Abstract Column‐integrated moist static energy (MSE) budgets underpin theories of tropical convection and circulation, yet in reanalyses and climate models the budget rarely closes; residuals routinely match the leading terms and mask physical insights. This study derives an MSE conservation law that is strictly consistent with GISS ModelE3 and elucidates why conventional diagnostics fail. Multiple intertwined factors—the breakdown of the product rule upon discretization, effects of mass‐filtering, mismatched flux and advective forms, numerical noise in diagnosed vertical velocity, asynchronous model output timing, and postprocessing including vertical interpolation and temporal averaging—leave significant residuals in both annual means and daily variability, even when raw 30‐min model output is used. Residuals are even larger over land and along coastlines. To tackle this obstacle, this study implements the “process increment method,” which accurately computes the column MSE flux divergence by calculating the change in column‐integrated internal energy, geopotential energy, and latent heats before and after applying the dynamics scheme. Furthermore, the calculated column flux divergence is decomposed into horizontal and vertical advective components. The most crucial finding is that vertical interpolation into pressure coordinates can introduce errors substantial enough to reverse the sign of vertical MSE advection in the warm‐pool regions. In ModelE3, native‐grid values show MSE import via vertical circulations, while values after interpolation into pressure coordinates indicate export. This discrepancy may prompt a reevaluation of vertical advection as an exporting mechanism and underscores the importance of precise MSE budget calculations.

Climate models are essential tools for understanding the Earth’s climate system, and undergo continual development to improve their skill across a range of metrics. Here we evaluate the new cloud representations and their radiative feedbacks in the NASA Goddard Institute for Space Studies (GISS)-E3 model relative to its predecessor GISS-E2.1. The new model now incorporates a considerable reworking of the cloud micro-and macrophysics, higher vertical resolution, and changes to moist turbulence. The results show substantial improvements in cloud climatology, primarily through a more realistic simulation of low-level clouds, especially stratocumulus, leading to improved shortwave and longwave cloud radiative effects. A machine learning (ML)-and observation-constrained tuning framework further improves tropical low-cloud regimes and Arctic cloud seasonality, making GISS-E3 competitive with CMIP6 models. However, persistent biases remain, including overly reflective tropical low clouds, excess high-cloud cover, and underestimated cloud cover over the Southern Ocean, highlighting limitations that parametric tuning alone cannot address. GISS-E3 exhibits a weakly positive net cloud feedback and an equilibrium climate sensitivity (ECS; 2.77◦C) comparable to GISS-E2.1. This behavior is linked to a weak non-low cloud feedback that lies near the lower bound of both the GISS-E3 calibrated-physics ensemble (CPE) and CMIP6 models, and outside observational constraints. In contrast, the CPE members that are more consistent with observations show higher ECS values (∼3.8◦C). These results highlight the importance of incorporating additional process-level observational constraints during both model development and tuning, particularly for high-and extratropical-cloud regimes, to better capture the mean state and cloud feedbacks.

A neural network (NN) surrogate of the NASA GISS ModelE atmosphere (version E3) is trained on a perturbed parameter ensemble (PPE) spanning 45 physics parameters and 36 outputs. The NN is leveraged in a Markov Chain Monte Carlo (MCMC) Bayesian parameter inference framework to generate a second posterior constrained ensemble coined a "calibrated physics ensemble," or CPE. The CPE members are characterized by diverse parameter combinations and are, by definition, close to top-of-atmosphere radiative balance, and must broadly agree with numerous hydrologic, energy cycle and radiative forcing metrics simultaneously. Global observations of numerous cloud, environment, and radiation properties (provided by global satellite products) are crucial for CPE generation. The inference framework explicitly accounts for discrepancies (or biases) in satellite products during CPE generation. We demonstrate that product discrepancies strongly impact calibration of important model parameter settings (e.g., convective plume entrainment rates; fall speed for cloud ice). Structural improvements new to E3 are retained across CPE members (e.g., stratocumulus simulation). Notably, the framework improved the simulation of shallow cumulus and Amazon rainfall while not degrading radiation fields, an upgrade that neither default parameters nor Latin Hypercube parameter searching achieved. Analyses of the initial PPE suggested several parameters were unimportant for output variation. However, many "unimportant" parameters were needed for CPE generation, a result that brings to the forefront how parameter importance should be determined in PPEs. From the CPE, two diverse 45-dimensional parameter configurations are retained to generate radiatively-balanced, auto-tuned atmospheres that were used in two E3 submissions to CMIP6.

