Improved Global Estimates of Terrestrial Evapotranspiration Using the MODIS and VIIRS Sensors

Abstract Terrestrial evapotranspiration (ET), as the link between water, energy, and carbon cycles, is key to understanding climate impacts on freshwater availability, agricultural yields, and forest mortality. Earth-observing sensors, such as the Moderate Resolution Imaging Spectroradiometer (MODIS) carried by NASA’s Earth Observing System (EOS), are well suited to the study of ET at regional to global scales. As the era of EOS comes to a close, we present a comprehensive calibration and validation of the latest MODIS MOD16 ET product as well as new ET estimates based on data from the Visible Infrared Imaging Radiometer Suite (VIIRS) sensor, which is very similar to MODIS and is carried by the Suomi NPP and NOAA-20 satellites, to continue the now 25-yr record of ET estimates from MOD16. Substantial updates and improvements to MOD16 are reported for the first time in over a decade of widespread use. Ground data sources including tower latent heat fluxes and plant traits are used in a Bayesian model-data fusion to constrain MODIS- and VIIRS-based ET estimates. Independent validation against sapflow and tower data shows that the resulting products have lower absolute bias and higher accuracy than the current MOD16 Collection 6.1 product. While MOD16 underestimates cropland ET, an intercomparison with the OpenET suite of models indicates that the improved MOD16 has the lowest error, lowest absolute bias, and highest model efficiency at noncropland sites. In combination with MODIS primary productivity estimates, these updates to MOD16 ensure the continuity of multidecadal ET and water-use efficiency estimates through 2030 or later.