DNN-MET: A deep neural networks method to integrate satellite-derived evapotranspiration products, eddy covariance observations and ancillary information

Accurate estimates of the spatiotemporal distribution evapotranspiration (ET) are essential for understanding terrestrial energy, carbon and water cycles. Station-based observations limited their spatial coverage whereas satellite-derived ET products exhibit large discrepancies uncertainties. Here we presented a Deep Neural Networks based Merging (DNN-MET) framework that combines information from products, eddy covariance (EC) ancillary surface properties to improve representation ET, especially in data-sparse regions. DNN-MET was implemented over Heihe River Basin (HRB) 2008 2015, performance eight input state-of-the-art (i.e., MOD16, ET-SEMI, ET-JPL, ET-MS, ET-HF, GLEAM, ETMonitor EB-ET) evaluated against 19 EC flux tower sites. The results showed improved HRB, decreased RMSE by 0.13 1.02 mm/day (14%-56%) when compared with products. also yielded superior derived other merging methods Random Forest, Bayesian model averaging simple method). When validated data-scarce regions, its remained better even training samples were 20% available An innovation our approach is building multivariate properties, incorporated geographical proximity effects autocorrelations into procedure, which can be used as “spatial knowledge engine” predictions. readily effectively applied elsewhere various hydrometeorological variables.