Interdecadal Dynamics of the North Pacific Ocean

Guillermo Auad Climate Research Division, Scripps Institution of Oceanography, La Jolla, California

Search for other papers by Guillermo Auad in
Current site
Google Scholar
PubMed
Close
Full access

We are aware of a technical issue preventing figures and tables from showing in some newly published articles in the full-text HTML view.
While we are resolving the problem, please use the online PDF version of these articles to view figures and tables.

Abstract

An isopycnal ocean model forced by NCEP–NCAR reanalysis wind stresses and heat fluxes is used to study the interdecadal variability of the Pacific Ocean in the 1958–97 period. A reasonable agreement is found between the model's modes of variability and those obtained by other researchers from both 100 years of observations and theoretical predictions. In agreement with previous observational work, decadal and interdecadal timescales have different descriptions, and from this study it is suggested that they indeed have different dynamics. This study focuses on the dynamics of the ocean's interdecadal variability, that is, of timescales of about 20 yr. The decadal timescale, that is, 10 yr, is briefly outlined and compared with previous studies. It is found that atmospheric heat fluxes play a key role in establishing the interdecadal SST pattern in the midlatitudinal North Pacific. These fluxes would excite a high baroclinic mode, igniting a series of events that move around the basin. In midlatitudes, interdecadal SSTs are most sensitive to the heat flux forcing along the eastern boundary north of about 30°N, in the western North Pacific at about 40°N, and along 20°N eastward of the date line; in the eastern North Pacific and north of 40°N, interdecadal pycnocline anomalies move across the Gulf of Alaska and toward the Aleutian Islands up to about the Kamchatka Peninsula, continuing to the southwest down to about 28°N. In their path, pycnocline oscillations induce SST changes in the Kuroshio–Oyashio Extension. On the other hand, in the eastern Tropics, the wind stress curl would induce interdecadal pycnocline oscillations that (between 10° and 20°N) propagate as Rossby waves, similar to those observed there for annual and interannual timescales, which, after crossing the date line, turn toward the north-northwest. All of these waves and/or events move or propagate within areas where the mean flow is of smaller amplitude than the phase speed in the direction of motion. In addition, the results presented here would suggest that a process similar to a servomechanism, and as envisaged by other authors, is present along 40°N, suggestive of an active ocean–atmosphere interaction over this area. Major differences are found between decadal and interdecadal dynamics.

Corresponding author address: Dr. Guillermo Auad, Climate Research Division, Scripps Institution of Oceanography, 9500 Gilman Drive, Dept. 0224, La Jolla, CA 92093-0224. Email: guillo@ucsd.edu

Abstract

An isopycnal ocean model forced by NCEP–NCAR reanalysis wind stresses and heat fluxes is used to study the interdecadal variability of the Pacific Ocean in the 1958–97 period. A reasonable agreement is found between the model's modes of variability and those obtained by other researchers from both 100 years of observations and theoretical predictions. In agreement with previous observational work, decadal and interdecadal timescales have different descriptions, and from this study it is suggested that they indeed have different dynamics. This study focuses on the dynamics of the ocean's interdecadal variability, that is, of timescales of about 20 yr. The decadal timescale, that is, 10 yr, is briefly outlined and compared with previous studies. It is found that atmospheric heat fluxes play a key role in establishing the interdecadal SST pattern in the midlatitudinal North Pacific. These fluxes would excite a high baroclinic mode, igniting a series of events that move around the basin. In midlatitudes, interdecadal SSTs are most sensitive to the heat flux forcing along the eastern boundary north of about 30°N, in the western North Pacific at about 40°N, and along 20°N eastward of the date line; in the eastern North Pacific and north of 40°N, interdecadal pycnocline anomalies move across the Gulf of Alaska and toward the Aleutian Islands up to about the Kamchatka Peninsula, continuing to the southwest down to about 28°N. In their path, pycnocline oscillations induce SST changes in the Kuroshio–Oyashio Extension. On the other hand, in the eastern Tropics, the wind stress curl would induce interdecadal pycnocline oscillations that (between 10° and 20°N) propagate as Rossby waves, similar to those observed there for annual and interannual timescales, which, after crossing the date line, turn toward the north-northwest. All of these waves and/or events move or propagate within areas where the mean flow is of smaller amplitude than the phase speed in the direction of motion. In addition, the results presented here would suggest that a process similar to a servomechanism, and as envisaged by other authors, is present along 40°N, suggestive of an active ocean–atmosphere interaction over this area. Major differences are found between decadal and interdecadal dynamics.

Corresponding author address: Dr. Guillermo Auad, Climate Research Division, Scripps Institution of Oceanography, 9500 Gilman Drive, Dept. 0224, La Jolla, CA 92093-0224. Email: guillo@ucsd.edu

1. Introduction

Several components make up the climate variability of the world ocean–atmosphere system. These components can have a preferred timescale, for example, El Niño events occur every 2–7 years or they can have a red spectrum. In reality, the picture is more complicated because in some scenarios the ocean responds with a preferred timescale to stochastic atmospheric forcing, and whether this response is linked, somehow, to coupled ocean–atmospheric deterministic modes is currently a topic of active research.

The basic ideas of oceanic–atmospheric interaction for subdecadal timescales were early envisaged by Bjerknes (1964) for the Atlantic Ocean, and unveiled in the Pacific by Latif and Barnett (1994). According to them, the atmospheric response to this anomaly involves a weakened Aleutian low that will adjust the underlying ocean, further increasing the SST positive anomaly: that is, a positive feedback. However, part of the atmospheric response also includes a wind stress curl anomaly, which tends to spin down the subtropical gyre, thus leading to an oscillatory-type behavior. The signature of the 20-yr cycle in sea surface temperature (SST) has a maximum in the western subpolar gyre between 45° and 50°N (e.g., Pierce et al. 2001), which would owe its existence to an increased poleward transport of positive SSTs by the Kuroshio Current and its extension. White and Barnett (1972) showed that ocean and atmosphere can be locked to each other by a continuous exchange of vorticity between both media. In the framework of their theory, coupled oceanic–atmospheric Rossby waves emerge as the linking process that allows this active air–sea interaction. There is more recent modeling evidence of oceanic–atmospheric coupling at interdecadal timescales in the North Pacific (e.g., Barnett et al. 1999; Pierce et al. 2001; Miller and Schneider 2000), and depending on the authors and data used, its timescale ranges from 18 to 30 yr [e.g., Robertson (1996) finds an 18-yr oscillation in his 500-yr coupled model run and a 30-yr signal in the Global Sea Ice Coverage and Sea Surface Temperature (GISST) dataset]. However, Frankignoul et al. (1997) argue in favor of a red spectrum with no preferred timescale and attribute those spectral peaks to too short time series.

The interdecadal dynamics in the North Pacific Ocean involves several oceanic and atmospheric components, and one of them, sea level pressure (SLP) variations in the Aleutian low region, leads to almost simultaneous changes in the westerlies (i.e., at around 40°N) and in the associated wind stress curl pattern over the whole midlatitudinal North Pacific with maximum amplitudes north of 40°N (Latif and Barnett 1996). These changes in the wind stress curl field have been associated (e.g., Miller and Schneider 2000) with two modes of decadal to subdecadal variability in the midlatitudinal Pacific Ocean (Deser and Blackmon 1995). The latter authors refer to one of them as the PNA (Pacific–North American pattern), typically a sea level pressure pattern thought to be forced from the Tropics, in which the stochastic atmospheric forcing, mainly through zonal wind stress anomalies, induces a maximum oceanic response in SST in the central North Pacific region at around 35°N (e.g., Miller and Schneider 2000). The other mode, the Pacific (inter) decadal oscillation (PDO; Mantua et al. 1997), typically defined in terms of SST, has its maximum amplitude in the western subpolar gyre (between 35° and 45°N depending on the dataset or model used), and has been associated with coupled phenomena between the ocean and atmosphere and has a typical timescale of 20–50 yr.

There are still many questions to be answered about the processes involved in the interdecadal dynamics of the North Pacific Ocean. However, several of them need to be answered first if one pretends to offer a reasonable description of the processes that make up the oceanic–atmospheric system: (i) it is still unknown how different oceanic regions are dynamically linked to each other for example, through the action of Rossby waves [the modeling study of Barnett et al. (1999) at least indicates that such a connection exists] and (ii) there is still debate in the scientific community as to whether there is a coupled response. In any case it will remain to be described how those oceanic and atmospheric feedback processes take place and as to whether which gyre, subpolar or subtropical, responds more vigorously to changes in SLP of the Aleutian low. Some ideas have been advanced on how the ocean connects back to the atmosphere (Peng et al. 1997), but this is still uncertain (Miller and Schneider 2000).

There is some consensus on how the atmosphere forces the ocean. First, SLP variations of the Aleutian low are almost immediately followed by changes in the wind stress and wind stress curl since for these timescales winds are dominantly geostrophic. From there, oceanic-only processes (wave propagation and/or advection) alter the heat budget of the western subpolar gyre and of the central North Pacific. As noted by Mantua and Hare (2002), very little is known about the PDO dynamics, to the point that even its geographical extent is still uncertain. Some progress has been made recently with the construction of proxy SST records from coral and tree-ring data (Linsley et al. 2000; Evans et al. 2001, respectively). These studies strongly suggest that the geographical extent of the PDO goes far beyond the midlatitudinal North Pacific Ocean, at least including the Tropics and the Southern Hemisphere. Studies on understanding the PDO are relevant due to the impact that regime shifts can have on society through changes in weather patterns (Cayan et al. 2001; Minobe 2000; Dettinger et al. 2000) and in the abundance and distribution of commercial biota species (Anderson and Piatt 1999; Beamish 1993; T. Baumgartner et al. 2003, personal communication).

