Abstract:Evapotranspiration (ET) is the major component of water cycle and energy balance in the soil-vegetation-atmosphere system. Accurate estimation of regional ET is crucial to agricultural water allocation and water resources management. Combining the MODIS data and the meteorological observations from national weather stations, this paper used the SEBAL (Surface Energy Balance Algorithms for Land) model of energy balance theory to estimate the ET pattern of agro-pastoral ecotone in Northwest China during the growing season in 2015 (April to October). The Penman-Monteith formula (P-M), together with the crop coefficients, was used to compare the ET results by the SEBAL model. The results show that the average absolute error between the SEBAL and the P-M ET values is 0.79mm/d, the root mean square error (RMSE) is 0.94mm/d, and R2=0.76. While slightly higher than the actual ET values, the SEBAL estimated ET values reasonably represent ET pattern in the study region. The range of daily average ET is 0.12-10.66mm/d, and the average daily ET is 4.31mm/d during the growing season. Spatially, the ET is higher in the southeastern and northeastern areas and lower in the western area of the study region. Statistical analysis of the ET and land surface parameters shows a significant positive correlation between the ET and NDVI and net radiation, and a negative correlation with the surface temperature and surface albedo. The daily ET values display a decreasing order with the following land use/cover types: cultivated land > forest land > unutilized land > grassland.