Multi-variable calibration of a semi-distributed hydrological model using streamflow data and satellite-based evapotranspiration

In this study, streamflow (Qs) and satellite-based actual evapotranspiration (ETa) are used in a multi-variable calibration framework to reproduce the catchment water balance. The application is for the HBV rainfall–runoff model at daily time-step for the Karkheh River Basin (51,000 km2) in Iran. Monte Carlo Simulation serves to estimate parameter values and to assess uncertainty for three calibration cases. In case one streamflow is used as the calibration target. In case two satellite-based ETa is used as calibration target. For both cases model performance is evaluated for the second variable that closes the water balance. In case three a preference-based multi-variable objective function is applied which is weighted for Qs and satellite-based ETa. For cloudy days a procedure is developed to complete the daily time series of satellite-based ETa that cover 4 years. Results on multi-variable calibration indicated satisfying results for both water balance terms. Results are compared against field observations and results of single-variable calibration. For cases one and two the second variable only is poorly simulated and resulted in poor reproduction of the water balance. The most important contribution of this work is that the catchment water balance is best reproduced when both Qs and satellite-based ETa serve as calibration target.