Based on a physical climate “core”, Earth system models (ESMs) simulate numerous complex relationships and feedbacks among the atmosphere, biosphere and cryosphere to model energy and mass transfer across domains (Séférian et al., 2019; Jones, 2020). ESMs are critical tools for studying the role of anthropogenic forcing on interacting biogeochemical systems, and for predicting cascades of physical and ecological responses to global warming (Steffen et al., 2018; Gillett et al., 2016; Zhang et al., 2024). Developed based upon a broad consultation process, the Coupled Model Intercomparison Phase 7 Earth System (CMIP7-ES) data request contains the requirements needed for the analysis of carbon, nitrogen, water and other cycles and their interactions with the physical climate, the biosphere and reservoirs (Juckes et al., 2025; Friedlingstein et al., 2025; Dunne et al., 2025; Jones et al., 2024). The data request is organized into scientific `opportunities' representing specific research foci that each contain a variable group representing the data needed to fulfill the scientific objectives of each opportunity (Mackallah et al., 2026; Data Request Task Team, 2025). While many of these variables were produced as part of CMIP6, the new variables will enable the evaluation of climate sensitivity to forcing, detection and attribution of climate change, and prediction of the likelihood of major tipping points under different emissions scenarios (Wunderling et al., 2023; Steffen et al., 2018). Moreover, while the CMIP6 data request included a large number of variables, most models did not output most variables meaning that many studies analysing multiple variables were often restricted to small subsets of models. By organising the CMIP7 data request around scientific opportunities, and omitting variables that were little used in CMIP6, the intention is that studies addressing these key opportunities will be able to draw on a much more complete set of CMIP7 data. Fulfillment of the CMIP7-ES data request will enable the study of a large number of terrestrial and marine ecological dynamics and their implications in the carbon cycle and climate system (Sanderson et al., 2024). This will allow for ESMs to be benchmarked against observations, serving both as validation of the processes represented by the models and their skill at predicting trajectories of global change (Fu et al., 2022; Collier et al., 2018).

Projecting Earth systems' responses to further forcing from anthropogenic emissions is at the centre of the CMIP7-ES data request. Emissions-driven, as opposed to concentration-driven, simulations that incorporate processes for additional release and removal of atmospheric carbon and reflect an advanced understanding of natural source-sink dynamics will explore the response of the climate system to human activities (Arora et al., 2020; Sanderson et al., 2024). Whereas in CMIP6, almost all scenario simulations were run with a single set of projected atmospheric CO₂ concentrations derived from an emulator (Meinshausen et al., 2011a, b; O'Neill et al., 2016), meaning that carbon cycle uncertainty was typically neglected when discussing uncertainties of the climate response to emissions. The new emissions-driven interactive CO₂ simulations proposed as part of the Scenario Model Intercomparison Project for CMIP7 will allow the effects of carbon cycle uncertainty on future projections to be fully characterised (Van Vuuren et al., 2026). As carbon dioxide removal (CDR) technologies are becoming more feasible and necessary to avoid the worst climate impacts, modelling a range of Shared Socioeconomic Pathways (SSPs) that reflect up-to-date industrial and agricultural emissions and include mitigation to net-zero and net-negative emissions is a priority (Van Vuuren et al., 2026; Riahi et al., 2017). For example, these simulations will include explicit representation of the carbon cycle effects of afforestation and reforestation in ESMs (Van Vuuren et al., 2026). An improved representation of the flux of CO₂, in particular into the oceans, will account for missing fluxes and reduce existing mass imbalances in models (Henson et al., 2022; Jones et al., 2016; Liddicoat et al., 2021; Planchat et al., 2023; Tang et al., 2025). Adding nitrogen and phosphorous cycling to terrestrial ESMs will improve estimates of photosynthesis (Gier et al., 2024), and adding methane emissions will ensure that major contributors to the greenhouse effect are represented (Davies-Bernard et al., 2020; Lovato et al., 2022; Sanderson et al., 2024). Progress on these aspects of ESMs will help create wholistic representations of the climate-carbon system and improve predictions of near- and long-term climate impacts in response to future emissions.