We investigate the influence of hydrometeor radiative effects on the winter (DJF) geopotential height anomaly (ZA) over the subtropical and tropical Pacific Oceans in both the Coupled Model Intercomparison Project phase 5 (CMIP5) and phase 6 (CMIP6) models, utilizing satellite observations from GPS radio occultation (RO). This study evaluates average ZA biases in historical climate simulations, focusing on the influence of how models calculate the radiative properties of frozen hydrometeors (cloud ice and falling ice). CMIP6 models are categorized based on their treatments of these radiative properties: separately (SON2), combined (SON1) and cloud ice only (NOS). NOS models exhibit overestimation of absolute ZA biases in the upper troposphere. In contrast, SON2 models reduce these biases by 30-80 m, while SON1 models show no such improvement compared to NOS. In fact, the spatially averaged ZA biases in SON1 are comparable to or larger than those in NOS, suggesting that the combined radiative effects of cloud ice and falling ice hydrometeors do not have the same impact as the separate treatment in SON2. These biases in CMIP6 also align with sensitivity tests conducted using CESM1-CAM5 and CESM2-CAM6, where turning off radiative effects of falling ice results in significantly larger ZA biases compared to simulations with these effects included. Overall, the progress from CMIP5 to CMIP6 remains somewhat limited, with improvements seen only in SON2 models. These findings suggest that independently treating cloud ice and falling ice radiative properties may be crucial for reducing variability among CMIP6 models and improving model accuracy.

We present the second generation of ROCKE-3D (Resolving Orbital and Climate Keys of Earth and Extraterrestrial Environments with Dynamics), a generalized three-dimensional general circulation model (GCM) for use in Solar System and exoplanetary simulations of rocky planet climates. ROCKE-3D version 2.0 is a descendant of ModelE2.1, the flagship Earth system model of the NASA Goddard Institute for Space Studies (GISS) used in the most recent Intergovernmental Panel on Climate Change (IPCC) assessments. ROCKE-3D is a continuous effort to expand the capabilities of GISS ModelE to handle a broader range of planetary conditions, including different atmospheric planet sizes, gravities, pressures, and rotation rates; more diverse chemistry schemes and atmospheric compositions; diverse ocean and land distributions and topographies; and potential basic biosphere functions. In this release we present updated physics and many more supported configurations which can serve as starting points to simulate the atmospheres of rocky terrestrial planets of interest. Two different radiation schemes are supported, the GISS radiation, valid only for atmospheres similar to that of modern Earth, and SOCRATES, which is more generalized but more computationally expensive. While ROCKE-3D can simulate a very wide range of planetary and atmospheric configurations, we describe here a small subset of them, with the goal of demonstrating the structural capabilities, rather than the scientific breadth, of the model. Three different atmospheric composition options (preindustrial Earth, the aerosol-free and ozone-free atmosphere used in ROCKE-3D 1.0, and an anoxic atmosphere with no aerosols), three ocean configurations (prescribed, Q-flux, and dynamic), and two resolutions are described: the medium resolution (4x5 degrees in latitude and longitude, previously used in ROCKE-3D 1.0) and the fine resolution, which has double the resolution in the atmosphere and 4 times the horizontal and 3 times the vertical resolution in the ocean. Finally, for the land surface hydrology, we have introduced generalized physics for arbitrary topography in the pooling and evaporation of water and river transport of water between grid cells, as well as for the vertical stratification of temperature in dynamic lakes. We quantify how the different component choices affect model results and discuss the strengths and limitations of using each component, together with how one can select which component to use. ROCKE-3D is publicly available, and tutorial sessions are available for the community, greatly facilitating its use by any interested group.

