A nonparametric test for change in variability using a proxy record with an application to ENSO

Abstract

A common problem in climate science is determining whether the pattern of variability in a particular process or variable has changed over time. When the modern observational record is short, this problem can be addressed by comparing its variability to that of an historical proxy record. In doing so, it is important to recognize that the statistical properties of the modern and proxy records are different. Here, a nonparametric test for a change in variability in this situation is described that accounts for this difference. The method is illustrated by testing for a change in ENSO variability using a record of an ENSO index over the period 1871–2007 and oxygen isotope records extracted from corals at Palmyra Island that cover four different periods spanning the past millennium. The results are mixed.