Outflow of the intermediate water in the East/Japan Sea through the Tsugaru Strait is inferred from the fact that profiling floats deployed below the sill depth in the central East/Japan Sea moved out to the Northwest Pacific Ocean through the Strait. Careful analysis of the float trajectories reveals that profiling floats parked at 350m, 700m, and 800m could be uplifted from depths below the sill of the Tsugaru Strait and moved over the 140m deep sill with a probability close to 100%. Application of an analytical model for the Bernoulli suction shows that intermediate waters can be aspirated over the sill depth of the Tsugaru Strait in late winter.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The East/Japan Sea (hereafter EJS) is a deep semi-enclosed marginal basin, which is connected with the North Pacific and the Okhotsk Sea through the Korea/Tsushima, Tsugaru, Soya, and Tatar Straits (Figure 1), whose sill depths are shallower than 140 m or less. The Tsushima Current of warm and saline water flows into the EJS through the Korea/Tsushima Strait and out through the Tsugaru and Soya Strait. The spatial structure and annual variation of the currents in the Tsugaru Strait have been recently revealed by ADCP measurements [Onishi et al., 2004], which indicate that it is the major outlet in terms of volume transport.
 The EJS possesses its own deep water which is formed through convective overturning in wintertime and comprises various water masses just like the open oceans [Kim et al., 2004]. The water masses formed in the northern part of the EJS fill up below the thermocline in the entire EJS including the southern part. As the depth of the permanent thermocline in the southern part of the EJS is about 200 m which is deeper than sill depths of not only the Tsugaru Strait but also the other straits, previous studies assumed that the whole water masses formed in the EJS should be isolated from the North Pacific and be consumed only within the EJS by dilution with surface water [Yanagi, 2002].
 Little attention has been paid to any possibility of the outflow of water below the permanent thermocline into the neighboring seas and the North Pacific except for a thin layer of the cold bottom water in the Korea Strait, originating from the northern EJS [Kim et al., 2006]. It is well known that this layer does not extend any farther out of the Korea Strait. Also, it should be noted that Kang et al.  recently noted the depletion of the deep water mass and the augmentation of the intermediate waters in the EJS and Kwon et al.  has shown the long-term change of the water masses. Heat and salt balances in the EJS require a priori correct water budget including any outflow of deep waters from the EJS. Thus, it is crucial to know whether any water below the thermocline finds a way out of the EJS.
 In this paper, we will provide the first observational evidence of intermediate water outflow through the Tsugaru Strait, using the trajectory data of profiling floats. Firstly, the detail characteristics of floats will be introduced, which are essential to demonstrate that the floats drift out with the intermediate water. Secondly, we will present a quantitative probability of which floats moved over the sill and went out of the EJS. Finally, an aspiration process will be proposed as a possible mechanism of the outflow.
2. Data and Result
2.1. Profiling Float Data in the East/Japan Sea
 The profiling float cycles vertically between a preset depth and the sea surface, changing its density periodically. It freely drifts for a pre-programmed time interval with the surrounding water at the preset depth (so called parking depth) and measures a temperature and salinity profile during its ascending to the sea surface. Then, while it broadcasts its data, position and time of the float at the surface are observed by satellites. It is important to know that a float drifts an unknown distance after it rises to the sea surface until its position is fixed by the satellite. Floats also drift away after their last contact with a satellite until they start to dive. Therefore, a special treatment of data such as proposed recently by Park et al.  is required in order to trace the float trajectory at the parking depth accurately, since a full surface trajectory is not immediately available.
 More than 100 floats have been deployed in the EJS since 1996 and most of them moved around in the Ulleung Basin and the Japan Basin. Eight floats approached the Tsugaru Strait within 100 km from the Tappi Sill (Table 1). Five floats passed through the Tsugaru Strait and moved out into the North Pacific. We easily identify from the satellite fixes that two floats, UW229 and 10029, in Table 1 passed over the Tappi sill while they drifted at the surface. However, trajectories of the three floats in Figure 2a have no surface fixes near the sill. We will examine carefully whether those floats moved over the sill while they were at depth, indicating uplift of deep water from a depth below the sill.
Table 1. Status of Floats Approaching Near the Tsugaru Strait
Shortest Distance From the Tappi Sill, km
Parking Depth, dbar
Cycle Period [(UP-time) + (DOWN-time)], hours
Passing Through the Tsugaru Strait
March 12, 1998
240 [10 + 230]
Over the sill
February 24, 2000
186 [18 + 168]
Over the sill
March 14, 2000
120 [18 + 102]
Over the sill
October 8, 2001
186 [18 + 168]
At the surface
May 8, 2001
120 [18 + 102]
At the surface
July 16, 2001
186 [18 + 168]
July 23, 2001
186 [18 + 168]
 It is remarkable to note that all three floats, UW213, SNU350, and 08173, passed through the Tsugaru Strait in late winter. However, two other floats, UW272 and UW288 that approached the sill in summer, never went out of the EJS (Table 1).
