Fuzzy Model Use For Evaluation Of Peak Outflow From A River Basin During Flash Flood

One of the most difficult tasks in field of operative hydrology is the prediction of occurrence and course of flash floods. It is difficult to predict torrential rainfalls because of their character. Usage of the nowcasting methods (weather forecast with two hour validity) holds hope. The torrential rainfall prediction should be followed by a suitable hydrological model able to estimate at least the resultant peak outflow. The hydrological models construction interferes with high measure of uncertainty, inherent in the rainfall prediction, rainfall-runoff process and its simulation. The way how to eliminate influence of uncertainty is using fuzzy logic and other artificial intelligence methods. The fuzzy model was compiled through the Fuzzy Logic Toolbox in the developmental environment of MATLAB. A model was calibrated with the help of genetic algorithm, neural networks and different optimization methods.