Sea Surface Temperature Variability at Futaoi Island
in the Tsushima Strait
 
Tomoharu Senjyu1
 
(received 1998/2/3, revised 1998/4/20, accepted 1998/4/30)
 
Abstract
 

The sea surface temperature (SST) variation at Futaoi Island in the Eastern Channel of the Tsushima Strait was investigated. In the summer two characteristic periods, a prevailing diurnal variation and dominant semi-diurnal variation, appear alternately in 5-7 day intervals. The variations tend to synchronize with the neap and spring tide periods; the semi-diurnal variation with a larger amplitude becomes dominant in the spring tide period, and the diurnal one with a smaller amplitude prevails in the neap tide. These fluctuations are caused by an internal tide excited from the external tide. The winter SST exhibits sawtooth-like fluctuations composed of a sudden increase followed by a gradual decrease with a dominant period of about 5 days. The variation indicates a periodic passing of warm water masses with a sharp front and gradual "tail", at its head and rear, respectively. The SST variation has a high coherency with the cross-strait wind fluctuation. This suggests that the sawtooth-like SST variation in winter is a kind of disturbance excited by the northwesterly monsoon burst in the Tsushima Strait.

 
(Key Words: Sea surface temperature, The Tsushima Strait, internal tide, warm water mass, monsoon burst)
 
1 National Fisheries University, 2-7-1, Nagata-honmachi, Shimonoseki-shi, Yamaaguchi, 759-6595 Japan (senjyu@fish-u.ac.jp)
 
Introduction
 
The Tsushima Strait is only the entrance of the Japan Sea for the Tsushima Current flowing from the East China Sea. Many oceanographers have studied the nearshore (or first) branch of the Tsushima Current in the Eastern Channel of the Tsushima Strait, which flows northeast to eastward along the Japanese coast. Kawabe (1982a) pointed out that the nearshore branch is found on the coastal side of the line where the main pycnocline intersects the bottom slope. Using a two-layer numerical model, Kawabe (1982b) concluded that the nearshore branch is a shelf-trapped current controlled by the topographic b-effect. Indeed, recent studies using ADCP (Acoustic Doppler Current Profiler) demonstrated a strong flow along the 100 m isobath around the strait (Kato, 1994a, b).

On the other hand, Tawara and Fujiwara (1985) revealed several temperature maxima and minima in the Eastern Channel of the Tsushima Strait by using sea surface temperature (SST) records of a ferry boat across the strait. The positions and shapes of these SST maxima are very changeable and usually persist only a week or so. Tawara and Fujiwara (1985) discussed the SST maxima in relation to the warm water intrusion from the East China Sea, and they speculated that the fluctuations are caused by the passing of patches of warm water in the surface layer.

To check the fish ground environment, SST has been measured at Futaoi Island in the Eastern Channel of the Tsushima Strait west off Shimonoseki, Japan (Fig.1) since 1988. Because the island is located on the flow path of the nearshore branch of the Tsushima Current, it is a suitable place for monitoring the Tsushima Current activity and disturbances generated in the Tsushima Strait. Indeed, Kato's (1994b) ADCP diagram shows a broad but strong current just north of Kyushu. This indicates the direct influence of the Tsushima Current on Futaoi Island.

Fig. 1 Location of the observation site. The lower panel is an enlarged map around Futaoi Island, as shown in a square in the upper panel. MJ and TU in the upper panel show locations of Moji and Cape Tsuyazaki, respectively. A solid square in the lower panel denotes the mooring site. In this study, we examined the SST variation at Futaoi Island, in particular focusing on the dominant frequency in the summer and winter seasons. The observation method and instruments used are shown in the next section. Section 3 describes the SST variation at Futaoi Island. Causes of fluctuations in the summer and winter seasons are discussed in Section 4. Finally, conclusions are given in Section 5.
 
Observation
 
The observation site is outside of a small bay north of Futaoi Island, where the water depth is 28 m (solid square in Fig. 1). There is a setnet for fishing at this location, and it is utilized as an observational platform; the mooring system is shown in Fig. 2. Fig. 2 Mooring system of the thermometer (left) and thermistor-chain (right). A self-record type thermometer (ALEC Electronics, Model AT-8000 or AT-32K) has been hung by a rope from setnet buoys on the sea surface since September 1988 (Fig. 2, left). The measurement depth is about 4 m. Note that the temperature at the constant depth can be recorded regardless of sea level variations due to the tide. Although the measurement is made in a subsurface layer at 4 m deep, we call it the sea surface temperature (SST). The time interval of the measurement is 30 minutes. To change the battery and recover the SST record, the thermometer is replaced every half year. This operation is useful to avoid the decay of measurement performance due to hulling of shellfish.

