Advantage of bulk lightning models for predicting lightning frequency over Japan

Abstract This study examined the performance of an explicit bulk lightning model coupled with a meteorological model for forecasting lightning by numerical weather prediction over Japan. The evaluation was conducted by comparing the lightning predicted by the explicit bulk lightning model, diagnosed empirically by the numerical model, and observed by ground base measurements. From the results, the bulk lightning model performed better in terms of lightning frequency than did the diagnostic scheme, which overestimated the lightning frequency, although there were no appreciable differences in the score of each method for the geographical distribution and time correlation compared with the observations. These results suggest that the explicit bulk lightning model is advantageous for predicting lightning frequency. The sensitivity of the simulated lightning to the choice of cloud microphysical model was also examined by using a two-moment and a one-moment bulk microphysical scheme. Sensitivity experiments on the choice of microphysical model indicated that the two-moment bulk scheme reproduced the observed lightning well, while the one-moment bulk scheme overestimated the lightning frequency. Analyses suggested that the overestimation of the lightning in the one-moment bulk scheme originated from active charge separation by riming electrification, in which graupel was produced more frequently and was assumed to fall faster. These results suggest that the explicit bulk lightning model with the two-moment bulk microphysical scheme offers an alternative to conventional lightning prediction methods. Graphical abstract