🇪🇺 EC-EARTH: a European community climate model coupling ECMWF’s IFS atmosphere, NEMO ocean & LPJ-GUESS vegetation. We’re upgrading LPJ-GUESS land processes and vegetation dynamics to better capture ecosystem–climate feedbacks. #ECEarth #ESM #ClimateModeling #ClimateScience #LPJGUESS

🌍🌊 CNRM-ESM: coupling climate physics & ocean biogeochemistry. Built by Météo-France/CNRM, it couples atmospheric dynamics (CNRM-CM) to NEMO–PISCESv2 ocean module capturing nutrient-plankton-carbon cycles. Our updates: boost atm. physics, ocean dynamics & carbon–nutrient cycles. #ClimateModeling

🇬🇧🌊 UKESM: The UK’s flagship #ESM linking atmosphere, ocean, land & ice with interactive carbon & chemistry cycles. What we do: upgrade ocean processes in HadGEM & NEMO, enhance JULES land surface, & improve climate–carbon feedbacks. #UKESM #ClimateModeling #ClimateScience

🌍🌐 ICON-ESM: Climate modeling on a whole new grid The first fully coupled #ESM on an icosahedral grid — no pole problem, uniform global resolution & ready for ML-powered #ClimateScience. What we do: add ML parameterisations & improve JSBACH land processes for more realistic vegetation dynamics.

🌍 Meet the #EarthSystemModels powering our project! Over the next weeks we’ll share 5 unique #ESMs — what makes them special, the science they enable, & how we’ll advance them. 🚀 ICON-ESM, UKESM, EC-EARTH, CNRM-ESM, IPSL-CM. #ClimateModeling #AI4Climate

💡 New paper! Climate extremes are impacting forests across Europe. Drought–heat compound events diminish resilience and prolong recovery times, with temperate and boreal forests lacking adaptations leading to physiological damage that increasing sensitivity. 👉 Find out more: buff.ly/w6ChLXa

💡 New paper! Droughts are complex hydrometeorological phenomena driven by large-scale land–atmosphere feedbacks. Using a new parametric calibration algorithm improves Deep Learning framework reliability in drought detection as showcased on European drought events. 👉 Find out more: buff.ly/llm8RRK

💡 New paper! Find out how recent significant improvements of #ESMValTool allows a more effective evaluation of complex #ESMs and high-resolution models while needing less computational resources. 👉 Join the discussion: buff.ly/CyycyeP

🚀 First GA for #AI4PEX in Lund! Last month partners from across Europe met to advance #EarthSystemModels with #AI, #ML & #EarthObservation - tackling ✨ Benchmarking 🛠 Syncing milestones 🌍 Boosting outreach 🤝 Hands-on collaboration. Big thanks to Lund University for hosting! 👉 www.ai4pex.org

🚀 Excited to share our team’s contributions at the ESA Living Planet Symposium next week in Vienna! Catch us across multiple sessions on carbon fluxes, EO data consistency, AI for climate science & more. #LPS25 🌍🛰️ #EarthObservation #EarthSystemModeling #CINEA

Fourth and last day of our #AI4PEX GA is again reserved for workshops. Moderated by Fangfei Lan from UNIL we spend the morning focusing on Neural Operator Learning. In the afternoon, we build our own hybrid model using the SINDBAD modeling framework, led by Nuno Carvalhais from MPG-BGC. #CINEA

AI4PEX GA Day 3 we are mainly focusing on the technical and methodological aspects: How to improve our benchmarking? And what can explainable AI do to improve our parametrizations? Highlight was the keynote by Dave Lawrence from UCAR on how #ArtificialIntelligence is entering the #ESMs development.

#AI4PEX GA Day 2: starting with in-depth scientific talks and posters on our main achievements so far, before discussing in more detail next steps for the atmosphere, ocean and land teams. In the evening we went to Kulturen Open Air Museum, immersing in the history of the region. #CINEA #HorizonEU

Today we're kicking off the 1st #AI4PEX General Assembly in Lund, Sweden, hosted by @lund-university.bsky.social. Within the next 4 days we will talk about achievements, discuss methods and plan research of next year, while getting closer, building trust and turning our team into a project family.

💡 New Paper! Clouds are crucial for climate modeling, affecting sunshine, heat, and rainfall. Traditional models use coarse grids, but new methods involve high-detail simulations and #MachineLearning to predict cloud cover and cloud content to improve #HybridModels. 👉 Learn more: buff.ly/uIN1R3O

#AI4PEX & #EXPECT joining forces Fantastic momentum at #EGU25! Teams from two #HorizonEU projects came together to explore synergies and plan future collaboration. Next steps? We're aiming for a joint hackathon and an ECR meeting to build stronger connections. #CINEA_EU #ESMs

AI4PEX @ #EGU: the past 2 days we heard a lot about Machine Learning applications and Climate Extremes by Ramon Fuentes Franco, Tom Beucler, Helena Reid, Svenja Seeber and Kai Cohrs.

