strengthen the analysis of 55 atmospheric warming in Chapter 2, Section 2.3.1.2. Do Not Cite, Quote or Distribute 1-73 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Assessments of the hydrological cycle in Chapters 2 and 8 are supported by longer time series and new 3 developments. Examples are new satellites (McCabe et al., 2017) and measurements of water vapor using 4 commercial laser absorption spectrometers and water vapor isotopic composition (Steen-Larsen et al., 2015; 5 Zannoni et al., 2019). Data products of higher quality have been developed since AR5, such as the multi- 6 source weighted ensemble precipitation (Beck et al., 2017), and multi-satellite terrestrial evaporation (Fisher 7 et al., 2017). Longer series are available for satellite-derived global inundation (Prigent et al., 2020). 8 Observations of soil moisture are now available via the Soil Moisture and Ocean Salinity (SMOS) and the 9 Soil Moisture Active Passive (SMAP) satellite retrievals, filling critical gaps in the observation of 10 hydrological trends and variability over land (Dorigo et al., 2017). Similarly, the Gravity Recovery and 11 Climate Experiment GRACE and GRACE-FO satellites (Tapley et al., 2019) have provided key constraints 12 on groundwater variability and trends around the world (Frappart and Ramillien, 2018). The combination of 13 new observations with other sources of information has led to updated estimates of heat storage in inland 14 waters (Vanderkelen et al., 2020), contributing to revised estimates of heat storage on the continents (von 15 Schuckmann et al., 2020; Chapter 7, Section 7.2.2.3). 16 17 The ongoing collection of information about the atmosphere as it evolves is supplemented by the 18 reconstruction and digitization of data about past conditions. Programs aimed at recovering information from 19 sources such as handwritten weather journals and ship logs continue to make progress, and are steadily 20 improving spatial coverage and extending our knowledge backward in time. For example, Brönnimann et al. 21 (2019) has recently identified several thousand sources of climate data for land areas in the pre-1890 period, 22 with many from the 18th century. The vast majority of these data are not yet contained in international 23 digital data archives, and substantial quantities of undigitized ship’s weather log data exist for the same 24 period (Kaspar et al., 2015). Since the AR5 there has been a growth of ‘citizen science’ activities to rapidly 25 transcribe substantial quantities of weather observations involving volunteers. Examples of projects include: 26 oldWeather.org, and SouthernWeatherDiscovery.org that both used ship-based logbook sources, and the 27 DRAW (Data Rescue: Archival and Weather) project, WeatherRescue.org, JungleWeather.org and the 28 Climate History Australia project, which recovered land-based station data from Canada, Europe, the Congo 29 and Australia respectively (e.g., Park et al., 2018; Hawkins et al., 2019). Undergraduate students have also 30 been recruited to successfully digitise rainfall data in Ireland (Ryan et al., 2018). Such observations are an 31 invaluable source of weather and climate information for the early historical period that continues to expand 32 the digital archives (e.g., Freeman et al., 2017) which underpin observational datasets used across several 33 Chapters. 34 35 Ocean 36 37 Observations of the ocean have expanded significantly since the AR5, with expanded global coverage of in 38 situ ocean temperature and salinity observations, in situ ocean biogeochemistry observations, and satellite 39 retrievals of a variety of EOVs. Many recent advances are extensively documented in a compilation by Lee 40 et al. (2019). Below we discuss those most relevant for the current assessment. 41 42 Argo is a global network of nearly 4000 autonomous profiling floats (Roemmich et al., 2019), delivering 43 detailed constraints on the horizontal and vertical structure of temperature and salinity across the global 44 ocean. Argo has greatly expanded since AR5, including biogeochemistry and measurements deeper than 45 2000 m (Jayne et al., 2017), and the longer timeseries enable more rigorous climate assessments of direct 46 relevance to estimates of ocean heat content (Chapter 2, Section 2.3.3.1 ; Chapter 7, section 7.2.2.2). Argo 47 profiles are complemented by animal-borne sensors in several key areas, such as the seasonally ice-covered 48 sectors of the Southern Ocean (Harcourt et al., 2019). 49 50 Most basin-scale arrays of moored ocean instruments have expanded since AR5, providing decades-long 51 records of the ocean and atmosphere properties relevant for climate, such as the El Niño-Southern 52 Oscillation (Chen et al., 2018), deep convection (de Jong et al., 2018) or transports through straits 53 (Woodgate, 2018). Key basin-scale arrays include transport-measuring arrays in the Atlantic Ocean, 54 continuing (McCarthy et al., 2020) or newly added since AR5 (Lozier et al., 2019), supporting the 55 assessment of regional ocean circulation (Chapter 9, section 9.2.3). Tropical ocean moorings in the Pacific, Do Not Cite, Quote or Distribute 1-74 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Indian and Atlantic oceans include new sites, improved capability for real time transmission, and new 2 oxygen and CO2 sensors (Bourlès et al., 2019; Hermes et al., 2019; Smith et al., 2019b). 3 4 A decade of observations of sea-surface salinity is now available via the SMOS and SMAP satellite 5 retrievals, providing continuous and global monitoring of surface salinity in the open ocean and coastal areas 6 for the first time (Vinogradova et al., 2019; Reul et al., 2020) (Chapter 9, Section 9.2.2.2). 7 8 The global network of tide gauges, complemented by a growing number of satellite-based altimetry datasets, 9 allows for more robust estimates of global and regional sea level rise (Chapter 2, Section 2.3.3.3; Chapter 9, 10 Section 9.6.1.3). Incorporating vertical land motion derived from the Global Positioning System (GPS), the 11 comparison with tide gauges has allowed the correction of a drift in satellite altimetry series over the period 12 1993–1999 (Watson et al., 2015; Chen et al., 2017), thus improving our knowledge of the recent acceleration 13 of sea level rise (Chapter 2, Section 2.3.3.3). These datasets, combined with Argo and observations of the 14 cryosphere, allow a consistent closure of the global mean sea level budget (Cross-Chapter Box 9.1 in 15 Chapter 9; WCRP Global Sea Level Budget, 2018). 16 17 Cryosphere 18 19 For the cryosphere, there has been much recent progress in synthesizing global datasets covering larger areas 20 and longer time periods from multi-platform observations. For glaciers, the Global Terrestrial Network for 21 Glaciers, which combines data on glacier fluctuations, mass balance and elevation change with glacier 22 outlines and ice thickness, has expanded and provided input for assessing global glacier evolution and its 23 role in sea level rise (Chapter 2, Section 2.3.2.3; Chapter 9, Section 9.5.1; Zemp et al., 2019). New data 24 sources include archived and declassified aerial photographs and satellite missions, and high-resolution (10 25 m or less) digital elevation models (Porter et al., 2018; Braun et al., 2019). 26 27 Improvements have also been made in the monitoring of permafrost. The Global Terrestrial Network for 28 Permafrost (Biskaborn et al., 2015) provides long-term records of permafrost temperature and active layer 29 thickness at key sites to assess their changes over time. Substantial improvements to our assessments of 30 large-scale snow changes come from intercomparison and blending of several datasets, for snow water 31 equivalent (Mortimer et al., 2020) and snow cover extent (Mudryk et al., 2020), and from bias corrections of 32 combined datasets using in situ data (Pulliainen et al., 2020; Chapter 2, Section 2.3.2.5; Chapter 9, Section 33 9.5.2). 34 35 The value of gravity-based estimates of changes in ice sheet mass has increased as the time series from the 36 GRACE and GRACE-FO satellites, homogenised and absolutely calibrated, is close to 20 years in length. 37 The ESA’s Cryosat-2 radar altimetry satellite mission has continued to provide measurements of the changes 38 in the thickness of sea ice and the elevation of the Greenland and Antarctic Ice Sheets (Tilling et al., 2018). 39 Other missions include NASA’s Operation IceBridge, collecting airborne remote sensing measurements to 40 bridge the gap between ICESat (Ice, Cloud and land Elevation Satellite) and the upcoming ICESat-2 laser 41 altimetry missions. Longer time series from multiple missions have led to considerable advances in 42 understanding the origin of inconsistencies between the mass balances of different glaciers and reducing 43 uncertainties in estimates of changes in the Greenland and Antarctic Ice Sheets (Bamber et al., 2018; 44 Shepherd et al., 2018a, 2020). Last, the first observed climatology of snowfall over Antarctica was obtained 45 using the cloud/precipitation radar onboard NASA’s CloudSat (Palerme et al., 2014). 46 47 Biosphere 48 49 Satellite observations have recently expanded to include data on the fluorescence of land plants as a measure 50 of photosynthetic activity via the Global Ozone Monitoring Experiment (Guanter et al., 2014; Yang et al., 51 2015) and OCO-2 satellites (Sun et al., 2017). Climate data records of Leaf Area Index (LAI), characterizing 52 the area of green leaves per unit of ground area, and the fraction of absorbed photosynthetically active 53 radiation (FAPAR) – an important indicator of photosynthetic activity and plant health (Gobron et al., 2009) 54 – are now available for over 30-years (Claverie et al., 2016). In addition, key indicators such as fire 55 disturbances/burned areas are now retrieved via satellite (Chuvieco et al., 2019). In the US, the National Do Not Cite, Quote or Distribute 1-75 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Ecological Observational Network (NEON) provides continental-scale observations relevant to the 2 assessment of changes in aquatic and terrestrial ecosystems via a wide variety of ground-based, airborne, and 3 satellite platforms (Keller et al., 2008). All these long-term records reveal range shifts in ecosystems 4 (Chapter 2, Section 2.3.4). 5 6 The ability to estimate changes in global land biomass has improved due to the use of different microwave 7 satellite data (Liu et al., 2015) and in situ forest census data and co-located lidar, combined with the 8 MODerate resolution Imaging Spectroradiometer (MODIS; Baccini et al., 2017). This has allowed for 9 improved quantification of land temperature (Duan et al., 2019), carbon stocks and human-induced changes 10 due to deforestation (Chapter 2, Section 2.2.7). Time series of Normalized Difference Vegetation Index 11 (NDVI) from MODIS and other remote sensing platforms is widely applied to assess the effects of climate 12 change on vegetation in drought-sensitive regions (Atampugre et al., 2019). New satellite imaging 13 capabilities for meteorological observations, such as the advanced multi-spectral imager aboard Himawari-8 14 (Bessho et al., 2016), also allow for improved monitoring of challenging quantities such as seasonal changes 15 of vegetation in cloudy regions (Miura et al., 2019; Chapter 2, section 2.3.4.3). 16 17 In the ocean, efforts are underway to coordinate observations of biologically-relevant EOVs around the 18 globe (Muller-Karger et al., 2018; Canonico et al., 2019) and to integrate observations across disciplines 19 (e.g., the Global Ocean Acidification Observing Network; Tilbrook et al., 2019). A large number of 20 coordinated field campaigns during the 2015/2016 El Niño event enabled the collection of short-lived 21 biological phenomena such as coral bleaching and mortality caused by a months-long ocean heatwave 22 (Hughes et al., 2018); beyond this event, coordinated observations of coral reef systems are increasing in 23 number and quality (Obura et al., 2019). Overall, globally coordinated efforts focused on individual 24 components of the biosphere (e.g., the Global Alliance of Continuous Plankton Recorder Surveys; Batten et 25 al., 2019) contribute to improved knowledge of the changing marine ecosystems (Chapter 2, Section 2.3.4.2). 26 27 Given widespread evidence for decreases in global biodiversity in recent decades related to climate change 28 and other forms of human disturbance (IPBES, 2019), a new international effort to identify a set of Essential 29 Biodiversity Variables is underway (Pereira et al., 2013; Navarro et al., 2017). 30 31 In summary, the observational coverage of ongoing changes to the climate system is improved at the time of 32 AR6, relative to what was available for AR5 (high confidence). 33 34 Paleoclimate 35 36 Major paleo reconstruction efforts completed since AR5 include a variety of large-scale, multi-proxy 37 temperature datasets and associated reconstructions spanning the last 2000 years (PAGES 2k Consortium, 38 2017, 2019; Neukom et al., 2019), the Holocene (Kaufman et al., 2020), the Last Glacial Maximum (Cleator 39 et al., 2020; Tierney et al., 2020b), the Mid-Pliocene Warm Period (McClymont et al., 2020), and the Early 40 Eocene Climate Optimum (Hollis et al., 2019). Newly compiled borehole data (Cuesta-Valero et al., 2019), 41 as well as advances in statistical applications to tree ring data, result in more robust reconstructions of key 42 indices such as Northern Hemisphere temperature over the last millennium (e.g., Wilson et al., 2016; 43 Anchukaitis et al., 2017). Such reconstructions provide a new context for recent warming trends (Chapter 2) 44 and serve to constrain the response of the climate system to natural and anthropogenic forcing (Chapters 3 45 and 7). 46 47 Ongoing efforts have expanded the number of large-scale, tree-ring-based drought reconstructions that span 48 the last centuries to millennium at annual resolution (Chapter 8; Cook et al., 2015; Stahle et al., 2016; 49 Aguilera-Betti et al., 2017; Morales et al., 2020). Likewise, stalagmite records of oxygen isotopes have 50 increased in number, resolution, and geographic distribution since AR5, providing insights into regional to 51 global-scale hydrological change over the last centuries to millions of years (Chapter 8; Cheng et al., 2016; 52 Denniston et al., 2016; Comas-Bru and Harrison, 2019). A new global compilation of water isotope-based 53 paleoclimate records spanning the last 2,000 years (PAGES Iso2K) lays the groundwork for quantitative 54 multi-proxy reconstructions of regional to global scale hydrological and temperature trends and extremes 55 (Konecky et al., 2020). Do Not Cite, Quote or Distribute 1-76 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Recent advances in the reconstruction of climate extremes beyond temperature and drought include 3 expanded datasets of past El Niño-Southern Oscillation extremes (Chapter 2, Section 2.4.2; e.g., Barrett et 4 al., 2018; Freund et al., 2019; Grothe et al., 2019) and other modes of variability (Hernández et al., 2020), 5 hurricane activity (e.g., Burn and Palmer, 2015; Donnelly et al., 2015), jet stream variability (Trouet et al., 6 2018), and wildfires (e.g., Taylor et al., 2016). 7 8 New datasets as well as recent data compilations and syntheses of sea level over the last millennia (Kopp et 9 al., 2016; Kemp et al., 2018), the last 20,000 years (Khan et al., 2019), the last interglacial period (Dutton et 10 al., 2015; Chapter 2, Section 2.3.3.3), and the Pliocene (Dumitru et al., 2019; Grant et al., 2019; Cross- 11 Chapter Box 2.4 in Chapter 2) help constrain sea level variability and its relationship to global and regional 12 temperature variability, and to contributions from different sources on centennial to millennial timescales 13 (Chapter 9, Section 9.6.2). 14 15 Reconstructions of paleoocean pH (Chapter 2, Section 2.3.3.5) have increased in number and accuracy, 16 providing new constraints on ocean pH across the last centuries (e.g., Wu et al., 2018), the last glacial cycles 17 (e.g., Moy et al., 2019), and the last several million years (e.g., Anagnostou et al., 2020). Such 18 reconstructions inform processes and act as benchmarks for Earth system models of the global carbon cycle 19 over the recent geologic past (Chapter 5, Section 5.3.1), including previous high-CO2 warm intervals such as 20 the Pliocene (Cross-Chapter Box 2.4 in Chapter 2). Particularly relevant to such investigations are 21 reconstructions of atmospheric CO2 (Hönisch et al., 2012; Foster et al., 2017) that span the past millions to 22 tens of millions of years. 23 24 Constraints on the timing and rates of past climate changes have improved since AR5. Analytical methods 25 have increased the precision and reduced sample-size requirements for key radiometric dating techniques 26 including radiocarbon (Gottschalk et al., 2018; Lougheed et al., 2018) and Uranium-Thorium dating (Cheng 27 et al., 2013). More accurate ages of many paleoclimate records are also facilitated by recent improvements in 28 the radiocarbon calibration datasets (IntCal20, Reimer et al., 2020). A recent compilation of global 29 cosmogenic nuclide-based exposure dates (Balco, 2020b) allows for a more rigorous assessment of the 30 evolution of glacial landforms since the Last Glacial Maximum (Balco, 2020a). 31 32 Advances in paleoclimate data assimilation (Chapter 10, Section 10.2.3.2) leverage the expanded set of 33 paleoclimate observations to create physically consistent gridded fields of climate variables for data-rich 34 intervals of interest (e.g., over the last millennium, Hakim et al. 2016) or last glacial period (Cleator et al., 35 2020; Tierney et al., 2020b). Such efforts mirror advances in our understanding of the relationship between 36 proxy records and climate variables of interest, as formalized in so-called proxy system models (e.g., 37 Tolwinski-Ward et al., 2011; Dee et al., 2015; Dolman and Laepple, 2018). 38 39 Overall, the number, temporal resolution, and chronological accuracy of paleoclimate reconstructions have 40 increased since AR5, leading to improved understanding of climate system processes (or Earth system 41 processes) (high confidence). 42 43 44 1.5.1.2 Threats to observational capacity or continuity 45 46 The lock-downs and societal outcomes arising due to the COVID-19 pandemic pose a new threat to 47 observing systems. For example, WMO and UNESCO-IOC published a summary of the changes to Earth 48 system observations during COVID-19 (WMO, 2020b). Fewer aircraft flights (down 75–90% in May 2020, 49 depending on region) and ship transits (down 20% in May 2020) mean that onboard observations from those 50 networks have reduced in number and frequency (James et al., 2020; Ingleby et al., 2021). Europe has 51 deployed more radiosonde soundings to account for the reduction in data from air traffic. Fewer ocean 52 observing buoys were deployed during 2020, and reductions have been particularly prevalent in the tropics 53 and Southern Hemisphere. The full consequences of the pandemic and responses will come to light over 54 time. Estimates of the effect of the reduction in aircraft data assimilation on weather forecasting skill are 55 small (James et al., 2020; Ingleby et al., 2021), potentially alleviating concerns about veracity of future Do Not Cite, Quote or Distribute 1-77 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 atmospheric reanalyses of the COVID-19 pandemic period. 2 3 Surface-based networks have reduced in their coverage or range of variables measured due to COVID-19 4 and other factors. Over land, several factors, including the ongoing transition from manual to automatic 5 observations of weather, have reduced the spatial coverage of certain measurement types including rainfall 6 intensity, radiosonde launches and pan evaporation, posing unique risks to datasets used for climate 7 assessment (WMO, 2017; Lin and Huybers, 2019). Ship-based measurements, which are important for ocean 8 climate and reanalyses through time (Smith et al., 2019c), have been in decline due to the number of ships 9 contributing observations. There has also been a decline in the number of variables recorded by ships, but an 10 increase in the quality and time-resolution of others (e.g., sea level pressure, Kent et al. 2019). 11 12 Certain satellite frequencies are used to detect meteorological features that are vital to climate change 13 monitoring. These can be disturbed by certain radio communications (Anterrieu et al., 2016), although 14 scientists work to remove noise from the signal (Oliva et al., 2016). For example, water vapour in the 15 atmosphere naturally produces a weak signal at 23.8 gigahertz, which is within the range of frequencies of 16 the 5G communications network (Liu et al., 2021). Concern has been raised about potential leakage from 5G 17 network transmissions into the operating frequencies of passive sensors on existing weather satellites, which 18 could adversely influence their ability to remotely observe water vapour in the atmosphere (Yousefvand et 19 al., 2020). 20 21 Threats to observational capacity also include the loss of natural climate archives that are disappearing as a 22 direct consequence of warming temperatures. Ice core records from vulnerable alpine glaciers in the tropics 23 (Permana et al., 2019) and the mid-latitudes (Gabrielli et al., 2016; Winski et al., 2018; Moreno et al., 2021) 24 document more frequent melt layers in recent decades, with glacial retreat occurring at a rate and geographic 25 scale that is unusual in the Holocene (Solomina et al., 2015). The scope and severity of coral bleaching and 26 mortality events have increased in recent decades (Hughes et al., 2018), with profound implications for the 27 recovery of coral climate archives from new and existing sites. An observed increase in the mortality of 28 larger, long-lived trees over the last century is attributed to a combination of warming, land use change, and 29 disturbance (e.g., McDowell et al., 2020). The ongoing loss of these natural, high-resolution climate archives 30 endanger an end in their coverage over recent decades, given that many of the longest monthly- to annually- 31 resolved paleoclimate records were collected in the 1960s to 1990s (e.g., the PAGES2K database as 32 represented in PAGES 2k Consortium, 2017). This gap presents a barrier to the calibration of existing 33 decades-to-centuries-long records needed to constrain past temperature and hydrology trends and extremes. 34 35 Historical archives of weather and climate observations contained in ship’s logs, weather diaries, observatory 36 logbooks and other sources of documentary data are also at jeopardy of loss from natural disasters or 37 accidental destruction. These include measurements of temperature (air and sea surface), rainfall, surface 38 pressure, wind strength and direction, sunshine amount, and many other variables back into the 19th century. 39 While internationally coordinated data rescue efforts are focused on recovering documentary sources of past 40 weather and climate data (e.g., Allan et al., 2011), no such coordinated efforts exist for vulnerable 41 paleoclimate archives. Furthermore, oral traditions about local and regional weather and climate from 42 indigenous peoples represent valuable sources of information, especially when used in combination with 43 instrumental climate data (Makondo and Thomas, 2018), but are in danger of being lost as indigenous 44 knowledge-holders pass away. 45 46 In summary, while the quantity, quality, and diversity of climate system observations have grown since AR5, 47 the loss or potential loss of several critical components of the observational network is also evident (high 48 confidence). 49 50 51 1.5.2 New developments in reanalyses 52 53 Reanalyses are usually the output of a model (e.g., a numerical weather prediction model) constrained by 54 observations using data assimilation techniques, but the term has also been used to describe observation- 55 based datasets produced using simpler statistical methods and models (see Annex I). This section focuses on Do Not Cite, Quote or Distribute 1-78 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 the model-based methods and their recent developments. 2 3 Reanalyses complement datasets of observations in describing changes through the historical record and are 4 sometimes considered as ‘maps without gaps’ because they provide gridded output in space and time, often 5 global, with physical consistency across variables on sub-daily timescales, and information about sparsely- 6 observed variables (such as evaporation) (Hersbach et al., 2020). They can be globally complete, or 7 regionally focussed and constrained by boundary conditions from a global reanalysis (Chapter 10, Section 8 10.2.1.2). They can also provide feedback about the quality of the observations assimilated, including 9 estimates of biases and critical gaps for some observing systems. 10 11 Many early reanalyses are described in Box 2.3 of Hartmann et al. (2013). These were often limited by the 12 underlying model, the data assimilation schemes and observational issues (Thorne and Vose, 2010; Zhou et 13 al., 2018). Observational issues include the lack of underlying observations in some regions, changes in the 14 observational systems over time (e.g., spatial coverage, introduction of satellite data), and time-dependent 15 errors in the underlying observations or in the boundary conditions, which may lead to stepwise biases in 16 time. The assimilation of sparse or inconsistent observations can introduce mass or energy imbalances 17 (Valdivieso et al., 2017; Trenberth et al., 2019). Further limitations and some efforts to reduce the 18 implications of these observational issues will be detailed below. 