Application of the 2σ Lightning Jump Algorithm Based on DBSCAN Cluster

Abstract

A DBSCAN (density-based spatial clustering of applications with noise) cluster of lightning data is proposed as the substitute for radar products to solve the problem of beam blockage in radar observation and the delay of radar products in service operations. Two lightning data, BJTLS (Beijing Total Lightning System) and upgraded National Lightning Positioning Network (DDW1), are used and the 2σ lightning jump algorithm is applied to perform severe weather nowcasting on 4 June and 12 June in 2022. The nowcasting effects of the strong convective cell identification method and the DBSCAN clustering method are further compared and analyzed. Based on a determined search radius (R) for neighboring lightning data and a determined minimum number of location results (number of minimum points) in R, the DBSCAN's clustering effect on lightning location data corresponds well with the strong convective radar echo. The ideal parameter combinations for BJTLS, R is 0.05, number of minimum points is 5; and for DDW1 data R is 0.22 and number of minimum points is 3. The results show that both methods and two kinds of data could effectively be used in severe weather nowcast. For BJTLS data, the effects of two methods are equivalent. The probability of detection, false alarm rate, critical success index and lead time of two methods are 100% and 100%, 11.9% and 13.3%, 88.1% and 86.7%, 38.9 min and 42.8 min, respectively. The 2σ lightning jump algorithm can be applied for nowcasting with lightning data, reducing the dependence on radar products. For DDW1 lightning data, compared with the identification method, the start time of the clustering method delays, leading to missing alarms. Since the flash rate of the DDW1 lightning data is low, there will be more missed cases if the flash rate threshold is set to trigger the lightning jump. But without the threshold, there will be more false alarms in operation. Therefore, BJTLS data is more suitable than DDW1 data for applying the 2σ lightning jump algorithm in the service operation. The detection efficiency of BJTLS in Beijing is high and it is necessary to further improve the detection efficiency of DDW1. In conclusion, the DBSCAN clustering method provides a new idea for the service operation of the 2σ lightning jump algorithm.

Keywords

• lightning location
• lightning jump
• severe weather
• nowcasting