Comprehensive evaluation of 0.25° precipitation datasets combined with MOD10A2 snow cover data in the ice-dominated river basins of Pakistan

A major portion of Pakistan's economy is based on cultivated lands which are irrigated from the supply of water from Upper Indus River Basins (UIB). Any change in UIB rivers flows may come with catastrophic events and therefore, will destructively affect Pakistan's economy. By aiming this scenario, an uneven and important climate variable (i.e., precipitation) obtained from different gridded and satellite datasets were used for its statistical and hydrological performance evaluation in UIB catchments for the period of 2000 to 2004. In addition, a bias corrected technique and snow cover product (MOD10A2) was also used to enhance the performance of precipitation data sets to obtain realistic discharge simulations. The results indicated that without correcting the biases from the datasets, only APHRODITE precipitation dataset showed higher correlation with observations compared to other precipitation datasets in Hunza River Basin (HRB) with correlation coefficient of (0.44) & and in Gilgit River Basin (GRB) (0.35), respectively. However, after applying bias correction technique (quantile mapping), the performance of precipitation datasets significantly improved. For GRB, correlation coefficient and root mean square values improved up to 48% & 55%, while for HRB up to 53% & 51%, respectively. Likewise, based on hydrological utility which was implied by the well-known hydrological model (snowmelt runoff model), bias corrected CHIRPS and APHRODITE precipitation datasets displayed best performance in simulating the discharge with Nash–Sutcliffe coefficient (0.82 & 0.90) & correlation coefficient (0.83 & 0.84) in HRB and (0.84 & 0.80) and (0.86 & 0.82) in GRB, respectively. Moreover, recalibration was also carried out to assess how the hydrological model can adjust and tolerate the errors of different precipitation data products. The results revealed that after adjusting the model parameters particularly coefficient of rainfall and coefficient of snow, the performance of data products significantly improved in terms of the difference in volumes against in situ measurements. Overall, this study may assist, provide guidelines and efficiently used for snowmelt runoff model coupled with different precipitation datasets for management of Indus River irrigation system of Pakistan.