1. Introduction
Water mass exchange between the tropical and subtropical Pacific Oceans plays an important role in the maintenance of pycnocline property in the tropical Pacific Ocean and is believed to affect interannual-to-decadal variability in the Tropics (Gu and Philander 1997; Zhang et al. 1999; Federov and Philander 2001; McPhaden and Zhang 2002). The exchange is accomplished by the so-called subtropical cell or STC (McCreary and Lu 1994) that connects the upper tropical and subtropical Pacific Oceans. Observational and modeling studies have advanced the knowledge about the pathways, strength, and mechanisms through which subtropical waters intrude into the equatorial Pacific Ocean (e.g., Fine et al. 1987; McPhaden and Fine 1988; McCreary and Lu 1994; Liu et al. 1994; Liu and Philander 1995; Lu and McCreary 1995; Rothstein et al. 1998; Huang and Liu 1999; Zhang et al. 1999; Harper 2000; Huang and Wang 2001). A relatively common picture emerges from these studies: waters subducted in the eastern part of the northern and southern subtropical gyres (if not too far east) travel westward and equatorward to the tropical Pacific Ocean where they join the equatorial undercurrent either through low-latitude western boundary currents (LLWBCs) or through the interior ocean. These waters upwell near the equator and then return to the subtropical gyres as Ekman transport. It has been established that pycnocline waters from the subtropics arrive at the Tropics via two pathways: through the western boundary via the LLWBCs and through the interior.
Meridional transports near 10°N and 10°S are frequently used to characterize tropical–subtropical exchange. These latitudes are equatorward of the bifurcation latitudes of the North and South Equatorial Currents that separate the tropical and subtropical gyres. Therefore, they cut across the equatorward-flowing LLWBCs and interior geostrophic flow in the pycnocline. Near 7°–12°N, Ekman pumping associated with the intertropical convergence zone (ITCZ) creates a “ridge” in the pycnocline that tends to limit the amount of equatorward geostrophic flow. There is no similar feature in the Southern Hemisphere. The importance of transport at 10°N is also reflected by the fact that a hydrographic section was occupied at that latitude as part of the World Ocean Circulation Experiment to understand exchange between the tropical and northern subtropical Pacific Ocean. Many studies suggest that mean boundary pycnocline transport is substantially larger than that through the interior near 10°N, but the difference is smaller near 10°S (cf. summary by Liu and Philander 2000).
However, less is known about the variability of boundary and interior exchange on interannual and decadal time scales that is potentially important to tropical heat content variability. Lukas (1988) found that the transport of the Mindanao Current (MC) inferred from tide gauge data was weaker before and stronger during the 1976 and 1982/83 El Niño events. Coles and Rienecker (2001) suggested that interior exchange had a seasonal cycle associated with seasonal variability of pycnocline thickness due to the seasonality of local Ekman pumping. Huang and Wang (2002) examined interannual variability of interior transport and found signals related to El Niño–Southern Oscillation (ENSO). McPhaden and Zhang (2002) analyzed hydrographic data during the past 50 years and found that pycnocline transport convergence into the tropical Pacific displays variations on multidecadal time scales. From the 1970s to the 1990s, there has been less and less subtropical pycnocline water coming to the Tropics. Together with the inference that the equatorial upwelling became weaker during this period due to weakening trade wind, they suggested that the STC had been slowing down since the 1970s. While the change of interior transport from the 1970s to the 1990s was significant, through the western boundary it was unclear (due to the relatively large uncertainty of LLWBC transport estimates).
Decadal temperature anomaly t′ has been observed to propagate from the subtropics toward the Tropics from the late 1970s to the late 1980s (Deser et al. 1996; Zhang et al. 1999). Gu and Philander (1997) hypothesized that the advection of decadal t′ originating from the subtropics by
Understanding υ′ is important to the assessment of the potential role of υ′
2. Model configuration and mean state
The OGCM of the Massachusetts Institute of Technology (Marshall et al. 1997a,b) is used for this investigation. The model configuration is the same as that adopted by Lee et al. (2002) except that the background mixing coefficients are somewhat smaller here. The model domain is global in the zonal direction and spans from 75°S to 75°N meridionally. The horizontal resolution is 1° × 1° poleward of 23°, telescoped to 1° × 0.3° in the Tropics. There are 46 vertical levels with a thickness of 10 m in the upper 150 m, gradually increasing to 400 m at depth. Two advanced mixing schemes are used by the model: the so-called K-Profile Parameterization (KPP) vertical mixing (Large et al. 1994) and the Gent and McWilliams (1990) and Redi (1982) (GM–Redi) isopycnal mixing. The model is forced at the surface by 12-hourly wind stress and daily heat and freshwater fluxes. These fluxes are based on the reanalysis product of the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR; Kalnay et al. 1996) except that the time averages are replaced by the Comprehensive Ocean–Atmosphere Data Set (COADS) product (da Silva et al. 1994). In addition to the imposed surface heat and freshwater fluxes, model SST and sea surface salinity are relaxed toward NCEP's SST and Levitus98 climatological mean salinity (Boyer and Levitus 1998), respectively, with time scales typically being 1–2 months.
After a 10-yr spinup from rest with Levitus98 climatological temperature and salinity (Boyer and Levitus 1998) forced by seasonal climatological forcings, the model is integrated using the forcings from 1980 to 2000. A more detailed description of the model configuration and comparison of the model state with satellite and in situ data are provided by Lee et al. (2002) and so will not be repeated here. However, in the latter part of this section, we compare the mean pycnocline structure and transports of the model with observational-based analyses, focusing primarily near 10°N and 10°S that are important to tropical–subtropical exchanges. For analysis of decadal variability, it would have been ideal to include the earlier decades (in particular, to capture the larger change in the mid-1970s). Unfortunately, we do not have the integration for those decades. The original objective of the 1980–2000 integration was to study seasonal-to-interannual variability during which much more data are available for validation [e.g., in situ data from the Tropical Ocean Global Atmosphere–Tropical Atmosphere Ocean (TOGA–TAO) arrays, satellite measurements of sea surface temperature, sea level, and wind].
As a way to assess the accuracy of the model simulation in terms of the variability of boundary and interior pycnocline flow, we also analyze a data assimilation product generated at the Jet Propulsion Laboratory (JPL) based on the same OGCM for the period of 1997–2000. The shortness of the assimilation period is simply due to availability of computational resources as adjoint OGCM at our resolution is fairly expensive to integrate. The data being assimilated include sea level anomalies measured by the TOPEX/Poseidon altimeter, surface fluxes that are also used as prior surface forcings for the model (i.e., the blended COADS and NCEP fluxes), and Levitus98 climatological mean temperature and salinity. The latter are given very weak weights because they represent temperature and salinity averaged over a much longer period than that of the assimilation. These data are assimilated into the OGCM using the so-called adjoint method (cf. Thacker and Long 1988; Lee and Marotzke 1998; Lee et al. 2000).
