Impact of Aleutian Low activity on the STMW formation in the Kuroshio recirculation gyre region



[1] To understand the formation of North Pacific subtropical mode water (STMW) in the Kuroshio recirculation gyre region, the cause of STMW thickness variation is investigated using temperature profiles in a historically archived data set. The thickness variation is predominantly controlled by the main thermocline depth (MTD). When the main thermocline deepens (shoals), the wintertime mixed layer depth can develop (not develop), and consequently, thicker (thinner) STMW is observed in summer. The large-scale atmospheric forcing controlling the MTD is explored using a wind-driven hindcast ocean model. The MTD variation stems primarily from a baroclinic response in the ocean to the Aleutian Low (AL) activity; especially, the meridional movement of the AL exerts a remarkable influence.

1. Introduction

[2] North Pacific subtropical mode water (STMW), characterized by low potential vorticity, is widely distributed in the Kuroshio recirculation gyre region [Masuzawa, 1969; Hanawa, 1987]. STMW is formed in a wintertime mixed layer (ML), the thickness of which is about a few hundred meters [e.g., Suga and Hanawa, 1990]. Numerous authors have studied the wintertime mixed layer depth (MLD) variation, and they have pointed out three major factors causing MLD variation to date. The first factor is the wintertime strong convection by the intense East Asian Wintertime Monsoon [e.g., Suga and Hanawa, 1995; Yasuda and Hanawa, 1997]. The second factor is the upper ocean stratification intensity during the preceding warm season [e.g., Qiu and Chen, 2006]; if the stratification intensity weakens (strengthens), the subsequent winter MLD becomes deeper (shallower). The third factor is a main thermocline depth (MTD) [Uehara et al., 2003], by which deeper ML is formed where the MTD is deeper, in association with anticyclonic eddies.

[3] Minobe [1999] found that a change in magnitude of the Aleutian Low (AL) has a dominant interdecadal (about 20-year) timescale. Since then, it has been reported that the AL influences STMW variation, especially in temperature [e.g., Taneda et al., 2000; Hanawa and Kamada, 2001; Yasuda and Kitamura, 2003; Sugimoto and Hanawa, 2007a]. Recently, Sugimoto and Hanawa [2009] demonstrated an existence of meridional movement of the AL on a decadal (about 10-year) timescale. However, it has not been clarified yet how the meridional movement of the AL affects STMW formation.

[4] In this study, in order to understand the STMW formation, we investigate the STMW thickness variation in the Kuroshio recirculation gyre region, and reveal what mechanisms preferentially control STMW formation. Further, we try to clarify the role of large-scale atmospheric forcing on STMW formation.

2. Data Set and Definition

[5] We use XBT, CTD, and XCTD temperature data archived as the World Ocean Database 2005 (WOD05) [Boyer et al., 2006] and at the Japan Oceanographic Data Center (JODC,, and temperature profiles from Argo floats [Oka et al., 2007]. To capture the STMW adequately, we use the profiles for which the maximum depth is greater than 450 m. To control data quality, we first remove duplicate profiles in the different data sources. For each profile, the measured temperature data are compared with all values measured in the same month within a 1° × 1° box that centers on the observation point; data are excluded if they fall outside of three standard deviations of the mean. Profiles with large temperature inversions (dT/dz < −0.1°C m−1) from the surface to 450 m are also removed. After quality control, temperature profiles are vertically interpolated onto a 1 m interval using a scheme presented by Akima [1970].

[6] We use wind stress curl (WSC) data calculated using the spatial derivative of wind stress of the Japanese Re-Analysis 25 years (JRA-25) [Onogi et al., 2007]. Because small-scale features tend to be exaggerated in the WSC field, we smooth the WSC field using a Gaussian filter with an e-folding scale of 200 km to highlight large-scale variations. Sugimoto and Hanawa [2009] showed that a change in magnitude of the AL is associated with the Pacific/North American (PNA) teleconnection pattern, which has a dominant signal in the central North Pacific in the wintertime pressure field, and a meridional movement of the AL is related to the West Pacific (WP) teleconnection pattern, which has a north–south dipole structure with a boundary at a latitude of about 40°N. In this study, as an indicator of AL activity, we use the PNA and WP indices given as rotated empirical orthogonal functions (REOFs) of monthly 700 hPa geopotential height anomalies based on the methodology by Barnston and Livezey [1987].

[7] We use the satellite-altimetry derived sea surface height (SSH) anomaly data set complied by the AVISO [Ducet et al., 2000]. We removed the steric height signals from the SSH signals according to the procedure by Stammer [1997] using the daily heat flux data of JRA-25.

