TOWARD A DYNAMIC-THERMODYNAMIC ASSIMILATION OF SATELLITE SURFACE-TEMPERATURE IN NUMERICAL ATMOSPHERIC MODELS

An assimilation technique is described in which satellite-observed sur face skin temperature tendencies are used in a model surface energy bu dget so that the predicted rate of temperature change in the model mor e closely agrees with the satellite observations. Both visible and inf rared GOES satellite data are used in the assimilation. The technique is based on analytically recovering surface moisture from similarity e xpressions derived from an evapotranspiration residual obtained as a d ifference between the unadjusted model evapotranspiration and the sate llite-inferred evapotranspiration. The technique has application in re gional-scale models where surface parameters such as root zone moistur e, soil moisture, etc., are unknown. It is assumed that the largest er ror in the surface energy budget is in the evapotranspiration term. Tw o tests are given for the technique, first, a one-dimensional test aga inst FIFE data and, second, a three-dimensional test over Oklahoma. In these cases the technique appears to correctly adjust the model respo nse to agree better with observations.