Changes in water properties and transports along 24°N in the North Pacific between 1985 and 2005



[1] We conducted a trans-Pacific hydrographic section along 24°N in 2005 to investigate the ocean structure and its changes from previous observations in 1985. We detected significant basin-average water property changes from 1985 to 2005. Apparent oxygen utilization increased below the thermocline by up to 6 μmol kg−1 around the density of the central mode water (around 600 m). It appeared that the North Pacific intermediate water (around 800 m) was less dense in 2005 than in 1985 because of warming. From the decrease of the zonal gradient of the temperature and salinity around the North Pacific deep water (2500–4000 m) and lower circumpolar deep water (<4000 m), we suggest that northward bottom water and southward deep water transports became weaker from 1985 to 2005, consistent with the speculation from the observed temperature increase of the bottom water along its main path in previous studies. Although these water property changes suggest a slowdown of the meridional overturn in the North Pacific and large transport changes in the deep layers (below 4000 m) are estimated from an inverse method, significant heat transport changes were not detected. The estimated temperature transport change of 0.1 PW between the two sections was mainly due to shallow overturn changes, especially changes in the Kuroshio. To describe variability due to the Kuroshio changes, we estimated mass and heat transport changes from long-term observations of the Kuroshio in the Okinawa Trough, and we determined decadal variability of temperature transports, which was consistent with the variability estimated from sea-surface flux data sets.

1. Introduction

[2] Ocean circulation, which transports heat, freshwater, nutrients, and gases, plays an important role in climate. Near the midlatitudes, poleward heat transport by the ocean circulation is about the same as atmospheric heat transport [e.g., Bryden and Imawaki, 2001]. During the World Ocean Circulation Experiment (WOCE), the WOCE Hydrographic Program (WHP) collected data from the ocean surface to the bottom along trans-ocean hydrographic sections. These data were carefully processed and highly accurate. On the basis of these data, transports of heat, salinity, and nutrients across the sections by the ocean circulation were estimated [e.g., Ganachaud and Wunsch, 2000].

[3] In the North Pacific, Bryden et al. [1991] estimated heat transport across 24°N to be 0.76 PW based on data from the WHP-P3 section (in 1985). Using data from long-term XBT and XCTD observations near 24°N, heat transport was estimated to be 0.83 PW [Roemmich et al., 2001]. Ganachaud and Wunsch [2003] estimated the transports across WOCE sections using an inverse method. Their inverse method solves mass and water property conservation equations to determine reference velocities for geostrophic velocities. The resulting transport estimates are consistent within all of the sections that they used in the world oceans. Their estimated value for heat transport across 24°N in the Pacific is 0.52 PW. Regarding changes in the transports, Bryden et al. [2005] reported that the meridional overturn across 25°N had decreased by 30% in the Atlantic from 1957 to 2004. However, using 1-year data from moored instruments deployed at the eastern and western boundaries along 26.5°N in the Atlantic, Cunningham et al. [2007] demonstrated that such changes could be explained by short-term variability. In the North Pacific, the interannual variability in heat transport was reported to be more than 0.4 PW during 1993–1999 based on XCTD and XBT data [Roemmich et al., 2001]. Regarding changes and variability in ocean heat content, Willis et al. [2004] estimated a warming rate in the upper 700 m of the world oceans from 1993 to 2003 of 0.86 W m−2. Levitus et al. [2005] reported a 0.2 W m−2 warming of the upper 3000 m from 1995 to 2003. We are not aware of any reports showing changes in meridional overturning in the North Pacific.

[4] In the shallower layers, changes in the ocean are often described in terms of water mass property changes. In the North Pacific, subtropical mode water (STMW) [Masuzawa, 1969] is divided into three classes on the basis of its formation area, formation process, and transport path. North Pacific subtropical mode water (NPSTMW), with a core density of σθ = 25.4 kg m−3, is the westernmost of the three, found from 150–160°E around 30°N [Suga and Hanawa, 1995]. Central mode water (CMW), with a core density of σθ = 26.2 kg m−3, is located around 160°W, 35°N [Suga et al., 1997]. Eastern subtropical mode water (ESTMW) is formed offshore of North America around 30°N [Hautala and Roemmich, 1998]. Mode waters are connected with climate change through sea-surface heat and fresh water fluxes. The change in the cross-sectional core area of NPSTMW along 137°E has a strong correlation with sea-surface fluxes in the Kuroshio Extension region and its recirculation gyre [Suga and Hanawa, 1995]. Suga et al. [2003] reported that the temperature and salinity of CMW abruptly increased from 1988 to 1989 and its density decreased. They suggest that these changes were caused by weakening of the Aleutian Low. Although variability in or near formation areas has been investigated in many studies, there are few reports about property changes in the subtropical region, to which the mode waters are subducted.