19 20 The methods used in the development of reanalyses have progressed since AR5 and, in some cases, this has 21 important implications for the information they provide on how the climate is changing. Annex I includes a 22 list of reanalysis datasets used in the AR6. Recent major developments in reanalyses include the assimilation 23 of a wider range of observations, higher spatial and temporal resolution, extensions further back in time, and 24 greater efforts to minimise the influence of a temporally varying observational network. 25 26 Atmospheric reanalyses 27 Extensive improvements have been made in global atmospheric reanalyses since AR5. The growing demand 28 for high-resolution data has led to the development of higher-resolution atmospheric reanalyses, such as the 29 Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA2; Gelaro et al., 30 2017) and ERA5 (Hersbach et al., 2020). There is a focus on ERA5 here because it has been assessed as of 31 high enough quality to present temperature trends alongside more traditional observational datasets (Chapter 32 2, Section 2.3.1.1) and is also used in the Interactive Atlas. 33 34 Atmospheric reanalyses that were assessed in AR5 are still being used in the literature, and results from 35 ERA-Interim (Dee et al., 2011, ~80 km resolution, production stopped in August 2019), the Japanese 55-year 36 Reanalysis (JRA-55) (Ebita et al., 2011; Kobayashi et al., 2015; Harada et al., 2016) and Climate Forecast 37 System Reanalysis (CFSR) (Saha et al., 2010) are assessed in AR6. Some studies still also use the 38 NCEP/NCAR reanalysis, particularly because it extends back to 1948 and is updated in near real-time 39 (Kistler et al., 2001). Older reanalyses have a number of limitations, which have to be accounted for when 40 assessing the results of any study that uses them. 41 42 ERA5 provides hourly atmospheric fields at about 31 km resolution on 137 levels in the vertical, as well as 43 land surface variables and ocean waves, and is available from 1979 onwards and is updated in near real-time, 44 with plans to extend back to 1950. A 10-member ensemble is also available at coarser resolution, allowing 45 uncertainty estimates to be provided (e.g., Chapter 2, Section 2.3). MERRA-2 includes many updates from 46 the earlier version, including the assimilation of aerosol observations, several improvements to the 47 representation of the stratosphere, including ozone, and improved representations of cryospheric processes. 48 All of these improvements increase the usefulness of these reanalyses (Hoffmann et al., 2019; Chapter 7, 49 Section 7.3). 50 51 Models of atmospheric composition and emission sources and sinks allow the forecast and reanalysis of 52 constituents such as O3, CO, NOx and aerosols. The Copernicus Atmosphere Monitoring Service (CAMS) 53 reanalysis shows improvement against earlier atmospheric composition reanalyses, giving greater confidence 54 for its use to study trends and evaluate models (e.g., Inness et al., 2019; Chapter 7, Section 7.3). 55 Do Not Cite, Quote or Distribute 1-79 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 The inter-comparison of reanalyses with each other, or with earlier versions, is often done for particular 2 variables or aspects of the simulation. ERA5 is assessed as the most reliable reanalysis for climate trend 3 assessment (Chapter 2, Section 2.3). Compared to ERA-Interim, the ERA5 forecast model and assimilation 4 system, as well as the availability of improved reprocessing of observations, resulted in relatively smaller 5 errors when compared to observations, including a better representation of global energy budgets, radiative 6 forcing from volcanic eruptions (e.g., Mt. Pinatubo: Allan et al., 2020), the partitioning of surface energy 7 (Martens et al., 2020) and wind (Kaiser-Weiss et al., 2015, 2019; Borsche et al., 2016; Scherrer, 2020). In 8 ERA5, higher resolution means a better representation of Lagrangian motion convective updrafts, gravity 9 waves, tropical cyclones, and other meso- to synoptic-scale features of the atmosphere (Hoffmann et al., 10 2019; Martens et al., 2020). Low-frequency variability is found to be generally well represented and, from 10 11 hPa downwards, patterns of anomalies in temperature match those from the ERA-Interim, MERRA-2 and 12 JRA-55 reanalyses. Inhomogeneities in the water cycle have also been reduced (Hersbach et al., 2020). 13 14 Precipitation is not usually assimilated in reanalyses and, depending on the region, reanalysis precipitation 15 can differ from observations by more than the observational error (Zhou and Wang, 2017; Sun et al., 2018; 16 Alexander et al., 2020; Bador et al., 2020), although these studies did not include ERA5. Assimilation of 17 radiance observations from microwave imagers which, over ice-free ocean surfaces, improve the analysis of 18 lower-tropospheric humidity, cloud liquid water and ocean surface wind speed have resulted in improved 19 precipitation outputs in ERA5 (Hersbach et al., 2020). Global averages of other fields, particularly 20 temperature, from ERA-Interim and JRA-55 reanalyses continue to be consistent over the last 20 years with 21 surface observational data sets that include the polar regions (Simmons and Poli, 2015), although biases in 22 precipitation and radiation can influence temperatures regionally (Zhou et al., 2018). The global average 23 surface temperature from MERRA-2 is far cooler in recent years than temperatures derived from ERA- 24 Interim and JRA-55, which may be due to the assimilation of aerosols and their interactions (see Chapter 2, 25 Section 2.3). 26 27 A number of regional atmospheric reanalyses (see Chapter 10, Section 10.2.1.2) have been developed, such 28 as COSMO-REA (Wahl et al., 2017), and the Australian Bureau of Meteorology Atmospheric high- 29 resolution Regional Reanalysis for Australia (BARRA) (Su et al., 2019). Regional reanalyses can add value 30 to global reanalyses due to the lower computational requirements, and can allow multiple numerical weather 31 prediction models to be tested (e.g., Kaiser-Weiss et al., 2019). There is some evidence that these higher 32 resolution reanalyses better capture precipitation variability than global lower resolution reanalyses (Jermey 33 and Renshaw, 2016; Cui et al., 2017) and are further assessed in Chapter 10, Section 10.2.1.2 and used in the 34 Interactive Atlas. 35 36 In summary, the improvements in atmospheric reanalyses, and the greater number of years since the routine 37 ingestion of satellite data began relative to AR5, mean that there is increased confidence in using 38 atmospheric reanalyses products alongside more standard observation-based datasets in AR6 (high 39 confidence). 40 41 Sparse input reanalyses of the instrumental era 42 43 Although reanalyses such as ERA5 take advantage of new observational datasets and present a great 44 improvement in atmospheric reanalyses, the issues introduced by the evolving observational network remain. 45 Sparse input reanalyses, where only a limited set of reliable and long observed records are assimilated, 46 address these issues, with the limitation of fewer observational constraints. These efforts are sometimes 47 called centennial-scale reanalyses. One example is the atmospheric 20th Century Reanalysis (Compo et al., 48 2011; Slivinski et al., 2021) which assimilates only surface and sea-level pressure observations, and is 49 constrained by time-varying observed changes in atmospheric constituents, prescribed sea surface 50 temperatures and sea ice concentration, creating a reconstruction of the weather over the whole globe every 3 51 hours for the period 1806–2015. The ERA-20C atmospheric reanalysis (covering 1900–2010; Poli et al., 52 2016) also assimilates marine wind observations, and CERA-20C is a centennial-scale reanalysis that 53 assimilates both atmospheric and oceanic observations for the 1901–2010 period (Laloyaux et al., 2018). 54 These centennial-scale reanalyses are often run as ensembles that provide an estimate of the uncertainty in 55 the simulated variables over space and time. Slivinski et al. (2021) conclude that the uncertainties in surface Do Not Cite, Quote or Distribute 1-80 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 circulation fields in version 3 of the 20th Century Reanalysis are reliable and that there is also skill in its 2 tropospheric reconstruction over the 20th century. Long-term changes in other variables, such as 3 precipitation, also agree well with direct observation-based datasets (Chapter 2, Section 2.3.1.3; Chapter 8, 4 Section 8.3.2.8). 5 6 Ocean reanalyses 7 8 Since AR5, ocean reanalyses have improved due to: increased model resolution (Zuo et al., 2017; Lellouche 9 et al., 2018; Heimbach et al., 2019); improved physics (Storto et al., 2019); improvements in the atmospheric 10 forcing from atmospheric reanalyses (see preceding subsection); and improvements in the data quantity and 11 quality available for assimilation (e.g., Lellouche et al., 2018; Heimbach et al., 2019), particularly due to 12 Argo observations (Zuo et al., 2019) (see Annex I). 13 14 The first Ocean Reanalyses Intercomparison project (Balmaseda et al., 2015) focussed on the uncertainty in 15 key climate indicators, such as ocean heat content (Palmer et al., 2017), thermosteric sea level (Storto et al., 16 2017, 2019), salinity (Shi et al., 2017), sea ice extent (Chevallier et al., 2017), and the AMOC (Karspeck et 17 al., 2017). Reanalysis uncertainties occur in areas of inhomogeneous or sparse observational data sampling, 18 such as for the deep ocean, the Southern Ocean and western boundary currents (Lellouche et al., 2018; Storto 19 et al., 2019). Intercomparisons have also been dedicated to specific variables such as mixed-layer depths 20 (Toyoda et al., 2017), eddy kinetic energy (Masina et al., 2017) of the polar regions (Uotila et al., 2019). 21 Karspeck et al. (2017) found disagreement in the Atlantic meridional overturning circulation (AMOC)` 22 variability and strength in reanalyses over observation-sparse periods, whereas Jackson et al. (2019) reported 23 a lower spread in AMOC strength across an ensemble of ocean reanalyses of the recent period (1993-2010) 24 linked to improved observation availability for assimilation. Reanalyses also have a larger spread of ocean 25 heat uptake than data-only products and can produce spurious overestimates of heat uptake (Palmer et al., 26 2017), which is important in the context of estimating climate sensitivity (Storto et al., 2019). The ensemble 27 approach for ocean reanalyses provides another avenue for estimating uncertainties across ocean reanalyses 28 (Storto et al., 2019). 29 30 While there are still limitations in their representation of oceanic features, ocean reanalyses add value to 31 observation-only based products and are used to inform assessments in AR6 (Chapters 2, 3, 7 and 9). 32 Reanalyses of the atmosphere or ocean alone may not account for important atmosphere-ocean coupling, 33 motivating the development of coupled reanalyses (Laloyaux et al., 2018; Schepers et al., 2018; Penny et al., 34 2019), but these are not assessed in AR6. 35 36 Reanalyses of the pre-instrumental era 37 38 Longer reanalyses that extend further back in time than the beginning of the instrumental record are being 39 developed. They include the complete integration of paleoclimate archives and newly available early 40 instrumental data into extended reanalysis datasets. Such integration leverages ongoing development of 41 climate models that can simulate paleoclimate records in their units of analysis (i.e., oxygen isotope 42 composition, tree ring width, etc.), in many cases using physical climate variables as input for so-called 43 ‘proxy system models’ (Evans et al., 2013; Dee et al., 2015). Ensemble Kalman filter data assimilation 44 approaches allow to combine paleoclimate data and climate model data to generate annually resolved fields 45 (Last Millenium Reanalysis, Hakim et al., 2016; Tardif et al., 2018) or even monthly fields (Franke et al., 46 2017). This allows for a greater understanding of decadal variability (Parsons and Hakim, 2019) and greater 47 certainty around the full range of the frequency and severity of climate extremes, allowing for better-defined 48 detection of change. It also helps to identify the links between biogeochemical cycles, ecosystem structure 49 and ecosystem functioning, and to provide initial conditions for further model experiments or downscaling 50 (see Chapter 2). 51 52 Applications of reanalyses 53 54 The developments in reanalyses described above mean that they are now used across a range of applications. 55 In AR6, reanalyses provide information for fields and in regions where observations are limited. There is Do Not Cite, Quote or Distribute 1-81 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 growing confidence that modern reanalyses can provide another line of evidence in describing recent 2 temperature trends (see Chapter 2, Tables 2.4 and 2.5) As their spatial resolution increases, the exploration of 3 fine-scale extremes in both space and time becomes possible (e.g., wind; Kaiser-Weiss et al., 2015). Longer 4 reanalyses can be used to describe the change in the climate over the last 100 to 1000 years. Reanalyses have 5 been used to help post-process climate model output, and drive impact models, however, they are often bias 6 adjusted first (e.g., Weedon et al., 2014). See Cross-Chapter Box 10.2 in Chapter 10. Copernicus Climate 7 Change Service (C3S) provides a bias adjusted dataset for global land areas based on ERA5 called WFDE5 8 (Cucchi et al., 2020) which, combined with ERA5 information over the ocean (W5E5; Lange, 2019), is used 9 as the AR6 Interactive Atlas reference for the bias adjustment of model output. 10 11 The growing interest in longer-term climate forecasts (from seasonal to multi-year and decadal) means that 12 reanalyses are now more routinely being used to develop the initial state for these forecasts, such as for the 13 Decadal Climate Prediction Project (DCPP; Boer et al., 2016). Ocean reanalyses are now being used 14 routinely in the context of climate monitoring, (e.g., the Copernicus Marine Environment Monitoring Service 15 Ocean State Report; von Schuckmann et al., 2019). 16 17 In summary, reanalyses have improved since AR5 and can increasingly be used as a line of evidence in 18 assessments of the state and evolution of the climate system (high confidence). Reanalyses provide 19 consistency across multiple physical quantities, and information about variables and locations that are not 20 directly observed. Since AR5, new reanalyses have been developed with various combinations of increased 21 resolution, extended records, more consistent data assimilation, estimation of uncertainty arising from the 22 range of initial conditions, and an improved representation of the atmosphere or ocean system. While noting 23 their remaining limitations, the WGI Report uses the most recent generation of reanalysis products alongside 24 more standard observation-based datasets. 25 26 27 1.5.3 Climate Models 28 29 A wide range of numerical models are widely used in climate science to study the climate system and its 30 behaviour across multiple temporal and spatial scales. These models are the main tools available to look 31 ahead into possible climate futures under a range of scenarios (see Section 1.6). Global Earth System Models 32 (ESMs) are the most complex models which contribute to AR6. At the core of each ESM is a GCM (General 33 Circulation Model) representing the dynamics of the atmosphere and ocean. ESMs are complemented by 34 regional models (see Chapter 10, Section 10.3.1) and by a hierarchy of models of lower complexity. This 35 section summarizes major developments in these different types of models since AR5. Past IPCC reports 36 have made use of multi-model ensembles generated through various phases of the World Climate Research 37 Programme (WCRP) Coupled Model Intercomparison Project (CMIP). Analysis of the latest CMIP Phase 6 38 (CMIP6, Eyring et al., 2016) simulations constitute a key line of evidence supporting this assessment report 39 (see Section 1.5.4). The key characteristics of models participating in CMIP6 are listed in Annex II. 40 41 42 1.5.3.1 Earth System Models 43 44 Earth system models are mathematical formulations of the natural laws that govern the evolution of climate- 45 relevant systems: atmosphere, ocean, cryosphere, land, and biosphere, and the carbon cycle (Flato, 2011). 46 They build on the fundamental laws of physics (e.g., Navier-Stokes or Clausius-Clapeyron equations) or 47 empirical relationships established from observations and, when possible, constrained by fundamental 48 conservation laws (e.g., mass and energy). The evolution of climate-relevant variables is computed 49 numerically using high performance computers (André et al., 2014; Balaji et al., 2017), on three-dimensional 50 discrete grids (Staniforth and Thuburn, 2012). The spatial (and temporal) resolution of these grids in both the 51 horizontal and vertical directions determines which processes need to be parameterised or whether they can 52 be explicitly resolved. Developments since AR5 in model resolution, parameterizations and modelling of the 53 land and ocean biosphere and of biogeochemical cycles are discussed below. 54 55 Model grids and resolution Do Not Cite, Quote or Distribute 1-82 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 The horizontal resolution and the number of vertical levels in ESMs is generally higher in CMIP6 than in 3 CMIP5 (Figure 1.19). Global models with finer horizontal grids represent better many aspects of the 4 circulation of the atmosphere (Gao et al., 2020; Schiemann et al., 2020) and ocean (Bishop et al., 2016; 5 Storkey et al., 2018), bringing improvements in the simulation of the global hydrological cycle (Roberts et 6 al., 2018). CMIP6 includes a dedicated effort (HighResMIP, Haarsma et al., 2016) to explore the effect of 7 higher horizontal resolution, such as ~50 km, ~25 km and even ~10 km (see 1.5.4.2 and Annex II, Table 8 AII.6). Improvements are documented in the highest resolution coupled models used for HighResMip 9 (Hewitt et al., 2017b; Roberts et al., 2019). Flexible grids allowing spatially variable resolution are more 10 widely used than at the time of the AR5 in the atmosphere (McGregor, 2015; Giorgetta et al., 2018) and in 11 the ocean (Wang et al., 2014; Petersen et al., 2019). 12 13 The number of vertical levels in the atmosphere of global models has increased (Figure 1.19) partly to enable 14 simulations to include higher levels in the atmosphere and better represent stratospheric processes (Charlton- 15 Perez et al., 2013; Kawatani et al., 2019). Half the modelling groups now use ‘high top’ models with a top 16 level above the stratopause (a pressure of about 1 hPa). The number of vertical levels in the ocean models 17 has also increased in order to achieve finer resolution over the water column and especially in the upper 18 mixed layer, and better resolve the diurnal cycle (Bernie et al., 2008) (see Chapter 3, Section 3.5 and Annex 19 II). 20 21 Despite the documented progress of higher resolution, the model evaluation carried out in subsequent 22 chapters shows that improvements between CMIP5 and CMIP6 remain modest at the global scale (Bock et 23 al., 2020; Chapter 3, Section 3.8.2). Lower resolution alone does not explain all model biases, for example, a 24 low blocking frequency (Davini and D’Andrea, 2020) or a wrong shape of the Intertropical Convergence 25 Zone (Tian and Dong, 2020). Model performance depends on model formulation and parameterizations as 26 much as on resolution (Chapter 3, Chapter 8, Chapter 10). 27 28 29 [START FIGURE 1.19 HERE] 30 31 Figure 1.19: Resolution of the atmospheric and oceanic components of global climate models participating in 32 CMIP5, CMIP6, and HighResMIP: (a) (b) horizontal resolution (km), and (c) (d) number of vertical 33 levels. Darker colour circles indicate high-top models (whose top of the atmosphere is above 50 km). The 34 crosses are the median values. These models are documented in Annex II. Note that duplicated models in 35 a modelling group are counted as one entry when their horizontal and vertical resolutions are same. For 36 HighResMIP, one atmosphere-ocean coupled model with the highest resolution from each modelling 37 group is used. The horizontal resolution (rounded to 10km) is the square root of the number of grid points 38 divided by the surface area of the Earth, or the number of surface ocean grid points divided by the area of 39 the ocean surface, for the atmosphere and ocean respectively. 40 41 [END FIGURE 1.19 HERE] 42 43 44 Representation of physical and chemical processes in ESMs 45 Atmospheric models include representations of physical processes such as clouds, turbulence, convection 46 and gravity waves that are not fully represented by grid-scale dynamics. The CMIP6 models have undergone 47 updates in some of their parameterization schemes compared to their CMIP5 counterparts, with the aim of 48 better representing the physics and bringing the climatology of the models closer to newly available 49 observational datasets. Most notable developments are to schemes involving radiative transfer, cloud 50 microphysics, and aerosols, in particular a more explicit representation of the aerosol indirect effects through 51 aerosol-induced modification of cloud properties. Broadly, aerosol-cloud microphysics has been a key topic 52 for the aerosol and chemistry modelling communities since AR5, leading to improved understanding of the 53 climate influence of short-lived climate forcers, but they remain the single largest source of spread in ESM 54 calculations of climate sensitivity (Meehl et al., 2020), with numerous parameterization schemes in use 55 (Gettelman and Sherwood, 2016; Zhao et al., 2018; Gettelman et al., 2019). See also Chapter 6, section 6.4. 56 The treatment of droplet size and mixed-phase clouds (liquid and ice) was found to lead to changes in Do Not Cite, Quote or Distribute 1-83 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 climate sensitivity (Annex VII: Glossary) of some models between AR5 and AR6 (Bodas-Salcedo et al., 2 2019; Gettelman et al., 2019; Zelinka et al., 2020, Chapter 7, Section 7.4). 3 4 The representation of ocean and cryosphere processes has also evolved significantly since CMIP5. The 5 explicit representation of ocean eddies, due to increased grid resolution (typically, from 1° to ¼°), is a major 6 advance in a number of CMIP6 ocean model components (Hewitt et al., 2017b). Advances in sea ice models 7 have been made, for example, through correcting known shortcomings in CMIP5 simulations, in particular 8 the persistent underestimation of the rapid decline in summer Arctic sea ice extent (Rosenblum and 9 Eisenman, 2016, 2017; Turner and Comiso, 2017; Notz and Stroeve, 2018). The development of glacier and 10 ice-sheet models has been motivated and guided by an improved understanding of key physical processes, 11 including grounding line dynamics, stratigraphy and microstructure evolution, sub-shelf melting, and glacier 12 and ice-shelf calving, among others (Faria et al., 2014, 2018; Hanna et al., 2020). The resolution of ice sheet 13 models has continuously increased, including the use of nested grids, sub-grid interpolation schemes, and 14 adaptive mesh approaches (Cornford et al., 2016), mainly for a more accurate representation of grounding- 15 line migration and data assimilation (Pattyn, 2018). Ice-sheet models are increasingly interactively coupled 16 with global and regional climate models, accounting for the height mass-balance feedback (Vizcaino et al., 17 2015; Le clec’h et al., 2019), and enabling a better representation of ice-ocean processes, in particular for the 18 Antarctic Ice Sheet (Asay-Davis et al., 2017). 19 20 Sea level rise is caused by multiple processes acting on multiple time scales: ocean warming, glaciers and ice 21 sheet melting, change in water storage on land, glacial isostatic adjustment (Chapter 9, Box 9.1) but no 22 single model can represent all these processes (Chapter 9, Section 9.6). In this report, the contributions are 23 computed separately (Chapter 9, Figure 9.28) and merged into a common probabilistic framework and 24 updated from AR5 (Church et al., 2013; Kopp et al., 2014; Chapter 9, Section 9.6). 25 26 Another notable development since AR5 is the inclusion of stochastic parameterizations of sub-grid 27 processes in some comprehensive climate models (Sanchez et al., 2016). Here, the deterministic differential 28 equations that govern the dynamical evolution of the model are complemented by knowledge of the 29 stochastic variability in unresolved processes. While not yet widely implemented, the approach has been 30 shown to improve the forecasting skill of weather models, to reduce systematic biases in global models 31 (Berner et al., 2017; Palmer, 2019) and to influence simulated climate sensitivity (Strommen et al., 2019). 