• Bettio, L., S. B. Power, and K. Walsh, 2003: The dynamics of ENSO in a coupled GCM. Proc. Int. Conf. on Earth System Modelling, Hamburg, Germany, Max Planck Institute for Meteorology, 327 pp.

  • Burgers, G., M. A. Balmaseda, F. C. Vossepoel, G. J. V. Oldenborgh, and P. J. van Leeuwen, 2002: Balanced ocean-data assimilation near the equator. J. Phys. Oceanogr., 32 , 25092519.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., and M. A. Alexander, 2001: Rossby waves in the tropical Pacific and their role in decadal thermocline variability. J. Phys. Oceanogr., 31 , 34963515.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., M. A. Alexander, and C. Deser, 2003: Why are there Rossby wave maxima in the Pacific at 10°S and 13°N? J. Phys. Oceanogr., 33 , 15491563.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., M. A. Alexander, C. Deser, and M. J. McPhaden, 2005: Anatomy and decadal evolution of the subtropical–tropical cells (STCs). J. Climate, 18 , 37393758.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., and M. G. Schlax, 1996: Global observations of oceanic Rossby waves. Science, 272 , 234238.

  • Davis, R. E., 1976: Predictability of sea surface temperature and sea level pressure anomalies over the North Pacific Ocean. J. Phys. Oceanogr., 6 , 249266.

    • Search Google Scholar
    • Export Citation
  • Dijkstra, H. A., and G. Burgers, 2002: Fluid dynamics of El Niño variability. Annu. Rev. Fluid Mech., 34 , 531558.

  • Fischer, G., 1965: On a finite-difference scheme for solving the primitive equations for a barotropic fluid with application to the boundary problem. Tellus, 17 , 405412.

    • Search Google Scholar
    • Export Citation
  • Folland, C. K., D. E. Parker, A. W. Coleman, and R. Washington, 1999: Large scale modes of ocean surface temperature since the late nineteenth century. Beyond El Niño: Decadal and Interdecadal Climate Variability, A. Navarra, Ed., Springer-Verlag, 73–102.

  • Folland, C. K., J. A. Renwick, M. J. Salinger, and A. B. Mullan, 2002: Relative influences of the Interdecadal Pacific Oscillation and ENSO on the South Pacific convergence zone. Geophys. Res. Lett., 29 .1643, doi:10.1029/2001GL014201.

    • Search Google Scholar
    • Export Citation
  • Galanti, E., and E. Tziperman, 2003: A midlatitudes–ENSO teleconnection mechanism via baroclinically unstable long Rossby waves. J. Phys. Oceanogr., 33 , 18771888.

    • Search Google Scholar
    • Export Citation
  • Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend on exchanges between the Tropics and extratropics. Science, 275 , 805807.

    • Search Google Scholar
    • Export Citation
  • Hazeleger, W., M. Visbeck, M. Cane, A. Karspeck, and N. Naik, 2001: Decadal upper ocean temperature variability in the tropical Pacific. J. Geophys. Res., 106 , 89718988.

    • Search Google Scholar
    • Export Citation
  • Jin, F. F., J. D. Neelin, and M. Ghil, 1994: El Niño on the Devil’s Staircase: Annual subharmonic steps to chaos. Science, 264 , 7072.

    • Search Google Scholar
    • Export Citation
  • Karspeck, A. R., R. Seager, and M. A. Cane, 2004: Predictability of tropical Pacific decadal variability in an intermediate model. J. Climate, 17 , 28422850.

    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., 1991: Can reflected extra-equatorial Rossby waves drive ENSO? J. Phys. Oceanogr., 21 , 444452.

  • Kleeman, R., J. J. P. McCreary, and B. A. Klinger, 1999: A mechanism for generating ENSO decadal variability. Geophys. Res. Lett., 26 , 17431746.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., L. Wu, and E. Bayler, 1999: Rossby wave–coastal Kelvin wave interaction in the extratropics. Part I: Low-frequency adjustment in a closed basin. J. Phys. Oceanogr., 29 , 23822404.

    • Search Google Scholar
    • Export Citation
  • Lysne, J., P. Chang, and B. Giese, 1997: Impact of extratropical Pacific on equatorial variability. Geophys. Res. Lett., 24 , 25892592.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • McGregor, S., N. J. Holbrook, and S. B. Power, 2004: On the dynamics of interdecadal thermocline depth and sea surface temperature variability in the low to mid-latitude Pacific Ocean. Geophys. Res. Lett., 31 .L24201, doi:10.1029/2004GL021241.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415 , 603608.

    • Search Google Scholar
    • Export Citation
  • Meinen, C. S., and M. J. McPhaden, 2000: Observations of warm water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. J. Climate, 13 , 35513559.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and A. Arakawa, 1976: Numerical methods used in atmospheric models. GARP Publication Series 17, 64 pp.

  • Miller, A. J., and N. Schneider, 2000: Interdecadal climate regime dynamics in the North Pacific Ocean: Theories, observations and ecosystem impacts. Prog. Oceanogr., 47 , 355379.

    • Search Google Scholar
    • Export Citation
  • Newman, M., G. P. Compo, and M. A. Alexander, 2003: ENSO-forced variability of the Pacific decadal oscillation. J. Climate, 16 , 38533857.

    • Search Google Scholar
    • Export Citation
  • Nonaka, M., S-P. Xie, and J. P. McCreary, 2002: Decadal variations in the subtropical cells and equatorial Pacific SST. Geophys. Res. Lett., 29 .1116, doi:10.1029/2001GL013717.

    • Search Google Scholar
    • Export Citation
  • Pacanowski, R. C., K. Dixon, and A. Rosati, 1991: The GFDL Modular Ocean Model users guide, version 1.0. GFDL Ocean Group Tech. Rep. 2, 376 pp.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. Academic Press, 289 pp.

  • Power, S. B., and R. Colman, 2006: Multi-year predictability in a coupled general circulation model. Climate Dyn., 26 , 247272.

  • Power, S. B., R. Kleeman, F. Tseikin, and N. R. Smith, 1995: A global version of the GFDL modular ocean model for ENSO studies. BMRC Tech. Rep., 18 pp.

  • Power, S. B., F. Tseikin, A. W. Coleman, and A. Sulaiman, 1998: A coupled general circulation model for seasonal prediction and climate change research. BMRC Research Rep. 66, 52 pp.

  • Power, S. B., T. Casey, C. Folland, A. Colman, and V. Mehta, 1999a: Inter-decadal modulation of the impact of ENSO in Australia. Climate Dyn., 15 , 319324.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., F. Tseitkin, V. Mehta, B. Lavery, S. Torok, and N. J. Holbrook, 1999b: Decadal climate variability in Australia during the twentieth century. Int. J. Climatol., 19 , 169184.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., M. Haylock, R. Colman, and X. Wang, 2006: The predictability of interdecadal changes in ENSO activity and ENSO teleconnections. J. Climate, 19 , 47554771.

    • Search Google Scholar
    • Export Citation
  • Rebert, J. P., J. R. Donguy, and G. Eldin, 1985: Relations between sea level, thermocline depth, heat content and dynamic height in the tropical Pacific Ocean. J. Geophys. Res., 90 , 1171911725.

    • Search Google Scholar
    • Export Citation
  • Rodgers, K. B., P. Friederichs, and M. Latif, 2004: Tropical Pacific decadal variability and its relation to decadal modulations of ENSO. J. Climate, 17 , 37613774.

    • Search Google Scholar
    • Export Citation
  • Timmermann, A., F. F. Jin, and J. Abshagen, 2003: A nonlinear theory for El Niño bursting. J. Atmos. Sci., 60 , 152165.

  • Tomczak, M., and S. J. Godfrey, 1994: Regional Oceanography: An Introduction. Pergamon Press, 422 pp.

  • Tourre, Y. M., C. Cibot, L. Terray, W. B. White, and B. Dewitte, 2005: Quasi-decadal and inter-decadal climate fluctuations in the Pacific Ocean from a CGCM. Geophys. Res. Lett., 32 .L07710, doi:10.1029/2004GL022087.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and J. W. Hurrell, 1994: Decadal atmosphere-ocean variations in the Pacific. Climate Dyn., 9 , 303319.

  • Trenberth, K. E., and T. J. Hoar, 1997: El Niño and climate change. Geophys. Res. Lett., 24 , 30573060.

  • Tziperman, E., L. Stone, M. A. Cane, and H. Jarosh, 1994: El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean-atmosphere oscillator. Science, 264 , 7274.

    • Search Google Scholar
    • Export Citation
  • Wang, X., F. F. Jin, and Y. Wang, 2003: A tropical recharge mechanism for climate variability. Part II: A unified theory for decadal and ENSO modes. J. Climate, 16 , 35993616.

