Physics-Guided Flow-Map Matching for Precipitation Nowcasting

2026-09-27

AI Summary

This study addresses the blurring of intense convective structures in precipitation nowcasting caused by deterministic models. We propose a physics-guided conditional flow matching method that pioneers the use of a frozen Lagrangian advection prior as guidance rather than simple superposition, effectively decoupling predictable advection from uncertain small-scale detail generation to prevent inherited blurriness. By integrating a conditional flow matching generative head with a four-step sampling technique, the approach efficiently synthesizes sharp stochastic details. Experimental evaluations demonstrate that the proposed model surpasses state-of-the-art methods on 18 out of 24 metrics across four radar benchmarks, achieving up to a 58.9% improvement in the Critical Success Index (CSI) for heavy rainfall thresholds.

Abstract

Precipitation nowcasting, generating future radar fields from past observations, is critical for flood warning and disaster response. It is also a demanding benchmark for spatiotemporal generative modeling, with chaotic dynamics, heavy-tailed intensities, and rare high-intensity structures that matter most. Deterministic models minimize a pixel loss and are driven toward the conditional mean, which blurs exactly those structures, while generative models that add a stochastic residual on top of a deterministic backbone inherit the same blur. We propose Physics-Guided Flow-Map Matching (PG-FMM), a conditional flow-map model that decouples predictable advection from uncertain small-scale detail. A frozen Lagrangian advection prior transports the radar field and supplies an explicit motion forecast, and a flow-map generative head, conditioned on the past frames and the prior rollout rather than summed onto it, produces sharp stochastic detail in four sampling steps. The prior serves only as guidance, so the head replaces blurred structure instead of inheriting it. Extensive experiments on four radar benchmarks show that PG-FMM outperforms state-of-the-art methods on 18 of 24 metrics, with the largest gains at heavy-rain thresholds, where the critical success index improves by up to 58.9%. The project page can be found at https://neurogica.github.io/PG-FMM.

Problem

• Precipitation nowcasting
• Spatiotemporal generative modeling
• Radar field prediction
• Deterministic blur

Innovation

• Physics-Guided Flow-Map Matching
• Precipitation Nowcasting
• Lagrangian Advection Prior
• Conditional Flow Model
• Spatiotemporal Generative Modeling