32 33 Representation of biogeochemistry, including the carbon cycle 34 Since AR5, more sophisticated land use and land cover change representations in ESMs have been 35 developed to simulate the effects of land management on surface fluxes of carbon, water and energy 36 (Lawrence et al., 2016), although the integration of many processes (e.g., wetland drainage, fire as a 37 management tool) remains a challenge (Pongratz et al., 2018). The importance of nitrogen availability to 38 limit the terrestrial carbon sequestration has been recognised (Zaehle et al., 2014; Chapter 5, Section 5.4) and 39 so an increasing number of models now include a prognostic representation of the terrestrial nitrogen cycle 40 and its coupling to the land carbon cycle (Jones et al., 2016a; Arora et al., 2020), leading to a reduction in 41 uncertainty for carbon budgets (Jones and Friedlingstein, 2020; Chapter 5, Section 5.1). As was the case in 42 CMIP5 (Ciais et al., 2013), the land surface processes represented vary across CMIP6 models, with at least 43 some key processes (fire, permafrost carbon, microbes, nutrients, vegetation dynamics, plant demography) 44 absent from any particular ESM land model (Chapter 5, Table 5.4). Ocean biogeochemical models have 45 evolved to enhance the consistency of the exchanges between ocean, atmosphere and land, through riverine 46 input and dust deposition (Stock et al., 2014; Aumont et al., 2015). Other developments include flexible 47 plankton stoichiometric ratios (Galbraith and Martiny, 2015), improvements in the representation of nitrogen 48 fixation (Paulsen et al., 2017), and the limitation of plankton growth by iron (Aumont et al., 2015). Due to 49 the long time scale of biogeochemical processes, spin-up strategies have been shown to affect the 50 performance of models used in AR5 (Séférian et al., 2016). 51 52 53 1.5.3.2 Model tuning and adjustment 54 55 When developing climate models, choices have to be made in a number of areas. Besides model formulation Do Not Cite, Quote or Distribute 1-84 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 and resolution, parameterizations of unresolved processes also involve many choices as, for each of these, 2 several parameters can be set. The acceptable range for these parameters is set by mathematical consistency 3 (e.g., convergence of a numerical scheme), physical considerations (e.g., energy conservation), observations, 4 or a combination of factors. Model developers choose a set of parameters that both falls within this range and 5 mimics observations of individual processes or their statistics. 6 7 An initial set of such choices is usually made by (often extensive) groups of modellers working on individual 8 components of the Earth system (e.g., ocean, atmosphere, land or sea ice). As components are assembled to 9 build an ESM, the choices are refined so that the simulated climate best represents a number of pre-defined 10 climate variables, or ‘tuning targets’. When these are met the model is released for use in intercomparisons 11 such as CMIP. Tuning targets can be one of three types: mean climate, regional phenomena and features, and 12 historical trends (Hourdin et al., 2017). One example of such a goal is that the climate system should reach a 13 mean equilibrium temperature close to observations when energy received from the sun is close to its 14 observed value. Whether tuning should be performed to accurate simulating long-term trends such as 15 changes in global mean temperature over the historical era, or rather be performed for each process 16 independently such that all collective behaviour is emergent, is an open question (Schmidt et al., 2017; 17 Burrows et al., 2018). 18 19 Each modelling group has its own strategy and, after AR5, a survey was conducted to understand the tuning 20 approach used in 23 CMIP5 modelling centres. The results are discussed in Hourdin et al. (2017), which 21 stresses that the behaviour of ESMs depends on the tuning strategy. An important recommendation is that the 22 calibration steps that lead to particular model tuning should be carefully documented. In CMIP6 each 23 modelling group now describes the three levels of tuning, both for the complete ESM and for the individual 24 components (available at https://explore.es-doc.org/ and in the published model descriptions, Annex II). The 25 most important global tuning target for CMIP6 models is the net top-of-the-atmosphere (TOA) heat flux and 26 its radiative components. Other global targets include: the decomposition of each of these TOA fluxes into a 27 clear sky component and a component due to the radiative effect of clouds, global mean air and ocean 28 temperature, sea ice extent, sea ice volume, glacial mass balance, global root mean square error of 29 precipitation. The TOA heat flux balance is achieved using a diversity of approaches, usually unique to each 30 modelling group. Adjustments are made for parameters associated with uncertain or poorly constrained 31 processes (Schmidt et al., 2017), for example the aerosol indirect effects, adjustments to ocean albedo, 32 marine dimethyl sulfide (DMS) parameterization, or cloud properties (Mauritsen and Roeckner, 2020). 33 34 Regional tuning targets include the meridional overturning circulation in the Atlantic Ocean, the Southern 35 Ocean circulation and temperature profiles in ocean basins (Golaz et al., 2019; Sellar et al., 2019); regional 36 land properties and precipitations (Mauritsen et al., 2019; Yukimoto et al., 2019), latitudinal distribution of 37 radiation (Boucher et al., 2020), spatial contrasts in TOA radiative fluxes or surface fluxes, and stationary 38 waves in the Northern Hemisphere (Schmidt et al., 2017; Yukimoto et al., 2019). 39 40 Even with some core commonalities of approaches to model tuning, practices can differ, such as the use of 41 initial drift from initialized forecasts, the explicit use of the transient observed record for the historical 42 period, or the use of the present-day radiative imbalance at the TOA as a tuning target rather than an 43 equilibrated pre-industrial balance. The majority of CMIP6 modelling groups report that they do not tune 44 their model for the observed trends during the historical period (23 out of 29), nor for equilibrium climate 45 sensitivity (25 out of 29). ECS and TCR are thus emergent properties for a large majority of models. The 46 effect of tuning on model skill and ensemble spread in CMIP6 is further discussed in Chapter 3, Section 3.3. 47 48 49 1.5.3.3 From global to regional models 50 51 The need for accurate climate information at the regional scale is increasing (Chapter 10, Section 10.1). 52 High-resolution global climate models, such as those taking part in HighResMIP, provide more detailed 53 information at the regional scale (Roberts et al., 2018). However, due to the large computational resources 54 required by these models, only a limited number of simulations per model are available. In addition to CMIP 55 global models, regional information can be derived using Regional Climate Models (RCMs) and Do Not Cite, Quote or Distribute 1-85 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 downscaling techniques, presented in Chapter 10 and the Atlas. RCMs are dynamical models similar to 2 GCMs that simulate a limited region and are forced with boudary conditions from a global simulation, often 3 correcting for biases (Chapter 10, Section 10.3 and Cross-Chapter Box 10.2, Annex II). This approach allows 4 the use of a higher resolution within the chosen domain, and thus better represent important drivers of 5 regional climate such as mountain ranges, land management and urban effects. RCMs resolving atmospheric 6 convection explicitly are now included in intercomparisons (Coppola et al., 2020) and used in Chapters 10, 7 11 and 12. Other approaches are also used to generate regional climate projections, such as statistical 8 downscaling (Maraun and Widmann, 2018; Chapter 10, Section 10.3). 9 10 The number of climate centres or consortia that carry out global climate simulations and projections has 11 grown from 11 in the first CMIP to 19 in CMIP5 and 28 for CMIP6 (see Section 1.5.4.2 and Annex II). 12 Regional climate models participating in the Coordinated Regional Downscaling Experiment (CORDEX) are 13 more diverse than the global ESMs (see Section 1.5.4.3 and Annex II) and engage an even wider 14 international community (Figure 1.20). 15 16 17 [START FIGURE 1.20 HERE] 18 19 Figure 1.20: A world map showing the increased diversity of modelling centres contributing to CMIP and 20 CORDEX. Climate models are often developed by international consortia. EC-Earth is shown as an 21 example (involving SMHI, Sweden; KNMI, The Netherlands; DMI, Denmark; AEMET, Spain; Met 22 Éireann, Ireland; CNR‐ISAC, Italy; Instituto de Meteorologia, Portugal; FMI, Finland), but there are too 23 many such collaborations to display all of them on this map. More complete information about 24 institutions contributing to CORDEX and CMIP6 is found in Annex II. 25 26 [END FIGURE 1.20 HERE] 27 28 29 1.5.3.4 Models of lower complexity 30 31 Earth System Models of Intermediate Complexity (EMICs) complement the model hierarchy and fill the 32 gap between conceptual, simple climate models and complex GCMs or ESMs (Claussen et al., 2002). 33 EMICs are simplified; they include processes in a more parameterized, rather than explicitly calculated, form 34 and generally have lower spatial resolution compared to the complex ESMs. As a result, EMICs require 35 much less computational resource and can be integrated for many thousands of years without 36 supercomputers (Hajima et al., 2014). The range of EMICs used in climate change research is highly 37 heterogeneous, ranging from zonally averaged or mixed-layer ocean models coupled to statistical-dynamical 38 models of the atmosphere to low-resolution 3-dimensional ocean models coupled to simplified dynamical 39 models of the atmosphere. An increasing number of EMICs include interactive representations of the global 40 carbon cycle, with varying levels of complexity and numbers of processes considered (Plattner et al., 2008; 41 Zickfeld et al., 2013; MacDougall et al., 2020). Given the heterogeneity of the EMIC community, modelers 42 tend to focus on specific research questions and develop individual models accordingly. As for any type of 43 models assessed in this report, the set of EMICs undergoes thorough evaluation and fit-for-purpose testing 44 before being applied to address specific climate aspects. 45 46 EMICs have been used extensively in past IPCC reports, providing long-term integrations on paleoclimate 47 and future timescales, including stabilization pathways and a range of commitment scenarios, with perturbed 48 physics ensembles and sensitivity studies, or with simulations targeting the uncertainty in global climate- 49 carbon cycle systems (e.g., Meehl et al., 2007; Collins et al., 2013). More recently, a number of studies have 50 pointed to the possibility of systematically different climate responses to external forcings in EMICs and 51 complex ESMs (Frölicher and Paynter, 2015; Pfister and Stocker, 2017, 2018) that need to be considered in 52 the context of this report. For example, Frölicher and Paynter (2015) showed that EMICs have a higher 53 simulated realized warming fraction (i.e., the TCR/ECS ratio) than CMIP5 ESMs and speculated that this 54 may bias the temperature response to zero carbon emissions. But, in a recent comprehensive multi-model 55 analysis of the zero CO2 emissions commitment, MacDougall et al. (2020) did not find any significant Do Not Cite, Quote or Distribute 1-86 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 differences in committed temperatures 90 years after halting emissions between EMICs and ESMs. While 2 some EMICs contribute to parts of the CMIP6-endorsed MIPs, a coordinated EMICs modeling effort similar 3 to the ones for the AR4 (Plattner et al., 2008) and AR5 (Eby et al., 2013; Zickfeld et al., 2013) is not in place 4 for IPCC AR6; however, EMICs are assessed in a number of chapters. For example, Chapters 4 and 5 use 5 EMICs in the assessment of long-term climate change beyond 2100 (Chapter 5, Section 5.5), zero-emission 6 commitments, overshoot and recovery (Chapter 4, Section 4.7), consequences of carbon dioxide removal 7 (CDR) on the climate system and the carbon cycle (Chapter 4, Sections 4.6 and Chapter 5, Section 5.6) and 8 long-term carbon cycle–climate feedbacks (Chapter 5, Section 5.4). 9 10 Physical emulators and simple climate models make up a broad class of heavily parametrized models 11 designed to reproduce the responses of the more complex, process-based models, and provide rapid 12 translations of emissions, via concentrations and radiative forcing, into probabilistic estimates of changes to 13 the physical climate system. The main application of emulators is to extrapolate insights from ESMs and 14 observational constraints to a larger set of emission scenarios (see Cross-Chapter Box 7.1 in Chapter 7). The 15 computational efficiency of various emulating approaches opens new analytical possibilities given that 16 ESMs take a lot of computational resources for each simulation. The applicability and usefulness of 17 emulating approaches are however constrained by their skill in capturing the global mean climate responses 18 simulated by the ESMs (mainly limited to global-mean or hemispheric land/ocean temperatures) and by their 19 ability to extrapolate skilfully outside the calibrated range. 20 21 The terms emulator and simple climate model (SCM) are different, although they are sometimes used 22 interchangeably. SCM refers to a broad class of lower-dimensional models of the energy balance, radiative 23 transfer, carbon cycle, or a combination of such physical components. SCMs can also be tuned to reproduce 24 the calculations of climate-mean variables of a given ESM, assuming that their structural flexibility can 25 capture both the parametric and structural uncertainties across process-oriented ESM responses. When run in 26 this setup, they are termed emulators. Simple climate models do not have to be run in ‘emulation’ mode, 27 though, as they can also be used to test consistency across multiple lines of evidence with regard to ranges in 28 ECS, TCR, TCRE and carbon cycle feedbacks (see Chapters 5 and 7). Physical emulation can also be 29 performed with very simple parameterisations (‘one-or-few-line climate models’), statistical methods like 30 neural networks, genetic algorithms, or other artificial intelligence approaches, where the emulator behaviour 31 is explicitly tuned to reproduce the response of a given ESM or model ensemble (Chapters 4, 5, and 7). 32 33 Current emulators and SCMs include the generic impulse response model outlined in Chapter 8 of the AR5 34 (AR5-IR (Supplementary Material 8.SM.11 of Myhre et al. (2013)), two-layer models (Held et al., 2010; 35 Rohrschneider et al., 2019; Nicholls et al., 2020), and higher complexity approaches that include upwelling, 36 diffusion and entrainment in the ocean component (e.g., MAGICC Version 5.3 (Raper et al., 2001; Wigley et 37 al., 2009), Version 6/7 (Meinshausen et al., 2011a); OSCAR (Gasser et al., 2017); CICERO SCM (Skeie et 38 al., 2017); FaIR (Millar et al., 2017b; Smith et al., 2018); and a range of statistical approaches (Schwarber et 39 al., 2019; Beusch et al., 2020b)). An example of recent use of an emulator approach is an early estimate of 40 the climate implications of the COVID-19 lockdowns (Forster et al. 2020; see Cross-Chapter Box 6.1 in 41 Chapter 6). 42 43 Since AR5, simplified climate models have been developed further, and their use is increasing. Different 44 purposes motivating development include: being as simple as possible for teaching purposes (e.g., a two- 45 layer energy balance model), as comprehensive as possible to allow for propagation of uncertainties across 46 multiple Earth System domains (MAGICC and others), or focus on higher complexity representation of 47 specific domains (e.g., OSCAR). The common theme in many models is to improve parameterisations that 48 reflect the latest findings in complex ESM interactions, such as the nitrogen cycle addition to the carbon 49 cycle, or tropospheric and stratospheric ozone exchange, with the aim of emulating their global mean 50 temperature response. Also, within the simple models that have a rudimentary representation of spatial 51 heterogeneity (e.g., four-box simple climate models), the ambition is to represent heterogeneous forcers such 52 as black carbon more adequately (Stjern et al., 2017), provide an appropriate representation of the forcing- 53 feedback framework (see e.g., Sherwood et al., 2015), investigate new parameterisations of ocean heat 54 uptake, and implement better representations of volcanic aerosol induced cooling (Gregory et al., 2016a). 55 Do Not Cite, Quote or Distribute 1-87 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 MAGICC (Wigley et al., 2009; Meinshausen et al., 2011a) and FaIR (Smith et al., 2018) were used in the 2 SR1.5 (IPCC, 2018) to categorize mitigation pathways into classes of scenarios that peak near 1.5°C, 3 overshoot 1.5°C, or stay below 2°C. The SR1.5 (Rogelj et al., 2018) concluded that there was a high 4 agreement in the relative temperature response of pathways, but medium agreement on the precise absolute 5 magnitude of warming, introducing a level of imprecision in the attribution of a single pathway to a given 6 category. 7 8 In this Report, there are two notable uses of simple climate models. One is the connection between the 9 assessed range of ECS in Chapter 7, and the projections of future global surface air temperature (GSAT) 10 change in Chapter 4, which is done via a two-layer model based on Held et al. (2010). It is also used as input 11 to sea level projections in Chapter 9. The other usage is the transfer of Earth system assessment knowledge 12 to Working Group III, via a set of models (MAGICC, FaIR, CICERO-SCM) specifically tuned to represent 13 the Working Group I assessment. For an overview of the uses, and an assessment of the related Reduced 14 Complexity Model Intercomparison Project (RCMIP), see Nicholls et al. (2020) and Cross-Chapter Box 7.1 15 in Chapter 7. 16 17 18 [START BOX 1.3 HERE] 19 20 Box 1.3: Emission metrics in AR6 WGI 21 Emission metrics compare the radiative forcing, temperature change, or other climate effects arising from 22 emissions of CO2 versus those from emissions of non-CO2 radiative forcing agents (such as CH4 or N2O). 23 They have been discussed in the IPCC since the First Assessment Report and are used as a means of 24 aggregating emissions and removals of different gases and placing them on a common (‘CO2 equivalent’, or 25 ‘CO2-eq’) scale. 26 AR5 included a thorough assessment of common pulse emission metrics, and how these address various 27 indicators of future climate change (Myhre et al., 2013). Most prominently used are the Global Warming 28 Potentials (GWPs), which integrate the calculated radiative forcing contribution following an idealized pulse 29 (or one-time) emission, over a chosen time horizon (IPCC, 1990a), or the Global Temperature-change 30 Potential (GTP), which considers the contribution of emission to the global-mean temperature at a specific 31 time after emission. Yet another metric is the Global Precipitation change Potential (GPP), used to quantify 32 the precipitation change per unit mass of emission of a given forcing agent (Shine et al., 2015). 33 As an example of usage, the Paris Rulebook [Decision 18/CMA.1, annex, paragraph 37] states that ‘Each 34 Party shall use the 100-year time-horizon global warming potential (GWP) values from the IPCC Fifth 35 Assessment Report, or 100-year time-horizon GWP values from a subsequent IPCC assessment report as 36 agreed upon by the ‘Conference of the Parties serving as the meeting of the Parties to the Paris Agreement’ 37 (CMA), to report aggregate emissions and removals of GHGs, expressed in CO2-eq. Each Party may in 38 addition also use other metrics (e.g., global temperature potential) to report supplemental information on 39 aggregate emissions and removals of GHGs, expressed in CO2-eq’. 40 Since AR5, improved knowledge of the radiative properties, lifetimes, and other characteristics of emitted 41 species, and the response of the climate system, have led to updates to the numerical values of a range of 42 metrics; see Chapter 7, Table 7.15. Another key development is a set of metrics that compare a pulse 43 emission of CO2 (as considered by GWP and GTP) to step-changes of emission rates for short-lived 44 components (i.e., also considering emission trends). Termed GWP* (which also includes a pulse component) 45 and Combined Global Temperature change Potential (CGTP), these metrics allow the construction of a near- 46 linear relationship between global surface temperature change and cumulative CO2 and CO2-equivalent 47 emissions of both short and long lived forcing agents (Allen et al., 2016; Cain et al., 2019; Collins et al., 48 2019). For example, the temperature response to a sustained methane reduction has a similar behaviour to the 49 temperature response to a pulse CO2 removal (or avoided emission). 50 In this Report, recent scientific developments underlying emission metrics, as relevant for Working Group I, Do Not Cite, Quote or Distribute 1-88 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 are assessed in full in Chapter 7, Section 7.6. In particular, see Box 7.3, which discusses the choice of metric 2 for different usages, and Section 7.6.1, which treats the challenge of comparing the climate implication of 3 emissions of short-lived and long-lived compounds. Also, the choice of metric is of key importance when 4 defining and quantifying net zeronet-zero greenhouse gas emissions; see Box 1.4 and Chapter 7, Section 5 7.6.2. Chapter 6 applies metrics to attribute GSAT change to short-lived climate forcer (SLCF) and long- 6 lived greenhouse gas (LLGHG) emissions from different sectors and regions (Section 6.6.2). 7 The metrics assessed in this Report are also used, and separately assessed, by Working Group III. See Cross- 8 Chapter Box 2: GHG emissions metrics and Annex B in Chapter 2 of the WGIII contribution to the AR6. 9 [END BOX 1.3 HERE] 10 11 12 1.5.4 Modelling techniques, comparisons and performance assessments 13 14 Numerical models, however complex, cannot be a perfect representation of the real world. Results from 15 climate modelling simulations constitute a key line of evidence for the present report, which requires 16 considering the limitations of each model simulation. This section presents recent developments in 17 techniques and approaches to robustly extract, quantify and compare results from multiple, independent 18 climate models, and how their performance can be assessed and validated. 19 20 21 1.5.4.1 Model ‘fitness for purpose’ 22 23 A key issue addressed in this report is whether climate models are adequate or ‘fit’ for purposes of interest, 24 that is, whether they can be used to successfully answer particular research questions, especially about the 25 causes of recent climate change and the future evolution of climate (e.g., Parker, 2009; Notz, 2015; Knutti, 26 2018; Winsberg, 2018). Assessment of a model’s fitness-for-purpose can be informed both by how the 27 model represents relevant physical processes and by relevant performance metrics (Baumberger et al., 2017; 28 Parker, 2020). The processes and metrics that are most relevant can vary with the question of interest, for 29 example, a question about changes in deep ocean circulation versus a question about changes in regional 30 precipitation (Notz, 2015; Gramelsberger et al., 2020). New model evaluation tools (Section 1.5.4.5) and 31 emergent constraint methodologies (Section 1.5.4.7) can also aid the assessment of fitness-for-purpose, 32 especially in conjunction with process understanding (Klein and Hall, 2015; Knutti, 2018). The broader 33 availability of large model ensemble may allow for novel tests of fitness that better account for natural 34 climate variability (see Section 1.5.4.2). Fitness-for-purpose of models used in this report is discussed in 35 Chapter 3 (Section 3.8.4) for the global scale, in Chapter 10 (Section 10.3) for regional climate, and in the 36 other chapters at the process level. 37 38 Typical strategies for enhancing the fitness-for-purpose of a model include increasing resolution in order to 39 explicitly simulate key processes, improving relevant parameterizations, and careful tuning. Changes to a 40 model that enhance its fitness for one purpose can sometimes decrease its fitness for others, by upsetting a 41 pre-existing balance of approximations. When it is unclear whether a model is fit for a purpose of interest, 42 there is often a closely-related purpose for which the evidence of fitness is clearer; for example, it might be 43 unclear whether a model is fit for providing highly accurate projections of precipitation changes in a region, 44 but reasonable to think that the model is fit for providing projections of precipitation changes that cannot yet 45 be ruled out (Parker, 2009). Such information about plausible or credible changes can be useful to inform 46 adaptation. Note that challenges associated with assessing model fitness-for-purpose need not prevent 47 reaching conclusions with high confidence if there are multiple other lines of evidence supporting those 48 same conclusions. 