Land surface temperature data assimilation and its impact on evapotranspiration estimates from the Common Land Model

A variational data assimilation algorithm for assimilating land surface temperature (LST) in the Common Land Model (COLM) is implemented using the land surface energy balance as the adjoint physical constraint. In this data assimilation algorithm, the evaporative fraction of soil and canopy is adjusted according to the surface temperature observations. The analysis results from COLM with the LST assimilation algorithm compare well with the field observations from AmeriFlux data at four sites with different land surface conditions. These results indicate that the surface temperature assimilation method is efficient and effective when only one observation for each day is available (e. g., an observation at 1400 local solar time). A sensitivity analysis for the COLM estimation of evapotranspiration (ET) is also carried out, and the impact of the surface temperature data assimilation on the ET estimates is assessed. It is found that except for time periods with a heavy rainfall event, the ET estimates with the surface temperature assimilation once per day during 0800-1600 local solar time compare better with the AmeriFlux ET measurements than do those without the LST data assimilation. This study implies that the data assimilation algorithm for assimilating regional-scale remote sensing LST data into COLM is promising.