Statistical analysis of tropical cyclone days in the Western North Pacific under the influence of ENSO and IPO

Abstract

This study investigates the statistical characteristics of tropical cyclone (TC) days in the Western North Pacific (WNP) from 1970 to 2023, with a focus on their distributional behavior and modulation by large-scale climate variability. Quantile–quantile (Q–Q) plots indicate that both average and total TC days follow positively skewed distributions, with the Gamma distribution providing a better fit than the Gaussian distribution, particularly in representing extreme events. We use bootstrap resampling to estimate confidence intervals and assess how the probability of various TC duration categories changes under different ENSO phases. Results show a marked increase in the occurrence of long-lived TCs (≥10 days) during strong El Niño events, especially in the southeastern WNP (Region IV), whereas strong La Niña years are associated with reduced long-duration activity. These probability shifts are consistent with observed variations in TC track density and environmental anomalies favorable to TC development during El Niño years.To further disentangle the influences of ENSO and the Interdecadal Pacific Oscillation (IPO), partial correlation analysis was applied. The results indicate a significant positive relationship between TC days and ENSO, independent of IPO effects, while no significant correlation was found between TC days and the IPO. This finding suggests that ENSO is the primary driver of interannual variability in TC duration across the WNP. Overall, the use of the Gamma distribution provides a robust framework for modeling TC activity, and the results underscore the critical role of ENSO in shaping TC behavior, with implications for forecasting and regional climate risk assessment.