This study is motivated by the recent findings of Tourre et al. (2001), who found, from 100 years of SST and SLP observations, that decadal (7–13-yr band) and interdecadal (periods longer than 13 yr) timescales have different descriptions, and are statistically independent. Their interdecadal SST signal is led by SLP by a few years, unlike the decadal band in which changes in both fields take place simultaneously. Their interdecadal signal thus has a well-defined structure and evolution and has, indeed, an important contribution to the SST variability at 15°N. They speculate that ENSO variability could be modulated by this signal, as noted earlier by Kirtman and Schopf (1998), while their decadal SST mode exhibits a structure and evolution that resembles those of ENSO. Our main goal is to investigate the dynamics behind those two timescales with particular emphasis on interdecadal oceanic processes in midlatitudes. However, we will also address a few aspects related to decadal variability since it is necessary to differentiate between both timescales.

A crucial task in this process will be to attempt to link, in one consistent picture, the different components of the interdecadal timescale that were studied with preferred detail by different authors. These include the Kuroshio Extension decadal variability (Deser et al. 1999), the Aleutian low system (Peng et al. 1997), the wind stress curl forcing of the North Pacific (Miller et al. 1998), and gyre-scale advection (Zhang and Levitus 1997). Comprehensive summaries of decadal and/or interdecadal variability in the Pacific Ocean were recently presented by Mantua and Hare (2002) and Schneider et al. (2002). This task, in turn, will hopefully lead to a better understanding of the Pacific climate variability, a key ingredient in any predictability study.

This paper is organized as follows. In section 2 we describe the numerical and statistical methods used in this study, in section 3 we present the results of applying those methods, and section 4 is left for discussions. In section 5 the conclusions are summarized.

2. Methods

a. Numerical methods: The OPYC model

The primitive equation ocean model, known as OPYC, was developed by Oberhuber (1993) and applied by Miller et al. (1994a,b, 1997, 1998), Cayan et al. (1995), and Auad et al. (1998a,b) to study monthly through decadal-scale ocean variations over the Pacific basin. Here we use an updated version of the model with higher resolution than the nominally 4° resolution grid used previously and a revised scheme for forcing with monthly mean fluxes. The grid extends from 67.5°S to 67.5°N and 119°E to 70°W, with periodic boundary conditions along the latitudes of the Antarctic Circumpolar Current. The model is constructed with 10 isopycnal layers (each with nearly constant potential density but variable thickness, temperature, and salinity) that are fully coupled to a bulk surface mixed-layer model and to a sea ice model with rheology. The resolution is 1.5° in the midlatitude open ocean with zonal resolution gradually increased to 0.65° resolution within a 10° band across the equator.

The forcing functions consist of a seasonal cycle and of National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis anomalies that are added to it. The monthly mean seasonal cycle are from various sources and are the same as Miller et al. (1994a,b). The wind stress climatology is derived from a combination of European Centre for Medium-Range Weather Forecasts (ECMWF) midlatitude fields and Hellerman–Rosenstein tropical climatology. The monthly mean seasonal cycle climatology of turbulent kinetic energy input to the mixed layer is estimated from the same datasets Oberhuber (1993). The monthly mean seasonal cycle climatology of total surface heat flux is computed during spinup (no anomalous forcing) by evaluating bulk formulas that use model SST with ECMWF-derived atmospheric fields (air temperature, humidity, cloudiness, etc.); the daily mean seasonal cycle is then saved (averaged over the last 10 yr of a 99-yr spinup) and subsequently used as specified forcing during the anomalously forced hindcasts.

Anomalous fields of wind stress, total surface heat flux, and TKE are then added to the respective mean seasonal cycles. Because there is no SST feedback to any of the anomalous forcing fields, the model is not constrained to reproduce the observed temperature variations. Near the equator, the anomalous heat fluxes are both poorly known and generally serve as a damping mechanism (but see Seager et al. 1995). Thus, Newtonian damping is employed within a 7° e-folding scale around the equator, where the SST anomalies are damped back to model climatology with 1–4-month timescales [see Barnett et al. (1991) for a map of the coupling coefficient]. The monthly forcing strategy of Auad et al. (1998a,b) was used to properly weight monthly mean forcing anomalies (Killworth 1996).

b. Statistical methods

After forcing the OPYC model with anomalous heat and momentum fluxes, different variables are averaged and saved every 15 days from 1958 to 1997. The mean climatologies are subtracted from these total fields in order to obtain the anomalous fields. Then, every time series at every grid point, for every variable (e.g., SST and SLP), is low passed (Kaylor 1977) to eliminate all oscillations with periods shorter than 6 yr.

Given the coupled nature of the signals that we aim to study, we take advantage of the properties of the singular value decomposition (SVD) technique (Bretherton et al. 1992) to separate our signal of interest. SVDs decompose the covariance between two fields in the same way that EOFs decompose the variance of one single field. Mathematically, SVDs are simply the “EOFs” of a nonsquare matrix. It is thus possible that a first SVD mode can account for say, 90% of the covariance, but that the variance of one or both of the variables involved in the computation could not be a major contributor to those variables' variances. In the present study, each mode's variance and covariance decreases with increasing mode number.

3. Results

a. Model to observations comparison

To partly validate our model solutions we use Reynolds SSTs (Reynolds and Smith 1994). Correlation coefficients and standard deviations were computed for model and observations for each grid point using SST winter anomalies. The results are displayed in Fig. 1 for periods longer than 6 yr. The spatially averaged 95% confidence level is (Davis 1976) 0.41. The value of the correlation coefficients shown in Fig. 1 does not change significantly, and a 10%–20% reduction is noticed on average if no bandpassing is applied. Low correlation areas do not necessarily imply a poor model performance but are sometimes associated with low sampling of the observations, for example, as in Auad et al. (2001) and Huddleston et al. (2003, manuscript submitted to Quart. J. Roy. Meteor. Soc.). The area without contours (top panel), at about 20°N, is a good example of this since the sampling density there is among the lowest in the North Pacific (see Auad et al. 2001, their Fig. 1). Emphasizing this concept is the fact that when Reynolds SSTs (which include satellite data in addition to ship data) were replaced by the Comprehensive Ocean–Atmosphere Data Set (COADS) or Scripps Institution of Oceanography (SIO) SSTs, the correlation coefficients were about 10%–20% lower depending on the location.

The standard deviation maps show, for both model and observations, maximum values in the 40°–50°N band west of 150°W and along the eastern margin of the basin. As noted by Pierce et al. (2001), from a suite of coupled models, they have a tendency to trap SST variance too close to the western margin of the basin, probably due to the poor resolution of western boundary dynamics. In fact, the middle panel of Fig. 1 is very similar to the spectral densities computed by them for their 20-yr signal in both model and observations (their Fig. 8). In agreement with Auad et al. (2001), the midlatitudinal model SSTs are an overestimate of the observed ones by a factor of about 1.5 on a spatial average. This can certainly be related to the model's tendency to slightly overestimate the amount of reemerged SST (Alexander et al. 1999). The correlations of Fig. 1 were about 15% higher when data from the period 1982–97 (more sampling density because of the availability of satellite data) were used, while heat storage correlations (upper 400 m) were analyzed by Auad et al. (1998a,b) for ENSO and decadal variability. Confirming the findings of Miller and Schneider (2000), the model SST variability in the central North Pacific (CNP) region (defined by them as 30°–40°N, 180°–210°E) leads the SST variability in the Kuroshio–Oyashio Extension (KOE) region (defined by them as 35°–40°N, 150°E–180°) by 2–4 years (Fig. 2).

Unlike coupled models, oceanic–atmospheric feedbacks are absent in ocean models. However, a proper modeling of the SST and mixed layer depth, assuming a decent realism in the atmospheric forcing functions, will result in implicit feedbacks (if any in nature). The above model/observation comparisons (Fig. 1), along with the results of the NCEP to COADS comparison of wind stresses and heat fluxes in midlatitudes (Auad et al. 2001) would suggest that, if feedback processes are present in nature, then they will be implicitly and approximately reproduced by the solutions of the ocean model.

b. Model mean flow in the KOE and vicinity

A description of the mean-flow structure in the KOE area and its vicinity is necessary in order to advance an understanding of the decadal and interdecadal dynamics of the North Pacific Ocean. It will be shown later that mean currents have a key role in establishing preferred paths of propagation in the interdecadal band. In addition, the KOE area will be shown to be a very important component of the decadal and interdecadal dynamics, though for different reasons in both bands. Based on the convergence latitude of the Kuroshio and Oyashio and/or on the location of the maximum SST variance, different authors have defined the Kuroshio–Oyashio Extension with different bounding latitudes and longitudes. For instance, Seager et al. (2001) define the KOE boundaries in the box 37.5°–42.5°N, 150°E–180°; Miller and Schneider (2000) in 35°–40°N, 150°E–180°; Qiu and Kelly (1993) in 30°–40°N, 141°–175°E; and Qiu (1995) in 25°–40°N, 136°E–180°, which also includes the southern recirculation area. Lysne and Deser (2002) define three different KOE areas for three different datasets according to the location of the maximum SST variance: 34°–44°N, 140°–175°E for the Navy Coastal Ocean Model (NCOM) data; 30°–41°N, 140°–175°E for the SIO dataset; and 31°–42°N, 140°–175°E for the World Ocean Atlas 1998 (WOA98) dataset.

