The authors demonstrate how to construct a network from a time series of U.S. hurricane counts and show how it can be used to identify unusual years in the record. The network links years based on a “line-of-sight” visibility algorithm applied to the time series plot and is physically related to the variation of hurricanes from one year to the next. The node degree is the number of links connected to a node. The authors find that the distribution of node degree is consistent with a random Poisson process. High hurricane-occurrence years that are surrounded by years with few hurricanes have many linkages. Of the environmental conditions known to affect coastal hurricane activity, they find years with little sunspot activity during September (peak month of the hurricane season) best correspond with the unusually high linkage years.
(a) Time series of U.S. hurricane counts from 1851–1870 with visibility lines drawn connecting the bars. (b) Network constructed from the visibility lines. The method is based on the work of Lacasa et al. .J. B. Elsner, T. H. Jagger and E. A. Fogarty. 2009. Visibility Network of United StatesHurricanes. Geophysical Research Letters, VOL. 36, L16702, doi:10.1029/2009GL039129. August.