AbstractRobust estimates of historical changes in aerosols are key for accurate constraints on climate sensitivity. Dry deposition is a primary sink of aerosols from the atmosphere. However, most global climate models do not accurately represent observed strong dependencies of dry deposition following turbulent transport on aerosol size. It is unclear whether there is a substantial impact of mischaracterized aerosol deposition velocities on historical aerosol changes. Here we describe improved mechanistic representation of aerosol dry deposition in the NASA Goddard Institute for Space Studies (GISS) global climate model, ModelE, and illustrate the impact on 1850–2000 changes in global aerosol burdens as well as aerosol direct and cloud albedo effects using a set of 1850 and 2000 time slice simulations. We employ two aerosol configurations of ModelE (a “bulk” mass‐based configuration and a configuration that more explicitly represents aerosol size distributions, internal mixing, and microphysics) to explore how model structural differences in aerosol representation alter the response to representation of dry deposition. Both configurations show larger historical increases in the global burdens of non‐dust aerosols with the new dry deposition scheme, by 11% in the simpler mass‐based configuration and 23% in the more complex microphysical configuration. Historical radiative forcing responses, which vary in magnitude from 5% to 12% as well as sign, depend on the aerosol configuration.

Marine oxygen plays a fundamental role in regulating the transfer of organic carbon and nutrients to their dissolved inorganic forms, serving as the terminal electron acceptor for heterotrophic respiration. Oxygen can become limiting to these processes in coastal and open-ocean oxygen deficient zones (ODZs). The maintenance of ODZs depends on the balance between physical processes such as ventilation and biogeochemical processes such as remineralization. These two processes act in opposing directions on ODZs, with ventilation responsible for transporting oxygen from the surface, where it is near saturation with the partial pressure of O-2 in the atmosphere, to deeper oxygen-depleted layers, and remineralization acting to consume oxygen through biogeochemical processes such as microbial degradation throughout the water column. Remineralization is represented in all CMIP6 models, but the actual parameterizations, as well as their magnitude, are widely varying. In this study, we examine the sensitivity to remineralization of the equatorial Pacific ODZ (O-2 <60 mu mol/kg) in a model; the NASA GISS Ocean Biogeochemical Model (NOBM-G) embedded in the NASA-GISS coupled atmosphere-ocean model. We find that increasing the remineralization rate shoals the ODZ onset depth (i.e. the regionally averaged upper bound of the ODZ), and decreases ODZ volume and regional-averaged thickness. Changes in biological consumption and vertical convergence of oxygen are identified as the primary contributors to changes in ODZ onset depth. ODZ thickness is mainly influenced by the shoaling of the bottom boundary of the ODZ, which decreases with maximum remineralization rate due to decreasing oxygen consumption and increasing horizontal oxygen convergence in deep waters. On the other hand, vertically-averaged ODZ area has a more complex, non-monotonic relationship with maximum remineralization rate. While these findings suggest an important role for remineralization in determining the shape of the ODZ in our model, the relative importance of remineralization vs. other physical parameterizations remains to be established. While our results reflect the structure of our ocean biogeochemical model, the relationship of remineralization to ODZ characteristics in other models should be examined to better inform model inter-comparisons.