Surface velocities from the ocean model were used to compute the mean flow fields (Fig. 3). An eastward flow is evident in the 30°–47°N band with the axis of the KOE being aligned along the 37°N parallel, approximately. The model Kuroshio and Oyashio converge at about 40°N, which is in fair agreement with available observations (e.g., Reed et al. 1994; Stabeno and Reed 1994). The model KOE axis is displaced about 2° to the north with respect to the real ocean but this is reasonable given the model resolution of 1.5°. Thus, based on the location of the eastward flow (Fig. 3) and on the location of maximum SST variance (Fig. 1, top panel), we define the KOE within 39°–44°N, 160°E–180°, which is within the boundaries given by the authors cited above. The eastward flow at and nearby the KOE is approximately 20% too broad and 50% slower when compared with the recent observations of Maximenko et al. (2002). The ocean model speeds are closer to the observed values than those obtained from coupled models and it will be shown later that, in order to obtain a reasonable representation of the interdecadal ocean's dynamics, both the mean flow and the difference between the mean flow and Rossby wave speeds need to be reasonably simulated.

c. SVD analysis of SSTs, wind stress curl, pycnocline depth, and surface currents

1) Basic modal description

The active participation of all heat budget components has shown that the North Pacific Ocean has a complex interplay of different physical processes from interannual (Auad et al. 1998a) to interdecadal (Auad et al. 1998b) timescales. The latter, plus the basin-scale averages described in the previous section, in the light of the findings of both Tourre et al. (2001) and White et al. (1997), were suggestive of an active coupling between ocean and atmosphere. It thus becomes necessary to better isolate the signal of interest (any dominant coherent mode with a timescale in the 18–30-yr range), which we know, given the complexity just mentioned, involves a suite of mutually interacting variables: oceanic and atmospheric.

The North Pacific (all analyses are from 10°S to 70°N) SST, wind stress curl, and pycnocline depth winter anomalies were low passed (Kaylor 1977) with a cutoff period of 6 yr, and detrended. We took advantage of the model architecture, specifically its isopycnal character, and integrated layer thicknesses from the surface down to the bottom of layer four to obtain a well-behaved proxy of pycnocline depth. In what follows, we will refer to this variable as “h4,” which in midlatitudes varies from about 150 m (subarctic) to 500 m (center of the subtropical gyre). The first two SVD temporal modes of the SST/h4 pair are shown in Fig. 4. The first mode explains 92% of the squared covariance, while the second mode explains 5%. However, the SST variance of the second mode is about half of the first mode variance. The correlation coefficients between time series, an indicator of the degree of coupling between both fields in each mode, is higher than 0.9 for both modes in Fig. 4. One hundred Monte Carlo simulations were carried out in which random noise time series, in the same number and length as those used in the analysis of our signal, were identically treated, that is, bandpassed, Hilbert transformed, and ultimately decomposed into SVD modes. The resulting 95% confidence levels for the first three modes are 20.2%, 18.7%, and 17.6%, respectively. Our approach is to obtain the portion of the variance of the wind stress curl and of the pycnocline depth (h4) that is significantly (statistically) coupled to the first mode SST. This will help to isolate the signal from the noise. The decomposition in singular values not only yielded orthogonal modes but, without any requirement being imposed, separated the first two modes by timescales. The first mode has a clear interdecadal timescale (roughly 20 yr), while the second mode has a timescale of approximately 10 yr.

2) The spatial structure of the modal oceanic fields: SST

Figure 5 shows the first SVD mode, amplitude, and phase of the SST/h4 pair (the same amplitude, phase, and time behavior were obtained for the first mode of the SST field when it was decomposed into SVD modes and paired to either h4, surface velocity, wind stress curl, surface heat fluxes, horizontal advection, or vertical mixing). A common feature to both fields is that, in midlatitudes mainly west of the date line, the phase changes mostly in the meridional direction. Figure 6 shows the time sequence of the reconstructed first SVD modes of the SST (left column), h4 (middle column), and of surface velocity (right column) fields. We have chosen to show the time frame January 1974–January 1984 (from January 1976 for the velocity field) to display, every two years, the changing features of the well-documented (e.g., Trenberth 1990; Miller et al. 1994a) 1976–77 climate “shift,” whose time span is 10 years and approximately covers one-half of a cycle of the first mode's timescale.

The warm, deep (h4) waters occupying the northern part of the KOE in 1974 are replaced by cooler and shallower water depths (Fig. 6, middle column), while along the eastern boundary, north of 30°N, SSTs change sign and are out of phase with those in the KOE. This warming along the U.S. and Canadian coasts is accompanied by warming conditions elsewhere in the basin except the KOE and the subpolar region west of the date line. Not only is there a good resemblance between the interdecadal SST spatial structure of our first SVD mode and that one of Tourre et al. (2001), but there is also a good correspondence with the EOF modes obtained by Miller et al. (1994b), Nakamura et al. (1997), and Deser and Blackmon (1995). This pattern also bears a close resemblance to that one estimated by Mestas-Nuñez and Enfield (1999), their REOF5, and to the PDO mode of Mantua et al. (1997).

3) The spatial structure of the modal oceanic fields: h4 (pycnocline depth)

The time evolution of the spatial pattern of the first h4 mode (Fig. 6, middle column) shows a progressive deepening of the pycnocline along the eastern boundary of the North Pacific Ocean. Simultaneously, the anomalously shallow area initially located north of 50°N moves westward and then toward the south until it collides with the anomalously shallow area, (almost) northward propagating, originally centered at 10°N. The evolution of h4 (Fig. 6, middle column) shows a westward propagation in the northernmost part of the basin, while its evolution is concomitantly followed by the SST field. This information is also present in the bottom panels of Fig. 5, where a zonal change of phase characterizes the area north of 50°N.

In the Tropics, a zonal change of the h4 phase is evident along the equator and in the northern part of it (at about 15°N). In this same area, large h4 amplitudes are also present. The interpretation of spatial phases is very complicated because they are all relative values that sometimes, mainly in large domains such as ours, go beyond one full cycle with its values given in the −180°/180° range. Thus, we computed Hovmöller diagrams, h4 contours in the time–space domain, to identify the correct sense of propagation in areas where h4 shows a large variance. Figures 7 and 8 show the h4 field in the zonal and meridional directions, respectively. Along 15° and 55°N, h4 changes phase from east to west with speeds of about 3.0 and 1.2 km day−1, respectively. The tropical wave is similar in its location and path to the Rossby waves described for interannual timescales by Kessler (1990), Xie et al. (2000), and Capotondi and Alexander (2001). According to them, these Rossby waves could either originate from Ekman pumping or from reflected Kelvin waves on the eastern boundary of the Tropics. Given that our model horizontal resolution does not favor a proper representation of Kelvin waves, we are more inclined to believe that the wind stress curl is generating them. The meridional Hovmöller diagrams (Fig. 8) show a southward displacement of phases (anywhere between the 160°E and 180° meridians) north of about 27°N and northward south of that latitude. Meridional speeds, southward or northward, range between 0.6 and 0.8 km day−1. All these senses of “propagation” take place on areas where the mean flow does not oppose or favors the sense of propagation.

The two deeper areas (positive values) initially centered at 45° and 20°N also move toward each other until they form just one blob of deeper-than-normal h4, January 1980, which significantly decreases in size by January 1984 and is centered at about 30°N. This mode has an interesting feature, that is, the development of deeper h4 in the eastern tropical Pacific in conjunction with those along the eastern boundary. However, these anomalies are smaller in comparison with those observed in the central and western North Pacific, and are thus of dubious significance. To interpret the interdecadal SVD mode in terms of Rossby waves we need to consider at least a second baroclinic mode, given that the mean flow seems to have some role in determining the paths followed by the anomalies. The dispersion relation for baroclinic Rossby waves, propagating on a moving environment, can be written (e.g., Pedlosky 1979)
i1520-0485-33-12-2483-e1
where ω is the wave frequency, K is the wavenumber vector, β is the meridional derivative of the Coriolis parameter, V is the mean surface velocity vector, Rd is the Rossby radius of deformation, and k and l are the zonal and meridional wavenumbers, respectively. Then, if the wave propagates normal to the mean flow, its effect is not felt on the wave while, if the wave propagates against it, it can only exist as long the frequency remains positive. In (1), effects due to horizontal and vertical gradients of the mean flow are ignored given that their contributions, for the current model parameters and settings, are small to a first approximation. Even smaller are the contributions from the vertical gradient of the anomalous flow.

We next evaluate (1) using a period of 20 yr for the wave frequency, zonal and meridional mean flows (from Fig. 3) of 6 and −1 cm s−1, respectively, a Rossby radius of 15 km (Ripa 1986), and a value of β = 1.7 × 10−11 m−1 s−1. Assuming that the wavelength is much longer than the Rossby radius we obtain a zonal wavenumber of k = −3.1 × 10−7 m−1 when a meridional wavelength of 3700 km is used (measured from Fig. 8). These numbers lead to zonal and meridional phase speeds of −3.2 and −0.65 cm s−1, respectively. The (almost) southward propagating anomalies (Fig. 8) are the ones responsible for the SST changes in the KOE area and, in fact, could play a role in the frontal displacements reported by Seager et al. (2001). The correlation coefficient between interdecadal KOE h4 and KOE SST was 0.84 (zero lag) and was the highest one found in the North Pacific basin. Not only did h4 have a typical depth that was closer to typical observations of pycnocline depth, but it was also the isopycnal depth to which the KOE SST was most sensitive.

