Editing IPCC:Wg1:Chapter:Chapter-1-comments (section)

=== 1.5.3 Climate Models ===

<div id="h2-29-siblings" class="h2-siblings"></div>

A wide range of numerical models is widely used in climate science to study the climate system and its behaviour across multiple temporal and spatial scales. These models are the main tools available to look ahead into possible climate futures under a range of scenarios ( [[#1.6|Section 1.6]] ). Global Earth system models (ESMs) are the most complex models that contribute to AR6. At the core of each ESM is a GCM (general circulation model) representing the dynamics of the atmosphere and ocean. ESMs are complemented by regional models (Section 10.3.1) and by a hierarchy of models of lower complexity. This section summarizes major developments in these different types of models since AR5. Past IPCC reports have made use of multi-model ensembles generated through various phases of the World Climate Research Programme (WCRP) Coupled Model Intercomparison Project (CMIP). Analysis of the latest CMIP Phase 6 (CMIP6; [[#Eyring--2016|Eyring et al., 2016]] ) simulations constitute a key line of evidence supporting this Assessment Report ( [[#1.5.4|Section 1.5.4]] ). The key characteristics of models participating in CMIP6 are listed in Annex II: Models.

<div id="1.5.3.1" class="h3-container"></div>

<span id="earth-system-models"></span>
==== 1.5.3.1 Earth System Models ====

<div id="h3-29-siblings" class="h3-siblings"></div>

Earth system models are mathematical formulations of the natural laws that govern the evolution of climate-relevant systems: atmosphere, ocean, cryosphere, land, and biosphere, as well as the carbon cycle ( [[#Flato--2011|Flato, 2011]] ). They build on the fundamental laws of physics (e.g., Navier–Stokes or Clausius–Clapeyron equations) or empirical relationships established from observations and, when possible, they are constrained by fundamental conservation laws (e.g., mass and energy). The evolution of climate-relevant variables is computed numerically using high-performance computers ( [[#André--2014|André et al., 2014]] ; [[#Balaji--2017|Balaji et al., 2017]] ), on three-dimensional discrete grids ( [[#Staniforth--2012|Staniforth and Thuburn, 2012]] ). The spatial (and temporal) resolution of these grids in both the horizontal and vertical directions determines which processes need to be parameterized or whether they can be explicitly resolved. Developments since AR5 in model resolution, parameterizations and modelling of the land and ocean biosphere and of biogeochemical cycles are discussed below.

<div id="1.5.3.1.1" class="h4-container"></div>

<span id="model-grids-and-resolution"></span>
===== ''1.5.3.1.1 Model grids'' ''and resolution'' =====

<div id="h4-10-siblings" class="h4-siblings"></div>

The horizontal resolution and the number of vertical levels in ESMs is generally higher in CMIP6 than in CMIP5 (Figure 1.19). Global models with finer horizontal grids better represent many aspects of the circulation of the atmosphere ( [[#Gao--2020|Gao et al., 2020]] ; [[#Schiemann--2020|Schiemann et al., 2020]] ) and ocean ( [[#Bishop--2016|Bishop et al., 2016]] ; [[#Storkey--2018|Storkey et al., 2018]] ), bringing improvements in the simulation of the global hydrological cycle ( [[#Roberts--2018|Roberts et al., 2018]] ). CMIP6 includes a dedicated effort (HighResMIP, [[#Haarsma--2016|Haarsma et al., 2016]] ) to explore the effect of higher horizontal resolution, such as ~50 km, ~25 km and even ~10 km ( [[#1.5.4.2|Section 1.5.4.2]] and Annex II, Table AII.6). Improvements are documented in the highest-resolution coupled models used for HighResMip ( [[#Hewitt--2017|]] [[#Hewitt--2017|Hewitt et al., 2017]] ; [[#Roberts--2019|Roberts et al., 2019]] ). Flexible grids allowing spatially variable resolution in the atmosphere ( [[#McGregor--2015|McGregor, 2015]] ; [[#Giorgetta--2018|Giorgetta et al., 2018]] ) and in the ocean ( [[#Wang--2014|Wang et al., 2014]] ; [[#Petersen--2019|Petersen et al., 2019]] ) are more widely used than at the time of the AR5.

