## 1. Introduction

In Katsumata (2016; hereinafter Part I), it was shown that temperature and salinity profiles and trajectory data from an Argo float can provide estimates of water velocity and density at the parking depth (typically 1000 dbar). With an appropriate definition of a mean field, eddies can be defined as the deviations from the mean, and their contribution to upper-ocean general circulation can be calculated. Part I examined eddy mass transport in the upper ocean in detail and found strong eddy transport along the Northern Hemisphere western boundary currents, in the southern Indian Ocean, and in the Southern Ocean.

Eddies play important roles in shaping large-scale circulation, both zonal and meridional, particularly in the Southern Ocean. The roles include diffusion of momentum and tracers (e.g., Klocker et al. 2012), meridional overturning circulation without a large-scale zonal pressure gradient [see Marshall and Speer (2012) for a review], and winter subduction (e.g., Sallée et al. 2010). In this Part II, we focus on two metrics of eddy activity in the Southern Ocean: form stress and length of streamlines.

Another quantity of interest is the length of streamlines around standing meanders. Using the equivalent barotropic nature of the Antarctic Circumpolar Current (ACC), Hughes (2005), in one of the rare observational works that are not regional, studied the vorticity balance in the Southern Ocean by using surface velocity observations from drifters. He demonstrated that these standing meanders can be explained as nondivergent barotropic Rossby waves trapped by the eastward flow of the ACC. Thompson and Naveira Garabato (2014) studied standing meanders in an eddy-resolving numerical simulation and found a link between the length of buoyancy contours and barotropic adjustment to increased wind forcing through vertical velocity. The lengthening can be interpreted as an increase in amplitude and/or wavelength of the trapped Rossby wave to accommodate changes in advection in response to the increase of wind forcing. Abernathey and Cessi (2014) also found lengthening of buoyancy contours with increased wind forcing in an idealized model of the ACC over a meridional ridge. In section 4, we investigated whether this perturbation to streamline length can be found in the eddy field estimated from the Argo data.

We note that because the ACC is generally equivalent barotropic (Killworth 1992; Kim and Orsi 2014) and in geostrophic balance, the buoyancy contours at constant depth are generally parallel to the streamlines, although recently some local departures from equivalent barotropic structure have been reported (Graham et al. 2012; Phillips and Bindoff 2014). Our discussion is necessarily qualitative because of the limitation (i.e., density) of the data, but we argue that eddies around seven major standing meanders tend to stretch the streamlines, particularly on the eastern flanks of meridional barriers. We summarize the discussion in section 5.

## 2. Data

Online locations of the data are listed in the acknowledgments section of this paper.

### a. Argo data

From two consecutive Argo float hydrographic profiles, an estimate of the velocity at the parking depth (mostly 1000 dbar) at the midpoint of the trajectory between the two profiles can be calculated. The estimated velocity is the displacement between the two surface positions divided by the surfacing time interval (mostly 10 days; Davis 1998, 2005). The estimated temperature and salinity profile at the midpoint between the two profiles is equated to the average of the two profiles. In the region south of 30°S, 253 990 such pairs of velocity and density estimates were calculated using the December 2015 version of the trajectory data, YoMaHa (Lebedev et al. 2007), and the Argo hydrography dataset. The uncertainty in eddy statistics has two components: the measurement uncertainty in each datum owing to uncertainties in the measurement technique and the sampling (or statistical) uncertainty because the number of data is not infinity. The measurement uncertainties for the velocity and density were 0.5 cm s^{−1} and 0.020 kg m^{−3}, respectively. Sampling uncertainty was estimated using the bootstrap method (Efron and Gong 1983) at a confidence interval of 68%. The reader is referred to Part I (section 2a) for more details of the Argo data processing and uncertainty estimates.

We estimated eddy statistics on a grid. A circle with a radius of 150 km was used to define a mean field centered on a grid point. This choice was made to separate two distinct length scales in the ACC dynamics, namely, standing meander scales of 300–500 km and transient eddy scales of 10–20 km (Hughes 2005). At the same time, we needed to have sufficient data density to carry out the quantitative analysis in section 3 and to resolve the spatial structure of eddy fluxes in section 4. In accordance with Davis (2005), the radius of the averaging circle was adjusted by a factor of *H*_{1} and *H*_{2} are the water depths on the circle and at the center satisfying *H*_{1} ≥ *H*_{2} > 0. These two depths were used to account for the longer correlation scale in the long-isobath direction than in the cross-isobath direction. The number of data was not sufficient for temporal analysis, and we averaged all data from July 1997 (first trajectory data) to December 2015 within the circle. Based on this averaging, we defined an eddy as a perturbation with a length scale less than approximately 300 km and a time scale shorter than decadal. This definition includes both transient and standing components, which could have been clearly separated in the output of a numerical model. The grid interval, which was 1° in latitude and longitude, was shorter than the averaging radius, and data from neighboring grids were statistically correlated. When the sampling error was calculated (e.g., in Fig. 5), this correlation was accounted for by reducing the degrees of freedom by the ratio of grid area to averaging area.

*n*= 1, 2, …,

*N*of velocity data

**u**

_{n}were found within an averaging circle centered at a grid point (

*i*,

*j*). The mean velocity at (

*i*,

*j*) is

### b. Satellite altimeter data

As discussed in the introduction, the sea surface height contour is a good proxy for the streamlines of the ACC (Killworth 1992; Kim and Orsi 2014). Following Sokolov and Rintoul (2007), sea level height anomalies from the AVISO (delayed time, merged gridded) products were added to the climatological dynamic heights referenced to 2500 dbar. The daily altimeter data were averaged from 1 January 2000 to 31 December 2015. The dynamic height was calculated from the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Atlas of Regional Seas 2009 (CARS2009) climatology (Ridgway et al. 2002) by using the International Thermodynamic Equation of Seawater—2010 (TEOS-10; IOC et al. 2010).

