A Preliminary Study On The Assimilation And Forecasting Scheme Of Regional Ensemble Forecasting In The Pre-flood Season In South China

With the occurrence of extreme weather more and more frequently,a single deterministic forecast can not meet the needs of small and medium scale extreme weather forecast in South China.In this paper,based on the conventional observation data such as surface sounding,satellite observation data,using the GSIV3.5(Gridpoint Statistical Interpolation Version 3.5)assimilation system And WRFV4.0(Weather Research Forecast Version 4.0)mesoscale numerical model,the NCEP(The National Centers for Environmental Prediction)global ensemble forecast system(GEFS)data are used to to achieve the regional ensemble forecast system(REFS)of data in the pre-flood season in South China through dynamic downscaling method,the multi-physical parameterization scheme combination and different assimilation schemes.The 32-day precipitation period in South China from May 15 th to June15 th,2019 was selected to carry out two sensitivity tests for regional ensemble forecasting:(1)Realization schemes of reginal ensemble forecast in difference way: the direct dynamic downscaling reginal ensemble forecast(REFS?SINGLE)and the multi-physical process parameterization scheme combination reginal ensemble forecast(REFS?MULTI);(2)Comparison of difference assimilation schemes: the hybrid assimilation test(Hybrid),the three-dimensional variational assimilation test(3DVar),the Ensemble Kalman Filter assimilation test(EnKF)and the control test without assimilation(Ctrl).The test results show that:(1)The regional ensemble forecast precipitation score and disturbance energy development obtained through the dynamic downscaling and the combination of multi-physical process parameterization schemes are better than the global ensemble forecast: The relationship between the ensemble spread of REFS precipitation and forecast error is better than that of GEFS.After 48 hours of integration,the perturbation energy of the REFS?MULTI and the REFS?SINGLE were 4.7 times and 6.3 times that of the GEFS,respectively.The higher the precipitation level,the better the TS score of the REFS is better than the GEFS;the REFS?MULTI is better than the REFS?SINGLE.In the 32-day test,,the AUC value of REFS is greater than that of GEFS in 28 days,and that of REFS?MULTI is greater than that of REFS?SINGLE in 22 days.It can be seen that the prediction skill of REFS is better than that of GEFS,and that of REFS?MULTI is better than that of REFS?SINGLE.(2)The analysis filed obtained by Hybrid is better than that by 3DVar and EnKF.In the prediction of wind field,temperature field and humidity field,the prediction error of Hybrid is less than 3DVar and EnKF in the early prediction period,but in the later period,the prediction error of 3DVar and EnKF is better than that of Ctrl and Hybrid.Similarly,as the development of the ensemble disturbance energy,Hybrid and Ctrl are better than 3DVar and EnKF in the early forecast,and 3DVar and EnKF are better than Hybrid and Ctrl in the middle and later forecast.From the 24-hour cumulative precipitation score,the assimilation test is better than the Ctrl.3DVar and EnKF are better than Hybrid,and 3DVar has a better score for the heavy and medium rain level,while EnKF has a better score for the heavy rain level and above.For the aggregate statistical test analysis,the AUC value of assimilation test is greater than that of Ctrl,and the AUC value of 3DVar is the largest in the range of 10 mm ? 100 mm cumulative precipitation,followed by EnKF.For the 125 mm precipitation threshold,the AUC value of EnKF is the best,followed by 3DVAR.The AUC of Hybrid is less than 3DVar and EnKF in each precipitation threshold.