Author(s): Yifan Li; Chendi Zhang; Marwan A. Hassan
Linked Author(s):
Keywords: Flash floods impact of induicng factors graph neural network (GNN) interpretation model the Hengduan Mountains region (HMR)
Abstract: The Hengduan Mountains region (HMR) in China faces significant threats from flash flood disasters but the impacts of inducing factors for flash floods have not been well understood. This study aims to quantify the impacts of single and integrated inducing factors, which provide insights to the formation mechanisms of flash floods. A framework, combining Graph Neural Network (GNN) and interpretation models, was established to obtain the flash flood regionalization and to quantify the single and integrated inducing factors’ impact. Integrated factors were constructed from five perspectives, i. e., geomorphology, climate, meteorology, soil, and vegetation, using principal component analysis (PCA). The results highlighted the importance of extreme rainfall and soil water conductive characteristics on the formation of flash floods.
Year: 2025