Tuesday 23th September 2025
Strengths and Limitations of Existing Earth System Models
Introduction
09:00–09:30 – Catalyst Talk I: Bjorn Stevens (MPI-M, Germany)
Opportunities for elucidating tropical climate dynamics with a new generation of km-scale climate models
Traditional climate models depend on statistical representations of processes that act on regional and smaller scales and which dominate the climate of tropical regions. These approximations are designed to limit extremes and necessarily neglect land-scape scales. While such simplifications do not seem to strongly impact the ability to inform us about global scale changes, it makes such models inadequate for the study of regional and coastal processes and extremes, as well as the tropics more generally. Thanks to advances in computing, km-scale global climate models, which are designed to explicitly represent processes on the regional scale, and explicitly simulate extremes, are coming into scientific use. In this talk I will present the status of this new modelling paradigm and the opportunities it offers for advancing understanding of the coastal systems, regional processes, extremes, and the tropics more generally.
09:30–10:00 – Catalyst Talk II: Veronika Eyring (DLR & University of Bremen, Germany)
AI-empowered Next-generation Multiscale Climate Modeling for Mitigation and Adaptation
Earth System Models (ESMs) are fundamental to understanding and projecting climate change. They have continued to improve, but systematic errors and large uncertainties in their projections remain. A large contribution to this uncertainty stems from the representation of processes such as clouds and convection that occur at scales smaller than the resolved model grid. This impacts the models’ ability to accurately project global and regional climate change, climate variability, and extremes. High-resolution, cloud resolving models with horizontal grid resolution of a few kilometers alleviate many biases of coarse-resolution models, but their high computational costs limit their applicability to run multiple decades and large ensembles. Yet short simulations from high-resolution models can serve as information to develop machine learning (ML)-based parametrisations that are then incorporated into hybrid (physics+ML) ESMs that promise to have significantly reduced systematic errors and enhanced projection capability compared to current ESMs. We argue that an AI-empowered multiscale approach is needed to improve climate information in the tropics and more generally on the global and regional scale, drawing on km-scale climate models and hybrid ESMs that include essential Earth system processes and feedbacks yet are still fast enough to deliver large ensembles for better quantification of internal variability and extremes. The talk presents progress in hybrid modelling work as well as key challenges and visions how to enhance climate modeling with ML (Eyring et al., 2024a,b).
References:
Eyring, V., P. Gentine, G. Camps-Valls, D. M. Lawrence, and M. Reichstein, AI-empowered next-generation multiscale climate modelling for mitigation and adaptation, Nat. Geosci., https://doi.org/10.1038/s41561-024-01527-w, 2024a.
Eyring, V., Collins, W.D., Gentine, P., Barnes, E.A., Barreiro, M., Beucler, T., Bocquet, M., Bretherton, C.S., Christensen, H.M., Gagne, D.J., Hall, D., Hammerling, D., Hoyer, S., Iglesias-Suarez, F., Lopez-Gomez, I., McGraw, M.C., Meehl, G.A., Molina, M.J., Monteleoni, C., Mueller, J., Pritchard, M.S., Rolnick, D., Runge, J., Stier, P., Watt-Meyer, O., Weigel, K., Yu, R., Zanna, L., Pushing the frontiers in climate modelling and analysis with machine learning, Nat. Clim. Chang., https://doi.org/10.1038/s41558-024-02095-y, 2024b.
Session 1: Global Coupled Earth System Models
This session analyses the current generation of global Earth System Models (ESMs), with emphasis on their capacity to represent land–ocean–atmosphere interactions and their potential for incorporating tropical coastal processes. The session explores the architectures and coupling strategies of ICON-ESM, ACCESS, Tai-ESM, POEM and IITM-ESM, examining the extent to which they support integration of physical, biogeochemical, and ecological components. It highlights how such models handle interactions across spatial and temporal scales, from large-scale circulation to regional and nearshore dynamics, and addresses the treatment of socio-ecological feedbacks and coastal boundary complexities. While some ESMs now enable limited nesting or coupling to higher-resolution modules, considerable challenges remain in resolving coastal–terrestrial exchange, biophysical feedbacks, and fine-scale bathymetric features. Guiding question: How can global coupled models capture land–atmosphere–ocean interactions, and what near-term steps would most improve their relevance for coastal dynamics without compromising large-scale skill?