<div id="_idContainer057" class="_idGenObjectStyleOverride-1"></div>

<!-- START IMG -->
<!-- IMG FILE -->
[[File:336d8e067ca4415dae38e7aaf9eb07bf IPCC_AR6_WGI_Figure_1_19.png]]
<!-- IMG TITLE + CAPTION -->

'''Figure 1.19 |''' '''Resolution of the atmospheric and oceanic components of global climate models participating in CMIP5, CMIP6 and HighResMIP:''' '''(a, b)''' horizontal resolution (km), and '''(c, d)''' number of vertical levels. Darker-colour circles indicate high-top models (in which the top of the atmosphere is above 50 km). The crosses are the median values. These models are documented in Annex II. Note that duplicated models in a modelling group are counted as one entry when their horizontal and vertical resolutions are the same. For HighResMIP, one atmosphere–ocean coupled model with the highest resolution from each modelling group is used. The horizontal resolution (rounded to 10 km) is the square root of the surface area of the Earth divided by the number of grid points, or the area of the ocean surface divided by the number of surface ocean grid points, for the atmosphere and ocean, respectively.
<!-- END IMG -->

The number of vertical levels in the atmosphere of global models has increased (Figure 1.19), partly to enable simulations to include higher levels in the atmosphere and better represent stratospheric processes ( [[#Charlton-Perez--2013|Charlton-Perez et al., 2013]] ; [[#Kawatani--2019|Kawatani et al., 2019]] ). Half the modelling groups now use ‘high-top’ models with a top level above the stratopause (a pressure of about 1 hPa). The number of vertical levels in the ocean models has also increased in order to achieve finer resolution over the water column and especially in the upper mixed layer and to better resolve the diurnal cycle ( [[IPCC:Wg1:Chapter:Chapter-3#3.5|Section 3.5]] and Annex II; [[#Bernie--2008|Bernie et al., 2008]] ).

Despite the documented progress of higher resolution, the model evaluation carried out in subsequent chapters shows that improvements between CMIP5 and CMIP6 remain modest at the global scale ( [[IPCC:Wg1:Chapter:Chapter-3#3.8.2|Section 3.8.2]] ; [[#Bock--2020|Bock et al., 2020]] ). Lower resolution alone does not explain all model biases, for example, a low blocking frequency ( [[#Davini--2020|Davini and D’Andrea, 2020]] ) or a wrong shape of the Intertropical Convergence Zone ( [[#Tian--2020|Tian and Dong, 2020]] ). Model performance depends on model formulation and parameterizations as much as on resolution (Chapters 3, 8 and 10).

<div id="1.5.3.1.2" class="h4-container"></div>

<span id="representation-of-physical-and-chemical-processes-in-esms"></span>
===== 1.5.3.1.2 Representation of physical and chemical processes in ESMs =====

<div id="h4-11-siblings" class="h4-siblings"></div>

Atmospheric models include representations of physical processes such as clouds, turbulence, convection and gravity waves that are not fully represented by grid-scale dynamics. The CMIP6 models have undergone updates in some of their parameterization schemes compared to their CMIP5 counterparts, with the aim of better representing the physics and bringing the climatology of the models closer to newly available observational datasets. Most notable developments are to schemes involving radiative transfer, cloud microphysics, and aerosols, in particular a more explicit representation of the aerosol indirect effects through aerosol-induced modification of cloud properties. Broadly, aerosol–cloud microphysics has been a key topic for the aerosol and chemistry modelling communities since AR5, leading to improved understanding of the climate influence of short-lived climate forcers, but they remain the single largest source of spread in ESM calculations of climate sensitivity ( [[#Meehl--2020|Meehl et al., 2020]] ), with numerous parameterization schemes in use (Section 6.4; [[#Gettelman--2016|Gettelman and Sherwood, 2016]] ; [[#Zhao--2018|Zhao et al., 2018]] ; [[#Gettelman--2019|Gettelman et al., 2019]] ). The treatment of droplet size and mixed-phase clouds (liquid and ice) was found to lead to changes in the climate sensitivity (Glossary) of some models between AR5 and AR6 (Section 7.4; [[#Bodas-Salcedo--2019|Bodas-Salcedo et al., 2019]] ; [[#Gettelman--2019|Gettelman et al., 2019]] ; [[#Zelinka--2020|Zelinka et al., 2020]] ).

