The Deep Convective Clouds and Chemistry (DC3) experiment seeks to understand the kinematic and microphysical controls on the lightning behavior of deep moist convection. This study utilized multiple dual-polarization Doppler radars across northern Alabama to quantify microphysical and kinematic properties and processes that often serve as precursors to lightning, such as the graupel echo volume, graupel mass, and convective updraft volume. The focus here was on one multicellular complex that occurred on 21 May 2012 in northern Alabama during DC3. The graupel echo volume and the graupel mass in the charging region correlated well with the total lightning flash rate (FR), and even better than the updraft volumes and maximum updraft velocities. The flash length scales (LS) and flash areas were generally anticorrelated to the FR, while it was correlated to the nonprecipitation ice volume. More specifically, the presence of smaller flashes was associated with a stronger lower positive charge region caused by larger graupel volumes, stronger updraft volumes, and stronger maximum updraft velocities while larger flashes occurred during lower FRs and were associated with a weakened lower positive charge region in combination with a stronger upper positive charge region, weaker updraft velocities, a smaller graupel volume and mass, and an increase in nonprecipitation ice volume.

Lightning flash rate parameterizations based on polarimetric and multi-Doppler radar inferred microphysical (e.g., graupel volume, graupel mass, 35 dBZ volume) and kinematic (e.g., updraft volume, maximum updraft velocity) parameters have important applications in atmospheric science. Although past studies have established relations between flash rate and storm parameters, their expected performance in a variety of storm and flash rate conditions is uncertain due to sample limitations. Radar network and lightning mapping array observations over Alabama of a large and diverse sample of 33 storms are input to hydrometeor identification, vertical velocity retrieval and flash rate algorithms to develop and test flash rate relations. When applied to this sample, prior flash rate linear relations result in larger errors overall, including often much larger bias (both over- and under-estimation) and root mean square errors compared to the new linear relations. At low flash rates, the new flash rate relations based on kinematic parameters have larger errors compared to those based on microphysical ones. Sensitivity of error to the functional form (e.g., zero or non-zero intercept) is also tested. When considering all factors (e.g., low errors including at low flash rate, consistency with past linear relations, and insensitivity to functional form), the flash rate parameterization based on graupel volume has the best overall performance.