Simulated LMA Source and Flash Detection Efficiency Using a New Interactive Tool

Detailed source and flash detection efficiencies for Lightning Mapping Arrays (LMA) are needed for observational and climatological work with the data. Simulations of the LMAs active during the Deep Convective Clouds and Chemistry campaign (DC3) were performed with a new Monte Carlo interactive tool. As with previous simulations, it propagated emissions from a given source point, added Gaussian observed timing errors to the retrieval times at each station and used a least squared algorithm to find the best solution for the source. This simulation added the ability to account for variable receiver thresholds to restrict the stations contributing to the solution based on received power, which allowed a better examination of the overall impacts of site selection. This simulation also used a new method for approximating the observed distribution of source powers by incorporating the effects of the digitizing window. The average errors were then calculated as has been done previously but with the addition of the probability of detection for a given source location. Based on a 17 month climatology of flashes over the West Texas Lightning Mapping Array (WTLMA), the distribution of flashes by number of sources was used to relate the source detection efficiency from the simulation to the most likely flash detection efficiency.

As observed in previous theoretical and observational studies, the average errors in azimuth, range and especially altitude of the source point solutions increased with increasing distance from the center of the network, with the standard deviations of these errors highly dependent on the station configuration and noise thresholds. The source and therefore flash detection efficiency also decreased with range as expected, but its centroid was offset from the center of the WTLMA when variable, observed receiver thresholds were used instead of uniform thresholds across the network. With the observed WTLMA thresholds, 95% of flashes could be detected to approximately 150 km from the center of the network, where average altitude errors were less than 0.4 km but the standard deviation of those errors approached 3 km. This resulted in a more realistic view of LMA network performance than a simulation without consideration of actual receiver thresholds.