Abstract

A three-dimensional ocean general circulation model, forced by idealized zonal winds, is used to investigate the effect of an abrupt intensification of westerly winds on the subduction process. Four experiments are carried out: 1) a control experiment with standard wind stress forcing, 2) an intensified winds experiment with wind stress that is larger in the region of the westerlies than the control, 3) an increased surface cooling experiment, and 4) an experiment with both intensified wind stress and surface cooling. Experiments 2 through 4, which contain surface anomalous forcing, are run from the steady state obtained in experiment 1, the control experiment. The results obtained for each of these runs are compared to the control experiment. A subarctic tracer injection experiment is also carried out to verify the differences in the subduction process of each of these experiments.

In the wind stress intensified experiment, an examination of the subsurface temperature field shows that negative temperature anomalies occupy the western portion of the southern part of the subtropical gyre, whereas in the surface cooling experiment, negative temperature anomalies occupy the eastern portion of the basin. The source of these negative temperature anomalies is not local since the forcing in the southern part of the subtropical gyre does not change from the control. A close analysis of the evolution of a subarctic surface tracer field indicates that the intensification of the wind stress increases the tracer concentrations, whereas surface cooling decreases the temperature in the region that contains the maximum tracer concentration.

In the standard case, the mixed layer is deep (shallow) in the northern (southern) part of the subtropical gyre. Between these two regions a mixed layer front, where the mixed layer depth changes drastically from north to south, exists. A water column with low potential vorticity that originates in the mixed layer penetrates into a subsurface layer from the point where an outcrop line and the mixed layer front intersect. This point is called the penetration point.

Intensified westerly winds bring about a deeper thermocline and shoaling subsurface isopycnals. These shoaling subsurface isopycnals are not predicted in classical models such as that of Luyten et al. The model experiment with intensified westerlies demonstrates that the penetration point shifts to the west. As a result, low potential vorticity water penetrates southwestward from the shifted penetration point and takes a more westward path. Therefore, the negative temperature anomalies appear in the southwestern part of the subtropical gyre. This study shows that the westward shift of the path of low potential vorticity water could cause the shoaling of subsurface isopycnal surfaces.

The intensification of the westerlies increases Ekman pumping and cools the ocean surface by enhancing sensible and latent heat flux. In the surface cooling experiment, the position of the outcrop lines moves southward significantly. This southward shift makes the subducted water colder and distributes it throughout the ventilated region of the southern part of the subtropical gyre.

The combined effect of wind intensification and surface cooling is approximately a linear combination of both experiments.

1. Introduction

It has often been stated that the ocean has enormous thermal inertia and hence has the unique ability to remove heat from the atmosphere in one location and return heat to the atmosphere in a distant location, on varying timescales. Evidence of this type of atmospheric and oceanic interaction has been pointed out by many authors. They noted that a sudden change in atmospheric circulation over the North Pacific during the mid 1970s caused a change in the variability of sea surface temperature (SST) and also of the interior ocean. The Aleutian Low was intensified from 1976 to 1988 (Nitta and Yamada 1989; Trenberth 1990; Yasuda and Hanawa 1997), as seen in Fig. 1. Figure 2 shows that the SST has also undergone a change during this same time period, which suggests that the variability of the SST may have been affected by the atmospheric changes (Tanimoto et al. 1993; Tanimoto et al. 1997; Yasuda and Hanawa 1997).

Fig. 1.

The wintertime wind stress fields averaged over each decade: (a) 1966 to 1975 and (b) 1976 to 1985, and (c) the difference between these two decades is shown (after Yasuda and Hanawa 1997).

Fig. 1.

The wintertime wind stress fields averaged over each decade: (a) 1966 to 1975 and (b) 1976 to 1985, and (c) the difference between these two decades is shown (after Yasuda and Hanawa 1997).

Fig. 2.

The wintertime sea surface temperature fields averaged over each decade: (a) 1966 to 1975, (b) 1976 to 1985, and (c) is the difference between these two decades (T. Yasuda and K. Hanawa 1996, personal communication).

Fig. 2.

The wintertime sea surface temperature fields averaged over each decade: (a) 1966 to 1975, (b) 1976 to 1985, and (c) is the difference between these two decades (T. Yasuda and K. Hanawa 1996, personal communication).

While it is understood that the variability of atmospheric circulation has a great influence on the SST field, the influence of changes in atmospheric circulation on subsurface layers of the ocean needs further study. Subduction, the process that carries surface water to the ocean’s interior, is likely responsible for conveying the variations of the thermal structure and the wind stress to subsurface layers.

Luyten et al. (1983), henceforth referred to as LPS, proposed the “ventilated thermocline” theory to reproduce the vertical structure of the midlatitude ocean. It has several layers with a specified surface density and wind stress curl in the steady nondissipative regime. The lower layers, where there is no direct wind forcing, conserve their potential vorticity. The solution has critical trajectories that separate the ventilated region, generated by Ekman pumping, from the eastern unventilated region, called the shadow zone, and the western unventilated region, called the pool region. In order to deal with continuous stratification, numerical models are required. Cox and Bryan (1984) studied the subduction process with a general circulation model (GCM) that applied a steady-state wind stress. They examined the balance of each term in the density equation to find that the overturning term is the most important one for the subduction process, and that the ventilated region is fed by water that is cooled in the western boundary outflow region.

Trajectory analysis (Cox and Bryan 1984; Inui and Hanawa 1997) and modeled tracer injection experiments (Inui and Hanawa 1997) suggest that it takes 10 to 15 years for the water particles to circulate throughout the ventilated region of the ocean. This timescale suggests that subduction affects the structure of the subsurface ocean on a decadal timescale. Temperature anomalies at the sea surface, which are affected by anomalous atmospheric conditions, are expected to be distributed in the subsurface layer of the lower latitudes at some time after the anomalies first appear. Recent observational studies of the North Pacific suggest that atmospheric variations affect the structure of the subsurface layers. (Yasuda and Hanawa 1997; Deser et al. 1996).

Watanabe and Mizuno (1994) speculated that the subduction process would affect the subsurface structure of the North Pacific. They selected a specific location at the sea surface and showed the correlation between the surface temperature and the temperature at a 400-m depth for a 2-yr time lag. The region of positive correlation extends southwestward in the subsurface layers and resembles the distribution of the shallow salinity minimum (SSM) (Reid 1973; Tsuchiya 1982; Talley 1985). To address water mass variability, North Pacific central mode water (NPCMW) and North Pacific subtropical mode water (NPSTMW) (Masuzawa 1969; Suga et al. 1997) are analyzed. NPCMW is distributed mainly between 30° and 40°N, and between 170°E and 150°W, which has a thermostad of 9°–13°C (Nakamura 1996; Suga et al. 1997). Yasuda and Hanawa (1997) suggested that the decadal change of two mode waters is related to changes of the westerlies. They also demonstrated that the formation of NPCMW during the decade of 1976–85 was enhanced and that the water was cooler than was in the previous decade of 1966–75. The decadal variability of mode waters can be attributed to the intensified westerlies after the mid 1970s. Distribution of the NPCMW or the SSM might be controlled significantly by the subduction process. Along with the results of the GCM experiments, these observational studies suggest that the change of the westerlies will affect the subsurface layers in the lower latitudes through the subduction process on a decadal timescale.

