【作者】 柳志慧； 王式功； 尚可政； 孔德兵； 赵文婧； 王昀；

【Author】 LIU Zhihui;WANG Shigong;SHANG Kezheng;KONG Debing;ZHAO Wenjing;WANG Yun;Lanzhou University College of Atmospheric Sciences;Meteorological Station of 95606 unit of the Chinese People’s Liberation Army;

【机构】 兰州大学大气科学学院甘肃省干旱与减灾重点实验室； 中国人民解放军95606部队气象台；

【摘要】 雷暴(冰雹)是一种常见的灾害性天气,不仅会造成人员伤亡,也会引起严重的经济损失。本文利用新疆阿克苏及邻近地区12个气象站1980~2013年雷暴(冰雹)资料,以及同期高空资料,统计了各站年均雷暴日数,对发生区域雷暴的环流形势进行分类,确定各天气型的自动识别指标。在此基础上,通过逐步回归方法,建立阿克苏及邻近地区区域雷暴概率回归预报模型,并对2013年进行试预报。结果表明:(1)区域雷暴的影响系统主要分为四种类型:巴湖低槽型、急流型、西北气流型和温度槽型。(2)各天气型自动识别指标为:巴湖低槽指数trough≥f0(f0=80);急流指数stream≥f1(f1=160),当stream不满足条件时,检验最大风速Umax≥f2(f2=20);西北气流指数airflow≥f3(f3=70);温度槽指数t≥f4(f4=1.5)。(3)利用天气型自动识别方法,对2002~2012年5~9月历史资料进行识别,样本数由1683d下降为876d,消空率为48%,漏报4d,漏报率为2.4%,对2013年5~9月历史资料进行识别,样本数由153d下降为80d,消空率为48%,无漏报。(4)基于NCEP/NCAR1°×1°再分析资料计算的物理量,筛选出合适的预报因子,通过逐步回归方法,建立阿克苏及邻近地区区域雷暴概率回归预报模型。对2002~2012年5~9月历史资料进行回代检验,平均TS评分为30.1%,对2013年5~9月进行试预报,平均TS评分为28.2%。更多还原

【Abstract】 The thunderstorm(hail) is a common severe weather, which leads to casualties and severe economic losses. Based on the thunderstorm(hail) observation data of 12 meteorological stations in Xinjiang Aksu and adjacent areas from 1980 to2013, the regional thunderstorm is defined as the day there is more than three stations with thunderstorm(hail) weather, by this definition, the regional thunderstorm concentrated in May to September. The weather circumfluence of 170 regional thunderstorms from May to September in 2002 ~ 2012 was analyzed and classified, and finally the weather situation was mainly divided into four types: The Balkhash lake low trough type, the jet type, the northwest airflow type and the temperature trough type. By constantly revise the critical value of automatic identification weather type to achieve optimal results, in the end the automatic identification indicators are determined, the indicator of Balkhash lake low trough type is trough≥f0(f0=80), the indicator of jet type is stream≥f1(f1=160) or Umax≥f2(f2=20), the indicator of northwest airflow type is airflow≥f3(f3=70), the indicator of temperature trough type is t≥f4,(f4=1.5). By this method, the result by using the data of2002 ~ 2012 from May to September showed: Samples are reduced to 876 from 1683, empty elimination rate is 48%, the omission is 4d and the non-response rate is 2.4%. The result by using the data of 2013 from May to September showed:Samples are reduced to 80 from 153, empty elimination rate is 48%, without omission. The method of automatic identification weather type has greatly reduced the sample number, and effectively improved the prediction. On this basis,based on the physical quantities by calculation using the NCEP/NCAR1° x 1° reanalysis data, to select some appropriate forecast factors, taking advantage of all samples were elected to four weather types from May to September in 2002~2012,by stepwise regression equation, the probability regression forecast model was established. Backing to generation of test with the historical data of 2002 ~ 2012 from May to September, the average TS score was 30.1%, the highest TS score was the lake low trough type, was 35.3%, the lowest TS score was the temperature trough type, was 22.0%, and the accuracy rate reached more than 65%. To try to forecast with historical data of 2013 from May to September, the average TS score was28.2%, the highest TS score was the jet type, was 33.3%, the lowest TS score was the temperature trough type, was 20.0%,and the accuracy rate reached more than 50%. The above results showed that the forecasting method in this paper can provide some technical support for the regional thunderstorm forecast in Aksu area of and adjacent areas.更多还原