VOLUME 34

JOURNAL OF PHYSICAL OCEANOGRAPHY

JULY 2004

Coastal Oyashio South of Hokkaido, Japan TOKIHIRO KONO Hokkaido Tokai University, Hokkaido, Japan

MICHAEL FOREMAN

AND

PETER CHANDLER

Institute of Ocean Sciences, Sidney, British Columbia, Canada

MAKOTO KASHIWAI Hokkaido National Fisheries Research Institute, Hokkaido, Japan (Manuscript received 11 December 2002, in final form 7 January 2004) ABSTRACT Coastal Oyashio Water (COW), the cold low-salinity water lying along the southeast coast of Hokkaido, sometimes flows into Funka Bay on the southwest coast in winter and spring. Because this water is low in density, a density current called the ‘‘Coastal Oyashio’’ was assumed to carry COW along these coasts. However, water property maps for April 1989, February 1992, February 1994, and February 1989 show that only for the latter month did COW extend as far west as Cape Erimo on the southwest coast. Steady-state flows were calculated using a baroclinic model and the density data collected in April 1989. Model results show that the Oyashio carries COW southwestward along the continental slope southeast of Hokkaido. It bifurcates south of Cape Erimo with a coastal portion turning around the cape but not always reaching the southwest coast because of the density structure along that coast. When the Oyashio crosses isobaths, the joint effect of baroclinicity and bottom relief (JEBAR) causes a barotropic flow to form a countercurrent on the southeast coast and eddies south off Cape Erimo and east off Tsugaru Strait. The countercurrent and the eddies are seen shoreward of the main current and downstream of the maximum JEBAR region. Thus, this model shows that the coastal Oyashio does not flow as a density current along the southeast coast. High-resolution SST maps collected in spring show eddies south of the cape, and confirm the model results. It is suggested that COW could be driven westward along the southwest coast by the northeasterly winds on the southeast coast.

1. Introduction Hokkaido, the northern island of Japan, has a narrow continental shelf and steep slope along its southeast coast (Fig. 1). The Oyashio flows southwestward on this slope as a western boundary current of the subarctic gyre (Kono 1997; Kono and Kawasaki 1997). Less dense water has been found coastward of the Oyashio and the associated currents were assumed to flow southwestward because of this density difference (Sugiura 1956; Ogasawara 1990). This water maintains a lower density throughout the year (Sugiura 1957), specifically being less saline than the Oyashio from March to May and warmer in October and November (Ogasawara 1990). Since its temperature and salinity are minimal and much less than the Oyashio in March, the water Corresponding author address: Dr. Tokihiro Kono, Department of Marine Science and Technology, School of Engineering, Hokkaido Tokai University Minamisawa, Minami-ku, Sapporo, Hokkaido 0058601, Japan. E-mail: [email protected]

q 2004 American Meteorological Society

may originate from melting ice in the Okhotsk Sea (Ohtani 1971; Ogasawara 1990). The water is called Coastal Oyashio Water (COW). The Tsugaru Warm Current flows out of the Japan Sea through Tsugaru Strait between Hokkaido and Honshu. It broadly extends eastward and northward to form an anticyclonic gyre in summer and autumn when its volume transport becomes large (Conlon 1982; Sugimoto and Kawasaki 1984; Kubokawa 1991). As seen from satellite infrared images in autumn, the current outflow pattern varies with a period of 10–20 days, sometimes reaching the southwest coast of Hokkaido (Yasuda et al. 1988). Forced by density differences, the warm saline water flows into Funka Bay (Isoda and Hasegawa 1997). Warm, saline water occupies Funka Bay from October to December and cold, low-salinity water is usually present from March to May (Ohtani 1971, 1979; Isoda and Hasegawa 1997). Cold, low-salinity, and low-density water has been observed along the southwest coast of Hokkaido (Shimizu and Isoda 1999). We hypothesize that this cold, low-salinity water is COW that has been

1477

1478

JOURNAL OF PHYSICAL OCEANOGRAPHY

VOLUME 34

FIG. 1. Bathymetry. Thick lines denote every 500-m isobath above 500 m while broken and dotted lines denote the 200- and 50-m depth contours.

transported from the southeast Hokkaido coast in winter when the Tsugaru Warm Current is weak. There are significant differences in the water properties of the subarctic COW and the Tsugaru Warm Current since the latter originates from the Kuroshio with modifications in the Japan Sea. Therefore, the distribution of COW and its temporal changes can have significant influences on fisheries along the southern coasts of Hokkaido. For example, walleye pollock eggs, which are spawned on the southwest coast of Hokkaido between January and March, are occasionally carried into Funka Bay by westward currents along the southwest coast (Isoda et al. 1998; Kono et al. 1998). Since this flow carries nutrient-rich water as well as the eggs, Funka Bay can be a good nursery ground for the larvae. In this study, horizontal distributions and transport of the COW are investigated using highly resolved maps that show the current patterns off the coasts south of Hokkaido and around Cape Erimo. A steady-state, finite-element model is used to describe coastal flows along the southeast and southwest coasts of Hokkaido.

Results from barotropic and baroclinic versions of this model are compared to examine how the density structure influences the current pattern. We also discuss wind influences on the current. 2. Horizontal distribution and transport path of the Coastal Oyashio water by the baroclinic current Vertical sections crossing the continental shelf and slope southeast of Hokkaido from January to May show both pycnoclines and haloclines that deepen shoreward (Kono and Kawasaki 1997). Though they change slightly from year to year, above the bottom of the pycnocline, temperature, salinity, and density are lower than 1.58– 3.58C, 33.0–33.2 psu, and 26.5 s t . This cold, low-salinity and low-density water is COW. To illustrate the horizontal distribution and transport of COW along the coasts of Hokkaido, we analyze temperature and salinity data observed by the Hokkaido National Fisheries Research Institute, the Hokkaido

JULY 2004

KONO ET AL.