    • Search Google Scholar
    • Export Citation
  • White, W. B., Y. M. Tourre, M. Barlow, and M. Dettinger, 2003: A delayed action oscillator shared by biennial, interannual, and decadal signals in the Pacific Basin. J. Geophys. Res., 108 .3070, doi:10.1029/2002JC001490.

    • Search Google Scholar
    • Export Citation
  • Wu, Z-J., R. Coleman, S. B. Power, X. Wang, and B. J. McAvaney, 2002: The El Niño Southern Oscillation response in the BMRC Coupled GCM. BMRC Research Rep. 91, 18 pp.

  • Zebiak, S. E., and M. A. Cane, 1987: A model El Niño–Southern Oscillation. Mon. Wea. Rev., 115 , 22622278.

  • View in gallery

    The first EOFs of the (a) unfiltered (11% variance) and (b) filtered (30% variance) BCM2.2 SST (respectively USST1 and FSST1) and (c), (d) their corresponding expansion coefficient time series.

  • View in gallery

    The SST signature of the IPO during its warm tropical Pacific phase. The contour interval is 0.04°C: negative contours are dashed (figure modified from Folland et al. 2002).

  • View in gallery

    The first EOFs of the (a) unfiltered thermocline depth anomalies from the SWM control simulation (11% variance), (b) filtered thermocline depth anomalies from the SWM control simulation (27% variance), (c) unfiltered BCM2.2 VAT (14% variance), and (d) filtered BCM2.2 VAT (25% variance). (e), (f) The corresponding expansion coefficient time series for the SWM results (black line) and the BCM2.2 results (gray line).

  • View in gallery

    The principal component time series for projections of (a) unfiltered EqWF; (b) filtered EqWF; (c) filtered OffEqWF; and (d) unfiltered OffEqWF, shown in black. The corresponding expansion coefficients for the SWM control simulation are shown in gray.

  • View in gallery

    Filtered SWM thermocline depth anomalies in the (a) Niño-3 and (b) Niño-4 regions.

  • View in gallery

    Filtered SWM thermocline depth variations in the (a) Niño-3 and (b) Niño-4 regions.

  • View in gallery

    Hovmöller plots of the unfiltered thermocline depth anomalies (m) along (a) 16°S and (b) 0° latitude generated in the OffEqWF experiment (expt Ib). Note the 16°S section has the x axis reversed so that it reads east-to-west from left-to-right. Contour intervals are provided every 8 m with minima at ±4 m, and every 0.4 m with minima at ±0.2 m, respectively; negative contours are dashed.

  • View in gallery

    Niño-3 region thermocline depth anomalies for each of the 17 SWM predictability experiments once the off-equatorial wind stress forcing was switched off (see legend inset).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 105 97 9
PDF Downloads 65 57 7

Interdecadal Sea Surface Temperature Variability in the Equatorial Pacific Ocean. Part I: The Role of Off-Equatorial Wind Stresses and Oceanic Rossby Waves

View More View Less
  • 1 Department of Physical Geography, Macquarie University, Sydney, New South Wales, Australia
  • | 2 Bureau of Meteorology Research Centre, Melbourne, Victoria, Australia
© Get Permissions
Full access

Abstract

The Australian Bureau of Meteorology Research Centre CGCM and a linear first baroclinic-mode ocean shallow-water model (SWM) are used to investigate ocean dynamic forcing mechanisms of the equatorial Pacific Ocean interdecadal sea surface temperature (SST) variability. An EOF analysis of the 13-yr low-pass Butterworth-filtered SST anomalies from a century-time-scale CGCM simulation reveals an SST anomaly spatial pattern and time variability consistent with the interdecadal Pacific oscillation. Results from an SWM simulation forced with wind stresses from the CGCM simulation are shown to compare well with the CGCM, and as such the SWM is then used to investigate the roles of “uncoupled” equatorial wind stress forcing, off-equatorial wind stress forcing (OffEqWF), and Rossby wave reflection at the western Pacific Ocean boundary, on the decadal equatorial thermocline depth anomalies.

Equatorial Pacific wind stresses are shown to explain a large proportion of the overall variance in the equatorial thermocline depth anomalies. However, OffEqWF beyond 12.5° latitude produces an interdecadal signature in the Niño-4 (Niño-3) region that explains approximately 10% (1.5%) of the filtered control simulation variance. Rossby wave reflection at the western Pacific boundary is shown to underpin the OffEqWF contribution to these equatorial anomalies. The implications of this result for the predictability of the decadal variations of thermocline depth are investigated with results showing that OffEqWF generates an equatorial response in the Niño-3 region up to 3 yr after the wind stress forcing is switched off. Further, a statistically significant correlation is found between thermocline depth anomalies in the off-equatorial zone and the Niño-3 region, with the Niño-3 region lagging by approximately 2 yr. The authors conclude that there is potential predictability of the OffEqWF equatorial thermocline depth anomalies with lead times of up to 3 yr when taking into account the amplitudes and locations of off-equatorial region Rossby waves.

Corresponding author address: Shayne McGregor, Department of Physical Geography, Macquarie University, Sydney 2109, Australia. Email: mcgregor@els.mq.edu.au

Abstract

The Australian Bureau of Meteorology Research Centre CGCM and a linear first baroclinic-mode ocean shallow-water model (SWM) are used to investigate ocean dynamic forcing mechanisms of the equatorial Pacific Ocean interdecadal sea surface temperature (SST) variability. An EOF analysis of the 13-yr low-pass Butterworth-filtered SST anomalies from a century-time-scale CGCM simulation reveals an SST anomaly spatial pattern and time variability consistent with the interdecadal Pacific oscillation. Results from an SWM simulation forced with wind stresses from the CGCM simulation are shown to compare well with the CGCM, and as such the SWM is then used to investigate the roles of “uncoupled” equatorial wind stress forcing, off-equatorial wind stress forcing (OffEqWF), and Rossby wave reflection at the western Pacific Ocean boundary, on the decadal equatorial thermocline depth anomalies.

Equatorial Pacific wind stresses are shown to explain a large proportion of the overall variance in the equatorial thermocline depth anomalies. However, OffEqWF beyond 12.5° latitude produces an interdecadal signature in the Niño-4 (Niño-3) region that explains approximately 10% (1.5%) of the filtered control simulation variance. Rossby wave reflection at the western Pacific boundary is shown to underpin the OffEqWF contribution to these equatorial anomalies. The implications of this result for the predictability of the decadal variations of thermocline depth are investigated with results showing that OffEqWF generates an equatorial response in the Niño-3 region up to 3 yr after the wind stress forcing is switched off. Further, a statistically significant correlation is found between thermocline depth anomalies in the off-equatorial zone and the Niño-3 region, with the Niño-3 region lagging by approximately 2 yr. The authors conclude that there is potential predictability of the OffEqWF equatorial thermocline depth anomalies with lead times of up to 3 yr when taking into account the amplitudes and locations of off-equatorial region Rossby waves.

Corresponding author address: Shayne McGregor, Department of Physical Geography, Macquarie University, Sydney 2109, Australia. Email: mcgregor@els.mq.edu.au

1. Introduction

The interdecadal sea surface temperature (SST) variability of the Pacific Ocean has been characterized by the third empirical orthogonal function (EOF) of the 13-yr lowpass-filtered near-global SST (Folland et al. 1999). This mode of interdecadal SST variability has been titled the interdecadal Pacific oscillation (IPO) due to its time scale of oscillation and symmetry about the Pacific Ocean equator (Folland et al. 1999, 2002; Power et al. 1999a). Folland et al. (2002) propose that the IPO is the Pacific-wide manifestation of the North Pacific principal mode of decadal variability named the Pacific decadal oscillation (PDO) by Mantua et al. (1997). The basinwide IPO SST pattern is very similar to the SST pattern associated with El Niño–Southern Oscillation (ENSO) events, although there are several distinct differences, in particular that the IPO warm SST anomalies reach the western tropical Pacific and extend farther away from the equator than is typical of ENSO events. Further, the IPO equatorial Pacific warming is accompanied by extensive cooling of the Pacific Ocean’s subtropical gyre circulations, and the time scale of fluctuations between the IPO’s warm and cold phases is approximately 15–30 yr (Folland et al. 1999).

ENSO research has identified an interdecadal signal in event amplitude and frequency that appears to be linked to the decadal variability of the Pacific Ocean (Trenberth and Hurrell 1994; Trenberth and Hoar 1997; Power et al. 1999a, b). Power et al. (1999a), for example, have shown that the IPO is statistically linked with changes in ENSO’s influence on Australian rainfall, surface temperature, river flow, and crop yield. When the IPO is in its negative (cool tropical Pacific) phase, the magnitudes of correlation coefficients between the Southern Oscillation index (SOI) and Australian climate variables are large. In contrast, when the IPO is in its positive (warm tropical Pacific) phase, the correlation coefficients between the SOI and Australian climate variables are weak. While this might suggest that the IPO plays a role in modulating ENSO impacts (Power et al. 1999a), this behavior could be due to the nonlinear relationship between ENSO and Australian rainfall; a larger La Niña SST anomaly tends to produce a larger increase in rainfall, whereas the magnitude of an El Niño SST anomaly is a poorer guide to how dry Australia will become (Power et al. 2006). Thus, the association between ENSO and Australian climate appears to be strong in decades dominated by La Niñas but weak in decades dominated by El Niños.

