Examining the impact of meteorological forcing uncertainty on land surface model-based evapotranspiration estimates

Obtaining accurate estimates of the amount of moisture leaving the land surface and required for vegetation demand is important for water resource management, agriculture, and meteorological and climate-based studies. Several studies have compared evapotranspiration (ET) formulations and methods, off-line and coupled model estimated ET and energy fluxes, like sensible heat, and satellite-based ET estimates (e.g., from MODIS). In this study we examine and evaluate how ET estimates are influenced by the uncertainty in meteorological forcings and differences in land surface and crop models. The forcing ensembles include a variety of datasets, such as operational (e.g., GDAS), reanalysis (e.g., MERRA-Land) and satellite-based (e.g., TRMM-3B42 precipitation). Experiments driven with individual forcings (e.g., operational-only, reanalysis-only, etc.) are compared against the ensemble-based simulations to see where they fall within the ensemble spread and are similar to the ensemble mean. In addition, independent in situ (e.g., FLUXNET) and satellite-based (e.g., MOD16) ET estimates are included to further evaluate and show spatially and seasonally how the simulations and products may vary or are similar. Different regions in the U.S. and Africa will be highlighted to emphasize what impacts, if any, data rich and sparse areas have on the ensemble-based precipitation and the resulting ET ensemble variances. Since ET can be significantly influenced by seasonal vegetation signals and crop presence, the models used in the study are also parameterized with time-varying vegetation parameters and crop extent maps to provide a more realistic land-surface representation in estimating the ET.