Application of ensemble Kalman filter in WRF model to forecast rainfall on monsoon onset period in South Vietnam

This paper presents some results of rainfall forecast in the monsoon onset period in South Vietnam, with the use of ensemble Kalman filter to assimilate observation data into the initial field of the model. The study of rainfall forecasts are experimented at the time of Southern monsoon outbreaks for 3 years (2005, 2008 and 2009), corresponding to 18 cases. In each case, there are five trials, including satellite wind data assimilation, upper-air sounding data assimilation, mixed data (satellite wind+upper-air sounding data) assimilation and two controlled trials (one single predictive test and one multi-physical ensemble prediction), which is equivalent to 85 forecasts for one trial. Based on the statistical evaluation of 36 samples (18 meteorological stations and 18 trials), the results show that Kalman filter assimilates satellite wind data to forecast well rainfall at 48 hours and 72 hours ranges. With 24 hour forecasting period, upper-air sounding data assimilation and mixed data assimilation experiments predicted better rainfall than non-assimilation tests. The results of the assessment based on the phase prediction indicators also show that the ensemble Kalman filter assimilating satellite wind data and mixed data sets improve the rain forecasting capability of the model at 48 hours and 72 hour ranges, while the upper-air sounding data assimilation test produces satisfactory results at the 72 hour forecast range, and the multi-physical ensemble test predicted good rainfall at 24 hour and 48 hour forecasts. The results of this research initially lead to a new research approach, Kalman Filter Application that assimilates the existing observation data into input data of the model that can improve the quality of rainfall forecast in Southern Vietnam and overall country in general.