Both the Greenland and Antarctic ice sheets have been melting at an accelerating rate over recent dec-ades. Meltwater from Greenland might be expected to initiate a climate response that is distinct, and perhaps different from, that associated with Antarctic meltwater. Which one might elicit a greater climate response, and what mecha-nisms are involved? To explore these questions, we apply climate response functions (CRFs) to guide a series of meltwater -perturbation experiments using a fully coupled climate model. In both hemispheres, meltwater drives atmospheric cooling, sea ice expansion, and strengthened Hadley and Ferrel cells. Greenland meltwater induces a slowdown of the Atlantic merid-ional overturning circulation (AMOC) and a cooling of the subsurface ocean in the northern high latitudes. Antarctic meltwater, instead, induces a slowdown of the Antarctic Bottom Water formation and a warming of the subsurface ocean around Antarctica. For melt rates up to 2000 Gt yr21, the climate response is rather linear. However, as melt rates increase to 5000 Gt yr21, the climate response becomes nonlinear. Due to a collapsed AMOC, the climate re-sponse is superlinear at high Greenland melt rates. Instead, the climate response is sublinear at high Antarctic melt rates, due to the halting of the northward expansion of Antarctic sea ice by warm surface waters. Finally, in the linear limit, we use CRFs and linear convolution theory to make projections of important climate parameters in response to meltwater scenarios, which suggest that Antarctic meltwater will become a major driver of climate change, dominating that of Greenland meltwater, as the current century proceeds.

Abstract Recent mass loss from ice sheets and ice shelves is now persistent and prolonged enough that it impacts downstream oceanographic conditions. To demonstrate this, we use an ensemble of coupled GISS‐E2.1‐G simulations forced with historical estimates of anomalous freshwater, in addition to other climate forcings, from 1990 through 2019. There are detectable differences in zonal‐mean sea surface temperatures (SST) and sea ice in the Southern Ocean, and in regional sea level around Antarctica and in the western North Atlantic. These impacts mostly improve the model's representation of historical changes, including reversing the forced trends in Antarctic sea ice. The changes in SST may have implications for estimates of the SST pattern effect on climate sensitivity and for cloud feedbacks. We conclude that the changes are sufficiently large that model groups should strive to include more accurate estimates of these drivers in all‐forcing historical simulations in future coupled model intercomparisons.

A bi-stable mode of the Atlantic Meridional Overturning Circulation (AMOC) is found in a 10-member ensemble simulation of the SSP2-4.5 scenario using the NASA GISS-E2-1-G climate model. Local feedbacks in the subpolar North Atlantic region in conjunction with internal variability in sea-ice transport and melt play a critical role in causing the divergent behavior of the AMOC in the ensemble members. While other fully coupled models have demonstrated the important role of surface freshening in leading to AMOC shutdown, either through hosing experiments or increased precipitation and greenhouse gas warming at high latitudes, in the GISS simulations, there are no external freshwater perturbations. This is the first time that a CMIP-class model has shown such a bifurcation across an initial condition ensemble.

Climate models project a future weakening of the Atlantic meridional overturning circulation (AMOC), but the impacts of this weakening on climate remain highly uncertain. A key challenge in quantifying the impact of an AMOC decline is in isolating its influence on climate, relative to other changes associated with increased greenhouse gases. Here we isolate the climate impacts of a weakened AMOC in the broader context of a warming climate using a unique ensemble of Shared Socioeconomic Pathway (SSP) 2-4.5 integrations that was performed using the Climate Model Intercomparison Project phase 6 (CMIP6) version of the NASA Goddard Institute for Space Studies ModelE (E2.1). In these runs internal variability alone results in a spontaneous bifurcation of the ocean flow, wherein 2 out of 10 ensemble members exhibit an entire AMOC collapse, while the other 8 members recover at various stages despite identical forcing of each en-semble member and with no externally prescribed freshwater perturbation. We show that an AMOC collapse results in an abrupt northward shift and strengthening of the Northern Hemisphere (NH) Hadley cell (HC) and intensification of the northern midlatitude eddy-driven jet. We then use a set of coupled atmosphere-ocean abrupt CO2 experiments spanning the range 1 times to 5 times CO2 (1x to 5xCO2) to show that this response to an AMOC collapse results in a nonlinear shift in the NH circulation moving from 2xCO2 to 3xCO2. Slab-ocean versions of these experiments, by comparison, do not capture this nonlinear behavior. Our results suggest that changes in ocean heat flux convergences associated with an AMOC collapse-while highly uncertain-can result in profound changes in the NH circulation and continued efforts to constrain the AMOC response to future climate change are needed.