49 50 51 1.5.4.2 Ensemble modelling techniques 52 53 A key approach in climate science is the comparison of results from multiple model simulations with each Do Not Cite, Quote or Distribute 1-89 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 other and against observations. These simulations have typically been performed by separate models with 2 consistent boundary conditions and prescribed emissions or radiative forcings, as in the Coupled Model 3 Intercomparison Project phases (CMIP, Meehl et al., 2000, 2007a; Taylor et al., 2012; Eyring et al., 2016). 4 Such multi-model ensembles (MMEs) have proven highly useful in sampling and quantifying model 5 uncertainty, within and between generations of climate models. They also reduce the influence on 6 projections of the particular sets of parametrizations and physical components simulated by individual 7 models. The primary usage of MMEs is to provide a well quantified model range, but when used carefully 8 they can also increase confidence in projections (Knutti et al., 2010). Presently, however, many models also 9 share provenance (Masson and Knutti, 2011) and may have common biases that should be acknowledged 10 when presenting and building on MME-derived conclusions (Boé, 2018; Abramowitz et al., 2019) (see 11 Section 1.5.4.6). 12 13 Since AR5, an increase in computing power has made it possible to investigate simulated internal variability 14 and to provide robust estimates of forced model responses, using Large Initial Condition Ensembles (ICEs), 15 also referred to as Single Model Initial condition Large Ensembles (SMILEs). Examples using GCMs or 16 ESMs that support assessments in AR6 include the CESM Large Ensemble (Kay et al., 2015), the MPI 17 Grand Ensemble (Maher et al., 2019), and the CanESM2 large ensembles (Kirchmeier-Young et al., 2017). 18 Such ensembles employ a single GCM or ESM in a fixed configuration, but starting from a variety of 19 different initial states. In some experiments, these initial states only differ slightly. As the climate system is 20 chaotic, such tiny changes in initial conditions lead to different evolutions for the individual realizations of 21 the system as a whole. Other experiments start from a set of well-separated ocean initial conditions to sample 22 the uncertainty in the circulation state of the ocean and its role in longer-timescale variations. These two 23 types of ICEs have been referred to as ‘micro’ and ‘macro’ perturbation ensembles respectively (Hawkins et 24 al., 2016). In support of this report, most models contributing to CMIP6 have produced ensembles of 25 multiple realizations of their historical and scenario simulations (see Chapters 3 and 4). 26 27 Recently, the ICE technique has been extended to atmosphere-only simulations (Mizuta et al., 2017), single- 28 forcer influences such as volcanic eruptions (Bethke et al., 2017) to regional modelling (Mote et al., 2015; 29 Fyfe et al., 2017; Schaller et al., 2018; Leduc et al., 2019) and to attribution of extreme weather events using 30 crowd-sourced computing (climateprediction.net; Massey et al., 2015). 31 32 ICEs can also be used to evaluate climate model parameterizations, if models are initialized appropriately 33 (Phillips et al., 2004; Williams et al., 2013), mostly within the framework of seamless weather and climate 34 predictions (e.g., Palmer et al., 2008; Hurrell et al., 2009; Brown et al., 2012). Initializing an atmospheric 35 model in hindcast mode and observing the biases as they develop permits testing of the parameterized 36 processes, by starting from a known state rather than one dominated by quasi-random short term variability 37 (Williams et al., 2013; Ma et al., 2014; Vannière et al., 2014). However, single-model initial-conditions 38 ensembles cannot cover the same degrees of freedom as a multi-model ensemble, because model 39 characteristics substantially affect model behaviour (Flato et al., 2013). 40 41 A third common modelling technique is the perturbed parameter ensemble (PPE; note that the abbreviation 42 also sometimes refers to the sub-category ‘perturbed physics ensemble’). These methods are used to assess 43 uncertainty based on a single model, with individual parameters perturbed to reflect the full range of their 44 uncertainty (Murphy et al., 2004; Knutti et al., 2010; Lee et al., 2011; Shiogama et al., 2014). Statistical 45 methods can then be used to detect which parameters are the main causes of uncertainty across the ensemble. 46 PPEs have been used frequently in simpler models, such as EMICs, and are being applied to more complex 47 models. A caveat of PPEs is that the estimated uncertainty will depend on the specific parameterizations of 48 the underlying model and may well be an underestimation of the ‘true’ uncertainty. It is also challenging to 49 disentangle forced responses from internal variability using a PPE alone. 50 51 Together, the three ensemble methods (MMEs, ICEs, PPEs) allow investigation of climate model uncertainty 52 arising from internal variability, initial and internal boundary conditions, model formulations and 53 parameterizations (Parker, 2013). Figure 1.21 illustrates the different ensemble types. Recent studies have 54 also started combining multiple ensemble types or using ensembles in combination with statistical analytical 55 techniques. For example, Murphy et al. (2018) combine MMEs and PPEs to give a fuller assessment of Do Not Cite, Quote or Distribute 1-90 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 modelling uncertainty. Wagman and Jackson (2018) use PPEs to evaluate the robustness of MME-based 2 emergent constraints. Sexton et al. (2019) study the robustness of ICE approaches by identifying parameters 3 and processes responsible for model errors at the two different timescales. 4 5 Overall, we assess that increases in computing power and the broader availability of larger and more varied 6 ensembles of model simulations have contributed to better estimations of uncertainty in projections of future 7 change (high confidence). Note, however, that despite their widespread use in climate science today, the cost 8 of the ensemble approach in human and computational resources, and the challenges associated with the 9 interpretation of multi-model ensembles, has been questioned (Palmer and Stevens, 2019; Touzé-Peiffer et 10 al., 2020). 11 12 13 [START FIGURE 1.21 HERE] 14 15 Figure 1.21: Illustration of common types of model ensemble, simulating the time evolution of a quantity Q 16 (such as global mean surface temperature). (a) Multi-model ensemble, where each model has its own 17 realization of the processes affecting Q, and its own internal variability around the baseline value (dashed 18 line). The multi-model mean (black) is commonly taken as the ensemble average. (b) Initial condition 19 ensemble, where several realizations from a single model are compared. These differ only by minute 20 (‘micro’) perturbations to the initial conditions of the simulation, such that over time, internal variability 21 will progress differently in each ensemble member. (c) Perturbed physics ensemble, which also compares 22 realizations from a single model, but where one or more internal parameters that may affect the 23 simulations of Q are systematically changed to allow for a quantification of the effects of those quantities 24 on the model results. Additionally, each parameter set may be taken as the starting point for an initial 25 condition ensemble. In this figure, each set has three ensemble members. 26 27 [END FIGURE 1.21 HERE] 28 29 30 1.5.4.3 The sixth phase of the Coupled Model Intercomparison Project (CMIP6) 31 32 The Coupled Model Intercomparison Project (CMIP) provides a framework to compare the results of 33 different GCMs or ESMs performing similar experiments. Since its creation in the mid-1990s, it has evolved 34 in different phases, involving all major climate modelling centres in the world (Figure 1.20). The results of 35 these phases have played a key role in previous IPCC reports, and the present Report assesses a range of 36 results from CMIP5 that were not published until after the AR5, as well as the first results of the 6th phase of 37 CMIP (CMIP6) (Eyring et al., 2016). The CMIP6 experiment design is somewhat different from previous 38 phases. It now consists of a limited set of DECK (Diagnostic, Evaluation and Characterization of Klima) 39 simulations and an historical simulation that must be performed by all participating models, as well as a wide 40 range of CMIP6-endorsed Model Intercomparison Projects (MIPs) covering specialized topics (Eyring et al., 41 2016) (see Figure 1.22). Each MIP activity consists of a series of model experiments, documented in the 42 literature (see Table 1.3) and in an online database (https://es-doc.org, see Pascoe et al. (2019) and Annex II). 43 44 The CMIP DECK simulations form the basis for a range of assessments and projections in the following 45 chapters. As in CMIP5, they consist of a ‘pre-industrial’ control simulation (piControl, where ‘pre-industrial’ 46 is taken as fixed 1850 conditions in these experiments), an idealized, abrupt quadrupling of CO2 47 concentrations relative to piControl (to estimate equilibrium climate sensitivity), a 1% per year increase in 48 CO2 concentrations relative to piControl (to estimate the transient climate response), and a transient 49 simulation with prescribed sea-surface temperatures for the period 1979–2014 (termed ‘AMIP’ for historical 50 reasons). In addition, all participating models perform a historical simulation for the period 1850–2014. For 51 the latter, common CMIP6 forcings are prescribed (Cross-Chapter Box 1.4, Table 2). Depending on the 52 model setup, these include emissions and concentrations of short-lived species (Hoesly et al., 2018; Gidden 53 et al., 2019), long-lived greenhouse gases (Meinshausen et al., 2017), biomass burning emissions (van Marle 54 et al., 2017), global gridded land use forcing data (Ma et al., 2020a), solar forcing (Matthes et al., 2017), and 55 stratospheric aerosol data from volcanoes (Zanchettin et al., 2016). The methods for generating gridded 56 datasets are described in (Feng et al., 2019). For AMIP simulations, common sea surface temperatures Do Not Cite, Quote or Distribute 1-91 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 (SSTs) and sea ice concentrations (SICs) are prescribed. For simulations with prescribed aerosol abundances 2 (i.e., not calculated from emissions), optical properties and fractional changes in cloud droplet effective 3 radius are generally prescribed in order to provide a more consistent representation of aerosol forcing 4 relative to earlier CMIP phases (Fiedler et al., 2017; Stevens et al., 2017). For models without ozone 5 chemistry, time-varying gridded ozone concentrations and nitrogen deposition are also provided (Checa- 6 Garcia et al., 2018). 7 8 Beyond the DECK and the historical simulations, the CMIP6-endorsed MIPs aim to investigate how models 9 respond to specific forcings, their potential systematic biases, their variability, and their responses to detailed 10 future scenarios such as the Shared Socioeconomic Pathways (SSPs, Section 1.6). Table 1.3 lists the 23 11 CMIP6-endorsed MIPs and key references. Results from a range of these MIPs, and many others outside of 12 the most recent CMIP6 cycle, will be assessed in the following chapters (also shown in Table 1.3). 13 References to all the CMIP6 datasets used in the report are found in Annex II, Table AII.10. 14 15 16 [START FIGURE 1.22 HERE] 17 18 Figure 1.22: Structure of CMIP6, the 6th phase of the Coupled Model Intercomparison Project. The centre shows 19 the common DECK (Diagnostic, Evaluation and Characterization of Klima) and historical experiments 20 that all participating models must perform. The outer circles show the topics covered by the endorsed 21 (blue) and other MIPs (red). See Table 1.3 for explanation of the MIP acronyms. (Expanded from Eyring 22 et al., 2016). 23 24 [END FIGURE 1.22 HERE] 25 26 27 [START TABLE 1.3 HERE] 28 29 Table 1.3: CMIP6-Endorsed MIPs, their key references, and where they are used or referenced throughout this report. 30 CMIP6-Endorsed MIP Long name Key references Used in chapters name Aerosols and Chemistry AerChemMIP Model Intercomparison (Collins et al., 2017) 4, 6, Atlas Project Coupled Climate Carbon C4MIP Cycle Model (Jones et al., 2016a) 4, 5, Atlas Intercomparison Project The Carbon Dioxide CDRMIP Removal Model (Keller et al., 2018) 4, 5, Atlas Intercomparison Project Cloud Feedback Model CFMIP (Webb et al., 2017) 4, 7, Atlas Intercomparison Project Coordinated Regional (Gutowski Jr. et al., 4, 8, 9, 10, 11, 12, CORDEX Climate Downscaling 2016) Atlas Experiment Detection and DAMIP Attribution Model (Gillett et al., 2016) 3, 10, Atlas Intercomparison Project Decadal Climate DCPP (Boer et al., 2016) 4, 8, Atlas Prediction Project Dynamics and (Gerber and Manzini, DynVarMIP Variability Model Atlas 2016) Intercomparison Project Do Not Cite, Quote or Distribute 1-92 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI Flux-Anomaly-Forced FAFMIP Model Intercomparison (Gregory et al., 2016b) 9, Atlas Project Geoengineering Model GeoMIP (Kravitz et al., 2015) 4, 5, 8, 12, Atlas Intercomparison Project Global Monsoons Model GMMIP (Zhou et al., 2016) 2,3,4, 10, Atlas Intercomparison Project High Resolution Model 3, 8, 9, 10, 11, HighResMIP (Haarsma et al., 2016) Intercomparison Project Atlas Ice Sheet Model ISMIP6 Intercomparison Project (Nowicki et al., 2016) 3, 7, 9, Atlas for CMIP6 Land Surface, Snow and (van den Hurk et al., LS3MIP 3, 9, 11, Atlas Soil Moisture 2016) Land Use Model LUMIP (Lawrence et al., 2016) 4, 6, Atlas Intercomparison Project Ocean Model (Griffies et al., 2016; OMIP 3, 9, Atlas Intercomparison Project Orr et al., 2017) Polar Amplification PAMIP Model Intercomparison (Smith et al., 2019a) 10, Atlas Project (Haywood et al., 2016; Jungclaus et al., 2017; Paleoclimate Modelling 2, 3, 7, 8, 9, 10, PMIP Otto-Bliesner et al., Intercomparison Project Atlas 2017; Kageyama et al., 2018) Radiative Forcing Model RFMIP (Pincus et al., 2016) 6, 7, Atlas Intercomparison Project Scenario Model 4, 5, 6, 9, 10, 12, ScenarioMIP (O’Neill et al., 2016) Intercomparison Project Atlas Sea Ice Model SIMIP (Notz et al., 2016) 4, 9, 12, Atlas Intercomparison Project Vulnerability, Impacts, VIACS AB Adaptation and Climate (Ruane et al., 2016) 12, Atlas Services Advisory Board Volcanic Forcings VolMIP Model Intercomparison (Zanchettin et al., 2016) 4, 8, Atlas Project 1 2 [END TABLE 1.3 HERE] 3 4 5 1.5.4.4 Coordinated Regional Downscaling Experiment (CORDEX) 6 7 The Coordinated Regional Downscaling Experiment (CORDEX, Gutowski Jr. et al., 2016) is an 8 intercomparison project for regional models and statistical downscaling techniques, coordinating simulations 9 on common domains and under common experimental conditions in a similar way to the CMIP effort. 10 Dynamical and statistical downscaling techniques can provide higher-resolution climate information than is 11 available directly from global climate models (Chapter 10, Section 10.3). These techniques require 12 evaluation and quantification of their performance before they can be considered appropriate as usable 13 regional climate information or be used in support of climate services. CORDEX simulations have been 14 provided by a range of regional downscaling models, for 14 regions together covering much of the globe Do Not Cite, Quote or Distribute 1-93 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 (Atlas, Figure Atlas.7), and they are used extensively in the AR6 WGI Atlas (Atlas.1.4; see also Annex II). 2 3 In support of AR6, CORDEX has undertaken a new experiment (CORDEX-CORE) where regional climate 4 models downscale a common set of global model simulations, performed at a coarser resolution, to a spatial 5 resolution spanning from 12 to 25 km over most of the CORDEX domains (Atlas, Box Atlas.1). CORDEX- 6 CORE represents an improved level of coordinated intercomparison of downscaling models (Remedio et al., 7 2019). 8 9 10 1.5.4.5 Model Evaluation Tools 11 12 For the first time in CMIP, a range of comprehensive evaluation tools are now available that can run 13 alongside the commonly used distributed data platform Earth System Grid Federation (ESGF, see Annex II), 14 to produce comprehensive results as soon as the model output is published to the CMIP archive. 15 16 For instance, the Earth System Model Evaluation Tool (ESMValTool; Eyring et al., 2020; Lauer et al., 2020; 17 Righi et al., 2020) is used by a number of chapters. It is an open-source community software tool that 18 includes a large variety of diagnostics and performance metrics relevant for coupled Earth System processes, 19 such as for the mean, variability and trends, and it can also examine emergent constraints (see Section 20 1.5.4.7). ESMValTool also includes routines provided by the WMO Expert Team on Climate Change 21 Detection and Indices for the evaluation of extreme events (Min et al., 2011; Sillmann et al., 2013) and 22 diagnostics for key processes and variability. Another example of evaluation tool is the CLIVAR 2020 23 ENSO metrics package (Planton et al., 2021). 24 25 These tools are used in several chapters of this report for the creation of the figures that show CMIP results. 26 Together with the Interactive Atlas, they allow for traceability of key results, and an additional level of 27 quality control on whether published figures can be reproduced. It also provides the capability to update 28 published figures with, as much as possible, the same set of models in all figures, and to assess model 29 improvements across different phases of CMIP (Chapter 3, Section 3.8.2). 30 31 These new developments are facilitated by the definition of common formats for CMIP model output (Balaji 32 et al., 2018) and the availability of reanalyses and observations in the same format as CMIP output 33 (obs4MIPs, Ferraro et al., 2015). The tools are also used to support routine evaluation at individual model 34 centres and simplify the assessment of improvements of individual models or generations of model 35 ensembles (Eyring et al., 2019). Note, however, that while tools such as ESMValTool can produce an 36 estimate of overall model performance, dedicated model evaluation still needs to be performed when 37 analysing projections for a particular purpose, such as assessing changing hazards in a given particular 38 region. Such evaluation is discussed in the next section, and in greater detail in later chapters of this Report. 39 40 41 1.5.4.6 Evaluation of process-based models against observations 42 43 Techniques used for evaluating process-based climate models against observations were assessed in AR5 44 (Flato et al., 2013), and have progressed rapidly since (Eyring et al., 2019). The most widely used technique 45 is to compare climatologies (long-term averages of specific climate variables) or time series of simulated 46 (process-based) model output with observations, considering the observational uncertainty. A further 47 approach is to compare the results of process-based models with those from statistical models. In addition to 48 a comparison of climatological means, trends and variability, AR5 already made use of a large set of 49 performance metrics for a quantitative evaluation of the models. 50 51 Since AR5, a range of studies has investigated model agreement with observations well beyond large scale 52 mean climate properties (e.g., Bellenger et al., 2014; Covey et al., 2016; Pendergrass and Deser, 2017; 53 Goelzer et al., 2018; Beusch et al., 2020a), providing information on the performance of recent model 54 simulations across multiple variables and components of the Earth system (e.g., Anav et al., 2013; Guan and 55 Waliser, 2017). Based on such studies, this Report assesses model improvements across different CMIP Do Not Cite, Quote or Distribute 1-94 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 DECK, CMIP6 historical and CMIP6-Endorsed MIP simulations, and of differences in model performance 2 between different classes of models, such as high- versus low-resolution models (see e.g., Chapter 3, Section 3 3.8.2). 4 5 In addition, process- or regime-oriented evaluation of models has been expanded since AR5. By focusing on 6 processes, causes of systematic errors in the models can be identified and insights can be gained as to 7 whether a mean state or trend is correctly simulated for the right reasons. This approach is commonly used 8 for the evaluation of clouds (e.g., Williams and Webb, 2009; Konsta et al., 2012; Bony et al., 2015; Dal 9 Gesso et al., 2015; Jin et al., 2017), dust emissions (e.g., Parajuli et al., 2016; Wu et al., 2016) as well as 10 aerosol-cloud (e.g., Gryspeerdt and Stier, 2012) and chemistry-climate (SPARC, 2010) interactions. Process- 11 oriented diagnostics have also been used to evaluate specific phenomena such as the El Niño Southern 12 Oscillation (ENSO, Guilyardi et al., 2016), the Madden–Julian Oscillation (MJO; Ahn et al., 2017; Jiang et 13 al., 2018), Southern Ocean clouds (Hyder et al., 2018), monsoons (Boo et al., 2011; James et al., 2015), and 14 tropical cyclones (Kim et al., 2018). 15 16 Instrument simulators provide estimates of what a satellite would see if looking down on the model 17 simulated planet, and improve the direct comparison of modelled variables such as clouds, precipitation and 18 upper tropospheric humidity with observations from satellites (e.g., Kay et al., 2011; Klein et al., 2013; 19 Cesana and Waliser, 2016; Konsta et al., 2016; Jin et al., 2017; Chepfer et al., 2018; Swales et al., 2018; 20 Zhang et al., 2018). Within the framework of the Cloud Feedback Model Intercomparison Project (CFMIP) 21 contribution to CMIP6 (Webb et al., 2017), a new version of the Cloud Feedback Model Intercomparison 22 Project Observational Simulator (COSP, Swales et al., 2018) has been released which makes use of a 23 collection of observation proxies or satellite simulators. Related approaches in this rapidly evolving field 24 include simulators for Arctic Ocean observations (Burgard et al., 2020) and measurements of aerosol 25 observations along aircraft trajectories (Watson-Parris et al., 2019). 26 27 In this Report, model evaluation is performed in the individual chapters, rather than in a separate chapter as 28 was the case for AR5. This applies to the model types discussed above, and also to dedicated models of 29 subsystems that are not (or not yet) part of usual climate models, for example, glacier or ice sheet models 30 (Annex II). Further discussions are found in Chapter 3 (attribution), Chapter 5 (carbon cycle), Chapter 6 31 (short-lived climate forcers), Chapter 8 (water cycle), Chapter 9 (ocean, cryosphere and sea level), Chapter 32 10 (regional scale information) and the Atlas (regional models). 33 34 35 1.5.4.7 Emergent constraints on climate feedbacks, sensitivities and projections 36 37 An emergent constraint is the relationship between an uncertain aspect of future climate change and an 38 observable feature of the Earth System, evident across an ensemble of models (Allen and Ingram, 2002; 39 Mystakidis et al., 2016; Wenzel et al., 2016; Hall et al., 2019; Winkler et al., 2019). Complex Earth System 40 Models (ESMs) simulate variations on timescales from hours to centuries, telling us how aspects of the 41 current climate relate to its sensitivity to anthropogenic forcing. Where an ensemble of different ESMs 42 displays a relationship between a short-term observable variation and a longer-term sensitivity, an 43 observation of the short-term variation in the real world can be converted, via the model-based relationship, 44 into an ‘emergent constraint’ on the sensitivity. This is shown schematically in Figure 1.23 (Eyring et al., 45 2019), see also Annex VII: Glossary. 46 47 Emergent constraints use the spread in model projections to estimate the sensitivities of the climate system to 48 anthropogenic forcing, providing another type of ensemble-wide information that is not readily available 49 from simulations with one ESM alone. As emergent constraints depend on identifying those observable 50 aspects of the climate system that are most related to climate projections, they also help to focus model 51 evaluation on the most relevant observations (Hall et al., 2019). However, there is a chance that 52 indiscriminate data-mining of the multi-dimensional outputs from ESMs could lead to spurious correlations 53 (Caldwell et al., 2014; Wagman and Jackson, 2018) and less than robust emergent constraints on future 54 changes (Bracegirdle and Stephenson, 2013). To avoid this, emergent constraints need to be tested ‘out of 55 sample’ on parts of the dataset that were not included in its construction (Caldwell et al., 2018) and should Do Not Cite, Quote or Distribute 1-95 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 also always be based on sound physical understanding and mathematical theory (Hall et al., 2019). Their 2 conclusions should also be reassessed when a new generation of MMEs becomes available, such as CMIP6. 