UBC Faculty Research and Publications

Changes in the leading ENSO modes associated with the late 1970s climate shift: Role of surface zonal.. An, Soon-Il; Ye, Zhengqing; Hsieh, William W. 2006-07-22

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


52383-Hsieh_AGU_2006_2006GL026604.pdf [ 1.35MB ]
JSON: 52383-1.0041773.json
JSON-LD: 52383-1.0041773-ld.json
RDF/XML (Pretty): 52383-1.0041773-rdf.xml
RDF/JSON: 52383-1.0041773-rdf.json
Turtle: 52383-1.0041773-turtle.txt
N-Triples: 52383-1.0041773-rdf-ntriples.txt
Original Record: 52383-1.0041773-source.json
Full Text

Full Text

Changes in the leading ENSO modes associated with the late 1970sclimate shift: Role of surface zonal currentSoon-Il An,1Zhengqing Ye,2and William W. Hsieh2Received 16 April 2006; revised 11 June 2006; accepted 13 June 2006; published 22 July 2006.[1] In this study, using the nonlinear principal analysis(NLPCA) technique, we demonstrated that the leading ElNino-Southern Oscillation (ENSO) mode in a physical basischanged since the late 1970s. The ENSO mode during thepre-1980s resembles the destabilized ‘ocean basin mode’,while during the post-1980s does ‘the recharge-mode’. Inparticular, for the pre-1980s, the surface zonal currentmainly acts as an intensifier of the ENSO, while that duringpost-1980s it plays a role in the transition of the ENSOcycle. The NLPCA results are reconfirmed by the eigenanalysis of the linearized intermediated ENSO model. Overa reasonable parameter range, the leading eigen modesassociated with the pre-1980s are completely separable fromthose associated with the post-1980s. The eigen structuresfor each decadal-period resemble the correspondingNLPCA patterns. Citation: An, S.-I., Z. Ye, and W. W.Hsieh (2006), Changes in the leading ENSO modes associatedwith the late 1970s climate shift: Role of surface zonal current,Geophys. Res. Lett., 33, L14609, doi:10.1029/2006GL026604.1. Introduction[2] In the late 1970s, an abrupt change of the climate statein the midlatitude as well as in the tropical Pacific wasobserved [Trenberth and Hurrel, 1994]. Together with suchclimate shift, the main characteristics of the interannualvariability over the tropical Pacific have changed [Wangand An, 2001; An and Wang, 2000; An et al., 2005]. On onehand, the multi-decadal variation in the climate state causesthe non-stationary characteristics of El Nin˜o-SouthernOscillation (ENSO) [e.g., An and Jin, 2000]. On the otherhand,itcanalsobeattributedtotheinternalnonlinearprocessof ENSO in a certain parameter regime [Timmermann et al.,2003; An and Jin, 2004]. Although it is still uncertainwhether the ENSO system is linear or nonlinear [e.g.,Philander and Fedorov, 2003; An and Jin, 2004], we viewthat the late 1970s climate shift is fairly enough to change theleading ENSO mode.[3] Recently, An et al. [2005] obtained the nonlinearENSO cycle in the evolution of tropical Pacific thermoclineanomalies using the nonlinear principal component analysis(NLPCA), which fits a closed curve to the data in the multi-dimensional PC (principal component) space. Interestingly,the shape of the closed curve, especially its asymmetry,significantly changed since the late 1970s, inferring achange of the leading ENSO mode. Furthermore, An andJin [2000] showed that the interdecadal change could leadto a quantitative change in the leading eigen mode of astripped-down version of the intermediate ENSO model. Asa further study over the previous works, here we presentobservational evidence on the interdecadal change in theleading ENSO mode and perform an eigen analysis of theintermediate complexity ENSO model.2. Combined NLPCA of the Sea Level andCurrent Anomalies[4] We have applied the combined principal componentanalysis (CPCA; a.k.a. Multi Singular Value Decompositionmethod) to monthly mean sea level height, sea surfacetemperature (SST) and upper-50m surface zonal currentanomalies. The anomalies were obtained by removing theclimatological monthly mean, and then a 7-point runningmean was applied. The data utilized are the University ofMaryland Simple Ocean Assimilation (SODA) data forJanuary 1958–December 2001 [Carton et al., 2000]. InFigure 1, the main features of sea level height, SST andsurface current’s variability are summarized by the twoleading modes: the first and second modes of sea levelheight are a zonal contrast mode (Figure 1c) and anapproximate zonal-symmetric mode (Figure 1d) with re-spect to the equator, respectively; those of SST are ahorseshoe pattern like the mature phase of ENSO(Figure 1a) and an equatorially-confined pattern like theinitial phase of ENSO (Figure 1b), respectively; those of thesurface current are a localized zonal/meridional contrastmode (Figure 1e) and an equatorial basin-wide mode(Figure 1f), respectively. Obviously, the first leading modeis highly correlated to the ENSO index, and the secondmode shows a lead/lag relationship to the ENSO index.