Efficient Assimilation of Precipitation: Results with TRMM/TMPA and Plans for GPM

Monday, 15 December 2014: 10:50 AM
Eugenia Kalnay1, Guo-Yuan Lien1, Tse-Chun Chen1 and Takemasa Miyoshi2, (1)University of Maryland, College Park, MD, United States, (2)RIKEN Advanced Institute for Computational Science, Kobe, Japan
Many attempts to assimilate precipitation observations in numerical models have been made, but they have resulted in little or no forecast improvement at the end of the precipitation assimilation. This is due to the nonlinearity of the model precipitation parameterization, the non-Gaussianity of precipitation variables, and the large and unknown model and observation errors.

We addressed these problems by using an EnKF (LETKF) that does not require linearization of Precipitation, and by Gaussian transformations of observed and model precipitation. In addition, several QC criteria were designed to reject precipitation observations that are not useful for the assimilation. OSSE experiments with the SPEEDY model, and a low-resolution version of the NCEP GFS was used to assimilate real TRMM/TMPA observations. Both the OSSEs and the real observation experiments gave encouraging results, indicating that both analyses and 5-day forecasts are improved by the assimilation of precipitation (Lien et al., 2013, Lien, 2014). For real observations, we found that it was necessary to apply quality control criteria to reject observations that are not useful for the assimilation. We found that the use of the ensemble forecast sensitivity to observations (EFSO) to analyze the effectiveness of precipitation assimilation gave results superior to the other QC methods. We plan to apply similar techniques to GPM precipitation observations in collaboration with Prof. Miyoshi of RIKEN, Japan.