基于雷电定位系统监测数据的雷暴云趋势预测

Nowcasting of the Thunderstorm Trend Based on Data Collected by Lightning Location System

以往的雷电预警研究无法给出特定区域落雷密度以及落雷频数随着雷暴生消演变的规律,而这是影响输电线路实时雷击跳闸概率的关键因素,对于输电线路雷击跳闸预警研究具有重要意义。首先将观测的目标区域按0.01×0.01划分为等面积网格,将每15 min雷电定位系统测量落雷数据按时间等分为3段;通过对每段落雷数据聚类分析剔除数据中的杂散点;在聚类云团识别的基础上,线性拟合3个连续时间段每个聚类云团质心位置,预测未来15 min内任意时间聚类云团质心移动规律;统计分析该时间段每个网格落雷密度并利用反距离加权插值法对所有网格在该时间落雷密度进行计算,获得聚类云团覆盖区域;最后,利用海南电网雷电定位系统历史数据对该算法进行验证。结果表明,该算法在15 min内聚类云团质心预测结果相对误差在20%以内,落雷预测区域与实际落雷区域基本吻合。

Previous studies of lightning prewarning failed to give the changes of the cloud-to-ground（CG） flash density and the CG frequency with the evolution and extinction of a thunderstorm in the concerned area, which had great influence in the real-time lightning trip probability study, with important significance in transmission line lightning trip prewarning. Firstly, the target area was divided into equal-area grids by 0.01°～ 0.01°, and then the data measured by lightning location system in every 15 minutes were divided into three sections, in each section, cluster analysis methods were used to exclude spurious points, at the basis of Cluster recognition, the cluster centroid motion model was built through linear fit and the cluster centroid for each cluster centroid of three successive time periods was predicted, and the thunderstorm cover range was learned in statistical analysis of each grid flash density at that time, while calculating the flash density of other grids with inverse distance weight approach; Finally, using historical data of Hainan power grid lightning location system, the algorithm was proved to be effective. The result shows that the relative error of cluster group centroid prediction within the future 15min is less than 20%, and a basic agreement between the actual lightning range and prediction range is validated.