The Initial Application And Advancement Of The Breeding Growth Mode Method In The Convective-permitting Ensemble Forecast

The convective-permitting ensemble forecast is regarded as an research of major significance.Meanwhile,there are also some difficulties.This paper firstly conducted a convective-permitting ensemble forecast experiment based on the Breeding Growth Mode(BGM)method and studied some key problems,which need to be solved in order to develop the convective-permitting ensemble forecast.Based on the experimental results and considering the local distribution and independence of perturbation,a new generation method of initial perturbation is put forward,which is named as the local BGM method.Then a new convective-permitting ensemble forecast experiment is conducted based on the local BGM method.The major conclusions are as followings.Firstly,the development rule and the effect of initial perturbation are studied in the convective-permitting ensemble forecast.It can be concluded that there is only 24 hours for the perturbation to reach nonlinear saturation in the convective-permitting ensemble forecast according to the analysis of growth rate,coefficient of association and growth mode of perturbed weather parameters.From the aspect of the distribution of perturbation pattern,the perturbation pattern will have significant change with the flow field on the lower level after saturating,while it is more stable on the higher level.Moreover,the perturbation can reflect the horizontal and vertical distribution of the uncertainty in atmosphere.It is reasonable to use 10-20 members to form the ensemble in the convective-permitting ensemble forecast.When the ensemble size increases from 1 member to 10 members,the forecast Root Mean Square Error(RMSE)will decrease rapidly.When the ensemble size increases from 10 members to 20 members,the forecast RMSE will decrease continually with slower rate.However,it will no longer decrease after the ensemble size of 20 members.There is little difference between the average spread of the ensemble sample of 10 members and 40 members,which means the ensemble spread cannot be improved through the increase of ensemble size.The Fractions Skill Score(FSS)shows that forecast result of precipitation of the ensemble of 10-20 members is better than that of 30-40 members.Secondly,some new optimization and evaluation method,including Probability Matched Mean(PMM),Neighborhood Ensemble Probability(NEP)and FSS,are introduced to overcome the disadvantage of traditional methods to generate ensemble average or probability forecast on the extreme or heavy precipitation.The experiment results show that the PMM method can overcome the disadvantage of traditional ensemble average and avoid the smoothing and underestimation to the extreme precipitation,which is suitable for the heavy rainfall.The ensemble average can well reflect the precipitation centers with the amount lighter,while the precipitation amount of the PMM method is a little heavier than the observation.It will generate better forecast result to use the ensemble average considering the light precipitation and the PMM method is more suitable than the former on the heavy precipitation level.These two methods can be applied together to provide guidance for operational forecast.The NEP method can reflect the mutual influence among grids on the domain to remedy the disadvantage of high-resolution forecast model on the grid scale.The FSS can transform the observation precipitation to continual fraction,which is more accurate to evaluate the forecast effect.Finally,the Local Breeding Growth Mode(LBGM)method,as a new method to generate initial perturbation,is put forward in this paper.The local distribution of perturbation is considered in this new method and the influence radius parameter is introduced.The growth of perturbation will be rescaled through calculating the ratio of the forecast RMSE of grids within the influence radius to that of the whole grids in the calculation domain.The results show that it can be observed significant local characteristics for the perturbation generated by the local BGM method.Furthermore,it can be concluded that the local BGM method can improve the ensemble spread and reduce the forecast RMSE effectively through the analysis of 5 perturbation and near-surface parameters respectively.At the same time,the ensemble spread will gradually increase and tend to be stable with the increase of influence radius.From the aspect of the ratio of ensemble spread to forecast RMSE,the local BGM method can improve the performance of ensemble forecast system.At the end,the evolvement of wind filed,precipitation tendency and maximum reflectivity is studied.The result shows that the local BGM method can generate the advantageous atmospheric environment for the development of severe convective weather and well forecast the evolvement of the squall-line and precipitation tendency.Generally speaking,as a new thinking of conducting ensemble forecast,the LBGM method may more suitable for the convective-permitting ensemble forecast with high resolution.