[5] North Pacific intermediate water (NPIW) is characterized by a salinity minimum at depths of 200–800 m. Wong et al. [1999] compared historical data from the 1960s with the data from the section along 24°N in 1985, and showed that the NPIW core had freshened by 0.017, and they suggested a strengthening of the atmospheric hydrological cycle. Because these mode waters and intermediate water are influenced by atmospheric changes, it is important to detail the water property changes in the ocean interior, to know how much climate variability is retained in the ocean. Although Kouketsu et al. [2007] reported that NPIW temperature increased in the western part of the subtropical gyre from the 1980s, through the 1990s and into the 2000s, based on a comparison between the WHP sections in the subtropical gyre and their revisits, the trans-Pacific property changes of NPIW were not described.

[6] For the deep layers in the ocean, changes are detected by comparisons of repeat WHP sections, since few other data are available at depths exceeding 3000 m. Fukasawa et al. [2004] found warming at depths exceeding 3000 m along the 47°N section in the North Pacific. Kawano et al. [2006b] showed warming by 0.005 to 0.01°C along the pathway of lower circumpolar deep water (LCDW) by comparing Pacific WHP sections with their revisits. Warming at depths exceeding 3000 m has been detected along the other repeat sections in the Pacific, and this warming may be significant as a trans-ocean change [Johnson et al., 2007]. Although these property changes suggest circulation changes in the deep layers, we are not aware of any estimates of deep transport decadal variability in the Pacific.

[7] For the present study, we conducted a 2005 revisit of the 1985 WHP-P3 section. By comparing the data from sections in 1985 and 2005, we detail property changes in water masses from the surface to the bottom. From these changes, we infer circulation changes. Furthermore, we re-estimate transports across 24°N in the North Pacific using an inverse method for a single section, and describe changes in these transports since the WHP observations. Since the shallow overturn at depths shallower than 700 m plays a major role in the northward heat transport in the North Pacific [e.g., Bryden et al., 1991; Talley, 2003; Talley et al., 2003], we estimate the heat transport variability by using a long-term observation section, known as the PN line, in the Okinawa Trough.

2. Data and Methods

[8] We used data from the WHP-P3 hydrographic section in 1985 (206 stations) and its revisit in 2005 (207 stations) along 24°N, from the western coast of North America to the East China Sea (Figure 1). The WHP-P3 data are available from the Climate Variability and Predictability/Carbon Hydrographic Data Office (CLIVAR/CCHDO; The station locations were the same in 1985 and 2005 except that we added one station at the intersection of WHP-P3 and WHP-P10 (149°E) in 2005, and five stations east of Hawaii in 1985 were shifted during the 2005 revisit to avoid a naval exercise area (Figure 1). The accuracies of temperature and salinity measurements are 0.001°C and 0.002, respectively. The accuracy of oxygen measurements is about 1 μmol kg−1 in the deep layer (detailed by Kawano et al. [2006b]). The accuracy of silicate measurements is 0.5% in the present study, based on the cruise report of WHP-P3, and a similar accuracy in the revisit. Traceability is a problem in comparing salinity, oxygen, and silicate values. Johnson et al. [1999] estimated the standard deviation of crossover differences for silicate among the WOCE Pacific sections to be about 2–3%. The use of standard sea water (SSW) batch corrections for WOCE expeditions leads to a significant reduction in the salinity variance at crossovers of WHP sections in the Pacific. To reduce apparent density differences in deep layers due to (noun?) biases from SSW batches, we applied the latest batch corrections [Kawano et al., 2006a] to the salinity data. Salinity values in this study are shown in the practical salinity scale 1978 (PSS-78).

Figure 1.

Locations of the observation stations in 1985 (open circles) and 2005 (solid circles). Depths deeper than 3000 m are shown by gray shading. The depth contour interval is 1000 m.

[9] To estimate mass transport by the Kuroshio in the Okinawa Trough, we use the long-term observation section designated as the PN line (Figure 2), which is located near the WHP-P3 section in the East China Sea. Observations along the PN line are carried out four times a year by the Nagasaki Marine Observatory, Japan Meteorological Agency (JMA).

Figure 2.

Location of the PN-line observation stations (solid circles) east of Okinawa. Open circles indicate WHP-P3 stations. The depth contour interval is 1000 m.

2.1. Changes in Potential Temperature and Salinity

[10] We describe water property changes on neutral density surfaces as calculated following Jackett and McDougall [1997]. If a water mass experiences remote warming, the density of the water mass become less than before. Changes in salinities of the water mass on a neutral density surface from before to after the warming are caused by the density changes of the water mass. Whether salinity (temperature) increases or decreases on a neutral density surface depends on the vertical profile of salinity at an observation station, even if the temperature of the water mass increases because of warming. To account for such situations, Bindoff and McDougall [1994] divided the changes in the relationship between potential temperature and salinity into the three processes that could occur in the water mass formation region: “pure warming”, “pure freshening” and “pure heaving”. In pure warming, the assumption is that only the heat flux from the atmosphere to the ocean increases, and that other forcing factors do not change. In pure freshening, only the freshwater flux increases. In pure heaving, it is assumed that neither the heat nor the freshwater flux has changed. In this case the water properties on the neutral density surface do not change and any apparent changes are due to vertical movement of the isopycnal surface. The relationships described by Bindoff and McDougall [1994] are summarized as following.