[5] The corresponding trajectory plot of the two leadingPCsassociatedwiththesealevelheightisshowninFigure1g.The virtual center of the tracks for the pre-1980s (1958–1979) is located at the origin, while that for the post-1980s(1980–2001) moves away from the origin, inferring apossible regime shift in the behavior of the two PCs.[6] From the first six leading CPCA modes, we extractedthe combined NLPCA modes; that is, a closed curve (basedon a neural network representation; see Hsieh [2004] fordetails) is used to fit the data in the 6-dimensional PC space(similar to An et al. [2005]). In Figure 2, the evolutionfeatures of SST, sea level height and surface zonal currentassociated with the first NLPCA mode on the equatorialband are shown. These modes explain 65.7% and 73.3% oftotal variance (of the 6 leading modes only, not of theoriginal data set) during the pre- and post-1980s, respec-tively, which are greater than those of the leading CPCAGEOPHYSICAL RESEARCH LETTERS, VOL. 33, L14609, doi:10.1029/2006GL026604, 2006ClickHereforFullArticle1Department of Atmospheric Sciences/Global Environment Laboratory,Yonsei University, Seoul, Korea.2Department of Earth and Ocean Sciences, University of BritishColumbia, Vancouver, British Columbia, Canada.Copyright 2006 by the American Geophysical Union.0094-8276/06/2006GL026604$05.00L14609 1of5mode (the CPCA mode 1 accounts for 49.5% and 56.3% ofthe total variance from the 6 leading PC2 during the pre-and post-1980s, respectively). For the pre-shift regime, thestanding-type oscillation in the sea level height with its nodearound the dateline and its center at the 120C176W is observed;the surface zonal current also shows a standing-type oscil-lation with a basin-wide pattern and its minimum at bound-aries; and SST anomalies shows a westward propagatingFigure 1. (a–f) The first and second principal component spatial pattern of (left) the SST, (middle) sea level height and(right) surface zonal current anomalies obtained from the Combined Principal Component Analysis (CPCA). Thepercentage of variance explanation appears at the top of each plot. (g) Trajectory plot of the principal components of thefirst and second PCs.Figure 2. Phase-longitude section of (left) the SST, (middle) sea level height, and (right) the surface zonal currentanomalies over the equatorial band (5C176S–5C176N) associated with the first NLPCA mode (a–c) for the pre-1980s and (d–f) forthe post-1980s. Phase generally increases with time.L14609 AN ET AL.: CHANGES IN THE LEADING ENSO MODES L146092of5tendency, especially in the eastern-to-central Pacific. Thesurface zonal current pattern is very similar to the so-called‘ocean basin mode’ [Neelin and Jin, 1993]. The ocean basinmode is a natural mode produced by the equatorial trappedwaves and their reflection at boundaries: the equatorialKelvin wave travels eastward until it hits the easternboundary where it is mostly reflected to the equatorially-trapped Rossby wave; this Rossby wave moves westwardand reflects at the western boundary as the Kelvin wave;thus a semi-closed loop completed by the traveling-waveswith 8–12 month periods is established in the tropicalPacific basin (with energy loss at the eastern boundarydue to the poleward traveling coastal Kelvin wave [Jin,1997]). Neelin and Jin [1993] mentioned that this oceanbasin mode can be destabilized by the zonal advectivefeedback (i.e., zonal advection of the mean zonal tempera-ture gradient by the anomalous zonal current). Once theocean basin mode becomes unstable, it could be modified asthe interannual mode, and have larger amplitude than theuncoupled ocean basin mode. An and Jin [2000] showedthat the zonal temperature gradient of the mean SST duringthe 1961–75 is relatively larger than that during 1981–95,and it caused the destabilization of the ocean basin modeduring the pre-1980s. In many respects, the leading NLPCAmode of the observations is consistent with the previousstudy.