Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model

Transcription

1 Journal of Oceanography Vol. 52, pp. 359 to Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model CHEOL-HO KIM and JONG-HWAN YOON Research Institute for Applied Mechanics, Kyushu University, Kasuga, Fukuoka 816, Japan (Received 29 September 1995; in revised form 10 November 1995; accepted 11 November 1995) The surface circulation in the Japan Sea is investigated using a 1.5 layer reduced gravity model. Historical observations suggest strongly that an anti-clockwise circulation is dominant in the subpolar region north of the Polar Front as a general feature. This anti-clockwise circulation as well as the branching of the Tsushima Warm Current was simulated well by incorporating the Na et al. (1992) s wind stress. The positive curl of the wind stress in the northern and the northwestern Japan Sea was found to play an important role in the formation of the subpolar gyre and the separation of the western boundary current (the East Korean Warm Current) in the Japan Sea. 1. Introduction One of the major features of the circulation in the Japan Sea is the existence of the Polar Front along about 40 N with warm and cold current regions in the south and north of it, respectively. As the oceanic condition of the warm current region is mainly characterized by the behavior of the Tsushima Warm Current, the path of the Tsushima Warm Current has been especially studied by many oceanographers since 1930 s (Kawai, 1974). Figure 1(a) shows the schematic pattern of the surface currents of the Japan Sea in summer suggested by Naganuma (1977). The Tsushima Warm Current bifurcates into two main branches, one of which flows along the Japanese coast (so called the Nearshore Branch) and the other one, the East Korean Warm Current (EKWC), along the Korean coast. The schematic map of the surface currents by Yarichin (1980) (Fig. 1(b)) shows the branching of the Tsushima Warm Current into two. The flow pattern of the EKWC is similar to that of Naganuma in general: the EKWC separates from the Korean coast at about N and a part of it flows southward as a countercurrent. The rest of the EKWC flows northeastward toward the Tsugaru Strait, forming the Polar Front with the northern cold water. According to the numerical investigations by Yoon (1982a, 1982b, 1991), the Nearshore Branch is controlled by the bottom topography in the shallow shelf region along the Japanese coast, while the EKWC is formed by the planetary β effect with a seasonally varying flow pattern due to the viscous and the diffusive effects on the western boundary layer. Between these two main currents, another current (the second branch) has been believed to exist on the offshore side of the Nearshore Branch (Shimomura and Miyata, 1957; Naganuma, 1977, 1985; Kawabe, 1982) while some other people suggested one meandering current instead of branching (Moriyasu, 1972; Ohwada and Tanioka, 1972). Contrary to the relatively abundant observations and dynamical explanations on the Tsushima Warm Current, the circulation in the cold current region is known very little until yet. Referring to the schematic map by Naganuma (1977) there can be seen some cyclonic gyres north of the Polar Front (Fig. 1(a)): two large cyclonic gyres locate in the Japan Basin and another smaller one in the north of 45 N. These cyclonic gyres are also suggested by Yarichin (Fig. 1(b)),

2 360 C.-H. Kim and J.-H. Yoon (a) (b) Fig. 1. Schematic surface current charts for the Japan Sea (a) in summer by Naganuma (1977) and (b) by Yarichin (1980). (C: cold water region, W: warm water region, H: relatively high temperature region, NB: Nearshore Branch, EKWC: East Korean Warm Current, LCC: Liman Cold Current, NKCC: North Korean Cold Current, EKB: East Korean Bay, JB: Japan Basin, P: Peter the Great Bay.) where one large gyre is seen in the whole Japan Basin with a smaller gyre to the north and south of it, respectively. So it is expected that the cyclonic gyres exist somewhere in the cold current region. The Siberian side of these cyclonic gyres has been called traditionally as the Liman Cold Current (LCC) since Schrenck (1873) had named it first (Kawai, 1974; Hahn, 1988). It has been accepted as one of the main currents in many historical schematic current charts proposed in Japan (Kawai, 1974). In reality, however, most of them were depicted only based upon tracer distributions without any direct current measurements. It seems that the existence of the LCC should be resorted to a few current measurements conducted by Russian investigators: it is reported that a part of the LCC, also known as the Shrank Current (according to Russian local terminology), flows southward from 51 N to about 47 N along the northern Siberian coast in all seasons (Ponomarev and Yurasov, 1994). This current seems to be followed by the Primorskoe Current (means the coastal current), which was observed to flow southwestward along the southern Siberian coast from 44 N to Peter The Great Bay in summer (Belinsky and Istoshin, 1950) and also in winter (Yarichin and Ryabov, 1994).

