A statistical lightning model: comparing global and regional detection networks over southern africa

Abstract

The World Wide Lightning Location Network (WWLLN) provides global coverage of lightning activity in near real time using a distributed array of Very Low Frequency (VLF) radio receivers. Although WWLLN achieves superior spatial coverage relative to regional detection systems, this advantage comes at the cost of comparatively low detection efficiency. This study documents a statistical technique for obtaining realistic flash rate densities from WWLLN data by employing satellite based lightning observations from the Lightning Imaging Sensor (LIS) and the Optical Transient Detector (OTD) as a reference standard. The WWLLN stroke rate densities were first converted to flash rate densities using an assumed multiplicity of 3.5 strokes per flash and then normalised against the merged LIS/OTD climatology through spatially varying seasonal correction factors computed on a 15° by 15° grid. The normalised WWLLN flash rate densities were validated against data from the South African Lightning Detection Network (SALDN), a regional system with high detection efficiency within the borders of South Africa. A chi squared (χ2) goodness of fit test was applied across multiple days during peak thunderstorm season (February 2007), and in every case the null hypothesis of equivalent spatial distributions could not be rejected (p = 1.00). The results confirm that the normalisation procedure successfully recovers realistic flash rate densities from WWLLN data, enabling the production of high temporal resolution global lightning maps that are comparable to satellite derived climatologies.