Sensitivity analysis and automatic calibration of a rainfall–runoff model using multi-objectives

The development of numerical model parameter sensitivity analysis (SA) has focused mainly on the selection of a single-objective measure of the distance between the model-simulated output and the data. However, practical experience with sensitivity analysis suggests that no single objective function is adequate to measure the ways in which the model fails to match the important characteristics of the observed data. In order to successfully measure parameter sensitivity of a numerical model, multiple criteria should be considered. Sensitivity analysis of a rainfall–runoff model is performed using the local sensitivity method (Morris method) and multi- objective analysis. Formulation of SA strategy for the MIKE/NAM rainfall–runoff model is outline. The SA scheme includes multiple objectives that measure different aspects of the hydrograph: (1) overall influence, and (2) parameter interaction. The SA is given as a set of Pareto ranks from a multi-objective viewpoint. In order to successfully calibrate the numerical model, multiple criteria should be considered. Multi-objective Differential Evolution (MODE) has proved effective in numerous such applications, where most of the techniques relying on the condition of Pareto efficiency to compare different solutions. We describe the performance of the population-based search algorithm (Non-dominated Sorting Differential Evolution (NSDE)) when applied to the rainfall–runoff model. The method has been applied for calibration of a test catchment and compared on validation data. The simulations show that the NSDE method possesses the ability to finding the optimal Pareto front and keeping a good diversity in the obtained front for the model calibration.