Our understanding of the physical mechanisms underpinning observed Pacific Ocean interdecadal SST variability is growing, but the question still remains as to whether the IPO operates independently of ENSO, and as such, many different hypotheses have been proposed to explain the variability [see Miller and Schneider (2000) and Power et al. (2006) for recent reviews]. Of the hypotheses proposed, two main types are most relevant to this paper. These are (i) that low-frequency extra-equatorial ocean exchanges with the equatorial Pacific generate SST variability in the Tropics on interdecadal time scales and (ii) that interdecadal SST variations in the tropical Pacific are primarily a consequence of higher frequency atmosphere–ocean dynamics in the equatorial region, in particular ENSO.

Several major hypotheses have been proposed concerning the role of extra-equatorial exchanges with the equatorial Pacific Ocean. (i) The first hypothesis, proposed by Gu and Philander (1997), suggests that the isopycnal (constant density) surfaces linking the midlatitude Pacific Ocean to the tropical Pacific allows the existence of a shallow meridional circulation. The mean meridional overturning circulation is proposed to transport temperature anomalies from the midlatitude Pacific to the Tropics where they alter SST on interdecadal time scales and create unstable atmosphere–ocean interactions to further amplify the signal. (ii) The second hypothesis, proposed by Kleeman et al. (1999), suggests that it is the anomalous advection of the mean temperature in this shallow meridional overturning circulation that is responsible for the tropical Pacific interdecadal variability. Support for this theory has been provided by the observational study of McPhaden and Zhang (2002). The modeling analysis of Capotondi et al. (2005) has further shown that the connection between meridional thermocline transport and equatorial SST variations holds at both interannual and decadal time scales. Nonaka et al. (2002) examined the relative importance of extra-equatorial winds on changes in meridional thermocline transport using an ocean GCM forced with observed wind stresses. They found that extra-equatorial winds (20°–8°S and 8°–25°N) are equally as important as equatorial winds (± 8° latitude) for decadal variations of equatorial SST anomalies, in agreement with the Kleeman et al. (1999) theory, while anomalous winds poleward of 20°S and 25°N do not play any significant role. (iii) The third tropical Pacific interdecadal variability hypothesis concerning oceanic processes linking extra-equatorial regions with the equatorial zone involves Rossby waves impinging on the western boundary, thereby creating coastal Kelvin waves that propagate equatorward and modulate the equatorial thermocline depth via eastward equatorial Kelvin wave propagation. This mechanism of decadal variability has been suggested in theoretical and modeling studies (Lysne et al. 1997; Liu et al. 1999; Hazeleger et al. 2001; Capotondi and Alexander 2001; Capotondi et al. 2003; Galanti and Tziperman, 2003). However, different authors have emphasized the importance of Rossby waves at different latitudes. For example, Capotondi et al. (2003) highlight the importance of Rossby waves at 13°N and 10°S, while the experimentation of Hazeleger et al. (2001) highlights the importance of Rossby waves poleward of 30° latitude. This mechanism has also been explained in a similar context to the delayed oscillator mechanism for ENSO by White et al. (2003) and Tourre et al. (2005), and the recharge/discharge paradigm by Wang et al. (2003).

High-frequency equatorial Pacific ocean–atmosphere (namely, ENSO) interactions are also identified within several different theories. For example, (i) Power and Colman (2006) have shown that ENSO-like interdecadal patterns can arise through random changes in ENSO activity, for example, in the relative number of El Nino and La Niña events in a given decade. They also showed that reddening by thermodynamic and dynamic ocean processes can help to explain the broader structure of decadal ENSO-like modes. (ii) Newman et al. (2003) proposed that ENSO forced variability could be reddened by ocean processes, thus leading to enhanced interdecadal variability in the extratropics. (iii) In the third theory, it is proposed that nonlinear mechanisms of ENSO may generate an apparent decadal signal in the Pacific Ocean (e.g., Jin et al. 1994; Tziperman et al. 1994; Timmermann et al. 2003; Rodgers et al. 2004).

Here, we focus on effectively testing the third hypotheses linking the extra-equatorial regions with the equatorial region via the propagation of planetary waves by seeking to (i) clarify the extent to which ocean dynamic equatorial variability is locally and remotely driven and (ii) determine the importance of Rossby waves incident on the western boundary in the remotely forced response. We address these questions by investigating the relative importance of equatorial wind stress forcing, off-equatorial wind stress forcing and oceanic Rossby wave reflections at the western Pacific boundary with a series of 1½ layer shallow-water model (SWM) experiments. In this study we define the equatorial region as the area within ±12.5° latitude and the off-equatorial region as those SWM latitudes outside of the equatorial region. However, first we seek to identify the importance of upper-ocean dynamics on the tropical Pacific Ocean decadal variability to determine the feasibility of using the SWM as an investigative tool. To this end, we compare the interannual and interdecadal SST variability evident in the Australian Bureau of Meteorology Research Centre (BMRC) CGCM with observations. The role of upper-ocean dynamics in this variability is then investigated using a linear first baroclinic-mode SWM forced with wind stresses from a century-time-scale CGCM simulation. The SWM includes only the important large-scale upper-ocean dynamics, specifically long Rossby waves and Kelvin waves.

This paper is organized as follows. Brief descriptions of the CGCM and SWM used in this study are provided in section 2. Section 3 details the CGCM simulation and the series of SWM experiments designed to investigate the role of equatorial wind stress forcing, off-equatorial wind stress forcing, and western boundary reflections generating the equatorial Pacific Ocean interdecadal SST and thermocline depth variability. Section 4 outlines the statistical techniques and filtering methods used to analyze the model outputs. Experiment results are presented in section 5, and the implications these results have for predictability is investigated and discussed in section 6. Discussion of the results and conclusions from the study are provided in section 7.

2. Models

a. Coupled general circulation model

The coupled general circulation model used in this study is the Australian BMRC coupled atmosphere–ocean–sea ice GCM version 2.2 (BCM2.2). BCM2.2 is designed for seasonal prediction and climate change research with a focus on the tropical Pacific. It has been shown to exhibit realistic interannual variability in the tropical Pacific and evidence of both Rossby and Kelvin wave propagation (Power et al. 1998; Bettio et al. 2003). Output from BCM2.2 has been used in a number of international intercomparisons including the Coupled Model Intercomparison Project (CMIP) Phase 2 (Power et al. 2006). The model incorporates the BMRC unified atmospheric general circulation model, which is a spectral model configured at the horizontal resolution of rhomboidal wave 21 (R21) with 17 vertical levels. The ocean model component is a global version of the Princeton University Geophysical Fluid Dynamics Laboratory Modular Ocean Model (Pacanowski et al. 1991; Power et al. 1995). The meridional grid spacing increases from 0.5° within 7° latitude of the equator to a maximum of 5.8° near the North Pole. The zonal spacing is constant at 2° over the entire globe. Full details of the model’s atmospheric, ocean, and sea ice components and some of the interdecadal variability evident in the model are described by Power et al. (2006) and Power and Colman (2006).

b. Shallow-water model

The shallow-water model (SWM), developed at Macquarie University for the MATLAB environment, is a 1½-layer model of the stratified ocean (e.g., Philander 1990, 106–108) resolved on a 1° × 1° spatial grid [the Arakawa C grid of Mesinger and Arakawa (1976)]. The upper and lower ocean density layers are separated by an interface (the pycnocline) that approximates the thermocline. The reduced gravity, g′, reflects the density difference between the upper and lower layers. We use the typical value of g′ = 0.03 m s−2 (Tomczak and Godfrey 1994, p. 37). The lower layer is assumed to be motionless and infinitely deep. We prescribe the mean depth of the upper well-mixed active layer as H = 300 m (Tomczak and Godfrey 1994, p. 37), so the first baroclinic-mode Kelvin wave speed c is 3 m s−1. Motion in the upper layer is driven by the applied wind stresses (per unit density), τ (m2 s−2), which are anomalies from the long-term monthly means (i.e., seasonal cycle removed). The associated response of the ocean is displayed by the vertical displacement of the pycnocline, η (m), and the horizontal components of the flow velocity. Considering that the pycnocline approximates the thermocline and that the thermocline depth is climatically well understood due to its role in helping to quantify the upper-ocean heat content (or warm-water volume), we use the term thermocline hereafter when discussing results from the SWM experiments. The ocean dynamics are described by the linear reduced-gravity form of the shallow-water equations detailed below [Eqs. (1)(3)]:
i1520-0442-20-11-2643-e1
i1520-0442-20-11-2643-e2
i1520-0442-20-11-2643-e3
where u and υ are the eastward and northward components of velocity respectively (m s−1) and t is time (s). The long Rossby wave speed CR(m s−1) is given by the equation, CR = β(c12/f2), where f (s−1) is the Coriolis parameter and β(m−1 s−1) is the derivative of f northward. The model time step is 2 h and Fischer’s (1965) numerical scheme is utilized for model time stepping. This model formulation permits Ekman pumping and Rossby and Kelvin wave propagation along the thermocline to be generated with appropriate large-scale wind stress forcing.