4) The spatial structure of the modal oceanic fields: The surface velocity field

In general terms, midlatitudinal warming (cooling) situations in SST are accompanied by deep (shallow) h4 depths. The comparison of these depths with the first mode surface velocity field (Fig. 6, right column) indicates that an important, if not dominant, geostrophic component is involved in the physical processes defining the temporal/spatial structure of the interdecadal mode. The velocity fields of the subtropical and subpolar gyres evolve and change their sense of circulation almost simultaneously. From Fig. 6 (right column), it is apparent that both gyres react with similar intensity, which is in line with the findings of Tourre et al. (1999) and White and Cayan (1998). The largest amplitudes of the zonal currents are located along 35°N between 150°E and 170°W and between 38° and 50°N from the coast to 150°E for the meridional flow (mostly the Oyashio area). These interdecadal modal flows are shown in Fig. 6 (right column) for the 1976–86 time frame. The most salient feature is the reversal from 1976 to 1986 of the subpolar gyre circulation, which goes from opposing the mean currents to be aligned with them.

5) Forcing

In this section we attempt first to identify areas of pycnocline forcing by surface heat fluxes and the wind stress curl and second to close the loop of oceanic processes that we found, from Figs. 68, to move around the KOE region from the eastern to the western North Pacific. The wind stress curl and heat flux forcing (Fig. 9) act to vertically displace the isopycnal surfaces. In some locations, as just off the U.S. and Canadian coasts, they counteract each other with heat fluxes having a dominant role; in the CNP and KOE both forcing agents reinforce each other, leading to a shallower pycnocline in the early 1980s (Figs. 6 and 9). In the central Pacific along 15°N, pycnocline depths have a sign that is not in line with either wind stress curl forcing or lagged (by 2–4 yr) heat flux forcing. The Bering Sea also shows a similar feature during the late 1970s and early 1980s, thus suggesting that in both areas the local perturbation of the pycnocline anomalies is of remote origin, as Figs. 7 and 10 also seem to suggest.

Given that these interdecadal pycnocline depths are sensitive to both forcing agents and that the KOE area is well known because of its SST's sensitivity to thermocline fluctuations, we estimated the lagged correlations between the interdecadal h4 field and KOE SSTs. Figure 10 shows that the maximum lagged correlation between interdecadal KOE SST and h4 takes place at zero lag in the KOE region, reaching a maximum value of 0.93. This partly confirms the sense of propagation found from the Hovmöller plots in Figs. 7 and 8; pycnocline fluctuations move around the KOE region in areas where the mean flow does not oppose the direction of propagation. They start on the eastern boundary, moving toward the Aleutian Islands through the Gulf of Alaska and Bering Sea. Before arriving at the Kamtchatka Peninsula, they turn toward the southwest down to the KOE region. The other pathway lies in the northern Tropics around 15°N from the eastern to the western boundary and then toward the northwest.

Sea level pressure and wind stress curl are two intimately related variables, and, in fact, Tourre et al.'s (2001) modal SLP and our modal wind stress curl (Fig. 9, left column) show similar features. In both analyzes, their observations and our modeling study, the increased cooling of the KOE and CNP regions (bottom panels in Figs. 6 and 9) is accompanied by an increased positive wind stress curl (Fig. 9, left column) or, equivalently, a strengthened SLP in the Aleutian low area. The maxima (minima) of the wind stress curl (SLP) then displaces to the southeast while SST anomalies show a simultaneous warming (on the eastern and then northern boundaries of the basin), also in agreement with their observations. This correspondence between SST and wind stress curl (SLP) is suggestive of active coupling and feedback between ocean and atmosphere (Peng and Witaker 1999), a feature also noted by Tourre et al. (2001) from observations. At the bottom of Fig. 8 (left column), we also notice an increased across-shore SST gradient off the U.S. and Canadian coasts. It is then worth investigating if a mechanism similar to that one proposed by White and Barnett (1972), and recently invoked by Pierce et al. (2001), is in operation, suggesting coupling between ocean and atmosphere through a vorticity exchange between both fluids. Their theory can be summarized by the following two vorticity equations, one for each media:
i1520-0485-33-12-2483-e2
where ψw and ψa are the vertically integrated oceanic and atmospheric streamfunctions and τ, Q, and β are the wind stress, the atmospheric heat flux (positive into the ocean), and the meridional gradient of the Coriolis parameter, respectively (all C are positive constants). They next parameterize the relationship between atmospheric heat fluxes and the zonal flow as
i1520-0485-33-12-2483-e4
which is a good approximation in the frontal areas of subpolar oceans. We now consider that ocean and atmosphere are coupled (i.e., that ψw = ψa) and, further, given the results of our SVD of the SST/surface velocity pair, that ocean currents flow approximately parallel to isotherms. This requisite combined with (2), (3), and (4) yields
i1520-0485-33-12-2483-e5

Figure 9 shows the time evolution of the first SVD mode wind stress curl (left column) and of the meridional gradient of the first SVD mode of −∇2SST (middle column) from January 1970 to January 1984 every two years. This first mode is well significant and explains 48% of the total squared covariance. As before, both modes are recovered if both variables are SVD paired to the SST field. North of about 30°N, subpolar areas, there is a good correspondence in sign between both fields, especially for areas where the variability is large. From Figs. 610 an interdecadal pathway was described that carries, on average, information from the eastern to the western boundary. We focus now on how the ocean carries information eastward and offer a possible explanation on how it forces the atmosphere along 40°–50°N. The interdecadal wind stress curl pattern of Fig. 9 (left column) changes almost simultaneously with the one of negative meridional gradient of the first SVD mode of ∇2SST (Fig. 9, middle column). Both variables evolve almost simultaneously in time, despite the noise that was generated in the computation of the meridional gradient of ∇2SST. The eastward propagation of the ∇2SST along the 40°–50°N band is not only seen in the coupled model experiments of Pierce et al. (2001), but also in their observations. We thus constructed the Hovmöller diagram of the wind stress curl, shown in Fig. 11. The top and bottom panels give eastward speeds of roughly 0.8 (SST Laplacian) and 1.7 km day−1 (wind stress curl). In a coupled system these speeds should be the same, but our ocean model lacks any explicit feedbacks. Our speed estimation for ∇2SST is 0.8 km day−1 and compares favorably with those estimated by Pierce et al. (2001) from SST observations, 0.9 km day−1, and from their coupled model, 1.8 km day−1. However, in this study, the area of eastward propagation is located more to the east (from 165° and eastward) than in Pierce et al. (2001).

These eastward phase speeds cannot exist at these low frequencies (i.e., 20–30-yr timescale) under the restrictions imposed by (1) with no mean flow. However, both model and real ocean have a nonzero mean flow, which allows eastward propagation for waves with k < R−1d. In the CNP, between 180° and 210°E, the mean flow is almost eastward with a speed of 6 cm s−1 approximately (from Fig. 3). Then, (1) becomes a third-order polynomial in k, whose only stable solution is an eastward propagating wave with k = 10−6 m−1. The resulting eastward phase speed is 0.9 km day−1 which favorably compares to the 0.8 km day−1 obtained above from our model, and to the 0.9 km day−1 obtained by Pierce et al. (2001) from observations.

The atmospheric heat fluxes (Fig. 9, right column) play a key role in establishing the SST patterns of Figs. 5 and 6 (left column). The evolution of both fields is similar with heat fluxes leading SSTs by 4–6 yr. The eastward displacement of the positive heat flux anomalies located initially in the CNP is probably related to the eastward displacement of ∇2SST; then negative anomalies take over. From 1976 through 1984 positive anomalies develop and move poleward along the eastern boundary from 40°N and northwestward to the Aleutian Island and Bering Sea area.

6) Differences between interdecadal and decadal variability

The second mode of our SST-h4 has a timescale of roughly 10 yr and its SST time evolution (Fig. 12) is similar to the modal SST obtained by Tourre et al. (2001). The SST and h4 phases progress to the west (Fig. 12, bottom panels), and this is not seen in the interdecadal (first) mode of either SST or h4. The main feature of this SST mode, in both our model and their observations, is the evolution of a dipole with centers located, in both model and observations, at 43°N, 160°E and at 40°N, 155°W (not shown). These two relative maxima switch signs about every five years, and might be partly responsible for the reported time lag of 4–6 yr (e.g., Miller and Schneider 2000) between the SSTs of the CNP and KOE areas. The amplitude maxima in the CNP area is reminiscent of recent descriptions of the PNA pattern (Deser and Blackmon 1995). The main difference between decadal and interdecadal variability is the path followed by the isopycnal perturbations after they are induced by the wind stress curl and/or surface heat fluxes. Figure 13 shows that decadal KOE SST are led by about 4 yr by decadal h4 variability in the CNP and eastern North Pacific, which was reported earlier from observations and coupled model experiments by Schneider et al. (2002). On the other hand, interdecadal variability goes around the KOE area, and in all cases, this propagation of information is favored by the mean flow direction. A schematic of the mean circulation of the North Pacific shows (Fig. 14) that the mean flow direction coincides with the path followed by the h4 anomalies reported in this article. The eastward wind drift along about 40°N also coincides with the path followed by the ∇2SST anomalies and wind stress curl.

The interdecadal and decadal modes and their corresponding paths of propagation are in line with the findings of Liu (1999), who describes two different pathways, that as noted before, are mostly determined by the relative size of the wave speed to the background flow. Specifically, the interdecadal mode would correspond to Liu's “A” mode, strongly influenced by atmospheric heat fluxes (cf. SST and Q patterns from Figs. 6 and 9) and the mean flow; on the other hand, our decadal mode resembles Liu's “N” mode, mostly forced by the wind stress curl in the eastern and central North Pacific and having a dominant westward propagation. We have computed the model's vertical structure functions (not shown) following the method of Ripa (1986) and found, in agreement with Liu's modes, that the first baroclinic mode, decadal timescale in this study, has maximum amplitude at 800 m while the second baroclinic mode, interdecadal timescale in this study, exhibits maximum amplitude at 400 m and is 60% slower than the first mode.