Fisheries Experimental Station (HFES), the Meteorological Agency, and the Iwate Fisheries Experimental Station. Except for the data from HFES (HFES 1997), these data are archived by the Japan Oceanographic Data Center in the Hydrographic and Oceanographic Department of the Japan Coast Guard (see online at http://www.jodc.go.jp/indexpj.html). Temperature, salinity, and density maps at 50-m depth for February and April 1989, February 1992, and February 1994 are shown in Figs. 2–5, respectively. Water colder than 2.08C, less saline than 33.1 psu, and less dense than 26.5 s t , all characteristics typical of COW, are denoted by the shaded areas. Water colder than 28C is coincident with water less than 33.1 psu for all the observation periods except February 1992. These waters are within 100 km, 30– 60 km, and 40–80 km of the southeast coast of Hokkaido for the respective periods February 1989, April 1989, and February 1994. They extend south of Cape Erimo, reaching latitudes of 418–418159N in February 1989 and February 1994, and exist as isolated features at around 418N, 1438409E in April 1989. In addition, the waters continue west from the cape to reach 1418E in February 1989, and can be seen between 1418 and 1428E on the southwest coast as a separate feature from those seen south of the cape in April 1989. In February 1994, COW extends to the west along the southwest coast from the cape but its westward limit is 1428E. In February 1992, water cooler than 28C occupies the coastal region southeast of Hokkaido within 180 km of the coast, while the area less than 33.1 psu is within 50–100 km. We analyze the transport path of COW in April 1989, the period when the horizontal sampling resolution is the highest. Using the density collected in April 1989, the distribution of the potential energy x is calculated as 0 rgz x5 dz, (1) r0 2h where h, r, r 0 , g are the water depth, the fluid density, the reference density, and the gravity, respectively; x corresponds to the transport streamfunction of the baroclinic current from the sea bottom to the surface, assuming no flow on the bottom (Mellor 1996). The contours of x follow the southeast coast of Hokkaido, suggesting a southwestward current along the continental shelf and slope (Fig. 6). These contours, and thus the main current axis, are concentrated on the shelf and slope with depths between 500 and 3000 m. A baroclinic current, the Tsugaru Warm Current (TWC), flows out from Tsugaru Strait and turns southward. These contours also suggest that the southwestward current off southeast Hokkaido bifurcates into two paths: an offshore part that turns south along the continental slope and a coastal part that turns around an anticyclonic eddy south of Cape Erimo, flows north to the coast, and then turns southward to be merged with the TWC. In addition there is also an eddy west of the TWC.

1479

E

FIG. 2. Horizontal distributions of temperature, salinity, and s t at 50-m depth in Feb 1989. Waters lower than 28C, 33.1 psu, and 26.5 kg m 23 are denoted by shaded areas in each map.

1480

JOURNAL OF PHYSICAL OCEANOGRAPHY

FIG. 3. As in Fig. 2 but for Apr 1989.

FIG. 4. As in Fig. 2 but for Feb 1992.

VOLUME 34

JULY 2004

KONO ET AL.

1481

FIG. 6. Distribution of the potential energy x (m 3 s 22 ) anomalies with respect to 28.4 3 10 4 m 3 s 22 . See text for the x calculation. Shaded areas denote the regions where water depths are between 500– 1000 and 2000–5000 m.

FIG. 5. As in Fig. 2 but for Feb 1994.

North of 418N, the southwestward current off southeast Hokkaido and its two bifurcations flow approximately parallel to the isobaths. The bifurcated current along the continental slope turns eastward at around 1438E between 418 and 408N crossing the isobaths. This could be the Oyashio front, the southern edge of the western subarctic gyre (Kono 1997). The contour patterns of this southwestward current and its two bifurcations also seem to correspond to the COW that was identified along the southeast and southwest coasts of Hokkaido and off Cape Erimo. However, this water is not distributed continuously from the southeast coast to the southwest coast as stated previously, suggesting that the coastal Oyashio might not flow steadily around Cape Erimo. This agrees with Kono et al. (1998), who used repeated observations west of 1428309E along the southwest coast between January to March in 1989–94 to show that the location of the western edge of COW can vary over periods of a few weeks. For all observations except February 1992, water less dense than 26.5 s t along the southeast coast of Hokkaido occupies a similar region to the water colder than 28C and less saline than 33.1 psu. However, for all the observational periods, this low-density water extends beyond this region to a larger area from the coast to at least latitudes south of 418309N and longitudes west of 1448E. In some cases, this area covers almost the entire map (February 1992) and all the area west of 1438309 (February 1994). Since both salinity and temperature tend to be high in these low-density areas, there seems to be overriding and/or mixing with the warm and saline water that originated from the Tsugaru Warm Current. Thus, although the COW and Tsugaru Warm Current may have similar densities, they can be distinguished by their particular temperatures and salinities.

