Using paleoclimate reconstructions to analyse hydrological epochs associated with Pacific decadal variability

Abstract

The duration of dry or wet hydrological epochs (run lengths) associated with positive or negative Inter-decadal Pacific Oscillation (IPO) or Pacific Decadal Oscillation (PDO) phases, termed Pacific decadal variability (PDV), is an essential statistical property for understanding, assessing and managing hydroclimatic risk. Numerous IPO and PDO paleoclimate reconstructions provide a valuable opportunity to study the statistical signatures of PDV, including run lengths. However, disparities exist between these reconstructions, making it problematic to determine which reconstruction(s) to use to investigate pre-instrumental PDV and run length. Variability and persistence on centennial scales are also present in some millennium-long reconstructions, making consistent run length extraction difficult. Thus, a robust method to extract meaningful and consistent run lengths from multiple reconstructions is required. In this study, a dynamic threshold framework to account for centennial trends in PDV reconstructions is proposed. The dynamic threshold framework is shown to extract meaningful run length information from multiple reconstructions. Two hydrologically important aspects of the statistical signatures associated with the PDV are explored: (i) whether persistence (i.e. run lengths) during positive epochs is different to persistence during negative epochs and (ii) whether the reconstructed run lengths have been stationary during the past millennium. Results suggest that there is no significant difference between run lengths in positive and negative phases of PDV and that it is more likely than not that the PDV run length has been non-stationary in the past millennium. This raises concerns about whether variability seen in the instrumental record (the last ∼100 years), or even in the shorter 300–400-year paleoclimate reconstructions, is representative of the full range of variability.