Comparison of Performance between Genetic Algorithm and SCE-UA for Calibration of SCS-CN Surface Runoff Simulation

摘要: Global optimization methods linked with simulation models are widely used for automated calibration and serve as useful tools searching cost-effective alternatives environmental management. A genetic algorithm (GA) shuffled complex evolution (SCE-UA) were the Long-Term Hydrologic Impact Assessment (L-THIA) model, which employs curve number (SCS-CN) method. The performance of two was compared by automatically calibrating L-THIA monthly runoff from 10 watersheds in Indiana. selected watershed areas ranged 32.7 to 5844.1 km2. SCS-CN values total five-day rainfall adjustment optimized, objective function Nash-Sutcliffe value (NS value). GA method rapidly reached optimal space until 10th generating population (generation), after generation solutions increased dispersion around space, called a cross hair pattern, because mutation rate increase. looping executions influenced model SCE-UA performed better case fewer loop than For most watersheds, using 50th when 5150 (one has 100 individuals). However, method, Optimized primary land use types nearly same methods, but those minor AMC somewhat different parameters did not significantly influence calculation function. is recommended cases takes long time user does have sufficient an program search best parameters. other cases, automatic calibration.