Annualized Flash Count Variations over the US from 2017 through 2023 and Their Relationship to NLDN Algorithms

We present an analysis of yearly cloud-to-ground (CG) flash counts over the continental U.S. from 2017 through 2023 using data from the U.S. National Lightning Detection Network (NLDN). The dominant feature of the data set is a roughly 28.5% decrease in CG flash counts in 2020 relative to the average over 2017 through 2019. This significant drop was followed by a gradual monotonic recovery to a 2023 count that was only about 4% lower than the 2017-2019 average. The CG flash counts in this study have all been obtained by processing the complete dataset with multiple intracloud (IC)/CG classification algorithms in order to remove the effects of various changes in the classification algorithm over the seven-year period, and to assess the impact of classification on the annualized temporal pattern. The data were fully reprocessed using two different classification algorithms: the currently operational one (became operational on 2021-05-27) and the pre-2015 method, which has been used in separate historical analysis going back prior to 2017. We show that the two classification algorithms preserve the major 2020 decrease in CG flashes and the 2021-23 recovery. In addition, we examine the effects of two different flash clustering algorithms and show that the effects of these are also much smaller than the large decrease and slow recovery in CG flashes. In addition, we investigate the effects of a simple two-parameter method of normalizing classifications using just two quantities (peak current and peak-to-zero time); that method has been recommended in past work. Finally, we use the current operational flash classification and clustering algorithms to investigate the total (IC+CG) flash counts over all seven years to see whether the large decrease in 2020 applies to total flashes or just CG flashes.