The procedure of the assimilation is as the following. First, a quadratic cost function is designed to penalize model–data misfit in sea level anomaly, surface fluxes, and time-mean temperature and salinity. The adjoint of the OGCM is then used to compute the partial derivatives of the cost function with respect to the control variables, chosen here to be initial state and surface fluxes averaged over a 10-day interval (close to the repeat cycle of the altimeter). The adjoint of the OGCM is constructed at JPL with the help of an automatic differentiation software called TAF marketed by FastOpt (more information available online at www.FastOpt.com). These partial derivatives (i.e., adjoint sensitivities) are used along with a conjugate-gradient algorithm to adjust the control variables so as to minimize the cost function. The assimilation product used for this analysis is available through a live-access data server (located online at http://eyre.jpl.nasa.gov/las). It is part of the collaborative effort by the Consortium “Estimating the Circulation and Climate of the Ocean” (ECCO; more information available online at http://ecco-group.org) funded under the National Ocean Partnership Program. An introduction of the consortium activity is given by Stammer et al. (2002, p. 289, 294–295). The assimilation differs from the simulation (for coincident years) primarily in terms of variability rather than time mean. This is because the time-mean constraint is purposely made to be very weak to accommodate the relatively short assimilation period. Comparison of variability between the assimilation and the simulation will be made in section 3. In the remaining part of this section, we discuss the mean pycnocline flow of the simulation in relation to previous observational analyses.
Figure 1 shows the zonal sections of averaged potential density from the model simulation and from Levitus98 data at 10°N and 10°S. The large-scale representation of the density structure is reasonable both in the western boundary and interior. The model's pycnocline is somewhat sharper than that of the data in the eastern part of the basin. The averaged depths of middle-to-lower pycnocline (e.g., σθ = 24–26.5 kg m−3) in the model are shallower than those in the data by about 10 m at 10°N and 20–30 m at 10°S. In the present study, pycnocline transport is defined as the volume transport between 22 and 26.5 potential-density (σθ) surfaces. When the σθ = 22 kg m−3 surface outcrops, we treat the depth of 50 m as the top of the pycnocline so as to exclude Ekman transport approximately (but nevertheless excluding some near-surface geostrophic flow as well).
To illustrate the spatial distribution of boundary and interior flow in the pycnocline, we define a “pseudo” streamfunction for pycnocline transport ψ(x, y) =
The western boundary and interior transports at 10°N are defined as pycnocline transports from the western boundary to 130°E and from this longitude to the eastern boundary, respectively. The separation of western boundary and interior at 10°S is chosen to be 159°E. The model's mean boundary and interior pycnocline transports are 15.7 and 5.6 Sv (1 Sv ≡ 106 m3 s−1) for 10°N and 9.2 and 10.1 Sv for 10°S. The boundary transport at 10°N is close to but somewhat larger than various estimates based on analyses of in situ data: 13 Sv by Lukas et al. (1991) based on ADCP measurements, 15 Sv by Wijffles (1993) based on an inverse model of hydrography, and 14 Sv by Huang and Liu (1999) based on the ocean analysis of temperature profile data using the NCEP ocean model. The interior transport at 10°N is close to the 5-Sv estimate by Johnson and McPhaden (1999) based on hydrography, and somewhat larger than the 4-Sv value obtained by Wijffles (1993) and Huang and Liu (1999).
Transports at 10°S are several Sv smaller than the observational-based analyses (Lindstrom et al. 1987; Butt and Lindstrom 1994; Johnson and McPhaden 1999; Huang and Liu 1999), especially that through the western boundary. The choice of the mean wind product is believed to affect pycnocline transport. Background mixings may also matter because they affect horizontal density gradient and thus vertical distribution of volume transport. In addition, the realism of island geometry off the coasts of New Guinea and New Ireland in our model is questionable because of the limited resolution. Lindstrom et al. (1987) and Butt and Lindstrom (1994) identified two LLWBCs in the South Pacific, the New Guinea, and New Ireland Coastal Undercurrents, carrying about 7 and 5 Sv of transport, respectively. In our solution, the distinction of these two currents is not clear because various straits in the region are not adequately represented. However, the boundary current transports reported by Lindstrom et al. (1987) and Butt and Lindstrom (1994) are based on fairly short temporal records. It is not clear whether the reported transports are representative of long-term averages. While we are working toward improving the model's mean state, it is the variability of pycnocline transport that is the focus of the present study. As will be discussed, the variability of pycnocline transport is primarily controlled by the variability of wind forcing. Therefore, the findings with regard to variability may not be strongly subject to the limitation of mean transports at 10°S.
3. Interannual-to-decadal variations of pycnocline transport
Anomalies of boundary, interior, and net pycnocline transports (i.e., with time mean removed) are shown in Fig. 3. Two features are seen: the variation of boundary transport tends to be anticorrelated to and generally has a smaller magnitude than that of interior transport. The correlation coefficients are −0.73 for 10°N and −0.93 for 10°S, with both being significant at 95% confidence level. The standard deviation of pycnocline transports through the boundary and interior as well as their sum are listed in Table 1. The compensation of interior flow by boundary flow at 10°S is larger than that at 10°N. Let σi and σn represent standard deviations of interior and net pycnocline transports, respectively. Then the ratio C = (σi − σn)/σi can be used to characterize the degree of compensation of interior flow by boundary flow. The value of C is 31% at 10°N and 62% at 10°S. The partial compensation of interior by boundary flow not only occurs on interannual but on decadal time scales as well. Figure 4 shows anomalies of boundary and interior transport after a 5-yr smoothing has been applied. There is less subtropical pycnocline water intrusion into the Tropics in the 1990s than in the 1980s. The opposite is true for boundary transport anomaly.
Based on analysis of hydrography data during the past 50 years, McPhaden and Zhang (2002) found that equatorward convergence of interior pycnocline flow into the 9°N–9°S band had been decreasing from the 1970s to the 1990s. A similar decrease was found in the convergence into the 15°N–15°S band. Although we do not have the simulation for the 1970s, the mean difference of interior transport between the 1990s and 1980s in our model, 2.9 Sv, is comparable to the estimate by McPhaden and Zhang (2002) of 3–4 Sv (with an uncertainty of 1–2 Sv). In their analysis, the estimated boundary transport was fairly uncertain (several Sv) due to insufficient data. In the present study, decadal difference in boundary transport at 10°N is −1.5 Sv (more subtropical pycnocline water intrusion into the Tropics in the 1990s). So approximately 50% of the decadal change of interior flow is compensated by an opposite change of boundary flow at this latitude. As will be seen in the next section (in conjunction with Fig. 10d), compensation on decadal time scales is found for the latitude band of approximately 9°–15°N. This finding prompts for additional measurements in the boundary to resolve decadal variability. On average, much of the ITF is fed by the MC. Does the 1.5-Sv decadal change in pycnocline transport associated with the MC affect the Indian Ocean through the ITF or does it influence in the tropical Pacific only? In our model, decadal change in ITF transport is only 0.1 Sv, suggesting that the effect of decadal variation in pycnocline flow at 10°N is on the tropical Pacific.