Some studies have investigated the response of the ocean to an abrupt change of wind stress curl. Anderson and Gill (1975) examined the linear adjustment of a midlatitude ocean on both the f plane and the β plane, showing that the long baroclinic Rossby waves play a key role in adjustment of the ocean in the latter case. Anderson and Killworth (1979) explored the propagation of nonlinear long Rossby waves. These studies, however, did not explain the effect of advection. Using an LPS model with two layers, Liu (1993a) investigated the thermocline response to a sudden change of Ekman pumping velocity and Liu (1993b) investigated the thermocline response to the varying Ekman pumping velocity with annual to decadal frequencies. He suggested that the difference in the dynamic response between the ventilated and shadow regions is that advection is dominant in the ventilated region while the propagation of Rossby waves is dominant in the shadow zone. Using a four-level model, Nakamura (1998) examined the response of the thermocline to a sudden change of the wind stress curl based on both advection and wave propagation. He showed that the temperature anomaly that is advected along its characteristics causes a horizontally shifted thermocline, resulting in a significantly weakened (weakly intensified) circulation in the northwestern (southeastern) subtropical gyre.

A sudden change of wind stress curl causes a sudden change in SST because the winds cool the ocean surface by removing heat from it via latent and sensible heat flux release. Even for a steady-state circulation, purely wind driven circulation is not enough to explain the subduction process, if we consider the mass flux between layers due to the cooling or heating effect. In an attempt to include thermal effects, Pedlosky et al. (1984) added a mixed layer to the LPS model, and Pedlosky (1986) used a 2-layer analytical model that allowed water mass exchange. Pedlosky (1986) assumed that the buoyancy flux, as a function of latitude alone, is proportional to the Ekman pumping. Furthermore, he found that surface heating shifts the ventilated zone westward compared to the purely wind driven case. Luyten and Stommel (1986) also investigated the effect of buoyancy flux using a 2-layer nonlinear model that allowed for exchange of water between the layers. There are also studies, using the continuously stratified model of Huang (1988), that include thermal effects using a mixed layer ocean model (Pedlosky and Robbin 1991;Williams 1991). These mixed layer studies have found that the amount of subducted water that penetrates into the thermocline through the bottom of the mixed layer tends to increase because the base of the mixed layer is more inclined compared to the case with purely wind driven circulation.

While these studies show the steady-state result, we need to also consider the response of the ocean to temporally varying forces. Using a 2-layer LPS model that includes the mixed layer, Liu and Pedlosky (1994) examined the thermocline temperature variability by moving the outcrop line. They found that varying the position of the outcrop line over decadal timescales caused a strong variation in the thermocline temperature.

The analytical models described above have shown the dynamic response of the ocean to a sudden change of the wind stress curl for an entire basin. The response of the ocean to a change of localized winds remains uncertain, although its effects are important for the thermal structure of the subsurface layers of the ocean. In addition, previous model studies had insufficient vertical resolution to adequately model the movement of the isopycnal surfaces or to include thermal effects.

This study utilizes an ocean general circulation model to evaluate the effect of the intensification of the westerlies on the subduction process. Our model is forced by idealized zonal winds over the entire basin. Intensification of the westerlies causes both intensified wind stress and surface cooling. To separate the effect of intensified wind stress and surface cooling on the subduction process four experiments are performed. The first is the standard case, which is driven by an idealized wind stress and an equilibrium temperature. The second intensifies the wind stress. The third increases surface cooling. And the fourth combines both intensified wind stress and surface cooling. Our purpose is to describe the differences among the four cases and to investigate the variations in the isopycnal surfaces and in water mass formation. While the perturbations are localized in the region of the westerlies, we will also investigate changes in the thermal structure of the lower latitudes where the forcing does not change locally. This study portrays the detailed mechanism of changes both from a dynamical and a descriptive perspective (water mass formation and distribution).

The remainder of this chapter is organized as follows:section 2 describes the ocean model, section 3 reveals the effect of the temperature field and the passive tracer field on the lower latitudes in each case, section 4 addresses the mechanisms responsible for the response of the ocean in the wind intensified case, section 5 addresses the mechanisms responsible for the response of the ocean in the cooling intensified case, and section 6 provides a summary and discussion.

2. Model description

A detailed description of the model used in this study is given in Inui and Hanawa (1997). We use the Geophysical Fluid Dynamics Laboratory Modular Ocean Model (MOM version 1.1). It is 40° wide extending from the equator to 60°N. The ocean is 3000 m deep with 24 vertical levels (Fig. 3a).

Fig. 3.

(a) The arrangement of grid points in the vertical direction. Boundary conditions used in this investigation, a meridional profile of (b) wind stress for the standard (SD) and cooling intensified (CI) cases, (c) wind stress for the wind intensified (WI) and combined (CB) cases, and (d) the specified temperature Ta for the cooling intensified (CI: denoted by solid line) and other cases (denoted by dashed line).

Fig. 3.

(a) The arrangement of grid points in the vertical direction. Boundary conditions used in this investigation, a meridional profile of (b) wind stress for the standard (SD) and cooling intensified (CI) cases, (c) wind stress for the wind intensified (WI) and combined (CB) cases, and (d) the specified temperature Ta for the cooling intensified (CI: denoted by solid line) and other cases (denoted by dashed line).

a. Boundary conditions

The model is forced by the zonal wind stress τ and by the surface heat flux Q. They are imposed at the ocean surface (z = 0) and are given by the boundary conditions

 
ρ0AMR (uz, υz) = (τ, 0),w = 0,ATV Tz = Q.
(1)

We used the following two zonal wind stress fields. In the standard case, τ is defined as

 
formula

where L is 40° and y is latitudinal extent. In the intensified case, τ is defined as

 
formula

where τSD is the wind stress field used in the standard case. The wind stress field is illustrated in Fig. 3b,c.

The heat flux is of the Haney (1971) type with the heat flux given by

 
Q = Q0(TaTSST),
(4)

where Ta is a specified atmospheric temperature, TSST is the model SST, and Q0 is a constant exchange rate that is set to be constant at 90 cm day−1, which assumes that ρCp = 1.0 cal °C−1 cm−3 (Cp is the heat capacity). This constant restores the surface condition to the defined value with a restoring time of approximately 44.4 days for a water depth of 4 m.

We used two atmospheric temperature fields. The first is a linear function of latitude (TaSD). The second is also linear except in the region of the westerlies between 30° and 50°N. In this region, there is a negative atmospheric temperature anomaly, which can be described as a sine curve that has minimum a value of −0.5°C at 40°N (TaCI). These two atmospheric temperature fields are illustrated in Fig. 3c.

b. Initial condition and integration

The model ocean is initially at rest and the temperature field is set to its zonally averaged climatological mean field (Levitus 1982). The equation of state ρ = ρ(T, S, z) is a third-order polynominal approximation to the UNESCO (1981) equation in the model. The salinity field remains at a constant value of 34.9 psu everywhere. After integrating 40 years with a 1-h time step and standard forcing (τSD and TaSD), the ocean model reaches a steady state except in its abyssal part.

