利用贵阳单站对流参数的强降水潜势预报方法研究  

Heavy rainfall potential forecast method with convective parameter of the single station in Guiyang

作　　者：易志学 兰世怀 

机构地区：[1]贵州省雷山县气象局,贵州雷山557100 [2]南京信息工程大学,江苏南京210044

出　　处：《贵州气象》2015年第1期1-8,共8页Journal of Guizhou Meteorology

基　　金：贵州省青年科技基金黔气科合QN[2011]13号.

摘　　要：该文利用2002-2011年10 a的常规观测资料,筛选出贵阳单站24 h强降水、12 h强降水及6h强降水个例.对大气温湿类、层结稳定度、热力及能量类物理指标与强降水进行相关性统计分析,确立了相关性较好的常用物理指标（IQ、K、mK、SWEAT、Tg）.通过分析降水前后其物理指标的变化特征,并利用成功指数建立指标预警阈值.再利用逐步消空法和BP神经网络集成方法对指标进行集成,得到相应的预报模型,从而建立了贵阳强对流天气未来0～6h,0～12h的潜势预报的预警预报指标,对贵阳地区降雨预报具有一定的指示作用.分析表明,利用逐步消空法建立的预报模型要优于BP神经网络集成的模型.The heavy rainfall cases of 24 hours, the heavy rainfall cases of 12 hours, and the heavy rainfall ca- ses of 6 hours, occurring in Urumqi a station, were selected from the conventional observation datum from 2002 to 2011. By statistics and analyzing the correlation of the atmospheric temperature and humidity, stratification stabili- ty, thermal and energy physical parameter and the heavy precipitation, the result show that the correlation with the physical parameters （IQ, K, mK, SWEAT, Tg） and heavy precipitation is better. Above analyzing characteristics of physical indexes before the precipitation breaking out, occurring and after rain, the threshold value of forecasting parameters having been confirming were set with the idea of opportunity index. By using the idea of stepwise de- creasing far method and the BP neural network integration, this physical parameter heaving been selecting are inte- grated , so the forecasting models are formed. Thus obtaining the Urumqi early warning parameters of strong convec- tive weather forecast of potential future 6 and 12 hours, it has a certain indication role on the forecasting of rainfall Urumqi area. Through the contrast analyzing, the forecasting model to establish with the stepwise decreasing far method is superior to the integration of the BP neural network model.

关 键 词：强降水 对流参数 预警阈值 逐步消空法 BP神经网络 预报集成 

分 类 号：P456[天文地球—大气科学及气象学]