Reducing uncertainty around marine and terrestrial carbon fluxes is critical to emissions-driven model deployment. The oceanic “pump” of carbon export represents a crucial sink for atmospheric carbon, removing ∼25 % of the carbon added to the atmosphere since the start of the industrial revolution (DeVries, 2022; Friedlingstein et al., 2025). The response of the marine carbon flux to additional warming is uncertain reflecting differences in the simulated dynamics and parameterizations in ocean biogeochemical modules (Rohr et al., 2023; Wilson et al., 2022). The “solubility pump” of CO₂ into dissolved inorganic carbon (DIC) accounts for the majority of carbon present in seawater, functioning as a result of the disequilibrium concentration of carbon in the oceans and atmosphere (Egleston et al., 2010; DeVries, 2022). The rate at which the ocean and atmosphere equilibrate and remove anthropogenic carbon depends on water pH levels and sea surface temperatures (Weiss, 1974). It is estimated that the oceans' buffering capacity has decreased substantially, but a large source of uncertainty remains in how seawater circulation will be affected by a warmer atmosphere, thereby influencing the ocean-atmosphere CO₂ flux (Archer, 2005; Liao et al., 2021). The “biological pump” of carbon export via marine organisms, both as DIC from dissolved carbonate skeletons and as larger particles that precipitate into sediments, sequesters anthropogenic carbon at a rate of approximately 10 gt (gigatons) of carbon per year (DeVries, 2022). Biological activity in the upper ocean is contingent on water chemistry (i.e. pH and nutrient availability), regional upwelling, and hemispheric-scale circulation patterns (Planchat et al., 2023). Improved modelling of trophic interactions in the upper ocean from primary producers to higher trophic levels, responses to changing water chemistry, and factors affecting carbon export to deeper ocean reservoirs are critical for estimating the magnitude of ocean carbon storage in the future (Wilson et al., 2022). Deliberately altering ocean water chemistry and circulation to enhance rates of carbon dissolution in seawater have been proposed as a method of reducing the greenhouse effect. More studies are needed to assess the feasibility, and estimate the impacts of ocean-based CDR on ecosystems, fisheries, and climate (Doney et al., 2025).

Over land, Jones et al. (2023) have shown that CMIP6 ESMs generally agree well with observations of regional mean fluxes and carbon stocks, although large spread across models is found. While improvements have been made in simulating terrestrial gross primary productivity (GPP), especially when including the nitrogen cycle, the net land-atmosphere carbon flux remains an uncertain component of ESMs (Gier et al., 2024). ESMs without an interactive nitrogen cycle tend to overestimate the effect of elevated CO₂ on the land carbon sink (Kou-Giesbrecht and Arora, 2023). A broader representation of the interactive nitrogen cycle and inclusion of other limiting factors such as phosphorus (Fleischer et al., 2019) are needed to better constrain the magnitude of the future land carbon sink (Kou-Giesbrecht and Arora, 2023; Gier et al., 2024). The spatial distribution of the land carbon sink is still poorly represented with an underestimation of the sink in the Northern Hemisphere, and poor agreement between ESMs and observations of carbon fluxes and stocks over permafrost-covered regions (Jones et al., 2023; Qiu et al., 2023). Representing permafrost biogeochemistry is crucial to better represent future carbon fluxes, including methane, and feedbacks between fluxes and climate (Kleinen et al., 2021; Schuur et al., 2022). Improvements in net land-atmosphere carbon fluxes are mainly attributed to improvements in the representation of GPP (Gier et al., 2024), while carbon turnover (including processes such as respiration, mortality and soil carbon decomposition) remains a large source of uncertainty (Koven et al., 2017; Canadell et al., 2021; Spafford and MacDougal, 2021; Pugh et al., 2020). With the exception of fire, tree mortality and demographic changes associated with climate-driven disturbances are poorly or not represented in ESMs resulting in uncertainties in the future response of the land carbon sink to climate change (Fisher and Koven, 2020; Pugh et al., 2020; Bonan et al., 2024). Although fire processes are represented in some ESMs, changes in fire regimes constitute a significant source of uncertainty in simulating the carbon cycle. CMIP6 models systematically underestimate the total burned area observed via satellites, instead tending to estimate an increase in burned area and fire emissions, contrary to observations (Zheng et al., 2021; Li et al., 2024). Running ESM simulations assuming different levels of fire prevalence and comparing the output to satellite records and charcoal reconstructions will help to constrain its role in the carbon cycle and improve predictions for future fire prevalence worldwide (Rabin et al., 2017; Li et al., 2024). Soil respiration represents another source of uncertainty in terrestrial ESMs, reflecting a need for better representations of below-ground processes in CMIP7 (Ito et al., 2020; Varney et al., 2022). Closing nitrogen and phosphorous cycles, modelling carbon fluxes from soil respiration and vegetation uptake, ecological responses to fire and drought, and the interactions between the land and ocean components of the biogeochemical cycles will all help to improve the representation of net primary production and its feedbacks to climate (Boysen et al., 2021; Song et al., 2021; Qiu et al., 2023). Finally, reducing uncertainties on land-use and land-cover changes (LULCC) and corresponding emissions will be crucial for emissions-driven runs. For example, Egerer et al. (2025) found large spread in afforested and reforested area in CMIP6 models forced by the same underlying LULCC scenarios. Differences in the consideration of land-use transitions (gross vs. net) and of management processes can result in large spread in LULCC fluxes (Arneth et al., 2017; Hartung et al., 2021). Models should clearly report the underlying assumptions for estimating LULCC transitions and fluxes.