3. Experimental design

a. BCM2.2 century time-scale simulation

A century time-scale BCM2.2 simulation was performed and the output data were analyzed to determine the important modes of interdecadal SST and vertically averaged temperature (VAT) to 300-m depth variability. This CGCM simulation was left to freely evolve as an internally coupled system between the atmosphere, ocean, and sea ice. Monthly averages of the CGCM wind stresses, SST, and VAT, were interpolated onto the 1° × 1° SWM grid to allow for comparison with the SWM results. The first EOF of both raw and 13-yr Butterworth low-pass filtered BCM2.2 SST and VAT variability are first compared with observations and then further analyzed against the series of SWM experiments.

b. SWM experiments

1) SWM control simulation with CGCM wind stress forcing

Anomlous monthly averaged wind stresses (from the average annual cycle) generated by BCM2.2 were used to force the SWM configured for the low to midlatitude Pacific Ocean (41°S–41°N, 120°E–68°W). The western (eastern) Pacific boundaries approximately follow the eastern coastlines of Asia, Australia, and the western Pacific islands (the western coastline of the Americas) while the north (south) model boundaries are at 41°N (41°S). The anomalous monthly wind stresses were linearly interpolated in time and updated at each time step throughout the 100-yr SWM experiments. The three SWM output fields, the thermocline displacement and two horizontal flow velocity components, were archived as 30-day snapshots.

The thermocline depth variability generated in the SWM control simulation is analyzed and compared with the corresponding BCM2.2 VAT variability. Here, we take advantage of the robust correlation between thermocline depth and upper ocean heat content in the tropical Pacific Ocean (to 15° latitude) (Rebert et al. 1985) to usefully compare results from both models. Comparison of the results from the SWM control simulation with the BCM2.2 simulation allows us to investigate whether large-scale wind-stress-forced upper-ocean dynamics play an important and significant role in generating the SST and VAT variations simulated in the much more sophisticated CGCM, and hence whether the SWM is an appropriate tool to investigate forcing mechanisms of the Pacific Ocean’s decadal variability.

2) Experiment I: The role of equatorial and off-equatorial forcing

To quantify the role of extra-equatorial interactions with the equatorial Pacific Ocean and anomalous equatorial wind stresses on the equatorial Pacific thermocline depth variability, we perform an SWM experiment consisting of two parts that essentially separate the Pacific Ocean basin into two regions. They are the equatorial zone, which is the area within ±12.5° latitude, and the extra-equatorial region, which includes those SWM latitudes outside of the equatorial zone. Part a, titled the “equatorial wind stress forcing” (EqWF) experiment, uses the same SWM configuration as in the control simulation, but only those anomalous wind stresses in the equatorial Pacific between ±12.5° latitude are retained (wind stress anomalies are tapered to zero between 10° and 15° latitude). Part b, hereafter titled the “off-equatorial wind stress forcing” (OffEqWF) experiment, also uses the same SWM configuration as in the control simulation, but here the anomalous wind stresses are retained in the extra-equatorial zone (those latitudes poleward of ±12.5° latitude) and wind stress anomalies are tapered to zero between 15° and 10° latitude. As in the control simulation, 30-day snapshots of the thermocline depth displacements are archived over the 100-yr simulation for both of these experiments. We note here that we are seeking to identify the role of wind stress forcing and the important regions of this forcing on ocean dynamic processes and we do not attempt to address the origin of the wind stress variability.

3) Experiment II: The role of western boundary interactions

The second SWM experiment has four components (experiments IIa–d) each using the same SWM configuration and forcing wind stresses as per the OffEqWF experiment (experiment Ib). This series of experiments was designed to allow us to investigate the role of oceanic Rossby waves reflected at the Pacific Ocean western boundary on the oceanic interactions between the extra-equatorial region and the equatorial zone. Reflection of these Rossby waves explicitly relies on the conservation of mass. Thus, to prevent wave reflection at the western Pacific boundary and, hence, evaluate the importance of wave reflection on the interdecadal equatorial thermocline depth anomalies, we introduce a sizeable linear damping term, Fd (m), into the calculations of the thermocline depth between the selected latitudes along the western boundary, Eq. (4). This term acts to damp out Rossby waves impinging on the boundary, thereby preventing the coastal Kelvin waves from transporting the incoming Rossby mass equatorward. The introduction of this term into the SWM calculations affects the instantaneous conservation of mass, causing it to oscillate around the original steady state. However, the integrated effect is such that the SWM mass is conserved in time:
i1520-0442-20-11-2643-e4
The damping term (Fd) is calculated using the method of Burgers et al. (2002) for the prevention of Kelvin wave propagation around their artificial north and south boundaries. Here, however, instead of calculating the linear damping term for all longitudes at a specific latitude, the damping term at the western boundary as a function of latitude was calculated by the equation:
i1520-0442-20-11-2643-e5
where is the thermocline depth in meters, t is time in seconds, ξwb is the western boundary longitude in degrees, and γ is the nondimensional damping constant. The damping coefficient, rk, is specified by the equation:
i1520-0442-20-11-2643-e6
where rk(0) = 2.5 × 10−6 and ξcrit = ξwb + 3.

The magnitude of the damping term is increased by increasing the value of the damping constant, γ, since Rossby waves tend to impinge effectively normal to the boundary rather than parallel, as occurs with Kelvin waves along the artificial north and south boundaries. The damping constant value was tested through a series of SWM experiments in a rectangular domain (not shown) and the damping constant was found to have an optimal value of ∼1.4. Experiments IIa–d incorporate this damping term between (a) 41° and 30°S, 41° and 30°N only; (b) 41° and 20°S, 41° and 20°N only; (c) 41° and 10°S, 41° and 10°N only; and (d) 41°S and 41°N (the entire western Pacific boundary). The 30-day snapshots of the thermocline depth anomalies from each experiment were analyzed and compared with the interdecadal thermocline depth anomalies produced in the OffEqWF experiment.

4. Analysis methods

For the present study, we investigate the variability of (i) SST and VAT in BCM2.2 and (ii) the thermocline depth from the series of SWM experiments. The unfiltered model anomalies of thermocline depth, SST, and VAT are from the long-term monthly means (i.e., seasonal cycle removed). Since the IPO is defined as the third EOF of the 13-yr lowpass filtered near-global SST (Folland et al. 1999; Power et al. 1999a), interdecadal variability in the model output is analyzed using the lowpass Butterworth-filtered anomalies, with a 13-yr cutoff. As we used the reduced variable domain (near-global scale reduced down to the Pacific basin) as input to the EOF analysis, we expect the IPO-like variability in both model outputs to be the first mode variability instead of the globally observed third mode. The correlation matrix was used in the EOF analysis of both unfiltered and filtered model outputs to provide equal weighting of the different variables analyzed and a sensible comparison of results between the model fields. To reduce computational costs, the EOF analysis was performed on 2° averages of the output fields.

Each SWM experiment was perturbed from rest (i.e., zero initial state) and, as the equatorial region thermocline depth variability includes the effects of both local and remote forcing and signals in the off-equatorial region (namely, Rossby waves) of the SWM domain (±40°) can take up to ∼10 yr to cross the Pacific Ocean basin, the initial 10 yr of each SWM experiment output have been discarded. For the purposes of comparison, the first 10 yr of the BCM2.2 SST and VAT outputs have also been removed.

5. Results

a. BCM2.2 century-time-scale simulation

The first EOF (EOF1) of the unfiltered BCM2.2 SST anomalies (USST1) accounts for approximately 11% of the Pacific Ocean (41°S–41°N, 120°E–68°W) SST variance, while the first EOF of the low-pass-filtered BCM2.2 SST anomalies (FSST1) accounts for approximately 30% of the Pacific Ocean filtered variance (Fig. 1). The spatial pattern of USST1 (Fig. 1a) is dominated by positive weighting in the central and eastern tropical Pacific. The spatial pattern of FSST1 (Fig. 1b) has positive weightings across the entire tropical Pacific Ocean between the equator and approximately 20° latitude. There is also a poleward extension of the tropical Pacific positive weighting along the eastern Pacific continental boundary. Analysis of the expansion coefficients accompanying the unfiltered (filtered) spatial pattern reveals an interannual (slow interdecadal scale) oscillation between positive and negative values. These oscillations describe the timing (phase) of the pattern variations seen in both spatial maps (Figs. 1a and 1b).

