During May and June 2003, the continental United States experienced highly persistent and anomalous circulation patterns and precipitation amounts. All-time high rainfall records were set in several eastern states on the basis of records dating back to 1895. The NASA/NCAR fvGCM (finite volume general circulation model), which uses hydrodynamics of Lin and Rood together with NCAR's CCM3 physics, is run daily on an experimental basis to make 10-day forecasts in real-time; it simulated the rainfall anomalies with discernible skill. The McRAS (Microphysics of clouds with Relaxed Arakawa-Schubert due to Moorthi and Suarez) cloud scheme has shown several promising features in simulating rainfall climatology with the fvGCM in which the NCAR CCM3 cloud scheme and radiation package were replaced with McRAS and the radiation package due to Chou and Suarez. A new set of 10-day forecasts were made for May and June 2003 with McRAS in the fvGCM to assess its performance against that of the NASA/NCAR fvGCM.

Nine monthly time-lagged forecasts were produced with the data for each month by averaging the entire month using the one through ninth day of each forecast, respectively. This can be viewed as an ensemble of 31 (30) forecasts for May (June) with each average having one through nine day lead-time. Comparison with the observations will delineate the differences in the skills of the two cloud schemes.

This work is a precursor to understanding the role of soil moisture in maintaining the circulation pattern and the persistent rainfall anomalies of the spring of 2003. The cloud scheme which best maintains the observed circulation patterns and rainfall amounts will be employed for assessing the role of initial soil moistures. Some key aspects of the influence of soil moisture and land-atmosphere interactions on the simulated climate of North America will be presented.