Beyond carbon cycle characterizations the CMIP7-ES data request will provide new insights into how changes in Earth's energy balance affect climate and biogeochemical cycling. In particular, it will enable study of how aerosol transport, deposition, and reactions with other heat-trapping gases within the atmosphere affect climate, the carbon cycle and ecosystems. How aerosol particulates and trace gases, stemming both from industrial processes and volcanism, modify global temperatures have been major source of uncertainty in transient simulations (Fyfe et al., 2021; Hansen et al., 2023; Clyne et al., 2021). Improvements in the representation of aerosol forcing over multiple phases of CMIP had a large impact on temperature trends derived from models. Exclusion of forcing from volcanoes in early phases of CMIP5 resulted in simulations of global temperature that were significantly greater than observed trends (Domingues et al., 2008; Schmidt et al., 2014). CMIP6 experiments brought models into better agreement with respect to natural (i.e. volcanic) forcing, and indicated that anthropogenic aerosol emissions likely switched from driving a cooling trend over the 20th century to driving a warming trend in the 21st century, due to improvements in air quality (Quaas et al., 2022; Fiedler et al., 2023; Bellouin et al., 2020; Zanchettin et al., 2022; Bauer et al., 2022). Although globally the cooling effect of aerosols has decreased (Bauer et al., 2022), the spatial distribution of these changes is determined by regional emissions and meteorological dynamics (Williams et al., 2022; Stier et al., 2024). Downscaled and high-resolution models are needed to tie improvements in air quality with local and regional temperature trends and weather events (Roberts et al., 2025). Although aerosol forcing as a physical process falls primarily within the Atmosphere theme (Dingley et al., 2025), there are also direct and indirect interactions between atmospheres, oceans, and land. As examples, how aerosols nucleate moisture in the atmosphere, thus affecting regional hydroclimate, or how transport of dust particles fertilizes remote ecosystems are potential research topics that could be addressed by the data request (Samset et al., 2024; Iles et al., 2024; Bellouin et al., 2020; Richardson et al., 2016; Persad, 2023).

Next-generation ESMs will deepen our understanding of how anthropogenic and natural forcing drive climate variability, affect ocean-atmosphere circulation, and alter the risk of extreme events. It remains poorly-understood how changes in Earth's radiative balance translate into internal climate variability and affect climate on local to regional spatial scales (Boer et al., 2016; Jain et al., 2023). The inaccurate representation of patterns of internal variability on timescales ranging from days to decades hinders adaptation efforts when the full range of values for critical climate variables such as daily temperature and precipitation are not well-constrained (Degroot et al., 2021; Laepple et al., 2023). Understanding the relationship between variability and forcing is a critical component of climate change detection and attribution (D&A), which is needed to diagnose how anthropogenic emissions affect dynamical systems, such as the jet stream, Atlantic meridional overturning circulation (AMOC), and other ocean-atmosphere circulation patterns (Gillett et al., 2016, 2025). The Detection and Attribution Model Intercomparison Project (DAMIP v2.0) will include simulations with only land use change prescribed and simulations with only aerosol changes prescribed, in which atmospheric CO₂ is interactively modulated to allow for the effects of biogeochemical feedbacks on the responses to individual forcings to be analysed (Gillett et al., 2025). Longer simulations will be used by the Paleoclimate Model Intercomparison Project (PMIP) for constraining natural and forced variability on decadal to millennial timescales (Kageyama et al., 2018). The data request will also address the research needs of geoengineering experiments such as the Geoengineering Model Intercomparison Project (GEOMIP), enabling important research into the potential effects of direct intervention into radiative forcing (i.e. solar radiation management) and carbon dioxide removal on physical climate, ecosystems, and society (Visioni et al., 2023).