### c. Climatological transport in the upper 1000 dbar

Climatological transport in the upper 1000 dbar was calculated from geostrophic velocities estimated from the climatological absolute pressure field *p*_{A}. The pressure *p*_{A} is the sum of the dynamic height referenced to 1000 dbar and the geostrophic pressure at 1000 dbar. The former was calculated from the temperature and salinity in the CARS2009 climatological data (Ridgway et al. 2002). The latter was a grid product G-YoMaHa estimated with an optimal interpolation method constrained by the Argo drift data (Katsumata and Yoshinari 2010).

### d. Atmospheric reanalysis

Sea surface wind stress was obtained from the ERA-Interim synoptic monthly means (Dee et al. 2011) and further averaged from January 2000 to December 2015.

## 3. Form stress at 1000-dbar depth

### a. Formulation

We employ the temporal residual mean (TRM) formulation (McDougall and McIntosh 2001) to separate the mean and eddy. In the TRM averaging, the basic nature of oceanic flows that adiabatic flow follows neutral density surfaces is preserved such that eddy flux across a mean neutral surface appears only when there are genuine diabatic processes. As a result, the total (mean plus eddy induced) advecting velocity is the same for all tracers and momentum. Under simple Eulerian averaging, in contrast, the advecting velocity is generally different for different tracers.

*f*is the Coriolis parameter,

*p*is the zonal pressure gradient,

_{x}*ρ*

_{0}is a typical density (1025 kg m

^{−3}), and we have neglected the third- and higher-order terms in the perturbation amplitude [

*O*(

*α*

^{3}) in McDougall and McIntosh 2001]. Here, the TRM velocity

**V**= (

*u*,

*υ*) is the horizontal component of the three-dimensional velocity

**U**= (

*u*,

*υ*,

*w*), and

*z*

_{a}when averaged over time. We used an approximate form of (5) correct to

*O*(

*α*

^{3}):

*ρ*is locally referenced potential density with subscripts denoting derivatives, and the isopycnal and diapycnal diffusivities are

*A*and

*εA*, respectively.

*H*(=1000 dbar) and the sea surface,

^{y}(

*z*= 0) = 0 was used (McDougall and McIntosh 2001, their section 8). The second term on the right-hand side of (8) is the eddy pressure perturbation multiplying the interfacial slope

*p*′ > 0 pushing an isopycnal interface shoaling eastward

*N*data of meridional velocity

*V*

_{n}and isopycnal depth

*z*

_{an}around a grid point as

### b. Form stress at 1000 dbar

The distribution of the form stress [(9)] is shown in Fig. 2. Several in situ observations are available for comparison with the interfacial form stress estimated with the Argo float data. Phillips and Rintoul (2000) measured velocity and temperature with four moorings near 50.5°S, 143.0°E from March 1993 to January 1995. The interfacial form stress at 1150 dbar was 0.51 N m^{−2}. This point measurement does not agree with the nearest grid value (−0.08 N m^{−2} at 50.5°S, 143.5°E) but is similar in magnitude to the grid values slightly upstream (0.17 N m^{−2} at 51.5°S, 142.5°E and 0.97 N m^{−2} at 51.5°S, 141.5°E). The difference might be due to the different averaging methods used, as suggested by the fact that using a 300-km averaging radius resulted in much better agreement (Part I).

Spatial distribution of zonal form stress [(9)] at 1000 dbar. The dashed lines show latitudinal circles at 56.5° and 62.5°S. Positive stress indicates a downward flux of eastward momentum. Form stresses with magnitude less than *ρ*_{0}*f*Δ*h*Δ*u* were not plotted, where Δ*h* and Δ*u* are the uncertainties in heave and in drift velocity (section 2a), respectively.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of zonal form stress [(9)] at 1000 dbar. The dashed lines show latitudinal circles at 56.5° and 62.5°S. Positive stress indicates a downward flux of eastward momentum. Form stresses with magnitude less than *ρ*_{0}*f*Δ*h*Δ*u* were not plotted, where Δ*h* and Δ*u* are the uncertainties in heave and in drift velocity (section 2a), respectively.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of zonal form stress [(9)] at 1000 dbar. The dashed lines show latitudinal circles at 56.5° and 62.5°S. Positive stress indicates a downward flux of eastward momentum. Form stresses with magnitude less than *ρ*_{0}*f*Δ*h*Δ*u* were not plotted, where Δ*h* and Δ*u* are the uncertainties in heave and in drift velocity (section 2a), respectively.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Strong form stress was localized to steep topographies such as the Drake Passage (60°W), Kerguelen Plateau (80°E), Macquarie Ridge (165°E), and Pacific Antarctic Ridge (150°W). This localization is similar to the zonal distribution of particle crossing (Thompson and Sallée 2012, their Fig. 3d) and the “shallow” topographic form stress in the state estimate model (Masich et al. 2015, their Fig. 6), although particle crossing and form stress are physically different quantities. That the form stress discussed in Masich et al. (2015) is a topographic form stress, not interfacial form stress, explains the difference between the significant contribution found at 1000 dbar here and the values that are nearly zero in their Fig. 5.

### c. Zonal momentum balance in a zonal band across the Drake Passage

*y*

_{S}) and 56.5°S (=

*y*

_{N}):

^{11}N. Roughly half, (8.1 ± 1.9) × 10

^{11}N, of this momentum was transferred downward across the 1000-dbar surface.

