An Application of Rank-Order Statistics to the Study of Hurricane Variability

Abstract

A method is presented for quantifying the temporal variability found in hurricanes. The method is based upon several well-known nonparametric tests that use ranking techniques; among these are tests devised by Kendall, Kruskal-Wallis and Jonckheere. The tests area used in an attempt to answer the following questions: 1) How variable are selected meteorological elements in time? 2) Are there significant differences among profiles? 3) Are there differences in variability between the interior and exterior portions of profiles? 4) Do the selected elements exhibit a trend in time? Data taken from Hurricane Debbie (1969) on two days are processed and a computational example is presented.