Streamflow Simulation in Ungauged Catchments Using the Regionalization Methods in Hamoun-Jazmourian River Basin

In this study, to simulate streamflow time series in ungauged catchments, the parameters of two HBV and IHACRES hydrological models were transfer from gauged (donor) to ungauged catchments using three main regionalization methods including Physical Similarity (PS), Multiple Regression (MR), Spatial Proximity (SP) and an integrated method (IDW-PS). This study was carried out using daily data including precipitation, temperature and evaporation related to at 21 sub-catchments in Hamoun-Jazmourian River Basin in southeast Iran over a 14-year period (2004-2016). Regionalization methods were studied under three modes: temporal (transferring across different periods), spatial (transferring between same calibration periods but different sites) and spatiotemporal (transferring across both different periods and sites). The results indicated that: (1) the more complex HBV model showed better results than the simple model, so that the average NSE coefficients in two different periods in the HBV model were 0.625, 0.57 and 0.5 for calibration, validation and the most appropriate regionalization method (physical similarity) respectively, while these values for IHACRES model were 0.57, 0.51 and 0.46, (2) multiple regression method with mean NSE coefficients equal to 0.2 and 0.24 for HBV and IHACRES models showed the worst regionalization results and (3) the HBV parameters related to snow and runoff components, were associated with highest and lowest uncertainties respectively, while for the IHACRES, the most and least robustness parameters were plant stress threshold factor, f and the proportion of slow flow to total flow, vs respectively.