Between these approaches, a persistent “gap between simulation and understanding” (Held 2005, see also Balaji et al. 2022) challenges our ability to transfer insight from conceptual models to reality, and distill the physical mechanisms underlying the behavior of state-of-the-art ESMs. This calls for a concerted effort to learn from the entire model hierarchy — or rather, model spectrum — to understand the differences and similarities across its various levels of complexity for increased confidence in climate predictions.
A diverse and well-integrated model ecosystem is also an indispensable prerequisite for the timely assessment of climate risks and effective decision-making. This places renewed emphasis on the concept of fit-for-purpose modeling, which is intrinsically linked to the climate model spectrum through the need to understand what levels of complexity are required and sufficient for a given scientific question or application. The climate community is increasingly interested in making models useful also beyond the academic domain (Mansfield et al., 2023).
In this session, we bring together contributions from all subfields of climate science that showcase how different modeling approaches advance our understanding of the Earth system, highlight inconsistencies in the model spectrum, and/or enable applications in climate impact projections. The key goal is to foster exchange between researchers working on different rungs of the model complexity ladder, focusing on process understanding. Contributions may employ dynamical systems models, physics-based low-order models, explainable machine learning, fast climate models and Earth System Models of Intermediate Complexity (EMICs), simplified or idealized setups of ESMs, CMIP models, and km-scale models. Processes and phenomena of interest include the Earth system response to transient forcing, tipping behavior, climate variability and extremes, and predictability.
As Earth warms, regional climate signals are accumulating. Some signals, for example, land warming more than the ocean and the Arctic warming the most, were expected and successfully predicted. Underlying this success was the application of physical laws across a climate model hierarchy under the assumption that large and small spatial scales are well separated. With additional warming, however, discrepancies between real-world signals and model predictions are accumulating, especially at regional scales. In this talk, we will highlight the emerging list of model-observation discrepancies in historical trends. We demonstrate how the climate model hierarchy can be used to understand the physical processes underlying these discrepancies. We argue that progress can be made by filling gaps in the hierarchy and making more process-informed observations.
How to cite: Shaw, T. and Kang, J.: Understanding regional discrepancies using the climate model hierarchy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13950, https://doi.org/10.5194/egusphere-egu26-13950, 2026.
The climate and carbon cycle interact in multiple ways. An increase in carbon dioxide in the atmosphere warms the climate through the greenhouse effect, but also leads to uptake of CO2 by the land and ocean sink, a negative feedback. However, the warming associated with a CO 2 increase is also expected to suppress carbon uptake, a positive feedback. This study addresses the question: “under what circumstances could the climate–carbon cycle system become unstable?” It uses both a reduced form model of the climate–carbon cycle system as well as the complex land model JULES, combined with linear stability theory, to show that: (i) the key destabilising loop involves the increase in soil respiration with temperature; (ii) the climate–carbon system can become unstable if either the climate sensitivity to CO2 or the sensitivity of soil respiration to temperature is large, and (iii) the climate–carbon system is stabilized by land and ocean carbon sinks that increase with atmospheric CO2 , with CO2-fertilization of plant photosynthesis playing a key role. For central estimates of key parameters, the critical equilibrium climate sensitivity (ECS) that would lead to instability at current atmospheric CO2 lies between about 11K (for large CO2 fertilization) and 6K (for no CO2 fertilization). Given the apparent stability of the climate–carbon cycle, we can view these parameter combinations as implausible. The latter value is close to the highest ECS values amongst the latest Earth Systems Models. We find that the stability of the climate–carbon system increases with atmospheric CO2 , such that the glacial CO2 concentration of 190 ppmv would be unstable even for ECS greater than around 4.5 K in the absence of CO2 fertilization of land photosynthesis.
How to cite: Clarke, J., Huntingford, C., Ritchie, P., Varney, R., Williamson, M., and Cox, P.: Conditions for instability in the climate–carbon cycle system, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12377, https://doi.org/10.5194/egusphere-egu26-12377, 2026.
The stability of the Atlantic Meridional Overturning Circulation (AMOC) under future climate change remains uncertain. While most climate models across the model hierarchy project a weakening or collapse under freshwater forcing, transient simulations under increasing CO2 levels also commonly show a weakening or even a collapse of the AMOC. However, longer equilibrium experiments---primarily conducted with lower-complexity models due to computational costs---show more varied responses to CO2 forcing. While most models show an initial weakening of the AMOC, some models equilibrate to a weak AMOC state only at very high CO2 levels, while others equilibrate to a stronger-than-present AMOC. One such model is the intermediate complexity model CLIMBER-X, which (in equilibrium) shows that the AMOC strengthens until at least 16 times preindustrial CO2 levels are reached. However, during the transient phase of increasing CO2, the AMOC weakens. This suggests that the AMOC's transient response may differ from its equilibrium behavior. This raises the question: can the AMOC collapse under rapid and high CO2 increase, even if a stable equilibrium state exists?
We show that the AMOC exhibits rate-dependent tipping; when CO2 increases fast enough and reaches sufficiently high levels, the AMOC can fully collapse. This occurs under very high forcing, starting from 7 times preindustrial CO2 levels and a rate of 2.0% ppm/yr CO2 increase. This collapse occurs despite the existence of a stable AMOC at equilibrium. By examining the physical processes through which the collapse occurs, we contribute to the understanding of the AMOC response in a warming climate. By also incorporating freshwater forcing, we assess the risks of rapid warming on the AMOC stability. Our results show that even models with a stable equilibrium AMOC under high CO2 levels can experience weakening during the transient phase or even collapse. This highlights the need to assess both the rate and magnitude of CO2 forcing when assessing the stability of the AMOC. While this effect occurs at very high CO2 levels in CLIMBER-X, the role of the rate of CO2 increase may become relevant at lower CO2 levels when combined with freshwater forcing. Our findings demonstrate that the AMOC can undergo rate-dependent tipping under rapid and high CO2increase, even if a stable AMOC exists at very high CO2 levels.
How to cite: Terpstra, S., Falkena, S., Bastiaansen, R., and von der Heydt, A.: Rate-dependent Tipping of the AMOC under CO2 increase in an Intermediate Complexity Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14266, https://doi.org/10.5194/egusphere-egu26-14266, 2026.
A collapse of the Atlantic Meridional Overturning Circulation (AMOC) would have strong consequences for the global climate system. Assessing whether the AMOC will collapse in the future is difficult since current Earth System Models (ESMs) have biases. An earlier study using an intermediate complexity Earth system model (EMIC) showed the potential effect of freshwater biases on AMOC stability. However, the used model has a limited ocean model with respect to the used resolution and processes represented compared to ESMs. Here, we supplement the EMIC simulations with simulations of an ocean-only model using the same resolution as is typically used in ESMs. This allows us to study the effect of ocean resolution on the physical mechanism controlling the effect of freshwater biases on AMOC stability. We find that both the intermediate complexity and the ocean-only model behave qualitatively similar. In both models freshwater biases influence AMOC stability where negative (positive) biases in the Indian Ocean tend to stabilize (destabilize) the AMOC, whereas the opposite applies to biases in the Atlantic Ocean. Based on the freshwater biases present in most ESMs, our results suggest that most ESMs have a too stable AMOC and might therefore underestimate the probability of an AMOC collapse under future emission scenarios.
How to cite: Boot, A. and Dijkstra, H.: The effect of freshwater biases on AMOC stability across the model complexity spectrum, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16747, https://doi.org/10.5194/egusphere-egu26-16747, 2026.
