LAUSR

Abstract To conduct accurate lightning prediction and nowcasting, it is important to clarify the relationship between electrification and microphysical properties in clouds. Using the dual-polarization Doppler radar with the hydrometeor classification algorithm and lightning observation, radar-derived microphysical properties related to lightning flash rate were investigated in terms of 10 isolated thunderstorm cases over the Kanto Plain, Japan, with 351 radar volume scans conducted every 5 min during the summer period. In this study, the ice-related particles within the 35 dBZ volume (V35IC) showed the highest correlation coefficient (r = 0.75) and the lowest normalized root mean square error (NRMSE = 8.3%) in relation to cloud-to-ground lightning (CG), and also best to intra-cloud lightning (IC, r = 0.69, NRMSE = 8.1%). The correlation between V35IC and CG flash rate was better than that of the mixed-phase (between 0 °C and − 40 °C) 35 dBZ volume (r = 0.64), without considering the microphysical particles. The second-best radar-derived microphysical property was the ice mass consisting of the radar grids of all the ice-related particles (r = 0.72). These results suggest that an effective lightning indicator in terms of a radar-derived property is the ice-related particle volume, in addition to the traditionally used graupel and hail. The ice-related particle volume in the colder region of the thunderstorm (air temperature lower than −10 °C) showed a higher correlation with CG and IC than that in the warmer region, especially in terms of IC. These results are consistent with the non-inductive charging theory that has been well accepted in previous studies. Following time-lag correlation analysis from 0 to −30 min every 5 min before thunderstorm initiation, it was found that the ice-related index of V35IC demonstrated the slowest decrease (r = 0.75–0.65 in 30 min) in the correlation coefficient with time lapse and may be suitable as a lightning nowcast indicator.