[7] For the post-1980s, the evolution features are quitedifferent from those for the pre-1980s (bottom plots inFigure 2). The standing feature in the sea level height turnsto the propagating feature; the basin-wide pattern of thesurface current disappears; and the westward moving ten-dency of SST changes to the eastward moving tendency[e.g., An and Wang, 2000]. The maximum surface current inthe eastern Pacific appears when the zonal gradient of thesea level height is at maximum, indicating that the surfacecurrent is mostly due to the local pressure gradient from themass difference rather than the surface wind stress. Forexample, the positive SST anomaly and the resulting con-vective atmospheric heat generate the positive wind stressanomaly in the western part of the center of the SSTanomaly, which may enforce the westerly zonal surfacecurrent. But the zonal current anomaly is negative as shownhere. Since the geostrophic current can be represented bythe gradient of the sea level height (or thermocline), theeffect of the zonal current in inducing the coupled mode iseasily adding into the effect of sea level height. An and Jin[2001], under the two-box approximation of the intermedi-ate ENSO model, theoretically proved this point, andmentioned that the two-box approximation could representthe ocean adjustment but not the ocean wave dynamics, thusonly the recharge mode is possible. In this regard, the ENSOmode inferred from the NLPCA pattern for the post-shiftregime is presumably the recharge mode due to the desta-bilization of the ocean adjustment mode [Jin, 1997]. Notethat the life time and amplitude for the positive zonalcurrent anomaly and those for the negative zonal currentanomaly are not the same. This asymmetric oscillation wasdominant during the post-1980s [An et al., 2005].[8] The phase relationships among the SST, sea levelheight, and surface zonal current appeared in the NLPCAmode give an idea on the role of the surface current for eachdecadal period. For the pre-1980s, as shown in Figure 2, thesea level height anomalies over the equatorial easternPacific are almost in-phase with the surface zonal currentanomalies. Since a positive (negative) surface zonal currentanomaly generates a positive (negative) SST anomalythrough the warm (cold) surface water advection by theeastward (westward) zonal current, the surface current playsa role in the intensification of this system. For the post-1980s, on the other hands, the surface zonal current anoma-lies in the equatorial Eastern Pacific lead the sea level heightanomalies by about a quarter of a cycle. It infers that thesurface zonal current plays a role of the transition mecha-nism in this oscillation. Thus, for each decadal period, therole of the surface zonal current in ENSO evolution is quitedistinctive.3. Eigen Analysis of the Intermediate ENSOModel[9] In the previous section, we presented the observa-tional evidence for the decadal changes of the leadingENSO mode. Here, we perform the eigen analysis of theintermediate ENSO model linearized with respect to theclimatological mean background states (for details, see Anet al. [2004]). The background states for the model arespecified based on the observed. To avoid model drifts awayfrom the original background state of the intermediatemodel when the climate states for 1981–95 or 1961–75are directly used, we add/subtract the difference between theannual-mean state for 1981–95 and that for 1961–75 into/from the original background state, which includes surfacewind, atmospheric surface divergence, SST, current, up-welling, and thermocline depth (approximately 20C176C iso-therm depth). For this calculation, again SODA data areutilized.[10] Here, the eigenvalues obtained from the linearizedintermediate ENSO model for each background state areplotted in Figure 3. To test the model behaviour for differentparameters, the eigen analysis is performed for variousvalues of the coupling coefficient, m. The coupling coeffi-cient, representing the strength of the dynamical/thermody-namical coupling between atmosphere and ocean in theFigure 3. Eigenvalues obtained from the linearizedintermediate ENSO model for the pre-1980s backgroundcondition (open square), and for the post-1980s backgroundcondition (open circle). The eigen spectrum is symmetricaround the zero frequency. The size of marker indicates thestrength of the coupling coefficient (m).L14609 AN ET AL.: CHANGES IN THE LEADING ENSO MODES L146093of5physical sense, is also a control parameter for the Hopf-bifurcation. The Hopf-bifurcation occurs at the boundarybetween stable and unstable regimes. A striking feature inFigure 3 is the clear separation between the leading eigenmodes for the post-1980s and those for the pre-1980s. Thisdistinction is maintained even if the coupling coefficientchanges. For the post-1980s, the leading eigen mode for m =1 has a frequency of about 0.3 yrC01(C243 years period). Asthe coupling coefficient increases, the growth rate signifi-cantly increases, and the frequency increases by a tinyamount. Over the range of the coupling coefficient (0.8 <m < 1.4, with m = 1 as the standard value), the leading modesare unstable oscillatory modes with interannual period. TheHopf-bifurcation point is located between m = 0.8 and 0.9.The leading mode corresponds to the recharge mode [Jin,1997].