High resolution simulations with the Collaborative Model for Multiscale Atmospheric Simulation (COMMAS) with a new two-moment microphysics scheme have reproduced the basic morphological and electrical evolution of the observed storm. Model precipitation initiates as raindrops, which freeze into graupel and lead to strong electrification and lightning. The early noninductive charge separation graupel and ice crystals in the temperature range of -5 to -15 C placed positive charge on graupel, producing a main negative charge region and lower positive charge region as inferred from the lightning observations. A positive charge screening layer also developed at the top of the cloud by ion attachment. The Hallett-Mossop ice multiplication process is the main source of ice crystals at this relatively high temperature range. Prediction of ice number concentration is crucial in tracking the size and growth of crystals. Lightning initiated between the main negative and lower positive regions at altitudes close to the observed initiation heights. Later in the simulation, and graupel higher in the storm gains negative charge noniductively to enhance the main negative charge and create an upper positive charge region, creating a normal tripole structure with associated intracloud flashes. A very similar charge structure was indicated by both charge analysis of lightning and a balloon sounding through the storm near the end of its lightning activity. A decrease in lightning source height can also be seen in the upper part of the storm as it dissipates and graupel falls out of the storm, supporting the observational inference of graupel as the negative charge carrier in that region.
One feature of the storm that has not been adequately explained is how it produced six negative cloud-to-ground lightning flashes in the first 8 minutes of lightning activity (determined from both 3-dimensional lightning mapping data and National Lightning Detection Network observations). The lightning-determined charge structure was therefore a dipole configuration with positive charge below negative charge. NLDN data suggest that the CG flashes were weak, with peak currents of -6 to -15 kA and 1 to 3 return strokes. The first intracloud flash was the seventh flash in the storm and was the first indication by lightning of an upper positive charge region. In the simulations, flashes that initiate between the lower charge region mostly remain intracloud flashes, whereas in the observed storm only a few did not contact ground at some point. This discrepancy is a point of ongoing research involving the simulated charge structure, charge sedimentation, and the physics of the lightning parameterization.