Scenarios with higher warming levels raise the likelihood of triggering tipping points within the Earth system (Schleussner et al., 2024; Ritchie et al., 2021; Armstrong McKay et al., 2022). Tipping elements of concern include the aridification of the Amazon basin, ratcheted loss of mass of the Greenland and West Antarctic ice sheets leading to rapid sea level rise, and a slowed AMOC due to the weakened oceanic thermal gradients (Wunderling et al., 2024). If any of these thresholds are reached within the next century it would result in widespread social and economic damage (Dietz et al., 2021). ESMs are the best source of information that we have about what tipping points might be reached under different warming scenarios. Building an understanding of the risk of tipping points, and of what cascades of impacts may result from them is crucial for building climate adaptation policies that accounts for uncertain but high-risk outcomes.

The substantial knowledge on Earth system dynamics is built on the incremental progress made in previous CMIP iterations, leading to the current potential to predict natural systems evolution and its connections with human activities and elucidate manifold scientific dimensions. Answering the following scientific questions are considered to be a high priority for CMIP7-ES. These questions flow from a series of opportunities (described in Sect. 4) proposed by members of the ES author group with community consultation, and are summarized in Fig. 1.

Cycles: how do the global carbon and other biogeochemical cycles respond to and feedback into changes in radiative forcing, and how does carbon cycle uncertainty contribute to uncertainty in projected warming? Which are the biogeochemical compounds that are still lacking or under-represented in exchanges and flows across ESMs realms?

The CMIP7-ES theme deals primarily with Earth system cycles and interactions across domains, not with the physical climate itself (i.e., atmospheric dynamics and general circulation), which falls under the Atmosphere & Ocean and Sea Ice themes (Li et al., 2025b; Dingley et al., 2025). Although related (e.g., marine ecosystems and fisheries), opportunities that deal with the social impacts of climate change were determined best suited for the Impacts and Adaptation theme (Ruane et al., 2025). Overlap exists with other themes in the area of ecological change and its associated impacts on biodiversity and ecosystem function. Any opportunity dealing chiefly with a single domain, i.e., cryosphere, land, or atmosphere, falls within other thematic areas as the ES theme emphasizes the transfer of energy and mass across domains.

The audience for the data request includes modelling centres with the expertise and capacity to generate ESM simulations, as well as a larger community of scientists and stakeholders who use model data for a variety of applications. Opportunities under the CMIP7-ES umbrella flow from the CMIP6 community of endorsed MIPs and research consortia, which represent both modelling centers and independent research activities (Eyring et al., 2016). These include the Coupled Climate-Carbon Cycle Model Intercomparison Project (C4MIP), Fire Model Intercomparison Experiment (FireMIP), Fisheries and Marine Ecosystem Model Intercomparison Project (FishMIP), Paleoclimate Model Intercomparison Project (PMIP), Geoengineering Model Intercomparison Project (GEOMIP), Tipping Point Model Intercomparison Project (TIPMIP), Detection and Attribution Model Intercomparison Project (DAMIP), Decadal Climate Prediction Project (DCPP), (ScenarioMIP) and others. The opportunities described below, with their attendant requests for data from the modelling centres, have been contributed by the members of these experiments. Fulfilment of the data requirements described below is critical for the advancement of research activities within these longstanding research groups.

CMIP7-ES data will also be useful to the community of climate model data users. These include researchers involved in data-model comparison of observational, remotely sensed, and paleoclimate data. For data-model comparisons, CMIP data serves as an independent source of climate information, and findings from these activities cyclically drive model development by identifying areas of data-model disagreement. Non-academic audiences for ES model data include non-profit organizations, climate policy makers and private sector stakeholders, for example insurers involved in catastrophe risk modelling (Stoffel et al., 2024). CMIP will continue to be an integral part of designing climate adaptation polity on national and international levels (IPCC, 2021), and is increasingly being used by state and local governments to guide municipal planning for future climate impacts.

A defining trait of an Earth system model is the ability to consider interactions between the physical climate system and the global carbon cycle. This coupling between carbon and climate can be represented in two main ways. The first is by allowing the atmospheric CO₂ concentration to be prognostic in response to imposed emissions with land and ocean carbon reservoirs freely exchanging CO₂ with the atmosphere, in an “emissions-driven” configuration. The second is to specify atmospheric CO₂ concentrations, such that fossil fuel CO₂ emissions are inferred as what is needed to balance out the change in carbon masses in the total land, ocean, and atmosphere system (a “concentration-driven” configuration) (see Fig. 2 in Jones et al., 2016). In both emissions-driven and concentration-driven configurations, it is necessary to quantify changes to all of the carbon pools in the biosphere in order to ensure carbon mass conservation. The central goal of this opportunity is thus to track all carbon throughout the Earth system, ensure a closed carbon budget, and allow for the calculation of compatible fossil fuel CO₂ emissions in concentration-driven experiments.

