TG-net: A Physically Interpretable Deep Learning Forecasting Model for Thunderstorm Gusts

Thunderstorm gusts are a common and hazardous type of severe convective weather, characterized by small spatial scales, short durations, and significant destructive power. They often lead to severe disasters, highlighting the critical importance of accurate forecasting for thunderstorm gusts. Previous studies have explored the environmental factors and temporal-spatial distribution characteristics of thunderstorm gusts, highlighting the need for improved forecasting methods. In recent years, artificial intelligence techniques have shown promise in enhancing thunderstorm gust forecasting accuracy, with various machine learning algorithms and models being developed. This paper proposes a multi-scale feature fusion module called Thunderstorm Gusts Block (TG-Block) and a deep learning (DL) model named Thunderstorm Gusts net (TG-net) based on Attention U-net and TG-TransUnet models, and employs interpretable methods such as Integrated Gradient, Deep Learning Importance Features, and Shapley Additive exPlanations to validate the model’s practical relevance and reliability. The analysis of feature importance underscores the model’s ability to capture key thermodynamic and multiscale weather characteristic information for thunderstorm gusts nowcasting. Currently, the model is running in real-time operation in Institute of Urban Meteorology.