Figure 2b shows bottom topography near the Tappi Sill and schematic diagrams of float trajectories at the sea surface and at depth. In order to clarify that a float can go over the sill, we have to know the positions (PDS and PAS) in principle where a float reaches the depth equal to the sill depth, 140 m, during its descent and ascent (Figure 2b). Since it is difficult to directly estimate the positions (PDS and PAS), we will show the possibility that the position of the Tappi Sill (PS) is located between PDS and PAS.
2.2. Time Information
 Besides of the float trajectory data obtained by Argos satellites, additional information on timing when a float starts to dive and arrives to the sea surface can be traced using the cycle period and cycle number of each float according to Park et al.  within an accuracy of tens of minutes, as follows:
where TDm and TAm are Descending-Start-Times and Ascending-End-Times at the mth cycle, ΔTcyc the cycle period of float operation, ΔTdown the “DOWN-time” which means the duration from starting to dive to starting to ascend (Table 1), ΔTasc the time from starting to ascend to arriving at the surface.
 We know the float ascending speed (typically 0.08 dbar/sec) as well as the deepest depth where a float parks using previous profile data so that we can estimate ΔTasc. Similarly, the time (TA − TAS) taking to ascend from 140 m depth to the surface can be also computed with the ascending rate, which is about 0.49 hours. Even though the time difference between TD and TDS cannot be directly computed, according to the descent rate test of Davis et al. , it takes about a half hour to pass the depth of 140 m after a float dives to the depth.
 Therefore, the time differences between TL and TDS and that between TAS and TF (ΔTLDS and ΔTASF, respectively) are
Table 2. Parameters and Probabilities for Tracking Subsurface Trajectory of Float
Probability of no outflow
PLDS (VLDS ≥ VLDSmin)
PASF (VASF ≥ VASFmin)
P = 1 − (PLS + PSF)
2.3. Extrapolation Using Surface Trajectory of Float
 The temporal integration of the surface velocities for the unknown drifting periods (TD-TL and TF-TA) could provide estimates of surfacing and diving positions as extrapolations from the surface trajectories. Closed symbols in Figure 2a show the extrapolated positions where the three floats surface and dive near the Tappi Sill according to Park et al. .
 In fact, PDS and PAS in Figure 2b are the positions that we have to know in order to demonstrate that the floats dive below the sill depth and yet go over the sill. Even though we have no further information on the vertical structure of current around the sill, we can compute the upper limit of the distance which a float may drift from PD to PDS and from PAS to PA under the assumption of homogeneous currents in depth. Maximum current speeds estimated at the surface are 0.85 m/s for 08173, 0.41 m/s for SNU350, and 0.91 m/s for UW213. PDS and PAS may be apart from PD and PA by only 1.5 km, 0.7 km, and 1.6 km in maximum, respectively, which are more than 10 times smaller than the distance from the sill (PS). Therefore, it is clearly evident that all floats sink down before they reach the sill and rise up after passing over the sill.
2.4. Probability of Float Movements Over the Tappi Sill
 In this section, we investigate the quantitative probability that the three floats could move over the sill following the bottom. There are two extreme cases that a float may pass through the Tsugaru Strait without huddling the sill, which are indicated as dotted lines in red and blue in Figure 2b. For the case of the red trajectory, if P*DS were placed at the eastern side of PS, a float would pass through above the sill while it is ascending or it is at the surface after ascending. For the case of the blue trajectory, if P*AS were at the western side of the PS, a float could go by the sill while it is at the surface before descending or it is descending. Thus, showing that a float trajectory does belong to neither case, we demonstrate that the float must move over the sill if none of two extreme cases are possible in reality.
 At this point, P*DS and P*AS cannot estimated from the trajectory data directly, but a minimum drifting speed of a float can be obtained, which is required for P*DS to be at the eastern side of the PS or P*AS to be at the western side of PS. In other words, the drifting speeds means a minimum value for a float to pass through the sill while it is at the depth shallower than the sill depth, which follows as
 If the mean speed of currents between PL and PS (VLDS) exceeds the minimum speed (VLDSmin) or that between PF and PS (VASF) exceeds VASFmin, a float just passes through the Tsugaru Strait while it is at a depth above the sill. The distances and the minimum velocities are tabulated in Table 2 for each float.
 The probability (Pno_overflow) with which the overflow from depths below the sill never happens are
 Therefore, the probability (Poverflow) that a float flows over the sill can be calculated as Poverflow = 1 − Pno_overflow.