Chair: Bjorn Stevens (MPI-M, Germany)
10:00–10:15 – Jan Härter (University of Potsdam, Germany)
A vision for tropical coastal modeling and observations
Tropical coasts are unique with an astounding ecological diversity at the interface between the oceanic, terrestrial and atmospheric Earth system components. The tropical coasts provide essential ecosystem services to 40% of the world’s population, constitute bottlenecks for international trade and experience heavy impacts of global environmental change. Yet, key processes at this interface of disciplines remain poorly understood. Such lack in understanding may be the result of an insufficient push to holistically modelling the relevant processes – in combination with the incorporation of local expertise and observational campaigns. We present a vision of how to model tropical coastal regions to tackle the combined effects of processes at these coasts using a numerical modelling approach combined with targeted observational campaigns.
10:15–10:30 – Nils Brüggemann (MPI-M, Germany), Arjun Kumar, David Nielsen, Mikael Karvinen, Fatemeh Chegini, Nuno Serra, Cathy Hohenegger, Tatian Ilyina, and Jochem Marotzke
Air-Sea Interactions of Tropical Cyclones in Kilometer-Scale Earth System Models
Tropical cyclones (TCs), with their enormous destructive potential, pose extreme risks to societies in coastal regions. Despite their severe impacts, many aspects of TC dynamics remain poorly understood. Recent advances in computational modeling now make it possible to gain new insights into TC behavior. In this study, I present results from global, non-hydrostatic, storm-resolving, and fully coupled simulations that enable a deeper understanding of the air–sea interactions and feedbacks shaping TC dynamics. In particular, I examine how small-scale and mesoscale oceanic turbulence influences TC intensity, how ocean biogeochemistry responds to TCs, and how the internal variability of TC frequency can be assessed. Finally, I discuss potential avenues for improving the representation of TCs in coupled high-resolution Earth System Models.
☕ 10:30–11:00 – Morning Coffee Break
11:00–11:15 – Georg Feulner (PIK, Germany)
Paleoclimate Modelling Perspectives on Coastal Systems
After a brief introduction into the philosophy and design of the Potsdam Earth Model (POEM) developed at the Potsdam Institute for Climate Impact Research (PIK), I will highlight examples from Earth’s history for which coastal dynamics and coastal ecosystems play an important role – and for which improved modelling capabilities linking global and coastal processes would be crucial.
11:15–11:30 – Praveen Veluthedathekuzhiyil (IITM, India), Swapna P., Sandeep N., Ayantika DC., Priya P., Umakanth U.
IITM-ESM v3: Advances in Tropical Climate Simulation and Bias Reduction
The IITM Earth System Model (IITM-ESM) is India’s primary coupled climate model participating in the Coupled Model Intercomparison Project (CMIP). Recent developments enhance the representation of tropical variability and global teleconnections, critical for South Asian monsoon prediction and climate impact studies. Key upgrades include: (i) a Triangular-Cubic-Octahedral (TCO) atmospheric grid for higher tropical resolution, (ii) refined ocean resolution over the tropics, (iii) improved land surface representation with 24 vegetation types using Land Use Land Cover data from the National Remote Sensing Centre (NRSC), India, and (iv) a basin-based river discharge scheme. These improvements yield more realistic simulations of monsoon intra-seasonal oscillations (MISOs), Atlantic Meridional Overturning Circulation (AMOC), and low-pressure system (LPS) propagation. Control simulations show better agreement with observations, notably reducing long-standing biases in rainfall distribution, temperature, winds, and ocean stratification. The upgraded IITM-ESM (v3) will contribute to CMIP7, providing a vital tropical perspective to the global modeling community.