The land and ocean variable groups for this opportunity are based on the variables defined following the CMIP6 C4MIP experiment described in Jones et al. (2016), their Fig. 5 (land carbon cycle) and Fig. 13 (ocean carbon cycle) (Table 2). On land, two new variables (Lmon.cGeologicStorage and Lmon.fHarvestToGeologicStorage), which were not present in the CMIP6 variable request, will allow tracking of carbon under intentional CDR such as bioenergy with carbon capture and storage (BECCS). In CMIP6, CDR fluxes were specified as forcings rather than simulated endogenously, and this is anticipated to be the case for most CDR methods in CMIP7 Fast Track experiments as well, with the exception of reforestation and afforestation fluxes (Van Vuuren et al., 2026). Moreover, some modelling centres are experimenting with CDR representation for other approaches (e.g., Sanderson et al., 2024), and these new variables will allow the reporting of these fluxes for models and scenarios that treat them prognostically.

In addition to the land and ocean carbon cycle variables, atmospheric mole fractions of CO₂, as well as total fluxes of CO₂ between the atmosphere and the land and ocean, are needed to close the global carbon budget, particularly for emissions-driven ESM scenarios.

An emergent linear and path-independent relationship between global warming and cumulative CO₂ emissions (Allen et al., 2009; Matthews et al., 2009; Zickfeld et al., 2009) underlies much of global climate policy, including the concept of a remained carbon budget for climate stabilization (IPCC, 2021). In concentration-driven ESMs, fossil fuel CO₂ emissions must be inferred as the difference in the total stock of carbon in the combined land, atmosphere, and ocean systems over time. Thus, one key goal of tracking the carbon cycles through each of these systems is to infer the implied emissions for a given CO₂ concentration forcing pathway, and to diagnose relationships such as the Transient Climate Response to cumulative CO₂ emissions (TCRE, Arora et al., 2020), which is the slope of the relationship between warming and cumulative emissions. Under emissions-driven ESM simulations, CO₂ emissions are specified rather than diagnosed, and it is necessary to have similar information about the response of the carbon cycle to the emissions. Such information will allow us to characterise the contribution of carbon cycle uncertainty to uncertainty in projections of future climate change under particular emissions scenarios. Moreover, this information could help us understand the processes contributing to, and help to narrow such an uncertainty using emergent constraints, based for example on changes in carbon pools over the historical period.

Earth system models simulate a broad suite of carbon cycle feedbacks in response to changing climate and atmospheric CO₂ emissions, which are central to the projections of Earth system change into the future. Across several generations of coupled carbon-climate models, the carbon cycle feedbacks act as a large source of uncertainty in these projections (Friedlingstein et al., 2006; Arora et al., 2013, 2020). Reducing the uncertainty in these feedbacks is thus a key goal in narrowing projections of climate change in response to future CO₂ emissions, especially in the context of evolving ESMs (e.g. inclusion of N and CH₄ cycles) and more systematic use of emission-driven experiments. One approach to reducing this uncertainty is systematic benchmarking of land and ocean models against a broad range of historical observations at site to global scales, so that model fidelity can be assessed and tracked over time (Collier et al., 2018; Fu et al., 2022; Gier et al., 2024) (Table 3). In particular, tier1 variable groups include metrics from diverse ES realms that would enable for a direct or as close as possible comparison with available observations.

Fire is the primary terrestrial ecosystem disturbance globally and a critical Earth system process (Bowman et al., 2009; Li and Lawrence, 2017; Li et al., 2024, 2025a). This opportunity enhances our understanding of past, present, and future fire changes and the role of fire in the Earth system. The proposed variables are divided into two categories: fire variables and fire driver/impact variables. The fire variables (Table 4), (i.e., burned area fraction and fire carbon emissions) are of the highest priority and essential for understanding fire behavior in ESMs. They serve two main aims: (a) analysing historical fire patterns and projecting future fire regimes under varying climate and socio-economic scenarios to inform long-term environmental planning and policy making; (b) benchmarking and evaluating fire simulations in coupled Earth system models, leading to improvements in future modelling systems to support more precise climate predictions and a deeper understanding of Earth system complexities. The fire driver and impact variables are currently used in Earth system modelling and strongly overlap with baseline variables. These include variables related to carbon, nitrogen, water, and energy cycles; vegetation distribution and structure; climate indicators (e.g., temperature, precipitation, wind speed, permafrost extent, active layer thickness, sea ice and snow coverage, sea surface temperature); and atmospheric circulation, composition, and chemistry. These variables will be analysed to: (a) assess the accuracy of models in capturing the relationship between fire and climate, ecosystems, and environmental factors; (b) understand the drivers of fire regime changes, the impacts of fire, and cross-sphere feedbacks between fire and various Earth system components. In addition, daily maximum temperature, precipitation, wind speed, and minimum relative humidity are required to calculate the Canadian Fire Weather Index (Quilcaille et al., 2023). FWI is a method to represent the impact of weather and is related to fire's drivers. These variables are available in biodiv_land_daily, CFMIP-daily, and AgModelExpandedDaily.