A 10-member ensemble simulation with the NASA GISS-E2-1-G climate model shows a clear bifurcation in the Atlantic meridional overturning circulation (AMOC) strength under the SSP2-4.5 extended scenario. At 26 & DEG;N, the bifurcation leads to 8 strong AMOC and 2 much weaker AMOC states, while at 48 & DEG;N, it leads to 8 stable AMOC-on and 2 nearly AMOC-off states, the latter lasting approximately 800 years. A variety of fully coupled models have demonstrated tipping points in AMOC through hosing experiments, i.e., prescribing sufficient freshwater inputs in the subpolar North Atlantic. In the GISS simulations, there are no external freshwater perturbations. The bifurcation arises freely in the cou-pled system and is the result of stochastic variability (noise-induced bifurcation) associated with sea ice transport and melt-ing in the Irminger Sea after a slowing of the greenhouse gas forcing. While the AMOC strength follows the near shutdown of the Labrador Sea deep convection initially, the Irminger Sea salinity and deep mixing determine the timing of the AMOC recovery or near collapse at 48 & DEG;N, which varies widely across the ensemble members. Other feedbacks such as ice-albedo, ice-evaporation, E -P , and the overturning salt-advection feedback play a secondary role that may enhance or reduce the primary mechanism which is ice melt. We believe this is the first time that a coupled climate model has shown such a bifurcation across an initial condition ensemble and might imply that there is a chance for significant and prolonged AMOC slow down due to internal variability of the system.SIGNIFICANCE STATEMENT: We believe this is the first time that divergent AMOC behavior has been reported for an ensemble of Earth system model simulations using identical climate forcing and no prescribed freshwater pertur-bations. This response is a manifestation of noise-induced bifurcation, enhanced by feedbacks, revealing the role stochastic (or intrinsic) variability may play in AMOC stability.

AbstractThe next‐generation global climate model from the NASA Goddard Institute for Space Studies, GISS‐E3, contains many improvements to resolution and physics that allow for improved representation of tropical cyclones (TCs) in the model. This study examines the properties of TCs in two different versions of E3 at different points in its development cycle, run for 20 years at 0.5° resolution, and compares these TCs with observations, the previous generation GISS model, E2, and other climate models. E3 shares many TC biases common to global climate models, such as having too few tropical cyclones, but is much improved from E2. E3 produces strong enough TCs that observation‐based wind speed thresholds can now be used to detect and track them, and some storms now reach hurricane intensity; neither of these was true of E2. Model development between the first and second versions of E3 further increased the number and intensity of TCs and reduced TC count biases globally and in most regions. One‐year sensitivity tests to changes in various microphysical and dynamical tuning parameters are also examined. Increasing the entrainment rate for the more strongly entraining plume in the convection scheme increases the number of TCs (though also affecting other climate variables, and in some cases increasing biases). Variations in divergence damping did not have a strong effect on simulated TC properties, contrary to expectations based on previous studies. Overall, the improvements in E3 make it more credible for studies of TC activity and its relationship to climate.

We developed a demographic vegetation model, BiomeE, to improve the modeling of vegetation dynamics and ecosystem biogeochemical cycles in the NASA Goddard Institute of Space Studies' ModelE Earth system model. This model includes the processes of plant growth, mortality, reproduction, vegetation structural dynamics, and soil carbon and nitrogen storage and transformations. The model combines the plant physiological processes of ModelE's original vegetation model, Ent, with the plant demographic and ecosystem nitrogen processes that have been represented in the Geophysical Fluid Dynamics Laboratory's LM3-PPA. We used nine plant functional types to represent global natural vegetation functional diversity, including trees, shrubs, and grasses, and a new phenology model to simulate vegetation seasonal changes with temperature and precipitation fluctuations. Competition for light and soil resources is individual based, which makes the modeling of transient compositional dynamics and vegetation succession possible. Overall, the BiomeE model simulates, with fidelity comparable to other models, the dynamics of vegetation and soil biogeochemistry, including leaf area index, vegetation structure (e.g., height, tree density, size distribution, and crown organization), and ecosystem carbon and nitrogen storage and fluxes. This model allows ModelE to simulate transient and long-term biogeophysical and biogeochemical feedbacks between the climate system and land ecosystems. Furthermore, BiomeE also allows for the eco-evolutionary modeling of community assemblage in response to past and future climate changes with its individual-based competition and demographic processes.

