Assessment of Processes Contributing to Multi-Day Statistical Prediction of Bomb Cyclones around Japan

Abstract This study investigates the relative importance of three potential predictors for the statistical prediction of bomb cyclone occurrence near Japan with a two-day lead time. The three predictors are a zonal wave train north of the Himalayas, a zonal wave train south of the Himalayas, and low sea level pressure (SLP) over southeast China, identified through lag-composite analyses based on bomb cyclone tracks provided by Kyushu University for 1996 − 2022. Logistic regression, which showed the best performance among three prediction methods examined, was used to assess the relative importance of the predictors. Prediction experiments were conducted using all three predictors, combinations of two predictors, and each predictor individually. The northern Himalayan wave train is the most important predictor of bomb cyclone occurrence near Japan, while the southern wave train and southeast China SLP provide additional predictive skill. The highest skill was achieved when all three predictors were included. Excluding either the southern wave train or southeastern China SLP resulted in only a slight decrease in prediction skill, whereas excluding the northern wave train substantially reduced the skill. These results clearly demonstrate the crucial role of the northern wave train in predicting bomb cyclone occurrence near Japan.