The Application Of Radar And Satellite Data Assimilation In A South China Rainstorm Forecast

With the development of numerical weather prediction(NWP),data assimilation(DA)has become a key method to improve the quality of initial fields.DA can effectively integrate the information of background fields and observation data to obtain optimal model initial conditions.At present,variational assimilation is still the mainstream DA scheme applied both at home and abroad.Radar and satellite data as the most widely used irregular observations which have high spatial and temporal resolution are very suitable to improve the initial fields of Short-term NWP.However,domestically studies on the application of assimilating radial velocity,reflectivity,and satellite observations are inadequate.In this paper,we use ARW-WRF high-resolution regional model and GSI gridpoint statistical interpolation to build a set of operational prediction systems,adopting 3D-Var method to achieve DA of Doppler Weather Radar(DWR)reflectivity and velocity,Himawari-8 meteorological satellite AHI data.Regarding problems of data application and forecasting effect of unconventional data,multiple sets of comparative experiments were designed to perform simulation analysis and assessment of DA results based on a case of severe precipitation process in southern China.Here are the main conclusions.During the heavy precipitation period in the coastal areas of Guangdong on 17-18 July 2017,the system can grasp the scope and intensity of precipitation and the simulation of radar reflectivity.With the help of assimilating radar and satellite data,the short-term precipitation prediction also shows improvement.Particularly,the assimilation of satellite data has better ability to analyze cloud-water materials and weather process on the sea significantly.DA of conventional data,radial velocity,reflectivity,and satellite data can all improve simulation tactics such as temperature,wind vector,humidity,and precipitation.However,their prediction results are not absolutely superior to the test for DA of certain types of data.For the prediction of precipitation,assimilation radial velocity data is better than the other two kinds of data in forecasting the magnitude of 50-100 mm precipitation within 12 h,but for the magnitude rating of over 100 mm precipitation,assimilating radar reflectivity and conventional data are better.And for 12 to 36 h rainfall prediction,the assimilation of Himawari-8 satellite data have more advantages.For the 12 h accumulated precipitation,the forecast results for each group of tests are mainly on the west coast of the Pearl River estuary,and there is no description of most parts of eastern Guangdong.For the prediction on temperature,humidity,and wind fields,compared with the control group,the trend of the time and height for each group is quite consistent.It should be noted that the assimilation tests of radial velocity and reflectivity play a major role in the 0-12 h.For 24-36 h,changes on every field are limited.Overall,the assimilation test with all data added at the same time performs relatively smoothly.Although the tests in each group seldom took the top spot,they were generally higher than the results without assimilation of any data and conventional observations,indicating that the mutual corrections of various data are beneficial to simulation of this precipitation process.Since precipitation in Southern China is greatly affected by circulatory systems such as typhoon and monsoon,its development mechanism is always complicated.By calculating cumulative precipitation and the mean error(ME)and root mean square error(RMSE)of several factors,the prediction improvement of each group on temperature,humidity,or wind is more consistent than the prediction improvement of precipitation.