1482

JOURNAL OF PHYSICAL OCEANOGRAPHY

VOLUME 34

3. Diagnostic model The previously described observations suggest that the coastal Oyashio bifurcates just south of Cape Erimo. Part of the current flows southward while the other part turns around the cape and flows to the northwest, sometimes extending west of 1428309E in January–March. However, the current pattern is not clear south of Cape Erimo and on the south coast of Hokkaido. We now attempt to clarify a steady-state flow pattern for the coastal Oyashio along these coasts using a three-dimensional baroclinic model that is forced with a density field computed from the observations described above. As we need to determine detailed flow patterns along the south coast of Hokkaido, a region where the bottom topography (as shown in Fig. 1) is complicated, we have chosen a variable-resolution, finite-element method that can capture the detailed bottom topography and provide a smooth representation of the coastline. Based on Lynch and Werner (1987), the linearized three-dimensional shallow-water equations are solved with conventional hydrostatic and Boussinesq assumptions and eddy viscosity closure in the vertical direction. When combined with suitable boundary conditions (described later), a three-dimensional velocity field and surface elevations are computed. The horizontal momentum equation is f 3V2

1 2

] ]V g N 5 2g=h 2 ]z ]z r0

E

0

=r dz

(2)

z

in which r(x, y, z) is the fluid density, r 0 is the reference density, h(x, y) is the free surface elevation, V(x, y, z) is the horizontal velocity with components u and y, h(x, y) is the bathymetric depth, f is the Coriolis vector, N is the vertical eddy viscosity, g is the gravity, (x, y) are the horizontal coordinates, positive eastward and northward, respectively, with (0, 0) at 408N, 1458E, and z is the vertical coordinate, positive upward with z 5 0 at the surface. Boundary conditions are applied as stress at the free surface and the bottom, respectively: N

]V 5 hC ]z

(z 5 0)

(3)

N

]V 5 kV ]z

(z 5 2h),

(4)

and

where hC is the atmospheric forcing and k is a linear bottom stress coefficient. In addition, we have the continuity equation ]W 1 =·V5 0 ]z

(5)

in which W(x, y, z) is the vertical velocity. The model region is bounded by the southern coast of Hokkaido, the northeastern coast of Honshu, and three straight lines connecting Cape Nosappu at

FIG. 7. Model domain with mesh and boundary conditions.

438209N, 1458509E and the Oshika Peninsula at 388159N, 1418309E, and offshore points at 428N, 1478E and 388N, 1468E. The triangular grid (Fig. 7) has 5215 nodes, 9215 elements, and resolution varying between 1.0 and 37 km. The water depth at each node was specified using archived data from the Japan Oceanographic Data Center. Vertically averaged velocities were specified along the northeastern boundary and at the mouth of Tsugaru Strait for the Oyashio and Tsugaru Warm Currents, respectively. Geostrophic velocities for January 1994 were calculated along the observational line near the northern boundary. They were computed with reference to moored current-meter data (Kono and Kawasaki 1997) and then integrated vertically. The integrated velocity crossing the northern boundary was then estimated by assuming no variation along isobaths. The computed smoothed velocities and transports across the continental shelf and slope are shown in Fig. 8. The volume transport of the Tsugaru Warm Current shows a seasonal change having a maximum in autumn and a minimum in spring (Toba et al. 1982; Sugimoto and Kawasaki 1984). The minimum transport of 0.7 3 10 6 m 3 s 21 (Toba et al. 1982) was applied as a boundary condition across the eastern mouth of Tsugaru Strait. Since we do not have a velocity distribution across the mouth, a linear change was assumed with a maximum velocity of 20.15 m s 21 at the deepest point and zero velocity at both sides of the boundary (Fig. 8). In addition, a geostrophic outflow condition is specified on the southern boundary (Naimie and Lynch 1993), in which neither the elevation nor transport are known, but it is assumed that a geostrophic balance exists between the two along this boundary. The three-dimensional density field was specified using observations carried out in April 1989, a period when the sampling was most dense off the south coast of Hokkaido; s t was interpolated using the kriging

JULY 2004

KONO ET AL.

FIG. 8. Averaged velocity and volume transport normal to the (top) northeastern boundary and (bottom) Tsugaru Strait boundary. The x axis is a distance from the southeast coast of Hokkaido and the northeast coast of Honshu, respectively. Transport was computed by integrating vertically averaged velocity out from these coasts. Locations of the boundaries are shown in Fig. 7.

method (Cressie 1990) at each of the following depths: 0, 20, 50, 100, 150, 200, 250, 300, 400, 600, and 800 m. 4. Flow patterns from the diagnostic models Two flow patterns were calculated using barotropic (homogeneous density) and baroclinic (April 1989) forcing and the same boundary conditions. In order to clarify the influence of the density structure and the bottom topography on the current field, wind forcing was not applied in either case. Following Foreman et al. (2000), the vertical eddy viscosity N was set to 0.029 m 2 s 21 . This assumes the formulation N 5 0.5Urms, where Urms , the root-mean-square vertically averaged current speed, was computed to be 0.24 m s 21 from the Kono and Kawasaki (1997) mooring observations. Following Lyard (1997), the linear bottom stress coefficient k was assumed to have the form k 5 0.003U brms , where U brms is the root-mean-square bottom current speed. With U brms 5 0.11 m s 21 this produced a value of 0.0003 m s 21 . Transport streamfunction c and current vectors of vertically averaged velocity are shown in Fig. 9. For the barotropic model and the area south of Hokkaido, the contours of c are almost parallel to the isobaths. In particular, southwestward flow along the continental shelf and slope southeast of Hokkaido bifurcates into two paths south of Cape Erimo. The current along the shelf and shelf break around the cape flows north-