3 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 361 The LCC is taken over by the North Korean Cold Current (NKCC) flowing southward along the northeast coast of Korea, which was fairly often observed during the period of s through the drift bottle experiments, hydrographic observations and some direct current measurements using current meter mostly south of 43 N (Nishida, 1935, 1936, 1938, 1942). Recently, this cold current system is identified at the sea surface through the satellite images, where the LCC and the NKCC sometimes manifest themselves by forming the small eddies between the cold current axis and the Siberian and Korean coast (Ostrovskii and Hiroe, 1994). Though no more information is available after 1940 s on the direct current measurement in the northeastern coastal waters of Korea, one can find more about the observational facts at the southern tip of the NKCC in the south of 38 N. Within 20 km from the coast near 38 N the NKCC flowing southward with the speed of cm/sec was observed at the surface layer in June through August (Lee and Chung, 1981; Lie and Byun, 1985). Acoustic Doppler Current Profiler (ADCP) measurement has shown that the vertical scale of the NKCC extended from the surface to about 150 m depth near the coast with the southward component of more than 10 cm/sec in June, 1992 (Shin et al., 1995). However, south of 37 N, it becomes very weak at the surface layer on account of the northward flowing EKWC and can be observed only within few kilometers from the coast (Lee and Chung, 1981). Meanwhile, it is already known from the oceanographic observations covering the northern Japan Sea that the offshore extension of the EKWC exhibits an eastward flow along the Polar Front (Uda, 1934; Ohwada and Tanioka, 1972; Nishiyama et al., 1992). Then it seems to be quite reasonable to suppose that this eastward flow forms an anti-clockwise recirculation together with the southward flowing cold currents in the continental side and the northward flowing Tsushima Warm Current along the west coast of Hokkaido (Akagawa, 1955; Hata, 1962; Hatakeyama et al., 1985). The image of the anti-clockwise circulation in the cold current region is supported by the recent CREAMS (Circulation Research of the East Asian Marginal Seas) observation in the Russian territory. The CREAMS Group found the minimum of dynamic height on the sea surface at the center of the cold water region (41 30 N, 135 E) (Kim, 1994). The CTD observations by Hirai (1994) also found the feature of an anti-clockwise circulation in the northern cold water region east of 133 E. These observations suggest the existence of an overall anti-clockwise circulation in the cold current region with at least one smaller-scale cyclonic gyre over the eastern Japan Basin. As for the formation of the LCC/NKCC, Seung (1992) has shown the importance of both wind and thermal forcing in a geometrically simplified domain using a quasi-geostrophic 1.5 layer model. Sekine (1986, 1991) also noticed the effect of wind forcing in a two layer model, but the wind stress field adopted in his model does not seem to present the typical curl of wind stress over the Japan Sea, as will be discussed later. The purpose of this paper is to consider the formation mechanism of the subpolar gyre in the Japan Sea under the wind forcing and the separation of the EKWC which is closely related to the former. We use a simple reduced gravity model, because it is expected to give clear interpretations for the numerical results compared to comprehensive ocean general circulation model. 2. Model The numerical model used here is a nonlinear 1.5 layer (reduced gravity) model in a spherical coordinate. Only motion of the upper layer is considered and the lower layer is assumed to be infinitely deep and motionless. Then the governing equations are

