Methods to Overcome Lightning Location System Performance Limitations on Spatial and Temporal Analysis: Brazilian Case

Abstract

One of the most interesting attributes of Lightning Location Systems (LLSs) data is that they can be analyzed in several ways according to the objectives of the study. However, the quality of the data is governed by the system performance and has some limitations when analyzed at different temporal/spatial scales, and these limitations will depend on the analysis method. This work focuses on approaches to minimize the variations associated with LLS performance. In this way, specific network configurations for the Brazilian Lightning Detection Network (BLDN) were obtained through the reprocessing of selected sensor data, resulting in three distinct datasets. Each dataset was then evaluated using different procedures: trimmed flash (exclusion of low current discharges), thunderstorm days (TDs), and thunderstorm hours (THs). The comparison between TDs obtained from the LLS and TDs available from surface stations shows consistent results with a good correlation of those datasets. An 11 -yr analysis of BLDN data also shows that improvement (over time) of the system sensitivity has led to the detection of an increasing number of low peak current events. By eliminating low peak current discharges (less than 19 kA), the sensitivity variation was significantly reduced, partially "normalizing" long-term performance. TDs and THs were the most effective method to normalize temporal variations of the lightning activity, overcoming most of the network performance variations. From the spatial perspective, TDs and THs also seem to produce the most reliable lightning distribution. These results might guide long-term temporal and spatial analysis of lightning data, providing a more stable approach that is independent of system performance.

Improvements on Lightning Density Estimation Based on Analysis of Lightning Location System Performance Parameters: Brazilian Case

“…Even though not the case in this paper, the RDEM was still not able to correct for temporal variations in network performance, because of computational complexity. In this case, other techniques (refer to [7]) could be applied.…”

“…The maximum density for this case is ∼ 12 flashes·km −2 · yr −1 . It is important to mention that the corrections obtained by most RDE models are not able to compensate for variations of the network's performance over time, 5 for which the approaches described by Bourscheidt et al [7] could be applied. Thus, this implies that the methods used here and in [7] are connected and may be applied together for a more complete understanding of the system's performance effects on the spatial and temporal analysis of lightning activity.…”

“…In Brazil, for example, many attempts were made to minimize such variations and a series of relative detection efficiency models (RDEMs) were developed [4]- [6]. In addition to these models, other approaches based on thunderstorm days/thunderstorm hours are suggested more recently to overcome the spatial and temporal variation in system performance and are shown to be very efficient, especially for long-term analysis [7]. However, with those methods, one is not able to retrieve quantitative lightning density information, which again requires the use of the RDEMs.…”

“…However, with those methods, one is not able to retrieve quantitative lightning density information, which again requires the use of the RDEMs. Thus, the methods described in this paper and described by [7] are complimentary. The LA for RINDAT, also for other networks that use the same technology, is produced using confidence ellipses.…”

Thunderstorm incidence in southeastern Brazil estimated from different data sources

“…Third, also from Fig. 6a, a large spot (the region in white in the state of São Paulo) of high flash density can be seen, coincident with the location of the city of São Paulo, the largest city of the country with a population larger than 10 million people, suggesting that the urban activity is affecting the thunderstorm activity (Naccarato et al, 2003;Pinto et al, 2004;Farias et al, 2008;Bourscheidt et al, 2012). The effects of orography and urban area are not evident in the low-resolution data in Figs.…”