Interdecadal Pacific Oscillation reconstructed from trans-Pacific tree rings: 1350–2004 CE.

The interdecadal Pacific Oscillation (IPO) represents the decadal mode of the El Niño-Southern Oscillation phenomenon. As such the IPO is one of the dominant modes of decadal climate variability on both sides of the Pacific Ocean basin. For this paper we utilized a newly developed tree-ring data network comprised of five multi-centennial Vietnamese cypress ring-width chronologies that range from around 11°N–23°N latitude. We combined these data with an average of tree-ring derived drought indices from the North American Drought Atlas over the south central USA, from a box that spans from 27.5°N–35.0°N, 85.0°–110.0°W and contains 55 gridpoints and more than 100 tree ring site chronologies, these 2 locations exhibit rainfall variations that are strongly, negatively correlated with each other and are representative of the influence of the IPO on terrestrial rainfall. The final reconstruction model, weighted most heavily on the three most southerly of the five Vietnamese cypress records, spans from 1350 to 2004, and explains nearly 57% of the variance in the original IPO data for the 5-month season of October–February. The reconstruction model passes all standard statistical tests using a split calibration-verification scheme. We reveal 15 positive and 15 negative phase shifts of the IPO prior to the period of instrumentation, suggesting that the IPO has been active for at least the past seven centuries with varying degrees of intensity. We compare our reconstruction with two related millennial records: the MacDonald and Case (Geophys Res Lett 32(8):L08703, 2005) tree ring-derived reconstruction of the Pacific Decadal Oscillation, and an ice core-derived reconstruction of the IPO from the Law Dome Ice core in Antarctica by Vance et al. (Geophys Res Lett 33(6):L06712, 2015). While there is good general agreement with the latter record, there are three key periods of the past where the two records are out of phase, and we explore the reasons for this disparity. Direct comparison with the related Tripole Index (TPI) shows weaker correlation, likely owing to the stronger relationship between our tree ring data and the equatorial Pacific region relative to the north and south regions of the Pacific that combine to comprise the TPI calculation. [ABSTRACT FROM AUTHOR]