A Comparison of Processing Methods for the Oklahoma Lightning Mapping Array

The Oklahoma Lightning Mapping Array (OKLMA) has consisted of 17 Very High Frequency (VHF) sensors in two separate clusters separated by roughly 150 km since 2012. This large footprint of sensors is expected to benefit the spatial coverage, sensitivity and accuracy of the network (e.g., Chmielewski & Bruning, 2016, https://doi.org/10.1002/2016jd025159; Koshak et al., 2018, https://doi.org/10.1175/jtech-d-17-0041.1). The unique configuration of sensors facilitates the evaluation of treating the two clusters as a single, unified network. Network operators may wish to do this when processing networks in close proximity or for temporary network extensions during field projects. The 0000-0600 UTC period on 16 June 2019 included millions of VHF sources and thousands of flashes from isolated storms and a mesoscale convective system. The 0300-0700 UTC period on 12 June 2019 contained a single storm of interest proceeding scattered activity. The VHF solutions and their grouped flashes from the unified OKLMA network are compared to those generated from the combination of the two clusters each operating as separate, independent networks. The unified network processing not only required the substantially longer processing time, but also resulted in lower sensitivity and more noise, suggesting that the theoretical model of high accuracy and sensitivity with a large LMA footprint may not apply to this configuration. Additionally, relatively few (< 10%) of the VHF sources observed by a single cluster of sensors matched a VHF source observed by the opposite cluster, suggesting that the sources mapped by each cluster of sensors were most often unique. Plain Language Summary The Oklahoma Lightning Mapping Array (OKLMA) can be treated as two separate networks of instruments or one large one. Theoretically, treating it as one large network, as has been done historically, should be better since the distances between stations and the large number of stations can better solve for where the lightning occurred in the cloud. We test both treating it as one large network and two separate networks during two periods of busy lightning activity. Treating the OKLMA as one large network took longer to compute and produced much worse results than treating it as the combination of two separate networks. The longer computation time was expected, but the worse results were not. In fact, it was relatively infrequent when the lightning signals recorded by each separate network were identical, which may be part of why the performance is worse because it is expected that all stations observe the same lightning signals using the historical treatment of a single large network.