1483

westward along the coast to reach Funka Bay, while the deeper branch continues southward along the slope. For the baroclinic model, the current on the southeast coast of Hokkaido also bifurcates into two paths. These paths are approximately the same as for the barotropic model at depths below 500 m. This implies that for the baroclinic model, these currents are generally controlled by the bathymetry on the continental shelf break and slope. In shallow waters, however, the current pattern is clearly different. The contours of c are concentrated on the continental shelf break and slope (500–2000-m depth) between 1448509 and 1438E southeast of Hokkaido and around 1418509E east off Tsugaru Strait. The contours and vectors of averaged velocity (Figs. 9 and 10) show countercurrents and circulation cells shoreward of these convergence regions. Eddies are seen around 1448209E on the continental shelf southeast of Hokkaido, on the shelf break around 1438E southwest of Cape Erimo, and around 1418409 and 1428159E east of Tsugaru Strait. The TWC that flows out from Tsugaru Strait extends along the southwest coast of Hokkaido, at least as far east as 1428309E. The coastal part of the bifurcated current that turns around the eddy south of the cape flows northwestward and then turns southward and is merged with the TWC. As a result, the baroclinic model current does not reach Funka Bay, unlike the barotropic model current. So the baroclinic model currents differ from the barotropic currents in some regions. To determine if these differences result from the baroclinic current component (e.g., the thermohaline velocity; Mellor 1996), common current patterns are sought in the distribution of potential energy x. As illustrated in the second section, x contours concentrated on the shelf and slope in the depth range from 500 to 3000 m indicate strong currents in this region. Eddies are seen south of Cape Erimo and east of Tsugaru Strait. In addition, the TWC extends along the coast as far east as 1428, then turns southward, and is merged with the coastal part of the bifurcated current. Accordingly, the current intensification southeast of Hokkaido, the formation of the eddies, and the control of the coastal paths of the bifurcated current can be ascribed to the baroclinic current component. Along the continental shelf and the shelf break southeast of Hokkaido, countercurrents are seen for the baroclinic model (Figs. 9 and 10), although the x contours suggest only a southwestward current along this coast (Fig. 6). The eastward and northward countercurrents associated with the eddies south of the cape and east of the strait are not as significant for the x distribution as for the baroclinic model. In addition, a counterclockwise eddy is seen near the southwest coast between 1428209 and 1438E, in which the x contour reaches the coast before turning northwestward along the coast. Since these current patterns are not present in the x distribution, a barotropic current must predominate in

1484

JOURNAL OF PHYSICAL OCEANOGRAPHY

FIG. 9. Transport streamfunction and vertically averaged currents for (top) barotropic model and (bottom) baroclinic model. Thick lines show streamfunction in intervals of 2 3 106 m 3 s 21 while thin lines denote subcontours in 2.5 3 10 5 m 3 s 21 intervals. Shaded areas denote the regions where water depths are between 500 and 1000 m and between 2000 and 5000 m.

VOLUME 34

JULY 2004

KONO ET AL.

1485

FIG. 10. (top) Transport streamfunction and (bottom) s t (kg m23) at 50-m depth both with current vectors at 50-m depth. Location of section shown in Fig. 11 is shown in the upper panel. Contours of the transport streamfunction are the same as in Fig. 9. Note that the vectors shown in the lower panel are denser than those in the upper.

the regions of the eddies and countercurrents for the baroclinic model. It should be noted that the barotropic currents are generated shoreward of the baroclinic current. This generation will be discussed later. As stated in the second section, low-salinity and lowdensity COW is seen along the southeast coast of Hokkaido. Figure 11 shows a vertical section (crossing the shelf and slope between 428519N, 1448E and 42839N, 1448299E) of the density used in the baroclinic model.

(See Fig. 10 for the precise location.) Water less dense than 26.5 s t is seen as a pycnocline along the coast at depths shallower than 75 m. This is COW. COW is denser shoreward, implying a southwestward baroclinic current along the shelf. This density structure along the coast corresponds to Sugiura (1957) and Ogasawara (1990). The baroclinic model showed an intensified southwestward current driven by the density structure along

1486

JOURNAL OF PHYSICAL OCEANOGRAPHY

VOLUME 34

coastal branch carrying COW turns around the eddy south of the cape. However, COW is not seen along the current path that flows northward along the 1428309E meridian after this turn. On the other hand, water colder than 28C and less saline than 33.0 psu is seen along the southwest coast of Hokkaido (Fig. 3), where a current is seen to flow eastward (Fig. 9 and 10). This must be ascribed to the particular snapshot of density that is used in the steady-state model and suggests major temporal changes and strong mixing along the southwest coast. 5. Joint effect of baroclinicity and bottom topography on the Coastal Oyashio

FIG. 11. Vertical section of s t (lines) and velocity component normal to the section (colored areas) from the baroclinic model; s t units are kilograms per cubic meter. Blue areas denote southwestward velocity and red northeastward. The vertical section is located between 428519N, 1448E and 42839N, 1448299E, as shown in Fig. 10.

the continental slope southeast of Hokkaido. In the vertical section crossing this current, the southwestward velocity gradually decreases downward, at least from the surface to 500 m, north of 428409N, and has a maximum at 428249N (Fig. 11). This continental slope current is the Oyashio as described by Kono and Kawasaki (1997). Northeastward (countercurrent) velocities are seen along the continental shelf and the shelf break in the depth range 150–400 m (Fig. 11). At 50-m depth, large southwestward velocities (the Oyashio) are seen to be almost parallel to the transport streamfunction (Fig. 10). To compare the current pattern with the density structure, we will examine the horizontal map (Fig. 10) and the vertical section (Fig. 11). COW is not carried southwestward along the continental shelf but rather is carried by the Oyashio along the continental slope. A vertical shear is not seen below the COW in the Oyashio, implying low density COW has not affected the velocity structure of this current. The Oyashio bifurcates south of Cape Erimo and the