The spatial pattern and expansion coefficient oscillation time scales in the unfiltered BCM2.2 SST results are in broad agreement with ENSO variability (Figs. 1a and 1c), although there are some differences. The spatial pattern reveals that the tropical Pacific SST anomalies are shifted farther west than observed, while the expansion coefficient time series indicates that the modeled ENSO is more biennial than is observed in the real world. Bettio et al. (2003) discuss the dynamics of ENSO in BCM2.2 as part of a detailed model investigation. They found that all major components of the two leading ENSO paradigms, the recharge/discharge oscillator and the delayed action oscillator (Dijkstra and Burgers 2002), are produced by BCM2.2. Overall, BCM2.2 has been shown to simulate ENSO-like SST variability across the equatorial Pacific well (Wu et al. 2002).

The spatial pattern of FSST1 (Fig. 1b) matches well with the observed IPO SST anomaly signature (Fig. 2). First, positive weightings in the tropical Pacific Ocean region of the FSST1 spatial pattern cover a very similar area to the observed IPO signature with both having maximum values on the equator at approximately 120°W. Also the positively weighted region at approximately 40°S, 120°W corresponds to similar weightings in the observed IPO SST signature. Finally, areas of zero and negative weighting are evident in the subtropical gyre regions of FSST1, which also correspond to areas of strong negative weighting apparent in the observed IPO pattern. We note that the amplitude of the negative weighting seen in the subtropical gyre regions of the observed IPO SST signature is not reproduced in FSST1. Observational records analyzed by Trenberth and Hurrell (1994) suggest that the Pacific Ocean interdecadal variability is initiated in the equatorial region as variations in tropical Pacific SST lead variations in the central North Pacific SST by approximately 3 months. As such, the observed variations in Pacific Ocean subtropical gyre SST are likely to be due to atmospheric and/or oceanic feedback processes following the initial tropical Pacific SST change. The fact that this feedback process in BCM2.2 does not produce as intense a SST variation in the gyre region is not expected to be of any real significance to our results, as we are investigating the mechanisms forcing the Pacific Ocean decadal variability and BCM2.2 does a good job in producing decadal SST variability in the tropical Pacific where observational records suggest the Pacific Ocean interdecadal variability is initiated. In essence, the spatial pattern and time series of FSST1 are consistent in overall shape and time scale with the observed IPO.

b. SWM experiments

1) SWM control simulation with CGCM wind stress forcing

EOF1 of the control simulation unfiltered SWM thermocline displacements (UC1) and low-pass-filtered thermocline displacements (FC1) can be seen in Fig. 3. UC1 (FC1) accounts for approximately 11% (27%) of the total variance of the unfiltered (filtered) thermocline depth anomalies. The spatial pattern of UC1 (Fig. 3a) has a positively weighted region covering the eastern tropical Pacific and a negatively weighted region covering the western tropical Pacific. The spatial pattern of FC1 (Fig. 3b) has a large positively weighted area covering the majority of the equatorial region and either side of the equator two bands of negative weighting exist that together make a chevron shape that is quasi symmetric about the equator. In satellite altimeter observations, this pattern is characteristic of Rossby wave propagation (e.g., Chelton and Schlax 1996). Analysis of the expansion coefficients accompanying the unfiltered (filtered) spatial pattern reveals an interannual (interdecadal) oscillation between positive and negative values in Fig. 3e (Fig. 3f).

EOF1 of both the unfiltered (UVAT1) and low-pass filtered (FVAT1) VAT anomalies in the BCM2.2 output respectively account for approximately 14% of the unfiltered and 25% of the filtered Pacific Ocean VAT variance (Figs. 3c,d). Comparison of the spatial patterns of UC1 with UVAT1 (cf. Figs. 3a and 3c) reveals several obvious similarities. Most notably, both patterns display an east–west tilt of the thermocline (and hence a zonal gradient in the upper-ocean warm-water volume) in the tropical Pacific. Comparison of the spatial patterns of FC1 with FVAT1 (cf. Figs. 3b and 3d) also reveals several obvious similarities. These dominant mode patterns arising from the two separate model outputs both have large positive weightings that extend meridionally over the majority of the tropical Pacific plus the chevron shaped bands of negative weighting in the extratropical Pacific Ocean.

Analysis of the corresponding expansion coefficients for UVAT1 (FVAT1) [Fig. 3e (Fig. 3f), dotted line], with UC1 (FC1) [Fig. 3e (Fig. 3f), solid line] shows considerable similarity between the two model outputs. The correlation coefficient between the two unfiltered (filtered) time series is 0.78 (0.51), significant above the 99% (95%) confidence level. Statistical significance of all correlation coefficients determined in this study take account of serial (auto-) correlation in the series based on the reduced effective number of degrees of freedom outlined by Davis (1976). The strong correspondence between the two first-mode EOF spatial patterns and the statistically significant correlations between the corresponding time series of each model output (Fig. 3) leads us to conclude that the SWM control simulation captures the important ocean dynamic processes that are fundamental to the generation of both interannual and interdecadal upper-ocean variability simulated in the sophisticated CGCM. As such, the SWM is a reasonable and computationally inexpensive investigative tool to explore forcing mechanisms of the Pacific Ocean interdecadal SST variability.

2) Experiment I: The role of equatorial and off-equatorial forcing

To identify the importance of wind stress forcing in the equatorial and off-equatorial regions on the Pacific Ocean climate variability, as represented by EOF1 of the unfiltered and filtered SWM control simulation results, we project the unfiltered and filtered experiment I results, parts a and b, onto the corresponding EOF1 spatial pattern of the SWM control simulation (Fig. 4). The unfiltered (filtered) results of experiment Ia, titled the EqWF experiment, project extremely well onto the unfiltered (filtered) EOF1 spatial pattern from the SWM control simulation (Figs. 4a and 4b). Comparison of the EOF1 time series produced by this projection with the corresponding EOF1 time series of the control simulation reveals a striking similarity, which is quantified by the correlation coefficient of 0.97 for the unfiltered expansion coefficients and 0.98 for the filtered expansion coefficients (both are significant above the 99% confidence level). The unfiltered experiment Ib results, titled the OffEqWF experiment, projected poorly onto the unfiltered EOF1 spatial pattern with a correlation coefficient of only 0.13 (Fig. 4d). In contrast, the filtered OffEqWF experiment results (experiment Ib) project reasonably well onto the filtered EOF1 spatial pattern of the SWM control simulation (Fig. 4c). Comparison of the filtered EOF1 time series produced by this projection with the filtered EOF1 time series from the SWM control simulation reveals a strong positive correlation of 0.67 (statistically significant above the 95% confidence level). However, the amplitude of the OffEqWF experiment expansion coefficients is noticeably smaller than that of the control simulation expansion coefficients.

Based on the results from these experiment I projections, we conclude that EqWF is most important in producing the interannual and interdecadal equatorial Pacific Ocean anomalies. However, it is clear that OffEqWF plays a smaller, but nevertheless statistically significant, role in Pacific Ocean interdecadal time-scale variability. The question still remains, however: What are the relative roles of EqWF and OffEqWF in determining the equatorial thermocline depth variability? To examine this, the filtered equatorial thermocline depth anomalies (hereafter thermocline depth anomalies are simply referred to as anomalies) generated in experiment I, parts a and b, are analyzed against the corresponding SWM control simulation equatorial anomalies. The equatorial zones for comparison are the Niño-3 (5°N–5°S, 150°–90°W) and Niño-4 (5°N–5°S, 160°E–150°W) regions.

Figure 5a confirms the results of the data projections above as the filtered EqWF anomalies in the Niño-3 region are in phase with, and account for a large portion of, the filtered control simulation variability. In contrast, the filtered OffEqWF Niño-3 region anomalies do not appear to be phase linked to the control and make up only a very small proportion of the amplitude variance of the control simulation anomalies. The filtered OffEqWF Niño-3 region anomalies have a variance of 0.02 m2 compared to the filtered control anomaly variance of 1.21 m2. The filtered EqWF Niño-3 anomalies are much more aligned to the control results with a variance of 1.20 m2 (refer to Table 1). Figure 5b shows that the filtered EqWF Niño-4 region anomalies are in phase with (r = 0.96, significant at the 99% level), and account for a large portion of, the filtered control simulation variability. This again illustrates the important contribution of EqWF on the interdecadal (filtered) anomalies in the Niño-4 region. However, the filtered OffEqWF Niño-4 region anomalies have a variance of 0.03 m2 compared to 0.29 m2 for those generated in the control simulation; that is, they explain about 10% of the control simulation variance in the Niño-4 region (refer to Table 1).

