Analysis on water balance in different land cover types at upper reaches of Heihe River basin in northwestern China using SWAT model

In this study, SWAT (Soil and Water Assessment Tool) model was applied to simulate the monthly runoff of Yingluoxia watershed, located at upper reaches of Heihe River basin in northwestern China for the period from 1995.1 to 2009.12. Results show that the Nash-Sutcliffe efficiency (NSE) value is 0.953 in calibration period (1998.1-2003.12) and 0.908 in validation period (2004.1-2009.12). Accordingly, the relative errors (RE) are -3.64% and 5.15%, and the coefficient of determination (R
²
) are 0.96 and 0.91, respectively. After analyzing water balance in different land cover types based on SWAT model, we draw conclusion that seven land cover types contribute to the total water yield as follow, Deserts with 0.1%, Steppes with 10.1%, Needleleaf Forestes with 5.0%, Scrubs with 10.0%, Meadows 60.6%, Alpine Vegetation 14.1% and No Vegetation with 0.1%. Further calculation shows that the average annual precipitation and actual evapotranspiration in the study area are 516.5 mm and 328.9 mm, and the runoff coefficient is 0.34.