Is Convective Initiation of Weak Synoptically Forced and Pulse Thunderstorms Dependent on Land Cover Classification?

Many studies have sought to investigate the relationship between land usage and convection; however, most of these studies focus on long term climate variability as a result of land cover change. We investigated the relationship between land usage and mesoscale convective initiation in an effort to aid operational efforts in forecasting “popcorn” convection. We used observational data obtained from the Multi-Radar/Multi-Sensor (MRMS) System to identify areas of convective activity and created an algorithm in MATLAB to identify mesoscale convection not generated from outflow boundaries. We also obtained a subset of data from National Land Cover Database (NLCD) 2019 Land Cover (CONUS) to identify land usage across our study domain. We re-gridded the NLCD data to match the MRMS data and used Sobel-Feldman operators to identify boundaries between one or more land cover classifications. From there, we used contingency table-based statistics to determine independence between the presence of a land cover boundary and the development of storms over the same grid point. Results indicate that convection is overall dependent on land usage with cultivated cropland boundaries representing the largest deviation from expected number of storms in Central Kentucky.