Comparison of thunderstorm characteristics between Changdu and Wuhan in China based on spatio-temporal clustering of LLS data

This paper proposes a method of thunderstorm clustering, identification and tracking based on lightning location data. The clustering and identification of thunderstorm is realized using density-based spatial clustering of applications with noise (DBSCAN). And thunderstorm tracking is realized by using moving speed threshold and thunderstorm cluster merging or splitting judgment. The thunderstorm characteristics of Changdu $(31.14^{\circ}N, 97.19^{\circ}E$, 3.36 km above sea level) and Wuhan $(30.60^{\circ}N, 114.31^{\circ}E$, 0.02 km above sea level) in China were compared by this method. Five parameters were analyzed, including the duration, area, flash frequency, median current amplitude and movement velocity of the thunderstorm cluster. The results show that: (1) the region of thunderstorm cluster obtained by this method coincides with the radar echo above 30dBZ, which can well represent the development trend of the thunderstorm cluster; (2) The duration time, area and flash frequency of thunderstorms in Wuhan were significantly higher than that in Changdu. The maximum area of thunderstorms in Wuhan was almost 6 times that in Changdu, while the positive lightning current amplitude was less than that in Changdu. The duration of Type A thunderstorms organized by convective cells or multi-cell storms in the same area is slightly longer than that of type B thunderstorms with little or no movement velocity, and type A thunderstorms also produce higher flash frequency.