Intensification of the westerlies causes both momentum intensification and surface cooling. In an attempt to separate the effects of momentum intensification and surface cooling, four experiments are carried out once a steady state is reached. In the standard case (SD), τSD and TaSD are used. In the wind intensified case (WI), τWI and TaSD are used, and in the cooling intensified case (CI), τSD and TaCI are used. Finally, in the combined case (CB), τWI and TaCI are used. The estimated timescale of the subduction process in the model, the time taken for surface water to feed the thermocline of the subtropical gyre, is about 15 years. To understand the physics in the upper ocean, the model is integrated for an additional 15 years for each case beginning at the termination of the initial 40-yr integration with the appropriate forcings.

3. Results

Features of our model ocean in the SD case are nearly identical to those of Cox and Bryan (1984). A detailed description of the results for the SD case is given in Inui and Hanawa (1997). A comparison of the temperature and tracer fields of the four experiments follows.

a. Temperature field

Figure 4 shows the temperature field for the SD case and the difference in temperature between the SD and the WI, CI, and CB cases at the first model level (z = −2 m). In the WI case, the western boundary current is intensified, which causes more warm water to be advected to the high latitudes compared to the SD case. Negative SST anomalies occupy a region centered on the boundary between subarctic and subtropical gyres, resembling the pattern of decadal changes observed in the North Pacific (Fig. 2). Enhanced Ekman transport brings colder surface water from the north, which creates negative SST anomalies. In the CI case, a negative SST anomaly extends from east to west along the zonal band from 30° to 50°N, which is the same region that the surface cooling given by boundary condition as TaCI was applied. The resultant SST anomaly field in the CI case has the same magnitude as that in the WI case, making it easy to compare the difference of the effect that the surface anomaly field has on the interior ocean.

Fig. 4.

Sea surface temperature fields (z = −2.0 m) for the (a) SD case, and in (b)–(d), the difference between each case and the SD case is shown by dashed lines for (b) WI, (c) CI, and (d) CB cases after 15 years since the start of each intensification. The contour intervals are 1°C for SST and 0.1°C for the difference. For the difference, regions lower than −0.1°C are shaded.

Fig. 4.

Sea surface temperature fields (z = −2.0 m) for the (a) SD case, and in (b)–(d), the difference between each case and the SD case is shown by dashed lines for (b) WI, (c) CI, and (d) CB cases after 15 years since the start of each intensification. The contour intervals are 1°C for SST and 0.1°C for the difference. For the difference, regions lower than −0.1°C are shaded.

Figure 5 shows the temperature field for the SD case and the difference between the SD case and each of the other three cases at the ninth level (z = −148.44 m). In the CI case, the region of negative temperature anomaly extends southwestward from the eastern part of the basin centered at 40°N. This figure implies penetration of colder water from the northeastern part of the subtropical gyre. Conversely, in the WI case, the region of the negative temperature anomaly lies in the southern part of the subtropical gyre. This negative temperature anomaly is larger than the SST anomaly. Since there is no surface cooling in this region, subduction of colder surface water in the region of the westerly winds must be responsible for the decrease in the temperature field.

Fig. 5.

Same as Fig. 4 for the temperature fields at the 9th level (z = −148.4 m).

Fig. 5.

Same as Fig. 4 for the temperature fields at the 9th level (z = −148.4 m).

Figure 6 shows the zonal temperature field at 27°N for each case together with their difference from the SD case. At this latitude, neither wind nor cooling is intensified locally. In the CI case, the core of colder water exists mainly in the eastern part of the basin, whereas in the WI case the colder water exists mainly in the western portion of the basin. In contrast to this colder region in the WI case, there is warming in other parts of the basin.

Fig. 6.

Zonal sections of temperature at 27°N for the (a) SD, (b) WI, (c) CI, and (d) CB cases after 15 years since the start of each intensification. In (b)–(d), the difference between each case and the SD case is also shown. The contour intervals are 1°C for SST and 0.1°C for the difference. For the difference, regions lower than −0.1°C are shaded and higher than +0.1 are stippled.

Fig. 6.

Zonal sections of temperature at 27°N for the (a) SD, (b) WI, (c) CI, and (d) CB cases after 15 years since the start of each intensification. In (b)–(d), the difference between each case and the SD case is also shown. The contour intervals are 1°C for SST and 0.1°C for the difference. For the difference, regions lower than −0.1°C are shaded and higher than +0.1 are stippled.

In the CB case, as seen in Figs. 4–6, the temperature field is approximately equal to a linear combination of both the WI and CI cases.

To verify that the surface cooling caused by an enhancement of Ekman transport in the WI case is not responsible for the cooling in the subsurface layers, another experiment is carried out with the standard wind stress and with the equilibrium SST obtained in the WI case (Fig. 7). This experiment indicates that the surface cooling created by the Ekman transport from the north is not the major cause of the subsurface cooling in the WI case (Fig. 7d). The cause of the cooling will be explored later.

Fig. 7.

(a) Sea surface temperature anomaly fields (z = −2.0 m) for the WI case. Temperature anomaly fields at the 9th level (z = −148.4 m) for the (b) WI, (c) CI, and (d) additional experiment case after 15 years since the start of each intensification. The contour intervals are the same as in Fig. 4.

Fig. 7.

(a) Sea surface temperature anomaly fields (z = −2.0 m) for the WI case. Temperature anomaly fields at the 9th level (z = −148.4 m) for the (b) WI, (c) CI, and (d) additional experiment case after 15 years since the start of each intensification. The contour intervals are the same as in Fig. 4.

The more simple LPS model indicates that the intensification of the westerly winds makes the thermocline deeper and its temperature warmer. However, our findings show that the upper thermocline temperature decreases in the WI case. Furthermore, compared to the CI case, the negative anomaly in the subsurface layer of the WI case occurs in a different location, which suggests that the origin of the negative anomaly is different. This process is investigated in detail in the next section.

b. Subarctic tracer fields

To elucidate latitudinal exchange of water, a passive tracer is added to the subarctic surface (whose circulation pattern is defined by the Sverdrup relation). Evolution of the tracer distribution demonstrates the effect of variations on water mass distribution and character. Although this model does not include the salinity change, the tracer experiment is still useful to visualize the subarctic water in the model because the subarctic water in the real ocean (e.g., SSM) penetrates into the subtropical gyre.

The equation for tracer concentration is given in the appendix. The calculation starts at the onset of each of the model experiments, at year 40, and has been done simultaneously as each experiment is run until it reaches a quasi-steady state with another 15 years. Throughout this integration, the tracer is restored to a value of 100 at the subarctic surface and to a value of zero at the surface of the other regions. The restoring timescale is the same as one used for the temperature and salinity so that the tracer concentration will represent the character of the water mass at the surface of the subarctic gyre. The tracer distribution after 15 years of integration for each run is compared, which is a quasi-steady state.

Figure 8 shows the subarctic surface tracer distribution on the isopycnal surface, σθ = 26.1, which corresponds to the main thermocline in the subtropical gyre for the SD and WI cases. Unlike the SD and CI cases, the tracer in the WI case reaches the western boundary. This suggests that the intensification of the westerly winds causes the tracer at the surface of the subarctic gyre to be distributed to the southwest of the subtropical gyre.

Fig. 8.

