Hydrologic regionalization by using self-organizing feature maps neural network
Xiaoming Xu · Journal of Hydraulic Engineering · 2005
The self-organizing feature maps (SOFM) neural network, which is widely used in pattern classification of data set, is applied to hydrological regionalization of Jiangsi and Fujian Provinces, China. The basic data obtained from 86 stations are used for calculation. The basic data include topographic and hydro-meteorological factors reflecting the characteristics of catchment. By using this method the number of cluster can be automatically identified. The result shows that three distinct hydrologic regions exist in the territory of these two provinces. The validity of this regional classification is verified by the calculated maximum flood discharge occurred in different hydrological stations.