4 362 C.-H. Kim and J.-H. Yoon u t + u u a cosφ λ + v u a φ uv a tan φ fv = g η a cosφ λ + τ λ ρ 0 H Ru + A h 2 h u 1+ tan2 φ u 2sin φ v a 2 a 2 cos 2, 1 φ λ ( ) v t + u v a cosφ λ + v v a φ + u2 tan φ a + fu = g a η φ + τ φ ρ 0 H Rv + A h 2 h v 1+ tan2 φ v + 2sin φ u a 2 a 2 cos 2, 2 φ λ ( ) η t + 1 a cosφ λ ( uh) + 1 vh cosφ a cosφ φ ( ) = γη + κ h ( ) 2 η. 3 Here u and v are the zonal (λ) and meridional (φ) components of velocity, while η is the sum of the displacements at the surface (η 1 ) and the interface (η 2 ), and H the upper layer thickness in motion (Fig. 2). The term f is the Coriolis parameter, a the radius of the earth, g the reduced gravity and ρ 0 the average density of the upper and the lower layer. τ λ and τ φ are the zonal and meridional components of wind stress, respectively. The coefficient R denotes the coefficient of Rayleigh drag, A h the coefficient of horizontal eddy viscosity and γ the coefficient of Newtonian damping. The coefficient κ corresponds to the coefficient of horizontal eddy diffusivity for density, which was introduced in the reduced gravity model by Masuda and Uehara (1992). The depth of the upper layer is taken as 200 m. The reduced gravity g is assumed to be a constant value of 2.0 cm/sec 2, considering the typical values of the Tsushima Warm Current water and the cold water beneath it, though significant spatio-temporal variations of density are expected at the surface layer. Since η 1 is very small compared to η 2 (η 1 /η 2 g /g where g is the Fig. 2. Definitions of model thickness and density. Thickness of the upper layer is decreased linearly toward the Japanese coast within 4 grid interval in order to show the topographic effect.

5 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 363 Table 1. The numerical experiments. Steady inflow means the inflow of a constant volume transport of m 3 /sec. In Experiment 2 model ocean is closed and flat. gravitational acceleration), η will be regarded as the interface displacement in the following. R and A h are /sec and cm 2 /sec which yield the proper order of magnitudes for the width of the western boundary layer in the present model: the Stommel layer (R/β) is 500 m and the Munk layer ((A h /β) 1/3 ) is 37 km. The coefficient κ is zero in Experiment 1 3, but different value other than zero is considered in Experiment 4 6. The damping coefficient γ controls the propagation of a long Rossby wave from the eastern boundary (Kawase, 1987). In the test experiment by Takaki (personal communication) when γ is as /sec (corresponding to about 100 days damping time), only eastern boundary layer appears, because Rossby wave damps down before reaching the western boundary. While in the case of /sec (order of 30 years damping time), only western boundary layer remains. The transit time of internal long Rossby wave, L/βλ R2, over the Japan Sea is about 1300 days (3.5 years) if L is taken as 1000 km and λ R (internal Rossby deformation radius, g H 0 /f ) as 21 km at 40 N. We adopt γ as /sec (about 3 years of damping time scale), which gives a proper order of magnitude in the present study. As for the wind forcing, we used monthly mean wind stress data which were computed from the twice daily weather map for the period of 1978 to 1987 by Na et al. (1992). Through the experiments except the closed ocean case (Experiment 2) the inflowing volume transport is fixed as a constant value of 2 Sv (10 6 m 3 /sec) with no horizontal shear of the northward velocity at the model inlet. At the Tsugaru and the Soya Straits the radiation condition is applied for the eastward component of velocity and the slippery condition for the northward component. No slip condition is imposed for the velocity (u, v) and no mass flux condition for η at the lateral boundaries. To incorporate the topographic effect, the inverted bottom topography is given along the Japanese coast as shown in Fig. 2: H 0 (the depth of the upper layer at rest) decreases linearly toward the coast. The finite differencing for equations (1) (3) is made on the Arakawa C grid (Mesinger and Arakawa, 1976). A potential enstrophy conserving scheme for a spherical grid is adopted following Arakawa and Lamb (1981), which improves the simulation of nonlinear aspects of the flow. Grid interval of the model ocean is 1/3 in both longitudinal and latitudinal directions. We did 6 cases of experiments, which are described in Table 1. After 7 years of numerical integration, the total kinetic and potential energy reached almost statistical steady state for each experiment. We will discuss only the results of the last year (10th year) in the following sections.