Next, it is important to note that decadal and interdecadal variability are also different in the heat budget components that control their physics. On a basin-scale average in the mixed layer and north of 15°N, the horizontal advection term is negligible for decadal timescale variability, while it plays a major role in the interdecadal band (Table 1). At this latitude, this contribution is very likely due to Rossby wave activity, through thermocline heave, for which geostrophic currents are normal to latitude circles. At this latitude, atmospheric heat fluxes can make an important contribution to the overall heat budget (e.g., Fig. 9 shows some amplitudes in the 1970s and 1980s), since the Newtonian damping of SST anomalies is constrained to the 7°S–7°N; band that is, at 5°N only one-half of the SST anomalies are damped.

On a regional scale, the heat budget components show that decadal-timescale SST force the overlying atmosphere (SST damping) in the KOE area (Fig. 15, bottom panel) in agreement with the reports of Schneider and Miller (2001) and Schneider et al. (2002). However, for interdecadal timescales, the damping of SSTs to the atmosphere seems to take place in the Bering Sea/Aleutian Island area (Fig. 15, top panel). We are not aware that the Bering Sea has been previously reported as an area of oceanic forcing. However, it is in, or nearby, this region where observations and models show that the different interdecadal atmospheric fields are more energetic (Aleutian low system and its associated wind stress curl). Similarly, the center of action of the decadal mode is located in the KOE and CNP regions, as seen in the modal wind stress curl pattern of this study and in the modal SLP pattern of Tourre et al. (2001), which have a similar description. The spatial structure of the interdecadal and decadal SST variances also contributes to clearly differentiate between both timescales. Decadal variability has maximum variance in the KOE and CNP areas, that is, in the two areas where the dipole is located. Interdecadal variability, instead, has maximum variability slightly north of the KOE area and along the eastern boundary between Canada and northern Mexico. The dominant forcing agent also differs between decadal and interdecadal timescales. The former is mostly driven by the wind stress curl (e.g., Schneider et al. 2002), while the latter by atmospheric heat fluxes that have maximum variability along the eastern boundary and in the western half of the KOE, that is, where the interdecadal SST also shows maximum variability.

4. Discussion

Even though there is a good agreement between our findings and those obtained from coupled models for decadal timescales (e.g, Schneider et al. 2002), some differences exist for interdecadal timescales as noted by Pierce et al. (2001) from a set of experiments with four different coupled models. For instance, ECHO2 apparently does not reproduce the observed interdecadal SST variability along the eastern boundary. The reasons for this disagreement might reside in how numerical models can simulate high-order baroclinic modes and their interaction with the mean flow. It is particularly interesting that ECHO2, on the other hand, properly simulates the main features of the decadal variability, being that this mode is almost independent of the mean flow.

The dynamics behind the horizontal displacement of the interdecadal pycnocline perturbations is still uncertain. However, this author is more inclined to interpret this mode in terms of wave propagation, given its similarities with Liu's A mode (i.e., forcing and vertical structure). Perhaps most important, its propagation path coincides with that one of the mean circulation in the North Pacific. We have found that the mean flow is very irregular in some areas along these pathways, which is not concomitantly reflected in the phase speeds obtained from the Hovmöller diagrams. In turn, these propagation speeds are different from mean flow speeds in most locations and from the speed at which the atmospheric forcing evolves.

The mean flow of the KOE would prevent a direct propagation of interdecadal perturbations from the generation area to the KOE; instead, two different interdecadal perturbations develop and propagate around this “forbidden” area of strong eastward flow. One of them, is a moving perturbation that propagates westward along 10°–20°N, while the other one, an originally northwestward propagating wave that moves from about 45°N on the eastern boundary toward the subarctic, up to about the Kamchatka Peninsula, and then turning southwestward toward the KOE region. When both interdecadal perturbations arrive at the western boundary, they slowly turn, following the approximate direction of the mean flow converging at around 25°–30°N.

The area occupied by the KOE would prevent any westward propagation as long as the mean flow to the east is faster than the phase speed to the west and the vortex tube stretching does not counter the mean flow effect in the dispersion relation. In models with weaker mean flows westward propagation could occur but that will not lead to realistic representations of the ocean dynamics. For interdecadal timescales, in the ocean model, since westward propagation is apparently inhibited by the mean flow, pycnocline depth perturbations move poleward (westward) along the basin's eastern (northern) boundaries and westward along the 10°–18°N. On both paths, the mean flow is, in general, in the same direction as the propagation direction. For decadal timescales westward propagation seems to originate from the eastern and central North Pacific, migrating to the KOE area. These waves are fast enough to counter the mean flow at the depth of the pycnocline and/or have the effects of vortex tube stretching canceling out the mean flow contribution; as noted by Liu (1999), these are probably first baroclinic Rossby waves.

The main shortcoming of this study is the relatively large ratio between the timescale of interest to the time series length. However, we are confident about the representativeness of our results, given the reasonable similitude in both the spatial structure and time evolution that exists between our first two SVD modes, decadal and interdecadal, of the SST-wind stress curl pair based on 40 years of data and the SVD modes obtained from 100 years of observations by Tourre et al. (2001). This comparison and also the one between our results and those of Liu (1999) suggest that, even though our study is closer to a case study, there is some representativeness in our findings.

We have focused our attention on the midlatitudinal North Pacific Ocean because of the poor sampling rate of the observations in the South Pacific Ocean. However, the model interdecadal SSTs in the Southern Hemisphere are about the mirror image of those in the Northern Hemisphere for interdecadal SST (T. Baumgartner et al. 2003, personal communication), a feature that is in line with the findings of Garreaud and Battisti (1999) on the interdecadal atmospheric symmetry in both hemispheres. The tropical band, on one hand, showed small amplitudes in the analyzed modal fields (SST, pycnocline depth, or wind stress curl), but qualitatively it is apparent that Rossby waves do exist in the northern part of the Tropics between 10° and 20°N. These waves are similar to those reported by Kessler (1990) in the annual and interannual bands. It is apparent from this study that interdecadal SSTs in the KOE are originated from Rossby waves having a meridional component much larger than the zonal one (i.e., lk), unlike the decadal timescale. Part of the confusion about the characterization of decadal and interdecadal SST variability can arise from the fact that some historical conclusions were drawn from analyzes that included both timescales and the fact that decadal and interdecadal SSTs have in common both a maxima in the KOE area and its vicinity and CNP SSTs leading KOE SSTs by about 3–6 yr.

In the interdecadal band, the Bering Sea area would be playing a similar role to that played by the KOE region in the decadal mode. In its southern half, SST anomalies are damped to the atmosphere, affecting the sea level pressure of the Aleutian low system. If verified from observations, this SST damping could play a major role in the life cycle of the interdecadal variability in the North Pacific. Of course, verification from observations and more specific numerical experiments need to be carried out to asses the role of the Bering Sea in the context of interdecadal variability in the North Pacific. The above points to the fact that in the real ocean the picture is very likely more complicated, as noted by Liu (1999), since many modes and timescales might be present at one time: for instance, even though surface net heat fluxes are dominant in determining the SST variability, the wind stress curl has also a role in at least perturbing the pycnocline depths in and near the Gulf of Alaska. Southeastward of this location, the diabatic forcing is dominant over the wind stress curl. Even though our model lacks any feedbacks between ocean and atmosphere, those mechanisms can be present in an implicit way, given the good correspondence between the model's decadal and interdecadal SSTs and those obtained from observations. This is also supported by the fact that the model's forcing functions, the NCEP wind stress and heat flux, showed good correspondence with observations in the extratropics (Auad et al. 2001).

The forcing of the interdecadal band is complicated. Wind stress curl and heat fluxes have maximum amplitudes in different areas, but in general the effects of the latter dominate the former, for example, at the eastern boundary, while in the KOE and CNP they reinforce each other during the 1970s and 1980s. After the regime shift of 1976–77, the warming of the areas occupied by the Aleutian Islands and the southern half of the Bering Sea led to both a warming of the atmosphere and to a reversal of the meridional gradient of the SST in the KOE area. Both situations, would combine to flip the sign of the atmospheric forcing in about 10–13 yr, thus leading to a quasi periodicity of 20–26 yr.

5. Conclusions

The results obtained in this study using an ocean model link the theoretical and observational findings on decadal and interdecadal variability in the North Pacific of Liu (1999) and Tourre et al. (2001), respectively. Very different approaches have been followed in all three papers, but a consistent picture emerges from them. We also obtained satisfactory results when comparing model SSTs on either a gridpoint-by-gridpoint basis or estimating the basin-scale SST average as in White et al. (1997).

In agreement with the observational work of Tourre et al. (2001), decadal and interdecadal variability are two different phenomena, and from this study it is suggested that different dynamics control them. It is possible that the only difference between both timescales lies in the dispersion properties of the first and second baroclinic modes of the Rossby waves generated in the northeastern corner of the basin. If the decadal waves have a faster westward phase speed, first baroclinic mode, they will be able to overcome the eastward mean flow of the KOE at the pycnocline depth and travel directly to this region. The Rossby waves in the interdecadal band, instead, have to go around the KOE because of their lower speed (second or third baroclinic modes), given the restrictions imposed by the mean flow field. The description of these two timescales and their pathways are in line with the theoretical findings of Liu (1999), who described a non-Doppler shift first baroclinic mode and an advective second baroclinic mode.