Subarctic surface tracer distribution on the isopycnal surface σθ = 26.1 for the (a) SD and (b) WI cases after 15 years since the start of each intensification. Shading interval is 5% for concentration. The outcrop line is indicated by a dashed line.

Fig. 8.

Subarctic surface tracer distribution on the isopycnal surface σθ = 26.1 for the (a) SD and (b) WI cases after 15 years since the start of each intensification. Shading interval is 5% for concentration. The outcrop line is indicated by a dashed line.

Figure 9 shows the relationship between tracer concentration and the corresponding temperature for a vertical section along 36°N, where the tracer begins to be subducted into the subtropical gyre. Each dot represents a value of tracer concentration and its corresponding temperature for each water particle. In the SD case (Fig. 9a), the water with the highest tracer concentration has temperature values ranging from 13° to 14°C. The maximum tracer concentration is around 18.5%. In the WI case (Fig. 9b), the highest tracer concentration occurs with temperatures of l3° to 14°C, which has a same range as the SD case. On the other hand, the maximum tracer concentration has increased to 23%. In the CI case (Fig. 9c), the highest tracer concentration occurs with temperatures of 12.5° to 13.5°C, which means that cooling intensification shifts the core temperature to lower values. The maximum tracer concentration in the CI case is similar to that in the SD case. In the CB case (Fig. 9d), the temperature range of water with highest concentration is from 12.5° to 13.5°C, and the maximum tracer concentration is around 20%. The range of temperature of the water with the highest tracer concentration shifts toward lower values, due to the intensification of surface cooling. On the other hand, the maximum tracer concentration is higher in the WI case than the SD case. Panel d in Fig. 9 indicates that the effects of both intensified winds and surface cooling affects the temperature fields of the combined case in a linear fashion.

Fig. 9.

The concentration of the subarctic tracer versus temperature at 36°N for the (a) SD, (b) WI, (c) CI, and (d) CB cases. The highest tracer concentrations occur with temperature ranges from 13.0° to 14.0°C in the SD case, (b) 13.0° to 14.0°C in the WI case, 12.5° to 13.5°C in the CI case, and 12.5° to 13.5°C for the CB case.

Fig. 9.

The concentration of the subarctic tracer versus temperature at 36°N for the (a) SD, (b) WI, (c) CI, and (d) CB cases. The highest tracer concentrations occur with temperature ranges from 13.0° to 14.0°C in the SD case, (b) 13.0° to 14.0°C in the WI case, 12.5° to 13.5°C in the CI case, and 12.5° to 13.5°C for the CB case.

The results shown in Figs. 8 and 9 are summarized as follows: through the circulation from higher to lower latitudes 1) wind intensification increases the tracer concentration, 2) cooling intensification leads to the shift of the core temperature in the region of the highest tracer concentration, and 3) the combination of the wind and the cooling intensification causes an increase in the tracer concentration and a shift of the core temperature. These results are useful for isolating both effects in an analysis of the water mass characteristics in the real ocean.

4. Mechanisms of change in the wind intensified case

a. The change of the isopycnal depth and its relation to the subsurface temperature field

The previous section illustrates that subsurface layers in the southern part of the subtropical gyre have negative temperature anomalies in response to intensified westerly winds. Based on the LPS model, the intensified winds cause the isopycnal surfaces to move downward, which results in an increased thermocline temperature. Contrary to the LPS model, the GCM experiment with intensified westerly winds, shown in the previous section, yields a decrease in the upper thermocline temperature. The reason for the different results in the WI case can be clarified using layer concepts. The isopycnal depth and layer thickness will be examined to deduce the origin of the anomalies seen in the case with the intensification of the westerly winds. Therefore, two isopycnal surfaces, corresponding to the top and bottom of the thermocline, were chosen to describe the temperature variability. Both surfaces outcrop in the region of the westerly winds. In this study, these two surfaces are called the subsurface and thermocline, respectively.

Figures 10 and 11 show the temporal change of the depth of the isopycnal defined by σθ = 26.4 (the thermocline) and σθ = 25.8 (the subsurface) for the WI case. From Fig. 10, it is evident that the intensified westerly winds result in a deepening of the thermocline. A deeper thermocline in the region of the westerlies, where the winds have been intensified, can also be predicted by classical LPS theory. Other regions also seem to become deeper, and we give a speculative explanation for this deepening in the section 6. Figure 11 shows the time series of the depth of the isopycnal given by σθ = 25.8. This isopycnal surface deepens in the northwestern part of the gyre, due to the enhanced western boundary current. In the region surrounding the northwestern corner of the subtropical gyre; however, the isopycnal depth gradually gets shallower with time at 18° longitude and 32°N, and slopes toward the southwest. Outside the region of the western boundary current the isopycnal becomes more shallow in contrast to the deeper thermocline predicted by classical LPS model.

Fig. 10.

Depth of the isopycnal surface (σθ = 26.4) for the WI case and its difference from the SD case after (a) 1 yr, (b) 2.5 yr, (c) 5 yr, and (d) 10 yr after the start of each intensifications. The contour interval is 20 m for the isopycnal depth. The difference is shown in color.

Fig. 10.

Depth of the isopycnal surface (σθ = 26.4) for the WI case and its difference from the SD case after (a) 1 yr, (b) 2.5 yr, (c) 5 yr, and (d) 10 yr after the start of each intensifications. The contour interval is 20 m for the isopycnal depth. The difference is shown in color.

Fig. 11.

Same as Fig. 10 for the depth of the isopycnal surface σθ = 25.8.

Fig. 11.

Same as Fig. 10 for the depth of the isopycnal surface σθ = 25.8.

Kubokawa and Inui (1999) identified the water path by plotting the minimum potential vorticity defined at each latitude on an isopycnal surface. They found that the path of the water with low potential vorticity on each isopycnal surface is an important factor for determining its baroclinic structure. It has also been shown that when the paths of water with low potential vorticity that are of different densities meet the upper waters must rise to conserve their potential vorticity. This is a plausible explanation for the subsurface shoaling. To verify the cause of the subsurface shoaling, we examine the path and origin of the water with low potential vorticity as in Kubokawa and Inui (1999).

Figure 12 shows the distribution of the mixed layer depth for the SD and WI cases 10 years after the onset of the the intensification of the westerly winds. The mixed layer depth is defined as the depth corresponding to a change in density that is 0.1σθ larger than of the surface value. In all cases, the mixed layer depth is deep in the northern part of the subtropical gyre and shallow in the southern part. In this study, the mixed layer front (MLF), a narrow zone where the mixed layer depth changes abruptly, extends from the southwest to the northeast from 20° to 30°N in the southern part of the subtropical gyre.

Fig. 12.

Mixed layer depth distribution (bold lines) and the outcrop line of the isopycnal σθ = 25.6 (thin line) in the (a) SD and (b) WI cases. The contour interval is 20 m for the mixed layer depth. The depths between 180 and 500 m are shaded and the depths below 500 m are not shown. The diamond denotes the penetration point where the mixed layer front and the outcrop line intersect.

Fig. 12.

Mixed layer depth distribution (bold lines) and the outcrop line of the isopycnal σθ = 25.6 (thin line) in the (a) SD and (b) WI cases. The contour interval is 20 m for the mixed layer depth. The depths between 180 and 500 m are shaded and the depths below 500 m are not shown. The diamond denotes the penetration point where the mixed layer front and the outcrop line intersect.