While OffEqWF appears to make a negligible contribution to the equatorial region anomalies, visual analysis of the filtered time series (Figs. 5a and 5b) shows that the anomaly differences between the EqWF experiment and the control simulation occur in phase with the OffEqWF experiment anomalies. Thus, while the OffEqWF anomalies explain only a small proportion of the total filtered SWM control simulation variance, the results are likely to be significant in terms of potential predictability. We note that the OffEqWF anomalies ignored any equatorial region coupling with the overlying wind stress, and it is highly likely that even these small perturbations of the thermocline (and thus SST) would be intensified by positive atmospheric feedbacks (in particular the Bjerknes feedback). We further note that the equatorial region wind stresses were generated in a CGCM. Hence, these stresses implicitly include the atmospheric feedbacks generated through OffEqWF equatorial Pacific SST variations, thus exaggerating the effects of the EqWF.

3) Experiment II: The role of western boundary interactions

In the previous section (experiment I) we showed that the impact of OffEqWF ocean dynamics on the equatorial thermocline depth is small, but by no means negligible. Here, we examine the role of western boundary interactions on these OffEqWF equatorial anomalies to aid the identification of the processes responsible for the low-frequency extra-equatorial ocean exchanges with the equatorial zone that act to modulate the equatorial Pacific thermocline depth. The filtered SWM equatorial anomalies produced in experiment II are presented here and compared with the corresponding anomalies generated in the OffEqWF experiment (see Fig. 6). Again, the equatorial zones used in the comparison are the Niño-3 and Niño-4 regions.

No discernable difference was seen between the anomalies for experiments IIa (not shown) and IIb (not shown) and the OffEqWF experiment (Figs. 6a–d, solid line) in either of the Niño-3 or Niño-4 regions (Fig. 6). This confirms that western boundary reflections poleward of ± 20° latitude have a negligible role to play in contributing to equatorial thermocline depth variations. In contrast, western boundary reflections within 20° latitude of the equator clearly play an important role in the OffEqWF Niño-3 and Niño-4 region thermocline depth variations. This is confirmed in Fig. 6 as the filtered Niño-3 and Niño-4 region anomalies from experiment IIc (Figs. 6a–d, dashed line) are significantly different from the corresponding OffEqWF experiment time series (Figs. 6a–d, solid line). Furthermore, the corresponding anomalies produced in experiment IId (Figs. 6a–d, dotted line) diverge even further from the OffEqWF experiment results (Figs. 6a–d, solid line).

The statistical similarity between experiments IIa, IIb, and the OffEqWF experiment are summarized in Table 2, as are the differences between experiment IIc, IId, and the OffEqWF experiment. Experiment IIc produces results with a filtered variance in the Niño-3 (Niño-4) region of 0.009 m2 (0.017 m2), while the corresponding filtered variance from the OffEqWF experiment is 0.020 m2 (0.030 m2). Given that western boundary reflections poleward of 20° latitude have a negligible role to play in feeding into Niño-3 and Niño-4 thermocline depth variations, we find that western boundary reflections produced at 20° to 10° latitude underpin approximately 50% of the filtered Niño-3 and Niño-4 region variance in the OffEqWF experiment. Furthermore, the filtered Niño-3 region variances resulting from experiments IId and IIc are different by 0.006 m2 while the filtered Niño-4 region variances from experiments IId and IIc are different by 0.009 m2. Hence, western boundary reflections between ±10° latitude are required for approximately 30% of the filtered OffEqWF Niño-3 and Niño-4 region thermocline depth variability. Thus, these results indicate that western boundary reflections are required for approximately 80% of the OffEqWF filtered Niño-3 and Niño-4 region thermocline depth anomaly variance.

6. Predictability

The experiment II results have established that processes at the Pacific Ocean western continental boundary appear to play a critical role in the transfer of mass generated from OffEqWF ocean dynamics into the equatorial Pacific, with Rossby wave reflection at the western Pacific boundary likely to be critical to this transfer. To confirm that Rossby waves generated in the extra-equatorial region transport this mass, we examine Hovmöller (time–longitude) plots of the thermocline depth variations at 16°S (Fig. 7a) and 16°N (not shown) for the last 90 years of the OffEqWF experiment. Thermocline depth variations at the equator are also examined (Fig. 7b). The 16°S section shows the dominance of westward propagating anomalies with estimated speeds of approximately 13 cm s−1, which is consistent with the Rossby wave speeds for this latitude (12.7 cm s−1). Upon reaching the western boundary, the majority of these waves appear to transfer their mass to the equatorial region (Fig. 7b). It must be noted that the equatorial response shown is produced by off-equatorial wind stress forcing in both hemispheres: thus western boundary reflections in the Northern Hemisphere and at different latitudes in the Southern Hemisphere all contribute to the OffEqWF thermocline depth variability of the equatorial region. This mass transfer is supported by the fact that the quantity of mass transferred (calculated at various times throughout the remaining 90 years of the SWM simulation) is consistent with expectations from linear theory (Kessler 1991). Additionally, an equatorial Hovmöller plot corresponding to the experiment with nonreflecting boundary conditions operating on the entire western boundary (experiment IId, figure not shown) demonstrates that thermocline depth variability in the equatorial zone is in effect absent. This reinforces the critical role that Rossby wave reflection plays at the western boundary and implies a theoretical predictability with lead times of months to years, depending on the latitude and longitude of Rossby wave generation.

Lagged correlations were calculated between the OffEqWF Niño-3 region anomalies and extra-equatorial zone anomalies along 150°W in the NH and 120°W in the SH. These longitudes were selected since most of the modeled Rossby waves that reach the western boundary at 16° latitude appear to form close to these meridians, thereby providing the longest lead times. The maximum correlation coefficient of 0.39 (0.23) corresponding to latitudes greater than 12°S (12°N) at the 120°W (150°W) meridian occurs at 14°S (16°N) where the off-equatorial anomalies lead the OffEqWF Niño-3 anomalies by 22(28) months, with the correlation coefficient significant above the 99% (99%) confidence level. These leads are more or less consistent with the time needed for oceanic Rossby waves, at the given latitude, to transfer their mass to the equatorial region via the western boundary, thereby modulating the equatorial region thermocline depth. We note, nevertheless, that the 22–28-month leads are not exclusive of further modulation of the equatorial thermocline by OffEqWF at longer time scales. To explore the maximum deterministic lead times of OffEqWF equatorial anomalies we carried out a further 17 OffEqWF SWM experiments. These experiments utilized the same SWM configuration and wind stresses as experiment Ib (the OffEqWF experiment). Here, however, the wind stresses used to force each experiment were respectively switched off at 10, 15, 20, . . . , 85, 90 yr. The amplitudes of the Niño-3 region anomalies were subsequently monitored to identify how long after the OffEqWF lapsed the upper-ocean dynamics would be able to modulate the equatorial thermocline depth. Figure 8 shows the Niño-3 region anomalies, after the wind stresses were systematically switched off, for each of the 17 experiments. Interestingly, the envelope of all of the predictability experiment Niño-3 region anomalies shows that OffEqWF can modulate equatorial region thermocline depths for up to approximately 3½ years after the wind stress forcing is switched off (Fig. 8).

The apparent OffEqWF predictability of the equatorial thermocline depth anomalies may not be of large importance given that the OffEqWF equatorial anomalies make up only a modest proportion of the total equatorial variability. However, it is quite possible that such relatively small perturbations of the thermocline depth (and SST) could be intensified by a positive atmospheric feedback (i.e., the Bjerknes feedback). Thus, in a coupled setting the importance of the predictable component may be enhanced. This might also help to explain the apparent predictability of decadal SST variations reported by Karspeck et al. (2004) using the intermediate complexity coupled model of Zebiak and Cane (1987).

7. Discussion and conclusions

In this paper the interannual and interdecadal SST variability represented in a 100-yr simulation of the BMRC CGCM (BCM2.2) was first “ground truthed” against observations to ensure consistency. The role of upper-ocean dynamics in this interdecadal variability was further explored using a linear first baroclinic-mode shallow-water model (SWM) forced with wind stresses from the century-time-scale BCM2.2 simulation, thus confirming the feasibility of using the SWM as an investigative tool to investigate the tropical Pacific Ocean’s decadal variability. The ocean dynamic mechanisms underpinning this decadal variability were then explored in response to changes in the regions of applied wind stress forcing. The relative importance of equatorial wind stress forcing (EqWF, within ±12.5° latitude), off-equatorial wind stress forcing (OffEqWF, those SWM latitudes polewards of ±12.5° latitude), and oceanic Rossby wave reflections at the western Pacific boundary were investigated. We note that, although we seek to identify the role of wind stress forcing regions on the fundamental large-scale ocean dynamics, we do not attempt to address the origin of the wind stress variability.