[11] Observed changes in potential temperature (θ) and salinity (S) have the approximate relationships

equation image

[12] Here, α and β are the thermal expansion and haline contraction coefficients, respectively. θ′∣z and S′∣z are the changes in θ and S on an isobar, θ′∣n and S′∣n are the changes on a neutral density surface, θz and Sz are the vertical gradients, and N′ is a neutral density surface depth change. Changes on isobars (θ′∣z, S′∣z) are calculated at the mean depths of neutral density surfaces.

[13] A set of conditions defining pure warming, freshening, and heave have been derived [Bindoff and McDougall, 1994]. For our purposes, the following correspondences are useful:warming:

equation image


equation image

[14] For this study, we present the regionally averaged property changes along with confidence levels. To estimate the 90% confidence levels using Student's t test, the effective numbers of degrees of freedom in property changes are needed. These numbers are estimated using integral spatial scales [von Storch and Zwiers, 2001], which are determined by autocovariances following Johnson and Doney [2006]. The resulting integral spatial scales (typical values are 100–300 km, not shown) generally decrease with increasing depth.

2.2. Inverse Method

[15] Inverse methods have been used to quantify the ocean circulation in many studies [e.g., Wunsch, 1978; Wunsch et al., 1983; Robbins and Bryden, 1994; Sloyan and Rintoul, 2001]. We used a simple version of the method with mass conservation equations to study the 1985 and 2005 sections. In our inverse method, we divided the water column into 15 layers on the basis of neutral density (γn) surfaces (Table 1), and estimated unknown reference velocities for geostrophic velocity profiles between stations by solving the mass conservation equations, accounting for dianeutral mass fluxes. The initial reference level, where the geostrophic velocity normal to the section is assumed to be zero, is γn = 28.0 kg m−3 (near 2500-m depth). This reference level is slightly different from 3000 m used by Bryden et al. [1991], but both levels are near the silicate maximum indicating old water. Mass transports estimated using a reference level of γn = 28.05 kg m−3 (near 3000-m depth) for the inversion do not change much (not shown). Because the layers in the Philippine Basin for γn > 28.0 kg m−3 are isolated by the Izu-Ogasawara Ridge, the mass conservation equations for that deep basin are analyzed separately from those for the eastern basins. Thus we solved 17 mass conservation equations considering dianeutral transports (for 15 trans-Pacific layers and 2 layers in the Philippine Basin). For the dianeutral transports of the mass conservation equations, weighting factors between dianeutral and horizontal cross-sectional transports were set to the value at which the condition number for solving the equations is at a minimum, following McIntosh and Rintoul [1997]. Furthermore, we applied the constraint that dianeutral mass fluxes are positive at each interface for γn > 27.30 kg m−3, since it is known that the North Pacific Ocean has no deep convection. These settings were also employed by Robbins and Bryden [1994]. Transport through the Bering Strait was set to 0.8 Sv [Coachman and Aagaard, 1988]. Since we focused on long-term changes in transport, the Ekman transport on 1° grids along the section was calculated using the NCEP wind stress [Kalnay et al., 1996] averaged from 1985–2005. The temperatures averaged over the upper 50 m from the World Ocean Atlas 2005 [Locarnini et al., 2006] are used in calculating the Ekman heat transport, assuming that the averaged temperatures represent the velocity weighted averaged temperatures for the Ekman transport.

Table 1. Layer Parameters Used for the Inverse Methoda
γnσ (kg m−3)Depth (m)θ (°C)
  • a

    Properties from observations in 2005 were averaged on each interface. σθ, σ2, and σ4 are potential densities with reference levels at 0, 2000, and 4000 m, respectively.

Surface 023.95
24.30σθ = 24.2910021.40
26.00σθ = 25.9534012.26
26.60σθ = 26.514908.42
27.00σθ = 26.876305.85
27.30σθ = 27.148004.62
27.60σθ = 27.4111103.50
27.80σθ = 27.5815502.51
27.95σ2 = 36.9121501.77
28.00σ2 = 36.9525101.54
28.03σ2 = 36.9828201.41
28.06σ4 = 45.8332701.28
28.08σ4 = 45.8637301.19
28.10σ4 = 45.8843101.11
28.12σ4 = 45.9050101.03