This opportunity is composed of a baseline set of variables that have already been widely used in CMIP6 and exist in most ESMs, along with a number of selected variable groups whose inclusion in the Data Request (Table 5) (see details in Appendix A) will extend the scientific purpose toward relevant ecosystem processes and downstream applications. The new variables requested will provide additional constraints on the ocean carbon pump both historically and under climate scenario projections.

The consistent simulation of the energy and water cycles by numerical models of the Earth system is fundamental as their flows across atmosphere, land, cryosphere and oceans tightly interact to shape the climate and its future changes (Trenberth, 2014). It is well established that global precipitation and evaporation changes are controlled by Earth's energy balance, while water vapour is a relevant gaseous absorber in the atmosphere that in turn plays a primary role in the global radiative budget (Allan et al., 2020). The very large latent energy fluxes also tie together Earth's energy and water cycles, meaning that they should be considered in concert. The growing volume of information provided by satellite Earth observation systems will increase our capability to understand and better constrain the uncertainties related to water and energy budgets in modelled historical changes (Stephens et al., 2023).

These two interconnected opportunities primarily rely on the production of the baseline_climate_variables group, which contains the main variables central to the implementation of the energy budget framework and the water cycle analyses as described in the Sixth IPCC Assessment Report (Forster et al., 2021; Douville et al., 2021) (Table 6). Similarly, the core variable groups requested by other ES opportunities (Sect. 4.1 and 4.2) could be further exploited to closely investigate the land surface heat and energy budgets. Complementary variable groups, namely seaice_budget_energy_monthly and seaice_budget_freshwater_monthly, address the need to specifically describe the flows of energy and water in sea ice to reach a better closure of the global budgets, while the int_ocean_budgets variables set was designed to refine the computation of oceanic budgets in the light of recent model advancements and to improve the comparison with observations on non-hydrostatic pressure levels.

The CMIP Rapid Evaluation Framework (REF) was created to evaluate and benchmark the CMIP7 Assessment Fast Track (CMIP7 AFT) simulations as soon as they are uploaded to the ESGF with metrics and diagnostics that are available through different open-source evaluation and benchmarking tools (Hoffman et al., 2025) (Table 7). This opportunity contains the set of variables that are needed for the planned diagnostics and metrics for the REF (CMIP Model Benchmarking Task Team, 2024). The suggested metrics/diagnostics for the REF to be available for all CMIP7 AFT experiments will allow basic evaluations. The exact selection of variables was also made consistent with the model evaluation diagnostics in Chapter 3 of the latest IPCC report (Eyring et al., 2021). Due to the fixed timeline for the CMIP7 AFT simulations there is only a short time period for the technical implementation of the REF, and therefore the available metrics and diagnostics in this first version of the REF will be limited to a temporal resolution of monthly mean data and about five metrics/diagnostics per realm based on a community selection. The realms were chosen specifically to be consistent with the realms used for the data request.

This opportunity has been submitted by the Tipping Point Modelling Intercomparison Project (TIPMIP) (Jones et al., 2025). As part of the CMIP7-ES theme, this opportunity comprises two variable groups which are also relevant to the Atmosphere, Land & Land Ice, Ocean & Sea Ice themes (Table 8).