Studies of climate sensitivity and feedbacks typically employ a suite of models with similar base climates but different model physics. Such an approach is useful for uncovering how changes to physical processes affect the climate response to changes in radiative forcing, but obscures the dependence of the climate response on the initial state of the climate itself. In order to better understand this dependence, we study the response to radiative forcing of two nearly identical configurations of the Community Earth System Model (CESM) with production-grade physics and resolutions that have dramatically different climates. The first, called Aqua, is completely covered with a uniform-depth ocean except for two 10º-wide polar continents to avoid the polar singularities in the ocean model. The second, Ridge, is identical to Aqua except for the presence of a thin ridge continent connecting the two polar caps. The ridge supports gyres in the ocean and leads to a warm, ice-free climate resembling a global Pacific Ocean, with a warm pool and cold tongue in the tropical ocean connected by a Walker circulation in the atmosphere. In contrast, the mean climate of Aqua is zonally symmetric and dominated by a global cold belt in the ocean driven by vigorous equatorial upwelling. The lack of gyres leads to a deep oceanic thermocline and reduces meridional heat transport, which allows for the development of persistent sea ice at high latitudes.
These two mean climates are perturbed by increasing atmospheric CO2 concentration at a rate of 1% per year until quadrupling. Aqua initially warms more slowly than Ridge, with the transient climate response (TCR) at doubling 23% smaller for Aqua than Ridge. After doubling, however, Aqua begins to warm faster than Ridge and Aqua’s global mean temperature surpasses Ridge’s at quadrupling. A linear feedback analysis is used to gain insight into the time-evolving responses of these two configurations to increased CO2 concentration. At all stages, Aqua’s net top-of-the-atmosphere heating is greater than Ridge’s. At early times, this is due to high clouds replacing low clouds in Aqua’s high latitudes, but decreasing surface albedo due to sea-ice loss eventually becomes a dominant factor. Aqua’s deep thermocline supports a higher ocean heat uptake (OHU) efficiency relative to Ridge that initially offsets these positive feedbacks and results in Aqua’s lower TCR. As CO2 concentration approaches quadrupling, the combined effects of declining OHU efficiency and a strengthening ice-albedo feedback drive Aqua’s warming to temperatures compatible to Ridge. In the century following quadrupling, Aqua warms several Kelvin more than Ridge.
These idealized systems can shed light on the fundamental aspects of Earth’s climate system—such as how the response to radiative forcing depends on the base climate—that might be obscured in more complex configurations.
How to cite: Pitt Wolfe, C., Ma, Y., Katavouta, A., Reed, K., and Williams, R.: Exploring state dependence of the climate response to radiative forcing using two idealized coupled climate models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12666, https://doi.org/10.5194/egusphere-egu26-12666, 2026.
The history of climate modelling is one of increasing complexity and increasing resolution, driven by and constrained by the available computational capacity. These models are widely used, directly and indirectly, to support policy and adaptation decisions across society. They are also used in academic studies across a range of disciplines to study the response of the climate system to future atmospheric greenhouse gas concentrations on multi-decadal timescales. These are extrapolatory endeavours in a non-stationary system without possibility of relevant verification.
There has been much research on individual and multi-model analyses in this context. Here I will instead discuss how the targets of our endeavours (particularly the support of societal decisions) demands a rethinking of our modelling activities. I will highlight the need to reflect on the minimum requirements for ensemble size and ensemble variety, and the role of a hierarchy of models in providing the best possible information to stakeholders across society.
These issues will be discussed in the light of a recent meeting on the foundations of climate change science attended by over 70 researchers across a variety of disciplines. The meeting was entitled “How to spend 15 billion dollars?: A workshop on how to make climate change modelling more robust and more useful to society.” It gathered expertise from disciplines as diverse as earth system modelling, integrated assessment modelling, philosophy, economics, maths, statistics and finance.
Here I will present the key messages coming out of this meeting alongside the themes presented in a recent essay on the subject, “A Model of Catastrophe”[1].
[1] Stainforth, D.A., “A Model of Catastrophe”, Aeon.co, 2025 (https://aeon.co/essays/todays-complex-climate-models-arent-equivalent-to-reality)
How to cite: Stainforth, D.: Designing Climate Change Modelling to Support Societal Decisions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8434, https://doi.org/10.5194/egusphere-egu26-8434, 2026.
Climate impact assessments increasingly rely on high-resolution climate and forcing datasets, under the premise that finer detail enhances both the accuracy and the policy relevance of projections. Systematic evaluations of when and where higher resolution data improve model outcomes remain limited, and it is still unclear whether increasing spatial resolution consistently enhances climate impact model performance across application areas, regions, and forcing variables. Here we show that improvements in climate input accuracy and impact model performance are most pronounced when moving from coarse (60 km) to intermediate (10 km) resolution, while further refinement to 3 km and 1 km provides more modest and inconsistent benefits. Using the cross-sectoral model simulations from the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP), we demonstrate that higher resolution substantially improves model skill in temperature-sensitive impact models and topographically complex regions, whereas precipitation-driven and low-relief systems show less consistency to increase performance with resolution. For temperature, both climate inputs and model outputs improved most strongly at the 60 km → 10 km transition, with diminishing gains at finer scales. A similar result emerged for precipitation, although some models even exhibited reduced performance when resolution increased beyond 10 km. These results highlight that optimal resolution depends on sectoral and regional context, and point to the need for improving model process representation and downscaling techniques so that added spatial detail can translate into meaningful performance gains. For data providers, this implies prioritizing investments in resolutions that maximize improvements where they matter most, while for modeling groups and users, it underscores the need for explicit benchmarking of resolution choices. More broadly, this work advances the design of consistent, efficient, and policy-relevant multi-sectoral climate impact assessments by clarifying when high-resolution data meaningfully enhance outcomes.
How to cite: Malle, J., Reyer, C., and Karger, D. and the ISIMIP modellers and sector coordinators: When and where higher-resolution climate data improve impact model performance, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5437, https://doi.org/10.5194/egusphere-egu26-5437, 2026.
Today's climate adaptation and mitigation planning tasks require rapid access to large ensembles of climate projections for a wide range of emissions scenarios, including overshoot scenarios. While Earth system models (ESMs) provide physically consistent projections, their high computational cost limits scenario exploration. Climate emulators - statistical or machine-learning-based models trained on ESM data to generate data replicating the ESMs behaviour for a multitude of emissions scenarios - are therefore proposed to deliver these projections efficiently. Here we present the novel modular SCALES–MESH emulator framework, combining physics-based regional projections with AI downscaling capabilities. The SCALES module translates projections of global mean surface air temperature into regional surface air temperature projections aggregated over the AR6-IPCC regions, while the MESH module performs spatio-temporal downscaling to gridded fields using a conditional score-based generative model. MESH is trained on multiple datasets and evaluated against parent ESMs using spatial, temporal, and distributional diagnostics. Results show that the emulator captures regional patterns, temporal variability, and probability distributions of emulated climate variables, including during warming and cooling phases of overshoot scenarios. We further demonstrate the potential for transfer learning across ESMs, pointing toward scalable multi-model and resolution-agnostic emulation. Together, SCALES–MESH enables rapid, flexible, and physically grounded exploration of climate futures, supporting decision-relevant climate risk assessment at unprecedented scope.
How to cite: Kain, V., Schwind, N., Högner, A., Shmuel, A., Nauels, A., Nicholls, Z., Zecchetto, M., and Schleussner, C.-F.: Cascaded score-based emulation of Earth system models for impact evaluation with SCALES-MESH , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9869, https://doi.org/10.5194/egusphere-egu26-9869, 2026.
The PRIME (Mathison et al. 2025) framework provides a fast response tool to look at climate impacts for up-to-date mitigation scenarios. PRIME combines the FaIR simple climate model and pattern scaling of Earth System Models (ESMs) with the JULES land surface model to quantify spatially resolved climate impacts. In addition, PRIME samples uncertainty from both the spatial patterns of CMIP6 ESMs and the probabilistic configuration of the latest version of FaIR.
