The Rainstorm Ensemble Forecast Based On BGM And Physical Perturbation

The rainstorms are frequently occured in our country.Currently most researchers in our country prefer to use the single numerical prediction method to forecast the rainstorms.However this prediction method has a great uncertainty,especially for the rainfall intensity and areas.Now the ensemble forecasting technology has already made great success in the systematic scales,but it still has some problems.For example, we haven’t grasp the structure of initial error exhausivly, so the perturbation methods need to be improved. In the past,most scholars focused their attentions on just one forecast method, without reasonably matching different methods by fully considering the respects of advantages and disadvantages. In this paper I performed a series of ensemble prediction tests of rainstorm on the base of previous studies.The main conclusions are as follows:(1) By analyzing the evolution of the error varying with time, we find that the moist convection system is a necessary condition to influence the growth and propagation of perturbation errors.In the spatial structure,the perturbation errors of potential temperature and wind fields have large values in which wet convection systems exists.The perturbation errors spread by time. Thus, the areas where perturbation errors rapidly grow lies in wet convection systems and large values of CAPE.The stratification instability and humidity has a decisive influence on wet convection system.The vertical wind shear and instability of environment condition has a significant influence on the intensity and path of rainstorm.(2) The initial disturbance design:The original Breeding Growth Modes get the high energy errors via implanting random perturbations at the initial time.Since the divergence is an important factor to the ensemble forecast, I considered whether can I enlarge the divergence through appropriate implanting spatial and temporal variation to BGM.This new method can not only promote the breeding,but also enlarge the divergence between the ensemble members under the same circumstance.Based on this purpose,I designed a rolling BGM and a regional BGM.The rolling BGM combine the time-log method with BGM.It can fully utilize the historical observation information.The regional BGM preformed perturbations with different intensity on the convection region.these two methods have increased comprehensive score than BGM.The regional BGM has a more obvious effect.(3) The model perturbation design:Due to the small divergence by perturbing the initial field alone,I added model perturbations in which I perturbed the tend of physical disturbance.Unlike the traditional parameters perturbation method,the new method has a great stability and reliability.I combined this method with regional BGM,which can improve the forecast skill of control forecast.This new method can get a greater improvement than perturb initial field alone,indicating that combine model perturbation with initial perturbation is feasible.