Analysis of Soil Moisture and Overland Flow Generation Using Cellular Automata and Self-Organizing Feature Maps

Soil moisture is a key component and has a major influence on the generation of overland flow. The space-time self-organizations of soil moisture and overland flow generation in Tarrawarra experimental catchment in Australian are analyzed here. The A-K network, which is the combination of ART neural network and Kohonen neural network, is used to identify the spatial pattern of soil moisture. The semivariograms are calculated for the clustering center of each identified pattern in order to find the structures of variance. The variety of overland flow generating area in catchment is analyzed by using a cellular automata which model the self-organizing incorporation in soil water balance including infiltration, rainfall and evaporation. In this paper, our goal is to find the possible occurrence of self-organization in hydrological process. Some initial results tend to approve the hypothesis.