A skillful self-evolving deep-learning framework for pluvial flood process forecasting in urban areas

doi:10.1016/j.jhydrol.2025.134593

A skillful self-evolving deep-learning framework for pluvial flood process forecasting in urban areas

2025-11-11

0

PRE

AI

DOI：10.1016/j.jhydrol.2025.134593

Original

Original request for help

Save

Abstract

En 中文

• A deep-learning self-evolving framework is proposed to develop surrogate models. • Two autoregressive CNN models update hourly flow depth and velocity maps. • The MAE is 0.7 mm for flow depth and 0.3 mm/s for flow velocity, respectively. • The model is 850× and 3000× faster than GPU and CPU-based numerical simulations.

A skillful self-evolving deep-learning framework for pluvial flood process forecasting in urban areas

Abstract

Journal

Organization