Interdecadal atmospheric heat flux variability is primarily responsible for the generation of pycnocline depth oscillations in the eastern North Pacific Ocean in midlatitudes. Very likely this is due to the excitation of a second or third baroclinic mode that has maximum amplitude at or near the surface. In midlatitudes, this leads to the horizontal displacement, starting at about 45°N, of pycnocline anomalies toward the north and west. Before arriving at the Kamchatka Peninsula the anomalies, following the mean flow, turn southwestward, which upon arrival in the KOE area leads to the creation of SST anomalies. This propagation continues up to about 28°N where they encounter another field of pycnocline anomalies moving northwestward. The latter were induced by the arrival of a westward-propagating Rossby wave along 15°N, probably generated by the wind stress curl in the northeastern part of the Tropics (10°–20°N). This part of the cycle, from atmospheric forcing at or near the eastern boundary to wave arrival to the KOE area, takes about 5–6 years. From this study alone it cannot be concluded that oceanic–atmospheric coupling/feedbacks are either absent or present. However, the present results along with the modeling and observational evidence of Pierce et al. (2001) would suggest that a mechanism similar to that anticipated by White and Barnett (1972) could be operating at and/or near 40°N. Further, our results here are consistent with the results of Latif and Barnett (1994, 1996) in that oceanic–atmospheric coupling takes place when the midlatitudinal wind stress curl responds to changes in the underlying SST field. Thus, the travel time from the KOE to the CNP/eastern North Pacific area is of about 5–7 years, leading to a quasiperiodicity of about 20–26 yr, at least in the 1958–97 time frame.

Five main differences found between decadal and interdecadal variability in the North Pacific Ocean are (i) the path followed by pycnocline oscillations from the eastern and central North Pacific to the KOE area (around and through the KOE for interdecadal and decadal variability, respectively), (ii) the horizontal advection term integrated north of 15°N in the mixed layer is important for the interdecadal heat budget but is a minor contributor in the decadal band, (iii) oceanic forcing is geographically differentiated for both timescales (SST anomalies are damped to the atmosphere in the KOE and Bering Sea regions in the decadal and interdecadal bands, respectively), (iv) maximum SST variability takes place west of the date line at 45°N and along the eastern boundary in the interdecadal band and in the CNP and KOE areas in the decadal band, and (v) the interdecadal band is mostly forced by atmospheric heat fluxes, unlike the decadal band which is mostly driven by the wind stress curl.

In summary, it remains to verify the existence in nature of the interdecadal oceanic mechanism described in this article, as well as to shed more light on its atmospheric component. Fundamental questions arise from this study, given the similarity found between the interdecadal SST pattern and that of the PDO: Does Liu's (1999) A mode describe the ocean dynamics of the PDO? If so, is this mode interacting with the atmosphere in line with the early ideas of White and Barnett (1972)? Related to this, it has been reported (Auad et al. 2003, manuscript submitted to J. Geophys. Res.) that PDO-like SST patterns not only are obtained as a result of simulating modern climate (e.g., the 1976–77 shift) but also when inferring the behavior of the North Pacific during the termination of the last ice age. Thus, it will be a key step toward a better understanding of the dynamics of large-scale climate changes to obtain long, reconstructed records of the PDO index at many locations over the last, say, 20 000 yr. Coupled models should be ideal for studying subdecadal coupled phenomena such as those reported in this article. However, the author believes that several improvements need to be made to them in order to obtain better representations of the mean flow and of the SST variance in the interdecadal band. The decadal band is better represented by coupled models since it is much less dependent on the mean flow. It was/is the author's intention to advance a different perspective, that is, from an ocean model forced by NCEP–NCAR reanalysis fluxes on interdecadal variability in the North Pacific Ocean. More research is needed in this direction, for example, the computation of the oceanic and atmospheric streamfunctions in order to verify the existence of coupled oceanic–atmospheric Rossby waves. It will be also important to attack the coupled problem from the perspective of the energetics of the system in order to quantify the importance of each process. The knowledge gained on these processes will leave us in a better position to construct more accurate forecasting systems to aid society in both planning and prevention.

Acknowledgments

Financial support was provided by the National Oceanic and Atmospheric Administration (NA17RJ1231 through the Experimental Climate Prediction Center and the Consortium for the Ocean's Role in Climate), the Department of Energy (W/GEC 00-006), and the National Science Foundation (OCE-00-82543). The views expressed herein are those of the author and do not necessarily reflect the views of NOAA or any of its subagencies. I especially thank Art Miller and Warren White for their extensive and insightful comments and reviews on the earlier drafts of this paper. I also thank two anonymous reviewers for their important suggestions and remarks that helped to significantly improve the quality of this paper.

REFERENCES

  • Alexander, M. A., C. Deser, and M. S. Timlin, 1999: The reemergence of SST anomalies in the North Pacific Ocean. J. Climate, 12 , 24192433.

    • Search Google Scholar
    • Export Citation
  • Anderson, P. J., and J. F. Piatt, 1999: Community reorganization in the Gulf of Alaska following ocean climate regime shift. Mar. Ecol. Prog. Ser., 189 , 117123.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, and W. B. White, 1998a: Simulation of heat storages and associated heat budgets in the Pacific Ocean, 1, El Niño-Southern Oscillation timescale. J. Geophys. Res., 103 , 2760327620.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, and W. B. White, 1998b: Simulation of heat storages and associated heat budgets in the Pacific Ocean, 2, Interdecadal timescale. J. Geophys. Res., 103 , 2762127635.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, J. O. Roads, and D. R. Cayan, 2001: Pacific Ocean wind stress and surface heat flux anomalies from NCEP reanalysis and observations: Cross-statistics and ocean model response. J. Geophys. Res., 106 , 2224922265.

    • Search Google Scholar
    • Export Citation
  • Barnett, T. P., M. Latif, E. Kirk, and E. Roeckner, 1991: On ENSO physics. J. Climate, 4 , 487515.

  • Barnett, T. P., D. W. Pierce, R. Saravanan, N. Schneider, D. Dommenget, and M. Latif, 1999: Origins of the midlatitude Pacific decadal variability. Geophys. Res. Lett., 26 , 14531456.

    • Search Google Scholar
    • Export Citation
  • Beamish, R. J., 1993: Climate and exceptional fish production off the west coast of North America. Can. J. Fish. Aquat. Sci., 50 , 22702291.

    • Search Google Scholar
    • Export Citation
  • Bjerknes, J., 1964: Atlantic air–sea interaction. Advances in Geophysics, Vol. 10, Academics Press, 1–82.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5 , 541560.

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

    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., A. J. Miller, T. P. Barnett, N. E. Graham, J. N. Ritchie, and J. M. Oberhuber, 1995: Seasonal-interannual fluctuations in surface temperature over the Pacific: Effects of monthly winds and heat fluxes. Natural Climate Variability on Decadal-to-Century Time Scales, National Academy Press, 133–150.

    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., A. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson, 2001: Changes in the onset of spring in the western United States. Bull. Amer. Meteor. Soc., 82 , 399415.

    • Search Google Scholar
    • Export Citation
  • 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
  • Deser, C., and M. L. Blackmon, 1995: On the relationship between tropical and North Pacific sea surface temperature variations. J. Climate, 8 , 16771680.

    • Search Google Scholar
    • Export Citation
  • Deser, C., M. A. Alexander, and M. S. Timlin, 1999: Evidence for a wind-driven intensification of the Kuroshio Current Extension from the 1970s to the 1980s. J. Climate, 12 , 16971706.

    • Search Google Scholar
    • Export Citation
  • Dettinger, M. D., D. S. Battisti, R. D. Garreaud, G. J. McCabe, and C. M. Bitz, 2000: Interhemispheric effects of interannual and decadal ENSO-like climate variations on the Americas. Present and Past Interhemispheric Climate Linkages in the Americas and their Societal Effects, V. Markgraf, Ed., Cambridge University Press, 1–16.

    • Search Google Scholar
    • Export Citation
  • Evans, M. N., M. A. Cane, D. P. Schrag, A. Kaplan, B. K. Linsley, R. Villalba, and G. M. Wellington, 2001: Support for tropically-driven Pacific decadal variability based on paleoproxy evidence. Geophys. Res. Lett., 28 , 36893692.

    • Search Google Scholar
    • Export Citation
  • Frankignoul, C., P. Muller, and E. Zorita, 1997: A simple model of the decadal response of the ocean to stochastic wind forcing. J. Phys. Oceanogr., 27 , 15331546.

    • Search Google Scholar
    • Export Citation
  • Garreaud, R. D., and D. S. Battisti, 1999: Interannual ENSO and interdecadal ENSO-like variability in the Southern Hemisphere tropospheric circulation. J. Climate, 12 , 21132123.

    • Search Google Scholar
    • Export Citation
  • Kaylor, R. E., 1977: Filtering and decimation of digital time series. Institute for Physical Science and Technology Tech. Rep. BN 850, University of Maryland, 42 pp.

    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., 1990: Observations of long Rossby waves in the northern tropical Pacific. J. Geophy. Res., 95 , 51835217.

  • Killworth, P., 1996: Time interpolation of forcing fields in ocean models. J. Phys. Oceanogr., 26 , 136143.

  • Kirtman, B. P., and P. S. Schopf, 1998: Decadal variability in ENSO predictability and prediction. J. Climate, 11 , 28042822.

  • Latif, M., and T. Barnett, 1994: Causes of decadal climate variability over the North Pacific and North America. Science, 266 , 634637.

    • Search Google Scholar
    • Export Citation
  • Latif, M., and T. Barnett, 1996: Decadal climate variability over the North Pacific and North America—Dynamics and predictability. J. Climate, 9 , 24072423.

    • Search Google Scholar
    • Export Citation
  • Linsley, B. K., G. M. Wellington, and D. P. Schrag, 2000: Decadal sea surface temperature variability in the subtropical South Pacific from 1726 to 1997 a.d. Science, 290 , 11451148.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 1999: Forced planetary wave response in a thermocline gyre. J. Phys. Oceanogr., 29 , 10361055.