The representation of ocean and cryosphere processes has also evolved significantly since CMIP5. The explicit representation of ocean eddies, due to increased grid resolution (typically, from 1° to ¼°), is a major advance in a number of CMIP6 ocean model components ( [[#Hewitt--2017|]] [[#Hewitt--2017|Hewitt et al., 2017]] ). Advances in sea ice models have been made, for example through correcting known shortcomings in CMIP5 simulations, in particular the persistent underestimation of the rapid decline in summer Arctic sea ice extent ( [[#Rosenblum--2016|Rosenblum and Eisenman, 2016]] , 2017; [[#Turner--2017|Turner and Comiso, 2017]] ; [[#Notz--2018|Notz and Stroeve, 2018]] ). The development of glacier and ice-sheet models has been motivated and guided by an improved understanding of key physical processes, including grounding line dynamics, stratigraphy and microstructure evolution, sub-shelf melting, and glacier and ice-shelf calving, among others ( [[#Faria--2014|Faria et al., 2014]] , 2018; [[#Hanna--2020|Hanna et al., 2020]] ). The resolution of ice-sheet models has continuously increased, including the use of nested grids, sub-grid interpolation schemes, and adaptive mesh approaches ( [[#Cornford--2016|Cornford et al., 2016]] ), mainly for a more accurate representation of grounding-line migration and data assimilation ( [[#Pattyn--2018|Pattyn, 2018]] ). Ice-sheet models are increasingly interactively coupled with global and regional climate models, accounting for the height–mass-balance feedback ( [[#Vizcaino--2015|Vizcaino et al., 2015]] ; [[#Le%20clec’h--2019|Le clec’h et al., 2019]] ), and enabling a better representation of ice-ocean processes, in particular for the Antarctic Ice Sheet ( [[#Asay-Davis--2017|Asay-Davis et al., 2017]] ).

Sealevel rise is caused by multiple processes acting on multiple time scales: ocean warming, glaciers and ice-sheet melting, change in water storage on land, and glacial isostatic adjustment (Box 9.1) but no single model can represent all these processes (Section 9.6). In this Report, the contributions are computed separately (Figure 9.28) and merged into a common probabilistic framework and updated from AR5 (Section 9.6; [[#Church--2013|Church et al., 2013]] ; [[#Kopp--2014|Kopp et al., 2014]] ).

Another notable development since AR5 is the inclusion of stochastic parameterizations of sub-grid processes in some comprehensive climate models ( [[#Sanchez--2016|Sanchez et al., 2016]] ). Here, the deterministic differential equations that govern the dynamical evolution of the model are complemented by knowledge of the stochastic variability in unresolved processes. While not yet widely implemented, the approach has been shown to improve the forecasting skill of weather models, to reduce systematic biases in global models ( [[#Berner--2017|Berner et al., 2017]] ; [[#Palmer--2019|Palmer, 2019]] ) and to influence simulated climate sensitivity ( [[#Strommen--2019|Strommen et al., 2019]] ).

<div id="1.5.3.1.3" class="h4-container"></div>

<span id="representation-of-biogeochemistry-including-the-carbon-cycle"></span>
===== 1.5.3.1.3 Representation of biogeochemistry, including the carbon cycle =====

<div id="h4-12-siblings" class="h4-siblings"></div>

Since AR5, more sophisticated land-use and land-cover change representations in ESMs have been developed to simulate the effects of land management on surface fluxes of carbon, water and energy ( [[#Lawrence--2016|Lawrence et al., 2016]] ), although the integration of many processes (e.g., wetland drainage, fire as a management tool) remains a challenge ( [[#Pongratz--2018|Pongratz et al., 2018]] ). The importance of nitrogen availability to limit the terrestrial carbon sequestration has been recognized (Section 5.4; [[#Zaehle--2014|Zaehle et al., 2014]] ) and so an increasing number of models now include a prognostic representation of the terrestrial nitrogen cycle and its coupling to the land carbon cycle ( [[#Jones--2016|Jones et al., 2016]] ; [[#Arora--2020|Arora et al., 2020]] ), leading to a reduction in uncertainty for carbon budgets (Section 5.1; [[#Jones--2020|Jones and Friedlingstein, 2020]] ). As was the case in CMIP5 ( [[#Ciais--2013|Ciais et al., 2013]] ), the land surface processes represented vary across CMIP6 models, with at least some key processes (fire, permafrost carbon, microbes, nutrients, vegetation dynamics, plant demography) absent from any particular ESM land model (Table 5.4). Ocean biogeochemical models have evolved to enhance the consistency of the exchanges between ocean, atmosphere and land, through riverine input and dust deposition ( [[#Stock--2014|Stock et al., 2014]] ; [[#Aumont--2015|Aumont et al., 2015]] ). Other developments include flexible plankton stoichiometric ratios ( [[#Galbraith--2015|Galbraith and Martiny, 2015]] ), improvements in the representation of nitrogen fixation ( [[#Paulsen--2017|Paulsen et al., 2017]] ), and the limitation of plankton growth by iron ( [[#Aumont--2015|Aumont et al., 2015]] ). Due to the long time scale of biogeochemical processes, how the models are initialized (spun up) strategies has been shown to affect their performance in AR5 ( [[#Séférian--2016|Séférian et al., 2016]] ).