6 364 C.-H. Kim and J.-H. Yoon 3. Results and Discussion At first, let s see the case where only inflow and outflow are allowed, but without wind forcing (Experiment 1). Figure 3 shows the distributions of horizontal velocity and interface displacement. The interface displacement η shows the horizontal distribution very similar to the flow pattern like a streamfunction. The Nearshore Branch flows steadily along the Japanese coast following the constant (inverted) bottom topography as a topographically controlled current. The EKWC branches away from the Nearshore Branch at the mouth of the Japan Sea and flows along the Korean coast up to 42 N, where it separates from the coast and flows offshore toward the Tsugaru Strait. It should be noted that the separation latitude is controlled by the latitude of the outflow. In Experiment 2 only wind forcing is applied without inflow and outflow, assuming that the model ocean is closed and flat. Then there appears a large and strong anti-clockwise circulation in the northern part of the Japan Sea in winter (Fig. 4(a)). It consists of a south/southwestward flow along the Siberian and Korean coast and an offshore return flow. This circulation is composed of two separate cyclonic gyres: one locates off the East Korean Bay, centered at about N, E and another in the northeastern Japan Basin, centered at N, 138 E. It is interesting to note that the former agrees well with the cyclonic gyre in the East Korean Bay shown in Fig. 1(b). The southward flow in this gyre, that is, the NKCC affects to about 36 N in the Korean coast in winter. The eastward return flow is formed at N in the central region, Fig. 3. Distributions of (a) horizontal velocity and (b) interface displacement (unit in meter) in Experiment 1.

7 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 365 which is corresponding to the position of the Polar Front in the central Japan Sea (Isoda et al., 1991). In summer, this anti-clockwise circulation system becomes weak and a southwestward flow appears along the Japanese coast (Fig. 4(b)), which will be discussed with the distribution of wind stress curl later in this section. Figure 5 shows the result of the case in which the inflow and outflow are given together with the wind forcing. In winter, the EKWC cannot flow farther to the north along the Korean coast on account of the strong NKCC flowing southward (Figs. 5(a) and 5(b)), while in summer it becomes strong as the former becomes weak (Figs. 5(c) and 5(d)). It separates from the Korean coast at about N in winter and N in summer. A general pattern of the Polar Front is formed along the latitude 40 N in the distribution of the interface displacement (Figs. 5(b) and 5(d)). The results of Experiments 2 and 3 are characterized by the presence of a basin-scale anticlockwise circulation with the two cyclonic gyres inside it. These features are different with the previous numerical results in that anti-clockwise circulation is confined mainly north of 42 N in winter, elongated east to west (Vasilev and Makashin, 1992) or even clockwise circulation is dominant in the northern half of the Japan Sea (Sekine, 1986, 1991). The reason is that the northern circulation simulated in the present model is largely dependent upon the characteristics of the adopted climatological wind stress field. Figure 6(a) shows the distribution of wind stress curl in January computed from the twice daily weather map for 10 years by Na et al. (1992), where maximum of positive curl appears at two locations in the northern Japan Sea. The positive curl Fig. 4. Distributions of horizontal velocity in (a) February and (b) September in Experiment 2. Volume transport is calculated at sections A, B, C and D.