At 10°S, a near-decadal signal is also seen in the interior pycnocline transport (Fig. 4b). Also evident is the compensation by boundary flow. However, the phase of the decadal signal of interior or boundary transport at 10°S leads the one at 10°N by about 3 years. The difference of interior pycnocline transport at 10°S between the 1990s and 1980s is 0.4 Sv. This is smaller than the corresponding estimate of about 3 Sv by McPhaden and Zhang (2002). Note that the error bars for their estimate in the Southern Hemisphere are larger than those in the Northern Hemisphere. Therefore, the difference from their estimate is not all that significant. Furthermore, as McPhaden and Zhang noted, there are differences in decadal variability among different wind products. It is not clear whether a different wind product would result in a larger decadal change or not.
Figure 5 shows zonal sections of meridional velocity of the LLWBCs at 10°N and 10°S averaged annually from 1997 to 2000. The MC is strongest in 1997, an El Niño year, and weaker from 1998 to 2000, which is more characterized by the La Niña condition. The core depth of the MC is shallower in 1997 than it is in 1998–2000. The NGCU also displays similar interannual changes as that of the MC. However, the phase lags that of the MC by about a year (strongest and shallowest in 1998 instead of in 1997). As will be discussed in section 4, this is related to the phase difference of wind stress curl anomalies near 10°N and 10°S. The generally shallower core depth of the LLWBCs during the predominant El Niño condition reflects the large-scale zonal tilt of the pycnocline as the westerly wind anomaly during El Niño raises and depresses the pycnocline in the western and eastern part of the basin, respectively. Vice versa for La Niña. The cause for the variation in the strength of the MC, as discussed in the next section, is an off-equatorial wind stress curl. Similar variations in the core depth and strength of the MC associated with the 1982/83 and 1986/87 El Niño and the subsequent La Niña events are also found (not shown). We focus only on the 1997–2000 period to facilitate the comparison with the data assimilation product.
To assess the accuracy of the model simulation, interannual changes of the LLWBCs are compared with those inferred from the data assimilation product discussed in section 2. Figure 6 shows the meridional velocity sections at 10°N(S) for the LLWBCs from 1997 to 2000 estimated by the assimilation (the counterpart of Fig. 5, but from the assimilation). The variability in the simulation is qualitatively similar to that in the assimilation. However, changes in the strength and core depth of the LLWBCs from El Niño to La Niña in the simulation are not as large as those inferred from the assimilation. This is because the simulation underestimates the interannual change of tropical zonal wind and consequently the variation of the zonal tilt of the pycnocline. In fact, the magnitude of sea level anomaly associated with the 1997 El Niño in the simulation is too small because the westerly wind anomaly in 1997 (from the NCEP product) is too weak. The assimilation of sea level data results in more realistic anomalies of westerly wind and sea level (and thus more reliable variability in the core depth and circulation of the pycnocline).
Decadal variation of the MC resulted from the simulation is shown in Fig. 7. It is stronger in the 1990s than it is in the 1980s. Its core depth is shallower in the 1990s when it is stronger. These features are somewhat similar to the difference in the characteristics of the MC El Niño and La Niña years as shown in Figs. 5 and 6. The easterly trade winds in the equatorial Pacific are known to be weaker in the 1990s than the 1980s. The pycnocline is expected to become shallower in the west and deeper in the east in the 1990s in response to the change of tropical wind. Therefore, the shallower core depth of the MC in the 1990s reflects large-scale tropical zonal wind in a way that is similar to interannual changes associated with El Niño and La Niña. The decadal change in the core depth of the MC is consistent with the decadal change in the tilt of pycnocline depth reported by McPhaden and Zhang (2002).
There are insufficient in situ data to characterize interannual variation of pycnocline flow via the LLWBCs. However, sea level observations such as those obtained by the TOPEX/Poseidon altimeter allow one to infer the variability of near-surface meridional geostrophic flow, which is relatively coherent with the flow in the deeper part of the pycnocline (as seen from Figs. 5 and 6). In fact, Lukas (1988) used in situ sea level data to examine the variability of the MC. He found that interannual change in the strength of MC was anticorrelated with the equatorial heat content during the 1976 and 1982 El Niño events.
Figure 8 shows time series of anomalous east–west difference in sea level across the western boundary and across interior for the model, TOPEX/Poseidon data, and the assimilation. The respective time averages for the period of 1997–2000 have been removed to facilitate the comparison with the assimilation product available for this period. They are proportional to the magnitude of near-surface geostrophic transport (with reversed sign in the Southern Hemisphere). Both the model and the data suggest that the variability of near-surface geostrophic flow in the boundary is 1) anticorrelated to and 2) generally smaller than that in the interior. The magnitude of variability in the model simulation is smaller than that of the data at 10°N. At 10°S, the model also underestimates the variability since 1997, but overestimates it during 1994–96. These differences are believed to be related to the fidelity of the wind stress product (NCEP) in representing actual wind variability. The differences between the assimilation and the data are much smaller. The similarity between the model simulation and the data indicates that the model has a reasonable skill in simulating the relative variability of pycnocline flow in the boundary and interior.
The above findings highlight two major differences between the mean and variability of tropical–subtropical mass exchange in terms of boundary and interior pathways. First of all, time-mean boundary and interior transports
4. Forcing mechanisms
Forcing mechanisms responsible for the variability of boundary and interior transports are examined in this section. In particular, we investigate processes responsible for the counteracting tendency of boundary and interior pycnocline transports and the larger variability of the latter. The importance of three possible mechanisms is evaluated: 1) forcing by equatorial zonal wind stress, 2) forcing by off-equatorial wind stress curl, and 3) effect of the Indonesian Throughflow.
Using a reduced-gravity model of the tropical Pacific forced by observed wind for the period of 1970–87, Zebiak (1989) found that meridional transports through the western boundary at 5°N and 5°S generally opposed the tendency of equatorial heat content, but were more than compensated by the interior transports. They suggested that the tendency was consistent with the reflection of equatorial long Rossby waves (Cane and Sarachik 1983). Equatorial westerly wind generated eastward-propagating, downwelling Kelvin waves. Associated with these waves was an eastward equatorial jet. Water carried by this jet returned poleward farther east in the interior. Meridional mode of westward-propagating, upwelling Rossby waves was also excited by the equatorial westerly wind. When these waves impinged on the western boundary, they produced convergence of flow toward the equator, which compensated for waters carried eastward by the equatorial jet. Anticorrelated boundary and interior transports at 5°N(S) were also found in a similar model by Springer et al. (1990) for the period of 1979–83. They suggested that the convergence of boundary flow appeared to be related to the reflection of equatorial Rossby waves only after the onset of the 1982/83 El Niño event, but not before. However, transports at 10°N and 10°S were not examined in these studies.
