WIND: Weather Inverse Diffusion for Zero-Shot Atmospheric Modeling

Deep learning has revolutionized weather forecasting, but many challenges remain, including climate modeling. Moreover, the current landscape remains fragmented: highly specialized models are typically trained individually for distinct tasks. To unify this landscape, we introduce WIND, a single pre-trained foundation model capable of replacing specialized baselines across a vast array of tasks. Crucially, in contrast to previous atmospheric foundation models, we achieve this without any task-specific fine-tuning. To learn a robust, task-agnostic prior of the atmosphere, we pre-train WIND with a self-supervised video reconstruction objective, utilizing an unconditional video diffusion model to iteratively reconstruct atmospheric dynamics from a noisy state. At inference, we frame diverse domain-specific problems strictly as inverse problems and solve them via posterior sampling. This unified approach allows us to tackle highly relevant weather and climate problems, including probabilistic forecasting, spatial and temporal downscaling, reconstruction of spatial fields from sparse observations and enforcing global dry air mass conservation. We further demonstrate how WIND can be applied to explore extreme weather events under prescribed out-of-distribution thermodynamic perturbations. By combining generative video modeling with inverse problem solving, WIND offers a computationally efficient alternative for AI-based atmospheric modeling.

AI weather models have made remarkable progress over the last few years. However, extending them to longer timescales and adapting them for climate applications remains an open challenge. Formulating climate-related tasks, such as zooming in on local weather details or filling in missing satellite measurements, is currently not done in a unified way, meaning individual models must be trained for each specific task. To solve this, we first show how to train a more stable atmospheric emulator utilizing AI video-generation techniques. We then show how various climate-related problems can be solved using a flexible framework that answers climate-science questions on the fly, without needing to retrain the model. This unified approach is an efficient tool for analyzing atmospheric dynamics. By replacing specialized tools with one flexible model, WIND can even be used for complex but important tasks like enforcing physical laws and simulating extreme storms under future climate change scenarios.