Zonal distribution of the form stress integrated between 56.5° and 62.5°S and in the upper 1000 dbar. The form stress was first integrated at each longitude and accumulated eastward from 0°, so that the value at 360° longitude was the form stress integrated in the box (8.1 × 10^{11} N). When integrated, the uncertainty due to measurement error was small (0.5 × 10^{11} N) compared to the uncertainty due to sampling (1.9 × 10^{11} N). The dotted line shows the similarly accumulated zonal wind stress (17.1 × 10^{11} N when integrated). The dashed–dotted line shows the planetary vorticity advection by the geostrophic component of the meridional velocity ^{11} N, unscaled, when integrated). The bottom topography meridionally averaged at each longitude is also shown (Kerguelen = Kerguelen Plateau, S.E.Ind. = South East Indian Ridge, Pac.A. = Pacific Antarctic Ridge, and Drake = Drake Passage).

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Zonal distribution of the form stress integrated between 56.5° and 62.5°S and in the upper 1000 dbar. The form stress was first integrated at each longitude and accumulated eastward from 0°, so that the value at 360° longitude was the form stress integrated in the box (8.1 × 10^{11} N). When integrated, the uncertainty due to measurement error was small (0.5 × 10^{11} N) compared to the uncertainty due to sampling (1.9 × 10^{11} N). The dotted line shows the similarly accumulated zonal wind stress (17.1 × 10^{11} N when integrated). The dashed–dotted line shows the planetary vorticity advection by the geostrophic component of the meridional velocity ^{11} N, unscaled, when integrated). The bottom topography meridionally averaged at each longitude is also shown (Kerguelen = Kerguelen Plateau, S.E.Ind. = South East Indian Ridge, Pac.A. = Pacific Antarctic Ridge, and Drake = Drake Passage).

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Zonal distribution of the form stress integrated between 56.5° and 62.5°S and in the upper 1000 dbar. The form stress was first integrated at each longitude and accumulated eastward from 0°, so that the value at 360° longitude was the form stress integrated in the box (8.1 × 10^{11} N). When integrated, the uncertainty due to measurement error was small (0.5 × 10^{11} N) compared to the uncertainty due to sampling (1.9 × 10^{11} N). The dotted line shows the similarly accumulated zonal wind stress (17.1 × 10^{11} N when integrated). The dashed–dotted line shows the planetary vorticity advection by the geostrophic component of the meridional velocity ^{11} N, unscaled, when integrated). The bottom topography meridionally averaged at each longitude is also shown (Kerguelen = Kerguelen Plateau, S.E.Ind. = South East Indian Ridge, Pac.A. = Pacific Antarctic Ridge, and Drake = Drake Passage).

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

*H*

_{E}is the depth of the bottom of the Ekman layer (0 <

*H*

_{E}< 1000 dbar). The eddy transport is defined by

That the sampling error was not negligible underscored the limitation imposed by the data density. However, the measurement error was an order of magnitude smaller than the integrated form stress.

A comparison between the wind and form stresses revealed the difference of their spatial scales: the wind stress increased smoothly, whereas the form stress increased at much smaller spatial scales concentrated near the topographic barriers. A closer look showed that the form stress “step” was located on the eastern flanks of the meridional ridges.

The ACC tends to surmount a topographic ridge by first deflecting north on the western flank and then back to the south on the eastern flank (Fig. 4), a pattern reproduced in simple models (Ward and Hogg 2011; Abernathey and Cessi 2014) and consistent with planetary potential vorticity *f*/*H* conservation in the Southern Hemisphere. This behavior explains the dashed–dotted line *ρf* in Fig. 3. Decreases of the planetary vorticity were found on the western flanks of topographic ridges and increases were found on the eastern sides (note

*f*< 0 here). As expected from (13), the integrated geostrophic Coriolis term was much smaller (0.7 × 10

^{11}N) than the other two terms, downward flux of eastward momentum [(8.1 ± 1.9) × 10

^{11}N] and wind input of eastward momentum (17.1 × 10

^{11}N), in agreement with the balance expressed by (13).

Interface height of a two-layer, quasigeostrophic analytical model of a wind-driven current over a meridional ridge [after Abernathey and Cessi (2014, their Fig. 10) © American Meteorological Society. Used with permission.].

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Interface height of a two-layer, quasigeostrophic analytical model of a wind-driven current over a meridional ridge [after Abernathey and Cessi (2014, their Fig. 10) © American Meteorological Society. Used with permission.].

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Interface height of a two-layer, quasigeostrophic analytical model of a wind-driven current over a meridional ridge [after Abernathey and Cessi (2014, their Fig. 10) © American Meteorological Society. Used with permission.].

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

The extent of the localization was examined by sorting grid values of the form stress (Fig. 5). Approximately the top 7% (1 × 10^{12} m^{2}) of the total area contributed 7.2 × 10^{11} N or 90% of the overall form stress (8.1 × 10^{11} N). Half of the total form stress (4.0 × 10^{11} N) was accounted for by only 2% (0.2 × 10^{12} m^{2}) of the total area.