Consistent with the simulations by these reduced-gravity models, counteracting tendencies between boundary and interior flow are seen at 5°N(S) in our model. Figure 9 compares boundary, interior, and net pycnocline transports at 5°N with the corresponding ones at 10°N. Net pycnocline flow across 5° and 10°N (Fig. 9c), being the lower branch of the STC at two different latitudes, are close to each other. It suggests that diapycnal convergence/divergence of pycnocline water and the change in the volume of pycnocline water between the latitudes due to along-isopycnal flow are small on interannual and decadal time scales. Interior transports between these two latitudes are also somewhat coherent (Fig. 9b). However, boundary transports across the two latitudes are not similar at all (Fig. 9a). On interannual time scales, there is little coherence at any time lag between boundary transports at 5° and 10°N. On decadal time scales, the tendencies of the two are actually opposite. If boundary transports at 5° and 10°N are both associated with the reflection of Rossby waves excited by equatorial zonal wind, one would expect the variability at 10°N to be substantially smaller than that at 5°N because the amplitude of equatorial Rossby waves decreases substantially from 5° to 10°N. In the model, however, the magnitudes of boundary transport variability at these two latitudes are comparable. These features suggest that the reflection of equatorial Rossby waves, which presumably affects boundary transports at 5°N significantly, may not be very important at 10°N.
To investigate changes in pycnocline circulation that affect transports at 5° and 10°N(S), interannual and decadal changes of pseudo streamfunction ψ(x, y) (defined in section 2) are shown in Fig. 10. Two patterns of anomalous circulation affect transports at 10°N(S): one that originates in the equatorial band and one that centers at latitudes near and poleward of 10°N(S). The former is more easily seen in the change from 1996 to 1997 (Fig. 10a), characterized by intense circulation anomaly in the equatorial band with maxima near 3°N and 4°S and with the intensity decreasing farther away from the equator. This is reminiscent of the effect of equatorial Rossby waves, which have maximum aplitude near these latitudes. On top of this pattern is an anomalous circulation centered between 10° and 15°N west of the date line. This off-equatorial pattern changes direction from 1997 to 1998 (Fig. 10b) and from 1998 to 1999 (Fig. 10c). Off-equatorial anomaly of circulation is also found in the Southern Hemisphere (more apparent in Figs. 10b,c). While that near 10°N is more zonally oriented, the one near 10°S is tilted in the northwest–southeast direction. These off-equatorial circulation anomalies have a larger influence on transports at 10°N(S) than the patterns centered in the equatorial band. Similar features of anomalous pycnocline circulation are found during the 1981–84 period (not shown) that features the development of another large El Niño and the subsequent La Niña events.
Change in pycnocline circulation from the 1980s to the 1990s (Fig. 10d) is characterized by an off-equatorial pattern centered between 10° and 15°N in the western to central Pacific. It resembles the difference between 1997 and 1998 (Fig. 11b, but with opposite direction). Figure 10d also explains why the decadal tendencies of boundary transport at 5° and 10°N are different. The anomalous interior flow across 10°N in the 1980s relative to the 1990s does not reach the equatorial band through the interior, but connects with anomalous boundary flow at 5°N.
The pycnocline circulation patterns indicate that, both on interannual and decadal time scales, change in off-equatorial circulation affects boundary transports at 10°N(S) significantly. The direction of off-equatorial anomalous circulation near the western boundary in the pycnocline is somewhat similar to that of full-depth integrated circulation in many cases (Fig. 11). This consistency suggests that, to some extent, the change in pycnocline circulation near the western boundary reflects the change of full-depth integrated horizontal circulation. As discussed in the following, the change in horizontal circulation is found to be primarily a result of Ekman pumping by off-equatorial wind stress curl in the western part of the basin.
Time-mean wind stress curl and its zonal average are presented in Figs. 12a,b. Their local maxima and minima occur near 10°N and 10°S, respectively, both resulting in local maxima in Ekman pumping (defined as wind stress curl divided by planetary vorticity that is negative in the Southern Hemisphere). While the curl pattern in the Northern Hemisphere is more zonally oriented, in the Southern Hemisphere it is tilted along the northwest–southeast direction. Interannual-to-decadal variability of the curl (Figs. 12c,d) has local maxima near 15°N and 10°S (Figs. 12c,d), primarily due to curl anomalies over the western part of the basin. These anomalies are caused by meridional movement and the change in the strength of the curl pattern. An example is shown in Fig. 13 for the period of 1997–99 both for the NCEP product (Figs. 13a,c,e) and for the second European Remote Sensing Satellite (ERS-2) scatterometer measurement (Figs. 13b,d,f). The band of maximum wind stress curl near the western Pacific near 10°N is tilted upward in 1997, downward in 1998, and upward again in 1999 (as marked by the black line segments). Changes in the pattern of wind stress curl resemble those of the ITCZ and the South Pacific convergence zone (SPCZ). A similar feature is seen on decadal time scales for the NCEP product (not shown).
The effect of wind on the ocean is accomplished by Ekman pumping, ∇ × (τ/f) = (∇ × τ + βτx/f)/f, where τ is wind stress and τx is its zonal component, f is planetary vorticity, and β = ∂f/∂y. We see that βτx/f contributes to Ekman pumping in addition to ∇ × τ. The contribution by βτx/f is very large near the equator where f is small, but drops off substantially away from the equator. We have computed the interannual-to-decadal standard deviation of βτx/f and found that it is 3–4 times smaller than that of ∇ × τ near 10°N and 10°S. Therefore, the following discussion focuses on the variability of ∇ × τ that is dominant near these latitudes.
To illustrate the variation of off-equatorial wind stress curl and the corresponding change in oceanic state, anomalies of wind stress curl and sea level for the period of 1997–99 (referenced to the mean of this period) are shown in Fig. 14. In 1997, the negative sea level anomaly near 10°–15°N and west of 160°E corresponds to the positive curl anomaly primarily west of the date line. The positive sea level anomaly west of 160°E developing from 1998 to 1999 corresponds to the negative curl anomaly during the same period. In the Southern Hemisphere, the sea level anomaly is related to wind stress curl anomaly of the same sign because planetary vorticity is negative south of the equator. Near 10°S, the sea level anomaly changes from being negative in 1998 to being positive in 1999, corresponding to predominately negative (positive) curl anomaly in 1998 (1999). The anomalies near 10°N tend to be more zonal while those near 10°S are tilted along the northwest–southeast direction. These characteristics are consistent with the orientations of the anomalous patterns of off-equatorial circulation shown in Fig. 10. The covariability of wind stress curl and sea level in the model is generally consistent with wind measurement by the ERS-2 scatterometer and sea level measurement by the TOPEX/Poseidon altimeter (Fig. 15). Nevertheless, the NCEP product underestimates the magnitude of wind stress curl anomaly near 10°N(S), giving rise to smaller variation of sea level.