2.3. Data From the PN Line

[16] The PN line is a hydrographic section in the Okinawa Trough near the WHP-P3 section (Figure 2). Observations along the PN line have been carried out since 1970. However, before 1987 the observations were carried out using Nansen bottles and bathythermographs, and with sampling densities different from recent conductivity-temperature-depth (CTD) profiler observations. Therefore an optimal interpolation method was used to regrid the earlier data to a vertical resolution of 1 m to coincide better with the recent CTD observation stations (Figure 2). The mean vertical structure for the section was calculated from the CTD data from 1990 to the most recent year (horizontal scale, 100 km; the vertical scale, 300 m). Assuming a Gaussian correlation function, 100-km horizontal and 300-m vertical correlation lengths were estimated from the deviations from the mean structure using a least squares method. The noise variance for the interpolation was assumed to be 1% of the signal variance. Estimates of transport using the converted pre-1987 data and the more recent data are comparable. We used this combined data set to estimate the long-term variability of the Kuroshio transports.

3. Results

3.1. Property Changes

[17] Trans-Pacific θ-S relationships along 24°N shifted between 1985 and 2005 (Figure 3). In the shallow layers, for γn < 26.6 kg m−3 (σθ ≃ 26.5 kg m−3), which outcrop at the sea surface in the North Pacific in winter, temperatures and salinities decreased on isopycnals from 1985 to 2005. The NPIW salinity minimum also decreased by 0.0045. It appears that the freshening by 0.017 observed from the 1960s to the WHP-P3 sampling in 1985 [Wong et al., 1999] continued during the 1990s, although to a lesser degree. In contrast, we detected a slight increase in the temperatures and salinities γn > 27.0 kg m−3 (σθ ≃ 26.9 kg m−3). θ and S increased for γn > 27.0 kg m−3. The maximum temperature increase (+0.028 K) was at γn ≃ 27.4 kg m−3 (σθ ≃ 27.3 kg m−3), where there was a corresponding increase of 0.004 in the salinity. Although Wong et al. [1999] described a decrease in salinity in the layers deeper than NPIW from the 1960s to 1985, that trend was reversed from 1985 to 2005.

Figure 3.

Longitudinal average θ-S relationships on neutral density surfaces at 24°N across the Pacific for 1985 (dashed line) and 2005 (solid line). Gray curves are contours for potential densities.

[18] The average apparent oxygen utilization (AOU) and silicate profiles over the trans-Pacific section changed from 1985 to 2005 (Figures 4 and 5) . AOU is calculated as the difference between the measured oxygen concentration and the oxygen saturation concentration, estimated following Garcia and Gordon [1992]. AOU increased for γn > 25.0 kg m−3, especially around γn = 26.6 kg m−3, where the magnitude of the increase reached 6 μmol kg−1. The increase in AOU means that the water mass around γn = 26.6 kg m−3 took longer to reach the section, or that ventilation reduced or stopped in this layer, assuming that changes in oxygen utilization rates (OUR) due to changes in biological activity were negligible. The increase in silicate may be due to water mass transport from a more northern silica-rich region, or due to the water mass experiencing more remineralization. Therefore the temperature and salinity decreases on the neutral density surfaces above the salinity minimum (upper NPIW and CMW) are more likely to be explained by water mass property changes than by an advection velocity increase from the north, where temperature and salinity are low.

Figure 4.

Profiles for AOU (μmol kg−1). (a) Longitudinal average on neutral density surfaces at 24°N across the Pacific for 1985 (dashed line) and 2005 (solid line). (b) The averaged change from 1985 to 2005 on the neutral density surfaces with 90% confidence intervals (gray bars).

Figure 5.

The same as Figure 4 but for silicate concentrations (μmol kg−1).

[19] Changes in each water property from 2005 to 1985 vary spatially along the hydrographic section (Figures 68) . A large cooling (and freshening) was detected west of Hawaii around γn = 25.8 kg m−3. This pattern may correspond to the freshening reported by Lukas [2001]. Salinity changes within the density range of NPSTMW in the western part of the section show NPSTMW freshening (Figure 7). Furthermore, around this density surface near the coast of North America, the temperature (and salinity) decrease is very large. These changes in ESTMW and the California Current system contribute to the large freshening of the whole section (see Figure 3).

Figure 6.

Vertical section at 24°N across the Pacific showing changes (2005 − 1985 [2005 values minus 1985 values]) in potential temperature (°C) on (a) isobars, (b) neutral density surfaces, and (c) neutral density surfaces in the deep layer (γn > 27.8 kg m−3).

Figure 7.

Vertical section at 24°N across the Pacific showing changes (2005 − 1985) in salinity on isobars.

Figure 8.

Vertical section at 24°N across the Pacific showing changes (2005 − 1985) in AOU (μmol kg−1) and silicate (μmol kg−1) on neutral density surfaces.