The transport streamfunction field from our baroclinic model shows countercurrents southeast of Hokkaido and eddies south of Cape Erimo, east of Tsugaru Strait, and along the southwest coast (Figs. 9 and 10). The horizontal pressure gradient force due to the density structure forms these eddies and countercurrents shoreward of the baroclinic currents as well as intensifying the current itself. It has been reported that, if a geostrophic transport referenced to the bottom crosses an isobath, a barotropic transport field should arise to restore mass balance (Shaw and Csanady 1983; Huthnance 1984; Csanady 1985; Mertz and Wright 1992). This phenomenon is known as JEBAR (joint effect of baroclinicity and bottom relief ). It has been examined for the Labrador Current (Lazier and Wright 1993), the Tsushima Warm Current (Isobe 1994), and the circulation in the Scotia, Maine, region (Hannah et al. 1996). We will clarify its role in the coastal Oyashio results computed by our baroclinic model. The momentum equation (1) can be separated into equations in the x and y directions as follows: 2fy 2

1 2

E

1 2

E

] ]u ]h g N 5 2g 2 ]z ]z ]z r0

and fu 2

] ]y ]h g N 5 2g 2 ]z ]z ]y r0

0

z

z

0

]r dz ]x

(6)

]r dz. ]y

(7)

We average Eqs. (5) and (6) from the bottom z 5 2h to the surface z 5 h using | h | K h to give k ]h g ] 2 f y 1 c x 2 u B 5 2g 2 h ]x r 0 ]x and k ]h g ] f u 2 c y 1 y B 5 2g 2 h ]y r 0 ]y

E

0

r dz 2

2h

E

0

2h

r dz 2

1 ]x h ]x (8) 1 ]x , h ]y (9)

where u , y and hc x , hc y are x and y components of

JULY 2004

averaged velocity and wind stress hC, respectively, and x, the potential energy, is given by Eq. (1). Since we are computing a mean flow, dh/dt 5 0 and ] ] (hu) 1 (h y ) 5 0. ]x ]y

(10)

Using Eq. (9), ]/]x(8) 2 ]/]y(7) gives the vorticity equation as follows:

1

2 1 2 ] y ] u ]x ]h ]x ]h 1k 1 22 1 2 5 2 . ]x h ]y ]y ]x ]y ]y ]x

f hu

]h21 ]h21 ]c y ]c x 1 hy 1h 2 ]x ]y ]x ]y

[

1487

KONO ET AL.

B

B

]

21

21

(11)

The first, second, and third terms in the left side of Eq. (10) represent the contributions of stretching, wind stress curl, and bottom stress curl to the vorticity. The baroclinic transport relative to the sea bottom, Vg , is calculated as

1

2

1 ]x ]x Vg 5 2 , . f ]y ]x

when a baroclinic current flows toward and away from a coast. He also discussed the more realistic cases when a front is drawn offshore or onshore across a linear bottom slope. For the case of the front being drawn offshore and using the parameters for our coastal Oyashio, we calculate the surface elevation, the velocity field, and the JEBAR term. We assume a straight coast along the y axis with a bottom whose depth h increases linearly in the positive x direction as shown in Fig. 13. There is a front between the waters whose densities are r 0 and r1 . The front is anchored on the bottom at x 5 X(y) and at the surface at x 5 X(y) 1 l. Here, the surface elevation is separated into steric and residual components as

h 5 h1 1 h 2 .

Based on Csanady (1985), the steric elevation h1 is defined as

h1 5 2

(12)

Then the right-hand side of Eq. (10) can be expressed as f Vg · =h 21 , meaning relative vorticity arises when the baroclinic transport referenced to the bottom crosses an isobath. This is the JEBAR effect. When the water column moves downstream, stretching or shrinking occurs to conserve potential vorticity. In Fig. 12, we show the respective distributions of the JEBAR term, the stretching term and the bottom stress term. The JEBAR term is positive and large between 1448159, 145859, and 1458159E along the southeast coast of Hokkaido, south of Cape Erimo, between 1418509 and 1428309E on the southwest coast and east of Tsugaru Strait. In these areas, the baroclinic transport suggested by the x contours flows to the deeper water. The stretching term tends to be positive and has a similar magnitude in the same areas while the bottom stress term is about one-tenth the size of the other two contributions in each of these same regions. So convergence tends to be balanced with the vorticity supplied by the JEBAR effect. Countercurrents are formed at 1438209–1448E on the southeast coast of Hokkaido. The eddies are around 1438E south of Cape Erimo, at 1428209–1438E off the southwest coast of Hokkaido, and around 1418409E east of Tsugaru Strait. All of these features are onshore and downstream of the significant convergence (stretching) areas forced by the JEBAR term as shown in Fig. 12. This phenomenon seems to agree with Csanady (1985) who showed that cross-isobath baroclinic flow is converted into barotropic flow at the slope. Applying his simple model to our coastal Oyashio current, we will now discuss how the eddies are formed. Using a diagnostic model with a surface-to-bottom front and a linear bottom slope, Csanady (1985) calculated the sea surface topography and the velocity field

(13)

E

0

2h

r dz 2 r0

E

`

x

r dh dx. r 0 dx

(14)

Substituting these equations for the elevation into the vertically averaged momentum Eqs. (7) and (8), assuming no wind forcing and ignoring the bottom stress in the x direction, we obtain the momentum equations for two cases: 