<div id="1.5.3.2" class="h3-container"></div>

<span id="model-tuning-and-adjustment"></span>
==== 1.5.3.2 Model Tuning and Adjustment ====

<div id="h3-30-siblings" class="h3-siblings"></div>

When developing climate models, choices have to be made in a number of areas. Besides model formulation and resolution, parameterizations of unresolved processes also involve many choices as, for each of these, several parameters can be set. The acceptable range for these parameters is set by mathematical consistency (e.g., convergence of a numerical scheme), physical considerations (e.g., energy conservation), observations, or a combination of factors. Model developers choose a set of parameters that both falls within this range and mimics observations of individual processes or their statistics.

An initial set of such choices is usually made by (often extensive) groups of modellers working on individual components of the Earth system (e.g., ocean, atmosphere, land or sea ice). As components are assembled to build an ESM, the choices are refined so that the simulated climate best represents a number of pre-defined climate variables, or ‘tuning targets’. When these are met the model is released for use in intercomparisons such as CMIP. Tuning targets can be one of three types: mean climate; regional phenomena and features; or historical trends ( [[#Hourdin--2017|Hourdin et al., 2017]] ). One example of such a goal is that when the simulated climate system receives energy from the sun in accordance with what we observe today, the resulting mean equilibrium temperature should also be close to observations. Whether tuning should be performed to facilitate accurate simulation of long-term trends such as changes in global mean temperature over the historical era, or rather be performed for each process independently such that all collective behaviour is emergent, is an open question ( [[#Schmidt--2017|Schmidt et al., 2017]] ; [[#Burrows--2018|Burrows et al., 2018]] ).

Each modelling group has its own strategy and, after AR5, a survey was conducted to understand the tuning approach used in 23 CMIP5 modelling centres. The results are discussed in [[#Hourdin--2017|Hourdin et al. (2017)]] , which stresses that the behaviour of ESMs depends on the tuning strategy. An important recommendation is that the calibration steps that lead to particular model tuning should be carefully documented. In CMIP6 each modelling group now describes the three levels of tuning, both for the complete ESM and for the individual components (available at https://explore.es-doc.org and in the published model descriptions, Annex II: Models). The most important global tuning target for CMIP6 models is the net top-of-the-atmosphere (TOA) heat flux and its radiative components. Other global targets include: the decomposition of the energy fluxes at TOA into a clear sky component and a component due to the radiative effect of clouds, global mean air and ocean temperature, sea ice extent, sea ice volume, glacial mass balance, and the global root mean square error of precipitation. The TOA heat flux balance is achieved using a diversity of approaches, usually unique to each modelling group. Adjustments are made for parameters associated with uncertain or poorly constrained processes ( [[#Schmidt--2017|Schmidt et al., 2017]] ), for example the aerosol indirect effects, adjustments to ocean albedo, marine dimethyl sulfide (DMS) parameterization, or cloud properties ( [[#Mauritsen--2020|Mauritsen and Roeckner, 2020]] ).

Regional tuning targets include: the AMOC, the Southern Ocean circulation, and temperature profiles in ocean basins ( [[#Golaz--2019|Golaz et al., 2019]] ; [[#Sellar--2019|Sellar et al., 2019]] ); regional land properties and precipitations ( [[#Mauritsen--2019|Mauritsen et al., 2019]] ; [[#Yukimoto--2019|Yukimoto et al., 2019]] ) ; latitudinal distribution of radiation ( [[#Boucher--2020|Boucher et al., 2020]] ); spatial contrasts in TOA radiative fluxes or surface fluxes; and stationary waves in the Northern Hemisphere ( [[#Schmidt--2017|Schmidt et al., 2017]] ; [[#Yukimoto--2019|Yukimoto et al., 2019]] ).