Over the CMIP6 historical period (1850–2014), aerosols provided the largest negative forcing compared to all other climate forcings via their ability to absorb or scatter solar radiation and alter clouds. Aerosols played an important role in counterbalancing warming by greenhouse gases (GHGs). Here we study aerosol forcing trends in the CMIP6 simulations of the NASA Goddard Institute for Space Studies (GISS) ocean‐atmosphere ModelE version 2.1 (GISS‐E2.1‐G) using a fully coupled atmospheric composition configuration, including interactive gas‐phase chemistry, and either an aerosol microphysical (MATRIX) or a mass‐based aerosol (OMA) module. Simulations of the CMIP6 historical period are analyzed as well as four Shared Socioeconomic Pathway (SSP) future scenarios for 2015–2100: SSP1‐2.6, SSP2‐4.5, SSP3‐7.0, and SSP5‐8.5. The main conclusion of this study is that aerosol forcing in the GISS model has reached its turning point, switching from globally increasing to a decreasing trend in the first decade of the 21st century. This result is robust, independent of which aerosol module or SSP scenario is used. Non‐linear aerosol‐cloud interactions dominate as a forcing agent over aerosol‐radiation interactions. Aerosols' ability to counterbalance GHG forcing on the global scale is today at a level comparable to that at the beginning of the last century. In the 1980s, the decade of largest global aerosol loads, aerosols balanced up to 80% of GHG forcing. As a consequence, global warming of the last decades, which is primarily driven by greenhouse gases, has been augmented by the effect of decreasing aerosol cooling in our model. By the end of this century, following the SSP scenarios, aerosols will only counterbalance 0%–20% of GHG forcing, depending on model and on scenario.

Daytime heat is often associated with reduced soil moisture and cloud cover, while nighttime heat is connected to high humidity and increased cloud cover. Due to these differing mechanisms, compound daytime and nighttime heat events may respond differently to major anthropogenic forcings (greenhouse gases, anthropogenic aerosols, land-use and land-cover change). Here, we use GISS ModelE2.1-G historical single-forcing runs from 1955 to 2014 to examine how individual anthropogenic forcings affect compound heat events—specifically warm daytime and nighttime temperatures compounded with dry precipitation or high humidity conditions. We show that greenhouse gases alone amplify the natural frequency of warm-dry events by 1.5–5 times and warm-humid events by 2–9 times in tropical and extratropical latitudes. Conversely, aerosols and land-use/land-cover change reduce the frequency of these events, resulting in more modest increases and in some regions, declines, in the historical ‘all-forcings’ scenario. Individually, aerosol effects are stronger and more widespread compared to land-use, oftentimes reducing the natural frequency of these events by 60%–100%. The responses of these compound events are primarily driven by changes in daytime and nighttime temperatures through large-scale warming via greenhouse gases and cooling from aerosols and land-use/land-cover change. However, changes in warm-dry events are amplified in regions with concurrent precipitation declines (e.g. Central America, Mediterranean regions) and warm-humid events are amplified by global concurrent humidity increases. Additionally, we find differences between daytime and nighttime compound responses in the historical experiment that can be traced back to the individual forcings. In particular, aerosols produce a greater cooling effect on daytime relative to nighttime temperatures, which notably results in a historical reduction of Northern Hemisphere daytime warm-dry events relative to natural conditions. Our analysis provides a more comprehensive understanding of the significant impacts of different anthropogenic climate forcings on daytime and nighttime warm-dry and warm-humid events, informing future risk and impact assessments.