The sea level at Moji and wind at Cape Tsuyazaki (MJ and TU in Fig. 1) are compared with the SST record to discuss the cause of its variations. Unfortunately, the wind and sea level data before 1995 are available at this time. Thus, we analyze the SST record before 1995 in this study, though the SST observation has been continued.

In addition to the above observation, a thermistor-chain with 11 sensors (AANDERAA TR-7) was moored on the setnet buoys (Fig. 2, right) to observe the variation of vertical thermal structure. The observation was carried out in the periods of April 25 to October 18, 1996 and October 22, 1996 to May 7, 1997. Although temperatures at 2 m intervals were obtained every hour, measurements at 1 and 5 m deep were noisy and erroneous in the latter observation period. Moreover, the temperatures at several dep at times, which wereprobablths exhibited inversionsycaused by the instrumental error of individual thermistors. Therefore, the data from the thermistor-chain is used in supplementary discussion.

Fig. 3 Time series of SST at Futaoi Island during the period from October 1, 1989 to September 30, 1990. A scale of the horizontal axis denotes 5 days, except for the last scale in October, December, January, February, March, May, July and August.  
SST variation at Futaoi Island
 
As an example of the SST record, the time series for one year during the period from October 1, 1989 to September 30, 1990 is shown in Fig. 3. Temperatures from 12:30 to 16:00 on February 6, 1990 lack data because of the thermometer replacement.

The SST at Futaoi Island exhibits a clear seasonal change; the temperature decreases gradually from September or October and shows a minimum of about 13oC in the period from February to March, though a somewhat higher SST (more than 15oC) was found in this observation period. After that, gradual warming begins from April and the SST reaches a maximum of about 29oC in the period from July to August.

Besides the seasonal variation, seasonal changes of the dominant frequency are recognizable in shorter variations. In the summertime (from late June to August), short period fluctuations with relatively large amplitudes (more than 1.0oC) are seen. On the other hand, periodic variations with a period of about 5 days are dominant in winter (from December to early April). The range of the SST fluctuations in winter is about 1.0oC. In spring and autumn, SST amplitude is much smaller than that in summer and winter (less then 0.5oC). Thus, the SST variations in the summer and winter seasons are examined in more detail.

A typical example of the summertime SST variation is shown in Fig. 4, which is the enlarged time series of Fig. 3 during the period of July 20 to August 4, 1990. Diurnal variations are dominant in the periods of July 20-23 and July 31 to August 4. Note that the maximum SST in a day appears at around midnight in the former period. This indicates that the diurnal variation is not caused by solar heating. (Since the observation was made at a depth of 4 m, the solar heating effect is feeble.) On the other hand, fluctuations of periods equal or less than semi-diurnal prevail in the period of July 24-28. The range of SST fluctuations in a day is about 2.0oC, which is larger than that in the periods of prevailing diurnal variation. These two characteristic periods, prevailing diurnal variation and dominant semi-diurnal variation, appear alternately in 5-7 day intervals. Similar SST variations can be seen in the period from mid-June to late September every year.
 

Fig. 4 A typical example of summertime SST variations measured from July 20 to August 4, 1990.
The enlarged time series in winter during the period of December 30, 1989 to January 14, 1990 is shown in Fig. 5. In contrast to the summertime record, variations with diurnal and semi-diurnal periods are not seen. Instead, periodic variations with periods of 3-5 days are prevalent. The time series exhibits sawtooth-like SST variations composed of a sudden increase followed by a gradual decrease. The typical rate of temperature increase is 1.0oC in 4-5 hours, but a steeper increase of more than 1.0oC in one hour occurs at times, such as on January 5-6 in Fig. 5. Before the sudden increase, the SST tends to decrease significantly. Similar variations appear every year from early December to late March.
 
Discussions
 

Summertime SST variation

To investigate the dominant frequency in summertime fluctuations, the spectrum density based on the SST record from June 5 to August 29, 1990 is calculated and shown in Fig. 6. Two spectrum peaks at 23.5 and 12.5 hours are prominent; they correspond to the tidal periods. Thus, we compared the SST variation at Futaoi Island with the sea level record at Moji   (MJ in Fig. 1)
 
 

Fig. 6 Spectrum density diagram in the summertime based on the SST record from June 5 to August 29, 1990.
  Figure 7 shows the sea level at Moji corrected by the atmospheric pressure (upper) and SST anomaly from 25-hours running mean values at Futaoi Island (lower) during the period of June-August, 1990. The amplitude of SST fluctuation tends to be large, and semi-diurnal variation is dominant in the spring tide period, while in the neap tide period, the SST amplitude is smaller than that in the spring tide and diurnal variation is prevalent.
 