3 As an example, Chapter 7 (Section 7.5.4) discusses and assesses recent studies where equilibrium climate 4 sensitivities (ECS) diagnosed in a multi-model ensemble are compared with the same models’ estimates of 5 an observable quantity, such as post-1970s global warming or tropical sea-surface temperatures of past 6 climates like the last glacial maximum or the Pliocene. Assessments of other emergent constraints appear 7 throughout later chapters, such as Chapter 4 (Section 4.2.5), Chapter 5 (Section 5.4.6) and Chapter 7 (Section 8 7.5.4). 9 10 11 [START FIGURE 1.23 HERE] 12 13 Figure 1.23: The principle of emergent constraints. An ensemble of models (blue dots) defines a relationship 14 between an observable, mean, trend or variation in the climate (x-axis) and an uncertain projection, 15 climate sensitivity or feedback (y-axis). An observation of the x-axis variable can then be combined with 16 the model-derived relationship to provide a tighter estimate of the climate projection, sensitivity or 17 feedback on the y-axis (adapted from Eyring et al. 2019). 18 19 [END FIGURE 1.23 HERE] 20 21 22 1.5.4.8 Weighting techniques for model comparisons 23 24 Assessments of climate model ensembles have commonly assumed that each individual model is of equal 25 value (‘model democracy’) and when combining simulations to estimate the mean and variance of quantities 26 of interest, they are typically unweighted (Haughton et al., 2015). This practice has been noted to diminish 27 the influence of models exhibiting a good match with observations (Tapiador et al., 2020). However, 28 exceptions to this approach exist, notably AR5 projections of sea ice, which only selected a few models 29 which passed a model performance assessment (Collins et al., 2013), and more studies on this topic have 30 appeared since the AR5 (e.g., Eyring et al., 2019). Ensembles are typically sub-selected by removing either 31 poorly performing model simulations (McSweeney et al., 2015) or model simulations that are perceived to 32 add little additional information, typically where multiple simulations have come from the same model. They 33 may also be weighted based on model performance. 34 35 Several recent studies have attempted to quantify the effect of various strategies for selection or weighting of 36 ensemble members based on some set of criteria (Haughton et al., 2015; Olonscheck and Notz, 2017; 37 Sanderson et al., 2017). Model weighting strategies have been further employed since AR5 to reduce the 38 spread in climate projections for a given scenario by using weights based on one or more model performance 39 metrics (Wenzel et al., 2016; Knutti et al., 2017; Sanderson et al., 2017; Lorenz et al., 2018; Liang et al., 40 2020). However, models may share representations of processes, parameterization schemes, or even parts of 41 code, leading to common biases. The models may therefore not be fully independent, calling into question 42 inferences derived from multi-model ensembles (Abramowitz et al., 2019). Emergent constraints (see 43 Section 1.5.4.5) also represent an implicit weighting technique that explicitly links present performance to 44 future projections (Bracegirdle and Stephenson, 2013). 45 46 Concern has been raised about the large extent to which code is shared within the CMIP5 multi-model 47 ensemble (Sanderson et al., 2015a). Boé (2018) showed that a clear relationship exists between the number 48 of components shared by climate models and how similar the simulations are. The resulting similarities in 49 behaviour need to be accounted for in the generation of best-estimate multi-model climate projections. This 50 has led to calls to move beyond equally-weighted multi-model means towards weighted means that take into 51 account both model performance and model independence (Sanderson et al., 2015b, 2017; Knutti et al., 52 2017). Model independence has been defined in terms of performance differences within an ensemble 53 (Masson and Knutti, 2011; Knutti et al., 2013, 2017, Sanderson et al., 2015b, 2015a, 2017; Lorenz et al., 54 2018). However, this definition is sensitive to the choice of variable, observational data set, metric, time 55 period, and region, and a performance ranked ensemble has been shown to sometimes perform worse than a Do Not Cite, Quote or Distribute 1-96 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 random selection (Herger et al., 2018a). The adequacy of the constraint provided by the data and 2 experimental methods can be tested using a calibration-validation style partitioning of observations into two 3 sets (Bishop and Abramowitz, 2013), or a ‘perfect model approach’ where one of the ensemble members is 4 treated as the reference dataset and all model weights are calibrated against it (Bishop and Abramowitz, 5 2013; Wenzel et al., 2016; Knutti et al., 2017; Sanderson et al., 2017; Herger et al., 2018a, 2018b). Sunyer et 6 al. (2014) use a Bayesian framework to account for model dependencies and changes in model biases. Annan 7 and Hargreaves (2017) provides a statistical, quantifiable definition of independence that is independent of 8 performance-based measures. 9 10 The AR5 quantified uncertainty in CMIP5 climate projections by selecting one realization per model per 11 scenario, and calculating the 5–95% range of the resulting ensemble (see Chapter 4, Box 4.1) and the same 12 strategy is generally still used in AR6. Broadly, the following chapters take the CMIP6 5–95% ensemble 13 range as the likely uncertainty range for projections 8, with no further weighting or consideration of model 14 ancestry and as long as no universal, robust method for weighting a multi-model projection ensemble is 15 available (Box 4.1, Chapter 4). A notable exception to this approach is the assessment of future changes in 16 global surface air temperature (GSAT), which also draws on the updated best estimate and range of 17 equilibrium climate sensitivity assessed in Chapter 7. For a thorough description of the model weighting 18 choices made in this Report, and the assessment of GSAT, see Chapter 4 (Box 4.1). Model selection and 19 weighting in downscaling approaches for regional assessment is discussed in Chapter 10 (Section 10.3.4). 20 21 22 1.6 Dimensions of Integration: Scenarios, global warming levels and cumulative carbon emissions 23 24 This section introduces three ways to synthesize climate change knowledge across topics and chapters. These 25 ‘dimensions of integration’ include (1) emission and concentration scenarios underlying the climate change 26 projections assessed in this report, (2) levels of global mean surface warming relative to the 1850-1900 27 baseline (‘global warming levels’), and (3) cumulative carbon emissions (Figure 1.24). All three dimensions 28 can, in principle, be used to synthesize physical science knowledge across WGI, and also across climate 29 change impacts, adaptation, and mitigation research. Scenarios, in particular, have a long history of serving 30 as a common reference point within and across IPCC Working Groups and research communities. Similarly, 31 cumulative carbon emissions and global warming levels provide key links between WGI assessments and 32 those of the other WGs; these two dimensions frame the cause-effect chain investigated by WGI. The closest 33 links to WGIII are the emissions scenarios, as WGIII considers drivers of emissions and climate change 34 mitigation options. The links to WGII are the geophysical climate projections from the Earth System Models 35 which the climate impacts and adaptation literature often uses as their starting point. 36 37 38 [START FIGURE 1.24 HERE] 39 40 Figure 1.24: The Dimensions of Integration across Chapters and Working Groups in the IPCC AR6 assessment. 41 This report adopts three explicit dimensions of integration to integrate knowledge across chapters and 42 Working Groups. The first dimension is scenarios, the second dimension is global-mean warming levels 43 relative to 1850-1900, and the third dimension is cumulative CO2 emissions. For the scenarios, illustrative 44 2100 end-points are also indicated (white circles). Further details on data sources and processing are 45 available in the chapter data table (Table 1.SM.1). 46 47 [END FIGURE 1.24 HERE] 48 49 50 The section is structured as follows: first, the scenarios used in AR6 are introduced and discussed in relation 51 to scenarios used in earlier IPCC assessments (Section 1.6.1). Cross-Chapter Box 1.4 provides an overview 52 of the new scenarios and how they are used in this report. Next, the two additional dimensions of integration 53 are introduced: global warming levels (Section 1.6.2) and cumulative emissions (Section 1.6.3). Net zero 8 Note that the 5–95% is a very likely range (See Box 1.1 on the use of calibrated uncertainty language in AR6), though if this is purely a multi-model likelihood range, it is generally treated as likely, in absence of other lines of evidence Do Not Cite, Quote or Distribute 1-97 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 emissions are discussed in Box 1.4. The relation between global warming levels and scenarios is further 2 assessed in Cross-Chapter Box 11.1 in Chapter 11. 3 4 5 1.6.1 Scenarios 6 7 A scenario is a description of how the future may develop based on a coherent and internally consistent set 8 of assumptions about key drivers including demography, economic processes, technological innovation, 9 governance, lifestyles and relationships among these driving forces (IPCC, 2000; Rounsevell and Metzger, 10 2010; O’Neill et al., 2014; see Section 1.6.1.1). Scenarios can also be defined by geophysical driving forces 11 only, such as emissions or abundances of greenhouse gases, aerosols, and aerosol precursors or land use 12 patterns. Scenarios are not predictions; instead, they provide a ‘what-if’ investigation of the implications of 13 various developments and actions (Moss et al., 2010). WGI investigates potential future climate change 14 principally by assessing climate model simulations using emission scenarios originating from the WGIII 15 community (Section 1.6.1.2). The scenarios used in this WGI report cover various hypothetical ‘baseline 16 scenarios’ or ‘reference futures’ that could unfold in the absence of any or any additional climate policies 17 (see Annex VII: Glossary). These ‘reference scenarios’ originate from a comprehensive analysis of a wide 18 array of socio-economic drivers, such as population growth, technological development, and economic 19 development, and their broad spectrum of associated energy, land use and emission implications (Riahi et al., 20 2017). With direct policy relevance to the Paris Agreement’s 1.5°C and ‘well below’ 2°C goals, this report 21 also assesses climate futures where the effects of additional climate change mitigation action are explored, 22 i.e., so-called mitigation scenarios (for a broader discussion on scenarios and futures analysis, see Cross- 23 Chapter Box 1, Table 1 in SRCCL, IPCC, 2019b). 24 25 For this Report, the main emissions, concentration and land use scenarios considered are a subset of 26 scenarios recently developed using the Shared Socioeconomic Pathways framework (SSPs) (Riahi et al., 27 2017; see Section 1.6.1.1 and Cross-Chapter Box 1.4). Initially, the term ‘SSP’ described five broad 28 narratives of future socio-economic development only (O’Neill et al., 2014). However, at least in the WGI 29 community, the term ‘SSP scenario’ is now more widely used to refer directly to future emission and 30 concentration scenarios that result from combining these socio-economic development pathways with 31 climate change mitigation assumptions. These are assessed in detail in WGIII (WGIII, Chapter 3; Cross- 32 Chapter Box 1.4, Table 1). 33 34 The WGI report uses a core set of five SSP scenarios to assist cross-Chapter integration and cross-WG 35 applications: SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5 (Cross-Chapter Box 1.4, Table 1). 36 These scenarios span a wide range of plausible societal and climatic futures from potentially below 1.5°C 37 best-estimate warming to over 4°C warming by 2100 (Figure 1.25). The set of five SSP scenarios includes 38 those in ‘Tier 1’ simulations of the CMIP6 ScenarioMIP intercomparison project (O’Neill et al., 2016; see 39 Section 1.5.4) that participating climate modelling groups were asked to prioritize (SSP1-2.6, SSP2-4.5, 40 SSP3-7.0 and SSP5-8.5), plus the low emission scenario SSP1-1.9. SSP1-1.9 is used in combination with 41 SSP1-2.6 to explore differential outcomes of approximately 1.5 and 2.0 °C warming relative to pre-industrial 42 levels, relevant to the Paris Agreement goals. Further SSP scenarios are used in this report to assess specific 43 aspects, e.g., air pollution policies in Chapter 6 (Cross-Chapter Box 1.4). In addition, the previous generation 44 of Representative Concentration Pathways (RCPs) is also used in this report when assessing future climate 45 change (Section 1.6.1.3; Cross-Chapter Box 1.4, Table 1). 46 47 Climatic changes over the 21st century (and beyond) are projected and assessed in subsequent chapters, 48 using a broad range of climate models, conditional on the various SSP scenarios. The projected future 49 changes can then be put into the context of longer-term paleoclimate data and historical observations, 50 showing how the higher emission and higher concentration scenarios diverge further from the range of 51 climate conditions that ecosystems and human societies experienced in the past 2000 years in terms of global 52 mean temperature and other key climate variables (Figure 1.26; see also Figure 1.5). 53 54 55 [START FIGURE 1.25 HERE] Do Not Cite, Quote or Distribute 1-98 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 Figure 1.25: Global mean surface air temperature (GSAT) illustrated as warming stripes from blue (cold) to red 3 (warm) over three different time periods. From 1750 to 1850 based on PAGES 2K reconstructions 4 (2017, 2019); from 1850 to 2018 showing the composite GSAT time series assessed in Chapter 2; and 5 from 2020 onwards using the assessed GSAT projections for each Shared Socio-economic Pathway (SSP) 6 (from Chapter 4). For the projections, the upper end of each arrow aligns with colour corresponding to the 7 95th percentile of the projected temperatures and the lower end aligns with the colour corresponding to the 8 5th percentile of the projected temperature range. Projected temperatures are shown for five scenarios 9 from ‘very low’ SSP1-1.9 to ‘very high’ SSP5-8.5 (see Cross-Chapter Box 1.4 for more details on the 10 scenarios). For illustrative purposes, natural variability has been added from a single CMIP6 Earth system 11 model (MRI ESM2). The points in time when total CO2 emissions peak, reach halved levels of the peak 12 and reach net-zero emissions are indicated with arrows, ‘½’ and ‘0’ marks, respectively. Further details 13 on data sources and processing are available in the chapter data table (Table 1.SM.1). 14 15 16 [END FIGURE 1.25 HERE] 17 18 19 While scenarios are a key tool for integration across IPCC Working Groups, they also allow the integration 20 of knowledge among scientific communities and across timescales. For example, agricultural yield, 21 infrastructure and human health impacts of increased drought frequency, extreme rainfall events and 22 hurricanes are often examined in isolation. New insights on climate impacts in WGII can be gained if 23 compound effects of multiple cross-sectoral impacts are considered across multiple research communities 24 under consistent scenario frameworks (Leonard et al., 2014; Warszawski et al., 2014; see also Chapter 11, 25 Section 11.8). Similarly, a synthesis of WGI knowledge on sea level rise contributions is enabled by a 26 consistent application of future scenarios across all specialised research communities, such as ice-sheet mass 27 balance analyses, glacier loss projections and thermosteric change from ocean heat uptake (e.g. Kopp et al., 28 2014; see Chapter 9). 29 30 31 [START FIGURE 1.26 HERE] 32 33 Figure 1.26: Historical and projected future concentrations of CO2, CH4 and N2O and global mean surface 34 temperatures (GMST). GMST temperature reconstructions over the last 2000 years were compiled by 35 the PAGES 2k Consortium (2017, 2019) (grey line, with 95% uncertainty range), joined by historical 36 GMST timeseries assessed in Chapter 2 (black line) – both referenced against the 1850-1900 period. 37 Future GSAT temperature projections are from CMIP6 ESM models across all concentration-driven SSP 38 scenario projections (Chapter 4). The discontinuity around year 2100 for CMIP6 temperature projections 39 results from the fact that not all ESM models ran each scenario past 2100. The grey vertical band 40 indicates the future 2015-2300 period. The concentrations used to drive CMIP6 Earth System Models are 41 derived from ice core, firn and instrumental datasets (Meinshausen et al., 2017) and projected using an 42 emulator (Cross-Chapter Box 7.1 in Chapter 7; Meinshausen et al., 2020). The colours of the lines 43 indicate the SSP scenarios used in this report (see Cross-Chapter Box 1.4, Figure 1). Further details on 44 data sources and processing are available in the chapter data table (Table 1.SM.1). 45 46 [END FIGURE 1.26 HERE] 47 48 49 In addition to the comprehensive SSP scenario set and the RCPs, multiple idealized scenarios and time-slice 50 experiments using climate models are assessed in this report. Idealized scenarios refer to experiments where, 51 for example, CO2 concentrations are increased by 1% per year, or instantly quadrupled. Such idealized 52 experiments have been extensively used in previous model intercomparison projects and constitute the core 53 ‘DECK’ set of model experiments of CMIP6 (see Section 1.5.4). They are, for example, used to diagnose the 54 patterns of climate feedbacks across the suite of models assessed in this report (Chapter 7). 55 56 In the following, we further introduce the SSP scenarios and how they relate to the Shared Socioeconomic 57 Pathways framework (Section 1.6.1.1), describe the scenario generation process (Section 1.6.1.2), and Do Not Cite, Quote or Distribute 1-99 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 provide a historical review of scenarios used in IPCC assessment reports (Section 1.6.1.3), before briefly 2 discussing questions of scenario likelihood, scenario uncertainty and the use of scenario storylines (Section 3 1.6.1.4). 4 5 6 1.6.1.1 Shared Socio-economic Pathways 7 8 The Shared Socioeconomic Pathways SSP1 to SSP5 describe a range of plausible trends in the evolution of 9 society over the 21st century. They were developed in order to connect a wide range of research 10 communities (Nakicenovic et al., 2014) and consist of two main elements: a set of qualitative, narrative 11 storylines describing societal futures (O’Neill et al., 2017a) and a set of quantified measures of development 12 at aggregated and/or spatially resolved scales. Each pathway is an internally consistent, plausible and 13 integrated description of a socio-economic future, but these socio-economic futures do not account for the 14 effects of climate change, and no new climate policies are assumed. The SSPs’ quantitative projections of 15 socio-economic drivers include population, gross domestic product (GDP) and urbanization (Dellink et al., 16 2017; Jiang and O’Neill, 2017; Samir and Lutz, 2017). By design, the SSPs differ in terms of the socio- 17 economic challenges they present for climate change mitigation and adaptation (Rothman et al., 2014; 18 Schweizer and O’Neill, 2014) and the evolution of these drivers within each SSP reflects this design. 19 Broadly, the five SSPs represent ‘sustainability’ (SSP1), a ‘middle of the road’ path (SSP2), ‘regional 20 rivalry’ (SSP3), ‘inequality’ (SSP4), and ‘fossil fuel intensive’ development (SSP5) (Cross-Chapter Box 1.4, 21 Figure 1) (O’Neill et al., 2017a). More specific information on the SSP framework and the assumptions 22 underlying the SSPs will be provided in the IPCC WGIII report (WGIII, Chapter 3; see also Box SPM.1 in 23 SRCCL (IPCC, 2019d)). 24 25 The SSP narratives and drivers were used to develop scenarios of energy use, air pollution control, land use 26 and greenhouse gas (GHG) emissions developments using integrated assessment models (IAMs) (Riahi et 27 al., 2017; Rogelj et al., 2018a). An IAM can derive multiple emission futures for each socio-economic 28 development pathway, assuming no new mitigation policies or various levels of additional mitigation action 29 (in the case of reference scenarios and mitigation scenarios, respectively (Riahi et al., 2017). By design, the 30 evolution of drivers and emissions within the SSP scenarios do not take into account the effects of climate 31 change. 32 33 The SSPX-Y scenarios and the RCP scenarios are categorized similarly, by reference to the approximate 34 radiative forcing levels each one entails at the end of the 21st century. For example, the ‘1.9’ in the SSP1-1.9 35 scenario stands for an approximate radiative forcing level of 1.9 W m-2 in 2100. The first number (X) in the 36 ‘SSPX-Y’ acronym refers to one of the five shared socio-economic development pathways (Cross-Chapter 37 Box 1.4, Figure 1; Table 1.4). 38 39 40 [START TABLE 1.4 HERE] 41 42 Table 1.4: Overview of different RCP and SSP acronyms as used in this report. Scenario Acronym Description ‘SSPX’ with X The shared socioeconomic pathway family, i.e., the socioeconomic developments with standing for the storylines regarding (among other things) GDP, population, urbanisation, economic shared collaboration, human and technological development projections that describe different socioeconomic future worlds in the absence of climate change and additional climate policy (O’Neill et pathway family (1, al., 2014). The quantification of energy, land use and emission implications in those 2, …, 5) storylines is not part of the SSPX narratives, but follows in a second step in which their climate outcomes are defined. This second step is dependent upon the IAM that is used for this quantification (Riahi et al., 2017) (see SSPX-Y) ‘RCPY’ with Y Representative Concentration Pathways (Moss et al., 2010; van Vuuren et al., 2011). Do Not Cite, Quote or Distribute 1-100 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI standing for These are GHG concentrations (Meinshausen et al., 2011b) , aerosol emissions approximate (Lamarque et al., 2011) and land use pattern time series (Hurtt et al., 2011) derived from radiative forcing several IAMs. The pathways were originally generated from specific sets of socio- level in 2100, at economic drivers, but these are no longer considered. Instead, these RCP emission and levels 2.6, 4.5, 6.0 concentration time series are used in combination with a range of socio-economic futures or 8.5. (see SSPX-RCPY). For example, the CMIP5 intercomparison (assessed in IPCC AR5) developed climate futures based on these emission and concentration pathways from the RCPs. The SSP and RCP Combination of the SSP socioeconomic pathway X with climate futures stemming from combination GCMs, AOGCMs or Earth system model runs that used the RCPY. This combination is ‘SSPX-RCPY’ widely used in the impact literature assessed by WGII (see for example the Special Issue with X and Y as on SSPs by van Vuuren et al. (2014) and the large literature collection in the above. International Committee On New Integrated Climate change assessment Scenarios database (ICONICS, 2021). These SSPX-RCPY scenarios differ from the SSPX-Y group (below) in that the respective socio-economic futures (SSPXs) and emission and concentration futures (RCPYs) were developed separately before being used in combination. ‘SSPX-Y’ with X SSPX-Y is the abbreviation for a scenario, where X is the numbering of the SSP and Y as above. socioeconomic family (1 to 5) that was used to develop the emission pathway, and the Y indicates the approximate radiative forcing value reached by the end of the century. The SSPX-Y scenarios span the nominal range from 1.9 to 8.5 W m2. A range of different IAMs were used to quantify the SSPX-Y scenarios, but each IAM quantified both the scenario-economic futures (energy use, land use, population etc) and various emission futures within the same IAM modelling framework, thus enhancing the consistency between the socio-economic backgrounds and their resulting emission futures. In contrast, the SSPX-RCPY framework combines the SSP socio-economic futures and RCP emission and concentration futures at random (see above). For more details, see Section 1.6.1.1. 1 2 [END TABLE 1.4 HERE] 3 4 5 This SSP scenario categorisation, focused on end-of-century radiative forcing levels, reflects how scenarios 6 were conceptualized until recently, namely, to reach a particular climate target in 2100 at the lowest cost and 7 irrespective of whether the target was exceeded over the century. More recently, and in particular since the 8 IPCC SR1.5 report focused attention on peak warming scenarios (Rogelj et al., 2018b), scenario 9 development started to explicitly consider peak warming, cumulative emissions and the amount of net 10 negative emissions (Rogelj et al., 2018b; Fujimori et al., 2019). 