Even with some core commonalities of approaches to model tuning, practices can differ, such as the use of initial drift from initialized forecasts, the explicit use of the transient observed record for the historical period, or the use of the present-day radiative imbalance at the TOA as a tuning target rather than an equilibrated pre-industrial balance. The majority of CMIP6 modelling groups report that they do not tune their model for the observed trends during the historical period (23 out of 29 groups), nor for ECS (25 out of 29). ECS and TCR are thus emergent properties for a large majority of models. The effect of tuning on model skill and ensemble spread in CMIP6 is further discussed in [[IPCC:Wg1:Chapter:Chapter-3#3.3|Section 3.3]] .

<div id="1.5.3.3" class="h3-container"></div>

<span id="from-global-to-regional-models"></span>
==== 1.5.3.3 From Global to Regional Models ====

<div id="h3-31-siblings" class="h3-siblings"></div>

The need for accurate climate information at the regional scale is increasing (Section 10.1). High-resolution global climate models, such as those taking part in HighResMIP, provide more detailed information at the regional scale ( [[#Roberts--2018|Roberts et al., 2018]] ). However, due to the large computational resources required by these models, only a limited number of simulations per model are available. In addition to CMIP global models, regional information can be derived using regional climate models (RCMs) and downscaling techniques, presented in [https://www.ipcc.ch/report/ar6/wg1/chapter/chapter-10 Chapter 10] and the Atlas. RCMs are dynamical models, similar to GCMs, that simulate a limited region and are forced with boudary conditions from a global simulation, often correcting for biases (Section 10.3, Cross-Chapter Box 10.2 and Annex II). This approach allows the use of a higher resolution within the chosen domain, and thus better represents important drivers of regional climate such as mountain ranges, land management and urban effects. RCMs resolving atmospheric convection explicitly are now included in intercomparisons ( [[#Coppola--2020|Coppola et al., 2020]] ) and are used in Chapters 10, 11 and 12. Other approaches, such as statistical downscaling, are also used to generate regional climate projections (Section 10.3; [[#Maraun--2018|Maraun and Widmann, 2018]] ).

The number of climate centres or consortia that carry out global climate simulations and projections has grown from 11 in the first CMIP to 19 in CMIP5 and 28 for CMIP6 ( [[#1.5.4.2|Section 1.5.4.2]] and Annex II). Regional climate models participating in the Coordinated Regional Downscaling Experiment (CORDEX) are more diverse than the global ESMs ( [[#1.5.4.3|Section 1.5.4.3]] and Annex II) and engage an even wider international community (Figure 1.20).

<div id="_idContainer059" class="_idGenObjectStyleOverride-1"></div>

<!-- START IMG -->
<!-- IMG FILE -->
[[File:a514699cf882e9bac13c3ca48d0a4efa IPCC_AR6_WGI_Figure_1_20.png]]
<!-- IMG TITLE + CAPTION -->

'''Figure 1.20 |''' '''World map showing the increased diversity of modelling centres contributing to CMIP and CORDEX.''' Climate models are often developed by international consortia. One such consortium, EC-Earth, is shown as an example under the label '''8 EU Cities''' (involving SMHI, Sweden; KNMI, The Netherlands; DMI, Denmark; AEMET, Spain; Met Éireann, Ireland; CNR‐ISAC, Italy; Instituto de Meteorologia, Portugal; and FMI, Finland). There are too many such collaborations to display all of them on this map. More complete information about institutions contributing to CORDEX and CMIP6 is found in Annex II.
<!-- END IMG -->

<div id="1.5.3.4" class="h3-container"></div>

<span id="models-of-lower-complexity"></span>
==== 1.5.3.4 Models of Lower Complexity ====

