Evapotranspiration Estimation for Tibetan Plateau Headwaters Using Conjoint Terrestrial and Atmospheric Water Balances and Multisource Remote Sensing

Satellite remote sensing combined with water balance calculations provides a promising approach to estimating evapotranspiration (ET), a critical variable in water-energy exchange. Here we compare ET estimates from terrestrial and atmospheric water balances, multisource remote sensing (AVHRR, GLEAM, and MOD16), and a land surface model (GLDAS NOAH) for headwaters on the Tibetan Plateau (TP), that is, headwaters of the Brahmaputra (HBR), Salween (HSR), Mekong (HMR), Yangtze (HYR), and Huang (Yellow; HHR) Rivers, for the 2003-2012 period. Results show that (1) ET estimated from terrestrial and atmospheric water balances agrees closely in three basins (HMR, HYR, and HHR) but has large discrepancies in the other two basins (HBR and HSR), mainly caused by uncertainties in the terrestrial water balance; (2) agreement between various ET products and water balance-derived ET baselines is highest for GLEAM in two basins (HMR and HYR) and GLDAS NOAH in another two basins (HSR and HHR); and (3) large discrepancies between water balance-derived ET and all ET products are found in the most glacierized HBR, which may reflect the importance of sublimation in the ET process. The decadal mean ET based on water balance-derived ET baselines is highest in the HHR (447 mm/year) and HSR (430 mm/year) and lowest in the HBR (238 mm/year), ranging from 51% to 78% of mean precipitation in the five TP headwaters. These findings have important implications for ET estimation on the TP headwaters, which greatly influences downstream water availability.