Satellite-based monitoring and hazard assessment of multiple flooding types in the Haihe river Basin induced by the July 2023 extreme rainstorm

Abstract

In the context of global climate change, extreme rainstorms are increasingly frequent. In complex terrain, a single rainstorm often triggers different types of flooding, including flash floods, river floods, and waterlogging, posing threats to socioeconomic and human safety. Most studies treat these floods uniformly, relying on a single data source and method, inadequately capturing their heterogeneity and causes. Focusing on the July 2023 extreme rainstorm and flood in the Haihe River Basin, this study developed a framework to monitor different types of flooding using multi-source satellite data and to analyse their driving factors and hazards. In the plains, we used Sentinel-1 and GF-3 SAR data to extract inundated areas of river floods and waterlogging. In the mountain areas, flash flood-affected areas were identified by combining Sentinel-2 NDVI changes with terrain and hydrological analyses. Random forest models compared driving factors among flood types and assessed flood hazards. Results show flash floods in mountain areas concentrate in low-lying river valleys, with slope contributing 24.33%. In the plains, river floods expanded progressively, while waterlogging displayed significant differences in response between urban and rural areas, with distance to rivers (27.88%) and maximum daily rainfall (13.84%) identified as dominant factors. From these findings, we generated a flood hazard distribution map that accurately identifies areas with high flood hazard. By classifying flood types and analysing driving factors, this study reveals the heterogeneity of flooding induced by a single extreme rainstorm. Furthermore, it offers scientific insights and a novel framework for managing floods under extreme rainfall.