The OffEqWF is found to be responsible for approximately 10% (1.5%) of the overall interdecadal equatorial thermocline depth variability in the Niño-4 (Niño-3) region. While being relatively small contributions to the total equatorial region variability, the OffEqWF unfiltered (filtered) Niño-3 region anomalies are correlated at 0.86 (0.94) with the OffEqWF unfiltered (filtered) Niño-4 region anomalies with the correlation coefficient significant above the 99% (99%) confidence level. Thus, this indicates that OffEqWF equatorial region thermocline depth variability is near zonal, which is consistent with the earlier work of Hazeleger et al. (2001). This is in stark contrast to the EqWF equatorial thermocline depth variability, which produces a correlation coefficient of 0.08 (0.44) between the unfiltered (filtered) Niño-3 region anomalies and the unfiltered (filtered) Niño-4 region anomalies. Of the near-zonal equatorial thermocline depth variability produced by OffEqWF experiment, approximately 80% of the anomaly variance is transferred from Rossby waves incident upon the western boundary. Analysis of the SWM flow velocity fields in the OffEqWF experiment reveals that the remaining 20% of the equatorial anomaly variance is transmitted to the equatorial region by meridional flow in the upper layer. These results are therefore consistent with studies that identify Rossby waves incident on the western Pacific boundary as a mechanism that links the extra-equatorial Pacific Ocean with the equatorial Pacific Ocean (e.g., Capotondi and Alexander 2001; Galanti and Tziperman 2003; Wang et al. 2003; White et al. 2003; McGregor et al. 2004).

While we note that the amplitude of the OffEqWF equatorial anomalies may be underestimated by our relatively simple model experiments, since we neglect atmospheric feedbacks in the equatorial region, the significance of these results is the potential for longer-term prediction of the equatorial region’s background state. In this study we find that oscillation peaks in the Niño-3 region thermocline depth anomalies can occur up 3 yr after the OffEqWF is switched off. This suggests that some of the variability in the equatorial thermocline generated by OffEqWF can be predicted up to 3 yr ahead. We also find that thermocline depth anomalies at 120°W, 14°S (150°W, 16°N) lead thermocline depth anomalies in the Niño-3 region by approximately 22 (28) months with a correlation coefficient of 0.39 (0.23), which is significant above the 99% (99%) confidence level. The correlation between thermocline depth at 120°W, 14°S (150°W, 16°N) and those of the Niño-3 region increases to 0.58 (0.59), significant at the 98% (99%) confidence level after applying 5-yr running mean to the thermocline depth anomalies. Hence, it might therefore be possible to identify a fraction of the observed Pacific Ocean equatorial thermocline depth variability produced by OffEqWF by taking into account the amplitudes and locations of off-equatorial region Rossby waves several years prior.

It is clear from our results and the results of Hazeleger et al. (2001) that Rossby waves forced by OffEqWF do not appear to produce enough variability in the equatorial region to be the sole source of the tropical Pacific Ocean decadal variability. Further to this, our results indicate that near-equatorial wind stresses are the dominant source of variability in the control simulation equatorial thermocline depth (i.e., when wind stresses are applied over the entire basin). Again, however, it must be noted that the EqWF used in this study were generated in the CGCM. Hence, these wind stresses implicitly incorporate atmospheric feedbacks that could be generated through OffEqWF equatorial Pacific SST variations. Thus, our results give an indication of the importance of EqWF for the interdecadal modulation of tropical Pacific Ocean thermocline depth variations, but they are unable to unambiguously quantify the role of the equatorial region wind stresses. Considering that Hazeleger et al. (2001) and the present study have shown that OffEqWF, without coupling in the equatorial region, has the ability to modulate the thermocline depth of the equatorial region zonally, and that Meinen and McPhaden (2000) have shown that the equatorial region’s warm-water volume (defined as the volume of water above the thermocline) plays an important dynamical role in the oscillation of ENSO events, the following questions are raised. Is OffEqWF equatorial region thermocline depth variability, with and without coupling, enough to affect ENSO events in any way? Second, could the decadal modulations of ENSO by OffEqWF ocean dynamics account for a proportion of the apparent dominant role of EqWF? We investigate these questions in Part II of this study.

In summary, our SWM control experiment confirms that the SWM forced with BCM2.2 wind stresses captures the important ocean dynamic processes fundamental to the interdecadal SST variability produced in BCM2.2. Results from the SWM wind stress experiments demonstrate that extra-equatorial Rossby waves generated in response to off-equatorial wind stress forcing are also responsible for a small but significant fraction of the total equatorial thermocline depth variability. We have further shown through experiment II that the reflection of Rossby waves at the western boundary within approximately 20° of the equator is required to transfer virtually all of the energy generated by the OffEqWF to the equatorial region. The implications of this result for the predictability of the decadal variations of thermocline depth were investigated, and we conclude that there is potential predictability of the OffEqWF equatorial thermocline depth anomalies with lead times of up to 3 yr when taking into account the amplitudes and locations of off-equatorial region Rossby waves.

Acknowledgments

A Macquarie University Ecosystem Design Scholarship and the Macquarie University Postgraduate Research Fund supported Shayne McGregor’s participation. Dr. Power’s participation was partially sponsored by the Australian Climate Change Science Program administered by the Australian Greenhouse Office. We thank Billy Kessler, Mike McPhaden, and two anonymous reviewers for their helpful comments. We would also like to thank the Australian Center for Advanced Computing and Communication (ac3) for the essential computer time allocations.

REFERENCES

  • Bettio, L., S. B. Power, and K. Walsh, 2003: The dynamics of ENSO in a coupled GCM. Proc. Int. Conf. on Earth System Modelling, Hamburg, Germany, Max Planck Institute for Meteorology, 327 pp.

  • Burgers, G., M. A. Balmaseda, F. C. Vossepoel, G. J. V. Oldenborgh, and P. J. van Leeuwen, 2002: Balanced ocean-data assimilation near the equator. J. Phys. Oceanogr., 32 , 25092519.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., and M. A. Alexander, 2001: Rossby waves in the tropical Pacific and their role in decadal thermocline variability. J. Phys. Oceanogr., 31 , 34963515.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., M. A. Alexander, and C. Deser, 2003: Why are there Rossby wave maxima in the Pacific at 10°S and 13°N? J. Phys. Oceanogr., 33 , 15491563.

    • Search Google Scholar
    • Export Citation
  • Capotondi, A., M. A. Alexander, C. Deser, and M. J. McPhaden, 2005: Anatomy and decadal evolution of the subtropical–tropical cells (STCs). J. Climate, 18 , 37393758.

    • Search Google Scholar
    • Export Citation
  • Chelton, D. B., and M. G. Schlax, 1996: Global observations of oceanic Rossby waves. Science, 272 , 234238.

  • Davis, R. E., 1976: Predictability of sea surface temperature and sea level pressure anomalies over the North Pacific Ocean. J. Phys. Oceanogr., 6 , 249266.

    • Search Google Scholar
    • Export Citation
  • Dijkstra, H. A., and G. Burgers, 2002: Fluid dynamics of El Niño variability. Annu. Rev. Fluid Mech., 34 , 531558.

  • Fischer, G., 1965: On a finite-difference scheme for solving the primitive equations for a barotropic fluid with application to the boundary problem. Tellus, 17 , 405412.

    • Search Google Scholar
    • Export Citation
  • Folland, C. K., D. E. Parker, A. W. Coleman, and R. Washington, 1999: Large scale modes of ocean surface temperature since the late nineteenth century. Beyond El Niño: Decadal and Interdecadal Climate Variability, A. Navarra, Ed., Springer-Verlag, 73–102.

  • Folland, C. K., J. A. Renwick, M. J. Salinger, and A. B. Mullan, 2002: Relative influences of the Interdecadal Pacific Oscillation and ENSO on the South Pacific convergence zone. Geophys. Res. Lett., 29 .1643, doi:10.1029/2001GL014201.

    • Search Google Scholar
    • Export Citation
  • Galanti, E., and E. Tziperman, 2003: A midlatitudes–ENSO teleconnection mechanism via baroclinically unstable long Rossby waves. J. Phys. Oceanogr., 33 , 18771888.

    • Search Google Scholar
    • Export Citation
  • Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend on exchanges between the Tropics and extratropics. Science, 275 , 805807.

    • Search Google Scholar
    • Export Citation
  • Hazeleger, W., M. Visbeck, M. Cane, A. Karspeck, and N. Naik, 2001: Decadal upper ocean temperature variability in the tropical Pacific. J. Geophys. Res., 106 , 89718988.

    • Search Google Scholar
    • Export Citation
  • Jin, F. F., J. D. Neelin, and M. Ghil, 1994: El Niño on the Devil’s Staircase: Annual subharmonic steps to chaos. Science, 264 , 7072.

    • Search Google Scholar
    • Export Citation
  • Karspeck, A. R., R. Seager, and M. A. Cane, 2004: Predictability of tropical Pacific decadal variability in an intermediate model. J. Climate, 17 , 28422850.

