Variational Data Assimilation of Satellite Radiance to a Mesoscale Rainstorm Simulation

With the development of the numerical weather forecasting technology, the models are becoming more and more mature, such that the quality of forecast depends more on the initial conditions. But the conventional observation is about 200 km away to each other, it can not describe the initial condition of atmosphere well to meet the need of numerical model. To improve these conditions with ATOVS satellite radiance data some researches use T213 products as the background based on GRAPES 3D-Var and MM5 models. The GRAPES (Global and Regional Assimilation and PrEdiction System) is a new generation hydrostatic/non-hydrostatic, multi-scale numerical model of China developed by Research Center for Numerical Meteorological Prediction of CAMS (Chinese Academy of Meteorological Sciences), CMA. GRAPES 3D-Var is the assimilating part of the system, it can assimilate many kinds of observations if the observation operator and observation error matrix is provided. MM5 model is the mesoscale prediction system developed by NCAR/PSU, it can be used for real-time studies, simulating or predicting the weather system such as mesoscale convective systems, fronts, land-sea breezes, mountain-valley circulations, and urban heat islands. A heavy rainstorm occurs in the middle-lower reaches of Yangze River and in the east of the southwest China from 22 to 24 June 2004. It brings serious disasters to the people in and out Hunan Province because of mud slide and flood caused by the rainstorm. So it is very urgent to pay attention to the research of such rainstorms as to mitigate the damage. The ATOVS detector of NOAA satellite consists of HIRS, AMSU-A, AMSU-B channels, of which the radiance of AMSU channels can penetrate the cloud to reach the satellite, so the AMSU radiance contain much information in the cloud which is very important to the initial condition of numerical model. In order to identify the effect of numerical weather prediction by assimilating satellite radiance data, two experiments are carried out by use of the GRAPES 3D-Var assimilation system and the MM5 numerical prediction model to simulate the storm, the NOATOVS experiment only assimilates the conventional sounding data, the ATOVS experiment simultaneously assimilates the conventional sounding data and the ATOVS radiance data. The assimilation results show that directly assimilating ATOVS radiance data can improve the temperature, humidity and wind fields within the troposphere. The temperature increment in the lower troposphere in the mid-lower reaches of Yangtze River and Korea peninsula is positive, the humidity increment in the sea area is positive. The assimilations are applied as the initial field of the MM5 model for prediction, which then undergo analysis for comparison. The comparisons between both the ATOVS and the NOATOVS experiments show that ATOVS experiment can not only simulate the weather situation well, but also describe the intensity and the location of the rainstorm better than the NOATOVS experiment. All these analyses imply that the simulations of mesoscale convective system and heavy local rainstorm may be improved to some extent through assimilating satellite radiance data directly, but further case studies are needed to verify it.