11 12 The SSP scenarios can be used for either emission- or concentration-driven model experiments (Cross- 13 Chapter Box 1.4). ESMs can be run with emissions and concentrations data for GHGs and aerosols and land 14 use or landcover maps and calculate levels of radiative forcing internally. The radiative forcing labels of the 15 RCP and SSP scenarios, such as ‘2.6’ in RCP2.6 or SSP1-2.6, are thus approximate labels for the year 2100 16 only. The actual global mean effective radiative forcing varies across ESMs due to different radiative 17 transfer schemes, uncertainties in aerosol-cloud interactions and different feedback mechanisms, among 18 other reasons. Nonetheless, using approximate radiative forcing labels is advantageous because it establishes 19 a clear categorization of scenarios, with multiple climate forcings and different combinations in those 20 scenarios summarized in a single number. The classifications according to cumulative carbon emissions (see 21 Section 1.6.3) and global warming level (see Section 1.6.2 and Cross-Chapter Box 7.1 on emulators in 22 Chapter 7) complement those forcing labels. 23 24 A key advance of the SSP scenarios relative to the RCPs is a wider span of assumptions on future air quality Do Not Cite, Quote or Distribute 1-101 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 mitigation measures, and hence emissions of short-lived climate forcers (SLCFs) (Rao et al., 2017; Lund et 2 al., 2020). This allows for a more detailed investigation into the relative roles of GHG and SLCF emissions 3 in future global and regional climate change, and hence the implications of policy choices. For instance, 4 SSP1-2.6 builds on an assumption of stringent air quality mitigation policy, leading to rapid reductions in 5 particle emissions, while SSP3-7.0 assumes slow improvements, with pollutant emissions over the 21st 6 century comparable to current levels (Cross-Chapter Box 1.4, Figure 2, Chapter 6, Figure 6.19). 7 8 One limitation of the SSP scenarios used for CMIP6 and in this Report is that they reduce emissions from all 9 the major ozone-depleting substances controlled under the Montreal Protocol (CFCs, halons, and 10 hydrochlorofluorocarbons (HCFCs)) uniformly, rather than representing a fuller range of possible high and 11 low emission futures (UNEP, 2016). Hydrofluorocarbon (HFC) emissions, on the other hand, span a wider 12 range within the SSPs than in the RCPs (Cross-Chapter Box 1.4, Figure 2). 13 14 The SSP scenarios and previous RCP scenarios are not directly comparable. First, the gas-to-gas 15 compositions differ; for example, the SSP5-8.5 scenario has higher CO2 concentrations but lower methane 16 concentrations compared to RCP8.5. Second, the projected 21st-century trajectories may differ, even if they 17 result in the same radiative forcing by 2100. Third, the overall effective radiative forcing (see Chapter 7) 18 may differ, and tends to be higher for the SSPs compared to RCPs that share the same nominal stratospheric- 19 temperature adjusted radiative forcing label. The stratospheric-temperature adjusted radiative forcings of the 20 SSPs and RCPs, however, remain relatively close, at least by 2100 (Tebaldi et al., 2021). In summary, 21 differences in, for example, CMIP5 RCP8.5 and CMIP6 SSP5-8.5 ESM outputs, are partially due to different 22 scenario characteristics rather than different ESM characteristics only (Chapter 4, Section 4.6.2). 23 24 When investigating various mitigation futures, WGIII goes beyond the core set of SSP scenarios assessed in 25 WGI (SSP1-1.9, SSP1-2.6, etc.) to consider the characteristics of more than 1000 scenarios (see Cross- 26 Chapter Box 7.1 in Chapter 7). In addition, while staying within the framework of socio-economic 27 development pathways (SSP1 to SSP5), WGIII also considers various mitigation possibilities through so- 28 called illustrative pathways (IPs). These illustrative pathways help to highlight key narratives in the literature 29 concerning various technological, social and behavioral options for mitigation, various timings for 30 implementation, or varying emphasis on different GHG and land use options. Just as with the SSPX-Y 31 scenarios considered in this report, these illustrative pathways can be placed in relation to the matrix of SSP 32 families and approximate radiative forcing levels in 2100 (see Cross-Chapter Box 1.4, Figure 1 and Working 33 Group III, Chapter 3). 34 35 No likelihood is attached to the scenarios assessed in this report, and the feasibility of specific scenarios in 36 relation to current trends is best informed by the WGIII contribution to AR6. In the scenario literature, the 37 plausibility of the high emissions levels underlying scenarios such as RCP8.5 or SSP5-8.5 has been debated 38 in light of recent developments in the energy sector. (see Section 1.6.1.4). 39 40 41 [START CROSS CHAPTER BOX 1.4 HERE] 42 43 Cross-Chapter Box 1.4: The SSP scenarios as used in Working Group I 44 45 Contributing Authors: Jan Fuglestvedt (Norway), Celine Guivarch (France), Chris Jones (UK), Malte 46 Meinshausen (Australia/Germany), Zebedee Nicholls (Australia), Gian-Kasper Plattner (Switzerland), 47 Keywan Riahi (Austria), Joeri Rogelj (UK/Belgium), Sophie Szopa (France), Claudia Tebaldi (USA/Italy), 48 Anne-Marie Treguier (France), and Detlef van Vuuren (Netherlands) 49 50 The new nine SSP emission and concentration scenarios (SSP1-1.9 to SSP5-8.5; Cross-Chapter Box 1.4, 51 Table 1) offer unprecedented detail of input data for climate model simulations. They allow for a more 52 comprehensive assessment of climate drivers and responses than has previously been available, in particular 53 because some of the scenarios’ time series, e.g., pollutants, emissions or changes in land use and land cover, 54 are more diverse in the SSP scenarios than in the RCPs used in AR5 (e.g., Chuwah et al., 2013) (Cross- 55 Chapter Box 1.4, Figure 2). Do Not Cite, Quote or Distribute 1-102 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 The core set of five SSP scenarios SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5 was selected in 3 this Report to align with the objective that the new generation of SSP scenarios should fill certain gaps 4 identified in the RCPs. For example, a scenario assuming reduced air pollution control and thus higher 5 aerosol emissions was missing from the RCPs. Likewise, nominally the only ‘no-additional-climate-policy’ 6 scenario in the set of RCPs was RCP8.5. The new SSP3-7.0 ‘no-additional-climate-policy’ scenario fills both 7 these gaps. A very strong mitigation scenario in line with the 1.5°C goal of the Paris Agreement was also 8 missing from the RCPs, and the SSP1-1.9 scenario now fills this gap, complementing the other strong 9 mitigation scenario SSP1-2.6. The five core SSPs were also chosen to ensure some overlap with the RCP 10 levels for radiative forcing at the year 2100 (specifically 2.6, 4.5, and 8.5) (O’Neill et al., 2016; Tebaldi et 11 al., 2021), although effective radiative forcings are generally higher in the SSP scenarios compared to the 12 equivalently-named RCP pathways (Cross-Chapter Box 1.4, Figure 1; Chapter 4, Section 4.6.2). In theory, 13 running scenarios with similar radiative forcings would permit analysis of the CMIP5 and CMIP6 outcomes 14 for pairs of scenarios (e.g., RCP8.5 and SSP5-8.5) in terms of varying model characteristics rather than 15 differences in the underlying scenarios. In practice, however, there are limitations to this approach (Section 16 1.6.1.1 and Chapter 4, Section 4.6.2). 17 18 19 [START Cross-Chapter Box 1.4, FIGURE 1 HERE] 20 21 Cross-Chapter Box 1.4, Figure 1: The SSP scenarios used in this report, their indicative temperature evolution 22 and radiative forcing categorization, and the five socio-economic storylines upon which they are built. The core 23 set of scenarios used in this report, i.e., SSP1-1.9, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, is shown together with 24 an additional four SSPs that are part of ScenarioMIP, as well as previous RCP scenarios. In the left panel, the indicative 25 temperature evolution is shown (adapted from Meinshausen et al., 2020). The black stripes on the respective scenario 26 family panels on the left side indicate a larger set of IAM-based SSP scenarios that span the scenario range more fully, 27 but are not used in this report. The SSP-radiative forcing matrix is shown on the right, with the SSP socioeconomic 28 narratives shown as columns and the indicative radiative forcing categorisation by 2100 shown as rows. Note that the 29 descriptive labels for the five SSP narratives refer mainly to the reference scenario futures without additional climate 30 policies. For example, SSP5 can accommodate strong mitigation scenarios leading to net zero emissions; these do not 31 match a ‘fossil-fueled development’ label. Further details on data sources and processing are available in the chapter 32 data table (Table 1.SM.1). 33 34 [END Cross-Chapter Box 1.4, FIGURE 1 HERE] 35 36 37 [START Cross-Chapter Box 1.4, TABLE 1 HERE] 38 39 Cross-Chapter Box 1.4, Table 1: Overview of SSP scenarios used in this report. The middle column briefly 40 describes the SSP scenarios and the right column indicates the previous RCP scenarios that most closely 41 match that SSP’s assessed global-mean temperatures (GSAT) trajectory. RCP scenarios are generally found 42 to result in larger modelled warming for the same nominal radiative forcing label (Chapter 4, Section 43 4.6.2.2). The five core SSP scenarios used most commonly in this report are highlighted in bold. Further SSP 44 scenarios are used where they allow assessment of specific aspects, e.g., air pollution policies in Chapter 6 45 (SSP3-7.0-lowNTCF). RCPs are used in this report wherever the relevant scientific literature makes 46 substantial use of regional or domain-specific model output that is based on these previous RCP pathways, 47 such as sea level rise projections in Chapter 9 (Section 9.6.3.1) or regional climate aspects in Chapters 10 48 and 12. See Chapter 4 (Section 4.3.4) for the GSAT assessment for the SSP scenarios and Section 4.6.2.2 for 49 a comparison between SSPs and RCPs in terms of both radiative forcing and global surface temperature. 50 SSPX-Y Description from an emission / Closest RCP scenarios scenario concentration and temperature perspective (Chapter 4, Table 4.2) SSP1-1.9 Holds warming to approximately 1.5°C Not available. No equivalently low RCP above 1850-1900 in 2100 after slight scenario exists. overshoot (median) and implied net zero Do Not Cite, Quote or Distribute 1-103 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI CO2 emissions around the middle of the century. SSP1-2.6 Stays below 2.0°C warming relative to RCP2.6, although RCP2.6 might be cooler 1850-1900 (median) with implied net for the same model settings. zero emissions in the second half of the century. SSP4-3.4 A scenario in between SSP1-2.6 and No 3.4 level of end-of-century radiative SSP2-4.5 in terms of end-of-century forcing was available in the RCPs. radiative forcing. It does not stay below Nominally SSP4-3.4 sits between RCP 2.6 2.0°C in most CMIP6 runs (Chapter 4) and RCP 4.5, although SSP4-3.4 might be relative to 1850-1900. more similar to RCP4.5. Also, in the early decades of the 21st century, SSP4-3.4 is close to RCP6.0, which featured lower radiative forcing than RCP4.5 in the first decades of the 21st century. SSP2-4.5 Scenario approximately in line with the RCP4.5 and, until 2050, also RCP6.0. upper end of aggregate NDC emission Forcing in the latter was even lower than levels by 2030 (see Section 1.2.2 and RCP4.5 in the early decades of the 21st Chapter 4, Section 4.3; SR1.5, (IPCC, century. 2018) , Box 1). SR1.5 assessed temperature projections for NDCs to be between 2.7 and 3.4°C by 2100 (Section 1.2.2; SR1.5 (IPCC, 2018); Cross- Chapter Box 11 in Chapter 11), corresponding to the upper half of projected warming under SSP2-4.5 (Chapter 4). New or updated NDCs by the end of 2020 did not significantly change the emissions projections up to 2030, although more countries adopted 2050 net zero targets in line with SSP1- 1.9 or SSP1-2.6. The SSP2-4.5 scenario deviates mildly from a ‘no-additional- climate-policy’ reference scenario, resulting in a best-estimate warming around 2.7°C by the end of the 21st century relative to 1850-1900 (Chapter 4). SSP4-6.0 The end-of-century nominal radiative RCP6.0 is nominally closest in the second forcing level of 6.0 W/m2 can be half of the century, although global mean considered a ‘no-additional-climate- temperatures are estimated to be generally policy’ reference scenario, under SSP1 lower in RCPs compared to SSPs. and SSP4 socioeconomic development Furthermore, RCP6.0 features lower narratives. warming than SSP4-6.0, as it has very similar temperature projections compared to the nominally lower RCP4.5 scenario in the first half of the century. SSP3-7.0 A medium to high reference scenario In between RCP6.0 and RCP8.5, although resulting from no additional climate SSP3-7.0 non-CO2 emissions and aerosols policy under the SSP3 socioeconomic are higher than in any of the RCPs. development narrative. SSP3-7.0 has particularly high non-CO2 emissions, including high aerosols emissions. SSP3-7.0- A variation of the medium to high SSP3-7.0-lowNTCF is between RCP6.0 lowNTCF reference scenario SSP3-7.0 but with and RCP8.5, as RCP scenarios generally Do Not Cite, Quote or Distribute 1-104 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI mitigation of CH4 and/or short-lived incorporated a narrow and comparatively species such as black carbon and other low level of SLCF emissions across the short-lived climate forcers (SLCF). Note range of RCPs. that variants of SSP3-7.0-lowNTCF differ in terms of whether methane emissions are reduced 9 (Chapter 4, Section 4.4 and Chapter 6, Section 6.6). SSP5-3.4 A mitigation-focused variant of SSP5-8.5 Not available. Initially, until 2040, similar OS that initially follows unconstrained to RCP8.5. (Overshoot emission growth in a fossil-intensive ) setting until 2040 and then implements the largest net negative CO2 emissions of all SSP scenarios in the second half of 21st century to reach SSP1-2.6 forcing levels in the 22nd century. Used to consider reversibility and strong overshoot scenarios in, e.g., Chapters 4 and 5. SSP5-8.5 A high reference scenario with no RCP8.5, although CO2 emissions under additional climate policy. Emission SSP5-8.5 are higher towards the end of the levels as high as SSP5-8.5 are not century (Cross-Chapter Box 1.4, Figure 2). obtained by Integrated Assessment Methane emissions under SSP5-8.5 are Models (IAMs) under any of the SSPs lower than under RCP 8.5. When used other than the fossil fueled SSP5 with the same model settings, SSP5-8.5 socioeconomic development pathway. may result in slightly higher temperatures than RCP8.5 (Chapter 4, Section 4.6.2). 1 2 [END Cross-Chapter Box 1.4, TABLE 1 HERE] 3 4 5 In contrast to stylized assumptions about the future evolution of emissions (e.g., a linear phase-out from year 6 A to year B), these SSP scenarios are the result of a detailed scenario generation process (see Sections 7 1.6.1.1 and 1.6.1.2). While IAMs produce internally-consistent future emission time series for CO2, CH4, 8 N2O, and aerosols for the SSP scenarios (Riahi et al., 2017; Rogelj et al., 2018a), these emission scenarios 9 are subject to several processing steps for harmonisation (Gidden et al., 2018) and in-filling (Lamboll et al., 10 2020), before also being complemented by several datasets so that ESMs can run these SSPs (Durack et al., 11 2018; Tebaldi et al., 2021). Although five scenarios are the primary focus of WGI, a total of nine SSP 12 scenarios have been prepared with all the necessary detail to drive the ESMs as part of the CMIP6 (Cross- 13 Chapter Box 1.4, Figure 1 and Table 2). 14 15 16 [START Cross-Chapter Box 1.4, TABLE 2 HERE] 17 18 Cross-Chapter Box 1.4, Table 2: Overview of key climate forcer datasets used as input by ESMs for 19 historical and future SSP scenario experiments. The data is available from the Earth System Grid Federation 20 (ESGF, 2021) described in Eyring et al. (2016). 21 9 The AerChemMIP variant of SSP3-7.0-lowNTCF (Collins et al., 2017) only reduced aerosol and ozone precursors compared to SSP3-7.0, not methane. The SSP3-7.0-lowNTCF variant by the Integrated Assessment Models also reduced methane emissions (Gidden et al., 2019), which creates differences between SSP3-7.0-lowNTCF and SSP3-7.0 also in terms of methane concentrations and some fluorinated gas concentrations that have OH related sinks (Meinshausen et al., 2020). Do Not Cite, Quote or Distribute 1-105 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI Climate Forcer Description CO2 emissions Harmonized historical and future gridded emissions of anthropogenic CO2 (emission-driven runs emissions (Hoesly et al., 2018; Gidden et al., 2019) are used instead of the only) prescribed CO2 concentrations. See Chapter 4 (Section 4.3.1). Historical and future Greenhouse gas surface air mole fractions of 43 species, including CO2, CH4, greenhouse gas N2O, HFCs, PFCs, halons, HCFCs, CFCs, SF6, NF3, including latitudinal concentrations gradients and seasonality from year 1 to 2500 (Meinshausen et al., 2017, 2020) Land use change and Globally gridded land use and land cover change datasets (Hurtt et al., 2020; management patterns Ma et al., 2020b) Biomass burning Historical fire-related gridded emissions, including SO2, NOx, CO, BC, OC, emissions NH3, NMVOCs, relevant to concentration-driven historical and future SSP scenario runs (van Marle et al., 2017). Stratospheric and Historical and future ozone dataset, also with total column ozone (CCMI, tropospheric ozone 2021). Reactive gas emissions Gridded global anthropogenic emissions of reactive gases and aerosol precursors, including CO, SOx, CH4, NOx, NMVOCs, or NH3 (Hoesly et al., 2018; Feng et al., 2020) Solar forcing Radiative and particle input of solar variability from 1850 through to 2300 (Matthes et al., 2017). Future variations in solar forcing also reflect long-term multi-decadal trends. Volcanic forcing Historical stratospheric aerosol climatology (Thomason et al., 2018), with the mean stratospheric volcanic aerosol prescribed in future projections. 1 2 [END Cross-Chapter Box 1.4, TABLE 2 HERE] 3 4 5 ESMs are driven by either emission or concentration scenarios. Inferring concentration changes from 6 emission time series requires using carbon cycle and other gas cycle models. To aid comparability across 7 ESMs, and in order to allow participation of ESMs that do not have coupled carbon and other gas cycles 8 models in CMIP6, most of the CMIP6 ESM experiments are so-called ‘concentration-driven’ runs, with 9 concentrations of CO2, CH4, N2O and other well-mixed GHGs prescribed in conjunction with aerosol 10 emissions, ozone changes and effects from human-induced land cover changes that may be radiatively active 11 via albedo changes (Cross-Chapter Box 1.4, Figure 2). In these concentration-driven climate projections, the 12 uncertainty in projected future climate change resulting from our limited understanding of how the carbon 13 cycle and other gas cycles will evolve in the future is not captured. For example, when deriving the default 14 concentrations for these scenarios, permafrost and other carbon cycle feedbacks are considered using default 15 settings, with a single time series prescribed for all ESMs (Meinshausen et al., 2020). Thus, associated 16 uncertainties (Joos et al., 2013; Schuur et al., 2015) are not considered. 17 18 The so-called ‘emission-driven’ experiments (Jones et al., 2016b) use the same input datasets as 19 concentration-driven ESM experiments, except that they use CO2 emissions rather than concentrations 20 (Chapter 4, Section 4.3.1; Chapter 5). In these experiments, atmospheric CO2 concentrations are calculated 21 internally using the ESM interactive carbon cycle module and thus differ from the prescribed default CO2 22 concentrations used in the concentration-driven runs. In the particular case of SSP5-8.5, the emission-driven 23 runs are assessed to add no significant additional uncertainty to future global surface air temperature (GSAT) 24 projections (Chapter 4, Section 4.3.1). However, generally, when assessing uncertainties in future climate Do Not Cite, Quote or Distribute 1-106 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 projections, it is important to consider which elements of the cause-effect chain from emissions to the 2 resulting climate change are interactively included as part of the model projections, and which are externally 3 prescribed using default settings. 4 5 6 [START Cross-Chapter Box 1.4, FIGURE 2 HERE] 7 8 Cross-Chapter Box 1.4, Figure 2: Comparison between the Shared Socio-economic Pathways (SSP) scenarios 9 and the Representative Concentration Pathway (RCP) scenarios in terms of their CO2, CH4 and N2O 10 atmospheric concentrations (panels a-c), and their global emissions (panels d-o). Also shown are gridded emission 11 differences for sulfur (panel p) and black carbon (panel q) for the year 2000 between the input emission datasets that 12 underpinned the CMIP5 and CMIP6 model intercomparisons. Historical emission estimates are provided in black in 13 panels d to o. The range of concentrations and emissions investigated under the RCP pathways is grey shaded. Panels p 14 and q adapted from Figure 7 in Hoesly et al. (2018). Further details on data sources and processing are available in the 15 chapter data table (Table 1.SM.1). 16 17 [END Cross-Chapter Box 1.4, FIGURE 2 HERE] 18 19 20 [END CROSS CHAPTER BOX 1.4 HERE] 21 22 23 1.6.1.2 Scenario generation process for CMIP6 24 25 The scenario generation process involves research communities linked to all three IPCC Working Groups 26 (Figure 1.27). It generally starts in the scientific communities associated with WGII and WGIII with the 27 definition of new socio-economic scenario storylines (IPCC, 2000; O’Neill et al., 2014) that are quantified in 28 terms of their drivers, i.e., GDP, population, technology, energy and land use and their resulting emissions 29 (Riahi et al., 2017). Then, numerous complementation and harmonisation steps are necessary for datasets 30 within the WGI and WGIII science communities, including gridding emissions of anthropogenic short-lived 31 forcers, providing open biomass burning emission estimates, preparing land use patterns, aerosol fields, 32 stratospheric and tropospheric ozone, nitrogen deposition datasets, solar irradiance and aerosol optical 33 property estimates, and observed and projected greenhouse gas concentration time series (documented for 34 CMIP6 through input4mips; Durack et al., 2018; Cross-Chapter Box 1.4, Table 2). 35 36 Once these datasets are completed, ESMs are run in coordinated model intercomparison projects in the WGI 37 science community, using standardized simulation protocols and scenario data. The most recent example of 38 such a coordinated effort is the CMIP6 exercise (Eyring et al., 2016; see also Section 1.5.4) with, in 39 particular, ScenarioMIP (O’Neill et al., 2016). The WGI science community feeds back climate information 40 to WGIII via climate emulators (Cross-Chapter Box 7.1 in Chapter 7) that are updated and calibrated with 41 the ESMs’ temperature responses and other lines of evidence. Next, this climate information is used to 42 compute several high-level global climate indicators (e.g., atmospheric concentrations, global temperatures) 43 for a much wider set of hundreds of scenarios that are assessed as part of IPCC WGIII assessment (WGIII 44 Annex C). The outcomes from climate models run under the different scenarios are then used to calculate the 45 evolution of climatic impact-drivers (Chapter 12), and utilized by impact researchers together with exposure 46 and vulnerability information, in order to characterize risk from future climate change to human and natural 47 systems. The climate impacts associated with these scenarios or different warming levels are then assessed as 48 part of WGII reports (Figure 1.27). 49 50 51 [START FIGURE 1.27 HERE] 52 53 Figure 1.27: A simplified illustration of the scenario generation process that involves the scientific communities 54 represented in the three IPCC Working Groups. The circular set of arrows at the top indicate the main 55 set of models and workflows used in that scenario generation process, with the lower level indicating the 56 datasets. Do Not Cite, Quote or Distribute 1-107 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 2 [END FIGURE 1.27 HERE] 3 4 5 1.6.1.3 History of scenarios within the IPCC 6 7 Scenario modelling experiments have been a core element of physical climate science since the first transient 8 simulations with a General Circulation Model in 1988 (Hansen et al., 1988; see Section 1.3). Scenarios and 9 modelling experiments assessed in IPCC reports have evolved over time, which provides a ‘history of how 10 the future was seen’. The starting time for the scenarios moves as actual emissions supersede earlier 11 emission assumptions, while new scientific insights into the range of plausible population trends, 12 behavioural changes and technology options and other key socioeconomic drivers of emissions also emerge 13 (see WGIII; Legget, 1992; IPCC, 2000; Moss et al., 2010; Riahi et al., 2017). Many different sets of climate 14 projections have been produced over the past several decades, using different sets of scenarios. Here, we 15 compare those earlier scenarios against the most recent ones. 16 17 Climate science research involving scenarios necessarily follows a series of consecutive steps (see Figure 18 1.27). As each step waits for input from the preceding one, delays often occur that result in the impact 19 literature basing its analyses on earlier scenarios than those most current in the climate change mitigation and 20 climate system literature. It is hence important to provide an approximate comparison across the various 21 scenario generations (Figure 1.28; Cross-Chapter Box 1.4, Table 1; Chapter 4). 