Research On Variational Assimilation Technique Of Radar Data Under Nudging Constraints

Many studies have shown that using three-dimensional variational(3DVar)to assimilate radar data can improve the precipitation forecast of severe convective weather.However,some studies have found that assimilating radar observation data into numerical prediction model may produce spurious precipitation and large errors in the location and magnitude of precipitation.One possible reason for this problem is that the cloud analysis scheme used in assimilating radar reflectivity data overestimates the humidity in the cloud.Another reason is the lack of proper balance in the dynamical and microphysical fields after assimilating radar data with 3DVar.In addition,the selection of background error covariance(BE)greatly affects the effect of radar data assimilation.To this end,based on the WRF(Weather Research and Forecasting Model)and its3DVar(WRFDA)system,this paper firstly calculates the BE with the stream function?and the nonequilibrium velocity potential?_uas the momentum control variables and the BE with the horizontal wind UV as the momentum control variables,respectively,using the NMC method.The effects of two groups of BE matrixes assimilating radar radial wind and reflectivity data on a dispersive convective precipitation forecast are compared.The results show that the difference of characteristic length scale between the two groups of control variables is mainly reflected in the difference of momentum control variables.After assimilating the real radar observations,the BE with??_uas momentum control variable produces a much larger influence range of wind field analysis increment than the BE with UV as momentum control variable,which produces a more localized analysis increment.The use of BE assimilated radar data with UV as the momentum control variable provides more improvement in the forecast of dispersive convective precipitation and significantly reduces the spurious precipitation forecast by BE assimilated radar data with??_uas the momentum control variable.Then,the impact of radar data assimilation frequency on the short-term precipitation forecast of a severe convective is explored.The sensitivity of 15 min and1 h interval assimilation frequency is tested,and the results show that higher radar data assimilation frequency(15 min interval)can improve the skill score,but it leads to greater precipitation overprediction bias.This may be due to the fact that the indirect radar reflectivity assimilation scheme overestimates the humidity in the convective region,and this wet bias gradually accumulates in the rapid cycle assimilation process,which eventually affects the effectiveness of precipitation forecasting.A diagnosis of the analysis fields shows that the higher-frequency cyclic assimilation produces excessive wind speed and water vapor convergence at low levels and exaggerated upward movement.Then,the large-scale constraint is imposed on the regional model by the spectral nudging technique,which nudging Global Forecast System(GFS)forecast field data during the forecast periods.The results demonstrate that the application of spectral nudging could significantly reduce the positional deviation of the precipitation forecast and the magnitude of overpredicted precipitation.Spectral nudging can effectively adjust large-scale circulation fields,thereby improving the conditions of water vapor convergence.Moreover,spectral nudging also improves the forecasts of surface variables such as wind,temperature,and humidity.Finally,the paper designs a nudging multi-time 3DVar analysis fields method,that is,using the observation nudging method to assimilate the analysis fields of multi-time 3DVar assimilated radar data.This method is applied to the assimilation experiments of two summer convective precipitation cases in Jiangsu and Anhui Province,and compared with the cyclic assimilation results of 3DVar only,the improvement of this method on convective precipitation forecasting is investigated.The results show that the range of water vapor convergence and vertical movement rising in the initial field generated by 3DVar is obviously larger,which leads to spurious and overpredicted precipitation.Compared with the 3DVar method,the root mean square error of horizontal wind,temperature and humidity in the initial field of the nudging multi-time 3DVar analysis fields method is significantly lower,and the water vapor convergence and vertical movement rising area are more matched with the radar observed convection.Thus,improving the prediction of radar echoes and precipitation,reducing the false precipitation,and alleviating the overestimation of precipitation.In particular,the improvement is more obvious for the forecast of heavy precipitation.