As demonstrated in previous studies, incorporating remotely sensed precipitation estimates into the initialization of a numerical weather prediction model can improve the depiction of the fields of moisture, latent heat flux, cloud cover, and other related parameters in the model forecasts, especially in regions where appropriate in situ observations are difficult to obtain. To that end, the Global Data Assimilation System (GDAS) of the Environmental Modeling Center's (EMC) Medium Range Forecast (MRF) model has been making limited use of instantaneous rain rate estimates derived from radiances from the Defense Meteorological Satellite Program (DMSP) Special Sensor Microwave/Imager (SSM/I).

This work represents an effort to improve the incorporation of SSM/I estimates into the GDAS. Two months of data (August-September 1999), including Stage III radar/raingauge estimates, Tropical Rainfall Measuring Mission (TRMM) radar/raingauge data, atoll raingauge data, and Oklahoma Mesonet data has been assembled and used to determine the relative error characteristics of the MRF forecasts and the SSM/I estimates against a common data set. The SSM/I estimates have consequently been incorporated into the GDAS using a three-dimensional variational technique (3-D VAR). The impacts on the MRF precipitation forecasts and on other related model fields are presented.