We present applications of this framework to explore impacts on both the earth system and potential impacts on societies, using new scenarios produced for CMIP7. From an earth system perspective, we use an updated configuration of JULES incorporating permafrost processes and fire to look at the impact of the northern high latitude net ecosystem balance. In terms of societal impacts, we simulate the potential impacts of climate change on agricultural drought of rain fed crops during the growing season. This analysis includes a quantification of the uncertainty derived from the global mean climate response and the spatial responses of ESMs. Results from PRIME will also be part of the FastMIP project.
How to cite: Hooke, D., Mathison, C., Burke, E., Jones, C., Gohar, L., and Wiltshire, A.: From scenarios to impacts – an emulation of regional climate impacts and their uncertainties using the CMIP7 mitigation scenarios , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19277, https://doi.org/10.5194/egusphere-egu26-19277, 2026.
Across the globe, today’s young generations will be more frequently exposed to climate extremes over their lifetime than earlier generations. Previous work has established this finding by combining simulations of historical and projected trends in climate extremes together with data on past and future demographic changes (Thiery et al. 2021, Grant et al. 2025). However, it has so far focused on a limited set of climate extreme indicators, using climate (impact) simulations from ISIMIP2 and demographics datasets that are now outdated, and did not fully assess uncertainty across the climate impact modelling chain.
We now build on this existing lifetime exposure framework and combine it with a chain of emulators constituted of a Simple Climate Model (SCM) and the Rapid Impact Model Emulator Extended (RIME-X, Schwind et al., submitted). RIME-X can translate the GMT distributions generated by an SCM for a given emission scenario into spatially explicit distributions of climate or climate impact indicators. It has already been used to produce projections for 40+ indicators derived from ISIMIP3 and other climate model simulations, and this list can be extended to further indicators whose evolution predominantly depends on the level of global warming and for which historical and future simulations are available.
We also update the lifetime exposure framework to consider more recent demographic data, and package it into a GitHub repository called dem4cli (short for ‘demographics for climate’) that will be made publicly available. We use spatially explicit population reconstructions and projections from the COMPASS project, and national-level life expectancy and cohort size estimates and projections from UNWPP2024.
This work delivers more robust calculations of lifetime exposure to changes in extremes or climate impacts, by leveraging the ability of the SCM-RIME-X emulator chain to represent both their forced response to emissions as well as the combined uncertainty arising from the GMT response to emissions, the local climate response to global warming, and interannual variability, in combination with updated demographic data. This new framework is designed to generate such policy-relevant information in a more flexible and systematic manner, as it can in theory be applied to any available emission or GMT trajectories, and extended to a broad range of climate hazards.
Thiery, W. et al. Intergenerational inequities in exposure to climate extremes. Science 374, 158–160 (2021)
Grant, L., Vanderkelen, I., Gudmundsson, L. et al. Global emergence of unprecedented lifetime exposure to climate extremes. Nature 641, 374–379 (2025)
Schwind et al. RIME-X v1.0: Combining Simple Climate Models, Earth System Models, and Climate Impact Models into a Unified Statistical Emulator for Regional Climate Indicators. Geoscientific Model Development (submitted)
How to cite: Lejeune, Q., Pietroiusti, R., Laridon, A., Schwind, N., Schleussner, C.-F., and Thiery, W.: Combining emulators and demographics: Building a flexible toolkit for lifetime exposure assessments, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16683, https://doi.org/10.5194/egusphere-egu26-16683, 2026.
The Global Overturning Circulation (GOC) is characterized by deep water formation in the subpolar North Atlantic, which feeds the southward-flowing branch of the Atlantic Meridional Overturning Circulation (AMOC). In contrast, the North Pacific lacks deep water formation and therefore does not host an analogous Pacific Meridional Overturning Circulation (PMOC). Proxy records, however, indicate that this asymmetric pattern of deep water formation has varied in the past, suggesting that a PMOC likely existed during earlier climate states. Recent studies further show that the development of a PMOC influences the future weakening of the AMOC: climate models that develop a PMOC in response to warming exhibit a stronger decline in AMOC strength. It therefore becomes important to understand under what circumstances a PMOC is likely to develop.
Here, we extend the pycnocline model of Gnanadesikan (1999) to a two-basin configuration, consisting of a narrow basin representing the Atlantic and a wide basin representing the Pacific. By including salinity as a prognostic variable, we find that this two-basin box model may exhibit three distinct overturning states under identical, longitudinally symmetric forcing: (1) an active narrow-basin sinking state, (2) an active wide-basin sinking state, and (3) a state with active sinking in both basins. Overturning states confined to a single basin are stabilized by the salt-advection feedback, whereas the state with sinking in both basins is maintained by a meridional temperature contrast. We find that this latter state becomes the preferred equilibrium when the interhemispheric temperature contrast increases, the northern gyre transport strengthens, and the hydrological cycle weakens. Moreover, we show that this state is more sensitive to high-latitude freshwater fluxes, indicating that a transition to such a state would enhance the projected future weakening of the AMOC. We verify these findings in an uncoupled global circulation model (MITgcm) with a simplified model geometry.
How to cite: Vanderborght, E. and Dijkstra, H.: Multi-stability of the Global Overturning Circulation: A Conceptual Approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7983, https://doi.org/10.5194/egusphere-egu26-7983, 2026.
Earth System Models of Intermediate Complexity (EMICs) allow for fast exploration of large-scale climate dynamics. These models thus enable the development and testing of large-ensemble-based techniques that would be too costly with more realistic climate models.
In this ongoing study we develop a rare event simulation setup to explore the possibility of a spontaneous collapse of the Atlantic Meridional Overturning Circulation (AMOC) in the FAMOUS model. FAMOUS is a low-resolution, coupled atmosphere-ocean general circulation model derived from the UK Met Office’s Unified Model specifically designed for efficient, long-duration and ensemble climate simulations. FAMOUS has previously been used to investigate the hysteresis of the Atlantic Meridional Overturning Circulation under freshwater hosing.
We apply a genealogical particle analysis (GPA) algorithm that is designed to probe the possibility of spontaneous AMOC transitions. The method initiates an ensemble of realisations in the "on"-state of the AMOC and clones ensemble members at regular intervals that are showing a low AMOC.
Contrary to recent results in another EMIC, a straightforward sampling based on the AMOC indicator does not result in any spontaneous transitions to the AMOC "off"-state. To improve the selection of potentially exceedingly rare trajectories, we therefore investigate statistical methods to identify physical variables that correlate with the state of the AMOC ahead of time, to be used as selection criteria in the GPA algorithm.
How to cite: Wouters, J., Hu, G., Bröcker, J., and Smith, R.: Investigating the possibility of rare spontaneous AMOC transitions in the intermediate complexity climate model FAMOUS., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20145, https://doi.org/10.5194/egusphere-egu26-20145, 2026.
The Atlantic Meridional Overturning Circulation (AMOC) is a critical component of the global climate system and its potential for abrupt collapse represents a significant tipping point. Our project investigates whether a persistent negative phase of the North Atlantic Oscillation (NAO), a dominant mode of atmospheric variability, can induce an AMOC collapse in the absence of external perturbations within the coupled PlaSim-LSG climate model of intermediate complexity. A control simulation establishes a baseline climatology, confirming that NAO variability leads AMOC fluctuations by approximately one year. To overcome the computational limitation of simulating rare events, we implement a rare event algorithm (GKLT) that efficiently biases the model toward trajectories with negative NAO conditions over 125-year simulations. The results reveal a fundamental bistability in the system. While persistent negative NAO forcing can trigger an AMOC collapse, the outcome is probabilistic: out of six independent ensemble simulations, four evolved entirely into a collapsed state (∼ 12 Sv), one remained entirely vigorous (∼ 23 Sv) and one split into both outcomes. A cluster-based analysis traces this divergence to the early amplification of small differences in North Atlantic heat fluxes, convection and sea-ice cover. These findings show that internal atmospheric variability alone can force the AMOC across a tipping point, highlighting the role of internal climate dynamics in shaping climate transitions.
