Improving Short‐Term Rainfall Forecasts by Assimilating Weather Radar Reflectivity Using Additive Ensemble Perturbations

To improve short-term rainfall forecasts through direct assimilation of radar reflectivity, atmospheric variables associated with rainfall should be modified based on their correlation with reflectivity. However, it is difficult to estimate such correlations. The ensemble Kalman filter can estimate the correlation by means of ensemble forecasts, although the estimation is limited to when rainfall is forecast by at least one member at analysis points. To assimilate reflectivity effectively even at points at which no rainfall is forecast, we suggest adding ensemble reflectivity perturbations, which are correlated with atmospheric variables, before ensemble Kalman filter assimilation. In the present study, this correlation is calculated in the whole computational domain including the rainfall regions. We apply this procedure to assimilation experiments with 1-km horizontal grid interval for two tornadic supercells that occurred on 6 May 2012 and on 2 September 2013, and we succeed in improving short-term rainfall forecasts by modifying wind, temperature, and water vapor. Plain Language Summary Short-term numerical forecasts can be improved by the new assimilation procedure of weather radar reflectivity suggested in this study. With this procedure, the initial atmospheric states of simulations are corrected based on more reasonable correlation with radar reflectivity.