Application Of Ensemble Forecasts On Severe Convective Weather Forecasting

Severe convective weather forecasting has been the essential and challenging task in the operational weather forecasting,because severe convective weather often occurs at small scales,accompined with rapid evolution and great destruction.By using in situ observations,sounding data,and the analysis and forecast data of the U.S.NCEP Global Ensemble Forecast System(GEFS),this thesis aims to study the application of ensemble forecasts on severe convective weather forecasting in the middle and lower reaches of the Yangtze River during the summer of 2012-2015.The environmental factors of physical quantities favoring severe convective weather have been identified,which facilitates the recognition of the occurence and development of severe weather.Using the ingredients-based forecast methodology and GEFS ensemble forecasts,the joint probability methodology of severe convective weather has been established and optimized in the study region.Forecast verification shows that the proposed joint probability approach is reliable and effective for severe convective weather forecasting.Further case studies indicate that the approach is influenced by the quality of ensemble forecasts,in terms of the prediction ability of the objective method.Firstly,the spatial and temporal distributions of three kinds of severe convective weather,i.e.rain storm,thunder storm and lightning storm,as well as the characteristics of physical quantities have been analyzed,which demonstrates that the frequencies of severe convective weather show distinct features for one-day and ten-day periods.The high frequency of severe convective weather tends to appear during late June to middle July and in the afternoon.Severe convective weather tends to appear more frequently in the mountainous and coastal regions,indicating that convection initiation is highly associated with the terrain effect in the formation mechanism of strong convection.Moreover,the impact of environmental factors on the occurrence and development of three types of convective storms has been analyzed.For the occurrence of short-time heavy rain,moisture,dynamic and thermal conditions are indispensable,while moisture and thermal conditions play a more vital role in the occurrence of thunder storm or lightning storm.The uncertainties in the calculation of parameters related to severe convective weather are discussed.For example,the assimilation and calculation schemes of the convective available potential energy(CAPE)can result in an error up to 30%.Considering all the requirements that the ingredients-based methodology involves,the stratified conditions of forecasting factors are proposed.Forecasting factors are filtered through Fisher linear discriminant and naive Bayes classification.To predict short-term heavy rain storms,the selected factors include precipitable water,equivalent potential temperature at 500 hPa,horizontal divergence at 925 hPa,while K Index,Showalter Index(SI),equivalent potential temperature at 850 hPa are recommended to predict lightning storm.Secondly,using the framework of ingredients-based methodology and joint probability,the joint probability methodology of severe convective weather has been proposed in the middle and lower reaches of the Yangtze River during the summer,based on the probability information provided by GEFS ensemble forecasts.The methodology is optimized accouting for the contribution of forecasting factors that improves forecast skill and discriminating ability.Statistical verification scores of deterministic and probabilistic forecasts exhibit that joint probability forecasting outperforms other objective forecast indices with stably higher accuracy in forecasting short-term heavy rain and lightning storms.For the 12-36 h lead times,the average threat score(TS)is about 0.12 and 0.3 for predicting short-term heavy rain and lightning storms,respectively.The highest scores(0.77 and 0.75)of area under relative characteristic curve(AUC)are obtained for the joint probability forecasts of short-term heavy rain and lightning storms,showing good discriminating ability among all objective forecasting methods.Finally,the severe convective weather event on 4-7 July,2013 has been investigated to analyze the forecast performance of GEFS.In general,GEFS can provide accurate forecasts of synoptic-scale fields,and the ensemble mean can reflect the evolution of the weather system.However,GEFS is under dispersive with much small ensemble spread,which leads to biased forecasts.For instance,geopotential height forecast at 500 hPa shows an obviously negative systematic error compared to the analysis field,which degrades the forecast quality.In this case,the joint probability forecast is superior to other objective forecasting methods.However,there still exists some weakness,such as over forecasts,the lack of forecasting capability of localized strong convection and the convective systems over heterogeneous terrains.The genesis,development,evolution and propagation of synoptic systems are also not ideal.Especially,due to low resolution and imperfect physical processes,the wind field and the divergence field are not well predicted in GEFS,which leads to the positional deviation and under dispersion in the forecasts of severe convective weather events.