]h 2

2 f ]x , 2f y 5 2g ]]xh 2 ghh 2

x,X f

r 0 2 r1 ]h f , r0 ]x

(15) x.X

and 

r 0 2 r1 dH f ]h k 1 g 2 1 y B, r0 dy ]y h

 2f u 5   g ]]yh g

2

1

gh f r 0 2 r1 ]h f k 1 y B, h r0 ]x h

x,X x . X, (16)

where the front is locally at the depth of h f and H f is the anchor depth; that is, H f 5 h(X). We substitute these momentum equations into the continuity Eq. (10) to give k ] 2h2 r 2 r1 ]h dH f ]h ]h2 1 0 1 5 0, 2 f ]x r 0 ]x dy ]x ]y

x,X (17)

and 2

]h ]h2 k ] 2h2 2 5 0, ]x ]y f ]x 2

x . X. (18)

We solve Eqs. (17) and (18) with parameters shown in Table 1 to obtain h 2 as shown in Fig. 14. For this calculation, we apply the boundary conditions, ]h 2 /]x

1488

JOURNAL OF PHYSICAL OCEANOGRAPHY

FIG. 12. Contributions to the vertically integrated vorticity (s 22 ) as computed by the baroclinic model. The JEBAR, stretching, and bottom stress terms are shown in the upper, middle, and lower panels, respectively. Red areas denote positive vorticity and blue negative. Lines denote the subcontours of the associated transport streamfunction in 2.5 3 10 5 m 3 s 21 intervals as shown in Fig. 9.

VOLUME 34

JULY 2004

KONO ET AL.

1489

FIG. 13. Schematic for the simple model in which a surface-tobottom front exists on a linearly sloping bottom.

5 0 at x 5 0 and at large x in comparison with X, and h 2 5 0 at y 5 0, respectively. These conditions mean that the net transport in the y direction from the coast to a large x equals the baroclinic current along the front. Using this h 2 , the transport streamfunction c is calculated by integrating hy out from the coast, assuming X (in kilometers) varies sinusoidally as X 5 15 sin[0.01p(y 2 50)] 1 50. The JEBAR term is also calculated according to Eq. (11) and is shown in Fig. 14 along with the c distribution; h 2 and c largely decrease offshore corresponding to the strong barotropic currents just shoreward of the JEBAR maximum (Fig. 14). The baroclinic current flows along the front in the negative y direction, but its transport is not large enough relative to the barotropic current to be shown in this figure. Current convergence is seen as a strong current along the front. A countercurrent arises shoreward and downstream of this maximum. Equation (17) means that the shrinking resulting from the JEBAR effect (the second term) generates the barotropic flow (the third term) and the bottom stress (the fourth term). From the c distribution shown in Fig. 14, TABLE 1. Parameters used in the surface-to-bottom front model. Density r 0 Density r1 Bottom slope s Front width l Corioli parameter f (for 428N)

26.5 kg m23 26.0 kg m23 100 m/30 km 30 km 9.76 3 1026 s21

FIG. 14. Distribution of the (top) residual elevation h 2, (middle) transport streamfunction c, and (bottom) JEBAR term in the vorticity equation with location of the front (dotted line) between h(X ) and h(X 1 l). Contour intervals are 5 3 10 23 m, 5 3 10 4 m 3 s 21 , and 5 3 10 211 s 22 , respectively.

1490

JOURNAL OF PHYSICAL OCEANOGRAPHY

VOLUME 34

FIG. 15. Monthly surface temperature maps based on data collected by Tohoku National Fisheries Research Institute. Contour interval is 18C with thick lines every 58C.

the offshore velocity component arises shoreward of the maximum JEBAR region and causes negative vorticity there. To satisfy both the sign of the vorticity and the velocity component onshore of the front, the counter-

current must be at a smaller y than the maximum region. This is called ‘‘in forward direction’’ by Csanady (1985). Therefore, an eddy or countercurrent having negative vorticity should be seen leftward of local max-

JULY 2004

KONO ET AL.

1491

FIG. 16. Time series of surface salinity for Jan and Feb 1992. Waters less saline than 32.0 psu are shaded.

ima in the JEBAR term toward the coast, that is, in the forward direction to the maximum area. This explains the countercurrent on the continental shelf southeast of Hokkaido, the eddy south off Cape Erimo, and the eddy east off Tsugaru Strait. However, there is an eddy with positive vorticity on the southwest shelf of Hokkaido between 1428159 and 1438E (Fig. 10) and it is located in a backward direction from the JEBAR maximum as shown in Fig. 12. The baroclinic current flows southeast along the coast into the deeper water and is converged (stretched) at around 1428109E (Fig. 12). Since the stretching area is near the coast, a northwestward current needs to be generated along the coast shoreward of the southeastward current to maintain the mass balance and the positive vorticity in this area. This is why a clockwise eddy is not seen in the forward direction. Just south of Cape Erimo, where the water is shallow and thus the bottom stress is effective as shown in Fig. 12, the current velocity is small and its pattern is obscure (Fig. 10). 6. Discussion Steady-state flows were investigated using a baroclinic model with a highly resolved density structure. Strong baroclinic currents were seen to carry COW flow along the continental slope. These currents are almost parallel to the isobaths along the southeast coast, while their westward extension can be limited by outflows of the Tsugaru Warm Current, which is forced by baroclinicity. Shoreward of this current, a barotropic current

arises to form eddies and countercurrents. Thus there is no current carrying the core COW southwestward along the southeast shelf. It must be stated that our model only shows a snapshot of conditions based on the density structure in April 1989. To validate our model results, we show highresolution surface temperature maps collected, calibrated, and archived by the Tohoku Fisheries Research Institute (see online at http://ss.myg.affrc.go.jp/kaiyo/ temp/temp.html). Figure 15 shows the maps based on data collected in May and July 1995–2002. In agreement with our model results, closed and semiclosed circulations having diameters of 100–200 km are seen south of Cape Erimo in all maps. In addition, none of the maps show a continuous band of cold water from the southeast coast of Hokkaido to the southwest coast. Based on the COW distribution maps and model results, we conclude that COW along the southeast coast of Hokkaido does not always reach Funka Bay. Kono et al. (1998) examined temporal changes in the distribution of the CO from repeated CTD observations between mid-January and mid-March for every year between 1989 and 1994. They showed that to the west of 1428309E along the southwest continental shelf of Hokkaido, the western limit of COW changed significantly over several days. As shown in Fig. 16, the water disappeared ten days after its arrival just east of the Funka Bay. Using salinity data from repeated observations, we created a time series of the western limit of water with salinity less than 33.0 psu. The eastward velocity as-

