Direct Assimilation Of Radar Reflectivity Data For The Simulation Of A Landfalling Typhoon

At present,there are few studies on direct assimilation of radar reflectivity factor for tropical cyclones.Semi-empirical method such as the so-called complex cloud analysis scheme is often used to assimilate radar reflectivity indirectly into the operational models.How to use advanced data assimilation methods to effectively assimilate the high-resolution radar reflectivity data and to obtain high-precision model initial fields to improve the intensity,structure and precipitation prediction for landfalling typhoons has become a frontier topic.In this study,using a convective scale WRF-GSI system(with horizontal resolution 3 km)and a reflectivity observation operator based on double-moment microphysics(Thompson)scheme,simulated radar reflectivity data are produced and then directly assimilated with EnKF every 12 minutes for 1 hour through Observing System Simulation Experiments(OSSEs)for the case of typhoon In-Fa(2021).By comparing these OSSEs with the control experiment without assimilating radar reflectivity data,we found that: Assimilation of reflectivity data has a slight improvement in the simulation of typhoon track,but obvious improvement in intensity and precipitation forecast,the positive impact lasts for 10-15 hours.Through the cycled assimilation,the final analysis can provide more accurate initial field with temperature field,vertical velocity and hydrometeor distribution closer to the “truth”.The background and analysis errors during the assimilation cycles are analyzed and the ensemble-based correlations between reflectivity and all analysis variables are examined.The results shows that the correlations between reflectivity and variables directly related to reflectivity operator(mixing ratios for rainwater,ice,graupel and number concentration of rainwater)are the highest,resulting in the largest and most effective adjust on these variables.Among the other variables indirectly related to reflectivity operator,temperature and vertical velocity(T and W)have also the obvious correlations with reflectivity,and can be adjusted effectively through the reliable background error covariance of EnKF.Moreover,sensitivity experiments are conducted to evaluate the impact of updating different state variables and using different localization-length.The results show that updating hydrometers only could adjust hydrometers well,but the positive impact of assimilation vanishes quickly during the forecast.Updating a full set of analysis variables and choosing an optimal localization-length are helpful to improve the numerical forecast ability for landfalling typhoon.