11:30–11:45 – Yolandi Ernst (University of the Witwatersrand, South Africa) Francois Engelbrecht, Jessica Steinkopf, Gregor Feig, Nolusindiso Ndara, Amukelani Maluleki
Tailoring an Earth system model to the African context
Africa contributes significantly to uncertainties in the global greenhouse gas budget. Savanna ecosystems, covering about 40-50 % of Africa, amplify this uncertainty due to the large seasonal and interannual variability over short time periods and unique drivers of change, such as fire, bush encroachment, and land use. Additionally, Africa is observationally poor, making it challenging to parametrise and validate models. To improve land-surface process characterisations in Africa, we coupled the conformal-cubic atmospheric model (CCAM) with a dynamic land-surface model, CABLE. With its flexible downscaling capabilities, this modelling system adopts a spectral nudging methodology and can be integrated as a quasi-uniform global model or a stretched-grid, variable-resolution regional model to obtain high resolution over a region of interest. We nudged the model using ERA5 reanalysis data, ensuring simulations were consistent with historical weather patterns as captured in the reanalysis outputs. This enabled the evaluation of several land-atmosphere fluxes and attributes of the simulated model outputs against Eddy covariance flux tower observations, as well as satellite data. In general, the modelling system does remarkably well at capturing the key attributes of the seasonal and diurnal cycles of certain land-atmosphere fluxes. Good agreement between satellite-derived and modelled soil moisture further supports the model’s capability. The skill of the model does vary by site, highlighting the importance of further validating the model across different biomes and climatic zones. Current efforts are directed to increasingly targeted in situ observations and variables, together with earth observation data, for improved calibration and model improvement to understand the underlying processes that drive carbon flux variations across the African region. Ultimately, the improvement of the land-atmosphere model will provide valuable input for addressing the biochemical component in an ocean model that is being developed in parallel.
11:45–12:00 – Knut Klingbeil (IOW, Germany)
Requirements for the ocean model component in a coastal ESM
The coastal ocean, ranging from the shelf break to estuaries, has a number of characteristics which discriminate it from the global ocean. Therefore, simply increasing the resolution in a global ocean model will not provide a proper coastal ocean model. Coastal ocean models need to reproduce physical processes on different scales and require specific numerical techniques. This talk highlights physical, numerical and technical aspects to be considered when coastal and estuarine dynamics should be represented in Earth System Models.
12:00–12:15 – Tilo Ziehn (CSIRO, Australia)
The Australian Earth System Model: A Southern Hemisphere perspective
The Australian Community Climate and Earth System Simulator (ACCESS) is a complex modelling system designed to support both operational weather forecasting and cutting-edge climate research. In this presentation, we introduce the Earth System Model variant, ACCESS-ESM, and highlight its recent applications in assessing future carbon budgets, exploring climate stabilisation pathways, and investigating scenarios involving temperature overshoot. Particular emphasis is placed on the critical role of the Southern Ocean in those simulations and on the broader implications of future climate change for the Southern Hemisphere.
🍽 12:15–13:30 – Lunch Break (lunch provided on site)
Session 2: Regional Models
This session examines the development and application of regional and nested ocean and biogeochemical models as a means to improve the representation of coastal and nearshore processes. It emphasises downscaling strategies, regional nesting of global models, and hybrid approaches that combine dynamical and statistical techniques to overcome resolution and boundary-condition constraints. The session addresses scientific and technical challenges, including data limitations, computational efficiency, and the need for realistic forcings and boundary interactions in highly dynamic coastal regions. The role of new modelling platforms (e.g. MPAS-Ocean, ICON-Coast), more established models (e.g. WRF-NEMO), and regional reanalyses is reviewed, along with emerging strategies to represent fine-scale physical and ecological features. Guiding question: How can regional models be more effectively nested within ESMs, and what benchmarks are most useful for representing tropical coastal and nearshore dynamics?
Chair: Alexandra Klemme (ZMT)
13:30–14:00 – Keynote: Yu-Heng Tseng (NTU, Taiwan)
Challenge of multi-scale Ocean-Atmosphere Coupled Modelling System
This talk addresses the challenges of building a multi-scale ocean–atmosphere coupled modeling system that connects coastal ocean models with the global Earth System Model (ESM). I will begin with early efforts in coupling coastal dynamics to basin- and global-scale circulation, then highlight applications that reveal both limitations and possible solutions. Major challenges include coupling strategies, non-hydrostatic effects of complex topography, and unresolved processes such as internal tides, internal waves, mesoscale and submesoscale eddies, and land–ocean exchanges through estuaries. A particularly critical issue is representing estuarine processes that cannot be explicitly resolved within land-ocean interaction. To address this, simplified parameterizations, such as estuary box models, are introduced within the ESM framework. These examples illustrate the broader challenge of bridging fine-scale coastal processes with large-scale climate dynamics, a key step toward improving predictions of the coupled ocean–atmosphere system.