1492

JOURNAL OF PHYSICAL OCEANOGRAPHY

VOLUME 34

FIG. 17. Wind direction percentage frequency for (top) Kushiro and Muroran, and comparison of westward speed of water less saline than 32.0 psu with northwestward and southeastward wind velocity component at (bottom left) Kushiro and (bottom right) Muroran, respectively. Wind velocity was averaged over the same period used for the speed estimates, namely, the time between salinity observations. See Fig. 1 for station locations.

sociated with this moving western limit was calculated and compared with the wind velocity, one of the largest driving forces. Wind data from Kushiro and Muroran were selected for comparison. These cities lie at the centers of the southeastern and southwestern coasts of Hokkaido, respectively. North-northeast and northwest wind directions are dominant at Kushiro and Muroran in February, the month when most of the repeated observations were carried out. The wind is almost parallel to each of the coasts at both the stations (Fig. 17). Figure 17 compares the northeastward and southeastward components of wind at Kushiro and Muroran with the eastward COW speed. The hourly wind components were averaged over the observation periods discussed earlier. The Kushiro winds are strongly correlated with the COW speeds with a confidence level of

99%, while the Muroran winds are not significant at the 95% level. Wind-influenced velocity fields are calculated using the same baroclinic model described above except for the addition of wind forcing hC of (20.086, 20.086) dyn cm 22 . This stress corresponds to a southwestward wind having a speed of 3 m s 21 , almost the same as the maximum velocity of the average southwestward component of wind at Kushiro (Fig. 18). Although a continuous southwestward current is not seen over the continental shelf southeast of Hokkaido, a southwestward current on the shelf break is seen to turn coastward at 1438359E and flow around Cape Erimo. The northwestward current is intensified between 1428209E and Cape Erimo where the anticlockwise eddy was seen for the no-wind model. This current pattern

JULY 2004

KONO ET AL.

1493

FIG. 18. Transport streamfunction and vertically averaged current for the baroclinic models (top) without wind forcing and (bottom) with forcing of (20.086, 20.086) dyn cm 22 . The upper panel is the same as in the lower panel of Fig. 9 and, except for the wind forcing, all the other conditions for the lower panel model run were the same as for the baroclinic model. Bottom topography is shaded and contour intervals and shaded areas are the same as in Fig. 9.

seems to correspond to the Fig. 16 salinity maps for 22–27 January, 31 January–4 February, and 4–7 February, periods when COW extended westward along the coast. The southwestward wind forcing raises the onshore gradient of the sea level, and could push COW westward along the southwest coast. When the wind forcing is weakened, the elevation field and the resulting velocity field along these coasts may turn back if the Oyashio and Tsugaru Warm Current are flowing steadily

in the offshore region. In such cases, if the COW is pushed back and as suggested by the salinity maps for 9–11 and 14–22 February, severe mixing of COW with the Tsugaru Warm Current water will occur. 7. Conclusions It was previously assumed that the coastal Oyashio was a buoyancy-forced current carrying cold and low-

1494

JOURNAL OF PHYSICAL OCEANOGRAPHY

salinity COW from the southeast coast of Hokkaido to the southwest coast, and sometimes as far west as Funka Bay. However, our model results show that barotropic currents due to the JEBAR effect arise along these coasts and control a southwestward current along the southeast coast. As a result, the model shows that the COW core is not carried along this southeast coast. On the other hand, the Oyashio carries COW offshore on the continental slope. This current turns around Cape Erimo but its northwestward extension is limited by the density structure along the southwest coast. Our model results agree with many observations that show an eddy south of Cape Erimo and that COW is not continuously distributed along the two southern coasts of Hokkaido. We suggest that COW is not carried continuously along the coasts because of a density structure that generates barotropic currents from the JEBAR effect and directly controls the Oyashio path. Wind forcing may drive the barotropic currents along the shelves and significantly vary the distribution of COW along the southwest shelf. However, it remains unclear what controls the density structure that was specified for our model. Since this structure mainly depends on paths and intensities of the Oyashio and Tsugaru Warm Currents, further investigation is needed to determine how these currents influence flow fields along the coastal region and how they are affected by changing winds. Acknowledgments. We thank one reviewer for encouraging us to show how JEBAR forms an eddy, Drs. Ito and Shimizu of the Tohoku Fisheries Research Institute for allowing us to use their sea surface temperature maps, and the Computer Center for Agriculture, Forestry and Fisheries Research MAFF for the use of their NEC SX4 computer in our model calculations. We also gratefully acknowledge the Hokkaido National Fisheries Research Institute, the Hokkaido Fisheries Experimental Station, the Japan Meteorological Agency, and the Iwate Fisheries Experimental Station for their efforts in collecting the data that were used in this study. The authors acknowledge financial and logistical support from the Fisheries Agency and the Ministry of Education, Culture, Sports, Science and Technology in Japan and Fisheries and Oceans Canada and encouragement from PICES, the North Pacific Marine Science Organization, under whose auspices this collaborative study was initiated. This research started when the first author belonged to Hokkaido National Fisheries Research Institute. REFERENCES Conlon, D. M., 1982: On the outflow modes of the Tsugaru Warm Current. La Mer, 20, 60–64. Cressie, N. A. C., 1990: The origins of kriging. Math. Geol., 22, 239–252. Csanady, G. T., 1985: ‘‘Pycnobatic’’ currents over the upper continental slope. J. Phys. Oceanogr., 15, 306–315. Foreman, M. G. G., R. E. Thomson, and C. L. Smith, 2000: Seasonal