Form stress accumulated after sorting. Estimated form stress in each grid was sorted in ascending order and integrated: 6.4 × 10^{12} m^{2} of grids had negative values and 8.9 × 10^{12} m^{2} of grids had positive values. These comprised 41% and 59% of the total area, respectively, between 56.5° and 62.5°S. Negative grids integrated to a form stress of −7.8 × 10^{11} N with sampling error estimated by the bootstrap method of (1.3/−1.1) × 10^{11}N. From this negative contribution, −2.2 × 10^{11} N came from those grids with values less than the observation error of Δ*h*Δ*υ*. Positive grids integrated to a form stress of 15.9 × 10^{11} N with sampling error estimated by the bootstrap method of (1.4/−1.6) × 10^{11} N. From this positive contribution, 2.7 × 10^{11} N came from those grids with values less than the measurement error of Δ*h*Δ*υ*. Overall form stress was 8.1 × 10^{11} N with sampling error of ±1.9 × 10^{11} N and measurement error of 0.5 × 10^{11} N.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Form stress accumulated after sorting. Estimated form stress in each grid was sorted in ascending order and integrated: 6.4 × 10^{12} m^{2} of grids had negative values and 8.9 × 10^{12} m^{2} of grids had positive values. These comprised 41% and 59% of the total area, respectively, between 56.5° and 62.5°S. Negative grids integrated to a form stress of −7.8 × 10^{11} N with sampling error estimated by the bootstrap method of (1.3/−1.1) × 10^{11}N. From this negative contribution, −2.2 × 10^{11} N came from those grids with values less than the observation error of Δ*h*Δ*υ*. Positive grids integrated to a form stress of 15.9 × 10^{11} N with sampling error estimated by the bootstrap method of (1.4/−1.6) × 10^{11} N. From this positive contribution, 2.7 × 10^{11} N came from those grids with values less than the measurement error of Δ*h*Δ*υ*. Overall form stress was 8.1 × 10^{11} N with sampling error of ±1.9 × 10^{11} N and measurement error of 0.5 × 10^{11} N.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Form stress accumulated after sorting. Estimated form stress in each grid was sorted in ascending order and integrated: 6.4 × 10^{12} m^{2} of grids had negative values and 8.9 × 10^{12} m^{2} of grids had positive values. These comprised 41% and 59% of the total area, respectively, between 56.5° and 62.5°S. Negative grids integrated to a form stress of −7.8 × 10^{11} N with sampling error estimated by the bootstrap method of (1.3/−1.1) × 10^{11}N. From this negative contribution, −2.2 × 10^{11} N came from those grids with values less than the observation error of Δ*h*Δ*υ*. Positive grids integrated to a form stress of 15.9 × 10^{11} N with sampling error estimated by the bootstrap method of (1.4/−1.6) × 10^{11} N. From this positive contribution, 2.7 × 10^{11} N came from those grids with values less than the measurement error of Δ*h*Δ*υ*. Overall form stress was 8.1 × 10^{11} N with sampling error of ±1.9 × 10^{11} N and measurement error of 0.5 × 10^{11} N.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

### d. Discussion

*z*= −

*B*(

*x*,

*y*):

_{t}→ 0), the two terms on the right-hand side balance. This balance in the entire water column is different from the balance in the upper 1000 dbar [(10)] in three ways: (i) the Coriolis term does not exist because there is no net meridional transport

^{y}= 0 at

*z*= 0 and −

*B*(McDougall and McIntosh 2001, their section 8).

Johnson and Bryden (1989) first estimated the magnitude of zonal momentum transferred vertically through a water column from temperature and velocity measurements from moorings in the Drake Passage. They found that the momentum flux was similar in magnitude to the wind stress and of the right sign to transfer eastward momentum downward. Quantitative discussion of the balance was not possible until the distribution of each term in the zonal momentum equation was estimated by numerical simulation. Marshall et al. (1993) used output from an eddy-resolving, quasigeostrophic, four-layer numerical simulation of the Southern Ocean to find a balance among the wind input, downward flux of momentum through each of the interfaces, and bottom topographic form stress. In that model, the northern boundary at 40°S was closed by a solid wall such that (15) held.

This constraint was removed by the Fine Resolution Antarctic Model (FRAM), which had an open northern boundary (Stevens and Ivchenko 1997). In the surface layer, the Coriolis and wind stress terms balanced. The balance was among the advection, Coriolis, and dissipation terms in the intermediate layers, where these three terms were at least an order of magnitude smaller than the Coriolis term in the surface and bottom layers that balanced the wind stress term and the bottom form stress term, respectively. This result corresponds to our result (Fig. 3) where (11) is the near-surface balance and (13) represents the balance of the intermediate layers. Keep in mind that we neglected the advection and dissipation terms because they were small. Stevens and Ivchenko (1997) identified the difference between the wind input and downward transfer of momentum as the planetary vorticity carried by the residual circulation [their (33)]. In their model, the contribution of the residual circulation was small around 1000 m (their Fig. 10), but our results showed that the Coriolis term that corresponds to their residual circulation term was as much as roughly half of the total wind input. This difference is not unexpected considering the different definitions of eddies and the heavy smoothing in our analysis.

Warren et al. (1996) criticized the form drag concept as “obscurantist physics” because the mass balance (15) is sufficient to assure the top-to-bottom momentum balance (14). Their formulation did not involve averaging and eddy mean decomposition, the result being that the eddy form stress was implicit. Use of the eddy explicit formulation (8) makes it apparent that the eddy form stress is not negligible in the upper-layer momentum balance, but it does not account for all momentum transfer from the surface to the bottom. The rest of the transfer is explained by the Coriolis term *y*) nor a storage form (i.e., ∂/∂*t*), it does not import/export zonal momentum across the integrating boundaries nor does it store the momentum. Consequently, there is some arbitrariness in its interpretation. The situation is confounded by the definition of the eddy form stress equations [(9) and (12)]: form stress is equal to vorticity carried by meridional eddy transport. Thus, it is possible to account for the eddy form stress as a term in the upper momentum balance equation [(10)] or in the mass balance equation [(15)]. If we consider (10) as the key equation, our result was that the zonal momentum was transferred downward by eddy form stress, and the magnitude of the zonal momentum transferred was approximately half the wind input. In contrast, the result can be paraphrased in terms of (15) that the meridional transport in the box consisted of southward eddy transport equal to approximately half of the northward surface Ekman transport.

