Accurate prediction of the structure and evolution of marine clouds remains a research challenge due to the lack of routine measurements over the ocean that are needed for model initial and boundary conditions. The DYCOMS II field program conducted in July 2001 off the southern California coast offers the possibility of using aircraft, drop sonde, and satellite data to evaluate regional and mesoscale models such as COAMPS and MM5 as well as for testing their ability to predict the structure and evolution of the marine coastal clouds. Preliminary simulations indicate that both models underpredicted the depth of the boundary layer as well as cloud coverage and vertical structure. A novel method has been developed using satellite and aircraft data to improve the model initial conditions and more accurately simulate the cloud field. The method is based on assimilation of the processed satellite data on cloud-top and sea-surface temperatures. The method provides more accurate initial conditions of the offshore marine layer and thermodynamic conditions over the ocean that lead to intense development of clouds. The improved simulations compare favorably with aircraft and drop sonde data during the nocturnal evolution of the marine stratocumulus along the southern California coast.