Performance evaluation and parameters sensitivity of a distributed hydrological model for a semi-arid catchment in India

Abstract

In present study, a distributed physics based hydrological model, MIKE SHE coupled with MIKE 11, is calibrated using multi-objective approach, i.e., minimization of error in prediction of stream flows and groundwater levels, using the data of eight years from 1991 to 1998 of Yerli sub-catchment $$(\hbox {area} = 15{,}881\,\hbox {km}^{2})$$ of upper Tapi basin in India. The sensitivity analyses of thirteen model parameters related with overland flow, unsaturated and saturated zones have been undertaken while simulating the runoff volume, peak runoff at catchment outlet and groundwater levels within the catchment with wide variations $$(\pm 50\%)$$ in the model parameters. The calibrated model has also been validated for prediction of stream flow and groundwater levels within the Yerli sub-catchment for period 1999–2004. The simulated results revealed that calibrated model is able to simulate hydrographs satisfactorily for Yerli sub-catchment (NSE $$=$$ 0.65–0.89, $$r=0.80{-}0.95$$ ) at daily and monthly time scales. The ground water levels are predicted reasonably satisfactorily for the plain area (RMSE $$=$$ 0.50–6.50 m) in the study area. The results of total water balance indicated that about 78% of water is lost from the system through evapotranspiration, out of which about 3.5% is contributed from the groundwater zone.