Our eddy explicit analysis gives a different answer to another question raised by Warren et al. (1996) regarding whether the ACC is relevant to the integrated zonal momentum balance or not. Figure 3 shows that the eddy form stress was enhanced on the lee side of the meridional ridges, which can be considered as an indirect effect of the ACC in the zonal momentum balance through generation of the eddies. We also note that confirmation of a steady balance is a different question from establishing a steady state. To fully understand the role of the ACC in the zonal momentum balance, it seems necessary to study a time-dependent problem, as was done by Ward and Hogg (2011). The result that the Coriolis term associated with the meridional transport is an important term in the momentum budget suggests that a closed northern boundary might lead to behavior of the modeled ACC quite different from that of the real ACC.

## 4. Streamline length at stationary meanders

### a. Formulation

*l*of a segment of a streamline around a point spanning a small azimuth of Δ

*θ*(Fig. 6) is related to the radius of curvature

*R*by

*R*as a substitute for

*l*.

Streamlines (thick lines) are given as *F*(*x*, *y*) = const. A local circle (dashed line) sharing a point and a tangent with the streamline defines the radius of curvature *R*. Length Δ*l* of an infinitesimal segment of the streamline is given by Δ*l* = *R*Δ*θ*.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Streamlines (thick lines) are given as *F*(*x*, *y*) = const. A local circle (dashed line) sharing a point and a tangent with the streamline defines the radius of curvature *R*. Length Δ*l* of an infinitesimal segment of the streamline is given by Δ*l* = *R*Δ*θ*.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Streamlines (thick lines) are given as *F*(*x*, *y*) = const. A local circle (dashed line) sharing a point and a tangent with the streamline defines the radius of curvature *R*. Length Δ*l* of an infinitesimal segment of the streamline is given by Δ*l* = *R*Δ*θ*.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

*F*(

*x*,

*y*) = const is given by

*F*is the streamfunction for the geostrophic transport in the top 1000 dbar such that

*u*and

*υ*are the transport per unit width in the top 1000 dbar in the zonal and meridional directions, respectively. Equation (17) becomes

*f*/

*β*=

*f*/(∂

*f*/∂

*y*) is assumed to be much longer than the typical meridional scales of standing meanders. As explained in section 2c, (18) can be evaluated using a climatology for the absolute geostrophic velocity field in the upper 1000 dbar. A perturbation to the curvature is

*u*, Δ

*υ*, Δ

*υ*

_{x}, Δ

*u*

_{y}, and Δ

*υ*

_{y}) as the eddy transport observed by Argo floats. The derivatives Δ

*υ*

_{x}, Δ

*u*

_{y}, and Δ

*υ*

_{y}were estimated by finite differencing. This means, as emphasized previously in section 2a, that the perturbation [(19)] measures the effect of both transient and standing eddies with a typical length scale of less than 300 km.

### b. Standing meanders and curvature of streamlines

An analytical solution (Abernathey and Cessi 2014) demonstrates that an eastward-flowing ACC surmounts a meridional ridge first by deflecting northward on the western flank and then gradually relaxing on the eastern flank (Fig. 4). In terms of curvature, this behavior is manifested as a zonal dipole with a negative (counterclockwise, anticyclonic) curvature on the western side and a positive (clockwise, cyclonic) one on the east. Seven such structures (Table 1) were identified in the Southern Ocean (Fig. 7). The magnitude and spatial distribution of curvature varied, but all positive peaks corresponded well to the topographic barriers shallower than 3000 m. From the definition (16), a measure of the magnitude of curvature in the box would be *κ* was calculated within the boxes (Table 1). The average was somewhat sensitive to the choice of the upper bound of uncertainty but was generally positive; a result consistent with the fact that overall the ACC is a cyclonic circulation.

Topographic barriers. Extent of boxes shown in Figs. 7 and 8, and averages of curvature (Fig. 7) and perturbation (Fig. 8) are shown. In the averaging, grid values of eddy transport in the upper 1000 dbar smaller than the measurement uncertainty Δ*h*Δ*u* (section 2a) were not used. Moreover, grid values smaller than 1 and 2 m^{2} s^{−1} were not used in calculating the values in the columns under >1 and >2 m^{2} s^{−1}, respectively.

Spatial distribution of curvature [(18)] of streamfunctions of geostrophic flow field in the upper 1000 dbar. Thick contours show depth contour of 3000 m. Thin contours show sea surface height representing fronts, as in Fig. 1. Color shading indicates curvature calculated by (18) from the geostrophic transport in the top 1000 dbar. Positive curvature is cyclonic or clockwise. Green boxes designate seven topographic barriers that ACC surmounts.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of curvature [(18)] of streamfunctions of geostrophic flow field in the upper 1000 dbar. Thick contours show depth contour of 3000 m. Thin contours show sea surface height representing fronts, as in Fig. 1. Color shading indicates curvature calculated by (18) from the geostrophic transport in the top 1000 dbar. Positive curvature is cyclonic or clockwise. Green boxes designate seven topographic barriers that ACC surmounts.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of curvature [(18)] of streamfunctions of geostrophic flow field in the upper 1000 dbar. Thick contours show depth contour of 3000 m. Thin contours show sea surface height representing fronts, as in Fig. 1. Color shading indicates curvature calculated by (18) from the geostrophic transport in the top 1000 dbar. Positive curvature is cyclonic or clockwise. Green boxes designate seven topographic barriers that ACC surmounts.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

