Mapping Regional Evapotranspiration in Cloudy Skies via Variational Assimilation of All-Weather Land Surface Temperature Observations

Variational data assimilation (VDA) approaches have been widely used to estimate evapotranspiration (ET) by assimilating remotely sensed thermal inferred-land surface temperature (TIR-LST) data from Geostationary and Polar Orbiting Weather Satellites. Key unknown parameters of the VDA approaches are neutral bulk heat transfer coefficient (C_HN, scale the sum of turbulent heat fluxes) and evaporative fraction (EF, represents the partitioning of available energy between sensible and latent heat fluxes). However, the TIR-LST data are mostly unavailable in cloudy skies, leading to the poor estimation of C_HN and EF, and consequently ET. Fortunately, the all-weather LST product (that contains both the TIR and microwave LST observations) provides LST observations even in cloudy conditions. In this study, both the all-weather LST data and MODIS TIR-LST products are assimilated into the VDA system to estimate ET. The performance of the VDA approach is tested over the Source Regions of Rivers (SRR) in Southwest China. The ET estimates are compared with the measurements at four sites in the SRR (namely, Dangxiong, Linzhi, Naqu, and Qomolangma). Results show that the ET retrievals are close to the observations. Overall, this study significantly advances the ability of VDA approaches to retrieve ET in cloudy skies.