    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., 1991: Can reflected extra-equatorial Rossby waves drive ENSO? J. Phys. Oceanogr., 21 , 444452.

  • Kleeman, R., J. J. P. McCreary, and B. A. Klinger, 1999: A mechanism for generating ENSO decadal variability. Geophys. Res. Lett., 26 , 17431746.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., L. Wu, and E. Bayler, 1999: Rossby wave–coastal Kelvin wave interaction in the extratropics. Part I: Low-frequency adjustment in a closed basin. J. Phys. Oceanogr., 29 , 23822404.

    • Search Google Scholar
    • Export Citation
  • Lysne, J., P. Chang, and B. Giese, 1997: Impact of extratropical Pacific on equatorial variability. Geophys. Res. Lett., 24 , 25892592.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis, 1997: A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Amer. Meteor. Soc., 78 , 10691079.

    • Search Google Scholar
    • Export Citation
  • McGregor, S., N. J. Holbrook, and S. B. Power, 2004: On the dynamics of interdecadal thermocline depth and sea surface temperature variability in the low to mid-latitude Pacific Ocean. Geophys. Res. Lett., 31 .L24201, doi:10.1029/2004GL021241.

    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415 , 603608.

    • Search Google Scholar
    • Export Citation
  • Meinen, C. S., and M. J. McPhaden, 2000: Observations of warm water volume changes in the equatorial Pacific and their relationship to El Niño and La Niña. J. Climate, 13 , 35513559.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and A. Arakawa, 1976: Numerical methods used in atmospheric models. GARP Publication Series 17, 64 pp.

  • Miller, A. J., and N. Schneider, 2000: Interdecadal climate regime dynamics in the North Pacific Ocean: Theories, observations and ecosystem impacts. Prog. Oceanogr., 47 , 355379.

    • Search Google Scholar
    • Export Citation
  • Newman, M., G. P. Compo, and M. A. Alexander, 2003: ENSO-forced variability of the Pacific decadal oscillation. J. Climate, 16 , 38533857.

    • Search Google Scholar
    • Export Citation
  • Nonaka, M., S-P. Xie, and J. P. McCreary, 2002: Decadal variations in the subtropical cells and equatorial Pacific SST. Geophys. Res. Lett., 29 .1116, doi:10.1029/2001GL013717.

    • Search Google Scholar
    • Export Citation
  • Pacanowski, R. C., K. Dixon, and A. Rosati, 1991: The GFDL Modular Ocean Model users guide, version 1.0. GFDL Ocean Group Tech. Rep. 2, 376 pp.

  • Philander, S. G. H., 1990: El Niño, La Niña, and the Southern Oscillation. Academic Press, 289 pp.

  • Power, S. B., and R. Colman, 2006: Multi-year predictability in a coupled general circulation model. Climate Dyn., 26 , 247272.

  • Power, S. B., R. Kleeman, F. Tseikin, and N. R. Smith, 1995: A global version of the GFDL modular ocean model for ENSO studies. BMRC Tech. Rep., 18 pp.

  • Power, S. B., F. Tseikin, A. W. Coleman, and A. Sulaiman, 1998: A coupled general circulation model for seasonal prediction and climate change research. BMRC Research Rep. 66, 52 pp.

  • Power, S. B., T. Casey, C. Folland, A. Colman, and V. Mehta, 1999a: Inter-decadal modulation of the impact of ENSO in Australia. Climate Dyn., 15 , 319324.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., F. Tseitkin, V. Mehta, B. Lavery, S. Torok, and N. J. Holbrook, 1999b: Decadal climate variability in Australia during the twentieth century. Int. J. Climatol., 19 , 169184.

    • Search Google Scholar
    • Export Citation
  • Power, S. B., M. Haylock, R. Colman, and X. Wang, 2006: The predictability of interdecadal changes in ENSO activity and ENSO teleconnections. J. Climate, 19 , 47554771.

    • Search Google Scholar
    • Export Citation
  • Rebert, J. P., J. R. Donguy, and G. Eldin, 1985: Relations between sea level, thermocline depth, heat content and dynamic height in the tropical Pacific Ocean. J. Geophys. Res., 90 , 1171911725.

    • Search Google Scholar
    • Export Citation
  • Rodgers, K. B., P. Friederichs, and M. Latif, 2004: Tropical Pacific decadal variability and its relation to decadal modulations of ENSO. J. Climate, 17 , 37613774.

    • Search Google Scholar
    • Export Citation
  • Timmermann, A., F. F. Jin, and J. Abshagen, 2003: A nonlinear theory for El Niño bursting. J. Atmos. Sci., 60 , 152165.

  • Tomczak, M., and S. J. Godfrey, 1994: Regional Oceanography: An Introduction. Pergamon Press, 422 pp.

  • Tourre, Y. M., C. Cibot, L. Terray, W. B. White, and B. Dewitte, 2005: Quasi-decadal and inter-decadal climate fluctuations in the Pacific Ocean from a CGCM. Geophys. Res. Lett., 32 .L07710, doi:10.1029/2004GL022087.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and J. W. Hurrell, 1994: Decadal atmosphere-ocean variations in the Pacific. Climate Dyn., 9 , 303319.

  • Trenberth, K. E., and T. J. Hoar, 1997: El Niño and climate change. Geophys. Res. Lett., 24 , 30573060.

  • Tziperman, E., L. Stone, M. A. Cane, and H. Jarosh, 1994: El Niño chaos: Overlapping of resonances between the seasonal cycle and the Pacific ocean-atmosphere oscillator. Science, 264 , 7274.

    • Search Google Scholar
    • Export Citation
  • Wang, X., F. F. Jin, and Y. Wang, 2003: A tropical recharge mechanism for climate variability. Part II: A unified theory for decadal and ENSO modes. J. Climate, 16 , 35993616.

    • Search Google Scholar
    • Export Citation
  • White, W. B., Y. M. Tourre, M. Barlow, and M. Dettinger, 2003: A delayed action oscillator shared by biennial, interannual, and decadal signals in the Pacific Basin. J. Geophys. Res., 108 .3070, doi:10.1029/2002JC001490.

    • Search Google Scholar
    • Export Citation
  • Wu, Z-J., R. Coleman, S. B. Power, X. Wang, and B. J. McAvaney, 2002: The El Niño Southern Oscillation response in the BMRC Coupled GCM. BMRC Research Rep. 91, 18 pp.

  • Zebiak, S. E., and M. A. Cane, 1987: A model El Niño–Southern Oscillation. Mon. Wea. Rev., 115 , 22622278.

Fig. 1.
Fig. 1.

The first EOFs of the (a) unfiltered (11% variance) and (b) filtered (30% variance) BCM2.2 SST (respectively USST1 and FSST1) and (c), (d) their corresponding expansion coefficient time series.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 2.
Fig. 2.

The SST signature of the IPO during its warm tropical Pacific phase. The contour interval is 0.04°C: negative contours are dashed (figure modified from Folland et al. 2002).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 3.
Fig. 3.

The first EOFs of the (a) unfiltered thermocline depth anomalies from the SWM control simulation (11% variance), (b) filtered thermocline depth anomalies from the SWM control simulation (27% variance), (c) unfiltered BCM2.2 VAT (14% variance), and (d) filtered BCM2.2 VAT (25% variance). (e), (f) The corresponding expansion coefficient time series for the SWM results (black line) and the BCM2.2 results (gray line).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 4.
Fig. 4.

The principal component time series for projections of (a) unfiltered EqWF; (b) filtered EqWF; (c) filtered OffEqWF; and (d) unfiltered OffEqWF, shown in black. The corresponding expansion coefficients for the SWM control simulation are shown in gray.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 5.
Fig. 5.

Filtered SWM thermocline depth anomalies in the (a) Niño-3 and (b) Niño-4 regions.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 6.
Fig. 6.

Filtered SWM thermocline depth variations in the (a) Niño-3 and (b) Niño-4 regions.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 7.
Fig. 7.

Hovmöller plots of the unfiltered thermocline depth anomalies (m) along (a) 16°S and (b) 0° latitude generated in the OffEqWF experiment (expt Ib). Note the 16°S section has the x axis reversed so that it reads east-to-west from left-to-right. Contour intervals are provided every 8 m with minima at ±4 m, and every 0.4 m with minima at ±0.2 m, respectively; negative contours are dashed.

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Fig. 8.
Fig. 8.

Niño-3 region thermocline depth anomalies for each of the 17 SWM predictability experiments once the off-equatorial wind stress forcing was switched off (see legend inset).

Citation: Journal of Climate 20, 11; 10.1175/JCLI4145.1

Table 1.

Variance (m2) of the unfiltered and filtered thermocline depth anomalies from expt I.

Table 1.
Table 2.

Variance (m2) of the unfiltered and filtered thermocline depth anomalies from expt II.

Table 2.
Save