How to cite: Magagna, A., Zappa, G., Cini, M., and Corti, S.: Simulating NAO-driven AMOC collapse in the PlaSim-LSG Climate Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21970, https://doi.org/10.5194/egusphere-egu26-21970, 2026.
The Atlantic Meridional Overturning Circulation (AMOC) is a crucial climate component, as its potential collapse would constitute a significant response to Earth’s changing climate. This critical transition has been the subject of numerous studies over the years, both from the aspect of climate modeling and dynamical systems theory. In the context of the latter, climate change is a process in which a complex, chaotic-like system possesses time-dependent parameters, in the form of e.g. the growing CO2 concentration. It has been known that such systems have a chaotic attractor which is also time-dependent, a so-called snapshot attractor. Such objects, and thus the systems they describe, can only be faithfully represented by a probability distribution over an ensemble of simulations, so-called parallel climate realizations.
Based on this probability distribution, we define a novel early warning indicator for crucial transitions such as an AMOC collapse. The AMOC is said to possess a multistable quasipotential landscape, and the collapse is a transition between stable states. We argue that, from the point of view of statistical physics, this is analogous to a phase transition, but in a non-adiabatic setting. As such, the variance of the distribution over the ensemble is expected to develop a local maximum around the transition point, giving rise to a potential early warning by identifying the preceding maximum of its derivative. This method is first demonstrated on a conceptual climate model, before the analysis is carried out on ensemble simulations from the ACCESS-ESM model. The analysis in the former case is simpler, while in the latter, one has to contend with the dependence of the AMOC strength on spatial coordinates, resulting in multiple early warning points for different depths and latitudes.
How to cite: Jánosi, D., Divinszki, F. T., Börner, R., and Herein, M.: Defining an early warning method for an AMOC collapse based on ensemble statistics, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5816, https://doi.org/10.5194/egusphere-egu26-5816, 2026.
The shape and extent of the Greenland Ice Sheet (GIS) during the Holocene remain a matter of considerable debate, with existing studies proposing a wide range of reconstructions. In this study, we aim to combine stable water isotopic information from ice cores with outputs from isotope-enabled climate models to investigate this problem. Directly exploring the space of possible ice sheet geometries through numerical simulations is computationally prohibitive. To address this challenge, we plan to develop a Gaussian process emulator that will serve as a statistical surrogate for the full climate model. The emulator will be trained on the results of a limited number of carefully designed simulations and will be used to enable fast, probabilistic predictions of model outputs at untried inputs. The inputs will consist of GIS morphologies, parameterized using a dimension-reduction technique adapted to the spherical geometry of the ice sheet. Using predictions from the emulator, we will explore the range of ice sheet morphologies that are compatible with available ice-core isotope measurements and other complementary observational data, including those collected during recent KANG-GLAC expeditions, with the goal of ultimately reducing uncertainty in reconstructions of Holocene GIS morphology.
How to cite: Malmierca Vallet, I., Sime, L. C., Voss, J., Fasoli, D., and Hogan, K.: Using Ice Cores and Gaussian Process Emulation to Recover Changes in the Greenland Ice Sheet During the Holocene, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17291, https://doi.org/10.5194/egusphere-egu26-17291, 2026.
Large-scale barotropic flow in the Arctic Ocean is strongly steered by the seafloor topography, yet how this geometry constrains free modes and facilitates inter-basin interactions remains unclear. Free modes conserve potential vorticity and at high latitudes the circulation pathway is enclosed by its sloping two-basin geometry. We begin by presenting a simple two-basin model, representing the Canadian and Eurasian basin respectively, with sloping boundaries and flat bottoms to explore simplified Arctic flow behaviour. Topographic Rossby waves are analytically obtained and the two basins are linked together via a mode-matching framework. We show free modes are tightly constrained to geometry, with basin-trapped dipole wave modes only emerging in certain geometric parameters. We then extend this to a more realistic, multiple-basin Arctic Ocean model that include the Nordic seas, and demonstrate the transmission and exchange of these topographic waves across these multiple sloping basins.
How to cite: Nguyen, M. D. T. and Johnson, E.: Barotropic waves in a sloping two- and multiple-basin Arctic ocean model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6930, https://doi.org/10.5194/egusphere-egu26-6930, 2026.
Offline aridity and drought indices have often implied widespread terrestrial drying under a warming environment, while Earth system models (ESMs) have projected modest changes in land-surface water fluxes. This persistent divergence has been typically attributed to missing vegetation physiological processes in offline frameworks. However, we here show that a more foundational cause is a structural inconsistency embedded in those diagnostics. Conventional potential evapotranspiration (PET) formulations can violate the assumption that precipitation (P) and atmospheric evaporative demand act as independent climatic constraints in the Budyko framework. Using open-water Penman and vegetation-responsive Penman–Monteith formulations forced by reanalysis data and ESM projections, we found that uncorrected PET strongly reflected land–atmosphere feedbacks, leading to pronounced negative P–PET correlations (-0.45 ± 0.29; mean ± s.d.). When PET was thermodynamically deflated, this dependence was largely removed (-0.02 ± 0.42), restoring consistency with the theoretical basis of Budyko-type diagnostics. This structural correction reduced inflation of the aridity index and substantially moderated projected evapotranspiration (ET) trends. Under a business-as-usual scenario, the trend of Budyko-based ET from uncorrected PET (+0.61 mm yr-2) exceeded that of CMIP6 ensemble mean (+0.28 mm yr-2) by more than a factor of two. CEP-deflated PET narrowed this discrepancy (+0.39 mm yr-2), while additional physiological adjustments provided comparatively smaller improvements. We suggest that violations of structural assumptions, rather than missing physiological processes alone, can play a central role in the divergence between offline aridity diagnostics and ESM hydrological projections.
Acknowledgement: This work was jointly supported by the National Research Foundation of Korea (NRF) grants funded by the Korea government (MSIT) (RS-2025-16070291 & RS-2024-00416443).
How to cite: Kim, D. and Choi, M.: Why offline aridity diagnostics overestimate future drying: the role of feedback-inflated evaporative demand, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6189, https://doi.org/10.5194/egusphere-egu26-6189, 2026.
Synthetic time series generation is an essential tool for robustly exploring different climate scenarios and their impacts. While sophisticated generation methods have been developed in the past, they often rely on physical and statistical assumptions and require extensive data for calibration and parameter estimation. We propose a straightforward method for time series generation based on constrained sampling of observations. This approach preserves the physical consistency between variables and maintains the short temporal structure present in the observation. We apply this procedure to generate temperature, precipitation, incoming solar radiation, and wind speed time series sampled from meteorological station observations. We obtain different sets of synthetic time series by constraining the mean temperature according to future scenarios provided by climate model projections. We show that the sampled time series preserve the multivariate dependence structure observed in both historical data and climate projections. While, by design, the method does not generate daily values beyond the observed range, it can simulate multi-day extremes that exceed those in the observational record, such as longer heatwaves. The approach is flexible and can be applied to other variables with other constraints, provided that a sufficiently long observational time series is available and the constraints are compatible with the observed data. The generation procedure may thus prove useful for studying potential future extremes and help in general downscaling tasks.