8 366 C.-H. Kim and J.-H. Yoon Fig. 5. Distributions of horizontal velocity and interface displacement (unit in meter) in (a), (b) February and (c), (d) September in Experiment 3.

9 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 367 Fig. 6. Curl of wind stress (a) in January and (b) of annual mean computed by Na et al. (1992). with the magnitude of more than dyn/cm 3 continues from October to next March almost at the same place as is inferred in the annual mean distribution (Fig. 6(b)) and spins up the ocean. In the rest of the season it becomes almost zero or turns to small negative value from the positive (Na et al., 1992). However, the wind stress curl calculated from the global data set such as Hellerman and Rosenstein s (1983) or Kutsuwada and Sakurai s (1982) which Sekine (1986, 1991) used does not show such a salient characteristic of wind in this marginal sea. From the distribution of interface displacement between the upper and the lower layers (Figs. 5(b) and 5(d)), it is certain that the lower cold water will form a dome-like distribution inside the corresponding northern and northwestern cyclonic gyres, respectively. Figure 7 shows the temporal change of interface displacement on the section along the latitude N. The cold water dome represented by the negative maximum of interface displacement starts to rise from December by the upward Ekman pumping of wind and develops strongly during February to April. At the same time, the interface displacement near 140 E shows the westward propagation of an internal Rossby wave. It seems that the Rossby wave propagation helps to weaken the eastern half of the cold water dome together with the spin-down process by the wind from May to September. The western flank of the cold water dome remains the same because it reached the western boundary already. The phase speed of internal Rossby wave is estimated to be about 1.2 cm/sec in Fig. 7. From October to next April, the negative curl of wind stress appears off the west coast of Japan though its magnitude is less than half of the northern positive maximum (Na et al., 1992), which is responsible for the southwestward flow along the Japanese coast during March to November (Fig. 4(b)). Overshooting of the WBC is one of the afflicting problems in the Ocean General Circulation Model. Even in the modeling of the Japan Sea the multi-level primitive equation model does not succeed in the proper separation of the WBC, resulting in an overshooting of the EKWC more

10 368 C.-H. Kim and J.-H. Yoon Fig. 7. Variation of interface displacement from 135 E to E along N latitude. Ordinate shows the year of numerical integration in the model. Negative values indicate a rising of the interface from no motion state (unit in meter). or less (Seung and Kim, 1993; Kim and Yoon, 1994). In our experiment, the EKWC separates nicely from the Korean coast with its seasonal variation. Considering the separation latitude in Experiment 1 (almost equal to the latitude of the Tsugaru Strait), it can be understood easily that the proper separation obtained in Experiment 3 is greatly indebted to the existence of the northwestern cyclonic gyre. Figure 8 shows the seasonal variations in the volume transports of the WBC at each section indicated in Fig. 4. The maximum transport of the cold currents driven by the wind reaches to about 2.6 Sv at the northernmost section D and about 1.4 Sv at the southernmost section A in mid-february, while the transport of the EKWC at section A is

11 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 369 Fig. 8. Volume transports of the cold currents in Experiment 2 at sections A, B, C and D indicated in Fig. 4 (unit in 10 6 m 3 /sec). Positive value indicates southward transport. estimated to be about 0.9 Sv in Experiment 1. The EKWC separates at the south of the latitude of section A from November to March (not shown here) when the NKCC is stronger than the EKWC in the volume transport at section A. In Experiments 1 to 3, we did not consider the effect of horizontal diffusion of interface displacement. When both the horizontal diffusion and the bottom friction are absent, the width of the western boundary layer is given as the Munk layer in the presence of lateral friction. Incorporation of horizontal diffusion, however, is known to change the structure of the western boundary layer into two different (viscous and diffusive) regimes according to the relative intensity of model parameters (Masuda and Uehara, 1992). Takaki et al. (1994) have shown in the fine-meshed reduced gravity model that the width of the western boundary layer diminished as the horizontal diffusion was increased in the viscous regime. This caused the strengthening of the nonlinearity of the WBC, thereby induced the overshooting of the WBC. When the horizontal diffusion became larger than the critical value, the western boundary layer changed into the diffusive regime. In Experiments 4 to 6, we introduced horizontal diffusion term with a different order of magnitude in κ but keeping other parameters unchanged. Figures 9(a) 9(c) show the distributions of horizontal velocity in February when κ is taken as 10 6, and 10 7 cm 2 /sec, respectively. Almost no difference is found in the separation latitude between the cases when κ = 10 6 cm 2 /sec and κ is zero (Fig. 5) except that the former is decreased about % in the magnitude of interface displacement of the northern cyclonic gyre compared to the latter. As κ is increased further, however, the northern and the northwestern cyclonic gyres become weaker and the separation position of the EKWC moves to north: from 37 N latitude when κ = 10 6 cm 2 / sec to N latitude when κ = 10 7 cm 2 /sec.