Figure 13 shows the potential vorticity (defined as q = −z) distribution on the isopycnal σθ = 25.6 for the SD and WI cases. The minimum potential vorticity defined at each latitude is also shown. First, we describe the results for the SD case. Along the outcrop line, the mixed layer is deepest and the potential vorticity is the lowest near the MLF (Fig. 12a). This path corresponds to a tongue of water with low potential vorticity that originated in this deep mixed layer region. It is found that the path of the water with low potential vorticity, on each isopycnal, starts from the point where the outcrop line and the MLF intersect, hereafter referred to as the penetration point. The penetration point is shown by a diamond in Fig. 12.

Fig. 13.

Potential vorticity distribution on the isopycnal surface of σθ = 25.6 for the (a) SD and (b) WI cases. The contour interval is 0.5 × 10−10 cm−1 s−1. The region with low potential vorticity, less than 3.0 × 10−10 cm−1 s−1 is shaded. Crosses denote the position of minimum potential vorticity at each latitude.

Fig. 13.

Potential vorticity distribution on the isopycnal surface of σθ = 25.6 for the (a) SD and (b) WI cases. The contour interval is 0.5 × 10−10 cm−1 s−1. The region with low potential vorticity, less than 3.0 × 10−10 cm−1 s−1 is shaded. Crosses denote the position of minimum potential vorticity at each latitude.

When the westerly winds are intensified (Fig. 13), a change in the path of the water with low potential vorticity and the position of the penetration point results. Compared to the SD case, the path of the water with low potential vorticity shifts to the west in the WI case. An examination of the position of the penetration point (Fig. 12) suggests that the penetration point also lies farther to the west. The westward shift of the penetration point causes the subduction of water with lower potential vorticity to be displaced to the west compared to the SD case. Therefore, the temperature anomalies dominate the southwestern part of the subtropical gyre.

According to Kubokawa and Inui (1999), the subsurface shoaling in the WI case could be caused by the intensification of the westerly winds, which makes the paths of the water with low potential vorticity, on each isopycnal surface, to be closer to one other. It is also possible that the subsurface shoaling may occur if the intensification of the westerly winds causes the layer thickness to become greater even though the position of the path of the water with low potential vorticity remains the same.

To examine whether the closeness of the paths of the water with low potential vorticity, on isopycnal surfaces, is responsible for causing the subsurface isopycnals to rise in the WI case, we will illustrate how the paths of different isopycnals intersect. Figure 14 shows the outcrop line and the path of a water column with minimum potential vorticity at each latitude on each isopycnal surface. Figure 14a shows how four paths with low potential vorticity water on each isopycnal surface circulate and extend to the southwestern part of the subtropical gyre in the SD case. Examination of Fig. 14b reveals that intensified westerly winds force these paths to move westward and alter the local potential vorticity without altering the horizontal potential vorticity gradient. This means that the change of the potential vorticity distribution is mainly caused by a shift in the position of the paths. This is the manner in which the paths of water with low potential vorticity respond.

Fig. 14.

The outcrop line and the position of minimum potential vorticity at each latitude on each isopycnal surfaces σθ = 25.4 (dot), 25.6 (cross), 25.8 (asterisk), 26.1 (circle) for the (a) SD and (b) WI cases. The region with low potential vorticity, less than 1 × 10−10 cm−1 s−1, is shaded.

Fig. 14.

The outcrop line and the position of minimum potential vorticity at each latitude on each isopycnal surfaces σθ = 25.4 (dot), 25.6 (cross), 25.8 (asterisk), 26.1 (circle) for the (a) SD and (b) WI cases. The region with low potential vorticity, less than 1 × 10−10 cm−1 s−1, is shaded.

An examination of the effect of increased layer thickness follows. Local observation of the isotherms reveals that the water with low potential vorticity penetrates to the south west, pushing up the 15°C through 17°C isotherms and pushing down the thermocline below the 13°C isotherm (Figs. 15a,c). An examination of the layer thickness reveals that there is little change compared to the change caused by the westward shift of the position of the water column, which has the greatest thickness. For example the thickest layer, located between 15° and 16°C at 30°N in the WI case (Fig. 15a), has little difference compared to the thickness between the same isotherms in the SD case. This means that in the WI case the increased layer thickness is not the primary cause for altering the temperature field. From these analyses, it is found that the westward shift of the penetration point is more effective on altering the subsurface temperature than two explanations based on Kubokawa and Inui (1999).

Fig. 15.

Zonal section of the temperature field at 30°N in the (a) WI and (b) CI cases, and at 27°N in the (c) WI and (d) CI cases. Solid and dashed line indicates the SD and the WI (CI) cases, respectively. Panels (a) and (c) illustrate deeper thermocline depth and shallower subsurface isopycnals, while panels (b) and (d) illustrate little change in thermocline depth and shallower subsurface isopycnals.

Fig. 15.

Zonal section of the temperature field at 30°N in the (a) WI and (b) CI cases, and at 27°N in the (c) WI and (d) CI cases. Solid and dashed line indicates the SD and the WI (CI) cases, respectively. Panels (a) and (c) illustrate deeper thermocline depth and shallower subsurface isopycnals, while panels (b) and (d) illustrate little change in thermocline depth and shallower subsurface isopycnals.

In an attempt to describe why the penetration point shifts westward, we will examine the difference of mixed layer depth between the SD and the WI cases in detail (Fig. 12). In the northwestern part of the subtropical gyre and north of the mixed layer front, the mixed layer becomes deeper. There is a northward shift of the mixed layer front in the WI case compared to the mixed layer front in the SD case. These two indicate that the mixed layer front has become steeper. A greater thickness results if the water column penetrates into the subsurface layers. The steeper mixed layer front implies that the subduction of water from the mixed layer into the interior region has been enhanced, which is consistent with the result obtained in Pedlosky and Robbins (1991). Moreover, a northward shift of the mixed layer front seems to accompany the westward shift of the penetration point.

The process discussed above can be summarized as follows. A deepening of the mixed layer and the northward shift of the mixed layer front result in a steeper inclination of the mixed layer front, although the mechanism for the northward shift of the MLF remain uncertain. When the mixed layer front becomes steeper, the water column must have lower potential vorticity in the region of the westerly winds. This water column moves southward and encounters an outcrop line at the penetration point that shifts to the west in the WI case compared to the SD case. Consequently, the path where the subsurface layer is thickest moves west rather than increasing the layer thickness. From the penetration point, a water column with low potential vorticity penetrates into the subsurface layer, taking a path that is shifted to the west. In Fig. 15a, the layers between 14° and 15°C, and between 15° and 16°C show this behavior. As a result, subsurface shoaling leads to lower temperatures in the subsurface. Finally, the region with negative temperature anomalies dominates the southwestern part of the subtropical gyre, while the surface density structure does not change significantly (Fig. 6b).

b. Enhanced transport of the subarctic tracer into the subtropical gyre

The previous section illustrates that the intensification of westerly winds causes the subarctic tracer to be distributed over a larger area with larger concentrations in the subtropical gyre (Figs. 8 and 9b).