Abstract This paper presents the response to anthropogenic forcing in the GISS‐E2.1 climate models for the 21st century Shared Socioeconomic Pathways emission scenarios within the Coupled Model Intercomparison Project Phase 6 (CMIP6). The experiments were performed using an updated and improved version of the NASA Goddard Institute for Space Studies (GISS) coupled general circulation model that includes two different versions for atmospheric composition: A non‐interactive version (NINT) with prescribed composition and a tuned aerosol indirect effect and the One‐Moment Aerosol model (OMA) version with fully interactive aerosols which includes a parameterized first indirect aerosol effect on clouds. The effective climate sensitivities are 3.0°C and 2.9°C for the NINT and OMA models, respectively. Each atmospheric version is coupled to two different ocean general circulation models: The GISS ocean model (E2.1‐G) and HYCOM (E2.1‐H). We describe the global mean responses for all future scenarios and spatial patterns of change for surface air temperature and precipitation for four of the marker scenarios: SSP1‐2.6, SSP2‐4.5, SSP4‐6.0, and SSP5‐8.5. By 2100, global mean warming ranges from 1.5°C to 5.2°C relative to 1,850–1,880 mean temperature. Two high‐mitigation scenarios SSP1‐1.9 and SSP1‐2.6 limit the surface warming to below 2°C by the end of the 21st century, except for the NINT E2.1‐H model that simulates 2.2°C of surface warming. For the high emission scenario SSP5‐8.5, the range is 4.6–5.2°C at 2100. Due to about 15% larger effective climate sensitivity and stronger transient climate response in both NINT and OMA CMIP6 models compared to CMIP5 versions, there is a stronger warming by 2100 in the SSP emission scenarios than in the comparable Representative Concentration Pathway (RCP) scenarios in CMIP5. Changes in sea ice area are highly correlated to global mean surface air temperature anomalies and show steep declines in both hemispheres, with the largest sea ice area decreases occurring during September in the Northern Hemisphere in both E2.1‐G (−1.21 × 106 km2/°C) and E2.1‐H models (−0.94 × 106 km2/°C). Both coupled models project decreases in the Atlantic overturning stream function by 2100. The largest decrease of 56%–65% in the 21st century overturning stream function is produced in the warmest scenario SSP5‐8.5 in the E2.1‐G model, comparable to the reduction in the corresponding CMIP5 GISS‐E2 RCP8.5 simulation. Both low‐end scenarios SSP1‐1.9 and SSP1‐2.6 also simulate substantial reductions of the overturning (9%–37%) with slow recovery of about 10% by the end of the 21st century (relative to the maximum decrease at the middle of the 21st century).

This proposal is emerging from GISS ModelE3 ESM development in the area of cloud physics, so we begin with an example of research needs/gaps from that work. Here, some of our greatest development concerns arise where we lack fundamental process-level understanding, as in ice formation. Namely, it is currently unclear what is the main process that is forming the majority of ice crystals in commonly occurring convection, apparently via secondary ice production at warm temperatures. We are keenly awaiting laboratory data for candidate mechanisms, which is not yet in hand to crucially establish their efficiency. Our progress is also hampered by a lack of uncertainty characterization in currently available measurements of ice crystal number size distributions. Furthermore, the same multiplication process may be responsible for a majority of ice crystals in many extratropical mixed-phase clouds, whose variable representation in CMIP6 ESMs may be a leading cause of differences in cloud phase feedback and ECS. Yet we have been required to deliver an ESM with the cloud physics knowledge at hand. The proposed grand challenge project is AI-enabled via application of machine learning (ML) to climate model and observational data streams (focal area 3), and applications include AI-guided observing system design and model/component/parameterization selection (areas 1 and 2). The project is structurally agnostic as to whether model or observing system components use AI approaches or not, but uncertainties must be estimated and propagatable in both.