P-39

Ensemble forecast sensitivity to the observations (EFSO) applied to precipitation data assimilation

G.-Y. Lien (RIKEN), E. Kalnay (University of Maryland), D. Hotta (Japan Meteorological Agency), and T. Miyoshi (RIKEN; University of Maryland)

The Forecast Sensitivity to the Observations (FSO) technique allows us to estimate the forecast error changes made by the assimilation of each individual observation. Because this method can estimate the impacts of all individual observations simultaneously, it is a very economical alternative to the computationally expensive observing system experiment (OSE), which requires running the entire data assimilation experiment many times. FSO can be computed under the ensemble data assimilation framework (EFSO; Kalnay et al. 2012; Ota et al. 2013). In this study we apply the EFSO method to precipitation data assimilation. The impacts of assimilating every TRMM Multisatellite Precipitation Analysis (TMPA) precipitation observation on the 6-hour GFS model forecasts are estimated during a one-year period. By performing statistics over this one-year sample of EFSO, we obtain an idea of the effectiveness of precipitation assimilation in terms of several factors, such as the geographic location, the observation value, and the number of precipitating members in the background ensemble. This approach can be an efficient way to develop appropriate quality control criteria for assimilating a new observing system.