基于EEMD分解的电场时序差分在雷电预警中的可行性分析

Feasibility Analysis on Electric Field Second Order Difference Based on EEMD Decomposition in Thunderstorm Warning

为了解决大气电场数据预报雷暴虚警率高的问题,将集成经验模态分解（EEMD）方法和二阶差分法结合应用于大气电场资料的分析,提出了一种雷电预警分析方法。该方法先用EEMD分解出晴天天气和雷暴天气大气电场的不同时间尺度变化分量,然后对包含雷电信号的高频模态分量IMF1（本征模态函数）进行二阶差分分析。晴天无雷暴发生时,地面大气电场的差分值集中在-0.5-0.5 k V/m3之间;雷暴过程中,差分大气电场出现剧烈变化,雷暴发生前,IMF1二阶差分量的增幅会明显变大,所对应的电场频率在0.016 5-0.045 5 Hz之间跳跃。经过仿真试验,结合雷达回波资料进行验证,得到雷电探测概率（probability of detection,POD）为85.1%,预警平均时间为30.2 min。

A lightning warning analysis method is presented,in which the Ensemble Empirical Mode Decomposition（ EEMD） method are the second order finite difference method are applied to the analysis of atmospheric electric field data,in order to solving the high false alarm rate of the thunderstorm warning by using atmospheric electric field data. In this method,the variscaled components of the atmospheric electric field in fair weather and thunderstorm are decomposed and the high frequency of IMF1（ intrinsic mode function） which contains the lightning signal is analyzed by the second order finite difference method. In fair weather,temporal difference electric data on the ground is concentrated on- 0. 5 - 0. 5 k V / m3. The temporal difference electric data in thunderstorm weather will increase significantly,before cloud-to-ground lighting occurring,the second order temporal difference electric data will increase sharply or the frequency of IMF1 will jump between 0. 016 5 - 0. 045 5 Hz. According to the simulation results combined with radar echo data for validation,the detection probability of warning is 85. 1% and the average time of warning is 30. 2 min.