Hourly precipitation analyses have been successfully assimilated into the Eta model since July 2001. They have made a significant improvement on the model's soil moisture field and a modest improvement on precipitation forecasts. However, the hourly analyses (NCEP Stage IV,supplemented by the more timely NCEP Stage II) tend to have a dry bias, and when they are assimilated directly into the model, over time the cumulative effect of small precipitation deficits can lead to noticeable dry bias in the model soil.

We have developed a method to correct this bias by comparing model precipitation received each day by the soil against a daily precipitation analysis (the NCEP/CPC 1/8 deg gauge analysis), and accumulate this daily deficit/surplus into a long-term precipitation budget history array for the model domain. We then use this budget history array to modify the hourly input for precipitation assimilation, with the aim of balancing out the budget deficit/surplus within the next 24 hours. The maximum amount of adjustment to the hourly precipitation input is limited to +/-20% of the original analysis value. This long-term budget check/adjustment approach has proven to be beneficial to the quality of the model's water cycle components and helpful in fine-tuning the model's soil moisture field.

The bias-adjustment of the hourly precipitation input was implemented into the Eta model on 16 Mar 2004. Detailed discussion of the approach and its impact on the model will be presented at the time of the AMS meeting.