宿州雷暴气候特征及其灾变G ANN预测模型研究

根据宿州市5个气象站1957-2008年雷暴日观测资料,利用数理统计方法分析了宿州雷暴的时空分布规律,以及雷暴与降水、温度的关系。结果显示：宿州属于多雷区,南部多于北部,南北相差3.21-4.65天;年际变化大,年累计雷暴日最多相差37天,初终雷及无雷期年际间振荡的幅度较为剧烈,52年来累计雷暴日线性减少的趋势明显,减少幅度为1.61-2.89天/10a;季分布以夏季最多,冬季最少;月分布呈单峰型,雷暴多集中在4-9月,以7月最多,12月最少;日分布呈单峰型,以16：00-18：00频率最大;多年平均月雷暴日数序列与相应的气温、降水量之间呈显著正相关。为了进一步预测雷暴的长期演变趋势,以埇桥为例建立了雷暴多发年份灰色人工神经网络组合预测模型,预测下一个雷暴多发年将发生在2025年。

Characteristics of Thunderstorm Weather and G ANN Predicting Model Based on the Thunderstorm Disaster Variation in Suzhou

According to the data of thunderstorm days of five weather stations in Suzhou from 1957 to 2008, the time space distribution of thunderstorms and the relationship of thunderstorms to rainfall and temperature in Suzhou are analyzed. The results show that Suzhou is an area attacked frequently by Lightning, and more lightning activities happen in the south part than in the north, with maximum difference being 3.21 to 4.65 days. There is great interannual variability of thunderstorm weather in Suzhou, and the annual cumulative number of thunderstorm days is up to 37 days. There is a linear decreasing trend in thunderstorm days with the range being 1.61 to 2.89 days per 10 years. Thunderstorm days exhibit a single peak type seasonal distribution, occur mostly in summer and least in winter, and centralized in the period from April to September, with maximum in July and minimum in December. The daily distribution is also the single peak type, with the maximum frequency from 16:00 to 18:00. The long range mean series of average thunderstorm days has significantly positive correlation with rainfall and temperature. In order to predict the trend of thunderstorms, taking Yongqiao as an example, the grey BP combined model is built, and the results show that the year of 2025 is the next year of thunderstorms occurring most frequently.