AI4PEX @ #EGU25 We started the week in Vienna with talks on heatwaves, wildfires and extreme evaporation by Dominik Schumacher, Lars Nieradzki & Yannis Markonis. On Day 2 we dived deeper into ocean modeling with Hongmei Li and model evaluation using the ESMValTool by Valeriu Predoi. #CINEA #ESM

🛠 Climate science needs sharp tools. Check out ESMValTool, MESMER, benchmark datasets & more at #EGU25. Predoi, Bauer, & Cohrs presenting powerful open-source tools, statistical methods, and new frameworks that make climate science more reproducible, transparent, and effective. #OpenScience

🌱 What’s up with carbon sinks, wildfires & ecosystem feedbacks in a changing climate? We’ve got you at #EGU25. Models, remote sensing & field data combined. Talks by Ilyina, Behncke, Nieradzik, Gomarasca & co. 🌍 #CarbonScience #ClimateFeedbacks #EGU25

🔥 Heatwaves, flash floods & weird weather? We’re digging into climate extremes at #EGU25 and how these events link to climate change. ⚡️ Catch talks by Schumacher, Ivanov, Seeber, and more! #ExtremeClimate #WeatherEvents #EGU25 #ClimateScience

🌍 AI is transforming climate research! Catch us at #EGU25 presenting cutting-edge work on applying #MachineLearning, hybrid models, and causal discovery in climate research. ⚡️ Don’t miss: Camps-Valls, Reid, Ouala, Beucler + more! #AI4Climate #ML4Science #EGU25 #ClimateAI

💡 New Preprint! How to improve plant physiological processes in #ESM? Check out the enhanced capability of our HYBRID-JSBACH4 model using feed-forward neural networks (FNN) pre-trained on observations capturing variability in stomatal conductance while reducing transpiration bias 👉 buff.ly/XdEyHd2

💡 New Paper! #MachineLearning improves #EarthSystemModels, but explaining its added value is tough. Inspired by climate model hierarchies, we propose using Pareto-optimal models to distill ML's added value in representing cloud cover, radiative transfer and tropical precipitation. 👉 buff.ly/iirtWkk

What are the climate models we are working on? Land model number 4: JSBACH4 is the land component of the ICON #ESM where we are focusing on phenology, stomatal conductance, and tree mortality. 👉 Learn more: buff.ly/xEiRG8c

📢The first 2 years of our project, we brought researchers & local decision-makers together in 2 World Café workshops to discuss climate science & policymaking. Now, we’re revisiting those conversations with a short summary — a chance to reflect on what was shared. 🗣️ Read it here: bit.ly/ESM2025WCS

What are the climate models we are working on? Land model number 3: ORCHIDEE (Organising Carbon and Hydrology In Dynamic Ecosystems), the land surface model of the IPSL #EarthSystemModels, where our focus is especially on the hydrological processes. 👉 Learn more: buff.ly/83X23ho #ClimateResearch

What are the climate models we are working on? Land model number 2: JULES (Joint UK Land Environment Simulator): a community land surface model coupled to the Met Office Unified Model where we will build high-dimensional emulators to calibrate land surface behavior. 👉Learn more: buff.ly/XkWSYOM

What are the climate models we are working on? Let’s start with the land models: LPJ-GUESS from Lund University, is a process-based dynamic vegetation-terrestrial ecosystem model where we focus on forest dynamics. 👉 Learn more: buff.ly/kJQxy5H #EarthSystemModels #ClimateResearch

🌍🌪️ How is climate change fueling extreme weather? Our concept paper dives into Extreme Event Attribution (EEA) — the key to linking weather events to climate change. While new methods are making it easier to track the impact of climate change, important challenges still remain. Paper linked 🔽

💡 New Paper! Have you ever wondered, in which ways #ArtificialIntelligence can help us to better predict and respond to #ExtremeEvents? 👉 Find out what #AI can already do, its challenges and opportunities: www.nature.com/articles/s41...

