Comparative study of IHACRES model optimisation schemes Available to Purchase

IHACRES, a popular catchment-scale rainfall-runoff model for simulating streamflow hydrographs, has been widely used by the hydrological community. The standard IHACRES modelling system consists of a calibration scheme based on the grid search method for estimating the required parameters. This research found that such a calibration approach is time consuming and not always reliable in finding the best parameter values. Alternative model calibration approaches – genetic algorithms (GAs), constraint non-linear optimisation (CNLO) and un-constraint non-linear optimisation (UCNLO) – are explored in this study. A comparison using visual inspection and statistical R² criterion was carried out to evaluate these parameter calibration approaches. It was found that significant improvements with the developed models could be achieved using CNLO, GA and UCNLO over the standard IHACRES method (with CNLO as the best in terms of calibration and validation accuracy). The paper also discusses the computational time issues of those schemes.