22 23 24 [START FIGURE 1.28 HERE] 25 26 Figure 1.28: Comparison of the range of fossil and industrial CO2 emissions from scenarios used in previous 27 assessments up to AR6. Previous assessments are the IS92 scenarios from 1992 (top panel), the Special 28 Report on Emissions Scenarios (SRES) scenarios from the year 2000 (second panel), the Representative 29 Concentration Pathway (RCP) scenarios designed around 2010 (third panel) and the Shared Socio- 30 economic Pathways (SSP) scenarios (second bottom panel). In addition, historical emissions are shown 31 (black line) (Chapter 5, Figure 5.5); a more complete set of scenarios is assessed in SR1.5 (bottom panel) 32 (Huppmann et al., 2018). Further details on data sources and processing are available in the chapter data 33 table (Table 1.SM.1). 34 35 [END FIGURE 1.28 HERE] 36 37 38 The first widely used set of IPCC emission scenarios was the IS92 scenarios in 1992 (Leggett et al., 1992). 39 Apart from reference scenarios, IS92 also included a set of stabilisation scenarios, the so-called ‘S’ 40 scenarios. Those ‘S’ pathways were designed to lead to CO2 stabilisation levels such as 350 ppm or 450 41 ppm. By 1996, those latter stabilisation levels were complemented in the scientific literature by alternative 42 trajectories that assumed a delayed onset of climate change mitigation action (Figure 1.28; Wigley et al., 43 1996). 44 45 By 2000, the IPCC Special Report on Emission Scenarios (SRES) produced the SRES scenarios (IPCC, 46 2000), albeit without assuming any climate-policy-induced mitigation. The four broad groups of SRES 47 scenarios (scenario ‘families’) A1, A2, B1 and B2 were the first scenarios to emphasize socio-economic 48 scenario storylines, and also first to emphasize other greenhouse gases, land use change and aerosols. 49 Represented by three scenarios for the high-growth A1 scenario family, those 6 SRES scenarios (A1FI, A1B, 50 A1T, A2, B1, and B2) can still sometimes be found in today’s climate impact literature. The void of missing 51 climate change mitigation scenarios was filled by a range of community exercises, including the so-called 52 post-SRES scenarios (Swart et al., 2002). 53 54 The RCP scenarios (van Vuuren et al., 2011) then broke new ground by providing low emission pathways 55 that implied strong climate change mitigation including an example with negative CO2 emissions on a large 56 scale, namely RCP2.6. As shown in Figure 1.28, the upper end of the scenario range has not substantially Do Not Cite, Quote or Distribute 1-108 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 shifted. Building on the SRES multi-gas scenarios, the RCPs include time series of emissions and 2 concentrations of the full suite of greenhouse gases and aerosols and chemically active gases, as well as land 3 use and land cover (Moss et al., 2010). The word ‘representative’ signifies that each RCP is only one of 4 many possible scenarios that would lead to the specific radiative forcing characteristics. The term pathway 5 emphasizes that not only the long-term concentration levels are of interest, but also the trajectory taken over 6 time to reach that outcome (Moss et al., 2010). RCPs usually refer to the concentration pathway extending to 7 2100, for which IAMs produced corresponding emission scenarios. Four RCPs produced from IAMs were 8 selected from the published literature and are used in AR5 as well as in this report, spanning approximately 9 the range from below 2°C warming to high (>4°C) warming best-estimates by the end of the 21st century: 10 RCP2.6, RCP4.5 and RCP6.0 and RCP8.5 (Cross-Chapter Box 1.4, Table 1). Extended Concentration 11 Pathways (ECPs) describe extensions of the RCPs from 2100 to 2300 that were calculated using simple rules 12 generated by stakeholder consultations; these do not represent fully consistent scenarios (Meinshausen et al., 13 2011b). 14 15 By design, the RCP emission and concentration pathways were originally developed using particular socio- 16 economic development pathways, but those are no longer considered (Moss et al., 2010). The different levels 17 of emissions and climate change represented in the RCPs can hence be explored against the backdrop of 18 different socio-economic development pathways (SSP1 to SSP5) (Section 1.6.1.1; Cross-Chapter Box 1.4). 19 This integrative SSP-RCP framework (‘SSPX-RCPY’ in Table 1.4) is now widely used in the climate impact 20 and policy analysis literature (e.g., ICONICS, 2021; Green et al., 2020; O’Neill et al., 2020), where climate 21 projections obtained under the RCP scenarios are analysed against the backdrop of various SSPs. 22 Considering various levels of future emissions and climate change for each socio-economic development 23 pathway was an evolution from the previous SRES framework (IPCC, 2000), in which socio-economic and 24 emission futures were closely aligned. 25 26 The new set of scenarios (SSP1-1.9 to SSP5-8.5) now features a higher top level of CO2 emissions (SSP5-8.5 27 compared to RCP8.5), although the most significant change is again the addition of a very low climate 28 change mitigation scenario (SSP1-1.9, compared to the previous low scenario, RCP2.6). Also, historically, 29 none of the previous scenario sets featured a scenario that involves a very pronounced peak-and-decline 30 emissions trajectory, but SSP1-1.9 does so now. The full set of nine SSP scenarios now includes a high 31 aerosol emission scenario (SSP3-7.0). The RCPs featured more uniformly low aerosol trajectories across all 32 scenarios (Cross-Chapter Box 1.4, Figure 2). More generally, the SSP scenarios feature a later peak of global 33 emission for the lower scenarios, simply as a consequence of historical emissions not having followed the 34 trajectory projected by previous low scenarios (Figure 1.28). 35 36 Over the last decades, discussions around scenarios have often focussed on whether recent trends make 37 certain future scenarios more or less probable or whether all scenarios are too high or too low. When the 38 SRES scenarios first appeared, the debate was often whether the scenarios were overestimating actual world 39 emissions developments (e.g., Castles and Henderson, 2003). With the strong emissions increase throughout 40 the 2000s, that debate then shifted towards the question of whether the lower future climate change 41 mitigation scenarios were rendered unfeasible (Pielke et al., 2008; van Vuuren and Riahi, 2008). Historical 42 emissions over 2000 to 2010 approximately track the upper half of SRES and RCP projections (Figure 1.28). 43 More generally, the global fossil and industrial CO2 emissions of recent decades tracked approximately the 44 middle of the projected scenario ranges (see Fig 1.28), although with regional differences (Pedersen et al., 45 2020). 46 47 48 1.6.1.4 The likelihood of reference scenarios, scenario uncertainty and storylines 49 50 In general, no likelihood is attached to the scenarios assessed in this Report. The use of different scenarios 51 for climate change projections allows to explore ‘scenario uncertainty’ (Collins et al., 2013; SR1.5; see also 52 Section 1.4.4). Scenario uncertainty is fundamentally different from geophysical uncertainties, which result 53 from limitations in the understanding and predictability of the climate system (Smith and Stern, 2011). In 54 scenarios, by contrast, future emissions depend to a large extent on the collective outcome of choices and 55 processes related to population dynamics and economic activity, or on choices that affect a given activity’s Do Not Cite, Quote or Distribute 1-109 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 energy and emissions intensity (Jones, 2000; Knutti et al., 2008; Kriegler et al., 2012; van Vuuren et al., 2 2014). Even if identical socio-economic futures are assumed, the associated future emissions still face 3 uncertainties, since different experts and model frameworks diverge in their estimates of future emission 4 ranges (Ho et al., 2019). 5 6 When exploring various climate futures, scenarios with no, or no additional, climate policies are often 7 referred to as ‘baseline’ or ‘reference scenarios’ (Section 1.6.1.1; Annex VII: Glossary). Among the five core 8 scenarios used most in this report, SSP3-7.0 and SSP5-8.5 are explicit ‘no-climate-policy’ scenarios (Gidden 9 et al., 2019; Cross-Chapter Box 1.4, Table 1), assuming a carbon price of zero. These future ‘baseline’ 10 scenarios are hence counterfactuals that include less climate policies compared to ‘business-as-usual’ 11 scenarios – given that ‘business-as-usual’ scenarios could be understood to imply a continuation of existing 12 climate policies. Generally, future scenarios are meant to cover a broad range of plausible futures, due for 13 example to unforeseen discontinuities in development pathways (Raskin and Swart, 2020), or to large 14 uncertainties in underlying long-term projections of economic drivers (Christensen et al., 2018). However, 15 the likelihood of high emission scenarios such as RCP8.5 or SSP5-8.5 is considered low in light of recent 16 developments in the energy sector (Hausfather and Peters, 2020a, 2020b). Studies that consider possible 17 future emission trends in the absence of additional climate policies, such as the recent IEA 2020 World 18 Energy Outlook ‘stated policy’ scenario (International Energy Agency, 2020), project approximately 19 constant fossil and industrial CO2 emissions out to 2070, approximately in line with the medium RCP4.5, 20 RCP6.0 and SSP2-4.5 scenarios (Hausfather and Peters, 2020b) and the 2030 global emission levels that are 21 pledged as part of the Nationally Determined Contributions (NDCs) under the Paris Agreement (Section 22 1.2.2; (Fawcett et al., 2015; Rogelj et al., 2016; UNFCCC, 2016; IPCC, 2018). On the other hand, the default 23 concentrations aligned with RCP8.5 or SSP5-8.5 and resulting climate futures derived by ESMs could be 24 reached by lower emission trajectories than RCP8.5 or SSP5-8.5. That is because the uncertainty range on 25 carbon-cycle feedbacks includes stronger feedbacks than assumed in the default derivation of RCP8.5 and 26 SSP5-8.5 concentrations (Ciais et al., 2013; Friedlingstein et al., 2014; Booth et al., 2017; see also Chapter 5, 27 Section 5.4). 28 29 To address long-term scenario uncertainties, scenario storylines (or ‘narratives’) are often used (Rounsevell 30 and Metzger, 2010; O’Neill et al., 2014) (see Section 1.4.4 for a more general discussion on ‘storylines’ also 31 covering ‘physical climate storylines’). Scenario storylines are descriptions of a future world, and the related 32 large-scale socio-economic development path towards that world that are deemed plausible within the 33 current state of knowledge and historical experience (WGIII; Section 1.2.3). Scenario storylines attempt to 34 ‘stimulate, provoke, and communicate visions of what the future could hold for us’ (Rounsevell and 35 Metzger, 2010) in settings where either limited knowledge or inherent unpredictability in social systems 36 prevent a forecast or numerical prediction. Scenario storylines have been used in previous climate research, 37 and they are the explicit or implicit starting point of any scenario exercise, including for the SRES scenarios 38 (IPCC, 2000) and the SSPs (e.g., O’Neill et al., 2017a). 39 40 Recent technological or socio-economic trends might be informative for bounding near-term future trends, 41 for example, if technological progress renders a mitigation technology cheaper than previously assumed. 42 However, short-term emission trends alone do not generally rule out an opposite trend in the future (van 43 Vuuren et al., 2010). The ranking of individual RCP emission scenarios from the IAMs with regard to 44 emission levels is different for different time horizons, e.g., 2020 versus longer-term emission levels; For 45 example, the strongest climate change mitigation scenario RCP2.6 was in fact the second highest CO2 46 emission scenario (jointly with RCP4.5) before 2020 in the set of RCPs and the strong global emission 47 decline in RCP2.6 only followed after 2020. Implicitly, this scenario feature was cautioning against the 48 assumption that short-term trends predicate particular long-term trajectories. This is also the case in relation 49 to the COVID-19 related drop in 2020 emissions. Potential changes in underlying drivers of emissions, such 50 as those potentially incentivised by COVID-19 recovery stimulus packages, are more significant for longer- 51 term emissions than the short-term deviation from recent emission trends (Cross-Chapter Box 6.1 on 52 COVID-19 in Chapter 6). 53 54 Do Not Cite, Quote or Distribute 1-110 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 1.6.2 Global warming levels 2 3 The global mean surface temperature change, or ‘global warming level’ (GWL), is a ‘dimension of 4 integration’ that is highly relevant across scientific disciplines and socioeconomic actors. First, global 5 warming levels relative to pre-industrial conditions are the quantity in which the 1.5°C and ‘well below 2°C’ 6 Paris Agreement goals were formulated. Second, global mean temperature change has been found to be 7 nearly-linearly related to a number of regional climate effects (Mitchell et al., 2000; Mitchell, 2003; Tebaldi 8 and Arblaster, 2014; Seneviratne et al., 2016; Li et al., 2020; Seneviratne and Hauser, 2020). Even where 9 non-linearities are found, some regional climate effects can be considered to be almost scenario-independent 10 for a given level of warming (Cross-Chapter Box 11.1 in Chapter 11; Chapter 4, Sections 4.2.4 and 4.6.1; 11 Chapter 8, Section 8.5.3; Chapter 10, Section 10.4.3.1). Finally, the evolution of aggregated impacts with 12 warming levels has been widely used and embedded in the assessment of the ‘Reasons for Concern’ (RFC) 13 in IPCC WGII (Smith et al., 2009; IPCC, 2014a). The RFC framework was further expanded in SR1.5 14 (2018), SROCC (2019) and SRCCL (2019) by explicitly describing the differential impacts of half-degree 15 warming steps (cf. King et al., 2017) (Section 1.4.4; Cross-Chapter Box 12.1 in Chapter 12). 16 17 In this Report, the term ‘global warming level’ refers to the categorisation of global and regional climate 18 change, associated impacts, emission and concentration scenarios by global mean surface temperature 19 relative to 1850-1900, which is the period used as a proxy for pre-industrial levels (see Cross-Chapter Box 20 11.1 in Chapter 11). By default, GWLs are expressed as global surface air temperature (GSAT; see Section 21 1.4.1; Cross-Chapter Box 2.3 in Chapter 2). 22 23 As the SR1.5 concluded, even half-degree global mean temperature steps carry robust differences in climate 24 impacts (see SR1.5,IPCC, 2018; Schleussner et al., 2016a; Wartenburger et al., 2017; see also Chapter 11). 25 This Report adopts half-degree warming levels which allows integration within and across the three WGs for 26 climate projections, impacts, adaptation challenges and mitigation challenges. The core set of - GWLs 1.5, 27 2.0, 3.0 and 4.0°C - are highlighted (Chapters 4, 8, 11, 12 and the Atlas). Given that much impact analysis is 28 based on previous scenarios, i.e., RCPs or SRES, and climate change mitigation analysis is based on new 29 emission scenarios in addition to the main SSP scenarios, these global warming levels assist in the 30 comparison of climate states across scenarios and in the synthesis across the broader literature. 31 32 The transient and equilibrium states of certain global warming levels can differ in their climate impacts 33 (IPCC, 2018; King et al., 2020). Climate impacts in a ‘transient’ world relate to a scenario in which the 34 world is continuing to warm. On the other hand, climate impacts at the same warming levels can also be 35 estimated from equilibrium states after a (relatively) short-term stabilisation by the end of the 21st century or 36 at a (near-) equilibrium state after a long-term (multi-decadal to multi-millennial) stabilisation. Different 37 methods to estimate these climate states come with challenges and limitations (Chapter 4, Section 4.6.1; 38 Cross-Chapter Box 11.1 in Chapter 11). First, information can be drawn from GCM or ESM simulations that 39 ‘pass through’ the respective warming levels (as used and demonstrated in the Interactive Atlas to this 40 report), also called ‘epoch’ or ‘time-shift’ approaches (Chapter 4, Sections 4.2.4 and 4.6.1) (Herger et al., 41 2015; James et al., 2017; Tebaldi and Knutti, 2018). Information from transient simulations can also be used 42 through an empirical scaling relationship (Seneviratne et al., 2016, 2018; Wartenburger et al., 2017) or using 43 ‘time sampling’ approaches, as described in James et al. (2017). Second, information can be drawn from 44 large ESM ensembles with prescribed SST at particular global warming levels (Mitchell et al., 2017), 45 although an underrepresentation of variability can arise when using prescribed SST temperatures (Fischer et 46 al., 2018a). 47 48 In order to fully derive climate impacts, warming levels will need to be complemented by additional 49 information, such as their associated CO2 concentrations (e.g., fertilization or ocean acidification), 50 composition of the total radiative forcing (aerosols vs greenhouse gases, with varying regional distributions) 51 or socioeconomic conditions (e.g., to estimate societal impacts). More fundamentally, while a global 52 warming level is a good proxy for the state of the climate (Cross-Chapter Box 11.1 in Chapter 11), it does 53 not uniquely define a change in global or regional climate state. For example, regional precipitation 54 responses depend on the details of the individual forcing mechanisms that caused the change (Samset et al., 55 2016), on whether the temperature level is stabilized or transient (King et al., 2020; Zappa et al., 2020), on Do Not Cite, Quote or Distribute 1-111 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 the vertical structure of the troposphere (Andrews et al., 2010), and, in particular, on the global distribution 2 of atmospheric aerosols (Frieler et al., 2012). Another aspect is how Earth system components with century 3 to millennial response timescales, such as long-term sea level rise or permafrost thaw, are affected by global 4 mean warming. For example, sea level rise 50 years after a 1°C warming will be lower than sea level rise 5 150 years after that same 1°C warming (Chapter 9). 6 7 Also, forcing or response patterns that vary in time can create differences in regional climates for the same 8 global mean warming level, or can create non-linearities when scaling patterns from one warming level to 9 another (King et al., 2018), depending on whether near-term transient climate, end of the century, 10 equilibrium climate or climate states after an initial overshoot are considered. 11 12 In spite of these challenges, and thanks to recent methodological advances in quantifying or overcoming 13 them, global warming levels provide a robust and useful integration mechanism. They allow knowledge from 14 various domains within WGI and across the three WGs to be integrated and communicated (Cross Chapter 15 Box 11.1). In this report, Chapters 4, 8, 11, 12 and the Atlas provide information specific to certain warming 16 levels, highlighting the regional differences, but also the approximate scalability of regional climate change, 17 that can arise from even a 0.5°C shift in global-mean temperatures. Furthermore, building on WGI insights 18 into physical climate system responses (Cross-Chapter Box 7.1 in Chapter 7), WGIII will use peak and end- 19 of-century global warming levels to classify a broad set of scenarios. 20 21 22 1.6.3 Cumulative CO2 emissions 23 24 The WGI AR5 (IPCC, 2013a) and the SR1.5 (IPCC, 2018) highlighted the near-linear relationship between 25 cumulative carbon emissions and global mean warming (Section 1.3; Section 5.5). This implies that 26 continued CO2 emissions will cause further warming and changes in all components of the climate system, 27 independent of any specific scenario or pathway. This is captured in the TCRE concept, which relates CO2- 28 induced global mean warming to cumulative carbon emissions (Chapter 5). This Report thus uses cumulative 29 CO2 emissions to compare the climate response across scenarios, and to categorise emission scenarios 30 (Figure 1.29). The advantage of using cumulative CO2 emissions is that it is an inherent emission scenario 31 characteristic rather than an outcome of the scenario-based projections, where uncertainties in the cause- 32 effect chain from emissions to atmospheric concentrations to temperature change are important. 33 34 There is also a close relationship between cumulative total greenhouse gas emissions and cumulative CO2 35 emissions for scenarios in the SR1.5 scenario database (IPCC, 2018; Figure 1.29). The dominance of CO2 36 compared to other well-mixed greenhouse gases (Figure 1.29; Chapter 5, Section 5.2.4) allows policymakers 37 to make use of the carbon budget concept (Chapter 5, Section 5.5) in a policy context, in which GWP- 38 weighted combinations of multiple greenhouse gases are used to define emission targets. A caveat is that 39 cumulative GWP-weighted CO2 equivalent emissions over the next decades do not yield exactly the same 40 temperature outcomes as the same amount of cumulative CO2 emissions, because atmospheric perturbation 41 lifetimes of the various greenhouse gases differ. While carbon budgets are not derived using GWP-weighted 42 emission baskets but rather by explicit modelling of non-CO2 induced warming (Chapter 5, Section 5.5; 43 Cross-Chapter Box 7.1 in Chapter 7), the policy frameworks based on GWP-weighted emission baskets can 44 still make use of the insights from remaining cumulative carbon emissions for different warming levels. 45 46 The same cumulative CO2 emissions could lead to a slightly different level of warming over time (Box 1.4). 47 Rapid emissions followed by steep cuts and potentially net-negative emissions would be characterised by a 48 higher maximum warming and faster warming rate, compared with the same cumulative CO2 emissions 49 spread over a longer period. As further explored in the WGIII assessment, one potential limitation when 50 presenting emission pathway characteristics in cumulative emission budget categories is that path 51 dependencies and lock-in effects (e.g. today’s decisions regarding fossil fuel related infrastructure) play an 52 important role in long-term mitigation strategies (Davis et al., 2010; Luderer et al., 2018). Similarly, high 53 emissions early on might imply strongly net negative emissions (Minx et al., 2018) later on to reach the same 54 cumulative emission and temperature target envelope by the end of the century (Box 1.4). This report 55 explores options to address some of those potential issues from a WGI perspective (see Chapter 5, Sections Do Not Cite, Quote or Distribute 1-112 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 5.5.2 and 5.6.2). 2 3 4 [START FIGURE 1.29 HERE] 5 6 Figure 1.29: The role of CO2 in driving future climate change in comparison to other greenhouse gases (GHGs). 7 The GHGs included here are CH4, N2O, and 40 other long-lived, well-mixed GHGs. The blue shaded area 8 indicates the approximate forcing exerted by CO2 in Shared Socio-economic Pathways (SSP) scenarios, 9 ranging from very low SSP1-1.9 to very high SSP5-8.5 (Chapter 7). The CO2 concentrations under the 10 SSP1-1.9 scenarios reach approximately 350 ppm after 2150, while those of SSP5-8.5 exceed 2000 ppm 11 CO2 in the longer term (through year 2300). Similarly to the dominant radiative forcing share at each 12 point in time (lower area plots), cumulative GWP-100-weighted GHG emissions happen to be closely 13 correlated with cumulative CO2 emissions, allowing policymakers to make use of the carbon budget 14 concept in a policy context with multi-gas GHG baskets as it exhibits relatively low variation across 15 scenarios with similar cumulative emissions until 2050 (inset panel). Further details on data sources and 16 processing are available in the chapter data table (Table 1.SM.1). 17 18 [END FIGURE 1.29 HERE] 19 20 21 [START BOX 1.4 HERE] 22 23 Box 1.4: The relationships between ‘net zero’ emissions, temperature outcomes and carbon dioxide 24 removal 25 Article 4 of the Paris Agreement sets an objective to ‘achieve a balance between anthropogenic emissions by 26 sources and removals by sinks of greenhouse gases’ (Section 1.2). This box addresses the relationship 27 between such a balance and the corresponding evolution of global surface temperature, with or without the 28 deployment of large-scale Carbon Dioxide Removal (CDR), using the definitions of ‘net zero CO2 29 emissions’ and ‘net zero greenhouse gas (GHG) emissions’ of the AR6 Glossary (Annex VII: Glossary). 30 31 ‘Net zero CO2 emissions’ is defined in AR6 as the condition in which anthropogenic CO2 emissions are 32 balanced by anthropogenic CO2 removals over a specified period. Similarly, ‘net zero GHG emissions’ is the 33 condition in which metric-weighted anthropogenic GHG emissions are balanced by metric-weighted 34 anthropogenic GHG removals over a specified period. The quantification of net zero GHG emissions thus 35 depends on the GHG emission metric chosen to compare emissions of different gases, as well as the time 36 horizon chosen for that metric. (For a broader discussion of metrics, see Box 1.3 and Chapter 7, Section 7.6, 37 and WGIII Cross-Chapter Box 2.) 