### c. Role of eddies

The role of eddies was calculated by identifying the perturbations in (19) as eddy transport. The result is shown in Fig. 8 after applying a 9-point smoother (3 × 3 grid points with a weight of 0.2 in the middle and of 0.1 at the other points) to suppress noise caused by spatial differentiation. Because of the spatial derivatives, the perturbation had smaller spatial scales than the curvature over the topographic barriers. Equation (19) does not allow for intuitive interpretations, and Fig. 7 shows that the eddies have effects that both increase and decrease the curvature *κ*. A decreased curvature *κ* means an increased *R*, the result being a lengthening of the streamline via (16). The average perturbation within the box was generally negative (Table 1) except for the South Scotia Ridge where missing data around topographies may have compromised the averaging (Fig. 8a). This negative average suggests that in most cases eddies increased the length of streamlines. A closer look shows negative peaks corresponding to the poleward eddy transport on the eastern flanks of the ridges, except for the Conrad Rise, which is more of a seamount than a ridge, the localized structure of which might be the reason for the lack of poleward eddy transport. As discussed in section 3, poleward eddy transport is important to downward transport of eastward momentum. The results in Fig. 8 suggest that poleward eddy transport also tends to lengthen streamlines.

Spatial distribution of perturbation Δ*κ* to curvature [(19)] by eddies indicated by color shading. Gray indicates areas where data are missing. Thin contours show sea surface height representing the fronts as in Fig. 7. Thick contours show depth contour at 3000 m. The region for each panel corresponds to the green boxes in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of perturbation Δ*κ* to curvature [(19)] by eddies indicated by color shading. Gray indicates areas where data are missing. Thin contours show sea surface height representing the fronts as in Fig. 7. Thick contours show depth contour at 3000 m. The region for each panel corresponds to the green boxes in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Spatial distribution of perturbation Δ*κ* to curvature [(19)] by eddies indicated by color shading. Gray indicates areas where data are missing. Thin contours show sea surface height representing the fronts as in Fig. 7. Thick contours show depth contour at 3000 m. The region for each panel corresponds to the green boxes in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

### d. Discussion

As stressed in section 2a, our definition of eddy inevitably included both transient and standing (with length scales less than about 300 km) components. A division between the two is possible in numerical simulations, and the contribution from each component depends on the setting. For example, in a four-layer idealized model of the ACC (Treguier and McWilliams 1990, their Fig. 9), the ratio of transient to standing eddy kinetic energy varied from 3 to 20 depending on the topography. In a three-layer quasigeostrophic model of the ACC (Hogg and Blundell 2006, their Fig. 11), the ratio in the top layer was closer to 2. In an attempt to extract the transient component, we redefined an eddy in a much smaller area than the area used in the main analysis: 0.4° in longitude (about 22 km at 60°S) and 0.1° in latitude (about 11 km). Admittedly, 11–22 km is still comparable to the Rossby deformation radius of the first mode, or typical eddy size, but 11 × 22 km^{2} was the smallest area to obtain a meaningful sample size; 30 grids had at least 5 data points within the standing meander regions defined in Fig. 7. We then calculated eddy kinetic energy as a measure of transient eddy activity. Figure 9 shows this “transient” eddy kinetic energy compared with the eddy kinetic energy (Fig. 1) from the nearest grid of the main analysis. The contribution of the transient component varied, but most of the kinetic energy was transient for strong eddies with kinetic energies >0.01 m^{2} s^{−1}, particularly from the Southeast Indian Ridge box. The sample size was very limited, but if this observation can be extrapolated to other boxes, the stretching of streamlines by eddies apparent in Fig. 7 would seem to imply that transient eddies tend to stretch the streamlines, a conclusion that is consistent with the eddy-resolving simulation of Thompson and Naveira Garabato (2014, their Fig. 11).

Comparison of eddy kinetic energy (abscissa) of approximately transient eddies (transient with spatial scales less than 22 km zonally and 11 km meridionally) and that (ordinate) of eddies used in the main analysis (transient with spatial scales less than 300 km). Colors indicate the box the eddy belongs to, defined in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Comparison of eddy kinetic energy (abscissa) of approximately transient eddies (transient with spatial scales less than 22 km zonally and 11 km meridionally) and that (ordinate) of eddies used in the main analysis (transient with spatial scales less than 300 km). Colors indicate the box the eddy belongs to, defined in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

Comparison of eddy kinetic energy (abscissa) of approximately transient eddies (transient with spatial scales less than 22 km zonally and 11 km meridionally) and that (ordinate) of eddies used in the main analysis (transient with spatial scales less than 300 km). Colors indicate the box the eddy belongs to, defined in Fig. 7.

Citation: Journal of Physical Oceanography 47, 9; 10.1175/JPO-D-17-0072.1

We also note that from 1997 to 2015, the period during which we averaged the Argo data, the westerly wind over the Southern Ocean as measured by the southern annular mode index was increasing (Swart and Fyfe 2012). Therefore, we hypothesize that the results summarized in Fig. 9 represent responses to an increase rather than a decrease of the wind.

## 5. Conclusions

The roles that eddies play in the dynamics of the ACC were studied by using hydrographic and trajectory data from Argo floats. We used two metrics: form stress across the 1000-dbar isopycnal and changes in the curvature of standing meanders. Eddy activities were very localized near major topographic barriers. The Coriolis forces associated with meridional transport and form stress contributed about equally to balancing the wind input in the circumpolar zonal momentum budget at the latitudes of the Drake Passage. The eddies, and particularly the transient components thereof, tended to lengthen the streamlines by increasing the radii of curvature of the standing meanders.

