For this study, 80 lightning-initiated wildfire cases were analyzed in order to explore lightning flash characteristics, underlying surface conditions, and local meteorology at the time of occurrence. These observations were compared with lightning flashes from the same event days which do not ignite wildfires. Preexisting land surface and meteorological conditions were used by analyzing data from the SPoRT Land Information System (SPoRT-LIS). Model parameters examined include: 0-10 cm soil moisture, green vegetation fraction, surface soil temperature, and total column relative soil moisture content. Lightning data from the National Lightning Detection Network (NLDN) was analyzed to determine what kinds of CG strikes are likely to cause wildfires in terms of polarity and current.
Statistical analyses were performed on all collected data to demonstrate differences between wildfire starting lightning flashes and other lightning flashes which did not initiate fires. This was mainly done using the Wilcoxon-Mann-Whitney rank sum test to determine the degree of independence between each of the lightning and surface observations. Results show that 0-10 cm relative soil moisture content and green vegetation fraction were the two LIS parameters which demonstrated the most utility in identifying lightning flashes which initiate wildfires.