In principle, the variability of sea level in the western Pacific reflects the integrated effect of wind stress curl across the basin. However, the anomaly of wind stress curl over the western part of the basin near 10°N(S) is much larger than that in the east such that the variability of sea level and pycnocline flow in the western Pacific are largely forced locally. In fact, boundary pycnocline transports at 10°N(S) are reasonably well correlated to “local” Sverdrup transports (with the sign reversed) computed from curl anomalies over the western part of the basin (Fig. 16), suggesting the important role of these wind stress curl anomalies in causing the variability of boundary flow. The magnitude of pycnocline transports are generally smaller than the local Sverdrup transports because the former is only part of the full-depth integrated transport inferred from the Sverdrup relation.
The coherent changes in pycnocline circulation, sea level, and wind stress curl indicate that local wind stress curl anomalies near 10°N(S) modify the strength of the horizontal circulation to create counteracting tendency of boundary and interior transport in the pycnocline. This effect is further examined through a sensitivity experiment. The control experiment is the last year of the spinup with seasonal forcing. A wind perturbation is introduced in the perturbation experiment. The zonal and meridional wind perturbation have the form of τx = ∂P/∂y and τy = ∂P/∂x, where P is a Gaussian “wind stress potential” with an e-folding scale of 5° and 100° in meridional and zonal directions, respectively. The pattern is centered at 10°N, 170°W. The effective magnitude of the wind stress curl perturbation is 2 × 10−8 N m−3, close to the observed variability of wind stress curl near 10°N. The perturbed boundary and interior transports integrated over the entire water column and in the pycnocline are coherent (Fig. 17), both showing a counteracting tendency that reflects the change in the strength of horizontal circulation. This is consistent with the fact that the directions of anomalous pycnocline and barotropic circulation near the western boundary generally agree with each other as discussed earlier in conjunction with Figs. 10 and 11. The perturbed pycnocline transports through the boundary and interior more or less cancel each other out.
In steady or equilibrium state, boundary transport would compensate the Sverdrup interior transport exactly. However, as one of the reviewers pointed out, transient boundary transport may not fully compensate interior transport because it takes time for Rossby waves generated in the interior to reach the western boundary. The near compensation of perturbed boundary and interior transports both over the full depth and in the pycnocline as shown by Fig. 17 suggests that the noncompensation due to transient Rossby wave adjustment might be small (insufficient to explain the noncompensation simulated by the model for the 1980–2000 period). Other sensitivity experiments have also been performed by reversing the sign of the perturbed wind stress curl and by changing the location of the perturbed curl to 10°S and examining the response of transports there. The near compensation between perturbed boundary and interior transports is consistently found, again indicating that Rossby wave adjustment time may not be the first-order process in explaining the noncompensated boundary and interior flow.
In the 1980–2000 model simulation, changes in boundary and interior pycnocline transports do not cancel each other out. There is a net pycnocline transport because interior flow is larger than that through the boundary. This suggests a second forcing mechanism of the pycnocline flow, in particular, one that affects the net pycnocline transport. On longer time scales, net pycnocline transport (the lower branch of the STC) should approximately balance the near-surface Ekman flow (the upper branch of the STC). Near-equatorial zonal wind is known to vary on interannual and decadal time scales associated with ENSO and decadal variability. This would affect the strength of the STC and thus the net pycnocline transport (i.e., the sum of boundary and interior pycnocline transports). Net pycnocline transport at 10°N(S) is compared with zonal wind stress averaged between the equator and 10°N(S) (Fig. 18). To facilitate the comparison in a comparable scale, the wind stress has been scaled by the values of f at 10°N(S) to produce transport unit (Ekman transport is wind stress divided by f). They are reasonably well correlated even on interannual time scales, reflecting the response of the lower branch of the STC to near-equatorial zonal wind on these time scales. Generally speaking, there is less equatorward net pycnocline transport during El Niño when the trade wind is more westerly, corresponding to weaker poleward Ekman flow. The opposite is true during La Niña.
To further examine the effect of near-equatorial zonal wind on net pycnocline transport, another perturbation experiment is performed in which the wind field is perturbed by a time-constant, globally uniform westerly wind. The magnitude of the westerly is close to the interannual-to-decadal standard deviation of zonal wind stress averaged over the Tropics. The choice of the uniform westerly wind isolates the effect of zonal wind without introducing off-equatorial curl. Both at 10°N and 10°S, perturbed interior transport is found to be larger than that of the boundary (Fig. 19), giving rise to a net pycnocline transport.
The dominance of interior transport in response to the uniform westerly wind perturbation can be explained by the following processes. The westerly wind perturbation generates positive sea level anomalies that propagate eastward as equatorial Kelvin waves. They reach the eastern boundary in a couple of months and are deflected poleward as coastal Kelvin waves that can easily reach 10°N and 10°S. The positive sea level anomalies reaching these latitudes also propagate westward as Rossby waves. While barotropic Rossby waves travel across the basin quickly, the much-slower baroclinic ones gradually spread the front of positive sea level anomaly toward the west. Therefore, the sea level anomaly is higher in the eastern than the western part of the basin, resulting in a poleward geostrophic flow that is against the equatorward Ekman flow caused by the westerly wind perturbation. Much of the perturbed geostrophic flow is in the interior because it takes time for the Rossby waves to travel from the eastern to the western boundaries.
In Fig. 19, there is a small perturbed boundary transport in opposite direction to the perturbed interior flow. This is consistent with the following argument. As discussed earlier, Ekman pumping is contributed by ∇ × τ as well as βτx/f. There is no ∇ × τ associated with uniform zonal wind perturbation, but βτx/f is not zero. Although its value falls off quickly away from the equator, it still has a small effect near 10°N and 10°S that acts effectively like a small ∇ × τ. From the first mechanism discussed earlier, this tends to create a small counteracting flow in the western boundary. Therefore, the equatorial zonal wind anomaly, without any change in off-equatorial curl, would also have some influence, although small, at latitudes as far as 10° away from the equator.
The variability of off-equatorial wind stress curl near 10°N(S) is correlated to near-equatorial zonal wind stress (Fig. 20). The general tendency is that when near-equatorial zonal wind is more westerly, wind stress curl near 10°N(S) tends to be more positive (negative), both creating an anomalously low sea level near the western boundary and an anomalous equatorward pycnocline flow in the boundary. Figure 21 illustrates the combined effect of the two forcings (the Northern Hemisphere example) in creating counteracting boundary and interior pycnocline transports, with the latter being dominant. In summary, off-equatorial wind stress curl modifies the strength of horizontal circulation to create counteracting boundary and interior transports in the pycnocline that have comparable magnitudes; the near-equatorial zonal wind affects the strength of the STC and thus the net pycnocline transport (mostly distributed in the interior). The combined effect is a larger interior transport than boundary transport.