To explore this feature we examine the vertical section of the tracer at 36°N where the tracer begins to subduct. Figure 16 illustrates the effect of the thermal structure on the tracer distribution. The west to east inclination of the isopycnal lines has changed. Compared to the SD case, the isopycnal lines have tilted downward in the western part of the basin and upward in the eastern part of the basin. This is consistent with the deeper mixed layer in the northwestern part of the subtropical gyre (Fig. 12). Along the isopycnal lines with σθ = 26.4 and σθ = 26.1, it is evident that convection penetrates deeper into the water column in the western part of the subtropical gyre in the WI case. This deeper penetration allows the tracer to penetrate deeper and causes the concentration of the tracer to be larger at depth in the WI case. The intensified winds cause convection to penetrate deeper than in the SD case and hence increase the tracer concentrations.

Fig. 16.

Zonal section of the subarctic surface tracer at 36°N for the (a) SD, (b) WI cases, and the (c) difference (WI − SD). The interval for shading is 5% for the tracer concentrations [(a) and (b)] and 2% for the difference (c). Potential density fields are also shown and the contour interval is 0.3σθ. The density field in (c) comes from the wind intensified case.

Fig. 16.

Zonal section of the subarctic surface tracer at 36°N for the (a) SD, (b) WI cases, and the (c) difference (WI − SD). The interval for shading is 5% for the tracer concentrations [(a) and (b)] and 2% for the difference (c). Potential density fields are also shown and the contour interval is 0.3σθ. The density field in (c) comes from the wind intensified case.

To examine other effects that may increase the transport from the subarctic to subtropical gyres, the change in the density stratification near the sea surface is examined in the subarctic gyre. Figure 17 shows the change in the convective region due to the intensification of the westerly winds. An examination of the difference in the potential vorticity (q = −z) distribution at the surface (averaged over 0–100 m) between the WI and SD cases shows that the region with higher potential vorticity is centered around the southwestern part of the subarctic gyre. To find which region is most responsible for this enhanced transport, a tracer is injected into the six regions shown in Fig. 18a. Examining the tracer distributions, we found that the distribution of tracer injected in region 3 in the WI case strikingly differs from that in the SD case (Figs. 18b,c). In the WI case, region 3 is characterized by higher potential vorticity. These results suggest that a change in the position of the subarctic convective region prevents the tracer from penetrating into deeper layers in the subarctic. As a result, more of the subarctic tracer flows into and is advected throughout the subtropical gyre.

Fig. 17.

Distribution of potential vorticity averaged over 0–100 m for (a) SD and (b) WI cases, and the (c) difference (WI − SD). The contour interval is 2.0 × 10−10 cm−1 s−1 for potential vorticity and is 1.0 × 10−10 cm−1 s−1 for the difference. Regions lower than 2.0 × 10−10 cm−1 s−1 in (a) and (b), and 1.0 × 10−10 cm−1 s−1 in (c) are shaded.

Fig. 17.

Distribution of potential vorticity averaged over 0–100 m for (a) SD and (b) WI cases, and the (c) difference (WI − SD). The contour interval is 2.0 × 10−10 cm−1 s−1 for potential vorticity and is 1.0 × 10−10 cm−1 s−1 for the difference. Regions lower than 2.0 × 10−10 cm−1 s−1 in (a) and (b), and 1.0 × 10−10 cm−1 s−1 in (c) are shaded.

Fig. 18.

(a) Schematic illustration of six boxes for the tracer experiment. Each box is 40° to 50°N latitude by 5° longitude. The horizontal distribution of tracer 3 at the ninth level (z = −148.4 m) for the (b) SD and (c) WI cases are shown. The interval for shading is 0.5% for the tracer concentrations.

Fig. 18.

(a) Schematic illustration of six boxes for the tracer experiment. Each box is 40° to 50°N latitude by 5° longitude. The horizontal distribution of tracer 3 at the ninth level (z = −148.4 m) for the (b) SD and (c) WI cases are shown. The interval for shading is 0.5% for the tracer concentrations.

The tracer reaches the western boundary region of the subtropical gyre in the wind intensified case due to the enhanced Ekman pumping velocity. Also the change of the convective region, both in the subtropical and subarctic gyres, enhances this process.

5. Mechanism of change in the cooling intensified case

The purpose of this section is to show the effect of surface cooling on thermocline structure. We explain the mechanisms that causes the negative temperature anomaly field (Fig. 5c), which differs from the WI case, and we show the relationship between the maximum tracer concentrations and the negative temperature anomalies (Fig. 9). Based on the LPS model, intensified surface cooling causes the isopycnal surfaces to move upward, which creates a decrease in the temperature in the interior ocean.

Using the same analysis as we did for the the WI case, we follow the depth of isopycnals corresponding to the thermocline and the subsurface (σθ = 26.4 and σθ = 25.8, respectively). As expected, the thermocline depth does not change significantly since there is no change in the wind stress curl. In contrast to this, the depth of the subsurface layers changes significantly and becomes shallower (Fig. 19). This is also demonstrated by the layer thickness between the 10° and the 16°C isotherms in Figs. 15b,c. The depth of the 10°C isotherm does not change significantly, while the 16°C isotherm moves upward. The behavior in the WI case is substantially different. The isopycnal depth gradually becomes shallower at the penetration point and from all points on the outcrop line of the ventilated region. They eventually extend throughout the ventilated region. Considering that a water column with lower potential vorticity, in the WI case, is confined to the southwestern part of the subtropical gyre, this subsurface shoaling in the CI case seems to result from a different mechanism.

Fig. 19.

Similar to Fig. 11, the change of the depth of the isopycnal depth σθ = 25.8, except for the cooling intensified case.

Fig. 19.

Similar to Fig. 11, the change of the depth of the isopycnal depth σθ = 25.8, except for the cooling intensified case.

An examination of the position of the outcrop line shows a significant southward shift, which corresponds to the negative temperature anomalies in the region of the westerlies (Fig. 4c). Figure 6c indicates that the negative anomalies occur between 14° and 15°C. In the northern part of the subtropical gyre, an examination of SST anomalies shows that the north to south gradient (∂TSST/∂y) has become larger in the south and smaller in the north. The change of the SST gradient causes the potential vorticity of the subducted water to increased (decreased) in the south (north). The origin of this negative temperature anomaly is between the north and the south.

Given the southward shift of the outcrop line, the response in the CI case is rather simple. Colder water penetrates from the surface of the westerly region and is distributed widely in the ventilated region. Since the tracer in the CI case penetrates approximately from the same place as in the SD case, a southward shift of the outcrop line causes the temperature of water particles to decrease. This is why maximum tracer concentrations are associated with decreased temperature.

6. Summary and discussion

A general circulation model has been used to investigate the response of the subduction process when the westerlies are intensified abruptly (Fig. 1). An intensification of the westerlies increases the Ekman pumping velocity and also cools the ocean by removing heat through sensible and latent heat flux. To identify the effects of both the wind stress and the cooling intensification on the subduction process, the region of the westerlies are driven by various anomalous forcings, shown in Fig. 3.

