Capturing typhoon‑driven causality: A dynamic graph convolutional network approach to interpretable storm surge forecasting

doi:10.1016/j.ress.2026.113193

Capturing typhoon‑driven causality: A dynamic graph convolutional network approach to interpretable storm surge forecasting

2026-07-22

0

PRE

AI

DOI：10.1016/j.ress.2026.113193

Original

Original request for help

Save

Abstract

En 中文

• Graph convolution integration enhances multi-station storm surge forecasting. • Graph structure learning infers dynamic inter-station causality during typhoons. • Interpretability is analyzed via graph visualization. • Physical propagation mechanism is shown to improve the forecasting model.

Keywords:

Storm surge

Dynamic graph structure learning

Interpretability

Deep learning

Causal inference

Capturing typhoon‑driven causality: A dynamic graph convolutional network approach to interpretable storm surge forecasting

Abstract

Journal

Organization