By using the Advanced Regional Prediction System (ARPS) and Ensemble Square Root Filter (EnSRF) method with flow dependent background error covariance, through assimilating multiple Doppler radar data, the strong convective weather occurred on June 23, 2013 was analyzed. Firstly, the composite radar reflectivity in assimilation tests and observation were compared to verify assimilation effects, moreover, by calculating the Root Mean Square Error (RMSE) and spread, the assimilation results were further evaluated quantitatively. Secondly, the impacts of EnSRF radar assimilation on thermal, dynamical, humidity and microphysical variables were investigated. Finally, the ensemble mean was simulated at a high resolution of 1 km. Results show that EnSRF can assimilate convective system similar to the observations and suppress the false echo. The RMSE of radial velocity and reflectivity reduce significantly. The assimilation experiment has larger range of radar reflectivity in vertical direction and has weaker intensity than that of the control experiment. In the convective region, the strength of surface perturbation potential temperature can decrease by 6 K and the relative humidity can increase by 30%. The rain, ice and snow mixing ratios increase significantly in the convective area. The simulation with the final analysis results can better simulate the severe storm.