<div id="h3-32-siblings" class="h3-siblings"></div>

'''Earth system models of intermediate complexity''' (EMICs) complement the model hierarchy and fill the gap between conceptual, simple climate models and complex GCMs or ESMs ( [[#Claussen--2002|Claussen et al., 2002]] ). EMICs are simplified; they include processes in a more parameterized, rather than explicitly calculated, form and generally have lower spatial resolution compared to the complex ESMs. As a result, EMICs require much less computational resource and can be integrated for many thousands of years without supercomputers ( [[#Hajima--2014|Hajima et al., 2014]] ). The range of EMICs used in climate change research is highly heterogeneous, ranging from zonally averaged or mixed-layer ocean models coupled to statistical-dynamical models of the atmosphere, to low-resolution three-dimensional ocean models coupled to simplified dynamical models of the atmosphere. An increasing number of EMICs include interactive representations of the global carbon cycle, with varying levels of complexity and numbers of processes considered ( [[#Plattner--2008|Plattner et al., 2008]] ; [[#Zickfeld--2013|Zickfeld et al., 2013]] ; [[#MacDougall--2020|MacDougall et al., 2020]] ). Given the heterogeneity of the EMIC community, modellers tend to focus on specific research questions and develop individual models accordingly. As for any type of models assessed in this Report, the set of EMICs undergoes thorough evaluation and fit-for-purpose testing before being applied to address specific climate aspects.

EMICs have been used extensively in past IPCC reports, providing long-term integrations on paleoclimate and future time scales, including stabilization pathways and a range of commitment scenarios, with perturbed physics ensembles and sensitivity studies, or with simulations targeting the uncertainty in global climate–carbon cycle systems (e.g., [[#Meehl--2007b|Meehl et al., 2007b]] ; [[#Collins--2013|Collins et al., 2013]] ). More recently, a number of studies have pointed to the possibility of systematically different climate responses to external forcings in EMICs and complex ESMs ( [[#Frölicher--2015|Frölicher and Paynter, 2015]] ; [[#Pfister--2017|Pfister and Stocker, 2017]] , 2018) that need to be considered in the context of this report. For example, [[#Frölicher--2015|Frölicher and Paynter (2015)]] showed that EMICs have a higher simulated realized warming fraction (i.e., the TCR/ECS ratio) than CMIP5 ESMs and speculated that this may bias the temperature response to zero carbon emissions. But, in a recent comprehensive multi-model analysis of the zero CO <sub>2</sub> emissions commitment, [[#MacDougall--2020|MacDougall et al. (2020)]] did not find any significant differences between EMICs and ESMs in committed temperatures 90 years after halting emissions. While some EMICs contribute to parts of the CMIP6-endorsed MIPs, a coordinated EMICs modelling effort similar to those carried out for AR4 ( [[#Plattner--2008|Plattner et al., 2008]] ) and AR5 ( [[#Eby--2013|Eby et al., 2013]] ; [[#Zickfeld--2013|Zickfeld et al., 2013]] ) is not in place for IPCC AR6; however, EMICs are assessed in a number of chapters. For example, Chapters 4 and 5 use EMICs in the assessment of long-term climate change beyond 2100 (Section 5.5); zero-emissions commitments, overshoot and recovery ( [[IPCC:Wg1:Chapter:Chapter-4#4.7|Section 4.7]] ); consequences of CO <sub>2</sub> removal (CDR) on the climate system and the carbon cycle (Sections 4.6 and 5.6); and long-term carbon cycle–climate feedbacks (Section 5.4).

'''Physical emulators and simple climate models''' make up a broad class of heavily parametrized models designed to reproduce the responses of the more complex, process-based models, and provide rapid translations of emissions, via concentrations and radiative forcing, into probabilistic estimates of changes to the physical climate system. The main application of emulators is to extrapolate insights from ESMs and observational constraints to a larger set of emissions scenarios (Cross-Chapter Box 7.1). The computational efficiency of various emulating approaches opens new analytical possibilities, given that ESMs take a lot of computational resources for each simulation. The applicability and usefulness of emulating approaches are however constrained by their skill in capturing the global mean climate responses simulated by the ESMs (mainly limited to global mean or hemispheric land/ocean temperatures) and by their ability to extrapolate skilfully outside the calibrated range.

