FuXi-Nowcast: Meet the longstanding challenge of convective initiation in nowcasting

2025-12-02

AI Summary

Nowcasting of convective initiation and high-impact weather (e.g., heavy precipitation, gusts) at 1–2-hour lead times remains a major challenge. This study proposes a multi-source fusion deep learning framework that integrates radar reflectivity, surface observations, HRLDAS, and FuXi-2.0 three-dimensional atmospheric fields to construct a Swin-Transformer-based multi-task nowcasting model. We introduce two novel components: a convection-signal enhancement module and a distribution-aware hybrid loss function—both designed to effectively mitigate echo intensity attenuation commonly observed in deep learning–based precipitation forecasts. The framework delivers joint 12-hour, 1-km-resolution extrapolation forecasts of radar reflectivity, precipitation, near-surface temperature, wind, and gusts over Eastern China. Experiments demonstrate consistent superiority over the operational CMA-MESO 3-km system across Critical Success Index (CSI) scores for reflectivity, precipitation, and gusts—with the most substantial gains observed for heavy precipitation forecasting—and significantly improved spatiotemporal accuracy in capturing convective initiation, evolution, and structural characteristics.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Accurate nowcasting of convective storms remains a major challenge for operational forecasting, particularly for convective initiation and the evolution of high-impact rainfall and strong winds. Here we present FuXi-Nowcast, a deep-learning system that jointly predicts composite radar reflectivity, surface precipitation, near-surface temperature, wind speed and wind gusts at 1-km resolution over eastern China. FuXi-Nowcast integrates multi-source observations, such as radar, surface stations and the High-Resolution Land Data Assimilation System (HRLDAS), with three-dimensional atmospheric fields from the machine-learning weather model FuXi-2.0 within a multi-task Swin-Transformer architecture. A convective signal enhancement module and distribution-aware hybrid loss functions are designed to preserve intense convective structures and mitigate the rapid intensity decay common in deep-learning nowcasts. FuXi-Nowcast surpasses the operational CMA-MESO 3-km numerical model in Critical Success Index for reflectivity, precipitation and wind gusts across thresholds and lead times up to 12 h, with the largest gains for heavy rainfall. Case studies further show that FuXi-Nowcast more accurately captures the timing, location and structure of convective initiation and subsequent evolution of convection. These results demonstrate that coupling three-dimensional machine-learning forecasts with high-resolution observations can provide multi-hazard, long-lead nowcasts that outperforms current operational systems.

Problem

• Accurate nowcasting of convective storms and initiation
• Predicts multiple weather variables at high resolution
• Outperforms operational models in capturing severe weather

Innovation

• Deep-learning system integrates multi-source observations with 3D atmospheric fields
• Convective signal enhancement module preserves intense convective structures
• Distribution-aware hybrid loss functions mitigate rapid intensity decay in nowcasts