Application feasibility of GIS-based neural network model in soil water analysis in the hilly Loess Plateau
基于GIS与神经网络的黄土丘陵区土壤水分模型研究

The main limiting factor of vegetation construction in hilly areas of Loess Plateau is soil moisture.The spatial distribution of soil moisture is influenced by meteorological factors,land-uses and complex topographies,making soil moisture-vegetation relationship pretty complex.This paper utilized data from soil moisture tests in the hilly Loess Plateau of Zhifanggou catchment to build a GIS-based BP neural network model.The model technique includes multiple factor influences on spatial distribution of soil moisture.Field data was used in training the net which then used to forecast soil moisture of the cachment.The forecast result is in good agreement with practical situations,showing the feasibility of GIS-based BP neural network in analyzing soil moisture distribution regulation in highly complex regional topographic terrains.