Radar Data Assimilation with the WRF-LETKF System and Its Preliminary Results of OSSEs on Typhoon Morakot (2009)

A Doppler radar data assimilation system implementing the local ensemble transform Kalman filter (LETKF) technique is first established upon the Weather Research and Forecasting (WRF) model in this study. The goal is to evaluate the capability of this system in improving the short-term forecasts for severe storms, especially the challenging quantitative precipitation forecasts (QPFs) over complex terrain. A series of tests using the Observing System Simulation Experiment (OSSE) strategy are performed to study the period when Typhoon Morakot (2009) passed through Taiwan and produced record-breaking rainfall. Ensemble members are initialized by perturbing the NCEP FNL data with WRFDA and simulated in three-level, two-way nested domains. Among the ensemble runs, the one with the most realistic precipitation is selected as the nature run, from which radial wind observations are extracted based on the configuration of Taiwan Central Weather Bureau (CWB) RCCG S-band Doppler radar.

The preliminary results show that the assimilation of the radial wind observations remarkably reduces the analysis errors of both the directly-related (e.g. horizontal winds) and indirectly-related (e.g. mixing ratios of hydrometeors) model variables. This indicates the reliability of the ensemble-based multivariate background error covariance. The 15-min assimilation cycle gives a better agreement between the root mean square error and the ensemble spread than the 7.5-min one does. After an assimilation period of three hours, the improvement in the forecast errors, including that of rainfall, can persist for 1 to 2 hours compared with the run without any assimilation. Details about the impact from the radar observations on the short-term forecast skill will be presented. We will also discuss the strategy of efficiently assimilating the radar data with high resolution in space and in time.