35Journal of Oceanography, Vol. 58, pp. 35 to 44, 2002ReviewKeywords:⋅ Regime shift,⋅ climate impacts,⋅ PDO,⋅ IPO,⋅ NPO,⋅ fishery oceanogra-phy.* Corresponding author. E-mail: [email protected] © The Oceanographic Society of Japan.The Pacific Decadal OscillationNATHAN J. MANTUA1* and STEVEN R. HARE21University of Washington, Joint Institute for the Study of the Atmosphere and Oceans, Seattle, WA 98195-4235, U.S.A.2International Pacific Halibut Commission, P.O. Box 95009, Seattle, WA 98195-2009, U.S.A.(Received 19 May 2001; in revised form 16 August 2001; accepted 16 August 2001)The Pacific Decadal Oscillation (PDO) has been described by some as a long-lived ElNiño-like pattern of Pacific climate variability, and by others as a blend of two some-times independent modes having distinct spatial and temporal characteristics of NorthPacific sea surface temperature (SST) variability. A growing body of evidence high-lights a strong tendency for PDO impacts in the Southern Hemisphere, with impor-tant surface climate anomalies over the mid-latitude South Pacific Ocean, Australiaand South America. Several independent studies find evidence for just two full PDOcycles in the past century: “cool” PDO regimes prevailed from 1890–1924 and againfrom 1947–1976, while “warm” PDO regimes dominated from 1925–1946 and from1977 through (at least) the mid-1990’s. Interdecadal changes in Pacific climate havewidespread impacts on natural systems, including water resources in the Americasand many marine fisheries in the North Pacific. Tree-ring and Pacific coral basedclimate reconstructions suggest that PDO variations—at a range of varying timescales—can be traced back to at least 1600, although there are important differencesbetween different proxy reconstructions. While 20th Century PDO fluctuations weremost energetic in two general periodicities—one from 15-to-25 years, and the otherfrom 50-to-70 years—the mechanisms causing PDO variability remain unclear. Todate, there is little in the way of observational evidence to support a mid-latitudecoupled air-sea interaction for PDO, though there are several well-understood mecha-nisms that promote multi-year persistence in North Pacific upper ocean temperatureanomalies.years. This situation, which originated with a stronglyanomalous winter in 1976–1977, has been termed a “cli-matic regime”, following a regime shift in 1977. The 1977change in Pacific climate was first reported by Nitta andYamada (1989) and Trenberth (1990), who described astep-like shift in the mean state of winter sea level pres-sure (SLP) in the North Pacific. Miller et al. (1994) pro-vided the first detailed depiction of the climatic changesand dubbed the 1976/77 North Pacific event a regime shift.Biologists noted dramatic late-1970’s changes inmuch of the biota around the North Pacific. Ebbesmeyeret al. (1991) quantified the change in 40 “environmen-tal” (climatic and biological) variables demonstrating astatistically significant step between 1976 and 1977 in acomposite of the time series. It was observations on Pa-cific salmon, however, specifically the catch history ofPacific salmon going back 70 years, that provided the most1. IntroductionClimate records from around the Pacific Basin con-tain evidence for strong interannual to interdecadal vari-ability, in special cases with remarkably large-scales(O(104 km)) of spatial coherence (NRC, 1998). El Niño/Southern Oscillation (ENSO) has long been known to bethe prominent source for hemispheric-scale interannualclimate variations for the Pacific and the global tropics(Rasmussen and Wallace, 1983). In the last two decadesof the 20th Century, the extratropical Pacific Ocean wasin an almost continuous El Niño-like state despite theabsence of tropical El Niño events in a majority of those36 N. J. Mantua and S. R. Haretantalizing evidence that a definite link existed betweeninterdecadal changes in North Pacific climate and NorthPacific fisheries. In a series of papers, Francis and Harefocused on Alaska salmon production and its link to cli-mate (Francis and Hare, 1994; Hare and Francis, 1995;Francis and Hare, 1997), arguing that Alaska salmon pro-duction was best characterized as alternating regimes,where the transition from one regime to another wasabrupt.The race to describe and understand interdecadalchanges in the Pacific accelerated through the 1990’s.Latif and Barnett (1996) provided a comparison of thelow-frequency variability in observations with that in theoutput from a coupled ocean/atmosphere model simula-tion, and proposed a mechanism for Pacific Decadal Vari-ability (PDV) with a near-20 year periodicity. Zhang etal. (1997) offered a series of analyses teasing apart sub-tle spatial differences between Pacific climate variabilityat interannual versus interdecadal time scales. Mantua etal. (1997) capitalized on the maturity of the rapidly evolv-ing research, synthesizing and extending research resultsfrom fishery, climate and hydroclimate studies, andlabeled the dominant pattern of PDV the Pacific(inter)Decadal Oscillation (PDO). Other studies haveused other names for what we call the PDO, for example:the Interdecadal Pacific Oscillation (IPO) of Power et al.(1997, 1999a), and the North Pacific Oscillation (NPO)of Gershunov and Barnett (1998).The collective body of research suggested that threemain characteristics distinguished PDO from ENSO: first,20th century PDO “events” persisted for 20-to-30 years,while typical ENSO events persisted for 6 to 18 months;second, the climatic fingerprints of the PDO were mostvisible in the extratropics, especially the North Pacific/North American sector, while secondary signatures ex-isted in the tropics, and the opposite was true for ENSO;and third, the mechanisms causing PDO variability werenot known, while causes for ENSO variability were rela-tively well-understood (Zhang et al., 1997; Mantua et al.,1997; NRC, 1998).A PDO index developed by Hare (1996) and Zhang(1996), also used by Mantua et al. (1997), is the leadingPC from an un-rotated EOF analysis of monthly, “re-sidual” North Pacific sea surface temperature (SST)anomalies, poleward of 20°N for the 1900–1993 periodof record (see lower panel of Fig. 1). “Residuals” are heredefined as the difference between observed anomalies andthe monthly mean global average SST anomaly (see Zhanget al., 1997). A remarkable characteristic of this index isits tendency