Abstract Recently, detailed spatio‐temporal analysis utilizing X‐band multi‐parameter radar‐derived 3D volume scan and total lightning data in Japan, have revealed the peak in‐cloud (IC) lightning occurs 5–10 min prior to maximum ground rainfall in individual thunderstorm cells during heavy rainfall events. This temporal relation holds the potential for improving real‐time monitoring and nowcasting of torrential rain causing natural disaster such as flash floods. However, cell interactions such as merging and splitting of the cells are often observed in multicellular thunderstorms, which cause abrupt fluctuation in the total lightning rate and make it difficult to track the storm evolution. To address this, we propose an area‐correction method based on the lightning activity (in terms of cell area ratio before and after merging/splitting) of individual cell. Preliminary observation results of two multicellular thunderstorm events with heavy rainfall (∼100 mm/hr) that exhibited cell merging and splitting during their life cycle are demonstrated. The area‐correction resulted in a reasonable and corresponding trend in the temporal behavior of IC lightning rate and ground precipitation volume, with the peak IC rate preceding the peak ground PV by 5–10 min. We also demonstrate a promising approach for short‐term prediction of ground rainfall with high accuracy (correlation coefficient between observed and predicted PV was 0.84–0.94), by means of a moving linear regression model, using recent IC observations. These results highlight the potential of total lightning for short‐term rainfall prediction in both isolated and multicellular thunderstorms.
Mondal et al. (2026) studied this question.