[8] In this study, STMW is defined as having a vertical temperature gradient of less than 1.5°C/100 m [Hanawa and Suga, 1995] with temperatures of 15°C–21°C [Oka, 2009]. The MLD in winter (February) is computed as the depth at which the temperature takes a 0.5°C lower value than that at 10 m [e.g., Qiu and Chen, 2006]. According to Qiu and Chen [2006], the upper ocean stratification intensity in warm season (July to September) is calculated as the mean value of the vertical temperature gradient calculated at every 1 m from the surface to 200 m. Here 200 m depth is selected as to cover the seasonal thermocline depth change adequately, but to exclude the influence of the main thermocline depth change. We use a 12°C isotherm as an indicator of the MTD because the 12°C isotherm is located at the middle part of the main thermocline in the western part of North Pacific subtropical gyre [Uehara et al., 2003].

[9] The analysis period of this study is 16 years, 1993–2008, because satellite observations started in October 1992. In the correlation analysis, the degrees of freedom are estimated by dividing the data length by the integral time scale calculated according to the method used by Davis [1976]. For that reason, the thresholds of the correlation coefficients obtained from various combinations mutually differ.

3. Results

3.1. Correlation Analysis in the Kuroshio Recirculation Gyre Region

[10] In this paper, the recirculation gyre (RG) is regarded as an area bounded on the west by the Izu ridge (approximately 141°E), on the east by the eastern edge of the quasi-stationary meander of the Kuroshio (150°E) [Mizuno and White, 1983], on the north by the Kuroshio Extension axis, and on the south by 31°N. Although the 15°C isotherm at 200 m is considered to be a good indicator of the KE axis [Kawai, 1972], in this study, we regard the 16°C isotherm as the KE axis in order to minimize the risk of misidentification.

[11] We investigate STMW in the warm season (July–September), because the STMW below the seasonal thermocline is isolated from atmospheric forcing and its water properties are preserved [e.g., Sugimoto and Hanawa, 2007b]. Figure 1a portrays a time series of STMW thickness, which reveals some long-term variation: it is thicker in the mid 2000s and thinner in the late 1990s. There is a high correlation (R = 0.69, the value of which exceeds a 1% significance level) between the mean thickness (black circles in Figure 1a) and the mean MLD in February (black circles in Figure 1b) when the MLDs reach the deepest value [e.g., Suga and Hanawa, 1990].

Figure 1.

Time series of (a) STMW thickness during the warm season (July–September), (b) MLD in February, (c) wintertime (December–February) mean net sea surface heat flux, (d) stratification intensity in the preceding warm season (July–September), and (e) MTD in the cold season (January–March) in the Kuroshio recirculation gyre (RG) region. Gray circles represent values observed in the RG region. Black circles represent mean values and the vertical bars present the standard deviations.

[12] Relations between the MLD and surface cooling (Figure 1c), stratification intensity (Figure 1d), and MTD (Figure 1e) are examined further. As listed in Table 1, the MLD has no significant correlations at a 1% significance level with both surface cooling and stratification intensity, but the MLD has the highest correlation with the MTD (R = 0.65). In order to quantitatively elucidate the impact on the MLD, we perform a multiple regression analysis with MLD as the dependent variable, using three explanatory variables of MTD, surface cooling, and stratification intensity which are normalized by unit standard deviations; three explanatory variables are approximately independent of each other (see Table 1). Consequently, a multiple regression coefficient for the MTD term is largest (17.70); the values of which are −9.76 for surface cooling term and −0.53 for stratification intensity term. These results imply that the development of MLD is predominantly controlled by the MTD.

Table 1. Correlation Coefficients Between MLD in February, Wintertime Surface Cooling or Heat Flux, Stratification Intensity in the Preceding Warm Season, and MTD in the Cold Season in the RGa
 Surface CoolingStratification IntensityMTD
  • a

    Bold type means value exceeding a 1% significance level. Winter is December–February, the warm season is July–September, and the cold season is January–March.

Surface cooling-−0.17−0.10
Stratification intensity -−0.05

3.2. Role of Oceanic Rossby Waves Excited by Aleutian Low Activity

[13] We investigate the behaviors of MTD in the large-scale spatial field. Figure 2a displays MTD anomaly obtained from data set gridded by applying a Gaussian filter with an e-folding scale of 200 km and 5 months. Long-term variations in the western region are obviously found; most signals appear to propagate from the eastern region. This westward propagation feature closely resembles the behavior of the oceanic Rossby wave observed from the smoothed SSH anomalies displayed in Figure 2b; a correlation coefficient between MTD anomaly and SSH anomaly at 145°E is remarkably high (R = 0.88), the value of which exceeds a 1% significance level.

Figure 2.

