A study of application of the precipitation data assimilation technique to numerical simulation of an excessive meiyu front rainfall event

With the heavy rainfall numerical forecast model AREM,three numerical experiments with and without the surface and sounding data assimilation were carried out respectively to simulate an excessive meiyu front rainfall event occurred in the middle and lower reaches of the Yangtze River on 29 June 2009, using the data assimilation system GRAPES 3DVAR and LAPS (Local Analysis Prediction System) with NCEP forecasts as the background field. Then the assimilation effects of the vertical profile data retrieved from the precipitation data 1DVAR method (by assimilating them in the GRAPES 3DVAR and LAPS respectively) were investigated. The results indicate that: （1） the rainfall forecasts from the sounding assimilation experiment using the GRAPES-3DVAR/LAPS are worse/better than that from the no data assimilation experiments. As to this excessive rain case, the background field modification by the GRAPES 3DVAR/LAPS data assimilation system gives a negative/positive effect on the rainfall forecast; （2） the assimilation of the vertical profiles retrieved from the precipitation data 1DVAR method in the GRAPES-3DVAR data assimilation system has an important effect on the physical field, with the air above the rain belt becoming more moist and warmer and the weather system configuration being more favorable to the precipitation and the development of the mesoscale system, which are more beneficial to simulating a location and intensity of the rain belt and mesoscale system structure closer to the observation. Therefore it to a large degree improved the precipitation simulation, with the location and intensity of 1 h, 6 h, 24 h accumulated precipitation and the mesoscale structure feature obviously improved; （3） the application of the precipitation data 1DVAR method in the LAPS system also improved the rainfall forecasting and made the rain belt location be closer to the observation. 