[11] For the pre-1980s, the Hopf-bifurcation, on the otherhands, is observed between m = 1.3 and 1.4. The leadingmodes in the sub-critical regime are a damped stationarymode, while those in the supercritical regime are the high-frequency westward-propagating oscillatory mode, resem-bling the ‘SST mode’ [Jin and Neelin, 1993]. When m > 1.6,the second leading modes with the quasi-biennual period arebrought up. This quasi-biennual mode becomes unstable form > 1.7.[12] To make the point more clearly, Figure 4 shows theeigen structure of SST, height (the height of this dynamicmodel can be a proxy of the thermocline depth), and surfacezonal currents associated with the leading eigen modes forthe standard value of the coupling coefficient (m = 1). Asmentioned previously, the leading mode for the pre-1980s isa damped stationary mode. The SST anomalies over theeastern Pacific are positively correlated with both thethermocline depth and surface zonal current anomalies inthe spatial pattern, consistent with the NLPCA mode of theobservations (Figure 2). As mentioned before, physically,both the positive (negative) thermocline depth anomaly andthe positive (negative) surface zonal current anomaly inducethe positive (negative) SST anomaly, through vertical andzonal thermal advection, respectively. Thus, as in theNLPCA result, the surface zonal current plays a role inthe intensification of the SST anomaly. However, the rigidin-phase relationship of this stationary mode is not realistic.More realistic modes can be captured in other parameterranges, as discussed in the next section.[13] For the post-1980s, the eigen structure of the leadingmode (about 3-year period) resembles the observed modefrom NLPCA (3.7-year period). The SST maximum isconfined over the eastern Pacific, and the evolution of thethermocline depth anomaly is similar to the observed. In theeastern Pacific, the surface zonal current anomaly leads theSST anomaly by a quarter cycle, indicating that the zonalcurrent anomaly plays a transition role in this oscillatorymode. The phase relationship among three variables infersthat the SST anomaly over the eastern Pacific is closelyrelated to the thermocline depth anomaly and that over thecentral Pacific is related to the surface zonal currentanomaly. This phase relationship has been observed in theNLPCA mode (see Figures 2d–2f). Overall, the eigenstructure of the leading mode obtained from the linearizedintermediate ENSO model is consistent with the NLPCAresult, when the corresponding background climate state isaccounted for. Thus, the observed evidence for the changein the leading mode associated with the late 1970s climateshift is reconfirmed by the eigen analysis of the dynamicalsystem.4. Concluding Remarks[14] In this study, we proposed that the decadal change inthe ENSO characteristics was rather the change of theleading eigen mode driven by the different physical conse-quence than that addressed by a statistical artifact. As aphysical consequence, the role of the surface zonal currentis clearly distinct in the NLPCA modes associated with thetwo regimes: for the pre-1980s, the surface zonal currenttends to be an intensifier of the system, while that for thepost-1980s plays a role in the ENSO transition. The eigenanalysis of the intermediate ENSO model reconfirmed thispoint such that over a reasonable parameter range (here, thecoupling coefficient is varied), the possible leading eigenmode associated with the pre-1980s is completely separablefrom that associated with the post-1980s. All these resultsprovide firm support for the decadal change of the leadingENSO mode.[15] One may argue that the leading eigen mode associ-ated with the pre-1980s for m = 1 is unrealistic so that it isFigure 4. Time-longitude dependence of the leading eigen mode (growth rate suppressed) along the equator (5C176S–5C176N)for m = 1.0: (a) SST, (b) height, and (c) surface zonal current for the pre-1980s condition; (d–f) as in Figure 4a–4c exceptfor the post-1980s condition. Y-axis indicates degrees in one cycle. The contour intervals of SST, height, and current for thepre-1980s (the post-1980s) condition are 0.1 C (0.5 C), 1 m (5 m), and 1 cm sC01(5 cm sC01), respectively.L14609 AN ET AL.: CHANGES IN THE LEADING ENSO MODES L146094of5not matched to the observations. A possible explanation forthis discrepancy is: Since we do not know the right value ofthe standard parameter (e.g., the coupling coefficient), ourchoice may not always be correct. For example, as shown inFigure 3, the leading eigen mode by changing the couplingstrength (m) relocates the stationary regime to the oscillatoryregime. In the oscillatory regime, although the period of theleading eigen mode is slightly shorter than the observed, theeigen structures of SST, thermocline depth, and zonalcurrent resembles the observed (not shown). In addition,the mean vertical stratification of the ocean and the con-tributions by the high-order baroclinic modes, which couldlead to change in the characteristics of the ENSO mode[Moon et al., 2004], are not considered here.