The concept of “tipping point” or “critical transition” appears in several fields of science: in ecology, for example, it appears in the form of mass extinction and rapid desertification (Scheffer et al., 2001; Arumugam et al., 2024); human physiology offers examples such as seizures and the abrupt increase in inflammatory response (Scheff et al., 2013); examples from the geosciences include potential events such as the dieback in the Amazon rainforest or the decline and collapse of polar ice sheets (Lenton et al., 2008). These seemingly disparate events share phenomenological attributes, which are understood to be defining characteristics of a tipping point or critical transition (Scheffer et al., 2009; Kuehn, 2011). A tipping point is reached when there is an abrupt qualitative change in the system's dynamics which is rapid compared to the system's normal evolution, and beyond this threshold the system enters a new dynamical state which is qualitatively different from the previous state. More recent studies have supported the existence of tipping points for certain Earth system components, example in Lenton et al. (2008) such as polar ice sheets (Bradley and Hewitt, 2024; Petrini et al., 2025), and the AMOC (van Westen et al., 2024; Ditlevsen and Ditlevsen, 2023). Wunderling et al. (2024) have also highlighted the possibility of interactions between these components of the Earth system, whereby one component crossing its tipping threshold could destabilize other components, triggering a so-called “tipping cascade”.

Given the fact that Earth system components are complex non-linear systems in their own right, that they are coupled to one another, and interact across many different spatio-temporal scales, giving a precise characterization of what critical transitions (or tipping events) could occur, or when they could occur, is at present very difficult (Ben-Yami et al., 2024). Yet the fact that such transitions might occur cannot be excluded, as several components of the Earth system may be susceptible to reaching a critical threshold beyond which amplifying feedbacks could result in abrupt and/or irreversible changes (Armstrong McKay et al., 2022). This could have far-reaching impacts on the global climate, ecosystems, and humankind. Recent assessments have highlighted the increasing risk for potential tipping events, in particular beyond 1.5 °C global warming, and also stressed the large uncertainties involved in any projection regarding the future occurrence of such events (IPCC AR6, Eyring et al., 2021). Addressing these uncertainties will necessitate a systematic effort to evaluate our understanding of Earth system dynamics and their evolution under sustained exogenous forcing, so that we may be better able to quantify the likelihood of tipping events and therefore the risks and impacts associated to them. This will be crucial for developing effective strategies to mitigate and adapt to the impacts of global environmental change.

The variables included in this opportunity will serve two major purposes: (1) the analysis of ESMs with respect to tipping points, and (2) serving as forcing input for uncoupled/offline component models (e.g. standalone ice sheet models) in follow-up tailored tipping experiments. There is a strong overlap between the variables in the opportunity and existing baseline_climate_variables, minimising additional computational costs while providing the base for a cross-domain analysis of tipping points.

There is empirical evidence from CMIP6 simulations of the existence of critical thresholds in ESMs, for example in rapid cooling events in the North Atlantic subpolar gyre (Swingedouw et al., 2021), decay and shutdown of the Atlantic meridional overturning circulation (Drijfhout et al., 2025), and dieback events in the Amazon rainforest (Parry et al., 2022), as well as more general abrupt shift events which could correspond to various tipping points (Terpstra et al., 2025). These phenomena were also reported in CMIP5 models by Drijfhout et al. (2025). Significant uncertainty remains around the mechanisms behind these regime shift events, as well as the critical thresholds that would trigger them once crossed.

This opportunity will allow the evaluation of key large-scale tipping elements such as ice-sheet collapse, permafrost carbon release, tropical forest dieback and shutdown of the AMOC, as well as possible feedbacks associated with each individual tipping element. Outputs will be used to evaluate the uncertainties associated with identifying the existence of tipping points in the biogeophysical Earth system; the critical thresholds and warming levels that may induce tipping; as well as the interactions and feedbacks between (possible) tipping elements. Outputs may be then used by impact models and other end-user groups to evaluate the downstream consequences of tipping points in the Earth system on human society.

As the magnitude of climate changes are strongly determined by cumulative emissions (Allen et al., 2009; Notz and SIMIP Community, 2020), differences between various emission scenarios will have relatively little impact in the near term. This makes the prediction of climate over the next few decades an initial value problem, and decadal predictions are created operationally (Smith et al., 2013). Such predictions are made not only for atmospheric variables, but across earth systems, including for ocean variables such as the AMOC (WMO, 2024), cryospheric variables such as sea ice concentration (WMO, 2024), and biogeochemical variables such as CO₂ uptake (Li et al., 2016; Gooya et al., 2024) (Table 9). Understanding the nature and limits of the predictability across the Earth system is key to delivering the maximum skill in these forecasts. The Decadal Climate Prediction Project (DCPP) defines experiments to allow the quality of climate prediction systems to be assessed through the use of hindcasts, as well as to assess as the inherent predictability of the Earth system (Boer et al., 2016). As part of the CMIP7 Fast Track, a prediction initialised in 2025 and comprising of 10 ensemble members is requested (dcppB-forecast-cmip6), but only from models who have also performed a hindcast.

