Spatial Modeling of Single Cell Cloud-to-Ground Lightning in Gwinnett County, GA

Though the meteorological phenomena of cloud-to-ground (CG) lightning is well studied, the spatial distribution of repeated storm events is less examined. This GIS-based study investigates the spatial variation of CG lightning from summer air mass (single-cell) thunderstorms in Gwinnett county, Georgia. Gwinnett county has previously been identified as a hotspot for repeated lightning activity. The dataset includes flashes from May-September 1995-2008, with spatial and temporal information. This study compares the frequency of flashes against 2010 and 2000 Census tract population data; DEM topography; road density data; interstate density; impervious surface density; 2001, 2005, and 2008 land cover; E-W grid number; and 1999 landuse data as explanatory variables. Where multicollinearity exists, only the most correlative explanatory variable is chosen. The analysis is repeated for 1 km2, 4km2, and 90 km2 grids. Pointwise correlation (G-function) provides evidence for clustering of lightning events using data over the entire period and for individual years. The amount of data as well as the grain of the explanatory variables justifies examining the data on a gridded level. Results from the initial 90 km2 grid indicates road density is a significant variable for all years (p<0.005), as well as each summer's flash individually (p<0.05). The significance of the E-W grid number indicates including spatial autocorrelation will improve the regression model. This spatial variation of lightning demonstrates possible effects of urban development in storm electrification, on a level easier to measure than air quality assessments.