38 39 Technical notes expanding on these definitions can be found as part of their respective entries in the Annex 40 VII: Glossary. The notes clarify the relation between ‘net zero’ CO2 and GHG emissions and the concept of 41 carbon and GHG neutrality, and the metric usage set out in the Paris Rulebook (Decision 18/CMA.1, annex, 42 paragraph 37). 43 44 A global net zero level of CO2, or GHG, emissions will be achieved when the sum of anthropogenic 45 emissions and removals across all countries, sectors, sources and sinks reaches zero. Achieving net zero CO2 46 or GHG emissions globally, at a given time, does not imply that individual entities (i.e., countries, sectors) 47 have to reach net zero emissions at that same point in time, or even at all (see WGIII, TS Box 4 and Chapter 48 3). 49 50 Net zero CO2 and net zero GHG emissions differ in their implications for the subsequent evolution of global 51 surface temperature. Net zero CO2 emissions result in approximately stable CO2-induced warming, but 52 overall warming will depend on any further warming contribution of non-CO2 GHGs. The effect of net zero 53 GHG emissions on global surface temperature depends on the GHG emission metric chosen to aggregate 54 emissions and removals of different gases. For GWP100 (the metric in which Parties to the Paris Agreement 55 have decided to report their aggregated emissions and removals), net zero GHG emissions would generally 56 imply a peak in global surface temperature, followed by a gradual decline (Chapter 7, Section 7.6.2; see also Do Not Cite, Quote or Distribute 1-113 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Chapter 4, Section 4.7.1 regarding the Zero Emission Commitment). However, other anthropogenic factors, 2 such as aerosol emissions or land use-induced changes in albedo, may still affect the climate. 3 4 The definitions of net zero CO2 and GHG should also be seen in relation to the various CDR methods 5 discussed in the context of climate change mitigation (see Chapter 5, Section 5.6, which also includes an 6 assessment of the response of natural sinks to CDR), and how it is employed in scenarios used throughout 7 the WG I and III reports, as described in Section 1.6.1. (See also WGIII, Chapters 3, 7 and 12.) 8 9 For virtually all scenarios assessed by the IPCC, CDR is necessary to reach both global net zero CO2 and net 10 zero GHG emissions, to compensate for residual anthropogenic emissions. This is in part because for some 11 sources of CO2 and non-CO2 emissions, abatement options to eliminate them have not yet been identified. 12 For a given scenario, the choice of GHG metric determines how much net CDR is necessary to compensate 13 for residual non-CO2 emissions, in order to reach net zero GHG emissions (Chapter 7, Section 7.6.2). 14 15 If CDR is further used to go beyond net zero, to a situation with net-negative CO2 emissions (i.e., where 16 anthropogenic removals exceed anthropogenic emissions), anthropogenic CO2-induced warming will 17 decline. A further increase of CDR, until a situation with net zero or even net-negative GHG emissions is 18 reached, would increase the pace at which historical human-induced warming is reversed after its peak 19 (SR1.5, IPCC, 2018). Net-negative anthropogenic GHG emissions may become necessary to stabilize the 20 global surface temperature in the long term, should climate feedbacks further affect natural GHG sinks and 21 sources (see Chapter 5). 22 23 CDR can be achieved through a number of measures (Chapter 5, Section 5.6, and SRCCL). These include 24 additional afforestation, reforestation, soil carbon management, biochar, direct air capture and carbon capture 25 and storage (DACCS), and bioenergy with carbon capture and storage (BECCS) (de Coninck et al., 2018, 26 SR1.5 Ch4; Minx et al., 2018; see also WGIII Chapters 7 and 12). Differences between Land Use, Land Use 27 Change and Forestry (LULUCF) accounting rules, and scientific book-keeping approaches for CO2 28 emissions and removals from the terrestrial biosphere, can result in significant differences between the 29 amount of CDR that is reported in different studies (Grassi et al., 2017). Different measures to achieve CDR 30 come with different risks, negative side effects and potential co-benefits – also in conjunction with 31 sustainable development goals – that can inform choices around their implementation (Fuss et al., 2018; Roe 32 et al., 2019, Chapter 5, Section 5.6). Technologies to achieve direct large-scale anthropogenic removals of 33 non-CO2 GHGs are speculative at present (Yoon et al., 2009; Ming et al., 2016; Kroeger et al., 2017; Jackson 34 et al., 2019). 35 36 [END BOX 1.4 HERE] 37 38 39 1.7 Final remarks 40 41 The assessment in this Report is based on a rapidly growing body of new evidence from the peer-reviewed 42 literature. Recently, scientific climate change research has doubled in output every 5–6 years; the majority of 43 publications deal with issues related to the physical climate system (Burkett et al., 2014; Haunschild et al., 44 2016). The sheer volume of published, peer-reviewed literature on climate change presents a challenge to 45 comprehensive, robust and transparent assessment. 46 47 The enhanced focus on regional climate in WGI AR6 further expands the volume of literature relative to 48 AR5, including non-English language publications sometimes presented as reports (‘grey’ literature), 49 particularly on topics such as regional observing networks and climate services. These factors enhance the 50 challenge of discovering, accessing and assessing the relevant literature. The international, multi-lingual 51 author teams of the IPCC AR6, combined with the open expert review process, help to minimise these 52 concerns, but they remain a challenge. 53 54 Despite the key role of CMIP6 in this Report (Section 1.5), the number of studies evaluating its results and 55 modelling systems remains relatively limited. At the time of publication, additional model results are still Do Not Cite, Quote or Distribute 1-114 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 becoming available. This reflects the need for close temporal alignment of the CMIP cycle with the IPCC 2 assessment process, and the growing complexity of coordinated international modelling efforts. 3 4 Indigenous and local knowledge includes information about past and present climate states. However, 5 assessing this knowledge, and integrating it with the scientific literature, remains a challenge to be met. This 6 lack of assessment capability and integration leads to most WGI chapters still not including Indigenous and 7 local knowledge in their assessment findings. 8 9 Spatial and temporal gaps in both historical and current observing networks, and the limited extent of 10 paleoclimatic archives, have always posed a challenge for IPCC assessments. A relative paucity of long-term 11 observations is particularly evident in Antarctica and in the depths of the ocean. Knowledge of previous 12 cryospheric and oceanic processes is therefore incomplete. Sparse instrumental temperature observations 13 prior to the industrial revolution makes it difficult to uniquely characterize a ‘pre-industrial’ baseline, 14 although this report extends the assessment of anthropogenic temperature change further back in time than 15 previous assessment cycles (Cross-Chapter Box 1.2, Chapter 7). 16 17 Common, integrating scenarios can never encompass all possible events that might induce radiative forcing 18 in the future (Section 1.4). These may include large volcanic eruptions (see Cross-Chapter Box 4.1 in 19 Chapter 4), the consequences of a major meteorite, smoke plumes following a conflict involving nuclear 20 weapons, extensive geo-engineering, or a major pandemic (Cross-Chapter Box 1.6). Scenario-related 21 research also often focuses on the 21st century. Post-2100 climate changes are not covered as 22 comprehensively, and their assessment is limited. Those long-term climate changes, potentially induced by 23 forcing over the 21st century (as in the case of sea level rise), are nevertheless relevant for decision making. 24 25 At the time of publication, the consequences of the COVID-19 pandemic on emissions, atmospheric 26 abundances, radiative forcing and the climate (see Cross-Chapter Box 6.1 in Chapter 6), and on observations 27 (Section 1.5.1), are not yet fully evident. Their assessment in this report is thus limited. 28 29 30 31 32 33 34 35 Do Not Cite, Quote or Distribute 1-115 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 Frequently Asked Questions 2 3 4 FAQ 1.1: Do we understand climate change better now compared to when the IPCC started? 5 Yes, much better. The first IPCC report, released in 1990, concluded that human-caused climate change 6 would soon become evident, but could not yet confirm that it was already happening. Today, evidence is 7 overwhelming that the climate has indeed changed since the pre-industrial era and that human activities are 8 the principal cause of that change. With much more data and better models, we also understand more about 9 how the atmosphere interacts with the ocean, ice, snow, ecosystems and land surfaces of the Earth. 10 Computer climate simulations have also improved dramatically, incorporating many more natural processes 11 and providing projections at much higher resolutions. 12 13 Since the first IPCC report in 1990, large numbers of new instruments have been deployed to collect data in 14 the air, on land, at sea and from outer space. These instruments measure temperature, clouds, winds, ice, 15 snow, ocean currents, sea level, soot and dust in the air, and many other aspects of the climate system. New 16 satellite instruments have also provided a wealth of increasingly fine-grained data. Additional data from 17 older observing systems and even hand-written historical records are still being incorporated into 18 observational datasets, and these datasets are now better integrated and adjusted for historical changes in 19 instruments and measurement techniques. Ice cores, sediments, fossils, and other new evidence from the 20 distant past have taught us much about how Earth’s climate has changed throughout its history. 21 22 Understanding of climate system processes has also improved. For example, in 1990 very little was known 23 about how the deep ocean responds to climate change. Today, reconstructions of deep ocean temperatures 24 extend as far back as 1871. We now know that the oceans absorb most of the excess energy trapped by 25 greenhouse gases and that even the deep ocean is warming up. As another example, in 1990, relatively little 26 was known about exactly how or when the gigantic ice sheets of Greenland and Antarctica would respond to 27 warming. Today, much more data and better models of ice sheet behaviour reveal unexpectedly high melt 28 rates that will lead to major changes within this century, including substantial sea level rise (see FAQ 9.2). 29 30 The major natural factors contributing to climate change on time scales of decades to centuries are volcanic 31 eruptions and variations in the sun’s energy output. Today, data show that changes in incoming solar energy 32 since 1900 have contributed only slightly to global warming, and they exhibit a slight downward trend since 33 the 1970s. Data also show that major volcanic eruptions have sometimes cooled the entire planet for 34 relatively short periods of time (typically several years) by erupting aerosols (tiny airborne particles) high 35 into the atmosphere. 36 The main human causes of climate change are the heat-absorbing greenhouse gases released by fossil fuel 37 combustion, deforestation, and agriculture, which warm the planet, and aerosols such as sulphate from 38 burning coal, which have a short-term cooling effect that partially counteracts human-caused warming. Since 39 1990, we have more and better observations of these human factors as well as improved historical records, 40 resulting in more precise estimates of human influences on the climate system (see FAQ 3.1). 41 42 While most climate models in 1990 focused on the atmosphere, using highly simplified representations of 43 oceans and land surfaces, today’s Earth system simulations include detailed models of oceans, ice, snow, 44 vegetation and many other variables. An important test of models is their ability to simulate Earth’s climate 45 over the period of instrumental records (since about 1850). Several rounds of such testing have taken place 46 since 1990, and the testing itself has become much more rigorous and extensive. As a group and at large 47 scales, models have predicted the observed changes well in these tests (see FAQ 3.3). Since there is no way 48 to do a controlled laboratory experiment on the actual Earth, climate model simulations can also provide a 49 kind of ‘alternate Earth’ to test what would have happened without human influences. Such experiments 50 show that the observed warming would not have occurred without human influence. . 51 52 Finally, physical theory predicts that human influences on the climate system should produce specific 53 patterns of change, and we see those patterns in both observations and climate simulations. For example, 54 nights are warming faster than days, less heat is escaping to space, and the lower atmosphere (troposphere) is 55 warming but the upper atmosphere (stratosphere) has cooled. These confirmed predictions are all evidence of Do Not Cite, Quote or Distribute 1-116 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 changes driven primarily by increases in greenhouse gas concentrations rather than natural causes. 2 3 4 [START FAQ 1.1, FIGURE 1 HERE] 5 6 FAQ 1.1, Figure 1: Sample elements of climate understanding, observations and models as assessed in the IPCC First 7 Assessment Report (1990) and Sixth Assessment Report (2021). Many other advances since 1990, such as key aspects 8 of theoretical understanding, geological records and attribution of change to human influence, are not included in this 9 figure because they are not readily represented in this simple format. Fuller explications of the history of climate 10 knowledge are available in the introductory chapters of the IPCC Fourth and Sixth Assessment Reports. 11 12 [END FAQ 1.1, FIGURE 1 HERE] 13 14 15 FAQ 1.2: Where is climate change most apparent? 16 17 The signs of climate change are unequivocal at the global scale and are increasingly apparent on smaller 18 spatial scales. The high northern latitudes show the largest temperature increase with clear effects on sea 19 ice and glaciers. The warming in the tropical regions is also apparent because the natural year-to-year 20 variations in temperature there are small. Long-term changes in other variables such as rainfall and some 21 weather and climate extremes have also now become apparent in many regions. 22 23 It was first noticed that the planet’s land areas were warming in the 1930s. Although increasing atmospheric 24 carbon dioxide concentrations were suggested as part of the explanation, it was not certain at the time 25 whether the observed warming was part of a long-term trend or a natural fluctuation – global warming had 26 not yet become apparent. But the planet continued to warm, and by the 1980s the changes in temperature had 27 become obvious or, in other words, the signal had emerged. 28 29 Imagine you had been monitoring temperatures at the same location for the past 150 years. What would you 30 have experienced? When would the warming have become noticeable in your data? The answers to these 31 questions depend on where on the planet you are. 32 33 Observations and climate model simulations both demonstrate that the largest long-term warming trends are 34 in the high northern latitudes and the smallest warming trends over land are in tropical regions. However, the 35 year-to-year variations in temperature are smallest in the tropics, meaning that the changes there are also 36 apparent, relative to the range of past experiences (see FAQ 1.2, Figure 1). 37 38 Changes in temperature also tend to be more apparent over land areas than over the open ocean and are often 39 most apparent in regions which are more vulnerable to climate change. It is expected that future changes will 40 continue to show the largest signals at high northern latitudes, but with the most apparent warming in the 41 tropics. The tropics also stand to benefit the most from climate change mitigation in this context, as limiting 42 global warming will also limit how far the climate shifts relative to past experience. 43 44 Changes in other climate variables have also become apparent at smaller spatial scales. For example, 45 changes in average rainfall are becoming clear in some regions, but not in others, mainly because natural 46 year-to-year variations in precipitation tend to be large relative to the magnitude of the long-term trends. 47 However, extreme rainfall is becoming more intense in many regions, potentially increasing the impacts 48 from inland flooding (see FAQ 8.2). Sea levels are also clearly rising on many coastlines, increasing the 49 impacts of inundation from coastal storm surges, even without any increase in the number of storms reaching 50 land. A decline in the amount of Arctic sea ice is apparent, both in the area covered and in its thickness, with 51 implications for polar ecosystems. 52 53 When considering climate-related impacts, it is not necessarily the size of the change which is most 54 important. Instead, it can be the rate of change or it can also be the size of the change relative to the natural 55 variations of the climate to which ecosystems and society are adapted. As the climate is pushed further away Do Not Cite, Quote or Distribute 1-117 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 from past experiences and enters an unprecedented state, the impacts can become larger, along with the 2 challenge of adapting to them. 3 4 How and when a long-term trend becomes distinguishable from shorter-term natural variations depends on 5 the aspect of climate being considered (e.g., temperature, rainfall, sea ice or sea level), the region being 6 considered, the rate of change, and the magnitude and timing of natural variations. When assessing the local 7 impacts from climate change, both the size of the change and the amplitude of natural variations matter. 8 9 10 [START FAQ 1.2, FIGURE 1 HERE] 11 12 FAQ 1.2, Figure 1: Observed variations in regional temperatures since 1850 (data from Berkeley Earth). Regions in 13 high latitudes, such as mid-North America (40ºN–64ºN, 140ºW–60ºW, left), have warmed by a larger amount than 14 regions at lower latitudes, such as Tropical South America (10ºS–10ºN, 84ºW–16ºW, right), but the natural variations 15 are also much larger at high latitudes (darker and lighter shading represents 1 and 2 standard deviations, respectively, of 16 natural year-to-year variations). The signal of observed temperature change emerged earlier in Tropical South America 17 than mid-North America even though the changes were of a smaller magnitude. (Note that those regions were chosen 18 because of the longer length of their observational record, see Figure 1.14 for more regions). 19 20 [END FAQ 1.2, FIGURE 1 HERE] 21 22 23 FAQ 1.3: What can past climate teach us about the future? 24 25 In the past, the Earth has experienced prolonged periods of elevated greenhouse gas concentrations that 26 caused global temperatures and sea levels to rise. Studying these past warm periods informs us about the 27 potential long-term consequences of increasing greenhouse gases in the atmosphere. 28 29 Rising greenhouse gas concentrations are driving profound changes to the Earth system, including global 30 warming, sea level rise, increases in climate and weather extremes, ocean acidification, and ecological shifts 31 (see FAQ 2.2, FAQ 7.1). The vast majority of instrumental observations of climate began during the 20th 32 century, when greenhouse gas emissions from human activities became the dominant driver of changes in 33 Earth’s climate (see FAQ 3.1). 34 35 As scientists seek to refine our understanding of Earth’s climate system and how it may evolve in coming 36 decades to centuries, past climate states provide a wealth of insights. Data about these past states help to 37 establish the relationship between natural climate drivers and the history of changes in global temperature, 38 global sea levels, the carbon cycle, ocean circulation, and regional climate patterns, including climate 39 extremes. Guided by such data, scientists use Earth system models to identify the chain of events underlying 40 the transitions between past climatic states (see FAQ 3.3). This is important because during present-day 41 climate change, just as in past climate changes, some aspects of the Earth system (e.g., surface temperature) 42 respond to changes in greenhouse gases on a time scale of decades to centuries, while others (e.g., sea level 43 and the carbon cycle) respond over centuries to millennia (see FAQ 5.3). In this way, past climate states 44 serve as critical benchmarks for climate model simulations, improving our understanding of the sequences, 45 rates, and magnitude of future climate change over the next decades to millennia. 46 47 Analyzing previous warm periods caused by natural factors can help us understand how key aspects of the 48 climate system evolve in response to warming. For example, one previous warm-climate state occurred 49 roughly 125,000 years ago, during the Last Interglacial period, when slight variations in the Earth’s orbit 50 triggered a sequence of changes that caused about 1°C–2°C of global warming and about 2–8 m of sea level 51 rise relative to the 1850-1900, even though atmospheric carbon dioxide concentrations were similar to 1850- 52 1900 values (FAQ 1.3, Figure 1). Modelling studies highlight that increased summer heating in the higher 53 latitudes of the Northern Hemisphere during this time caused widespread melting of snow and ice, reducing 54 the reflectivity of the planet and increasing the absorption of solar energy by the Earth’s surface. This gave 55 rise to global-scale warming, which led in turn to further ice loss and sea level rise. These self-reinforcing 56 positive feedback cycles are a pervasive feature of Earth’s climate system, with clear implications for future Do Not Cite, Quote or Distribute 1-118 Total pages: 215 Final Government Distribution Chapter 1 IPCC AR6 WGI 1 climate change under continued greenhouse gas emissions. In the case of sea level rise, these cycles evolved 2 over several centuries to millennia, reminding us that the rates and magnitude of sea level rise in the 21st 3 century are just a fraction of the sea level rise that will ultimately occur after the Earth system fully adjusts to 4 current levels of global warming. 5 6 Roughly 3 million years ago, during the Pliocene Epoch, the Earth witnessed a prolonged period of elevated 7 temperatures (2.5°C–4°C higher than 1850-1900) and higher sea levels (5–25 m higher than 1850-1900), in 8 combination with atmospheric carbon dioxide concentrations similar to present-day. The fact that Pliocene 9 atmospheric carbon dioxide concentrations were similar to present, while global temperatures and sea levels 10 were significantly higher, reflects the difference between an Earth system that has fully-adjusted to changes 11 in natural drivers (the Pliocene) and one where greenhouse gases concentrations, temperature, and sea level 12 rise are still increasing (present-day). Much about the transition into the Pliocene climate state – in terms of 13 key causes, the role of cycles that hastened or slowed the transition, and the rate of change in climate 14 indicators such as sea level – remain topics of intense study by climate researchers using a combination of 15 paleoclimate observations and Earth system models. Insights from such studies may help to reduce the large 16 uncertainties around estimates of global sea level rise by 2300, which range from 0.3 m to 3 m above 1850- 17 1900 (in a low-emissions scenario) to as much as 16 m higher than 1850-1900 (in a very high-emissions 18 scenario that includes accelerating structural disintegration of the polar ice sheets). 19 20 While present-day warming is unusual in the context of the recent geologic past in several different ways 21 (see FAQ 2.1), past warm climate states present a stark reminder that the long-term adjustment to present- 22 day atmospheric carbon dioxide concentrations has only just begun. That adjustment will continue over the 23 coming centuries to millennia. 24 25 26 [START FAQ 1.3, FIGURE 1 HERE] 27 28 FAQ 1.3, Figure 1: Comparison of past, present and future. Schematic of atmospheric carbon dioxide 29 concentrations, global temperature, and global sea level during previous warm periods as compared to 1850-1900, 30 present-day (2011-2020), and future (2100) climate change scenarios corresponding to low-emissions scenarios (SSP1- 31 2.6; lighter colour bars) and very high emissions scenarios (SSP5-8.5; darker colour bars). 32 33 34 [END FAQ 1.3, FIGURE 1 HERE] 35 36 37 Do Not Cite, Quote or Distribute 1-119 Total pages: 215