A Space-Time Transformer for Precipitation Nowcasting

TL;DR

This paper introduces SaTformer, a space-time transformer model that uses full space-time attention to improve precipitation nowcasting from satellite data, addressing limitations of traditional models and previous AI approaches.

Contribution

The paper presents SaTformer, a novel space-time transformer architecture for weather prediction, and introduces techniques to handle long-tailed precipitation data, achieving state-of-the-art results.

Findings

First place in NeurIPS Weather4Cast 2025 challenge.

Effective reformulation of precipitation regression as classification.

Techniques to address label imbalance in precipitation datasets.

Abstract

Meteorological agencies around the world rely on real-time flood guidance to issue life-saving advisories and warnings. For decades traditional numerical weather prediction (NWP) models have been state-of-the-art for precipitation forecasting. However, physically-parameterized models suffer from a few core limitations: first, solving PDEs to resolve atmospheric dynamics is computationally demanding, and second, these methods degrade in performance at nowcasting timescales (i.e., 0-4 hour lead-times). Motivated by these shortcomings, recent work proposes AI-weather prediction (AI-WP) alternatives that learn to emulate analysis data with neural networks. While these data-driven approaches have enjoyed enormous success across diverse spatial and temporal resolutions, applications of video-understanding architectures for weather forecasting remain underexplored. To address these gaps, we propose \textbf{SaTformer}: a video transformer built on full space-time attention that skillfully forecasts extreme precipitation from satellite radiances. Along with our novel architecture, we introduce techniques to tame long-tailed precipitation datasets. Namely, we reformulate precipitation regression into a classification problem, and employ a class-weighted loss to address label imbalances. Our model scored \textbf{first place} on the NeurIPS Weather4Cast 2025 ``Cumulative Rainfall'' challenge. Code and model weights are available: \texttt{\href{github.com/leharris3/w4c-25}{github.com/leharris3/satformer}}