The Study Of Stochastic Physics Perturbation In Storm Scale Ensemble Forecast

A weather forecast system is very sensitive to the model error. Particularly, the uncertainty in sub-grid parametrization process has the essential effect on the accuracy of weather forecast system. As the ensemble forecasting has made great progress in the medium-term numerical prediction, therefore the storm-scale introduction of ensemble prediction system technology has become a necessary choice. Due to small time scale, fast and strong nonlinear development of the storm-scale system, the traditional medium-range ensemble forecast method is obsolete. This paper focuses on the model perturbation of ensemble forecasting, introducing Stochastic Perturbed Parameterization Tendencies(SPPT) scheme, Stochastic Kinetic-Energy Backscatter(SKEB) scheme and Mixed Model Perturbation(SKEB+SPPT) scheme to the storm-scale ensemble forecast system.Based on the above consideration, using WRF model to simulate a severe convection weather process in AnHui Province on 31 May 2014.This paper evaluates the performance of ensemble forecast and analyses the characteristics of stochastic perturbation and kinetic energy evolvement.Results shows:(1)In this case, the 60 km length scale and 3 h decorrelation time scale in SPPT has the best performance both in the skill of probability prediction and probability of precipitation(POP). It is found that there is a close link between the selection of the length and time scale in SPPT and the scale of synoptic system as well as the forecast time length. Within limits, keeping the time scale as short as possible and length scale as long as possible will achieve better prediction effect.(2)The Mixed Model Perturbation scheme increases(reduces) the spread and accuracy(the forecast error) of SPPT scheme only or SKEB scheme only. In 500 hPa, 850 hPa, the spread of the surface, the temperature, the zonal wind and the water vapor mixing ratio of the Mixed Model Perturbation scheme increase dramatically, and the forecast error of them has been brought under effective control. In addition, the accuracy of forecast improves to various extent.(3)Compared to the SPPT scheme only and SKEB scheme only, the Mixed Model Perturbation scheme improves their prediction of precipitation, with the mistaking and missing forecast of precipitation remarkably reducing. The tendency and intension of the ensemble members of the Mixed Model Perturbation reflect the tendency and intension of the real precipitation better.(4)The perturbation spatial distribution of the Mixed Model Perturbation scheme is similar to that of the SPPT scheme at the beginning of forecast. As forecast time goes on,the perturbation spatial distribution is transformed and is similar to that of SKEB.(5)The kinetic energy perturbation of three model perturbation schemes shows that, in 1—24 h integration, the kinetic energy of them in different scale grows with the increase of forecast time, then approaching saturation during 18—24 h. In the prophase and metaphase of forecast, the kinetic energy perturbation of the Mixed Model Perturbation scheme is obviously bigger than that of SPPT only or SKEB only in all scales,indicating that the combination of the two stochastic perturbation schemes can efficiently complement the missing energy in different scales. At a later stage, the kinetic energy perturbation of these three model perturbation scheme inclines to unanimously, which means that the Mixed Model Perturbation scheme combines the SPPT and SKEB in a reasonable way.