  • Lysne, J., and C. Deser, 2002: Wind-driven thermocline variability in the Pacific: A model–data comparison. J. Climate, 15 , 829845.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., and S. R. Hare, 2002: The Pacific decadal oscillation. J. Oceanogr., 58 , 3544.

  • 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
  • Maximenko, N. A., P. P. Niiler, G. G. Panteleev, T. Yamagata, and D. B. Olson, 2002: Near-surface dynamical structure of the Kuroshio Extension derived from drifter and altimetry data. Exchanges, 7 , 14.

    • Search Google Scholar
    • Export Citation
  • Mestas-Nuñez, A., and D. B. Enfield, 1999: Rotated global modes of non-ENSO sea surface temperature variability. J. Climate, 12 , 27342746.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., and N. Schneider, 2000: Interdecadal climate regime dynamics in the North Pacific Ocean: Theories, observations and ecosystem impacts. Progress in Oceanography, Vol. 47, Pergamon, 355–379.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., D. R. Cayan, T. P. Barnett, N. E. Graham, and J. M. Oberhuber, 1994a: The 1976–77 climate shift of the Pacific Ocean. Oceanography, 7 , 2126.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., D. R. Cayan, T. P. Barnett, N. E. Graham, and J. M. Oberhuber, 1994b: Interdecadal variability of the Pacific Ocean: Model response to observed heat flux and wind stress anomalies. Climate Dyn., 9 , 287302.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., W. B. White, and D. R. Cayan, 1997: North Pacific thermocline variations on ENSO timescales. J. Phys. Oceanogr., 27 , 20232039.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., D. R. Cayan, and W. B. White, 1998: A westward-intensified decadal change in the North Pacific thermocline and gyre-scale circulation. J. Climate, 11 , 31123127.

    • Search Google Scholar
    • Export Citation
  • Minobe, S., 2000: Spatio-temporal structure of the pentadecadal variability over the North Pacific. Progress in Oceanography, Vol. 47, Pergamon, 381–408.

    • Search Google Scholar
    • Export Citation
  • Nakamura, H., G. Lin, and T. Yamagata, 1997: Decadal climate variability in the North Pacific during recent decades. Bull. Amer. Meteor. Soc., 78 , 22152225.

    • Search Google Scholar
    • Export Citation
  • Oberhuber, J. M., 1993: Simulation of the Atlantic circulation with a coupled sea ice–mixed layer–isopycnal general circulation model. Part I: Model description. J. Phys. Oceanogr., 23 , 808829.

    • Search Google Scholar
    • Export Citation
  • Pedlosky, J., 1979: Geophysical Fluid Dynamics. Springer-Verlag, 624 pp.

  • Peng, S., and J. S. Whitaker, 1999: Mechanisms for determining the atmospheric response to midlatitude SST anomalies. J. Climate, 12 , 13931408.

    • Search Google Scholar
    • Export Citation
  • Peng, S., W. A. Robinson, and M. P. Hoerling, 1997: The modeled atmospheric response to midlatitude SST anomalies and its dependency on background circulation states. J. Climate, 10 , 971987.

    • Search Google Scholar
    • Export Citation
  • Pierce, D. W., T. P. Barnett, N. Schneider, R. Saravanan, D. Dommenget, and M. Latif, 2001: The role of ocean dynamics in producing decadal climate variability in the North Pacific. Climate Dyn., 18 , 5170.

    • Search Google Scholar
    • Export Citation
  • Qiu, B., 1995: Why is the spreading of the North Pacific Intermediate Water confined on density surfaces around σ = 26.8? J. Phys. Oceanogr., 25 , 168180.

    • Search Google Scholar
    • Export Citation
  • Qiu, B., and K. A. Kelly, 1993: Upper-ocean heat balance in the Kuroshio–Oyashio Extension region. J. Phys. Oceanogr., 23 , 20272041.

    • Search Google Scholar
    • Export Citation
  • Reed, R. K., J. D. Schumacher, P. J. Stabeno, C. H. Pease, A. W. Kendall Jr., and P. Dell'Arciprete, 1994: Alaskan Stream inflow and effects in the Bering Sea. Eos, Trans. Amer. Geophys. Union, 75 , 342.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses. J. Climate, 7 , 929948.

  • Ripa, P., 1986: Evaluation of vertical structure functions for the analysis of oceanic data. J. Phys. Oceanogr., 16 , 223232.

  • Robertson, A. W., 1996: Interdecadal variability in a multicentury climate integration. Climate Dyn., 12 , 227241.

  • Schneider, N., and A. J. Miller, 2001: Predicting western North Pacific Ocean climate. J. Climate, 14 , 39974002.

  • Schneider, N., A. J. Miller, and D. W. Pierce, 2002: Anatomy of North Pacific decadal variability. J. Climate, 15 , 586605.

  • Seager, R., M. B. Blumenthal, and Y. Kushnir, 1995: An advective atmospheric mixed layer model for ocean modeling purposes: Global simulation of surface heat fluxes. J. Climate, 8 , 19511964.

    • Search Google Scholar
    • Export Citation
  • Seager, R., Y. Kushnir, N. H. Naik, M. A. Cane, and J. Miller, 2001: Wind-driven shifts in the latitude of the Kuroshio–Oyashio Extension and generation of SST anomalies on decadal timescales. J. Climate, 14 , 42494265.

    • Search Google Scholar
    • Export Citation
  • Stabeno, P. J., and R. K. Reed, 1994: Circulation in the Bering Sea basin by satellite-tracked drifters: 1986–1993. J. Phys. Oceanogr., 24 , 848854.

    • Search Google Scholar
    • Export Citation
  • Tourre, Y., W. B. White, and Y. Kushnir, 1999: Evolution of interdecadal variability in sea level pressure, sea surface temperature and upper-ocean temperature over the Pacific Ocean. J. Phys. Oceanogr., 29 , 15281541.

    • Search Google Scholar
    • Export Citation
  • Tourre, Y., B. Rajagopalan, Y. Kushnir, M. Barlow, and W. B. White, 2001: Patterns of coherent decadal and interdecadal climate signals in the Pacific basin during the 20th century. Geophys. Res. Lett., 28 , 20692072.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 1990: Recent observed interdecadal climate changes in the Northern Hemisphere. Bull. Amer. Meteor. Soc., 71 , 988993.

    • Search Google Scholar
    • Export Citation
  • White, W. B., and T. P. Barnett, 1972: A servomechanism in the ocean/atmosphere system of the midlatitude North Pacific Ocean. J. Phys. Oceanogr., 2 , 372381.

    • Search Google Scholar
    • Export Citation
  • White, W. B., and D. R. Cayan, 1998: Quasi-periodicity and global symmetries in interdecadal upper ocean temperature variability. J. Geophys. Res., 103 , 2133521354.

    • Search Google Scholar
    • Export Citation
  • White, W. B., J. Lean, D. Cayan, and M. Dettinger, 1997: Response of global upper ocean temperature to changing solar irradiance. J. Geophys. Res., 102 , 32553266.

    • Search Google Scholar
    • Export Citation
  • Xie, S. P., T. Kunitani, A. Kubokawa, M. Nonada, and S. Hosoda, 2000: Interdecadal thermocline variability in the North Pacific for 1958–1997: A GCM simulation. J. Phys. Oceanogr., 30 , 27982813.

    • Search Google Scholar
    • Export Citation
  • Zhang, R. H., and S. Levitus, 1997: Structure and cycle of decadal variability of upper-ocean temperature in the North Pacific. J. Climate, 10 , 710727.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(top) Correlation coefficients between model and observed SSTs. The data were low passed for periods longer than 6 yr. The contour interval (CI) is 0.1, and the 90% confidence level (Davis 1976) is 0.49 on a spatial average. (middle and bottom) The SST standard deviation for model and observations, respectively. The CI are 0.2 and 0.1 for the middle and bottom panels, respectively. Data from the 1958–97 time frame were used

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 2.
Fig. 2.

Regional averages of the model SST in the KOE (solid) and CNP (dashed) regions. The CNP region leads by about 2–4 yr. The time series were low passed with an 8-yr running mean filter

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 3.
Fig. 3.

Annually averaged model mean flow in the KOE region. The contours represent the amplitude with a CI of 2 cm s−1

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 4.
Fig. 4.

Singular value decomposition modal time series between the pair SST and h4: (a) time evolution of the first coupled mode between SST (solid line) and h4 (dashed line) and (b) time evolution of the second coupled mode between SST (solid line) and h4 (dashed line)

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 5.
Fig. 5.

(top) Amplitude and (bottom) phase of the first SVD mode of the SST/h4 pair. Contour interval for amplitudes is 0.01 and for phases is 30°

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 6.
Fig. 6.

Reconstructed time evolution of the first SVD modes of (left) SST, (middle) h4, and (right) surface velocity. Panels in each column are two years apart and go from Jan 1974 (top three panels) to Jan 1984 (bottom three panels) covering 0.5 cycle approximately. This time frame clearly shows the changes attained by the midlatitudinal North Pacific after the 1976 climate shift (Miller et al. 1994). Since tropical ocean currents have the tendency to dominate the SVD first modes, we computed the SVD for the SST/surface velocity pair north of 20°N. The time series and explained covariance are very similar to the ones of Fig. 5. The CI for the left and middle columns are 0.2°C (plus the 0.1°C contour) and 10 m, respectively

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 7.
Fig. 7.

Hovmöller diagrams (lon–time) for pycnocline depth contours (h4) along (top) 15° (CI = 4 m) and (bottom) 55°N (CI = 8 m)

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 8.
Fig. 8.

Hovmöller diagrams (time–lat) of the h4 field along the (left) 160°E, (middle) 170°E, and (left) 180° meridians (CI = 4 m)

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 9.
Fig. 9.