This study was limited by the number of data. If more float data were available southward of the ACC (e.g., Klatt et al. 2007), it would have been possible to discuss the momentum balance south of the ACC, including the subpolar wind-driven gyres. Recent studies have pointed out the importance of the westward wind near Antarctica in controlling ACC dynamics (e.g., Stewart and Thompson 2012; Zika et al. 2013). The data density also limited our study to analysis of mixed transient and standing eddies. The discussion of streamlines was necessarily qualitative because of the lack of spatial resolution.

Despite its importance in both zonal and meridional global ocean circulation, the Southern Ocean is a challenging region for numerical simulations because of the occurrence of subgrid-scale physical phenomena such as eddy mixing, deep-water production, and ice–ocean interactions. Though limited by data density, the present study offers a benchmark against which eddy parameterizations in numerical simulations can be compared for improvements. This remark also applies to the quasi-global analysis presented in Part I.

## Acknowledgments

Argo data were collected and made freely available by the International Argo Program and the national programs that contribute to it (http://www.argo.ucsd.edu; http://argo.jcommops.org). The Argo Program is part of the Global Ocean Observing System. The list of data and their online resource locations are as follows: Argo hydrographic data (snapshot as of December 2015, http://doi.org/10.17882/42182#42348;

The Ssalto/Duacs altimeter products were produced and distributed by the Copernicus Marine and Environment Monitoring Service (CMEMS; http://marine.copernicus.eu/copernicus.eu). Computing software for the International Thermodynamic Equation of Seawater—2010 (TEOS-10) is available online (at http://www.teos-10.org/software.htm). I very much appreciate the community efforts to maintain the availability of these data. This work was supported by the Japan Society for the Promotion of Science (KAKENHI) Grant Number 24540475.

## REFERENCES

Abernathey, R., and P. Cessi, 2014: Topographic enhancement of eddy efficiency in baroclinic equilibration.

,*J. Phys. Oceanogr.***44**, 2107–2126, doi:10.1175/JPO-D-14-0014.1.Davis, R. E., 1998: Preliminary results from directly measuring middepth circulation in the tropical and South Pacific.

,*J. Geophys. Res.***103**, 24 619–24 639, doi:10.1029/98JC01913.Davis, R. E., 2005: Intermediate-depth circulation of the Indian and South Pacific Oceans measured by autonomous floats.

,*J. Phys. Oceanogr.***35**, 683–707, doi:10.1175/JPO2702.1.Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system.

,*Quart. J. Roy. Meteor. Soc.***137**, 553–597, doi:10.1002/qj.828.Efron, B., and G. Gong, 1983: A leisurely look at the bootstrap, the jackknife, and cross-validation.

,*Amer. Stat.***37**, 36–48, doi:10.2307/2685844.Firing, Y. L., T. K. Chereskin, D. R. Watts, and M. R. Mazloff, 2016: Bottom pressure torque and the vorticity balance from observations in Drake Passage.

,*J. Geophys. Res. Oceans***121**, 4282–4302, doi:10.1002/2016JC011682.Graham, R. M., A. M. de Boer, K. J. Heywood, M. R. Chapman, and D. P. Stevens, 2012: Southern Ocean fronts: Controlled by wind or topography?

,*J. Geophys. Res.***117**, C08018, doi:10.1029/2012jc007887.Hogg, A. M. C., and J. R. Blundell, 2006: Interdecadal variability of the Southern Ocean.

,*J. Phys. Oceanogr.***36**, 1626–1645, doi:10.1175/JPO2934.1.Hughes, C. W., 2005: Nonlinear vorticity balance of the Antarctic Circumpolar Current.

,*J. Geophys. Res.***110**, C11008, doi:10.1029/2004jc002753.IOC, SCOR, and IAPSO, 2010: The International Thermodynamic Equation of Seawater—2010: Calculation and use of thermodynamic properties. Intergovernmental Oceanographic Commission, Manuals and Guides 56, 220 pp. [Available online at http://www.teos-10.org/pubs/TEOS-10_Manual.pdf.]

Johnson, G. C., and H. L. Bryden, 1989: On the size of the Antarctic Circumpolar Current.

,*Deep-Sea Res.***36A**, 39–53, doi:10.1016/0198-0149(89)90017-4.Katsumata, K., 2016: Eddies observed by Argo floats. Part I: Eddy transport in the upper 1000 dbar.

,*J. Phys. Oceanogr.***46**, 3471–3486, doi:10.1175/JPO-D-16-0150.1.Katsumata, K., and H. Yoshinari, 2010: Uncertainties in global mapping of Argo drift data at the parking level.

,*J. Oceanogr.***66**, 553–569, doi:10.1007/s10872-010-0046-4.Killworth, P. D., 1992: An equivalent-barotropic mode in the Fine Resolution Antarctic Model.

,*J. Phys. Oceanogr.***22**, 1379–1387, doi:10.1175/1520-0485(1992)022<1379:AEBMIT>2.0.CO;2.Kim, Y. S., and A. H. Orsi, 2014: On the variability of Antarctic Circumpolar Current fronts inferred from 1992–2011 altimetry.

,*J. Phys. Oceanogr.***44**, 3054–3071, doi:10.1175/JPO-D-13-0217.1.Klatt, O., O. Boebel, and E. Fahrbach, 2007: A profiling float’s sense of ice.

,*J. Atmos. Oceanic Technol.***24**, 1301–1308, doi:10.1175/JTECH2026.1.Klocker, A., R. Ferrari, J. H. LaCasce, and S. T. Merrifield, 2012: Reconciling float-based and tracer-based estimates of lateral diffusivities.

