Assessing evapotranspiration variability in contiguous United States: Comparing the various remote-sensed observations

Abstract:

Evapotranspiration (ET) couples the water cycle and energy budget of hydrological processes. Understanding the components of ET variability and their spatial distribution is essential for improving hydrological simulations, quantifying ET observation uncertainties and supporting water resources management under climate change. Although advances in monitoring hydrological components have been made, how to use various existing observations to obtain a better knowledge about ET variability remains a challenging task. This study adopts a system approach to analyze ET variability in contiguous United States, considering the factors of climatic forcing fluctuations and catchment storage dynamics. We apply an ET variance decomposition framework (Zeng and Cai, 2015) to calculate monthly ET variance based on climatic forcing (i.e., precipitation and potential ET) and GRACE terrestrial storage change data. We quantify the various sources of ET variance, in terms of variances of precipitation, potential ET and terrestrial storage and their covariances, and obtain a spatial map of its primary and secondary controlling factors in the in contiguous United States. Furthermore, the estimated ET variance is compared to two existing ET products (e.g., MODIS-based remote sensing and FLUEXNET-based interpolation). It is found that FLUXNET-based interpolation is systematically smaller than the estimated ET variance with less deviation; while the MODIS-based ET agrees with estimated ET variance with larger uncertainty. The decomposition framework provides not only an independent estimation of ET variance but also a method to assess the uncertainty of existing ET products.