(a) Longitude-time diagram of MTD anomaly averaged for a zonal band of 31°–34°N, the MTD of which is gridded on 1° (longitude) × 1° (latitude) by applying a Gaussian filter with an e-folding scale of 200 km and 5 months (m). Positive (negative) values mean the deep (shallow) anomaly. (b) As in Figure 2a but for the satellite-altimetry SSH anomaly smoothed by the identical Gaussian filter (cm). (c) As in Figure 2a but for the wind-driven hindcast model (SSH model) (cm).

[14] The results described above show the importance of SSH variations in elucidating the low-frequency variations of the MTD. Using 1½-layer reduced gravity model, which can reproduce large-scale baroclinic ocean responses to surface wind forcing, we evaluate the extent to which large-scale atmospheric forcing explains the observed SSH variation. Under a long-wave approximation, the linear vorticity equation governing the model is

equation image

where h signifies the sea surface height (SSH) anomaly, cR denotes the speed of the long baroclinic Rossby waves, g′ stands for the reduced gravity (2.7 cm s−2), and ρ0 represents the reference density, f is the Coriolis parameter, equation image is the unit vector in the vertical direction, and equation image · × equation image represents the vertical component of the curl of the wind stress vector. By integrating equation (1) from the eastern boundary (xe) along the baroclinic Rossby wave characteristic, we obtain the following equation:

equation image

We use wind stress data to hindcast the h(x, y, t) field from equation (2). Along the eastern boundary, h(xe, y, t) = 0 is assumed [Fu and Qiu, 2002]. The baroclinic Rossby wave speed cR is estimated by AVISO altimeter data as a function of latitude as in the work by Qiu [2003]: cR = 3.83 cm s−1 as a mean value of 31°–34°N.

[15] Figure 2c displays a time-longitude diagram of the modeled SSH. Most SSH signals in the west can be traced from the central North Pacific. The SSH model clearly represents the low-frequency variation and is very similar to the observed SSH signals in Figure 2b; a correlation coefficient at 145°E is 0.63 (the value of which exceeds a 1% significance level) and the model explains about 80% of the variance in the observation.

[16] We examine whether or not the variations are caused by AL activities of two types: meridional movement of the AL (WP forcing) and change in the magnitude of the AL (PNA forcing). Figures 3a and 3b present results obtained using wind stress fields regressed to the WP index and PNA index in respective months: WP-SSH model and PNA-SSH model. The WP-SSH model shows a dominant long-term variation, with behavior resembling that of the SSH model displayed in Figure 2c. The correlation coefficient is 0.52 at 145°E (the value of which exceeds a 1% significance level), and the variance of WP-SSH model accounts for about 42% of the total variance of the SSH model. On the other hand, the PNA-SSH model shows no low-frequency variation and the correlation coefficient with the SSH model is low (R = 0.20 at 145°E). However, the contribution of PNA forcing should not be neglected, because, by adding the PNA-SSH model output to the WP-SSH model output, the correlation coefficient is improved slightly (R = 0.58) and further the variance is about 55% of that of SSH model. These results show that the baroclinic response to the AL activity causes changes in the MTD; especially, the meridional movement of the AL (WP forcing) imparts a remarkable influence.

Figure 3.

As in Figure 2c but for (a) WP-SSH model, (b) PNA-SSH model, and (c) sum of WP-SSH model and PNA-SSH model (cm). See the text for WP-SSH and PNA-SSH models.

4. Summary and Remarks

[17] To elucidate STMW formation, we investigated STMW thickness variations in the RG region using temperature data archived a WOD05 and JODC and temperature profiles from Argo floats. Thickness variations were predominantly associated with MTD variations. The role of large-scale atmospheric forcing on the MTD was explored using a wind-driven hindcast ocean model. Consequently, MTD variations primarily result from the baroclinic response to the AL activity; especially, the meridional movement of the AL (WP forcing).

[18] The STMW temperature is also expected to be associated with the MTD. However, its relation was not significant (not shown here). This fact constitutes evidence supporting results from prior studies showing that the Kuroshio transport controls the STMW temperature [Hanawa and Kamada, 2001; Yasuda and Kitamura, 2003].

[19] The salient contribution of this study is that it reveals the strong influence of the AL meridional movement on STMW formation with some time lag (about three years). The meridional movement of AL has been known to have a dominant decadal (about 10-year) timescale [see Sugimoto and Hanawa, 2009]. We believe that results of this study will be a step toward the understanding of decadal-scale STMW variation.


[20] The authors express their sincere gratitude to the members of the Physical Oceanography Group at Tohoku University for their useful discussion. The authors were financially supported by the Global Center-Of-Excellence (GCOE) Program at Tohoku University. Comments from two anonymous reviewers were particularly helpful for improving this manuscript.