Study On ATOVS Radiance Data Assimilation In WRF-EnSRF System

Application of the weather satellite data in the operation system is acknowledged as one of the most important aspects of improving the accuracy of numerical weather forecast. The study of radiance data direct assimilation in the meso-scale numerical forecast field in China is lagging behind other developed countries so that it is still in the preliminary stage. So far, the variational method is mostly used, but as the development of four-dimensional assimilation, ensemble kalman filter got more and more attention and been studied a lot. How to assimilate satellite data using ensemble kalman filter method, which is a new way to improve the model initial field and forecast accuracy, is a very significant research subject.Firstly, using WRFDA as the data pre-process module, the research introduced a new assimilation scheme in the WRF-EnSRF system, which can implement the direct assimilation of ATOVS radiance data. Then, the radiance data are used in the rainstorm simulation experiments in Guangdong region on6-7June2008to test the capability of the assimilation scheme and its performance in the rainstorm simulation. The main results are as follows:(1) The single observation experiment shows that after assimilating the observation, it can modified the model initial field surrounding the observation position in the horizontal, meanwhile, the vertical modification is larger in the mid-lower troposphere which is in accordance with the peak height of the weighting function.(2) Through two groups of sensitivity experiments, it is found that the experiment using60ensemble members and correlation function used for localization in the horizontal falls to zero at1000km is the closest to the truth field in variables and precipitation simulation.(3) Compare between ATOVS radiance data assimilation experiment and no assimilation experiment shows that assimilating can modify the temperature、 humidity and wind of the initial field, which enriches the temperature information in the upper troposphere and humidity information in the mid-lower troposphere. On the other hand, in the observation space, the brightness temperature calculated by the analysis is closer to the real observation than the one calculated by the background, the mean and standard deviation of departure between observational and simulated brightness temperature both decreased after assimilation.(4) Both experiments can simulate the main rainfall area and period, but the results of the assimilation experiment are much better, it is not only obviously improved the position intensity and range of the strong precipitation center, which are similar with the real precipitation, but also show advantage in the threat score of precipitation grades of the storm and big storm.