Application Research Of Convection-allowing Ensemble Forecasts Based On Neighborhood In The Squall Line

As one of the research hotspots of numerical weather prediction,convectionallowing ensemble forecast is significant and valuable in improving forecasting accuracy of severe convective weather.This paper firstly conducted a convectionallowing ensemble forecast experiment for a squall line process based on the traditional Breeding Growth Mode(BGM)method,and analyzed the forecast results for the application of the classic initial perturbation method.Then,combined with the strongly local characteristics of convective-scale weather systems,the post-processing and initial perturbation methods of convection-allowing ensemble forecast are improved around the neighborhood idea.The evaluation of precipitation forecasts and the comparison of ensemble forecasts are carried out respectively.The major conclusions are as follows:Firstly,the probability matched mean(PMM)method is introduced to overcome the disadvantage of traditional ensemble mean(EM)method as EM smooths out the forecast values at high magnitudes and reduces the forecasting capability for severe convective weather.And ensemble forecast's influence on the simulation of structural characteristics of a squall line,deterministic precipitation forecast and scoring verification are researched.The results of precipitation forecast and scoring verification show that the forecasting accuracy of ensemble forecast is higher than that of control forecast.The EM method is more accurate at light rainfall events,and the PMM method has obvious advantages in forecasting techniques of heavy rainfall events.As an improved ensemble mean method,the PMM method can not only preserve EM's better precipitation distribution,but also avoid its smoothing and underestimation to the extreme precipitation.According to the analysis of structural characteristics of the squall line system,compared with the control forecast,the intensity of cold pool obtained by the ensemble forecast increases,the range and intensity of thunderstorm high pressure at the rear of the squall line increases,the band structure of hydrometeor is clearer,the wind field is more organized,and the interaction between the near-surface cold pool and the vertical wind shear in the squall line system can be simulated,which have obvious improvement on the structural characteristics of the squall line.Secondly,the current neighborhood probability(NP)method mainly considers the spatial displacement error in high-resolution precipitation forecasts,but the problem of the forecast time exceeding or lagging behind the observed field has not been properly solved.Therefore,a temporal factor is introduced into the NP method,and precipitation probability forecasts are evaluated in different spatial-temporal neighborhoods based on the improved NP method and the fractions skill score(FSS),combined with the relative operating characteristic(ROC)curve.The results indicate that a spatial scale of 15?45 km neighborhood radius can effectively rectify the displacement error of precipitation and optimize the forecast effect for such squall line processes.And the spatial scale of 15?30 km exhibits a better forecasting capability for smaller-scale extreme precipitation events.Within the temporal neighborhood that makes sense,the larger the temporal scale,the better the precipitation probability forecast effect for light rainfall.And the forecasting skill of heavy rainfall increases with the decrease of temporal scale.Temporal scale and rainfall intensity in convective-scale precipitation forecast are inter-related,and the temporal uncertainty of different magnitudes of precipitation can be captured by different temporal scales.Meanwhile,the spatial and temporal scales have interdependent influences on precipitation forecast effect.The improved NP method can simultaneously reflect the spatial and temporal uncertainties of convective-scale precipitation forecasts in the high-resolution model,achieving a comprehensive assessment of spatial-temporal neighborhoods and providing probabilistic forecast results that match different magnitudes of precipitation.Finally,given the strongly local nature of convective-scale weather systems,the neighborhood idea is introduced into the traditional BGM method to obtain the Local Breeding Growth Mode(LBGM)method.And the comparative experiment and applicability test of convection-allowing ensemble forecast are carried out.The LBGM method considers the local interaction of the convective-scale weather system in the perturbation growth process,so that the perturbations after cultivation are featured with more local characteristics.The LBGM method results in the perturbation distribution exhibiting characteristics more evident of flow dependence than the BGM method,and an initial perturbation with greater definite kinetic significance is derived.Information entropy can well measure the amount of local information contained in the perturbation distribution,verifying the fact that the LBGM method,compared with the BGM method,can increase the local information associated with the initial perturbation.With regard to the ensemble spread,the perturbation physical quantities obtained by the LBGM method are larger than the BGM method both in the breeding and forecasting phase,thereby solving the problem of insufficient ensemble spread obtained by the BGM method.In the meanwhile,the forecast root mean square error of each perturbation physical quantity can be further reduced,thereby improving the forecast effect of ensemble forecast to some extent.It can be seen from the structural characteristics and precipitation forecast that the results from the LBGM method are closer to actual situation than from the BGM method.The applicability test of another squall line shows that the LBGM method can better forecast the morphology distribution of radar echoes and precipitation,initially verifying the applicability of the new LBGM method in convection-allowing ensemble forecast.This provides a reference for further development of initial perturbation methods applied to convection-allowing ensemble forecast.