How to cite: Lehning, M., Milojevic, T., and Rivoire, P.: The future is in the past? A flexible resampling approach to generate multivariate time series, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16369, https://doi.org/10.5194/egusphere-egu26-16369, 2026.
The differentiability of numerical climate models exhibits many advantages over non-differentiable models. Differentiable climate models would be able to optimize parameters and quickly solve for climate equilibrium. They can also be used to find unstable climate equilibrium states that are impossible to identify in time-forwarding models. Differentiability also enables sensitivity studies, such as the impact of initial conditions on predictions, which is the key concept in the 4-dimensional variational method. Finally, differentiable ability also integrates well with the trending data-driven artificial intelligence model, such as NeuralGCM.
Currently, physics-based differentiable coupled climate models are still rare. Some existing ones include: ECMWF Integrated Forecasting System (ECMWF-IFS) and Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). The high scientific value of such a tool warrants development of further differentiable modelling systems.
In this work, we present jax-esm, a differentiable coupler for models written in Python with the JAX framework. JAX is a Python library developed by Google that builds on NumPy and adds automatic differentiation and just-in-time (JIT) compilation. It has been used to develop atmospheric models such as NeuralGCM and jax-gcm. In this example, we couple jax-gcm, a JAX-based atmosphere intermediate model, to a slab ocean model. We demonstrate the optimization of ocean mixed-layer depth and solving for climate equilibrium through differentiability.
How to cite: Hsu, T.-Y., Watson-Parris, D., and Feulner, G.: Jax-esm: a differentiable coupler for jax-based Earth system models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7143, https://doi.org/10.5194/egusphere-egu26-7143, 2026.
The Witwatersrand Basin on the Kaapvaal Craton hosts the world’s largest gold province, with the vast majority of gold concentrated in the 2.90–2.79 Ga Central Rand Group, whereas the slightly older 2.95–2.91 Ga West Rand Group is largely barren despite comparable sedimentary characteristics. This contrast has been attributed to intensified chemical weathering during Central Rand Group times, which promoted enhanced gold mobilisation from the Archaean hinterland. However, the climatic and environmental drivers of this weathering intensification remain poorly constrained. To address this, we investigated Mesoarchaean climate controls using the Planet Simulator (PlaSim), an Earth system model of intermediate complexity. We conducted 140 PlaSim simulations to quantify the climatic sensitivity to atmospheric greenhouse gas concentrations, continental surface area, surface albedo, and land configuration. CO₂-equivalent concentrations (3–30 %), land coverage (8–28 %), and albedo (0.15–0.30) were systematically varied across different land distributions (equatorial, polar and spread over different latitudes).
Next to the well-known effect of global warming under increased greenhouse gas concentrations, our results show that increasing continental area generally results in global cooling due to the higher albedo of land surfaces relative to oceans, particularly when land was concentrated at low latitudes. This cooling effect becomes pronounced once land exceeds approximately 13 % of Earth’s surface. At high latitudes, land has minimal climatic impact because of the low incoming radiation angle that leads to less absorption. Exceptions are noted under conditions of low greenhouse gas concentrations and low surface albedo, at which limited land growth could slightly enhance warming. Among the tested land positions, the scenario with land spread over different latitudes resulted in the highest climate sensitivity.
Overall, our results indicate that land distribution alone was unlikely to have caused global warming during the Mesoarchaean, and this climatic influence was probably dampened by a more rapid carbon cycle at that time. Instead, elevated atmospheric greenhouse gas levels emerge as the dominant driver of warming and enhanced chemical weathering. The climatic transition around ~2.9 Ga may further reflect the emergence of extensive low-albedo mafic or ultramafic surfaces and/or the latitudinal migration of the Kaapvaal Craton into a more radiatively sensitive, low-latitude zone. These combined factors likely contributed to intensified weathering, increased gold leaching, and the gold megaevent responsible for the formation of the Witwatersrand ores.
How to cite: Wasitschek, L., Frimmel, H. E., Hiby, N., and Pollinger, F.: Modelling Mesoarchaean climate: Economic implications , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20975, https://doi.org/10.5194/egusphere-egu26-20975, 2026.
The radiative balance of the early Earth was governed by other boundary conditions than today, including a fainter Sun, elevated greenhouse gas concentrations, and a smaller land surface area. Although the role of atmospheric composition in sustaining habitable surface temperatures during the Mesoarchaean has been extensively investigated, especially to solve the faint young Sun paradox, the climatic impact of land position and distribution under varying albedo remains comparatively underexplored.
We therefore assess how variations in land-surface albedo, land fraction, and land distribution could have modulated Mesoarchaean climate states. Using the Planet Simulator (PlaSim), an intermediate-complexity climate model, we conducted 195 simulations spanning CO₂-equivalent forcing levels of 3–10 % (30,000–100,000 ppm). Land-surface albedo was varied between 0.15 and 0.30, land area between 8 % and 28 %, and idealised land distributions were prescribed, including diagonal, staggered, and mid-latitude configurations. Ocean albedo was held constant at 0.144 to isolate the climatic impact of continental reflectivity.
Across all simulations, global mean temperature responds strongly and non-linearly to both land fraction and land albedo. At low land albedo (0.15) and low to intermediate CO₂-equivalent forcing (3–5 %), increasing land area produces a slight warming trend, despite minimal differences between land and ocean reflectivity. This behaviour indicates that land–ocean contrasts in surface energy partitioning and effective heat capacity can modify global climate even when shortwave albedo contrasts are small. Sensitivity increases abruptly as land albedo rises from 0.20 to 0.25. Beyond this threshold, modest increases in land area result in pronounced global cooling, consistent with a regime shift in the radiative balance. This non-linear response is most prominent at low to intermediate CO₂-equivalent forcing and becomes progressively muted at higher forcing (10 %), where greenhouse effects dampen the temperature response to surface reflectivity changes. The pattern occurs across all land configurations but is amplified when landmasses occupy equatorial to mid-latitudes, where insolation is highest and albedo exerts maximum leverage, whereas high-latitude land has a comparatively weaker effect.
These findings highlight that land surface characteristics such as albedo and distribution were critical for early Earth’s climate. Even under strongly greenhouse-forced atmospheres, surface properties significantly altered the planetary energy budget. Recognising such sensitivities is essential for reconstructing Archaean climate states and assessing the potential for climatic stability under reduced solar luminosity.
How to cite: Hiby, N., Wasitschek, L., and Frimmel, H. E.: Surface albedo as a first-order control on Mesoarchaean climate (PlaSim), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21556, https://doi.org/10.5194/egusphere-egu26-21556, 2026.
Reduced-complexity models (RCMs) are a critical tool in climate science. Their computational efficiency enables applications beyond the reach of more complex models, including uncertainty quantification, the integration of multiple lines of evidence via ensemble constraining, and running large scenario sets in the span of a few days. Thanks to these capabilities, RCMs played important roles in previous IPCC assessments, and are poised to play an important role in the upcoming Seventh Assessment Report (AR7). A key example is evaluating the climate response to the thousands of emissions scenarios in the peer-reviewed literature created with integrated assessment models. However, whether/which RCMs are suitable for performing such a task is contingent on their ability to faithfully emulate the behaviour of more complex models and observed climate change.
The Reduced-Complexity Model Intercomparison Project (RCMIP) was established to assess this capability, as well as to better understand inter-RCM differences (Nicholls et al., 2020; Nicholls et al., 2021). Here, we introduce the protocol for the third and latest phase, RCMIP3. This phase focuses on two priorities. First, it provides a common set of observational benchmarks to be optionally used for ensemble constraining prior to submission, with the objective of mitigating discrepancies arising from different calibration methodologies and facilitating a clearer assessment of intrinsic model differences. Second, it requests an expanded set of variables and experiments from modelling teams to enable a more thorough evaluation of the carbon cycle representation in these models – a key gap in previous RCMIP phases. Additionally, RCMIP3 includes many of the experiments in the “Assessment Fast Track" (AFT) of the Coupled Model Intercomparison Project Phase 7 (CMIP7). As a result, RCMIP3 will improve our understanding of future model differences under these experiments, in addition to providing the community with valuable early projections.