The proposed forcing mechanism can explain not only interannual variability, but decadal variability of boundary and interior pycnocline flow at 10°N as well. This is because equatorial zonal wind stress and off-equatorial wind stress curl covary both on interannual and decadal time scales. On interannual time scales, when equatorial zonal wind stress is more westerly (e.g., associated with El Niño), the ITCZ tends to shift equatorward while its intensity changes. These changes result in a curl anomaly near 10°N. A similar behavior is seen on decadal time scales. Equatorial zonal wind stress is more westerly in the 1990s than it is in the 1980s (solid curve in Fig. 22a). Associated with this is a more positive difference of wind stress curl between 10° and 15°N (Fig. 22b) and a cyclonic anomaly of pycnocline circulation in the western Pacific centering near 12°N (Fig. 10d). These features resemble the difference between El Niño and La Niña years. For instance, near-equatorial zonal wind in 1997 was more westerly than that in 1998 west of 140°W (dashed curve in Fig. 22a); this corresponds to a more positive curl in 1997 west of the date line (Fig. 22c) and a more cyclonic pycnocline circulation in the western Pacific centered near 12°N in 1997 than in 1998 (i.e., the pattern seen in Fig. 10b but with reversed direction, because the difference between 1998 and 1997 is shown in Fig. 10b). The anomaly of westerly wind in the 1990s (relative to the 1980s) spans from 150°E to 110°W while that associated with the 1997 El Niño is west of 140°W (Fig. 22a). A similar difference in zonal extent is seen for the anomalies of wind stress curl between interannual and decadal time scales (Figs. 22b,c). The anomalous patterns of pycnocline circulation on interannual and decadal time scales (Fig. 10) also display such a difference in zonal extent. For the Southern Hemisphere, the decadal variation of wind stress curl is smaller (Fig. 22b), consistent with the lack of decadal variability in boundary and interior pycnocline flow seen in Fig. 4b (if 1990 is chosen to be the separation between the two decades).
According to the proposed forcing mechanism, the relative contributions of off-equatorial wind stress curl and near-equatorial zonal wind stress would dictate the relative variability of boundary and interior transports. The larger the off-equatorial wind stress curl, the larger the change in horizontal recirculation and thus the stronger the compensation between the boundary and interior flow. On the other hand, the larger the near-equatorial zonal wind stress, the more dominant the variability of the interior flow becomes, reflecting the change in meridional circulation. Temporal standard deviation of local wind stress curl near 10°N is smaller than that near 10°S by 20%–40% (primarily contributed by interannual variability), depending on whether the western Pacific or the entire basin is being considered (Fig. 12). On the other hand, the temporal standard deviation of near-equatorial zonal wind stress north of the equator is larger than that south of the equator by 10%–30%, depending on the latitude band being considered [e.g., 0°–10°, 0°–5°, or 5°–10°N(S)]. These relative magnitudes of off-equatorial wind stress curl and near-equatorial zonal wind imply that the change of horizontal circulation near 10°S would be relatively more dominant than that near 10°N. This is consistent with the larger compensation between the boundary and interior pycnocline near 10°S rather than near 10°N.
The above discussion focuses on wind forcings. The possible role of the ITF is discussed in the following. Using the same OGCM with somewhat different background mixing coefficients, Lee et al. (2002) found that the time-mean strength of the NGCU is reduced significantly when the ITF is blocked off. This is because the circulation loop that goes around Australia has been cut off. The effect on time-mean strength of the MC, however, is relatively small because the MC is north of this circulation loop. Does the ITF affect the variability of tropical–subtropical exchange?
To address this issue, we analyze boundary and interior transports obtained from the two simulations performed by Lee et al. (2002) with and without the ITF (Fig. 23). The counteracting tendency between the boundary and interior flow and the larger magnitude of the latter are found to be insensitive to the presence or absence of the ITF. Even the decadal signal at 10°N is unaffected by the ITF. At 10°S, the net pycnocline transport drifts away when the ITF is blocked off. The open–closed ITF experiments are somewhat idealized because the wind field is assumed to be unaffected by the presence (or absence) of the ITF. In reality, longer-term variation in the ITF may change the air–sea heat flux, especially in the southern Indian Ocean (Hirst and Godfrey 1993; Lee et al. 2002), which may modify the wind over the Pacific Ocean through atmospheric teleconnection to affect tropical–subtropical exchange in that ocean. This is beyond the scope of the present study because it requires sensitivity experiments using a coupled ocean–atmospheric model.
5. Summary and conclusions
Interannual-to-decadal variations of tropical–subtropical mass exchange in the Pacific Ocean are investigated using a near-global ocean general circulation model for the period of 1980–2000 along with sea level and wind measurements by satellite altimeter and scatterometer and a data assimilation product in the 1990s. The analysis focuses on the variability of pycnocline transports through the western boundary and interior across 10°N and 10°S.
In contrast to time-mean exchange where boundary and interior pycnocline transports are both equatorward, the variations of boundary and interior pycnocline transports are found to be generally anticorrelated to each other. Moreover, the variation of boundary pycnocline transport is smaller than that of the interior. This is again different from the time-mean exchange where the boundary transport at 10°N is substantially larger than that through the interior. Interannual variations of boundary and pycnocline transports are consistent with near-surface geostrophic flow inferred from sea level data collected by the TOPEX/Poseidon altimeter. Equatorward interior pycnocline flow is weaker in the 1990s than it is in the 1980s, consistent with recent analysis of observational data. Nevertheless, approximately half of it is compensated by an opposite change in the boundary flow. This result indicates that the interior exchange in the North Pacific is more important to interannual and decadal variability in the tropical Pacific despite the fact that the western boundary current (the Mindanao Current) carries more subtropical pycnocline water to the Tropics than the interior does on the time mean.
The counteracting tendency of the boundary and interior flow and the more dominant variation of the latter are attributed to the combined effect of variations associated with off-equatorial wind stress curl primarily over the western Pacific and near-equatorial zonal wind stress. These two forcings are correlated on interannual and decadal time scales. The former changes the strength of the horizontal circulation and results in a variation of the boundary pycnocline flow that is opposite in direction and comparable in magnitude to that of the interior pycnocline flow. The latter primarily affects the strength of the shallow meridional overturning circulation with a primarily interior pycnocline flow opposing the surface Ekman flow. The proposed forcing mechanisms also provide a possible explanation of the differences in interannual variability between 10°N and 10°S. The large compensation larger compensation of the boundary and interior flow at 10°S on interannual time scales is consistent with the larger interannual variability of wind stress curl near 10°S. Namely, the variation in the strength of horizontal circulation may dominate near 10°S. The more dominant variability of interior than boundary transport at 10°N is consistent with the larger variation of zonal wind stress in the northern tropical Pacific. In other words, the variation in the strength of the meridional circulation may be more important near 10°N. Although the presence or absence of the Indonesian Throughflow affects time-mean boundary transport at 10°S significantly, it does not affect the relative variability of boundary and interior pycnocline transports.
