Diagnosing stroke, flash, and storm scale lightning variability using Lightning Differential Space

We introduce the Lightning Differential Space (LDS) framework for multiscale, data-driven characterization of cloud-to-ground (CG) lightning, in which consecutive stroke intervals are mapped into a two-dimensional space spanned by their spatial and temporal derivatives. Using Earth Networks Total Lightning Network (ENTLN) observations, we analyze CG strokes during peak lightning seasons (2020–2021) across three climatically distinct regions: the Amazon (tropics), the Eastern Mediterranean Sea (subtropics), and the northern U.S. Great Plains (mid-latitudes).

The LDS topography reveals a robust and regionally consistent “allowed” and “forbidden” zones, with dominant clusters separating intra-flash successive strokes from inter-flash intervals at thundercloud and cloud-system scales. While the overall structure is stable across regions, systematic shifts in cluster location and separability reflect contrasting convective environments, including differences in characteristic inter-event times and system-scale distances.

We further introduce a Current Ratio LDS, which projects the ratio of absolute peak currents between successive strokes onto the same stroke interval coordinates. This diagnostic acts as a statistical partitioning tool that sharply distinguishes intervals likely to contain flash-initiating strokes (where the succeeding stroke tends to be stronger) from intervals dominated by subsequent strokes within multi-stroke flashes. Across all regions, a distinct short time interval feature (< ~0.02 s) spans distances from sub-kilometer to hundreds of kilometers, suggesting rare near-simultaneous remote CG events and motivating renewed investigation of long-range thunderstorm coupling (teleconnection).

Overall, the LDS framework (combining number distribution and current ratio information) provides a scalable pathway for extracting coherent multiscale lightning behavior from large network datasets, with direct relevance for evaluating model representations of stroke and flash processes and for developing diagnostics supporting probabilistic monitoring and nowcasting.