Evaluation Of The Forecast For The Southwest Vortex By GRAPES-REPS And Improve Method On ETKF Initial Perturbation Scheme

The southwest vortex is a meso-scale weather system with important influence on the precipitation weather in China.At present,there are many researches on the development of the southwest vortex,or the three-dimensional structure,the temperature and humidity structure,but there are few studies on the pathway of southwest vortex.China has established a regional ensemble forecasting system based on GRAPES-Meso model.However,there still lack an identification method of southwest vortex for grid analysis data.It is not clear that what would the model predict on the mesoscale weather systems such as the southwest vortex,and yet there hasn't constructed a reasonable initial perturbation for the uncertainty in the southwest vortex ensemble forecast.In order to systematically study the prediction ability of GRAPES Regional Ensemble Predict System(GRAPES-REPS)to the southwest vortex,firstly this paper designs an identification method of southwest vortex which is suitable for high resolution grid analysis data,and the rationality and validity of the method are verified through a series of tests.Using this method to evaluate GRAPES regional ensemble forecast on the southwest vortex path prediction.Besides,the effect of GRAPES-REPS on the southwest vortex was also evaluated by analyzing the southwest vortex intensity,precipitation and so on.Since the southwest vortex is a shallow weather system in middle and low level,this paper will increase the vertical levels of observation space in the ETKF initial disturbance scheme to solve the problem of insufficient observation information in medium and low levels.And the five tests of different observation variables for calculate the projection matrix are designed,this paper attempts to improve the prediction technique of GRAPES regional ensemble forecast on the southwest vortex by improving the ETKF initial perturbation.Get the following main conclusions:The recognition method designed in this paper(HVW method)can effectively identify the southwest vortex process in high precision grid data,the coincidence rate with the grid point of the actual is 87.5%,reduce the false rate of subjective analysis of grid date.Compared with the weather chart,the hit rate is close to 80%,which fully demonstrates that the method can effectively capture the southwest vortex,the effectiveness and rationality of the HVW method in southwest vortex are verified.With the numerical prediction gradually turned to refine the forecast,the massive data emerges,the application of this method can greatly reduce the forecaster's analysis workload.The evaluation of prediction on southwest vortex by GRAPES-REPS shows: accuracy of GRAPES on the southwest vortex keeps a high score,and the false rate is slightly larger than the missing rate.And the forecast for occurrence and development of the southwest vortex by GRAPES-REPS performs well,most ensemble members can predict these processes,but the occurring time is earlier than observation.The forecasted moving path for southwest vortex by GRAPES-REPS in 24 h is reasonable,and the ensemble average forecast is much better than control member,the path would turn north after 24 h in the eastward moving cases.Besides,the forecast for intensity of the southwest vortex is too strong,which behaves as higher center positive vorticity and lower potential height.In 24 h prediction,the rainfall area match the observation very well,after 24 h,since the pathway of the eastward moving southwest vortexes turn north,the prediction of strong rainfall area turn the same.The results of the improvement experiment of ETKF initial disturbance scheme show that,increasing the vertical levels of observation space in ETKF initial disturbance scheme can significantly improve the forecasting effect.In addition,the results of the five tests of different observation variables combination shows that as the variables involved in the calculation increase,the time required for ? to stabilize is longer and the final state of the amplification factor is smaller.The analysis of the root mean square error and the spread of the variables in the whole simulation region shows that the combination of Test-1(U,V)is optimal and Test-2 is the suboptimal scheme.The analysis of the path prediction of a typical southwest vortex process shows that increasing the observation variables of the projection matrix makes the forecast of the center of the southwest vortex more concentrated and the error is further reduced.The analysis of the precipitation shows that the increase the vertical levels in the observation space(Test-1)has improved the northward of the rainfall zone.Increasing the projection matrix observation variables will further increase the probability of precipitation in central Guizhou,but also can lead large area of probability on precipitation forecast lies north.Combined with path prediction and precipitation drop forecast,Test-5 has the smallest error of path prediction and the maximum value of probability of precipitation prediction.Test-1 and Test-2 are slightly inferior to Test-5,and the prediction of precipitation are better than Test-5.