Various aspects related to the variability of boundary and interior pycnocline flow of the model are compared with satellite data, inferences from in situ observation, and data assimilation. The accuracy of the variability of pycnocline flow simulated by the model depends much on how realistic the wind product is. Smaller variability of the NCEP wind in the late 1990s compared to satellite scatterometer measurement results in generally weaker variability of boundary and interior flow simulated by the model in comparison with satellite altimeter data. To the first order, however, the model has fidelity in describing the relative variability of boundary and interior pycnocline flow.
The sensitivity experiments with open and closed Indonesian Throughflow channels are performed with the same code used for the default model integration but with different mixing coefficients. Specifically, the KPP background vertical diffusion and GM–Redi background mixing coefficient are both twice as large; the KPP background viscosity is an order of magnitude larger. The similarities of the boundary and interior transports simulated by these sensitivity experiments to those simulated by the default integration indicate that our findings are not sensitive to these mixing coefficients. We do not have a higher-resolution integration to evaluate how the findings depend on model resolution. Given the first-order agreement of the model and data in terms of the variability of zonal sea level slope associated with boundary and interior flow, we believe that our conclusions are relatively robust. Higher resolution is expected to further improve the consistency between the model and the data.
We have also examined seasonal variability of the boundary and interior transports. Similar to the interannual-to-decadal variations, the seasonal anomaly of interior transport is also larger in magnitude and is somewhat compensated by the boundary transport. The anticorrelation between the boundary and interior transport is −0.64 at 10°N and −0.25 at 10°S. In comparison, those for interannual-to-decadal time scales are −0.73 10°N and −0.93 at 10°S. On the seasonal time scale, while the mechanisms discussed in this paper may also play some role, there could also be other (more transient) processes involved. Seasonal exchange is beyond the scope of this study as our focus here is on interannual-to-decadal exchanges that have a potentially more direct impact on ENSO and its decadal modulation.
Acknowledgments
The research described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (NASA). The supercomputing was performed on an SGI-2000 of the JPL Supercomputing Project and an SGI-3000 of NASA Ames Research Center.
REFERENCES
Boyer, T. P., and S. Levitus, 1998: Objective analysis of temperature and salinity for the world ocean on a 1/4° grid. NOAA Atlas NESDIS, Washington, DC, 62 pp.
Butt, J., and E. Lindstrom, 1994: Currents off the east coast of New Ireland, Papua New Guinea, and their relevance to regional undercurrents in the western equatorial Pacific Ocean. J. Geophys. Res., 99 , 12503–12514.
Cane, M. A., and E. S. Sarachik, 1983: Seasonal heat transport in a forced equatorial baroclinic model. J. Phys. Oceanogr., 13 , 1744–1746.
Coles, V. J., and M. M. Rienecker, 2001: North Pacific subtropical–tropical gyre exchange in the thermocline: Simulations with two isopycic OGCMs. J. Phys. Oceanogr., 31 , 2590–2611.
da Silva, A., A. C. Young, and S. Levitus, 1994: Atlas of Surface Marine Data. NOAA Atlas NESDIS 6, 83 pp.
Deser, C., M. A. Alexander, and M. S. Timlin, 1996: Upper-ocean thermal variations in the North Pacific during 1970–1991. J. Climate, 9 , 1840–1855.
Fedorov, A. V., and S. G. Philander, 2001: A stability analysis of tropical ocean–atmosphere interactions: Bridging measurements and theory for El Niño. J. Climate, 14 , 3086–3101.
Fine, R. A., W. H. Peterson, and H. G. Ostlund, 1987: The penetration of tritium into the tropical Pacific. J. Phys. Oceanogr., 17 , 553–564.
Gent, P., and J. McWilliams, 1990: Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr., 20 , 150–155.
Gu, D., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend on exchanges between the Tropics and extratropics. Science, 275 , 805–807.
Harper, S., 2000: Thermocline ventilation and pathways of tropical–subtropical water mass exchange. Tellus, 52A , 330–345.
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 , 8971–8988.
Hirst, A. C., and J. S. Godfrey, 1993: The role of Indonesian throughflow in a global ocean GCM. J. Phys. Oceanogr., 23 , 1057–1086.
Huang, B., and Z. Liu, 1999: Pacific subtropical–tropical thermocline water exchange in the National Centers for Environmental Prediction ocean model. J. Geophys. Res., 104 , 11065–11076.
Huang, R. X., and Q. Wang, 2001: Interior communication from the subtropical to the tropical oceans. J. Phys. Oceanogr., 31 , 3538–3500.
Johnson, G. C., and M. J. McPhaden, 1999: Interior pycnocline flow from the subtropical to the equatorial Pacific Ocean. J. Phys. Oceanogr., 29 , 3073–3089.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kleeman, R., J. P. McCreary, and B. Klinger, 1999: A mechanism for generating ENSO decadal variability. Geophys. Res. Lett., 26 , 1743–1746.
Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys., 32 , 363–403.
Lee, T., and J. Marotzke, 1998: Seasonal cycles of meridional overturning and heat transport of the Indian Ocean. J. Phys. Oceanogr., 28 , 923–943.
Lee, T., J-P. Boulanger, A. Foo, L-L. Fu, and R. Giering, 2000: Data assimilation by an intermediate coupled ocean–atmosphere model: Application to the 1997–1998 El Niño. J. Geophys. Res., 105 , 26063–26087.
Lee, T., I. Fukumori, D. Menemenlis, Z. Xing, and L-L. Fu, 2002: Effects of the Indonesian throughflow on the Pacific and Indian Oceans. J. Phys. Oceanogr., 32 , 1404–1429.
Lindstrom, E., R. Lukas, R. Fine, E. Firing, S. Godfrey, G. Meyers, and M. Tsuchiya, 1987: The western equatorial Pacific Ocean circulation study. Nature, 330 , 533–537.
Liu, Z., and S. G. H. Philander, 1995: How different wind stress patterns affect the tropical subtropical circulations of the upper ocean. J. Phys. Oceanogr., 25 , 449–462.
Liu, Z., and S. G. H. Philander, 2000: Tropical–extratropical oceanic exchange. Ocean Circulation and Climate: Observing and Modelling the Global Ocean, G. Siedler, J. Church, and J. Gould, Eds., AP International Geophysical Series, Vol. 77, Academic Press, 247–257.
Liu, Z., S. G. H. Philander, and R. C. Pacanowski, 1994: A GCM study of tropical–extratropical upper ocean water exchange. J. Phys. Oceanogr., 24 , 2606–2623.
Lu, P., and J. P. McCreary, 1995: Influence of the ITCZ on the flow of thermocline water from the subtropical to the equatorial Pacific Ocean. J. Phys. Oceanogr., 25 , 2076–2088.