[16] We have noticed a sort of inconsistency betweenthe observed surface zonal current and that of the eigen-structure over the central Pacific (Figures 2f and 4f),particularly in their evolution features with respect to theevolution of SST anomaly. We concluded that it is due tothe model’s deficiency for simulating the undercurrent. Aswe know, strong zonal flow, the so-called ‘undercurrent’,exists in the equatorial subsurface. The undercurrent mayinfluence the surface zonal current through the momen-tum diffusion or nonlinear momentum flux [Philander,1990]. The undercurrent is weaker during the El Ninoyear. So, the anomalous undercurrent becomes negative.This anomalous easterly momentum is reflected to thesurface zonal current, thus the surface zonal currentanomaly tends to be negative as well. The model couldnot capture this phenomenon.[17] Acknowledgments. S.-I. An is supported by the SRC program ofKorea Science and Engineering Foundation, and the Brain Korea 21project. Z. Ye and W. Hsieh are supported by the Natural Sciences andEngineering Research Council of Canada.ReferencesAn, S.-I., and F.-F. Jin (2000), An eigen analysis of the interdecadalchanges in the structure and frequency of ENSO mode, Geophy. Res.Lett., 27, 1573–1576.An, S.-I., and F.-F. Jin (2001), Collective role of thermocline and zonaladvective feedbacks in the ENSO mode, J. Clim., 14, 3421–3432.An, S.-I., and F.-F. Jin (2004), Nonlinearity and asymmetry of ENSO,J. Clim., 17, 2399–2412.An, S.-I., and B. Wang (2000), Interdecadal change of the structure of theENSO mode and its impact on the ENSO frequency, J. Clim., 13, 2044–2055.An, S.-I., A. Timmermann, L. Bejarano, F.-F. Jin, F. Justino, Z. Liu, andA. W. Tudhope (2004), Modeling evidence for enhanced El Nin˜o–Southern Oscillation amplitude during the Last Glacial Maximum,Paleoceanography, 19, PA4009, doi:10.1029/2004PA001020.An, S.-I., W. W. Hsieh, and F.-F. Jin (2005), A nonlinear analysis of theENSO cycle and its interdecadal changes, J. Clim., 18, 3229–3239.Carton, J., G. Chepurin, X. Cao, and B. Giese (2000), A simple ocean dataassimilation analysis of the global upper ocean 1950–95. part I: Meth-odology, J. Phys. Oceanogr., 30, 294–309.Hsieh, W. W. (2004), Nonlinear multivariate and time series analysis byneural network methods, Rev. Geophys., 42, RG1003, doi:10.1029/2002RG000112.Jin, F.-F. (1997), An equatorial ocean recharge paradigm for ENSO. part II:A stripped-down coupled model, J. Atmos. Sci., 54, 811–829.Jin, F.-F., and J. D. Neelin (1993), Modes of interannual tropical ocean-atmosphere interaction—A unified view. part I: Numerical results,J. Atmos. Sci., 50, 3477–3503.Moon, B., S. Yeh, B. Dewitte, J. Jhun, I. Kang, and B. P. Kirtman (2004),Vertical structure variability in the equatorial Pacific before and after thePacific climate shift of the 1970s, Geophys. Res. Lett., 31, L03203,doi:10.1029/2003GL018829.Neelin, J. D., and F.-F. Jin (1993), Modes of interannual tropical ocean-atmosphere interaction—A unified view. part II: Analytical results in theweak-coupling limit, J. Atmos. Sci., 50, 3504–3522.Philander, G. H. (1990), El Nino, La Nina, and the Southern Oscillation,293 pp., Elsevier, New York.Philander, G. H., and A. Fedorov (2003), Is El Nin˜o sporadic or cyclic?,Annu. Rev. Earth Planet. Sci., 31, 579–594.Timmermann, A., F.-F. Jin, and J. Abshagen (2003), A nonlinear theory forEl Nino bursting, J. Atmos. Sci., 60, 152–165.Trenberth, K. E., and J. W. Hurrel (1994), Decadal atmosphere-oceanvariations in the Pacific, Clim. Dyn., 9, 303–319.Wang, B., and S.-I. An (2001), Why the properties of El Nino changedduring the late 1970s, Geophy. Res. Lett., 28, 3709–3712.C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0C0S.-I. An, Department of Atmospheric Sciences/Global EnvironmentLaboratory, Yonsei University, 134 Shinchon-dong, Seodaemun-gu, Seoul120-749, Korea. (sian@yonsei.ac.kr)W. W. Hsieh and Z. Ye, Department of Earth and Ocean Sciences,University of British Columbia, 6339 Stores Road, Vancouver, BC, CanadaV6T1Z4.L14609 AN ET AL.: CHANGES IN THE LEADING ENSO MODES L146095of5


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items