Surprisingly Different Outcomes from Two Different Processing Methodologies for the Oklahoma Lightning Mapping Array

Abstract:

A Lightning Mapping Array typically consists of several Very High Frequency (VHF) sensors spaced 10s of kilometers apart in a single cluster to triangulate VHF sources emitted by lightning channels in space and time, as the Oklahoma Lightning Mapping Array (OKLMA) was when it was installed in 2003. In 2013 a second cluster of sensors was added to the OKLMA roughly 150 km away from the initial sensors. Ideally, in addition to increasing the spatial coverage of a network, such spacing of sensors also improves network sensitivity and accuracy, especially of VHF sources observed by a mixture of sensors (Chmielewski and Bruning 2016, Koshak et al. 2018). The unique configuration of the OKLMA provides an opportunity to examine whether separated clusters perform as expected in one network. The 0000-0600 UTC period on 16 June 2019 included millions of VHF sources and thousands of flashes from both an isolated storm and a quasi-linear convective system which passed through the OKLMA domain and produced a broad range of flash sizes. Triangulating VHF sources from two separate, independent networks using the standard LMA processing technique was expected to result in lower accuracy and sensitivity than allowing all sensors from both clusters to triangulate the best VHF source solutions, especially in the central OKLMA area where sensors from each cluster should theoretically observe the same VHF sources. However, the opposite was observed. Additionally, very few (<2%) of the VHF sources observed by either cluster of sensors when treated independently could be identified as matching a VHF source observed by the opposite cluster of stations, suggesting that the idealized model of accuracy and sensitivity does not apply well to the OKLMA. This resulted in not only substantially longer processing times as expected but also lower sensitivity and more noisy solutions when both clusters were treated as part of one larger network. We will detail some of the differences resulting from the different processing techniques and how this affects the resolved flashes as these storms evolved and moved across the domain.