Assessment of Cloud Lightning Detection by the U.S. National Lightning Detection Network using Video and Lightning Mapping Array Observations

The U.S. National Lightning Detection Network™ (NLDN) underwent an upgrade aimed at increasing the detection efficiency for intra-cloud (IC) lightning flashes starting in early 2013. In order to evaluate this upgrade, we employed observations obtained during field programs carried out near a wind farm in Kansas (central U.S.) during the summers of 2012 and 2013. These observations included optical measurements obtained using two standard-speed automatically-triggered video cameras employed during the 2012 and 2013 storm seasons, and Very High Frequency (VHF) mapping system data from a short-baseline Lightning Mapping Array (LMA) available during the 2013 storm season. Time and location of cloud-to-ground (CG) strokes and IC pulses reported by the NLDN, as well as lightning magnetic field waveforms from nearby NLDN sensors, were used in our analysis. Using the standard-speed video data as ground truth, we found that the NLDN detected 40.7% (134 out of 329) of IC flashes in 2013 with a 92.5% (124 out of 134 flashes) classification (IC versus CG) accuracy. In 2012 (prior to the upgrade) the IC flash detection efficiency was 28.6% (28 out of 98 flashes detected) and the classification accuracy was 89.3% (25 out of 28 flashes classified accurately as IC). In addition to using the video data to identify lightning flash type (IC versus CG), the LMA data were used to classify flashes as IC or CG for cases where the video data provided ambiguous flash type identification. This analysis excludes a period during late May through early June 2013 when a network-wide communication problem lowered the detection efficiency of IC flashes. Interestingly, CG flash DE remained above 95% during this period.

The VHF LMA data provides a description of the spatial and temporal evolution of a flash. For flashes identified as IC discharges, we examined whether there was any correlation between the temporal and spatial behaviors of the LMA-reported VHF sources associated with a flash and low frequency IC pulses reported by the NLDN. Our results support the findings of previous studies that most of the reported cloud pulses occur during the initial breakdown period of a flash when channels (as defined by VHF sources) extend from the main negative to the upper positive charge region. Following the initial breakdown period, additional vertical channels may develop between the main negative and upper positive charge regions (as seen from VHF sources) during which the NLDN also reports some IC pulses. Finally, at later stages in a flash, the NLDN may report pulses with no associated VHF sources, indicating no new vertical channel development. These NLDN events may reflect currents flowing through pre-existing channels that were formed earlier in the flash. Additionally, we found that flashes with longer horizontal extent have a higher chance of being reported by the NLDN. In this report, we will summarize our findings to date, and discuss our continued work in this area to explore the relationship between other features of IC flashes observed at VHF and NLDN IC pulse detection. Candidate features include the maximum height, spatial distribution, flash duration, and the maximum vertical velocity of the initial upward leaders.