Using Precipitation Assimilation to Improve Simulations of Regional Climate Extremes

Daily precipitation analyses are being assimilated in a regional climate simulation bounded by the National Centers for Environmental Prediction / Dept. of Energy global reanalysis (NCEP/DOE R2) in order to improve simulations of not only precipitation but also other variables and processes. The assimilated rain rates were obtained from the Climate Prediction Center observed precipitation analysis for the continental United States. R2 precipitation was used over the rest of the regional model domain. These daily rain rates were interpolated to the regional model time-steps, and were continuously assimilated during the entire regional simulation, which covered the period July 1st, 1986 to December 31st, 1996. The precipitation assimilation (PA) scheme adjusts the model vertical specific humidity profiles based on the difference between the "observed" and the model predicted precipitation. A regional climate control experiment without precipitation assimilation was also performed. R2 also provides the initial and the boundary conditions for both regional experiments. The NCEP North American Regional Reanalysis (NARR) was used to evaluate the PA and the control simulations as well as the R2 performance. We focus here on the impact for two extreme events, the 1988 drought and the 1993 flood during the summertime (June, July). As it was expected, the PA precipitation was highly correlated with the observations as well as NARR for both summers, and therefore was significantly better than the corresponding precipitation from the control simulation and NCEP-R2. The PA evaporation patterns also compared relatively better with NARR patterns. The dynamic fields were also improved compared to the control for both summers. Overall PA was able to better reproduce the differences in the upper- and lower-level winds than the control simulation, which thus brought the regional simulation more in line with the bounding R2.