Probabilistic Precipitation Nowcasting with Rectified Flow Transformers
This work provides an incremental improvement in precipitation nowcasting accuracy for meteorologists and weather-dependent industries by better capturing uncertainty in extreme weather events.
This paper addresses the challenge of probabilistic precipitation nowcasting, which is crucial for various domains and safety-critical in extreme weather. The authors introduce FREUD, a rectified flow transformer model that achieves state-of-the-art performance on the SEVIR benchmark by employing a frame-wise encoder and a unified decoder for efficient, uncertainty-preserving compression of spatio-temporal weather data.
Accurate weather forecasts are essential across various domains and are safety-critical in extreme weather conditions. Compared to simulation-based forecasting, data-driven approaches show greater efficiency, enabling short-term, high-resolution nowcasting. In particular, diffusion models proved effective in weather nowcasting due to their strong probabilistic foundation. However, existing methods rely on deterministic compression to reduce the complexity of high-dimensional weather data, limiting their ability to capture uncertainty in the decoding process. In this work, we introduce $\textbf{FREUD}$, a $\textbf{Fr}$ame-wise $\textbf{E}$ncoder and $\textbf{U}$nited $\textbf{D}$ecoder model based on rectified flow transformers for efficient compression of spatio-temporal weather data. Frame-wise encoding enables continuous forecast updates, while the unified video decoder ensures temporal consistency. Our uncertainty-preserving first stage allows us to capture aleatoric uncertainty via ensembling, which is particularly beneficial for extreme weather events with high decoding variability. We achieve state-of-the-art performance in precipitation nowcasting with a compact latent-space rectified flow transformer on the SEVIR benchmark and show further performance gains by model and test-time scaling. Code available here: https://github.com/CompVis/weather-rf