 

The multi-layer observation with the thermistor-chain accurately demonstrates the transition of vertical thermal structure in the subsurface layer from neap to spring tide periods. Figure 8 shows the time series of temperatures at depths from 3 to 21 m in the period of July 20 to August 4, 1996. (Since the thermistor at a depth of 1 m showed unrealistic SST values in this observation, it is eliminated in Fig. 8.) The first quarter and full moon occurred on July 24 and 30, respectively. Relatively strong stratification is found in the neap tide period (July 23-26); the temperature ranges about 4.0oC between the top and bottom layers. Though the top layer (3 m deep) exhibits diurnal variation, semi-diurnal fluctuations are dominant below the depth. Since the larger temperature gradient in the vertical is illustrated by larger spacing between temperatures at each layer in this figure, we can see that the thermocline lay between 7 and 15 m deep and oscillated semi-diurnally in this period; this indicates the existence of an internal tide. On the other hand, in the spring tide period (July 29-August 1), the range of temperature fluctuation with the semi-diurnal period increases with depth and is about 2.0oC near 21 m in depth. Occasionally, almost homogeneous water occupies through the water column, such as 18:00-24:00 on July 30 and 06:00-11:00 on August 1; in these periods the temperature difference between the top and bottom layers is less than 1.0oC. This indicates a large displacement of the thermocline of more than 18 m.
 

Fig. 8 Variation of the vertical thermal structure measured with the thermistor-chain in the period from July 20 to August 4, 1996. Circles on the horizontal axis denote moon ages. Temperatures at a depth of 1 m lack data.   Generally, internal tides are generated at steep topography by the running up or down of external tidal flows. Since the bottom topography near the observation site is relatively steep (see Fig. 1), the internal tide might excite around the island. The vertical flow due to the running up or down of tidal currents oscillates the thermocline in both neap and spring tide periods, but the amplitude in the latter is larger than that in the former. Therefore, the SST amplitude tends to be large in the spring tide period, while weaker vertical flow generates smaller amplitudes of internal tide in the neap tide period. Since the tidal inequality makes larger and smaller vertical displacements of the thermocline in a day and the smaller one does not reach the upper layer, only the diurnal variation is recorded on the thermometer at the shallow depth. This is the reason why the semi-diurnal variation does not appear in the SST record in the neap tide period.
 

Wintertime SST fluctuation

The spectrum density based on the SST record during the period of November 5, 1989 to January 29,1990 is shown in Fig. 9a in order to display the dominant frequency in wintertime fluctuations. In contrast to the summertime spectrum density (Fig. 6), tidal period components are not dominant. Instead, a clear peak exists at 5.0 days, which corresponds to the sawtooth-like fluctuations, as shown in Fig. 5.
 

Fig. 9 Spectrum density diagrams of SST at Futaoi Island (a) and cross-strait wind speed at Cape Tsuyazaki (b) based on the records from November 5, 1989 to January 29, 1990.
  The sawtooth-like fluctuations can be observed in the wintertime only (see Fig. 3). The period in which the sawtooth-like fluctuations prevail almost coincides with that of strong monsoons blowing from the northwest to the southeast in the Tsushima Strait. Figure 10 shows time series of along (NE-SW) and cross-strait (NW-SE) wind components for three months of November 1989 to January 1990, obtained at Cape Tsuyazaki (TU in Fig. 1). Bursts of more than 8.0 ms-1 of the northwesterly monsoon occurred in 3-10 day intervals, though no remarkable trend was found in the along strait component. Figure 9b is the spectrum density of the cross-strait wind component. The leading three peaks are 9.1, 3.3, and 5.0 days. In addition, two sub-peaks are recognizable at 2.0 and 1.0 days; the latter probably corresponds to the land and sea breeze.

The coherency between the SST at Futaoi Island and cross-strait wind at Cape Tsuyazaki is shown in Fig. 11. The 5.0 days component, which is the dominant one in both SST and cross-strait wind, appears as a coherence-squared peak, and its phase lag is +49.4 degrees. This reveals that the northwesterly wind leads the SST increase by 16.5 hours. The high coherency between them suggests that the sawtooth-like SST fluctuations in winter are a kind of disturbance generated by the northwesterly monsoon burst in the Tsushima Strait, though the dynamics of the process are not yet clear.
 

