WADEPre: A Wavelet-based Decomposition Model for Extreme Precipitation Nowcasting with Multi-Scale Learning

2026-02-02

AI Summary

This study addresses the challenge of accurately forecasting extreme precipitation in nowcasting, a task hindered by regression-to-the-mean bias induced by heavy-tailed distributions and the lack of spatial locality in Fourier-based methods. To overcome these limitations, the work introduces discrete wavelet transform into precipitation nowcasting for the first time, proposing a dual-branch neural architecture that separately models low-frequency advection and high-frequency convection in the wavelet domain. A dynamic reconstruction module is designed to effectively fuse multi-scale information, complemented by a multi-scale curriculum learning strategy that enhances training stability and improves the model’s ability to capture extreme events. Evaluated on the SEVIR and Shanghai radar datasets, the proposed method significantly outperforms existing approaches, achieving superior accuracy and structural fidelity in predicting both large-scale trends and fine-scale transient features.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

The heavy-tailed nature of precipitation intensity impedes precise precipitation nowcasting. Standard models that optimize pixel-wise losses are prone to regression-to-the-mean bias, which blurs extreme values. Existing Fourier-based methods also lack the spatial localization needed to resolve transient convective cells. To overcome these intrinsic limitations, we propose WADEPre, a wavelet-based decomposition model for extreme precipitation that transitions the modeling into the wavelet domain. By leveraging the Discrete Wavelet Transform for explicit decomposition, WADEPre employs a dual-branch architecture: an Approximation Network to model stable, low-frequency advection, isolating deterministic trends from statistical bias, and a spatially localized Detail Network to capture high-frequency stochastic convection, resolving transient singularities and preserving sharp boundaries. A subsequent Refiner module then dynamically reconstructs these decoupled multi-scale components into the final high-fidelity forecast. To address optimization instability, we introduce a multi-scale curriculum learning strategy that progressively shifts supervision from coarse scales to fine-grained details. Extensive experiments on the SEVIR and Shanghai Radar datasets demonstrate that WADEPre achieves state-of-the-art performance, yielding significant improvements in capturing extreme thresholds and maintaining structural fidelity. Our code is available at https://github.com/sonderlau/WADEPre.

Problem

• extreme precipitation nowcasting
• heavy-tailed precipitation
• regression-to-the-mean bias
• spatial localization
• transient convective cells

Innovation

• Wavelet-based decomposition
• Extreme precipitation nowcasting
• Multi-scale learning
• Dual-branch architecture
• Curriculum learning