Errors in the modeling of surface processes profoundly affect mesoscale model simulations. Some of these dominant errors are due to uncertainty in the initial specification of soil moisture and parameterization errors in the radiation budgets and soil heat flow associated with subsurface processes. To minimize these types of errors in mesoscale model simulations, we developed a technique to perform a continuous assimilation of GOES retrievals of skin (ground) temperature, insolation, and surface albedo. Using this technique we adjust soil temperature and moisture to minimize errors in the surface thermodynamic and dynamical fields. In this technique, the differences between the modeled and GOES-derived skin temperature tendencies are used to estimate the errors in the modeled surface heat fluxes. These errors in the surface heat fluxes are referred to as adjustment heat fluxes. The adjustment latent heat flux is then used in a land surface model to minimize the errors in the specification of soil moisture by introducing additional flux terms in the prognostic equations of soil moisture. Further, both these adjustment heat fluxes are used to reduce the errors in the predicted skin/soil temperature. Using the Mesoscale Model, Version 5 (MM5), this technique was tested in the NOAH land surface model. We have performed four numerical simulations using the MM5. In the first simulation no GOES data were assimilated; In the second simulation, only GOES-derived skin temperature tendencies were assimilated; In the third, errors in the modeled soil heat flux are reduced by using the GOES-derived insolation; In the fourth simulation, GOES-derived albedo and insolation data were assimilated. Evaluation results of our technique using the MM5 model for a case study will be presented.