[20] West of Hawaii there is a pattern of decreasing temperature and salinity for γn < 26.9 kg m−3 and increasing for γn > 26.9 kg m−3 (Figure 7). Kouketsu et al. [2007] also reported this pattern and speculated that it is caused by warming of the NPIW core. AOU and silicate increased with time from 140°E to 170°W for 26.0 < γn < 27.1 kg m−3 (Figure 8). These changes in AOU and silicate suggest that the temperature decrease above the salinity minimum cannot be explained by an advection velocity increase from the north.

[21] In the deep layer below 3000 m, from 160°E to 170°W, temperature increased, as Kawano et al. [2006b] also reported. In contrast, from 140°E to 160°E, temperature in the deep layer of 3000–5000 m decreased, while the temperature increased near the bottom (below 5000 m) (Figure 6). Furthermore, no significant salinity change was detected from 160°E to 170°W, but a small positive change in salinity was detected between 140°E and 160°E at depths of 3000–5000 m (Figure 7). The density increase around 4000 m from 165°E to 170°W is smaller than that from 140°E to 165°E. From the thermal wind relationships, these density changes can cause northward transport changes above 4000 m and southward changes below.

[22] Since the patterns of water property changes vary zonally, we divided the section into three parts: the eastern part, from Hawaii to North America; the central part, from the Izu Ridge to Hawaii; and the western part, from Okinawa to the Izu Ridge. Components of the water property changes defined in equation (1) were determined for each of the three parts (Figures 911), as were the θ-S relationships (Figure 12).

Figure 9.

Components of changes in θ and S (from equation (1)) for the western part of the section at 24°N across the Pacific (between Okinawa and Izu Ridge). (red) Changes on isobars (αθ′∣z, βS′∣z); (blue) changes on neutral density surfaces (αθ′∣n, βS′∣n); and (green) effects of changes in depths of neutral density surfaces (Nαθz, NβSz). Error bars are 90% confidence intervals.

Figure 10.

The same as Figure 9 for the central part of the section (between Izu Ridge and Hawaii).

Figure 11.

The same as Figure 9 for the eastern part of the section (east of Hawaii).

Figure 12.

The area-averaged θ-S curves on neutral density surfaces for 1985 (dashed lines) and 2005 (solid lines) from (a) Okinawa to Izu Ridge, (b) Izu Ridge to Hawaii, and (c) Hawaii to North America, along 24°N in the Pacific. Gray curves are contours for potential densities.

[23] In the western part of the section between Okinawa and the Izu Ridge, heave is mostly counterbalanced by changes on isobars in all layers (see Figure 9). Changes on neutral density surfaces around the salinity minimum for 26.6 kg m−3 < γn < 27.2 kg m−3 exceed the 90% confidence levels (Figure 9). Salinity at the salinity minimum decreased (Figure 12a). Changes on neutral density surfaces in the intermediate layers appear to be caused by “pure freshening” based on the relationships defined in equations (2) and (3). The fresh tongue of NPIW, which extends from east to west in the western part of the section, stretches and shrinks annually [Nakano et al., 2005]. The apparent freshening of this water mass (Figure 9) may result from such behavior of this tongue. Since the changes on neutral density surfaces are insignificant for γn > 27.6 kg m−3 (Figure 9), changes in the Philippine Basin are caused mainly by heave.

[24] Although heave is balanced by changes on isobars for γn < 26.0 kg m−3 in the central part of the section (from the Izu Ridge to Hawaii) changes on neutral density surfaces are large compared to the other terms in the intermediate layer (see Figure 10). AOU increased substantially in this region (see Figure 8), suggesting an increase in the transit time of fresh cold water from the subarctic gyre due to advection decrease from north or northward displacement of outcrop positions of these layers which increase the transit length. Thus the decrease in salinity and temperature above the salinity minimum (see Figure 10) cannot be explained by an increase in advection velocity from the north alone, where salinity and temperature are low. The pattern of water property changes around the salinity minimum (Figure 12b) can be explained by “pure warming”, since θ changes on the neutral density surfaces are negative above the salinity minimum and positive below the salinity minimum, which is consistent with equation (2), and θ changes on isobars are significantly positive (Figure 10). Above the salinity minimum around γn = 26.6 kg m−3, “pure freshening” effect may not be negligible, since θ changes on isobars are not significant (Figure 10). These interpretations support the speculation by Kouketsu et al. [2007] that water property changes around the salinity minimum were caused mainly by warming.

[25] For γn > 27.4 kg m−3, temperature and salinity increased significantly (Figure 10). Even though the salinity batch corrections [Kawano et al., 2006a] decrease the salinity values from the 2005 observations, the increases of the salinities as well as temperatures on neutral density surfaces are also considerable. Changes in the deep layers (Figure 6c, 165°E–170°W) may be caused by water property changes, since heave is insignificant (Figure 10). Since the bottom layer of this central part of the section is the main path of LCDW [e.g., Kawano et al., 2006b], the temperature and salinity increases suggest a decrease in LCDW advection. This interpretation is consistent with the silicate increase in the bottom layer (Figure 8), since silicate concentrations increase northward near the bottom in the central part of the North Pacific [see Mantyla and Reid, 1983, Figure 2e].

