Evaluation and Uncertainty Estimation of Multi-Source Evapotranspiration Analyses over the Continental United States

Land evapotranspiration (ET, a sum of direct and bare soil evaporation, evaporation from canopy interception, and transpiration from vegetation) is a central process in the climate system and an essential component of the water, energy, and carbon cycles. ET governs land surface-atmosphere interactions of water vapor and heat energy, and it affects regional and global carbon cycles. ET is regularly monitored for practical applications such as detecting flash drought and managing agricultural crops. However, the availability of high-quality, in-situ ET observations is limited. Therefore, the ET simulated by land surface models or retrieved from satellite data is used as a good alternative. In this study, modeled and remotely-sensed ET data are inter-compared and evaluated against high-quality in-situ tower ET measurements.

The monthly ET products produced from five satellite retrievals (ALEX, GLEAM, Gridded FLUXNET, MODIS, SSEBop) and eight NLDAS land surface models (Mosaic, Noah2.8, SAC, VIC4.0.3, CLSM, Noah3.6, Noah-MP3.6, VIC4.1.2) are evaluated over the continental United States. The eddy covariance measurement ET and water balance ET (WBET) were selected as validation data at point and basin scales. It should be noted that the spatial-scale mismatch and energy non-closure error is an issue that cannot be ignored when tower-measured ET is used as a reference data set to assess products with coarser spatial resolution. Low spatial and temporal resolution (basin scale, annual) of the water balance method hampers its application for evaluating high temporal and spatial resolution ET. Additionally, these ET products contain many uncertainties coming from model structures, model parameters, and input data. Therefore, the Three Cornered Hat (TCH) method combined with WBET and tower-measured ET is used to estimate the relative uncertainties of these products at basin scales (i.e., for basins from 12 River Forecast Centers). The investigation for this work is underway and some preliminary results are presented in this study.