12 370 C.-H. Kim and J.-H. Yoon Fig. 9. Distributions of horizontal velocity computed with different magnitude of κ. (a) Experiment 4 (κ = 10 6 cm 2 /sec), (b) Experiment 5 (κ = cm 2 /sec), (c) Experiment 6 (κ = 10 7 cm 2 /sec).

13 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 371 Under the assumption of quasi-geostrophic dynamics, the dissipation of vorticity by horizontal diffusion is given as κ( f 2 /c 2 ) 2 η (Masuda and Uehara, 1992). So, the dissipation is more effective at high latitudes and at less stratified waters. It is expected in Experiments 5 and 6 that the enhanced horizontal diffusion dissipates effectively the vorticity of the northwestern (and also northern) cyclonic gyre, which weakens the southward flowing NKCC. 4. Conclusion As is reviewed in Section 1, many observational facts combine to give us a picture of a basinwide anti-clockwise circulation in the northern Japan Sea with smaller scale cyclonic gyre inside. Using a simple reduced gravity model, we successfully simulated this northern circulation system including the separation of the EKWC, the Nearshore Branch following the Japanese coast and the general feature of the Polar Front. The formation of the cyclonic gyres north of the Polar Front is due to the positive curl of the wind stress affecting steadily on the corresponding sea areas in winter. The wind stress field also plays a very important role to the separation of the EKWC and the southwestward shift of the Polar Front in the continental side. It is a little difficult to examine in detail the separation problem of the WBC in this kind of some coarse-resolution models. But present numerical model solution suggests that the separation position is determined at first order by the competition between the strengths of the opposing WBC as is shown in the example of the EKWC and the NKCC. Thermohaline forcing is known to be another candidate for the formation of the northern cyclonic gyre under an idealized condition of numerical model (Kim and Chung, 1989; Seung and Kim, 1989; Seung, 1992). As a future numerical investigation, more quantitative analysis should be made on the respective contributions of wind and buoyancy flux to the circulation in the Japan Sea based upon realistic field data. Acknowledgements This work is a part of the program, Circulation Research of the East Asian Marginal Seas, supported by a grant by the Ministry of Education, Science and Culture, Japan. We express our hearty thanks to Prof. Takematsu for his encouragement during this research. We also thank Prof. Na in Hanyang Univ. of Korea for providing the wind stress data. We appreciate the comments of the two anonymous reviewers and JO editor for improving the manuscript. References Akagawa, M. (1955): On the oceanographical conditions of the North Japan Sea (west of the Tsugaru-Straits) in summer (part 2). J. Oceanogr. Soc. Japan, 11, 1 7 (in Japanese with English abstract). Arakawa, A. and V. R. Lamb (1981): A potential enstrophy and energy conserving scheme for the shallow water equations. Mon. Wea. Rev., 109, Belinsky, N. A. and Y. V. Istoshin (1950): Primorskoe current as observed in expedition of the boat Rossinante in Proc. Central Inst. Weather Pred., 17, (in Russian). Hahn, S. D. (1988): Koreanology from an Oceanological Point of View. Haejosa, Seoul, 538 pp. (in Korean). Hata, K. (1962): Seasonal variation of the volume transport in the northern part of the Japan Sea. J. Oceanogr. Soc. Japan, 20th Anniversary Vol., (in Japanese with English abstract). Hatakeyama, Y., S. Tanaka, T. Sugimura and T. Nishimura (1985): Surface currents around Hokkaido in the late fall of 1981 obtained from analysis of satellite images. J. Oceanogr. Soc. Japan, 41, Hellerman, S. and M. Rosenstein (1983): Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr., 13, Hirai, M. (1994): Oceanic structure of the Liman Current region in summer. Fall Meeting of the Oceanogr. Soc. Japan, (in Japanese).