,*J. Mar. Res.***70**, 569–602, doi:10.1357/002224012805262743.Lebedev, K., H. Yoshinari, N. A. Maximenko, and P. W. Hacker, 2007: YoMaHa’07: Velocity data assessed from trajectories of Argo floats at parking level and at the sea surface. IPRC Tech. Note 4, 16 pp.

Marshall, J., and K. Speer, 2012: Closure of the meridional overturning circulation through Southern Ocean upwelling.

,*Nat. Geosci.***5**, 171–180, doi:10.1038/ngeo1391.Marshall, J., D. Olbers, H. Ross, and D. Wolf-Gladrow, 1993: Potential vorticity constraints on the dynamics and hydrography of the Southern Ocean.

,*J. Phys. Oceanogr.***23**, 465–487, doi:10.1175/1520-0485(1993)023<0465:PVCOTD>2.0.CO;2.Masich, J., T. K. Chereskin, and M. R. Mazloff, 2015: Topographic form stress in the Southern Ocean State Estimate.

,*J. Geophys. Res. Oceans***120**, 7919–7933, doi:10.1002/2015JC011143.McDougall, T. J., and P. C. McIntosh, 2001: The temporal-residual-mean velocity. Part II: Isopycnal interpretation and the tracer and momentum equations.

,*J. Phys. Oceanogr.***31**, 1222–1246, doi:10.1175/1520-0485(2001)031<1222:TTRMVP>2.0.CO;2.Munk, W. H., and E. Palmén, 1951: Note on the dynamics of the Antarctic Circumpolar Current.

,*Tellus***3**, 53–55, doi:10.3402/tellusa.v3i1.8609.Phillips, H. E., and S. R. Rintoul, 2000: Eddy variability and energetics from direct current measurements in the Antarctic Circumpolar Current south of Australia.

,*J. Phys. Oceanogr.***30**, 3050–3076, doi:10.1175/1520-0485(2000)030<3050:EVAEFD>2.0.CO;2.Phillips, H. E., and N. L. Bindoff, 2014: On the nonequivalent barotropic structure of the Antarctic Circumpolar Current: An observational perspective.

,*J. Geophys. Res. Oceans***119**, 5221–5243, doi:10.1002/2013JC009516.Redi, M. H., 1982: Oceanic isopycnal mixing by coordinate rotation.

,*J. Phys. Oceanogr.***12**, 1154–1158, doi:10.1175/1520-0485(1982)012<1154:OIMBCR>2.0.CO;2.Ridgway, K. R., J. R. Dunn, and J. L. Wilkin, 2002: Ocean interpolation by four-dimensional weighted least squares—Application to the waters around Australasia.

,*J. Atmos. Oceanic Technol.***19**, 1357–1375, doi:10.1175/1520-0426(2002)019<1357:OIBFDW>2.0.CO;2.Sallée, J.-B., K. Speer, S. Rintoul, and S. Wijffels, 2010: Southern Ocean thermocline ventilation.

,*J. Phys. Oceanogr.***40**, 509–529, doi:10.1175/2009JPO4291.1.Sokolov, S., and S. R. Rintoul, 2007: Multiple jets of the Antarctic Circumpolar Current south of Australia.

,*J. Phys. Oceanogr.***37**, 1394–1412, doi:10.1175/JPO3111.1.Stevens, D. P., and V. O. Ivchenko, 1997: The zonal momentum balance in an eddy-resolving general-circulation model of the Southern Ocean.

,*Quart. J. Roy. Meteor. Soc.***123**, 929–951, doi:10.1002/qj.49712354008.Stewart, A. L., and A. F. Thompson, 2012: Sensitivity of the ocean’s deep overturning circulation to easterly Antarctic winds.

,*Geophys. Res. Lett.***39**, L18604, doi:10.1029/2012gl053099.Swart, N. C., and J. C. Fyfe, 2012: Observed and simulated changes in the Southern Hemisphere surface westerly wind-stress.

,*Geophys. Res. Lett.***39**, L16711, doi:10.1029/2012GL052810.Thompson, A. F., and J.-B. Sallée, 2012: Jets and topography: Jet transitions and the impact on transport in the Antarctic Circumpolar Current.

,*J. Phys. Oceanogr.***42**, 956–972, doi:10.1175/JPO-D-11-0135.1.Thompson, A. F., and A. C. Naveira Garabato, 2014: Equilibration of the Antarctic Circumpolar Current by standing meanders.

,*J. Phys. Oceanogr.***44**, 1811–1828, doi:10.1175/JPO-D-13-0163.1.Treguier, A. M., and J. C. McWilliams, 1990: Topographic influences on wind-driven, stratified flow in a

*β*-plane channel: An idealized model for the Antarctic Circumpolar Current.,*J. Phys. Oceanogr.***20**, 321–343, doi:10.1175/1520-0485(1990)020<0321:TIOWDS>2.0.CO;2.Ward, M. L., and A. M. Hogg, 2011: Establishment of momentum balance by form stress in a wind-driven channel.

,*Ocean Modell.***40**, 133–146, doi:10.1016/j.ocemod.2011.08.004.Warren, B. A., J. H. LaCasce, and P. E. Robbins, 1996: On the obscurantist physics of “form drag” in theorizing about the circumpolar current.

,*J. Phys. Oceanogr.***26**, 2297–2301, doi:10.1175/1520-0485(1996)026<2297:OTOPOD>2.0.CO;2.Zika, J. D., J. Le Sommer, C. O. Dufour, A. Naveira-Garabato, and A. Blaker, 2013: Acceleration of the Antarctic Circumpolar Current by wind stress along the coast of Antarctica.

,*J. Phys. Oceanogr.***43**, 2772–2784, doi:10.1175/JPO-D-13-091.1.