The presentation will outline the RCMIP3 protocol and highlight the types of analyses it enables, along with preliminary results. By explicitly comparing RCM outputs with both ESM simulations and observations, RCMIP3 aims to strengthen the linkage across the climate-model hierarchy as well as evaluating and showcasing the suitability of RCMs for climate assessment.
Nicholls, Z., Meinshausen, M., Lewis, J., Corradi, M.R., Dorheim, K., Gasser, T., Gieseke, R., Hope, A.P., Leach, N.J., McBride, L.A., Quilcaille, Y., Rogelj, J., Salawitch, R.J., Samset, B.H., Sandstad, M., Shiklomanov, A., Skeie, R.B., Smith, C.J., Smith, S.J., Su, X., Tsutsui, J., Vega-Westhoff, B. and Woodard, D.L. 2021. Reduced Complexity Model Intercomparison Project Phase 2: Synthesizing Earth System Knowledge for Probabilistic Climate Projections. Earth’s Future. 9(6), https://doi.org/10.1029/2020EF001900.
Nicholls, Z.R.J., Meinshausen, M., Lewis, J., Gieseke, R., Dommenget, D., Dorheim, K., Fan, C.-S., Fuglestvedt, J.S., Gasser, T., Golüke, U., Goodwin, P., Hartin, C., Hope, A.P., Kriegler, E., Leach, N.J., Marchegiani, D., McBride, L.A., Quilcaille, Y., Rogelj, J., Salawitch, R.J., Samset, B.H., Sandstad, M., Shiklomanov, A.N., Skeie, R.B., Smith, C.J., Smith, S., Tanaka, K., Tsutsui, J. and Xie, Z. 2020. Reduced Complexity Model Intercomparison Project Phase 1: introduction and evaluation of global-mean temperature response. Geoscientific Model Development. 13(11), pp.5175–5190, https://doi.org/10.5194/gmd-13-5175-2020.
How to cite: Romero-Prieto, A., Sandstad, M., Sanderson, B. M., Nicholls, Z. R. J., Steinert, N. J., Gasser, T., Mathison, C., Kikstra, J., Aubry, T. J., Tanaka, K., Weber, K., and Smith, C.: Reduced Complexity Model Intercomparison Project Phase 3: protocol and preliminary results, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2310, https://doi.org/10.5194/egusphere-egu26-2310, 2026.
OSCAR is an open-source reduced-complexity Earth system model designed to probabilistically emulate the coupled climate–carbon–chemistry system with low computational cost. Following a preivously published evaluation of OSCAR v3.1 against observations and CMIP6 Earth system models, we present OSCAR v4, which incorporates a range of structural, numerical, and methodological improvements. Key developments include enhanced numerical stability, modularization of the code to allow running submodels independently, revised and streamlined modules, and recalibration using the latest AR6, CMIP6, and TRENDY datasets. Monte Carlo sampling has been improved using continuous probability distributions, and the constraining strategy now leverages Latin-hypercube sampling combined with probability integral transforms to provide more robust probabilistic ensembles compatible with observations. Alongside core model improvements, OSCAR v4 will introduce a suite of user-oriented functionalities and a full online documentation, facilitating broader adoption and reproducibility.
We illustrate the performance of OSCAR v4 through participation in the Reduced Complexity Model Intercomparison Project (RCMIP) phase 3 exercise. This benchmarking demonstrates the model’s ability to reproduce the spread of global temperature and carbon-cycle responses observed in more complex Earth system models, while providing rapid, policy-relevant probabilistic projections. Given it's level of complexity, OSCAR v4 is positioned as a versatile tool bridging comprehensive Earth system models and the simpler reduced-complexity approaches for large-scale climate assessments.
How to cite: Gasser, T., Zhu, B., Liu, X., Zhang, D., Lai, Y., and Shrivastav, G.: The compact Earth system model OSCAR v4, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18639, https://doi.org/10.5194/egusphere-egu26-18639, 2026.
Here we present an extended version of the forcing-driven and overshoot-aware spatial impacts emulator METEOR, which now includes functionality to emulate monthly outputs which include seasonality and natural variability, with the option to produce large distribution ensembles for a point, regional average or spatial domain. The philosophy of METEOR entails fast training on few and widely available datasets, sufficiently fast to be run on-the-fly and removing the need to archive large datasets and allowing interactive coupling with integrated assessment frameworks to simulate impacts directly. METEOR1.5 introduces a state dependent seasonal model and an autoregressive spatial, state-dependent noise model which can produce distributions of realisations conforming to the climatic trends and distributional properties of the emulated model Integrated impact modules allow the direct emulation of human and ecological stressors which are computed from easily retrained emulated climates to answer regional questions.
How to cite: Sandstad, M., Sanderson, B., Steinert, N., and Mittal, S.: METEOR 1.5 a spatial emulator for fast and relevant responses to impact questions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4264, https://doi.org/10.5194/egusphere-egu26-4264, 2026.
Integrating physical, socio-economic, and technological perspectives is indispensable for addressing climate mitigation challenges. While directly coupling state-of-the-art Earth System Models (ESMs) and Integrated Assessment Models (IAMs) offers a way to explore feedbacks between these domains, doing so with full-complexity models remains computationally prohibitive. This is particularly true for cost-effective intertemporal optimization IAMs due to fundamental operational differences: while ESMs perform forward simulations, such IAMs optimize over time. Consequently, direct coupling would require numerous computationally intensive iterations to converge, a complication further compounded by the stochastic nature of ESMs.
To overcome the barriers to coupling ESMs and IAMs, we employ their reduced-complexity representations (i.e., emulators). We couple an IAM emulator representing 9 distinct IAMs (Xiong et al. 2025) with an ESM emulator, FaIR, representing 66 ESM configurations (Smith et al. 2024a). Using this coupled ESM-IAM emulator framework in an optimization setting, we calculate cost-effective pathways that achieve the temperature targets of the Paris Agreement with and without overshoot.
Our preliminary results indicate that the uncertainty ranges for such pathways are significantly larger than previously estimated. Our results also have implications for target setting; we show how pathways differ when IAMs optimize directly for a temperature target – a capability IAMs traditionally lack. Instead, IAMs typically rely on temperature proxies, such as carbon budgets (or their corresponding carbon price pathways), which do not necessarily provide an accurate representation of the temperature target. Furthermore, this study offers advanced insights into the dynamics of climate-economy interactions, providing a roadmap for future efforts to couple full-complexity models.
References
Xiong, W., Tanaka, K., Ciais, P., Johansson, D. J. A., & Lehtveer, M. (2025). emIAM v1.0: an emulator for integrated assessment models using marginal abatement cost curves. Geosci. Model Dev., 18(5), 1575-1612. doi:10.5194/gmd-18-1575-2025
Smith, C., Cummins, D. P., Fredriksen, H. B., Nicholls, Z., Meinshausen, M., Allen, M., . . . Partanen, A. I. (2024). fair-calibrate v1.4.1: calibration, constraining, and validation of the FaIR simple climate model for reliable future climate projections. Geosci. Model Dev., 17(23), 8569-8592. doi:10.5194/gmd-17-8569-2024
How to cite: Tanaka, K., Weiwei, X., Allen, M., Cain, M., Jenkins, S., Mathison, C., Patel, V., Smith, C., and Tachiiri, K.: Coupled ESM-IAM Emulator: Exploring Uncertainties in Temperature Target Pathways, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14741, https://doi.org/10.5194/egusphere-egu26-14741, 2026.
