Study On Groud-Based Radar Data Assimilation Through Ensemble Kalman Filter On Landing Typhoon

In recent decades,the forecast error of typhoon track has been decreasing year by year while the progress of its intensity and precipitation is very slow,which is partly due to the relative low skill of numerical model in typhoon prediction.Due to the strong non-linearity of the typhoon's evolution,the small deviation of the initial field is easy to be doubled in the integration process,which makes the forecast result deviate from the actual weather condition.The data assimilation process combines the model background field and the observation data based on certain mathematical theory to obtain the analysis field,which can provide the accurate forecasting initial condition for the numerical model and improve the model forecasting ability to a certain extent.The ground-based Doppler radar can detect the fine structure of the landing typhoon kernel.It has large potential to improve the numerical prediction of landing typhoon in China assimilating high resolution Doppler radar data.Chinese ground-based Doppler radar data assimilation through Ensemble Kalman Filter in typhoon research still need more study,which is a very meaningful task.PSU-WRF-EnKF data assimilation system developed by Pennsylvania State University is used in this study.Assimilation effect was tested in three landing typhoon cases,which is Mujigae(1522),Meranti(1010)and Rammasun(1409),respectively.It is found that cycling radar data assimilation can significantly improve the typhoon path,strength,structure and precipitation simulation.By cycling data assimilation,the typhoon position of the analysis field is close to the observation,which makes the track error of typhoon landing point less than 10 km.Three typhoons on average,After assimilating radar data by about 8h,the typhoon track error reduce apparently.The averaged track error of three typhoon cases in assimilation window can reduce by 40 km to 20 km.After the assimilation path error can be less than 10 km.The typhoon intensity can be reduced to less than 10 hPa,and the typhoon can be significantly enhanced by the radar data assimilation.In the typhoon Mujigae case study,300 hPa warm core of Mujigae is significantly strengthened after data assimilation as well as typhoon eye contraction.Mujigae's convection asymmetric structure is closer to observation compared to reflactivity observation.Assimilation increments of 300 hPa potential temperature and 850 hPa vorticity are gradually concentrated in the typhoon kernel with increasing cycling.The data cycling assimilation improves the prediction of typhoon precipitation to a certain extent,and the more time of assimilation data,the more the precipitation forecast TS score increases.The radar data are divided into three parts and assimilated individually in sensitivity experiment according distance to typhoon center,which is located in 0-100 km,100-200 km and more than 200 km from the typhoon center.Only assimilating radar data in 0-100 km region can get the similar improvement in track,intensity and structure of Mujigae compared to all data assimilated.The improvement is not very apparently in sensitive experiment which only assimilating data in 100-200 km or outside 200 km.Track and intensity error can be reduced to less than 5km and 5hPa with several assimilation cycle of data within 100 km,indicating that the core data is the key to correct the model background.The number of data in 0-100 km is usually only 50% of the total number(in accordance with the typhoon case).Only assimilate this part of the data can reduce computing time by half compared to assimilating all data with similar assimilation improvement.