The SST anomalies in Fig. 4b show that the intensification of the westerly winds causes a similar pattern to the one seen in the decadal change in the North Pacific (Fig. 2). The SST anomalies in Fig. 4c show that the intensification of cooling causes colder water to be distributed at the surface in the zone defined by the equilibrium temperature field. The subsurface fields (top of the thermocline) in lower latitudes, where the forcing does not change locally, have negative anomalies in the wind intensified and cooling intensified cases. However the location of the anomalies differs between the two cases (Figs. 5 and 6).

To explain the results of the experiments, the change in the two isopycnal layer depths, corresponding to the subsurface (σθ = 25.8) and the thermocline (σθ = 26.4), is examined. Both of these lines outcrop in the region of the westerlies. In the wind intensified case, the subsurface isopycnal depth becomes shallower (Fig. 11), while the thermocline becomes deeper (Fig. 10). Also when comparing the wind intensified case to the standard case, a water column with lower potential vorticity (greater thickness) in the northwestern part of the subtropical gyre is shifted to the west (Fig. 13). This water penetrates southwestward from the penetration point that has shifted westward where the mixed layer front and the outcrop line intersect (Fig. 12). This leads to subsurface shoaling in response to the intensification of the westerlies and results in negative temperature anomalies in lower latitudes (Figs. 15a,c).

In the cooling intensified case, the subsurface isopycnal depth becomes shallower; however, the thermocline depth does not change significantly (Fig. 19). This is caused by the southward shift of an outcrop line which allows colder water to penetrate into the ventilated region on an isopycnal surface (Fig. 19). This process of penetration is different from that in the wind intensified case. However, similar to the wind intensified case, the cooling intensified case contains water with lower potential vorticity (greater thickness) in the subsurface temperature field of the low latitudes (Figs. 15b,d).

The mechanisms described above conclusively explains the difference between the wind stress and surface cooling effect. Moreover, evolution of a subarctic tracer distribution, which helps elucidate water mass distribution in the real ocean, is also useful to illustrate and to visualize the circulation from higher to lower latitudes.

When the westerly winds are intensified more tracer is distributed into the subtropical gyre (Fig. 8). This does not occur in the standard and cooling intensified cases. Tracer concentration increases in the wind intensified case, and the temperature decreases where the highest tracer concentration occurs in the cooling intensified case (Fig. 9). In the wind intensified case, the enhanced Ekman transport causes greater tracer subduction. This subduction is further enhanced by the repositioning of the convective region in both the northwestern part of the subtropical gyre and in the southern part of the subarctic gyre (Fig. 17). In the northwestern part of the subtropical gyre, deeper convection causes the tracer to enter into a deeper part of that region (Fig. 16). In the southern part of the subarctic gyre, shallower convection prevents the tracer from entering into a deeper part of that region (Fig. 18). This also explains the cause of the increased tracer concentration in the wind intensified case. Lower core temperature is caused by the southward shift of an outcrop line in the cooling intensified case (Fig. 12). In the combined case, changes in temperature and the subarctic tracer are explained by the linear combination of the effects of wind and cooling intensification.

The subarctic tracer injection experiments have shown that increased westerly winds supply more subarctic water to the subtropical gyre. This fact suggests that the surface water in higher latitudes (SSM) would be distributed farther to the southwest in the ventilated region of the real ocean, if the westerly winds were increased.

Watanabe and Mizuno (1994) speculated that the subduction process carries an anomalous surface temperature to the subsurface ocean. The present study seems to support this idea based on the timescale of the subduction process and the analysis of the changes in isopycnal depth.

The decadal change of the mode waters (North Pacific subtropical mode water and central mode water) shows an increase of mass and a decrease of core temperature (Yasuda and Hanawa 1997) when the westerlies are strong. Figure 20a shows zonal temperature sections at 37°N in the spring during the decades 1966–75 and 1976–85. The shaded region, with small vertical temperature gradient, was enhanced during the latter decade, which is consistent with the increased concentration of the subarctic tracer in the wind intensified case of this study. Figure 20b shows a scatterplot of temperature versus temperature gradient at 20°–40°N and 130°E–130°W for the summer of 1966–75 (left) and 1976–85 (right). There are two peaks where the temperature gradient is low located near 17° and 11°C in the previous decade, and near 16° and 10°C in the latter decade. The upper and lower peaks correspond to the NPSTMW and NPCMW, respectively. A water mass with a low temperature gradient has a colder core temperature in the latter decade, which is consistent with the result in the cooling intensified case in this study. It suggests that it is possible to identify the effects of intensification of wind and cooling on water mass formation and distribution in the North Pacific.

Fig. 20.

(a) Zonal temperature sections at 37°N in spring for 1966–75, 1976–85. (b) Scatter diagram of temperature versus vertical temperature gradient diagram for the summer temperature profiles in the region between 20°–40°N and 130°E–130°W for 1966–75 (left) and 1976–85 (right).(After T. Yasuda and K. Hanawa 1996, personal communication).

Fig. 20.

(a) Zonal temperature sections at 37°N in spring for 1966–75, 1976–85. (b) Scatter diagram of temperature versus vertical temperature gradient diagram for the summer temperature profiles in the region between 20°–40°N and 130°E–130°W for 1966–75 (left) and 1976–85 (right).(After T. Yasuda and K. Hanawa 1996, personal communication).

This study provides a detailed dynamical description of the response of the ocean to the intensified westerlies and identifies the specific effects associated with wind stress and cooling intensification. Previous studies addressed the effects of wind stress patterns on the tropical–subtropical circulation (e.g., Liu 1994; Liu and Philander 1995) but did not include mixed layer processes (Liu 1993a,b). Therefore, a close examination of the mixed layer distribution in this model presents important new aspects.

Kubokawa and Inui (1999) have found that the path of water with low potential vorticity on each isopycnal surface plays an important role in determining the thermocline structure. The path is formed by joining the points of minimum potential vorticity defined at each latitude that lead from the penetration point to the southwestern part of the subtropical gyre. In this study, the increased westerly winds cause the subsurface isopycnal surfaces to rise, a feature that is not predicted by the classical LPS model. The proximity of the paths of water with low potential vorticity on different isopycnal surfaces is not the dominant process for this subsurface shoaling (Fig. 14). Although the potential vorticity becomes lower in the wind intensified case compared to the standard case, due to the deeper mixed layer, the subsurface layer thickness on each isopycnal surface is not sufficiently altered. Rather, the westward shift of the penetration points causes the water with low potential vorticity to move farther to the west and dominate the southwestern part of the subtropical gyre. As a result, the subsurface isopycnal surfaces rise because of the westward shift of the water column with low potential vorticity.

In contrast to the complicated mechanisms in the wind intensified case, those in the cooling intensification case are fairly simple. This process could be explained by the LPS model. The same Ekman pumping velocity as that in the SD case forces the colder water, originating in a southward shift of an outcrop line, to penetrate into and to be distributed throughout the ventilated region. Furthermore, although the SST anomalies formed in the region of the westerlies are the same as those in the wind intensified case, subsurface temperature anomalies in the lower latitudes are significantly weaker. This occurs where there is no anomalous forcing.

In conclusion, we identify the effects of an abrupt wind stress and surface cooling intensifications in the region of the westerlies on the subduction process. In our model, the temperature of the equatorial region increases when the westerly winds are intensified. Debate remains as to whether or not this feature exists in the real ocean. What controls decadal changes of the equatorial region is important when considering global decadal changes.