VOLUME 34

current simulations for the western continental margin of Vancouver Island. J. Geophys. Res., 105, 19 665–19 698. Hannah, C. G., J. W. Loder, and D. G. Wright, 1996: Seasonal variation of the baroclinic circulation in the Scotia, Maine, region. Buoyancy Effects on Coastal and Estuarine Dynamics, C. T. Friedrichs, Ed., Coastal and Estuarine Studies, Vol. 53, Amer. Geophys. Union, 7–29. Hokkaido Fisheries Experimental Station, 1997: Record of Oceanographic Observations No. 5, Apr 1989–Mar 1990. 205 pp. Huthnance, J. M., 1984: Slope currents and ‘‘JEBAR.’’ J. Phys. Oceanogr., 14, 795–810. Isobe, A., 1994: Seasonal variation of the vertically averaged flow caused by the Jebar effect in the Tsushima Strait. J. Oceanogr., 50, 617–633. Isoda, Y., and K. Hasegawa, 1997: Heat budget of Funka Bay (in Japanese with English abstract). Umi to Sora, 3 (4), 93–101. ——, M. Shimizu, A. Ueoka, Y. Matsumo, K. Ohtani, and T. Nakatani, 1998: Interannual variations of oceanic conditions related to the Walleye Pollock population around the Pacific Sea area, south of Hokkaido (in Japanese with English abstract). Bull. Jpn. Soc. Fish. Oceanogr., 62, 1–11. Kono, T., 1997: Modification of the Oyashio Water in the Hokkaido and Tohoku areas. Deep-Sea Res., 44, 669–688. ——, and Y. Kawasaki, 1997: Results of CTD and mooring observations southeast of Hokkaido. Part 1: Annual velocity and transport variations in the Oyashio. Bull. Hokkaido Nat. Fish. Res., 61, 65–81. ——, K. Watanabe, K. Yabuki, and Y. Hamatzu, 1998: Temporal changes in distribution of Walleye Pollock eggs south of Hokkaido, Japan. Mem. Fac. Fish. Hokkaido Univ., 45, 52–55. Kubokawa, A., 1991: On the behavior of outflows with low potential vorticity from a sea strait. Tellus, 43A, 168–176. Lazier, J. R. N., and D. G. Wright, 1993: Annual velocity variations in the Labrador Current. J. Phys. Oceanogr., 23, 659–678. Lyard, F. H., 1997: The tides in the Arctic Ocean from a finite element model. J. Geophys. Res., 102, 15 611–15 638. Lynch, R., and F. Werner, 1987: Three-dimensional hydrodynamics on finite elements. Part I: Linearized harmonic model. Int. J. Numer. Methods Fluids, 7, 871–909. Mellor, G. L., 1996: Introduction to Physical Oceanography. AIP Press, 260 pp. Mertz, G., and D. G. Wright, 1992: Interpretations of the JEBAR term. J. Phys. Oceanogr., 22, 301–305. Naimie, C. E., and D. R. Lynch, 1993: FUNDY5 User’s Manual. Numerical Methods Laboratory, Dartmouth College, 40 pp. Ogasawara, J., 1990: Physics on the east and south coasts of Hokkaido (in Japanese). Coastal Oceanography of Japanese Islands, H. Kunishi, Ed., Tokai University Press, 839 pp. Ohtani, K., 1971: Studies on the change of the hydrographic conditions in the Funka Bay Part II: Characteristics of the waters occupying the Funka Bay (in Japanese with English abstract). Bull. Fac. Fish. Hokkaido Univ., 22, 58–66. ——, 1979: Water exchange of the Funka Bay (in Japanese). Bull. Coastal Oceanogr., 17, 50–59. Shaw, P., and G. T. Csanady, 1983: Self-advection of density perturbations on a sloping continental shelf. J. Phys. Oceanogr., 13, 769–782. Shimizu, M., and Y. Isoda, 1999: Flow structure of the Coastal Oyashio on the shelf area of Hidaka Bay (in Japanese with English abstract). Bull. Coastal Oceanogr., 36, 163–169. Sugimoto, T., and Y. Kawasaki, 1984: Seasonal variation of Tsugaru Warm Current and its dynamics (in Japanese). Bull. Coastal Oceanogr., 22, 1–11. Sugiura, J., 1956: A note on the current branches in the Oyashio area (in Japanese with English abstract). J. Oceanogr. Soc. Japan, 12, 117–119. ——, 1957: On the Oyashio Current in the sea adjacent to Hokkaido. Oceanogr. Magazine, 9, 133–142. Toba, Y., K. Tomizawa, Y. Kurasawa, and K. Hanawa, 1982: Seasonal and year-to-year variability of the Tsushima–Tsugaru Warm Current system with its possible cause. La Mer, 20, 41–51. Yasuda, I., K. Okuda, and M. Hirai, 1988: Short-term variations of the Tsugaru Warm Current in Autumn (in Japanese with English abstract). Bull. Tohoku Reg. Fish. Res. Lab., 50, 153–191.