Reconstructed time series of the first modes of the (left) wind stress curl (CI = 10−8 N m−2) of the meridional gradient of the first mode of the (middle) ∇2SST (CI = 10−13 °C m−3) and of the (right) surface heat flux (CI = 4 W m−2). All SVDs yield the first modes paired to SST or to wind stress curl (i.e., same spatiotemporal structure/behavior)

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 10.
Fig. 10.

Lagged correlations between pycnocline depth (h4) and KOE SST (both low passed for periods longer than 12 yr). The time lag is marked in every panel over the U.S. map and goes anticlockwise from 9 to 0 yr with h4 leading KOE SST. The 90% confidence levels were computed according to Davis (1976) and are 0.60 on a spatial average. Deeper-than-normal pycnocline depths are defined as positive. Contour labels go from −0.9 to 0.9 every 0.2 but including the zero correlation contour

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 11.
Fig. 11.

Hovmöller diagrams (lon–time) of (top) ∇2SST along 40°N (CI = 10−12 °C m−2) and (bottom) the wind stress curl along 45°N (CI = 10−8 N m−2)

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 12.
Fig. 12.

Amplitudes and phases of the second SVD mode (decadal timescale) of the SST–h4 pair for (left) SST and (right) h4. CIs for amplitudes are 0.1 and for phases are 30°. Both bottom panels show a westward propagation of the phase

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 13.
Fig. 13.

Lagged correlations between the h4 field and the KOE SST in the decadal band (6–13 yr). The time lag is marked in every panel over the continental U.S. map and goes from 4 to 0 yr h4 leading KOE SST. The 90% confidence levels were computed according to Davis (1976) and are 0.57 on a spatial average. Contours equal and higher, in absolute value, than 0.4 are shown. The arrows denote the time progression sense of the lag toward the zero-lag correlation. A positive h4 represents a deeper than normal pycnocline depth

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 14.
Fig. 14.

Schematic of the mean currents of the North Pacific Ocean (Stabeno and Reed 1985). These paths are very similar to the propagation paths found in this study (map taken from the Internet at http://www.pmel.noaa.gov/bering/pages/npmap4.html).

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Fig. 15.
Fig. 15.

First mode (i.e., interdecadal timescale) of the surface heat flux (solid line) vs first-mode SST (dashed line) averaged for the (top) Bering Sea area. (bottom) As in the top panel but for the second mode (i.e., decadal timescale) averaged in the KOE area

Citation: Journal of Physical Oceanography 33, 12; 10.1175/1520-0485(2003)033<2483:IDOTNP>2.0.CO;2

Table 1.

Heat budget comparison for decadal and interdecadal timescales (averaged north of 15°N)

Table 1.
Save
  • Alexander, M. A., C. Deser, and M. S. Timlin, 1999: The reemergence of SST anomalies in the North Pacific Ocean. J. Climate, 12 , 24192433.

    • Search Google Scholar
    • Export Citation
  • Anderson, P. J., and J. F. Piatt, 1999: Community reorganization in the Gulf of Alaska following ocean climate regime shift. Mar. Ecol. Prog. Ser., 189 , 117123.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, and W. B. White, 1998a: Simulation of heat storages and associated heat budgets in the Pacific Ocean, 1, El Niño-Southern Oscillation timescale. J. Geophys. Res., 103 , 2760327620.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, and W. B. White, 1998b: Simulation of heat storages and associated heat budgets in the Pacific Ocean, 2, Interdecadal timescale. J. Geophys. Res., 103 , 2762127635.

    • Search Google Scholar
    • Export Citation
  • Auad, G., A. J. Miller, J. O. Roads, and D. R. Cayan, 2001: Pacific Ocean wind stress and surface heat flux anomalies from NCEP reanalysis and observations: Cross-statistics and ocean model response. J. Geophys. Res., 106 , 2224922265.

    • Search Google Scholar
    • Export Citation
  • Barnett, T. P., M. Latif, E. Kirk, and E. Roeckner, 1991: On ENSO physics. J. Climate, 4 , 487515.

  • Barnett, T. P., D. W. Pierce, R. Saravanan, N. Schneider, D. Dommenget, and M. Latif, 1999: Origins of the midlatitude Pacific decadal variability. Geophys. Res. Lett., 26 , 14531456.

    • Search Google Scholar
    • Export Citation
  • Beamish, R. J., 1993: Climate and exceptional fish production off the west coast of North America. Can. J. Fish. Aquat. Sci., 50 , 22702291.

    • Search Google Scholar
    • Export Citation
  • Bjerknes, J., 1964: Atlantic air–sea interaction. Advances in Geophysics, Vol. 10, Academics Press, 1–82.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5 , 541560.

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

    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., A. J. Miller, T. P. Barnett, N. E. Graham, J. N. Ritchie, and J. M. Oberhuber, 1995: Seasonal-interannual fluctuations in surface temperature over the Pacific: Effects of monthly winds and heat fluxes. Natural Climate Variability on Decadal-to-Century Time Scales, National Academy Press, 133–150.

    • Search Google Scholar
    • Export Citation
  • Cayan, D. R., A. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson, 2001: Changes in the onset of spring in the western United States. Bull. Amer. Meteor. Soc., 82 , 399415.

    • Search Google Scholar
    • Export Citation
  • 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
  • Deser, C., and M. L. Blackmon, 1995: On the relationship between tropical and North Pacific sea surface temperature variations. J. Climate, 8 , 16771680.

    • Search Google Scholar
    • Export Citation
  • Deser, C., M. A. Alexander, and M. S. Timlin, 1999: Evidence for a wind-driven intensification of the Kuroshio Current Extension from the 1970s to the 1980s. J. Climate, 12 , 16971706.

    • Search Google Scholar
    • Export Citation
  • Dettinger, M. D., D. S. Battisti, R. D. Garreaud, G. J. McCabe, and C. M. Bitz, 2000: Interhemispheric effects of interannual and decadal ENSO-like climate variations on the Americas. Present and Past Interhemispheric Climate Linkages in the Americas and their Societal Effects, V. Markgraf, Ed., Cambridge University Press, 1–16.

    • Search Google Scholar
    • Export Citation
  • Evans, M. N., M. A. Cane, D. P. Schrag, A. Kaplan, B. K. Linsley, R. Villalba, and G. M. Wellington, 2001: Support for tropically-driven Pacific decadal variability based on paleoproxy evidence. Geophys. Res. Lett., 28 , 36893692.

    • Search Google Scholar
    • Export Citation
  • Frankignoul, C., P. Muller, and E. Zorita, 1997: A simple model of the decadal response of the ocean to stochastic wind forcing. J. Phys. Oceanogr., 27 , 15331546.

    • Search Google Scholar
    • Export Citation
  • Garreaud, R. D., and D. S. Battisti, 1999: Interannual ENSO and interdecadal ENSO-like variability in the Southern Hemisphere tropospheric circulation. J. Climate, 12 , 21132123.

    • Search Google Scholar
    • Export Citation
  • Kaylor, R. E., 1977: Filtering and decimation of digital time series. Institute for Physical Science and Technology Tech. Rep. BN 850, University of Maryland, 42 pp.

    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., 1990: Observations of long Rossby waves in the northern tropical Pacific. J. Geophy. Res., 95 , 51835217.

  • Killworth, P., 1996: Time interpolation of forcing fields in ocean models. J. Phys. Oceanogr., 26 , 136143.

  • Kirtman, B. P., and P. S. Schopf, 1998: Decadal variability in ENSO predictability and prediction. J. Climate, 11 , 28042822.

  • Latif, M., and T. Barnett, 1994: Causes of decadal climate variability over the North Pacific and North America. Science, 266 , 634637.

    • Search Google Scholar
    • Export Citation
  • Latif, M., and T. Barnett, 1996: Decadal climate variability over the North Pacific and North America—Dynamics and predictability. J. Climate, 9 , 24072423.

    • Search Google Scholar
    • Export Citation
  • Linsley, B. K., G. M. Wellington, and D. P. Schrag, 2000: Decadal sea surface temperature variability in the subtropical South Pacific from 1726 to 1997 a.d. Science, 290 , 11451148.

    • Search Google Scholar
    • Export Citation
  • Liu, Z., 1999: Forced planetary wave response in a thermocline gyre. J. Phys. Oceanogr., 29 , 10361055.

  • Lysne, J., and C. Deser, 2002: Wind-driven thermocline variability in the Pacific: A model–data comparison. J. Climate, 15 , 829845.

    • Search Google Scholar
    • Export Citation
  • Mantua, N. J., and S. R. Hare, 2002: The Pacific decadal oscillation. J. Oceanogr., 58 , 3544.

  • 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
  • Maximenko, N. A., P. P. Niiler, G. G. Panteleev, T. Yamagata, and D. B. Olson, 2002: Near-surface dynamical structure of the Kuroshio Extension derived from drifter and altimetry data. Exchanges, 7 , 14.

    • Search Google Scholar
    • Export Citation
  • Mestas-Nuñez, A., and D. B. Enfield, 1999: Rotated global modes of non-ENSO sea surface temperature variability. J. Climate, 12 , 27342746.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., and N. Schneider, 2000: Interdecadal climate regime dynamics in the North Pacific Ocean: Theories, observations and ecosystem impacts. Progress in Oceanography, Vol. 47, Pergamon, 355–379.

    • Search Google Scholar
    • Export Citation
  • Miller, A. J., D. R. Cayan, T. P. Barnett, N. E. Graham, and J. M. Oberhuber, 1994a: The 1976–77 climate shift of the Pacific Ocean. Oceanography, 7 , 2126.

    • Search Google Scholar
    • Export Citation
  • Miller,