The terms ‘emulator’ and ‘simple climate model’ (SCM) are different, although they are sometimes used interchangeably. SCM refers to a broad class of lower-dimensional models of the energy balance, radiative transfer, carbon cycle, or a combination of such physical components. SCMs can also be tuned to reproduce the calculations of climate-mean variables of a given ESM, assuming that their structural flexibility can capture both the parametric and structural uncertainties across process-oriented ESM responses. When run in this setup, they are termed emulators. Simple climate models do not have to be run in ‘emulation’ mode, though, as they can also be used to test consistency across multiple lines of evidence with regard to ranges in ECS, TCR, TCRE and carbon cycle feedbacks (Chapters 5 and 7). Physical emulation can also be performed with very simple parameterizations (‘one-or-few-line climate models’), statistical methods like neural networks, genetic algorithms, or other artificial intelligence approaches, where the emulator behaviour is explicitly tuned to reproduce the response of a given ESM or model ensemble (Chapters 4, 5 and 7).

Current emulators and SCMs include the generic impulse response model outlined in [[IPCC:Wg1:Chapter:Chapter-8|Chapter 8]] of AR5 (AR5-IR; Supplementary Material 8.SM.11 of [[#Myhre--2013|Myhre et al., 2013]] ), two-layer models ( [[#Held--2010|Held et al., 2010]] ; [[#Rohrschneider--2019|Rohrschneider et al., 2019]] ; [[#Nicholls--2020|Nicholls et al., 2020]] ), and higher-complexity approaches that include upwelling, diffusion and entrainment in the ocean component (e.g., MAGICC Version 5.3 ( [[#Raper--2001|Raper et al., 2001]] ; [[#Wigley--2009|Wigley et al., 2009]] ); Version 6/7 ( [[#Meinshausen--2011a|Meinshausen et al., 2011a]] ); OSCAR ( [[#Gasser--2017|Gasser et al., 2017]] ); CICERO SCM ( [[#Skeie--2017|Skeie et al., 2017]] ); FaIR ( [[#Millar--2017a|Millar et al., 2017a]] ; [[#Smith--2018|Smith et al., 2018]] ); and a range of statistical approaches ( [[#Schwarber--2019|Schwarber et al., 2019]] ; [[#Beusch--2020b|Beusch et al., 2020b]] ). An example of recent use of an emulator approach is an early estimate of the climate implications of the COVID-19 lockdowns (Cross-Chapter Box 6.1; [[#Forster--2020|Forster et al., 2020]] ).

Since AR5, simplified climate models have been developed further, and their use is increasing. Different purposes motivating development include: being as simple as possible for teaching purposes (e.g., a two-layer energy balance model); being as comprehensive as possible to allow for propagation of uncertainties across multiple Earth system domains (MAGICC and others); or focusing on higher-complexity representation of specific domains (e.g., OSCAR). The common theme motivating many models is to improve parameterizations that reflect the latest findings in complex ESM interactions – such as the nitrogen cycle addition to the carbon cycle, or tropospheric and stratospheric ozone exchange – with the aim of emulating their global mean temperature response. Also, within the simple models that have a rudimentary representation of spatial heterogeneity (e.g., four-box simple climate models), the ambition is to represent heterogeneous forcers such as black carbon more adequately ( [[#Stjern--2017|Stjern et al., 2017]] ), provide an appropriate representation of the forcing–feedback framework (e.g., [[#Sherwood--2015|Sherwood et al., 2015]] ), investigate new parameterizations of ocean heat uptake, and implement better representations of volcanic aerosol-induced cooling ( [[#Gregory--2016a|Gregory et al., 2016a]] ).

MAGICC ( [[#Wigley--2009|Wigley et al., 2009]] ; [[#Meinshausen--2011a|Meinshausen et al., 2011a]] ) and FaIR ( [[#Smith--2018|Smith et al., 2018]] ) were used in IPCC SR1.5 ( [[#IPCC--2018|IPCC, 2018]] ) to categorize mitigation pathways into classes of scenarios that peak near 1.5°C, overshoot 1.5°C, or stay below 2°C. The SR1.5 ( [[#Rogelj--2018b|Rogelj et al., 2018b]] ) concluded that there was ''high agreement'' on the relative temperature response of pathways, but ''medium agreement'' on the precise absolute magnitude of warming, introducing a level of imprecision in the attribution of a single pathway to a given category.

In this Report, there are two notable uses of simple climate models. One is the connection between the assessed range of ECS in Chapter 7, and the projections of future global surface air temperature (GSAT) change in Chapter 4, which is done via a two-layer model based on [[#Held--2010|Held et al. (2010)]] . It is also used as input to sea level projections in Chapter 9. The other usage is the transfer of Earth system assessment knowledge to WGIII, via a set of models (MAGICC, FaIR, CICERO-SCM) specifically tuned to represent the WGI assessment. For an overview of the uses, and an assessment of the related Reduced Complexity Model Intercomparison Project (RCMIP), see [[#Nicholls--2020|Nicholls et al. (2020)]] and Cross-Chapter Box 7.1.

