Analysis and bias correction of land surface temperature in the GEOS-5 global atmospheric modeling and assimilation system

The land surface skin temperature (Tskin) is central to the surface energy, water, and radiation balances, and in an atmospheric model improved skin temperature states are expected to yield improved temperature and humidity profiles. In this study, Tskin estimates retrieved from geostationary Earth orbiting satellite observations are assimilated into the Goddard Earth Observing System Model, version 5 (GEOS-5) atmospheric modeling and assimilation system to correct the modeled land surface temperature biases. The Tskin observations are assimilated every three hours globally, excepting high latitudes, using a model-bias aware ensemble Kalman filter-based land data assimilation system. Prior to assimilation, the geostationary Tskin observations are themselves bias-corrected, using an empirical bias correction function based on 2 m atmospheric temperature (T2m) observations and the Tskin-T2m relationship modeled by the GEOS-5 system. The impact of the assimilation is then evaluated by examining the impact on the modeled surface temperatures, land surface fluxes, and low-level temperature and humidity. Ultimately, it is also hoped that improved Tskin states in GEOS-5 will also lead to enhanced assimilation of atmospheric radiances from surface-sensitive channels.