To meet Paris-aligned climate goals and minimize temperature overshoot and its impacts, rapid and deep reductions in greenhouse-gas emissions from fossil-fuel combustion are required. Climate risk projections are strongly affected by uncertainty in anthropogenic aerosol effective radiative forcing (ERF) and by the co-evolution of air-pollutant emissions under decarbonization pathways. Because running large Earth System Model (ESM) ensembles remains computationally expensive for uncertainty quantification and broad policy-scenario exploration, reduced-complexity climate emulators are needed for efficient, transparent, and observation-connected assessments.
Here we develop an aerosol extension to the simple climate model (SCM) FaIR that emulates aerosol ERF from global anomalies in aerosol optical depth (ΔAOD) relative to a pre-industrial baseline for different species. Aerosol ERF is computed using a constrained parameterization that separates aerosol–radiation and aerosol–cloud interactions, with key parameters represented probabilistically and constrained by observational and model-based lines of evidence.
To emulate ΔAOD from emissions pathways, we implement an interpretable mapping calibrated to CMIP6 ESM output. An effective linear relationship between emission and burden anomalies is fitted using a single parameter that aggregates yield and lifetime effects. In a second step, we fit an effective optical parameter linking burden perturbations to ΔAOD. This produces model-dependent parameter distributions that enable propagation of both parametric uncertainty and between-model spread. In addition, we implement an integrated-assessment-model-based relationship linking air-pollutant emissions to CO₂ emissions under different air-quality policy stringencies, interpolated into a continuous air-quality parameter suitable for exploring uncertainty and its interaction with decarbonization trajectories.
We perform Monte Carlo ensembles sampling aerosol-ERF parameters, CMIP6-calibrated aerosol–AOD mappings, air-quality policy stringency, and net-zero timing, and evaluate impact-relevant climate risk metrics including peak warming, probability of remaining below 1.5 °C, threshold crossing year, overshoot duration, and warming rates computed over multiple near-term and decadal windows. Preliminary results show strong dependence of peak temperature outcomes on net-zero timing, while threshold-based metrics and warming rates exhibit pronounced sensitivity to air-quality assumptions, consistent with a partial loss of aerosol cooling under stricter pollution controls. Overall, the results indicate non-linear interactions between decarbonization timing, air-quality stringency, and warming-rate responses. The emulator provides a scalable basis for robust climate risk screening and for coupling SCM trajectories to impact assessments.
Keywords: Climate Change, Mitigation, Aerosols, Effective Radiative Forcing, Climate Emulators, Climate Modeling, CMIP6 Calibration, Air-quality Policy, Overshoot
How to cite: Arzola Röber, T., Bruckner, T., and Quaas, J.: Accounting for Aerosols in Climate Mitigation Pathways, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19612, https://doi.org/10.5194/egusphere-egu26-19612, 2026.
Parsimonious emulator models (PEMs) trained on Earth system models (ESMs) can be very useful when information about global quantities like global mean surface temperature (GMST) and ocean heat content (OHC) are sought. Here, I use data over several millennia from ESM runs extracted from the LongRunMip repository to construct and test PEMs for GMST and net incoming radiation flux.
For the GMST, I consider a linear impulse response in the form of a superposition of three decaying exponentials, comprising three weight coefficients and three characteristic decay times to be estimated by least square fitting to ESM runs with abrupt step function forcing. The model fit is good on all time scales, and the fitted model seems to perform even better for smoother forcing scenarios. This sugggests that the six model parameters represent essential features of each ESM.
Data for radiation flux, and its decomposition in longwave and reflected shortwave, are combined with GMST to produce Gregory plots. By fitting parabolic curves to these plots, I obtain a simple analytic expression for the evolution of the feedback parameter λt), the radiation fluxes, and the resulting increase in OHC.
From these PEMs we can easily compare the global performance of different ESMs under different forcing scenarios. For instance, a comparison of the GISS-E2-R and CESM104 models exhibit equilibrium climate sensitivities (ECSs) of 3.4 and 2.4 K, respectively. The main reason for the difference is very different albedo feedbacks in the two models. Resulting total feedback parameter λ(t) drops from 2.1 to 1.0 Wm-2 K-1 in GISS-E2-R and from 1.4 to 0.6 Wm-2 K-1 in CESM104. The OHC grows at nearly the same rate in the two models during the first millenium, but GISS saturates earlier and at lower final OHC.
How to cite: Rypdal, K.: Parsimonious models emulating millennium-long Earth system model simulations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12585, https://doi.org/10.5194/egusphere-egu26-12585, 2026.
Uncertainties in conventional (GCM) climate models, defined as the structural spread among com-
peting models, have increased for the first time in the latest AR6 report despite an exponential increase
in the modern computation power. The root problem is that these models are based in the weather
regime, that is, they spend unnecessary effort in calculating irrelevant weather details. This project
aims to produce precise regional projection using the Half Order Energy Balance Equation (HEBE): a
half order fractional derivative generalization of the standard Energy Balance Equation (EBE). HEBE
has the advantage of being a direct consequence of the continuum heat equation combined with energy-
conserving surface boundary conditions. A previous paper used Fractional EBE (FEBE) to model Earth
climate projections through 2100 on a global scale, and it yields significantly smaller uncertainty com-
pared to the CMIP6 MME. This project builds on a similar methodology, enhancing climate projection
with additional regional details and upgraded precision. The current results show that the parametric
uncertainty in HEBE’s temperature response is smaller than the internal variability at most locations,
at the exceptions of the high memory deep ocean regions near Pacific. HEBE’s regional hindcast ac-
curately reproduces ERA5 2mT series’ deterministic and stochastic patterns of regional temperature.
The global hindcast is also validated by various reanalysis datasets and instrumental records. The
direct year to year relative uncertainty (ratio between 90% confidence interval and best estimate) is
stable across time and marker scenarios, with most regions projecting values below 0.5 by 2100. On a
global scale, the parametric uncertainty in HEBE’s response temperature is negligible (±0.03K by 2100
using the SSP2-4.5 marker scenario). This effectively shows that HEBE’s projection is more precise
than its competitors even without taking period averages. The exceedingly low global uncertainty was
constrained by the large amount of regional information when taking the global averages. It should be
noted that the cited parametric uncertainty does not take into account systematic biases in HEBE and
in the input datasets. The most important source should be any errors in the forcings, especially con-
cerning aerosols. HEBE aims to provide a compelling and physically grounded alternative to complex
deterministic multi-model ensembles, offering a more precise, efficient, and interpretable means of pro-
jecting regional climate changes in the coming century. This positions it as a potentially valuable tool
for policy-relevant projections and adaptation planning, thereby showing the pertinency of fractional
derivative and Bayesian framework in atmospheric sciences.
How to cite: Wang, Y., Lovejoy, S., Lebiadowski, D., and Clarke, D.: Low Uncertainty Regional Climate Projections without Irrelevant Weather Details, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15472, https://doi.org/10.5194/egusphere-egu26-15472, 2026.
Traditional earth system model ensembles provide valuable information on climate extremes. However, their limited size often underrepresents rare high-impact events, restricting the ability to explore extreme outcomes and large-scale anomaly patterns. Using the climate emulator MESMER, trained on CMIP6 models, together with the risk assessment platform CLIMADA, we assess population exposure to annual maximum daily temperatures and asset exposure to annual maximum daily precipitation.
MESMER generates virtually unlimited, spatially explicit, global climate realizations for any scenario defined by emission or global-mean-temperature trajectories. This allows us to characterize the spread of potential outcomes and associated spatial patterns, identify rare high-impact realizations, compare results with standard CMIP6 ensembles, or explore custom scenarios beyond existing model experiments.
