The precipitation assimilation compares the model-produced precipitation against the observed values, and modifies the model's temperature, moisture and cloud fields accordingly during the 12-hour pre-forecast data assimilation period. The method has been tested extensively using the NCEP hourly multi-sensor precipitation analysis and have been shown to greatly improve the soil moisture field while providing a small improvement to the precipitation forecast.
The NCEP hourly precipitation analysis is not manually quality-controlled. Recently the manually QC'd multi-sensor hourly precipitation analyses (using similar algorithm as in the NCEP hourly analysis) from the 12 River Forecast Centers became available, and NCEP has started to mosaic them into a national hourly product. The manual QC of the data makes the operational implementation of the precipitation assimilation more feasible.
The implementation would be completed at the conference time, and results from real-time parallel tests and retrospective tests for different seasons will be presented. Planned Future implementation includes the assimilation of GOES precipitation estimates over the ocean, and the assimilation of GOES cloud data to improve the model's cloud field.
Supplementary URL: