Estimation and validation of high-resolution evapotranspiration products for an arid river basin using multi-source remote sensing data

Accurate estimation of evapotranspiration (ET) at high spatial resolution is crucial for drought monitoring and water resources management, but currently available remote sensing ET products generally have coarse spatial resolution (≥1000 m). To estimate ET at a high spatial resolution, Landsat images, Global Land Surface Satellite (GLASS), Moderate Resolution Imaging Spectroradiometer (MODIS), and meteorological forcing data were integrated, and the surface energy balance (SEBS) model was employed to calculate the 16-day average ET at 30 m resolution for China's Tarim River Basin, spanning from 2009 to 2018. The results indicated that the average 16-day ET estimates correlated well with ground observations for land and water surfaces (root mean square error (RMSE) for land = 0. 92 mm day−1, RMSE for water = 1. 63 mm day−1, mean bias for land = 0. 3 mm day−1, mean bias for water = 0. 52 mm day−1). Cross validation with GLASS, ETMonitor, and Penman-Monteith-Leuning (PMLV2) ET datasets revealed an overall increasing trend for all four products (PMLV2 = 6. 277 mm year−1, GLASS = 2. 185 mm year−1, ETMonitor = 3. 258 mm year−1, SEBS = 1. 441 mm year−1), demonstrating good spatial consistency. The consistent increasing pixels were primarily distributed in the northern, southwestern, and southeastern mountainous regions, accounting for 22. 8%, while 0. 29% of the consistent decreasing pixels were mainly concentrated in the central desert and mountain-front oasis areas. Inconsistent pixels accounted for 76. 9%, with 2. 34% of the inconsistent decreasing pixels exhibiting a scattered distribution, while 37. 28% of the inconsistent increasing pixels were mainly found in the central desert and some oasis areas. Furthermore, SEBS ET trend analysis indicated that the oasis area experienced more pronounced changes than the mountainous and desert areas during the 2009–2018 period. The SEBS ET estimated in this study can provide high-precision data support and a reference for future research on the water resources management.