14 372 C.-H. Kim and J.-H. Yoon Isoda, Y., S. Saitoh and M. Mihara (1991): SST structure of the Polar Front in the Japan Sea. p In Oceanography of Asian Marginal Seas, ed. by K. Takano, Elsevier. Kawabe, M. (1982): Branching of the Tsushima Current in the Japan Sea. Part II: Numerical experiment. J. Oceanogr. Soc. Japan, 38, Kawai, H. (1974): Transition of current images in the Japan Sea. p In The Tsushima Warm Current Oceanic Structure and Fishery, ed. by Fishery Soc. Japan, Koseisha-Koseikaku, Tokyo (in Japanese). Kawase, M. (1987): Establishment of deep ocean circulation driven by deep-water production. J. Phys. Oceanogr., 17, Kim, C.-H. and J.-H. Yoon (1994): Circulation of the Japan Sea as seen in the prognostic numerical model. Kaiyo Monthly, 26, (in Japanese). Kim, K. (1994): Circulation of the East Sea and CREAMS. Kaiyo Monthly, 26, (in Japanese). Kim, Y. E. and J. Y. Chung (1989): Numerical study of the circulation in the Japan Sea. I. Case of closed basin. J. Oceanol. Soc. Korea, 24, (in Korean with English abstract). Kutsuwada, K. and K. Sakurai (1982): Climatological maps of wind stress field over the north Pacific Ocean. Oceanogr. Mag., 32, Lee, J. C. and W. Chung (1981): On the seasonal variations of surface current in the eastern sea of Korea (August 1979 April 1980). J. Oceanol. Soc. Korea, 16, Lie, H.-J. and S.-K. Byun (1985): Summertime southward current along the east coast of Korea. J. Oceanol. Soc. Korea, 20, Masuda, A. and K. Uehara (1992): A reduced-gravity model of the abyssal circulation with Newtonian cooling and horizontal diffusion. Deep Sea Res., 39, Mesinger, F. and A. Arakawa (1976): Numerical methods used in atmospheric models. GARP Pub. Ser. 17, Vol. 1, World Meteorol. Organ., Geneva, 64 pp. Moriyasu, S. (1972): The Tsushima Current. p In Kuroshio Its Physical Aspects, ed. by H. Stommel and K. Yoshida, Univ. Tokyo Press. Tokyo. Na, J.-Y., J.-W. Seo and S.-K. Han (1992): Monthly-mean sea surface winds over the adjacent seas of the Korean Peninsula. J. Oceanol. Soc. Korea, 27, Naganuma, K. (1977): The oceanographic fluctuations in the Japan Sea. Mar. Sci. (Kaiyo Kagaku), 9, (in Japanese with English abstract). Naganuma, K. (1985): Fishing and oceanographic conditions in the Japan Sea. Umi to Sora, 60, Nishida, K. (1935): Results of the drift bottle experiments in the Japan Sea. Umi to Sora, 15, (in Japanese). Nishida, K. (1936): Results of the current measurements in the southern sea of Tyosen and the western part of the Japan Sea, on board the R.M.S. Misago-maru in June Sept., Annual Rept. Hydro. Obs. Fish. Exp. Sta., 7, (in Japanese). Nishida, K. (1938): Oceanographic investigations during June July, 1933, in the Japan Sea offshore along the east coast of Tyosen on board the R.M.S. Misago-maru. Annual Rept. Hydro. Obs. Fish. Exp. Sta., 8, (in Japanese). Nishida, K. (1942): Research conducted during December, 1934, on board the R.M.S. Misago-maru, in the neighbouring seas of the Korean Gulf. Annual Rept. Hydro. Obs. Fish. Exp. Sta., 9, (in Japanese). Nishiyama, K., M. Inagawa and T. Mizuno (1992): Circulation of the Japan Sea. Kaiyo Monthly, 24, (in Japanese). Ohwada, M. and K. Tanioka (1972): Cruise report on the simultaneous observation of the Japan Sea in October Oceanogr. Mag., 23, Ostrovskii, A. and Y. Hiroe (1994): The Japan Sea circulation as seen in satellite infrared imagery in autumn Proc. CREAMS 94, Fukuoka, Ponomarev, V. I. and G. I. Yurasov (1994): The Tartar (Mamiya) Strait Currents. J. Korean Soc. Coast. Ocean Eng., 6, Sekine, Y. (1986): Wind-driven circulation in the Japan Sea and its influence on the branching of the Tsushima Current. Prog. Oceanogr., 17, Sekine, Y. (1991): A numerical experiment on the seasonal variation of the oceanic circulation in the Japan Sea. p In Oceanography of Asian Marginal Seas, ed. by K. Takano, Elsevier. Seung, Y. H. (1992): A simple model for separation of East Korean Warm Current and formation of North Korean Cold Current. J. Oceanol. Soc. Korea, 27, Seung, Y. H. and K. Kim (1989): On the possible role of local thermal forcing on the Japan Sea circulation. J. Oceanol. Soc. Korea, 24,