💡 New Paper! Check out how to better represent the formation of thin low-level clouds in atmospheric models! Our #MachineLearning model enhances vertical detail in temperature and humidity profiles to improve cloud fraction. 👉 Learn more: https://buff.ly/3CMZoKu

🚀 Registration Now Open for the @ELLISforEurope Summer School: AI for Earth & Climate Sciences! 🌍🤖 Join us in Jena, Germany from September 1–5, 2025 📅 Apply now through March 31. Notifications by April. 🔗 www.ellis-jena.eu/summer-schoo... 🎯 @uni-jena.de, @MPI-BGC, @carlzeissstiftung.bsky.social

💡 New Preprint! What are the benefits of an online versus offline learning approach? Explore our application of online training approach to learning subgrid parameterisations of baroclinic ocean #eddies. 👉 Find out more: https://buff.ly/3WtdEPk #ESMs

💡 New Preprint! Find out how recent significant improvements of ESMValTool allows a more effective evaluation of complex #ESMs and high-resolution models while needing less computational resources. 👉 Join the discussion: gmd.copernicus.org/preprints/gm... @schlunma.bsky.social

💡 New Paper! How to improve #DeepLearning submodels for hybrid numerical modelling systems? Ouala et al. showcase an efficient and practical online learning approach using Euler Gradient Approximation for #HybridModels. 👉 Learn more: https://buff.ly/3C7uwUU

💡 New Paper! 🌡️ The observed temperature record is key for understanding #climatechange, but new findings suggest early 20th-century ocean temperatures were underestimated by ~0.26°C. Correcting this cold bias would result in a better projections. 👉 Find out more: https://buff.ly/4h6WyOU

🌍 Deadline approaching: If you are working with #AI and #MachineLearning in the context of the #CarbonCycle, submit your abstract to #EGU25 session: “AS 5.6 - Machine Learning for Carbon Cycle Research”. 📅 Deadline: January 15, 2025 🔗 Session Details: https://buff.ly/4adg3mG

@uni-hamburg.de is hiring a Postdoc for the #AI4PEX project focusing on biogeochemical impacts of ocean #ExtremeEvents with high-resolution #ESMs and #MachineLearning. Start date: 01.04.2025 (3 years) 📅 Application deadline: 06.02.2025 👉 More info and application:

*POSTDOC JOB* within the vibrant project @ai4pex.bsky.social Apply to join our dynamic team at Universitaet Hamburg focusing on biogeochemical impacts of ocean extreme events with high-resolution ESM and ML. Details here: www.uni-hamburg.de/stellenangeb... 🧪 🌊

👉 Only one week left to submit your abstract to #EGU25. If you are working on #AI for climate predictions consider submitting your abstract to session CL5.9: "AI-driven Forecasting for Weather, Climate, and Extreme Events" Deadline: January 15th, 2025: https://buff.ly/4gLhThe

#AI4PEX research focus 9: Coupling. We will couple, tune and evaluate the new #ML -enhanced component models of the land, ocean and atmosphere to #ESMs as more realistic component models are the first step towards improved projections of the future #EarthSystem.

#AI4PEX research focus 8: Projections AI4PEX aims to deliver improved #climateprojections of changes in the regimes of #ExtremeEvents and associated impacts from the next generation of #ESMs #ArtificialIntelligence -enhanced and grounded on #EarthObservations.

#AI4PEX research focus 6: Uncertainty Quantification To overcome limitations of traditional manual calibration methods, such as high computational costs and slow iterative processes, we improve uncertainty quantification by leveraging on Perturbed Parameter Ensemble. #EarthSystemModels #Emulation

#AI4PEX research focus 5: Land We build #hybridmodels representing short-to-long-term responses of #TerrestrialEcosystems to changes in climate and atmospheric CO2; improve the vegetation response to water and heat stress, and the temperature sensitivity of decomposition in soils.

#AI4PEX research focus 4: #Ocean We will improve the representation of ocean-related feedbacks on climate dynamics by (I) joint reconstruction of ocean heat and C fluxes (II) parameterization of ocean mesoscale processes (eddies) (III) emulation of ocean biogeochemistry

#AI4PEX research focus 3: Atmosphere We will build new data-driven, yet physics-aware, ML-based #hybridmodels to better represent #CloudFeedbacks and related processes that occur on scales smaller than the ESMs’ model grid.

#AI4PEX research focus 2: Model Development To develop better #EarthSystemModels we will expand the use of #EarthObservation in combination with #MachineLearning. We focus on explainability of #AI models and causality to better identify model-observations mismatch.

#AI4PEX research focus 1: #EarthObservation We leverage existing in-situ observations and EO data-streams to generate new upscaled #MachineLearning -based products, and cross-check with the Copernicus land, ocean and atmospheric datasets. #EarthSystemModels #climatescience

How does #AI4PEX plan to overcome uncertainties in #ESMs and improve #climateprojections in the future? Over the next days, we will introduce you to the main research tasks of our project. 🚀 Stay tuned!