Fig.10 Time series of wind speeds at Cape Tsuyazaki: along-strait (upper, NE is positive) and cross-strait components (lower, NW is positive). The bold lines denote 24-hours running mean values.
  A wintertime record of the thermistor-chain during the period of December 1-16, 1996 is shown in Fig. 12. The stratification in the winter is very weak compared with that in the summer; the temperature difference between the top (3 m) and bottom layers (21 m) is less than 1.0oC. As the thermocline does not exist, the influence of the internal tide is very weak. A sudden temperature increase of more than 1.0oC occurred from December 5 to 6 at all the depths. This reveals a passing of a mass of warm water having a vertical scale of more than 21 m. After that, a gradual decrease follows at each layer. Note that the rate of temperature decrease is almost the same at each layer. This indicates that the temperature decreases are not caused by a cooling from the sea surface. From the manner of temperature variation, a warm water mass with a sharp front and "tail" at its head and rear, respectively, is inferred.
  Fig. 11 Coherency between SST at Futaoi Island and cross-strait wind based on the records from November 5, 1989 to January 29, 1990: coherence-squared (left) and phase (right). The position of 5.0 days is shown with an arrow and a circle in coherence-squared and phase diagrams, respectively.   Fig. 12 Time series of a vertical thermal structure during the period of December 1-16, 1996. Temperatures at depths of 1 and 5 m lack data.
 
Concluding remarks
 
The SST variation at Futaoi Island in the Eastern Channel of the Tsushima Strait was investigated with a focus on the summertime and wintertime fluctuations. In the summertime, two characteristic periods, dominant diurnal variation and prevailing semi-diurnal variation, appear alternately in 5-7 day intervals. The variation tends to synchronize with the neap and spring tide periods; semi-diurnal variations with larger amplitudes are dominant in the spring tide period, and diurnal variations with smaller amplitudes prevail in the neap tide. These fluctuations are caused by the internal tide excited from the external tide.

The wintertime SST exhibits sawtooth-like fluctuations with a dominant period of about 5 days. The variation reveals a periodic passing of warm water masses with a sharp thermal front at its head and a gradual tail at its rear. Since the SST variation has a high coherency with the cross-strait wind fluctuation, it is suggested that the sawtooth-like SST variation in the winter season is a kind of disturbance excited by the northwesterly monsoon burst in the Tsushima Strait.

In this study, we discussed the short period fluctuations with periods from semi-diurnal to a few days. However, longer period fluctuations, in particular of about 20 days, seem to be present throughout the year (Fig. 3). Indeed, the wintertime spectrum (Fig. 9a) exhibits a broad peak at around 20 days. Since the period is commonly observed in the Kuroshio front fluctuations in the East China Sea (Qiu et al., 1990), this may be an essential feature of the Tsushima Current, as pointed out by Tawara and Fujiwara (1985). More extensive SST measurements are being performed in the Tsushima Strait, and new discoveries will arise from the results in the near future.
 

Acknowledgements
 
I would like to thank the Futaoi Island Fisheries Cooperative Association for their kind cooperation in the observation field. Prof. S. Sugihara, Dr. A. Isobe and Mr. T. Watanabe are to be acknowledged for their encouragement. The wind and sea level data used in the study were offered from the Research Institute for Applied Mechanics of Kyushu University and the Japan Oceanographic Data Center, respectively. I would also like to offer special respect to the late Prof. S. Tawara, who started the SST observations at Futaoi Island.
 
References

Kato, O. (1994a) Short-term fluctuations of the Tsushima Current in waters northwest of Yamaguchi Prefecture. J. Oceanogr., 50, 51-64.

Kato, O. (1994b) Structure of the Tsushima Current in the southwestern Japan Sea. J. Oceanogr., 50, 317-338.

Kawabe, M. (1982a) Branching of the Tsushima Current in the Japan Sea. Part I. Data analysis. J. Oceanogr. Soc. Japan, 38, 95-107.

Kawabe, M. (1982b) Branching of the Tsushima Current in the Japan Sea. Part II. Numerical experiment. J. Oceanogr. Soc. Japan, 38, 183-192.

Qiu, B., T. Toda and N. Imasato (1990) On Kuroshio front fluctuations in the East China Sea using satellite and in situ observational data. J. Geophys. Res., 95(C10), 18191-18204.

Tawara, S. and T. Fujiwara (1985) Sea surface temperature distribution and its variability across the Tsushima Strait. J. Oceanogr. Soc. Japan, 41, 49-55.