15 Modeling of the Wind-Driven Circulation in the Japan Sea Using a Reduced Gravity Model 373 Seung, Y. H. and K. Kim (1993): A numerical modeling of the East Sea circulation. J. Oceanol. Soc. Korea, 28, Shimomura, T. and K. Miyata (1957): The oceanographical conditions of the Japan Sea and its water systems, laying stress on the summer of Bull. Japan Sea Reg. Fish. Res. Lab., 6, (in Japanese). Shin, H.-R., S.-K. Byun and C. Kim (1995): The characteristics of structure of warm eddy observed to the northwest of Ullungdo in J. Oceanol. Soc. Korea, 30, (in Korean with English abstract). Takaki, K., J.-H. Yoon and M. Takematsu (1994): The effect of diffusion coefficient to the separation of the western boundary currents. Proc. CREAMS 94, Fukuoka, Uda, M. (1934): The results of simultaneous oceanographical investigations in the Japan Sea and its adjacent waters in May and June, J. Imp. Fish. Exp. Sta., 5, (in Japanese with English abstract). Vasilev, A. S. and V. P. Makashin (1992): Ventilation of the Japan Sea waters in winter. La mer, 30, Yarichin, V. G. (1980): Study state of the Japan Sea circulation. p In Problems of Oceanography, ed. by V. Pokudov, Hydrometeoizdat, Leningrad (in Russian). Yarichin, V. and O. Ryabov (1994): Current field structure of the Japan Sea in February March Proc. CREAMS 94, Fukuoka, Yoon, J.-H. (1982a): Numerical experiment on the circulation in the Japan Sea, Part I. Formation of the East Korean Warm Current. J. Oceanogr. Soc. Japan, 38, Yoon, J.-H. (1982b): Numerical experiment on the circulation in the Japan Sea, Part III. Mechanism of the nearshore branch of the Tsushima Current. J. Oceanogr. Soc. Japan, 38, Yoon, J.-H. (1991): The seasonal variation of the East Korean Warm Current. Rep. Res. Inst. Appl. Mech. Kyushu Univ., 38,