Flash Flood Peak Estimation in Small Mountainous Catchments Based on Distributed Geomorphological Unit Hydrographs Using Fuzzy C-Means Clustering

Abstract

The peak flow of a flash flood is an important factor concerning flood risk management. Empirical approaches based on measured hydrological data at gauging sites can be applied to estimate flood peaks; however, they are not suitable for catchments with limited available rainfall and streamflow data, especially ungauged small mountainous catchments. This paper presents a novel method for estimating flash flood peaks in small mountainous catchments based on derived distributed geomorphologic unit hydrographs (DGUHs) using fuzzy c-means (FCM) clustering. Data from 25,130 small mountainous catchments in Gansu Province, located in northwest China, are selected for this study. The results show that (1) the derived DGUHs can reflect the strong nonlinearity of the hydrologic process over small mountainous catchments; (2) the FCM clustering method performs well in the clustering of the DGUHs, and the optimal number of clusters in this case is insensitive to the excess rainfall intensity; and (3) empirical formulas based on catchment characteristics (including drainage area, length, and the longitudinal gradient of main channel networks) can estimate flash flood peaks accurately in small mountainous catchments.