Discovering Optimal Triplets for Assessing the Uncertainties of Satellite-Derived Evapotranspiration Products

Information relating to errors in evapotranspiration (ET) products, including satellite-derived ET is critical their application but often challenging obtain, with a limited number of flux towers available for the sufficient validation measurements. Triple collocation (TC) methods can assess inherent uncertainties above products using just three independent variables as triplet input. However, both severity which violate assumptions zero error correlations and corresponding impact on estimation are unknown. This study proposed cross-correlation analysis approach discover optimal regard providing most reliable estimation. All possible triple solutions same product were first evaluated by extended (ETC), among optimum was selected based correlation between ETC-based in-situ-based metrics, correspondingly, statistic experiment ranked triplets demonstrated how valid all pixels product. Six popular (MOD16, PML_V2, GLASS, SSEBop, ERA5, GLEAM) that produced 2003 2018 cover China’s mainland chosen experiment, estimates compared measurements from 23 in-situ towers. The findings suggest (1) there exists an input TC other collocating inputs together least; (2) characteristics six varied significantly across China, GLASS performing best (median error: 0.1 mm/day), followed GLEAM, MOD16 below 0.2 while PML_V2 SSEBop had slightly higher median (0.24 mm/day 0.27 mm/day, respectively); (3) removing seasonal variations signals has substantial enhancing accuracy estimations.