14:00–14:15 – Bodo Ahrens (Goethe University Frankfurt, Germany)
Coupled Regional Climate Modelling Systems in MED-CORDEX: Atmospheric Perspective
The WCRP Mediterranean Coordinated Regional Downscaling Experiment (Med-CORDEX) has significantly advanced the implementation and application of fully coupled regional climate modelling systems, which integrate a regionally closed water cycle. This study highlights key atmospheric impacts stemming from model coupling, particularly in the simulation of extreme weather events such as medicanes, the dynamics of near-coastal ocean-atmosphere interactions, and the influence of grid resolution and model ensemble configurations. These examples underscore the added value provided by coupled systems in enhancing the fidelity of regional climate projections. Additionally, the presentation includes an outlook on forthcoming developments within MED-CORDEX and reflects on the broader implications for other regions and coordinated modelling initiatives.
14:15–14:30 – Dante Espinoza (IMARPE, Peru)
Physical-biogeochemical modeling in the Northern Humboldt Current Ecosystem: Present trends and future scenarios
The Northern Humboldt Current Ecosystem (NHCE) is among the most productive coastal upwelling systems worldwide, sustained by equatorward winds that bring cold, nutrient-rich waters to the surface. Its functioning is strongly modulated by El Niño and the climate change. To investigate present coastal trends and potential future responses, we employed the regional coupled physical–biogeochemical model CROCO–PISCES. Model performance depends critically on the quality of wind products and the specification of open boundary conditions. The model reproduces main NHCE features, including coastal cooling, enhanced surface productivity, and progressive deoxygenation, with remote forcing identified as the dominant driver. Under high-emission climate scenarios toward the end of the twenty-first century, most dynamically downscaled Earth System Models project surface warming, increased surface productivity, and reduced subsurface productivity, while projections for dissolved oxygen remain uncertain. Recent advances in biogeochemical modeling of coastal bays are also discussed.
14:30–14:45 – Mia Sophie Specht (MPI-M, Germany)
Resolving Tropical Instability Waves: Seasonal Dynamics of Shear Instabilities and Coupled Ocean-Atmosphere Responses in high-resolution simulations
Tropical Instability Waves (TIWs) are a major source of intraseasonal variability in the tropical oceans. Characterized by sharp sea surface temperature fronts, they strongly influence the exchange between the ocean surface, its interior, as well as the overlying atmosphere. Using high-resolution ICON-Ocean simulations, we find that TIW-driven shear instabilities at the wave fronts extend below the mixed layer into the thermocline and follow a pronounced seasonal cycle. They peak in boreal summer, when TIW-induced shear combines with background shear from the northern South Equatorial Current (nSEC). High-resolution coupled IFS/FESOM simulations further demonstrate that the influence of TIWs is not confined to the ocean: they alter the atmospheric boundary layer, modify its height, and affect circulation above. These results highlight that a realistic representation of TIWs in climate models is crucial, not only to resolve oceanic variability, but also to capture their broader role in shaping regional ocean-atmosphere dynamics.
14:45 – 15:00 – Patrice Brehmer (IRD / SRFC, Senegal) Elodie Martinez, Thomas Gorgues, Hervé Demarcq, Shunya Koseki, Etienne Pauthenet, Keerthi Madhavan-Girijakumari, Ndague Diogoul, Aldo Affenou, Yoba Kande, Saliou Faye, Ndague Diogoul, Abdoulaye Sarré, Noel Keenlyside.
Coupling high-resolution Earth system model outputs and data-driven methods to project ecosystem change in the Canary Current upwelling system
Projecting ecosystem change in tropical coastal upwelling systems requires approaches capable of capturing fine-scale physical–biological interactions. In this study, we integrate high-resolution simulations from the nextGEMS project with Earth observation products and empirical models to assess past and projected changes in marine productivity and fish distributions in the southern Canary Current Large Marine Ecosystem (sCCLME). We employ a deep learning approach (UNet) trained on MODIS and OC-CCI chlorophyll-a data, using physical predictors (SST, sea surface height, eddy kinetic energy, wind stress) from nextGEMS simulations to reconstruct spatially and temporally continuous Sea Surface Chlorophyll-a (SSC) fields. In parallel, Generalised Additive Models (GAMs) based on long-term sea survey data (1995–2015) are used to investigate how environmental drivers shape the distribution and biomass of micronekton and small pelagic fish schools. Results show contrasting trends in micronekton biomass and fish school density, suggesting spatial reorganisations in trophic structure under changing ocean conditions. The combination of storm-resolving model outputs with data-driven ecological reconstruction demonstrates the potential of hybrid approaches to improve representation and forecasting of tropical coastal ecosystems, especially in data-limited regions. This work contributes to broader efforts to bridge the gap between global Earth system simulations and regional ecosystem applications.