Acknowledgments

Most of the numerical calculations were carried out on an NEC SX-3/44R computer at the Computing Center of Tohoku University. During the course of this study, we had useful discussions with members of Hokkaido University. Drs. Atushi Kubokawa and Shan-Ping Xie have provided invaluable comments. Discussions with Drs. Julian P. McCreary and Zhengyu Liu were also appreciated. Dr. Kim Van Scoy, Dr. Monica Stephenson, and Ms. Mary Kennedy helped us in improving our paper. Two reviewers have contributed to the quality of our paper. We expressed our thanks to all of them. In writing this paper, the first author was partially supported by the NSF sponsored Physical Oceanography Program in a grant to the Department of Atmospheric and Oceanic Sciences, University of Wisconsin—Madison.

REFERENCES

REFERENCES
Anderson, D. L. T., and A. E. Gill, 1975: Spin-up of a stratified ocean, with applications to upwelling. Deep-Sea Res., 22, 583–596.
——, and P. D. Killworth, 1979: Non-linear propagation of long Rossby waves. Deep-Sea Res., 26, 1033–1050.
Cox, M. D., and K. Bryan, 1984: A numerical model of the ventilated thermocline. J. Phys. Oceanogr., 14, 674–687.
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.
Haney, R. L., 1971: Surface thermal boundary condition for ocean circulation models. J. Phys. Oceanogr., 1, 241–248.
Huang, R. X., 1988: On the boundary value problems of the ideal-fluid thermocline. J. Phys. Oceanogr., 18, 619–641.
Inui, T., and K. Hanawa, 1997: A numerical investigation of effects of a tilt of the wind stress curl line on subduction process. J. Phys. Oceanogr., 27, 897–908.
Kubokawa, A., and T. Inui, 1999: Subtropical countercurrent in an idealized ocean GCM. J. Phys. Oceanogr., 29, 1303–1313.
Levitus, S., 1982: Climatological Atlas of the World Ocean. NOAA Prof. Paper No. 13, U.S. Department of Commerce, 173 pp.
Liu, Z., 1993a: Thermocline forced by varying Ekman pumping. Part I: Spinup and spin down. J. Phys. Oceanogr., 23, 2505–2522.
——, 1993b: Thermocline forced by varying Ekman pumping. Part II: Annual and decadal Ekman pumping. J. Phys. Oceanogr., 23, 2523–2540.
——, 1994: A simple model of the mass exchange between the subtropical and tropical ocean. J. Phys. Oceanogr., 24, 1153–1165.
——, and J. Pedlosky, 1994: Thermocline forced by annual and decadal surface temperature variation. J. Phys. Oceanogr., 24, 587–608.
——, and S. G. H. Philander, 1995: How different stress patterns affect the tropical–subtropical circulations of the upper ocean. J. Phys. Oceanogr., 25, 449–462.
Luyten, J., and H. Stommel, 1986: Gyres driven by combined wind and buoyancy flux. J. Phys. Oceanogr., 16, 1551–1560.
——, J. Pedlosky, and H. Stommel, 1983: The ventilated thermocline. J. Phys. Oceanogr., 13, 292–309.
Masuzawa, J., 1969: Subtropical Mode Water. Deep-Sea Res., 16, 463–472.
Nakamura, H., 1996: A pycnostad on the bottom of the ventilated portions in the central subtropical North Pacific: Its distribution and formation. J. Oceanogr., 52, 171–182.
——, 1998: Simple model prediction of horizontal temperature fields in the subtropical–subpolar system caused by sudden change in wind stress curl. J. Phys. Oceanogr., 28, 1578–1597.
Nitta, T., and S. Yamada, 1989: Recent warming of tropical sea surface temperature and its relationship to the Northern Hemisphere circulation. J. Meteor. Soc. Japan, 67, 375–383.
Pedlosky, J., 1986: The buoyancy and wind-driven ventilated thermocline. J. Phys. Oceanogr., 16, 1077–1087.
——, and P. Robbin, 1991: The role of the finite mixed-layer thickness in the structure of the ventilated thermocline. J. Phys. Oceanogr., 21, 1018–1031.
——, W. Smith, and J. R. Luyten, 1984: On the dynamics of the coupled mixed layer-thermocline system and the determination of the oceanic surface density. J. Phys. Oceanogr., 14, 1159–1171.
Reid, J. L., 1973: The shallow salinity minima of the Pacific Ocean. Deep-Sea Res., 20 (Suppl.), 51–68.
Suga, T., Y. Takei, and K. Hanawa, 1997: Thermostad distribution in the North Pacific subtropical gyre: The subtropical mode water and the central mode water. J. Phys. Oceanogr., 27, 140–152.
Talley, L. D., 1985: Ventilation of the subtropical North Pacific: The shallow salinity minimum. J. Phys. Oceanogr., 15, 633–649.
Tanimoto, Y., N. Iwasaka, K. Hanawa, and Y. Toba, 1993: Characteristic variations of sea surface temperature with multiple timescales on the North Pacific. J. Climate, 6, 1153–1160.
——, ——, and ——, 1997: Relationships between the sea surface temperatures, the atmospheric circulation and air–seas fluxes on multiple time scales. J. Meteor. Soc. Japan, 75, 831–849.
Trenberth, K. E., 1990: Recent observed interdecadal climate changes in the Northern Hemisphere. Bull. Amer. Meteor. Soc., 71, 988–993.
Tsuchiya, M., 1982: On the Pacific upper-water circulation. J. Mar. Res., 40 (Suppl.), 777–799.
UNESCO, 1981: Tenth report of the joint panel on oceanographic tables and standards. UNESCO Marine Science Tech. Paper 36.
Watanabe, T., and K. Mizuno, 1994: Decadal changes of the thermal structure in the North Pacific. Int. WOCE Newsletter, 15, 10–13.
Williams, R. G., 1991: The role of the mixed layer in setting the potential vorticity of the main thermocline. J. Phys. Oceanogr., 21, 1803–1814.
Yasuda, T., and K. Hanawa, 1997: Decadal changes in the mode waters in the midlatitude North Pacific. J. Phys. Oceanogr., 27, 858–870.

APPENDIX

The Equation for Tracer Concentrations

In the tracer-injection experiment, an additional equation for tracer concentrations Tr is used. It is

 
Trt + Γ(Tr) = FTr + sources,

where

 
formula

The advection operator Γ is defined as

 
formula

The constants are AATV (vertical mixing coefficient for tracer), ATH (horizontal mixing coefficient for tracer), and a (the radius of the earth). The range of Tr is from 0 to 100. The “sources” term can be nonzero only at the surface of the subarctic region, which is located north of the line defined by ψ = 0. It is given by

 
formula

where κ is set to be constant at 0.02 day−1.

In the case of additional experiments, it is also given by

 
formula

where Tri is the tracer concentration in the region i (i = 1, . . . , 6).

Footnotes

Corresponding author address: Dr. Tomoko Inui, Department of Atmospheric and Oceanic Sciences and Center for Climatic Research, University of Wisconsin—Madison, 1225 W. Dayton Street, Madison, WI 53706.