 In order to compute the probabilities in (4a) and (4b), we utilize the current velocity data estimated from satellite tracked drifter trajectories which passed through the Tsugaru Strait from 1990 to 2005, whose drogue loss indexes have been quality-controlled by AOML (Atlantic Oceanographic and Meteorological Laboratory). The whole drifter trajectories are interpolated using the method reported by Bograd et al.  and the current velocities have been calculated every 3 hours. The total number of current speed samples reaches 460 and the mean current speed is about 0.6 m/s (Figure 3).
 The histogram of the surface current speeds in Figure 3b is well fitted into a Rayleigh function with correlation of 0.94. Also, the fitted Rayleigh cumulative function has a high correlation of 0.99 with the observational cumulative histogram. According to this probability distribution function, PLDS and PASF in (4a) have been computed and listed in Table 2.
 The possibilities in Table 2 indicate that the current speeds near the Tappi Sill are mostly smaller than the minimum velocities (VLDSmin and VASFmin). That is, the surface currents near the sill are not strong enough to let the floats pass through the sill before they dive below the sill depth, with possibility of 99.3%∼99.9%. Indeed, the maximum current speed measured by the drifters reaches only about 1.4 m/s which is substantially smaller than the minimum speeds required for a float to pass the sill depth. Therefore, we can conclude that all three floats, SNU350, UW213, and 08173, must sink down below the sill depth and then go over the sill. There is an extremely small chance that these floats moved out of the EJS by drifting at the surface. The mean and tidal currents in the Tsugaru Strait reported by Onishi et al.  typically are maximum values of 0.5 m/s and 0.2∼0.3 m/s (for dominant constituents of M2, K1, and O1) in winter, respectively, which also support the probability estimated from surface drifters in this paper.
 An ideal model known as Bernoulli Suction which was used for the Strait of Gibraltar and the Alboran Sea by Stommel et al.  can be applied for the outflow of deep water from the EJS and its seasonal dependence. Later, Seim and Gregg  introduced a simple dynamical formulation for the Bernoulli Suction under a constant stratification and showed that the depth of fluid below sill depth which could be aspirated, h1 − H2, is a function of U2/N (Figure 4a). According to Seim and Gregg  the depth of the offshore streamline (h1) can be formulated as
where N2 = gΔρ/ρ0H1, g is the gravitational acceleration, ρ0 the surface density, and Δρ is the density difference from top to bottom in the deep basin.
Figure 4b is an example of potential density profile right near the Tappi Sill which was obtained in March, 2000, the late winter. Average buoyancy frequency (N) is about 5 × 10−3s−1. Supposing that current speed in the strait is about 1.4 m/s which is the maximum of the current velocities in Figure 3, h1 is computed as about 380 m. So, fluid deeper by 240 m than the sill depth of 140 m can go over the sill and enter the Tsugaru Strait. On the other hand, the typical buoyancy frequency in summer reaches about 2 × 10−2 s−1, which is 4 times larger than that in winter, and h1 becomes about 170 m under the same speed of surface current. Therefore, the late winter is the most favorable season for deep water flows out to the North Pacific because of the weak vertical stratification.
 As mentioned in section 2.1, the floats are designed to follow isobaric surfaces. So, the floats of UW213 and 08173 with the parking depth of 800 dbar might be denser than the surrounding water even though they grounded in front of the sill at the ocean bottom of 300 m depths (in-situ density at 300 dbar: 1028.7 kg/m3; density at 800 dbar: 1031.0 kg/m3). Also there is a possibility that some sediment like mud could be attached on surface of the float (D. Swift, personal communication, University of Washington, 2007). It should be noted that these behaviors could make a float denser than expected and prevents it from going over the sill with the surrounding fluid. However, all floats move to shallower regions (shallower than about 300 m) while they are at the surface before going over the sill. Indeed, the subsurface drifts of the floats (UW213 and SNU350) near the mouth of the Tsugaru Strait either following isobaths or hardly move around at depth before moving toward the sill at the surface. Even though the floats may be heavier than the surrounding bottom water, it is conjectured that they may slide or walk over the ocean bottom to traverse the sill by advection within strong currents at the bottom.
 We thank Steve Riser and Dana Swift for kindly providing profiling float data and valuable discussion. In addition special thanks go to Ray Schmitt and Young-Oh Kwon for their comments and discussions. This study was partly supported by the project of ARGO for Data Assimilation project in the Meteorological Research Institute/KMA, the Korea EAST-1(East Asian Seas Time series East Sea) Program from the Ministry of Maritime Affairs and Fisheries, and Korea Research Foundation Grant (KRF-2005-070-C00142). The first author was partly supported by the Ministry of Education, Korea, through the BK21 project.