This opportunity was submitted by DCPP, and incorporates two different variable groups that both span multiple themes. The essential variable group will allow analysis of key climate aspects of the decadal forecast, including modes of climate variability such as the Pacific Decadal Oscillation and AMOC that have substantial low-frequency components. The wider opportunity pushes beyond mean climates to assess the predictability of climate extremes.

Activities associated with CMIP7 will close critical gaps in our knowledge of the Earth system and its response to anthropogenic perturbation. This will set the stage for research in the event of significant mitigation efforts to slow the rate of warming. At present, there are still many sources of uncertainty in ESM simulations surrounding interactions among emissions, radiative forcing, and elemental cycling. In ocean models, there is uncertainty in how marine primary productivity drives to carbon storage, in particular surrounding the role of biogeochemical and ecological processes in deep ocean sediments (DeVries, 2022). Coastal processes, which have highly heterogeneous carbon flux dynamics, are not well-represented in global climate models (Laruelle et al., 2018; Dai et al., 2022). Increased spatial resolution and improvements in biogeochemical modules may reduce uncertainty surrounding the land-ocean continuum and clarify the unique features of the ocean carbon pump in coastal regions (Bourgeois et al., 2016; Friedlingstein et al., 2025). In terrestrial models, soil and plant respiration, their interactions with the nitrogen cycle and response to warming and rising CO₂ are sources of model disagreement (Jones et al., 2016). Together, differences in the parametrization of carbon-climate feedbacks over land account for an order of magnitude greater uncertainty than ocean carbon-climate feedbacks (Arora et al., 2020) representing uncertainty in land use and biophysical processes. The possibility of CDR raises questions regarding carbon cycle responses to rapid drawdown in atmospheric CO₂ concentrations under net-zero or net negative scenarios (Van Vuuren et al., 2026; Koven et al., 2022). Variables requested here will also support analysis of the carbon uptake effects of reforestation and afforestation, which will be simulated explicitly in ScenarioMIP simulations in CMIP7 (Van Vuuren et al., 2026). As efforts to better represent CDR processes continue to improve in ESMs (e.g., Sanderson et al., 2024), it will be critical to update data requests so that assessment of the realism of these processes, as well as consistency between ESMs and the integrated assessment models used to generate scenarios, can be assessed. This will require sustained efforts of the CDR community to identify key variables relevant to CDR processes alongside new model developments. How changes in radiative forcing, not only from CO₂but also associated with aerosols (e.g. dust, industrial pollutants and trace gases) are transported from point-source and distributed, and how this feeds back in to climate is another source of uncertainty in warming trajectories, especially at local to hemispheric spatial scales (Hansen et al., 2025; Zhao et al., 2022; Bauer et al., 2022). In the event of interference with solar forcing via solar radiation management having a strong understanding of how aerosols propagate through Earth systems will help anticipate potential effects. Regardless of future mitigation efforts, model downscaling, regional climate modelling, and detection and attribution will link anthropogenic forcing to impacts with increased specificity. Finally, tying ecosystem changes, severe weather events, and atmospheric dynamics to satellite observations will help track impacts in real time and serve as external validation of model performance.

Sustained efforts in these areas will address the needs of a diverse community of stakeholders. The scientific community has an interest in basic climate research, but CMIP7 should also support provision of ESM data to a variety of non-academic sectors (Lea et al., 2024). Climate policymakers have relied on scenario-based simulations to predict some probable range of outcomes that can be built into policy and planning for decades (IPCC, 2021; Durack et al., 2025a, b). While this has been true at the national and international levels, industry and local and regional governments represent new users of ESM output. For example, fire risk predictions serve the insurance industry by providing scientific basis for risk assessment models. In the context of marine ecosystems, the biological carbon pump of carbon sequestration in microorganisms is related to trophic dynamics at higher levels. Thus, modelling primary and secondary producers in the surface ocean supports fisheries management (Blanchard et al., 2024). Participants in international carbon markets, such as those mandated under the European Union Emissions Trading scheme could use CMIP7 variables that model rates of carbon cycling. Carbon sequestration data from soils, forests and oceans and coastlines (i.e. “blue carbon”) could be leveraged for use in these markets, although constraints remain regarding output resolution and flux uncertainty (Hilmi et al., 2021; Michaelowa et al., 2023). Improvements in model water budgets for fresh water are useful for water managers (Shao et al., 2023; Onyutha et al., 2021). Regional climate models and dynamical downscaling of coarser-resolution products will allow for CMIP7 data to be used on socially and politically-relevant spatial scales, for example cities, municipalities, and states.