Deploying climate models with real-world fidelity

Further progress in AI+climate modeling "Applying the ACE2 Emulator to SST Green's Functions for the E3SMv3 Global Atmosphere Model". Building on ACE2 model which uses our spherical Fourier neural operator (SFNO) architecture, this work shows that ACE2 can replicate climate model responses to sea surface temperature perturbations with high fidelity at a fraction of the cost. This accelerates climate sensitivity research and helps us better understand radiative feedbacks in the Earth system. Background: The SFNO architecture was first used in training FourCastNet weather model, whose latest version (v3) has state-of-art probabilistic calibration. AI+Science is not just about blindly applying the standard transformer/CNN "hammer". It is about carefully designing neural architectures that incorporate domain constraints like geometry and multiple scales, while being expressive and easy to train. SFNO accomplishes both: it incorporates multiple scales, and it respects the spherical geometry and this is critical for success in climate modeling. Unlike short-term weather, which requires only a few autoregressive steps for rollout, climate modeling requires long rollouts with thousands or even greater number of time steps. All other AI-based models fail for long-term climate modeling including Pangu and GraphCast which ignore the spherical geometry. Distortions start building up at the poles since the models assume domain is a rectangle, and they lead to catastrophic failures. Structure matters in AI+Science!

What if climate simulations could be fast, accurate and interpretable? 🌍⚡ This paper explores exactly that.” In the series of Exploring #SciML Papers from our research wing Vizuara Technologies Private Limited , We will now discuss paper by Tirtha Tilak Pani et al. Climate modeling sits at the intersection of science, policy, and urgent global action. Traditional Earth System Models (ESMs) offer detailed, realistic simulations, but they demand immense computational resources — making rapid scenario exploration difficult. On the other hand, simplified models like Energy Balance Models (EBMs) are efficient but often miss key dynamics. 🔍 What we did: Our work proposes a novel framework that bridges this gap by integrating Scientific Machine Learning (SciML) with classical climate–carbon cycle models to achieve both high accuracy and mechanistic interpretability. Key highlights include: 1) Built a hybrid modeling framework using Universal Differential Equations (UDEs) that embeds physical laws with neural networks. 2) Achieved <0.2% error across key climate variables , outperforming purely data-driven Neural ODEs and classic statistical baselines like ARIMA and VAR. 3) Employed symbolic regression to extract interpretable equations, ensuring transparency and explainability, critical for science-based policy decisions. 4) Designed for computational efficiency, enabling fast exploration of emission scenarios under limited data conditions. 📊 Why it matters: This research shows that with the right SciML integration, we can rapidly simulate climate dynamics, maintain physical fidelity, and generate interpretable insights, all without the computational burden of full-scale Earth system models. This has promising implications for climate risk assessment, policy testing, and scenario planning. 🔗 Read the full paper here ➤ https://lnkd.in/dd5suuFf Huge thanks to everyone involved, looking forward to feedback and collaboration! 🙌 Attaching the paper below as well. #ClimateScience #MachineLearning #ScientificML #ClimateModeling #DataScience #UDE #NeuralODE #AAAI #AAAIWorkshop #AI #ML #Climate #SciML

Google Research has released a new version of NeuralGCM, an AI powered climate model. The code, training data, and model checkpoints are now fully open source. This release focuses on improving the accuracy of rainfall simulations and adds new stochastic models. NeuralGCM is different from traditional climate models because it is trained using real-world rainfall measurements. The updated model addresses shortcomings of previous versions by improving the accuracy of rainfall simulations. It more realistically captures heavy rainfall events and daily rainfall patterns, outperforming even specialized, high-resolution models in test simulations.  Updated paper: https://lnkd.in/erV3BC_K  Original Paper: https://lnkd.in/esGzS_gN  Code: Apache V2 Model license: CC BY-SA 4.0 NeuralGCM Documentation: https://lnkd.in/eTuJgxSq NeuralGCM GitHub Repository: https://lnkd.in/eFu46tu5 Inference demo notebook: https://lnkd.in/eyusJpRC Checkpoints modification notebook: https://lnkd.in/e9kRvuYJ Credits for the update: Janni Yuval, Ian Langmore, Dmitrii Kochkov, Stephan Hoyer.

2026 - starting the year strong 💪 My colleagues at Google Research published a new paper in Science Advances that marks a significant step forward for large-scale precipitation forecasts. We’ve trained our hybrid AI-physics model, NeuralGCM, directly on NASA satellite observations to simulate global precipitation with a 40% average error reduction over land compared to leading climate models in multi-year runs. Precise precipitation forecasting is one of the "holy grails" of climate science—and it’s notoriously difficult because clouds are at smaller scales than traditionally modeled ☁️. Precipitation forecasts are so relevant in multiple scenarios: it's about knowing whether a farmer should plant seeds today or if a city needs to prepare for a 100-year storm. Here is why this development is a game-changer: ☁️ Smarter Tuning (compared to traditional models): Traditional models rely on fixed equations (parameterizations) that are difficult to tune perfectly for every scenario and rarely utilize the vast data available. NeuralGCM uses neural networks that are trained "online"—meaning they learn to work in harmony with the large-scale physics solver. ☁️ Learning Directly from Observations (compared to other hybrid models or ML models): While most AI models learn from "reanalysis" data (a mix of observations and model physics that can carry biases), NeuralGCM is trained directly on NASA satellite data. This allows the model to align its precipitation predictions with the best available record of actual rainfall. ☁️ Capturing Extremes:  NeuralGCM is significantly better at capturing extreme precipitation which traditional models often under-predict. ☁️ Correcting the Clock:  While many models predict peak rain too early in the day , NeuralGCM accurately reproduces the timing of peak precipitation, especially in complex regions like the Amazon. ☁️ Real-World Application:  This isn’t just theoretical. This past summer, a partnership with the University of Chicago and the Indian Ministry of Agriculture used NeuralGCM to provide AI-based monsoon forecasts for 38 million farmers. AI is learning the "parameterizations" of complex small-scale physics (like cloud formation) that have baffled traditional models for decades. A huge congratulations to Janni Yuval, Stephan Hoyer, Dmitrii Kochkov, Ian Langmore, Michael Brenner, Lizzie Dorfman, Olivia Graham, and the entire team for pushing the boundaries of what's possible for our planet’s resilience. Read the full story on the Google Research blog: https://lnkd.in/ga8V5jq8 Paper: https://lnkd.in/g3wfG4q2