We illustrate spatial and temporal patterns of exposure for temperature and precipitation extremes, highlighting contrasting regional responses and how highly impactful outcomes can emerge from climate variability.
How to cite: Pierini, L., Kropf, C., Gudmundsson, L., Seneviratne, S. I., and Bresch, D. N.: Revealing Probabilistic Patterns of Climate Extremes and Impacts Through Emulator-Based Risk Analysis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12408, https://doi.org/10.5194/egusphere-egu26-12408, 2026.
Estimates of economic damages from climate change in Europe depend on temperature projections, and they are thereby subject to scenario uncertainty and model uncertainty, as well as damage function uncertainty. An additional, often implicit source of uncertainty is the projected, yet poorly constrained, weakening of the Atlantic Meridional Overturning Circulation (AMOC), which would lower European temperatures. Here, I explicitly quantify the contribution of AMOC uncertainty to total damage uncertainty, with AMOC uncertainty comprising uncertainty about future AMOC developments as well as uncertainty about the cooling pattern that would follow an AMOC weakening. I combine a newly developed pattern-scaling-type emulator of the European cooling response to AMOC weakening — calibrated for different Earth system models (ESMs) — with temperature projections from multiple ESMs and emissions scenarios, alongside several damage functions. This allows me to decompose the total uncertainty in European economic damages into different drivers and estimate the share attributable to the behaviour of the AMOC.
How to cite: Schaumann, F.: Quantifying AMOC Uncertainty in European Climate Damage Projections, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8370, https://doi.org/10.5194/egusphere-egu26-8370, 2026.
Reservoirs are increasingly recognized as significant sources of greenhouse gas (GHG) emissions, yet their future emissions under climate change remain poorly quantified. This study evaluates the impact of climate change on net GHG emissions from Feitsui Reservoir, a major water supply reservoir in northern Taiwan, using an integrated modeling approach.
We utilized the multisite Weather Generator (multiWG) to generate future climate projections for three Shared Socioeconomic Pathways (SSP126, SSP245, SSP585) across four 20-year periods (2021-2040, 2041-2060, 2061-2080, 2081-2100), with 1995-2014 as the baseline. A Random Forest model (NSE = 0.8637) was trained to predict reservoir inflow based on temperature and precipitation data. These inflows were input into the G-RES model to calculate net GHG emissions in CO₂-equivalent units, including contributions from both CO₂ and CH₄.
Results reveal that reservoir GHG emissions will increase under all climate scenarios, with magnitude strongly dependent on emission pathways. Under the low-emission scenario (SSP126), emissions increase by 5.2-8.8% across all periods. The intermediate scenario (SSP245) shows moderate increases of 5.4-18.4%. The high-emission scenario (SSP585) demonstrates dramatic escalation, particularly in the late century (2081-2100), where emissions reach 1259.6 gCO₂e/m²/yr—a 45.8% increase. These findings underscore the critical need to consider climate impacts in reservoir management and carbon accounting frameworks.
How to cite: Yeh, F.-W. and Tung, C.-P.: Assessing Climate-Driven Greenhouse Gas Emissions from Feitsui Reservoir Using G-RES Under Multiple SSP Scenarios, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16498, https://doi.org/10.5194/egusphere-egu26-16498, 2026.
Greenhouse gas emission metrics are widely used for comparing climate impacts of different gases and for guiding mitigation policy. Conventional metrics such as GWP100 perform well for representing the warming effects of long-lived gases which behave like CO₂ but poorly for short-lived climate pollutants (SLCPs). Methane (CH4) is the most important SLCP and has been the main focus of alternative metrics. GWP* was developed to more accurately capture impact on global warming, particularly from stable and declining CH4 emissions which are not well served by GWP100. This means that GWP* better connects emissions pathways to long-term temperature targets (Cain et al., 2022). Previous studies optimised GWP* for CH4 for a limited range of scenarios up to 2100. However, future mitigation pathways involve a wider range of gases and transition speeds, overshoot behaviour, and long-term stabilization beyond this period. In addition, highly radiatively efficient fluorinated gases are increasingly important in mitigation strategies yet have not been demonstrated with the GWP* framework. In this study, we systematically test the performance of GWP* across an expanded set of emissions scenarios, including rapid mitigation, delayed action, and prolonged temperature overshoot pathways, and extend the analysis to multi-century time horizons with an optimisation of the flow term of GWP* (Mastropierro et al., 2025). We further develop and evaluate a generalized formulation of GWP* for fluorinated gases with diverse atmospheric lifetimes. The outcomes examine the performance of GWP* under realistic transition pathways and its representation of temperature responses for fluorinated gases. This work supports the development of more physically consistent multi-gas emission metrics for climate targets, carbon budgeting, and policy design, as it is a simple tool to calculate how much global warming is added or avoided by increasing or cutting SLCPs such as F-gases.
Cain, M., Jenkins, S., Allen, M.R., Lynch, J., Frame, D.J., Macey, A.H., Peters, G.P. Methane and the Paris Agreement temperature goals. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 380 (2022). https://doi.org/10.1098/rsta.2020.0456
Mastropierro, M., Tanaka, K., Melnikova, I. et al. Testing GWP* to quantify non-CO2contributions in the carbon budget framework in overshoot scenarios. npj Clim Atmos Sci 8, 101 (2025). https://doi.org/10.1038/s41612-025-00980-7
How to cite: Cain, M., Patel, V., Mastropierro, M., Tanaka, K., Jenkins, S., and Allen, M.: Applying GWP* to Long-Term Climate Pathways and Fluorinated Gases, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20117, https://doi.org/10.5194/egusphere-egu26-20117, 2026.
Posters virtual: Thu, 7 May, 14:00–18:00 | vPoster spot 1b
EGU26-18820 | ECS | Posters virtual | VPS23
Global climate dynamics in a highly parameterized radiative-convective-macroturbulent energy balance modelThu, 07 May, 14:00–14:03 (CEST) vPoster spot 1b
We present a one-layer global energy balance climate model with highly parameterized radiation, convection, and large-scale atmosphere/ocean macroturbulence. Planetary heat content is parameterized by a 2D in latitude-longitude layer characterized by a temperature field and a uniform constant heat capacity. Radiation is parameterized by mean-annual zonal average top-of-atmosphere solar irradiance. Radiative heating and cooling are parameterized by a uniform constant albedo and Stefan-Boltzmann emission with uniform constant emissivity. Convection is parameterized by a temperature threshold for convection which restricts the layer from warming beyond the threshold, effectively cooling the layer. Macroturbulence is parameterized by 2D barotropic turbulence forced at small scales and damped by Rayleigh friction. Energy conservation is maintained by balancing the convective cooling of the layer with the turbulent kinetic energy forcing, resulting in tropical forcing, while the frictional loss of kinetic energy is balanced by frictional heating of the layer. The parameterized energy transforming processes are characterized by timescales, which, for Earth-like planets, are ordered as tradiation > tmacroturbulence > tconvection.
We investigate the model’s equilibrium climate state in terms of the meridional heat transport (MHT), the resulting zonally averaged temperature profile, and their fluctuations by simulating the system over many radiation times. For Earth-like parameters, despite the model’s extremely simplified dynamics, our simulations reveal a MHT profile comparable to the observed, annually averaged MHT on Earth, featuring a maximum in the mid-latitudes of approximately 5PW, a form of Bjerknes compensation.
How to cite: van Kan, A., Weiss, J., and Knobloch, E.: Global climate dynamics in a highly parameterized radiative-convective-macroturbulent energy balance model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18820, https://doi.org/10.5194/egusphere-egu26-18820, 2026.