[26] In the eastern part of the section, from Hawaii to North America, the NPIW salinity minimum is not clearly defined (Figure 12c), with shallow salinity minimum at 26.0 < σθ < 26.5 kg m−3 in this region [Yuan and Talley, 1992]. Changes caused by heave are smaller than changes on neutral density surfaces and isobars for γn < 27.4 kg m−3 (Figure 11). Changes for γn < 27.0 kg m−3 may be affected by the changes in the advection on neutral density surfaces. Strong negative temperature and salinity anomalies in summer–fall 2002 around the pycnocline (30–150 m) on the Oregon and Vancouver Island continental shelves have been reported [e.g., Freeland et al., 2003]. These anomalies remained at least through 2005 [Georicke et al., 2005]. The changes west of 170°W for γn < 26.6 kg m−3 reported here may be strongly affected by this recent invasion of subarctic water. Changes for γn > 27.4 kg m−3 may be caused by “pure warming” processes, since salinity changes on the neutral density surfaces are counterbalanced by salinity changes from heave effects. However, the evidence is not conclusive because the balance between heave effects and changes on neutral density surfaces is the determining factor for temperature changes (Figure 11a), and the uncertainty of heave effects on salinity is large.

3.2. Transport Changes

[27] Warming of the water masses of 26.0 < γn < 27.3 kg m−3 is indicative of a temperature increase in the southward flow in the interior ocean. NPIW and CMW in the central part of the section were influenced by warming processes, and the pattern corresponding to these changes is predominant in the section-averaged θ-S relationship (Figure 3). Furthermore, changes in water properties suggest that northward transport of LCDW and southward transport of NPDW may have weakened. The overturning circulation and northward transport of mass and heat may have decreased because of these changes.

[28] We estimated mass and heat transports using inverse methods and compared the results for the 1985 and 2005 sections. The estimated overturn stream function, that Antarctic intermediate water (AAIW) and LCDW transported northward and NPIW and NPDW transported southward, was similar to that in previous studies [e.g., Bryden et al., 1991; Ganachaud and Wunsch, 2000; Talley, 2003] (Figure 13a). The northward estimated transports in specific density categories are not much different from 1985 to 2005, although we inferred an overturn slowdown from water property changes.

Figure 13.

Mass transports for 1985 (dashed lines) and 2005 (solid lines) by (a) density category and (b) depth . The density categories are defined only by density, without accounting for specific water properties, and the categorization is similar to the one used by Talley [2003]: surface, from the surface to γn = 26.0 kg m−3; NPIW, γn = 26.0–27.3 kg m−3; AAIW, γn = 27.3–27.6 kg m−3; NPDW1, γn = 27.6–27.95 kg m−3; NPDW2, γn = 27.95–28.03 kg m−3; NPDW3, γn = 28.03–28.08 kg m−3; LCDW, γn = 28.08 kg m−3 to the bottom.

[29] Changes in mass transport are large in the deep layers (Figure 13b). Northward transport in the bottom layers (below 5000 m) and southward transport of the deep layers (3000–5000 m) both decreased between 1985 and 2005. This suggests that geostrophic velocity in the deep layers has changed substantially, as is inferred from the property changes there. Kawano et al. [2006b] suggested that the bottom water warming in the Pacific is not due to the direct warming of LCDW but rather due to changes in the pressure field along the LCDW pathway that are propagated by Kelvin and Rossby waves from the Antarctic. Numerical model analysis shows a rapid propagation of these changes from the source region of LCDW to the North Pacific [Nakano and Suginohara, 2002]. Transport changes in the deep layers may be due to these pressure field changes, which could reduce the amount of cold bottom water north of 24°N.

[30] Temperature transports changes corresponded to the mass transport changes. The total temperature transport (T) across 24°N is defined as

equation image

[31] Here, cp is the specific heat capacity of seawater, ρ is the density, and v is the velocity estimated from the inverse method. Total temperature transport decreased from 0.42 PW in 1985 to 0.31 PW in 2005. This total change of approximately 0.1 PW was caused by a transport change in the surface layers (not shown).

[32] Following previous studies [e.g., Bryden et al., 1991; Wilkin et al., 1995], we divided the temperature transport (T) into two components for this study, the temperature transport by the Kuroshio (a western boundary current; Tk) and that by the interior ocean (Ti):

equation image

[33] Here, equation image and equation image are longitudinal integrations over the Okinawa Trough and from Okinawa to the west coast of North America, respectively; Vk and Vi are mass transports in the Kuroshio and the interior ocean, respectively; and θk and θi are average temperatures weighted by mass transports in the Kuroshio and interior ocean, respectively.