<div id="box-1.3" class="h2-container box-container"></div>

'''Box 1.3 | Emissions Met''' '''rics in AR6 WGI'''

<div id="h2-30-siblings" class="h2-siblings"></div>

Emissions metrics compare the radiative forcing, temperature change, or other climate effects arising from emissions of CO <sub>2</sub> against those from emissions of non-CO <sub>2</sub> radiative forcing agents (such as CH <sub>4</sub> or N <sub>2</sub> O). They have been discussed in the IPCC since the First Assessment Report and are used as a means of aggregating emissions and removals of different gases and placing them on a common (‘CO <sub>2</sub> equivalent’, or ‘CO <sub>2</sub> -eq’) scale.

AR5 included a thorough assessment of common pulse emissions metrics, and how these address various indicators of future climate change ( [[#Myhre--2013|Myhre et al., 2013]] ). Most prominently used are the global warming potentials (GWPs), which integrate the calculated radiative forcing contribution following an idealized pulse (or one-time) emission, over a chosen time horizon ( [[#IPCC--1990a|IPCC, 1990a]] ), or the global temperature change potential (GTP), which considers the contribution of emissions to the global-mean temperature at a specific time after emission. Yet another metric is the global precipitation change potential (GPP), used to quantify the precipitation change per unit mass of emission of a given forcing agent ( [[#Shine--2015|Shine et al., 2015]] ).

As an example of usage, the Paris Rulebook [Decision 18/CMA.1, annex, paragraph 37] states that

Each Party shall use the 100-year time-horizon global warming potential (GWP) values from the IPCC Fifth Assessment Report, or 100-year time-horizon GWP values from a subsequent IPCC assessment report as agreed upon by the ‘Conference of the Parties serving as the meeting of the Parties to the Paris Agreement’ (CMA), to report aggregate emissions and removals of GHGs, expressed in CO <sub>2</sub> -eq. Each Party may in addition also use other metrics (e.g., global temperature potential) to report supplemental information on aggregate emissions and removals of GHGs, expressed in CO <sub>2</sub> -eq.

Since AR5, improved knowledge of the radiative properties, lifetimes and other characteristics of emitted species, and the response of the climate system, have led to updates to the numerical values of a range of metrics (Table 7.15). Another key development is a set of metrics that compare a pulse emission of CO <sub>2</sub> (as considered by GWP and GTP) to step-changes of emission rates for short-lived components (i.e., also considering emissions trends). Termed GWP* (which also includes a pulse component) and combined global temperature change potential (CGTP), these metrics allow the construction of a near-linear relationship between global surface temperature change and cumulative CO <sub>2</sub> and CO <sub>2</sub> -eq emissions of both short- and long-lived forcing agents ( [[#Allen--2016|Allen et al., 2016]] ; [[#Cain--2019|Cain et al., 2019]] ; [[#Collins--2020|Collins et al., 2020]] ). For example, the temperature response to a sustained methane reduction has a similar behaviour to the temperature response to a pulse CO <sub>2</sub> removal (or avoided emission).

In this Report, recent scientific developments underlying emissions metrics, as relevant for WGI, are assessed in full in Section 7.6. In particular, see Box 7.3, which discusses the choice of metric for different usages, and Section 7.6.1, which treats the challenge of comparing the climate implication of emissions of short-lived and long-lived compounds. Also, the choice of metric is of key importance when defining and quantifying net zero GHG emissions (Box 1.4 and Section 7.6.2). [[IPCC:Wg1:Chapter:Chapter-6|Chapter 6]] applies metrics to attribute GSAT change to short-lived climate forcer (SLCF) and long-lived GHG emissions from different sectors and regions (Section 6.6.2).

The metrics assessed in this Report are also used, and separately assessed, by WGIII. See Cross-Chapter Box 2 and Annex B in [[IPCC:Wg1:Chapter:Chapter-2|Chapter 2]] of the WGIII contribution to AR6.

<div id="1.5.4" class="h2-container"></div>

<span id="modelling-techniques-comparisons-and-performance-assessments"></span>