A Hemispheric Analysis of Aerosol Particle-Lightning Relationships

Abstract

The relationship between predicted aerosol mass concentrations from NASA’s Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2) and observed lightning from the World Wide Lightning Location Network (WWLLN) and NASA’s Geostationary Lightning Mapper (GLM) are investigated from the years 2010 to 2021 and 2018 to 2021 respectively. Two-dimensional convective available potential energy (√CAPE)-precipitation histograms are made, allowing for the comparison of lightning in high and low aerosol regimes on a pixel-by-pixel basis, which showed a possible saturation effect of an aerosol impact on lightning. When controlling for thermodynamic factors, an increase in fine aerosol concentrations (PM2.5) is generally linked with an increase in lightning stroke density, while an increase in coarse aerosol concentrations is associated with different responses amongst the two lightning products. A regional analysis shows positive enhancements with increased fine aerosol concentrations as well, while an increase in coarse aerosol concentrations is associated with different responses depending on the region. A convolutional neural network (CNN) is used to develop a better parameterization for lightning stroke density using √CAPE, precipitation, aerosol concentrations and other meteorological variables. Preliminary results show a general underestimation of lightning stroke density by the CNN, regardless of the combination of variables used. The inclusion of predicted aerosol concentrations to the CNN appears to be less important than the inclusion of thermodynamic variables.