[34] If volume north of 24°N is conserved, then:

equation image

[35] The interior-ocean average temperatures are 14.6°C for 1985 and 14.4°C for 2005. The difference in average interior ocean temperature weighted by the mass transport is about 0.2°C. This difference corresponds to a temperature transport of 0.02 PW, assuming that transport in the interior ocean balances that in the Kuroshio. This result suggests that changes in the Kuroshio dominate the large changes in the temperature transports in this estimate.

3.3. Variability of the Kuroshio

[36] In the previous subsection, variability of temperature transport was not estimated from the trans-Pacific section data, although changes in circulation were suggested. In this subsection, we use the mean temperature in the interior ocean from the west coast of North America to Okinawa (θi), together with mass-transport-weighted data from long-term hydrographic observations of the Kuroshio on the PN line, to estimate the variability of temperature transports across 24°N caused by variability of the Kuroshio. The average interior ocean temperature (θi) was set at 14.5°C based on our two estimates using the inverse method. We calculated the temperature transports by the Kuroshio and its average temperature, assuming that the mass transport of the Kuroshio at the PN line compensates for the southward return flow in the interior ocean (equations (4) and (5)). Furthermore, to show long-term variability and allow comparison of our results with those of Kawai et al. [2008], we calculated 5-year running means for the estimated transports.

[37] The long-term average Kuroshio transport of about 26 × 109 kg s−1 (Figure 14a) is consistent with results from the inverse method (26 × 109 kg s−1 for 1985 and 28 × 109 kg s−1 for 2005). Transport variability is the same as that reported by Kawabe [1995]. The estimated temperature transport in this study decreased in the 1980s and increased in the 1990s. The amplitude of these changes (about 0.1 PW) is similar to the decadal variability estimated from the various sea-surface flux data sets [Kawai et al., 2008] in the 1990s (Figure 14c), although estimates based on flux data are larger than estimates from the present study for around 1985. Since changes in Ekman transport are important to the changes in the temperature transports for the 1980s, as estimated by Kawai et al. [2008] using results from ocean general circulation model, the difference between their estimate and the one in this study may be due to the assumed constant interior ocean circulation. The variabilities of temperature and mass transports in this study are similar to those of the averaged temperature, weighted by mass transport (Figure 14). The differences from the 1990s to the 2000s in mass transports and averaged temperature are 1.5 × 109 kg s−1 and 0.5°C, and their contributions to the temperature transport variability are about 0.03 PW and 0.07 PW, respectively. Thus temperature observations in the Kuroshio are important for estimating temperature transport.

Figure 14.

Variability of the Kuroshio in the Okinawa Trough (PN line). (a) Mass transport; (b) average temperature; (c) temperature transport. Error bars represent standard errors for the 5-year running mean.

[38] On the other hand, interannual and seasonal variability of the Kuroshio temperature transports is very large, as indicated by the standard error of 0.05 PW for the 5-year running means (Figure 14). This large variation over short timescales includes a large seasonal variability, which is estimated to be about 0.2 PW based on the PN-line hydrographic data and current measurement data [Ichikawa and Chaen, 2000]. The temperature transports in April 1985 and January 2006 were estimated to be 0.41 PW (0.61 PW in July 1985) and 0.36 PW (0.32 PW in April 2006), respectively. Since these values are similar to estimates from the original observations in 1985 and 2005, which included observations near the PN line in June 1985 and January 2006, the differences in total temperature transports in a comparison between observations in 1985 and 2005 are mainly due to Kuroshio transport differences.

4. Summary and Discussion

[39] We conducted revisit observations in 2005 of a 1985 WOCE trans-Pacific section at 24°N to observe and interpret changes in water properties. In the shallow layers of 26.0 < γn < 27.1 kg m−3, the observed temperature decrease (increase) on neutral density surfaces above (below) the salinity minimum was caused by warming, and there was a slight freshening by 0.0045 around the NPIW salinity minimum. Around γn = 27.3 kg m−3, temperature and salinity increased on the neutral density surfaces. AOU increased for γn > 25.0 kg m−3 (σθ ≃ 24.9 kg m−3) to the bottom. In particular, the section average AOU increase was 6 μmol kg−1 around γn = 26.6 kg m−3 (σθ ≃ 26.5 kg m−3). This suggests that the water mass sampled in 2005 was older than that sampled in 1985, and that the temperature and salinity decreases at 26.0 < γn < 26.8 kg m−3 (σθ = 25.9–26.7 kg m−3) were not caused by an increase in transport from the subarctic gyre. Furthermore, this water mass in 2005 was less dense than in 1985 because of warming and it contributed to transports at shallower depths.

