Event-Aware Spatiotemporal Precipitation Forecasting with Geographic Context and Physics-Guided Regularization

2026-10-08

AI Summary

This study addresses the challenges of spatial heterogeneity, imbalanced sample distributions, and skill degradation over lead time in hourly precipitation forecasting by proposing a spatiotemporal prediction framework that integrates geographic representation, event-aware learning, and physical regularization. Methodologically, explicit geographic encoding is introduced to characterize spatial heterogeneity, while an event-aware loss function is designed to mitigate data imbalance. Furthermore, a weakly asymmetric physical constraint grounded in the atmospheric water budget is formulated to suppress unphysical forecast decay. Experimental results demonstrate that the proposed framework substantially enhances the detection capability for moderate-to-heavy precipitation and alleviates systematic underestimation. Notably, it exhibits superior forecasting accuracy and robust stability under extended lead times and data-sparse scenarios.

Abstract

Hourly precipitation forecasting involves several distinct statistical challenges, including spatially varying predictor-precipitation relationships, a strongly imbalanced precipitation distribution, and progressive degradation of precipitation event skill with increasing lead time. We develop a spatiotemporal forecasting framework that addresses these challenges through three complementary components: explicit geographic representation, event-aware learning for imbalanced precipitation, and weak asymmetric regularization derived from the atmospheric water budget. The physical information is treated as an asymmetric constraint rather than as an additional prediction target, designed to discourage physically unsupported precipitation attenuation without replacing the data-driven forecast. Experiments using ERA5 data across regional, enlarged domain, and spatial subset settings show that explicit geographic information improves spatial field prediction, while event-aware learning provides the most consistent gains in detecting moderate and heavy precipitation events. Physical regularization has a more selective effect, mainly reducing systematic underprediction over the enlarged domain while improving longer lead precipitation event prediction in the spatial subset experiment. These results indicate that the benefit of physical guidance depends on the available data regime and becomes most apparent when data-driven precipitation information deteriorates with increasing lead time.

Problem

• precipitation forecasting
• spatial heterogeneity
• class imbalance
• lead time degradation
• spatiotemporal prediction

Innovation

• spatiotemporal forecasting
• event-aware learning
• physics-guided regularization
• geographic representation
• precipitation forecasting