Uncertainty in the initial specification of soil moisture profoundly affects mesoscale model simulations. Further, the errors in the estimated initial soil moisture are persistent and are not reduced within a typical case study/episode. To minimize errors in mesoscale model simulations, we developed a technique to perform a continuous data assimilation of surface-layer temperature and water vapor mixing ratio and adjustment of soil temperature and moisture to maintain dynamically consistent surface-atmosphere coupling. In this technique, first we used the analyzed surface data to directly assimilate surface-layer temperature and water vapor mixing ratio. We then used the difference between the observations and model predictions to calculate adjustments to the surface fluxes of sensible and latent heat. These adjustment heat fluxes were used to calculate a new estimate of the ground temperature, thereby affecting the predicted surface fluxes in the subsequent time step. We also used the adjusted latent heat fluxes to calculate changes in soil moisture using an inverse method in three land surface schemes. This indirect data assimilation of ground temperature and soil moisture was applied simultaneously with the direct assimilation of surface data in the model's surface layer, thereby maintaining greater consistency between the surface layer mass-field variables and the ground temperature and moisture. This technique was tested for three different land surface models: a diagnostic scheme based on land use and prespecified soil moisture availability, Ma, following Carlson and Boland as used in the Mesoscale Model Version 5 (MM5); a prognostic scheme suggested by Noilhan and Planton that is being used in MM5 and many other mesoscale models; and another prognostic scheme suggested by Chen and Dudhia that is also being used in MM5. In the first case, we showed that direct and indirect assimilation of surface observations can be efficiently used to modulate the Ma term and thus develop more realistic soil moisture feedback in the Carlson and Boland scheme. For the second (Noilhan and Planton) and third (Chen and Dudhia) schemes, our assimilation approach was used to estimate adjustments for equitable evaporation and transpiration which were then translated into corresponding soil moisture adjustments in the soil moisture prognostic equations. Evaluation results of our technique using the MM5 model for a case study will be presented.