Long-term variability in the El Niño/Southern Oscillation and associated teleconnectionsMICHAEL E. MANN and RAYMOND S. BRADLEYDepartment of Geosciences, University of MassachusettsAmherst, Massachusetts 01003 U.S.A.MALCOLM K. HUGHESLaboratory of Tree Ring Research, University of ArizonaTucson, Arizona 85721 U.S.A.AbstractIntroductionDataTable 1 Proxy, historical, and long instrumental data records used. Entries give description/type...Table 1 (Continued) Proxy, historical, and long instrumental data records used. Entries give desc...Table 1 (Continued) Proxy, historical, and long instrumental data records used. Entries give desc...Fig. 1 Distribution of multiproxy network back to (a) 1650, (b) 1750, and (c) 1820 (all proxy ind...Fig. 2 Distribution of restricted multiproxy network used for Niño-3 reconstruction back to (a) 1...Fig. 3 Distribution of the (1082) nearly continuous available land air/sea surface temperature gr...MethodFig. 4 Empirical orthogonal functions of the five leading eigenvectors of the 1902–93 global temp...Fig. 5 Annual-mean PCs of the five leading eigenvectors of the 1902–93 global temperature data fr...Table 2 Calibration and verification statistics for global reconstructions. Resolved variance sta...Table 2 (Continued)Fig. 6 Empirical orthogonal functions of the three leading eigenvectors of the 1902–93 restricted...Fig. 7 Annual-mean PCs of the three leading eigenvectors of the 1902–93 restricted tropical Pacif...Table 3 Calibration and verification statistics for tropical Pacific SST reconstructions, focusin...Fig. 8 Comparison of raw and reconstructed/calibrated (a) NH and (b) Niño-3 series during the 190...Fig. 9 Histogram of calibration residuals for (a) NH and (b) Niño-3 series. A Gaussian (normal) p...Fig. 10 Spectrum of raw instrumental (solid) and reconstructed/calibrated (dashed) Niño-3 series ...Fig. 11 Niño-3 correlation map during1902–80 calibration period based on (top) raw data and (bott...Fig. 12 Pattern of calibration (top, based on 1902–80 data) and verification (bottom, based on 18...Fig. 13 Reconstructed Niño-3 series along with the (negative and rescaled) SOI series (dot-dashed...Temperature reconstructionsFig. 14 Comparison of the proxy-reconstructed temperature anomaly pattern for 1941 with raw data,...Fig. 15 Time reconstructions (RPCs-solid) and 1902–93 raw data (PCs-dotted) for the first five gl...Fig. 16 Time reconstructions (RPCs-solid) and 1902–93 raw data (PCs-dotted) for the three eigenve...Fig. 17 Reconstructed Niño-3 index back to 1650, along with instrumental series from 1902–93. The...Fig. 18 Comparison of reconstructed Northern Hemisphere (NH) mean annual temperature series and N...Fig. 19 Niño-3 temperature correlation map in successive 50-year periods between 1650 and present...Fig. 20 Evolutive spectrum of Niño-3 index (top) and evolutive multivariate spectrum of global te...Fig. 21 Canonical pattern of the interannual ENSO signal in the temperature reconstructions as it...Fig. 22 Canonical pattern of the interdecadal ENSO-like signal in the temperature reconstructions...Fig. 23 Frequency of extreme warm or cold ENSO events per 80-year period, based on the reconstruc...Fig. 24 Frequency of unusually (5-year) long warm ENSO “spells” per 80-year period, for the recon...Fig. 25a Global temperature anomaly patterns for selected QN92 documented strong (S) and very str...Fig. 25a Global temperature anomaly patterns for selected QN92 documented strong (S) and very str...Fig. 26 Global temperature anomaly patterns for two reconstructed cold ENSO (La Niña) years. (i) ...ConclusionsAcknowledgmentsReferences321Long-termvariabilityintheElNiño/SouthernOscillationandassociated teleconnectionsMICHAEL E. MANN and RAYMOND S. BRADLEYDepartment of Geosciences, University of MassachusettsAmherst, Massachusetts 01003 U.S.A.MALCOLM K. HUGHESLaboratory of Tree Ring Research, University of ArizonaTucson, Arizona 85721 U.S.A.AbstractWe analyze global patterns of reconstructed surface temperature for insights into thebehavior of the El Niño/Southern Oscillation (ENSO) and related climatic variability during thepast three centuries. The global temperature reconstructions are based on calibrations of a large setof globally distributed proxy records, or “multiproxy” data, against the dominant patterns ofsurface temperature during the past century. These calibrations allow us to estimate large-scalesurface temperature patterns back in time. The reconstructed eastern equatorial Pacific “Niño-3”areal-mean sea surface temperature (SST) index is used as a direct diagnostic of El Niño itself,while the global ENSO phenomenon is analyzed based on the full global temperature fields. Wedocument low-frequency changes in the base state, amplitude of interannual variability, andextremes in El Niño, as well as in the global pattern of ENSO variability. Recent anomalousbehavior in both El Niño and the global ENSO is interpreted in the context of the long-termreconstructed history and possible forcing mechanisms. The mean state of ENSO, its globalpatterns of influence, amplitude of interannual variability, and frequency of extreme events showconsiderable multidecadal and century-scale variability over the past several centuries. Many ofthese changes appear to be related to changes in global climate, and the histories of external forcingagents, including recent anthropogenic forcing.IntroductionThe instrumental record of roughly the past century provides many key insights into thenature of the global El Niño/Southern Oscillation (ENSO) phenomenon. For example, classicinstrumental indices of ENSO, such as the Southern Oscillation Index (SOI) or Niño-3 easterntropical Pacific sea surface temperature (SST) index, demonstrate concentration of interannualvariability in the interannual 3–7 year period band, (2) interdecadal modulation in the frequency322and intensity of ENSO episodes, and (3) demonstrable (though variable) influences of ENSO onextratropical storm tracks influencing the climates of North America and other regions remotefrom the tropical Pacific (see, e.g., the reviews of Philander 1990; Diaz and Markgraf 1993; Allanet al. 1996; and references within). There is little instrumental evidence, however, to describe thenature of variability prior to the twentieth century. A variety of theoretical, dynamical modelssupport most of the features of ENSO evident in the instrumental record, including the