Sensitivity Analysis of a Conceptual HBV Raınfall-Runoff MODEL Using Eumetsat Snow Covered Area Product

Abstract:

HBV is a conceptual hydrological model extensively used in operational hydrological forecasting and water balance studies. In this study, we apply the HBV model on the upper Euphrates basin in Turkey, which has 10 624 km2 area. The Euphrates basin is largely fed from snow precipitation whereby nearly two-thirds occur in winter and may remain in the form of snow for half of the year. We analyze individual sensitivity of the parameters by calibrating the model using the Multi-Objective Shuffled Complex Evolution (MOSCEM) algorithm. The calibration is performed against snow cover area (SCA) in addition to runoff data for the water years 2009, 2010, 2011, 2012 and 2013. The SCA product has been developed in the framework of the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT), Satellite Application Facility on Support to Operational Hydrology and Water Management (H-SAF) Project. The product is generated by using data from Spinning Enhanced Visible and InfraRed Imager (SEVIRI) instrument making observations from a geostationary satellite Meteosat Second Generation (MSG). In the previous study evaluation of the model was done with commonly used statistical performance metrics (Nash-Sutcliffe) for high and low flows, volume error and root mean square error (RMSE). In this study signature metrics, which are based on the flow duration curve (FDC) are used to see the performance of the model for low flows. In order to consider a fairly balanced evaluation between high and low flow phases we divided the flow duration curve into segments of high, medium and low flow phases, and additionally into very high and very low phases. Root mean square error (RMSE) is used to evaluate the performance in these segments. The sensitivity analysis of the parameters around the calibrated optimum points showed that parameters of the soil moisture and evapotranspiration (FC, beta and LPrat) have a strong effect in the total volume error of the model. The parameters from the response and transformation routines (LSUZ, K1, K0 and bmax) have a significant influence on the peak flows. It is observed that the parameters of snow routine (Tmelt, CSF and DDF) have strong effect in high flows and total volume. The parameters FC, K0, K1 And K2 are found to have effect on low flows from the signature metrics.