Lukas, R., 1988: Interannual fluctuations of the Mindanao Current inferred from sea level. J. Geophys. Res., 93 , 6744–6748.
Lukas, R., E. Firing, P. Hacker, P. L. Richardson, C. A. Collins, R. Fine, and R. Gammon, 1991: Observations of the Mindanao Current during the Western Equatorial Pacific Ocean Circulation study. J. Geophys. Res., 96 , 7089–7104.
Marshall, J., A. Adcroft, C. Hill, L. Perelman, and C. Heisey, 1997a: A finite-volume, incompressible Navier–Stokes model for studies of the ocean on parallel computers. J. Geophys. Res., 102 , 5753–5766.
Marshall, J., C. Hill, L. Perelman, and A. Adcroft, 1997b: Hydrostatic, quasi-hydrostatic and nonhydrostatic ocean modeling. J. Geophys. Res., 102 , 5733–5752.
McCreary, J. P., and P. Lu, 1994: Interaction between the subtropical and equatorial ocean circulations: The subtropical cell. J. Phys. Oceanogr., 24 , 466–497.
McPhaden, M. J., and R. A. Fine, 1988: A dynamical interpretation of the tritium maximum in the central equatorial Pacific. J. Phys. Oceanogr., 18 , 1454–1457.
McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415 , 603–608.
Redi, M., 1982: Oceanic isopycnal mixing by coordinate rotation. J. Phys. Oceanogr., 12 , 1154–1157.
Rothstein, L. M., R. H. Zhang, A. J. Busalacchi, and D. Chen, 1998: A numerical simulation of the mean water pathways in the subtropical and tropical Pacific. J. Phys. Oceanogr., 28 , 322–343.
Schneider, N., A. J. Miller, M. A. Alexander, and C. Deser, 1999: Subduction of decadal North Pacific temperature anomalies: Observations and dynamics. J. Phys. Oceanogr., 29 , 1056–1070.
Springer, S. R., M. J. McPhaden, and A. J. Busalacchi, 1990: Oceanic heat content variability in the tropical Pacific during the 1982–1983 El Niño. J. Geophys. Res., 95 , 22089–22101.
Stammer, D., C. Wunsch, I. Fukumori, and J. Marshall, 2002: State estimation in modern oceanographic research. Eos, Trans. Amer. Geophys. Union, 83 , 289–295.
Thacker, W. C., and R. B. Long, 1988: Fitting dynamics to data. J. Geophys. Res., 93 , 1227–1240.
Wijffels, S. E., 1993: Exchange between hemispheres and gyres: A direct approach to the mean circulation of the equatorial Pacific. Ph.D. thesis, Publ. WHOI-93-42, MIT–WHOI Joint Program, 294 pp.
Zebiak, S., 1989: Oceanic heat content variability and El Niño cycles. J. Phys. Oceanogr., 19 , 475–486.
Zhang, R. H., L. M. Rothstein, and A. J. Busalacchi, 1999: Interannual and decadal variability of the subsurface thermal structure in the Pacific Ocean: 1961–90. Climate Dyn., 15 , 703–717.
Zonal sections of time-mean potential density: (a) model at 10°N, (b) Levitus98 data at 10°N, (c) model at 10°S, and (d) Levitus98 data at 10°S. Contour unit is kg m−3
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Time-mean pseudo streamfunction of pycnocline transport (see text for definition). Contour unit is Sv
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Anomalies of pycnocline transports at (a) 10°N and (b) 10°S
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Same as Fig. 3 except that a 5-yr smoothing has been applied
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Zonal sections of annually averaged meridional velocity of the model at (a), (b), (c), (d) 10°N and (e), (f), (g), (h) 10°S from 1997 to 2000. Contour unit is cm s−1
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Same as Fig. 5 except that they are from a data assimilation product
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Zonal sections of meridional velocity at 10°N averaged over (a) the 1980s, (b) the 1990s, and (c) their difference. Contour unit is cm s−1
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Anomalies of east–west sea level differences across the western boundary and across the interior at (a) 10°N and (b) 10°S for model simulation, assimilation, and TOPEX/Poseidon data
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Comparison of (a) boundary, (b) interior, and (c) net pycnocline transports at 10°N with those at 5°N
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Interannual and decadal changes of pycnocline transport pseudo streamfunction: (a) from 1996 to 1997, (b) from 1997 to 1998, (c) from 1998 to 1999, and (d) from the 1980s to the 1990s. Contour increment is 1 Sv
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Interannual and decadal changes of barotropic transport streamfunction: (a) from 1996 to 1997, (b) from 1997 to 1998, (c) from 1998 to 1999, and (d) from the 1980s to the 1990s. Contour increment is 2 Sv
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
(a) Time-mean and (c) interannual-to-decadal variability of wind stress curl and (b), (d) their respective corresponding zonal averages. The unit is 10−7 N m−3
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Annually averaged wind stress curl computed from NCEP and ERS-2 scatterometer from 1997 to 1999. The solid line segments highlight the meridional shift of the band of maximum curl in time
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Anomalies of annually averaged (a), (c), (e) wind stress curl and (b), (d), (f) sea level from 1997 to 1999
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Same as Fig. 14 except that the wind stress curl and sea level are obtained from the measurements by ERS-2 scatterometer and TOPEX/Poseidon altimeter, respectively
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Comparison of anomalous boundary pycnocline transport at (a) 10°N and (b) 10°S with local Sverdrup transport computed from wind stress curl averaged over the western Pacific
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Perturbed transports due to an imposed wind stress curl anomaly centered near 10°N. Perturbed boundary and interior transports are comparable
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Comparison of the anomaly of net pycnocline transport at (a) 10°N and (b) 10°S with tropical zonal wind stress [divided by values of f at 10°N(S) to produce transport unit]
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Perturbed pycnocline transports at (a) 10°N and (b) 10°S due to an imposed globally uniform westerly wind. Perturbed interior transport is more dominant
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Comparison of local Sverdrup transport computed from wind stress curl near (a) 10°N and (b) 10°S with tropical zonal wind stress [divided by values of f at 10°N(S) to produce transport unit]
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Schematic illustration of the combined effect of off-equatorial wind stress curl and near-equatorial zonal wind stress in causing counteracting boundary and interior pycnocline transports and a larger magnitude of the latter (e.g., the Northern Hemisphere)
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Changes of (a) zonal wind stress averaged over 10°N–10°S and (b), (c) tropical wind stress curl between 1997 and 1998 and between the 1990s and the 1980s. Note the difference in color scale between (b) and (c)
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Anomalies of pycnocline transports at (a), (c) 10°N and (b), (d) 10°S with the ITF channels (a), (b) open and (c), (d) closed
Citation: Journal of Climate 16, 24; 10.1175/1520-0442(2003)016<4022:IVOTEI>2.0.CO;2
Interannual-to-decadal std dev of pycnocline transport (Sv) through the western boundary and interior