AOSC Departmental Seminar
March 10, 2016

Temporal Clustering of Tropical Cyclone Occurrence

Patrick Harr
National Science Foundation & Naval Postgraduate School

Over the western North Pacific Ocean during June through October, a tropical cyclone occurs once every 6 days on average.  However, there are extended periods of inactivity and extended periods of heightened activity.  Several physical mechanism have been linked with periods activity and inactivity.  These mechanisms vary in spatial and temporal scales that include the global scale and slowly-varying Madden Julian Oscillation (MJO), synoptic-scale Kelvin waves, and localized effects from pre-existing tropical cyclones.  While these factors may impact the daily probability of tropical cyclone formation, their contribution to extended periods of inactivity or activity in a statistically-significant context of temporal clustering of tropical cyclone occurrence has not been examined in detail.

In this study, a temporal cluster index is based on a measure of the temporal variability between successive tropical cyclone formations and the significance of the index is based on the degree of departure from an underlying uniform multinomial distribution.  This clustering algorithm is capable of examining the cyclic occurrence of a phenomenon.  Instances of statistically significant temporal clustering are then examined with respect to physical mechanisms such as the MJO and Kelvin waves to determine the amount of variance in temporal clustering that could be explained by the various physical mechanisms.