[40] In the deep layer, based on temperature and salinity changes and the thermal wind, we suggest that northward LCDW transport and southward NPDW transport became weaker, agreeing with the speculation by Kawano et al. [2006b] that was based on an observed temperature increase along the main path of the LCDW. Although the property changes suggest a slowdown of the meridional overturning in the North Pacific, changes in temperature transport were not due to this change in deep circulation. The temperature transport change was related to the shallow overturn, including the Kuroshio, which has a large variability as shown in previous studies [e.g., Bryden et al., 1991; Talley, 2003]. Variability of temperature transport estimated using the interior ocean mean temperature and the Kuroshio transports along the PN line was consistent with an estimate based on sea-surface flux data from the 1990s. Variability is approximately 0.1 PW.

[41] An AOU increase detected west of Hawaii around γn = 26.6 kg m−3 was similar to that observed previously in the subarctic North Pacific [Emerson et al., 2004], but in the subarctic the decrease was much larger. Emerson et al. [2004] suggested that the AOU decrease detected on σθ = 26.6 kg m−3 resulted from this density surface ceasing to outcrop in the North Pacific in the latter decades of the 20th century. This explanation may apply to the AOU decrease we detected along 24°N. However, the effects of changes in OUR on observed changes in AOU is unknown. Mecking et al. [2006] reported that the difference of OUR from 1985 to 2000 reduces to less than 1 μmol kg−1 year−1 only in the eastern part of the section (east of 127°W) along 24°N. An AOU increase in the subtropical gyre similar to that observed here was investigated using a numerical model that accounted for biochemical processes, and it was concluded that changes in circulation were the primary cause of the AOU changes in the lower thermocline [Deutsch et al., 2005, 2006]. Their results indicated an AOU decrease in the central and western regions of the subtropical gyre from the 1980s to the 1990s. This decrease was not observed in the present study, possibly due to the lack of some transport processes in the model, such as eddy fluxes. There are large horizontal differences over the section in the pattern of changes in the θ-S relationship as well as in AOU. Although it was argued that changes in the intermediate layer of the central part of the section could not be caused by changes in advection velocity from the subarctic, it is not clearly known whether the observed pattern is only due to property changes of CMW and NPIW or due to changes in the circulation patterns of CMW and NPIW. In particular, more detailed information about NPIW modification and transport processes is needed.

[42] Changes in AOU and silicate concentrations in the deep layer are too small to be detected because of their low traceability. Although the changes may have biases resulting from the analysis of them, the zonal pattern of change is not negligible. There was a relatively large AOU increase from 140°E–150°E, and silicate concentrations increased to the west of Hawaii and decreased to the east of Hawaii (Figure 15). Robbins and Bryden [1994] suggested that transport of oxygen and nutrients along this section is affected by the relationship between the deep anticyclonic gyre and along-section variations in concentrations. The changes we observed may reflect changes in the anticyclonic circulation. This hypothesis appears especially relevant for silicate transport, with a clear silicate maximum that originates from the subarctic gyre in the eastern deep layer [e.g., Talley and Joyce, 1992] and a large east-to-west difference along neutral density surfaces. A numerical model suggests that the deep anticyclonic circulation is caused by vertical velocity convergence [Ishizaki, 1994]. Therefore the large decrease of transport in the bottom layer (see Figure 13b) may indicate a weakening of this convergence resulting in a reduced deep anticyclonic circulation, which could explain the observed changes in the pattern of silicate concentrations. However, the timescale of the changes in transports in this study is unknown, since the changes were revealed through comparisons of the section at only two time points. Using repeat sections in the Atlantic near 24°N, interannual variability of the overturn could not be conclusively detected [Cunningham et al., 2007]. Furthermore, no statistically significant trend was found in the North Pacific by data assimilation [Wunsch and Heimbach, 2006]. Thus we plan to carry out an assimilation experiment for near-bottom water property changes to investigate the variability of deep transports and the relationships between circulation changes and water property changes.

Figure 15.

Vertical section of changes (2005 − 1985) in AOU (μmol kg−1) and silicate (μmol kg−1) on neutral-density surfaces in the deep layer along 24°N across the Pacific.

[43] Temperature transports for the sections in 1985 and 2005, estimated using inverse methods, were 0.42 PW and 0.31 PW, respectively; smaller than the estimate of Bryden et al. [1991] but similar to the inverse estimate by Ganachaud and Wunsch [2000]. The estimated temperature transport may be smaller than the result of Bryden et al. [1991] since the inverse methods constrain the meridional overturn, and the small difference from Ganachaud and Wunsch [2000] may be due to the wind stress data used or to differences in the chosen constraints. The estimates from the present study are almost within the range of uncertainty of the inverse method used by Ganachaud and Wunsch [2000], and our estimates may have similar uncertainties.


[44] The authors are grateful to the captain and the crews of R/V Mirai. We thank Y. Kawai, K. Katsumata, Y. Kumamoto, and A. Nagano for their useful comments and discussions. Detailed comments made by the anonymous reviewers helped improve an early version of the manuscript.