Variability of Upper Ocean Heat Balance in the Shikoku Basin during the Ocean Mixed Layer Experiment (OMLET)

Transcription

1 Journal of Oceanography, Vol. 59, pp. 619 to 627, 2003 Variability of Upper Ocean Heat Balance in the Shikoku Basin during the Ocean Mixed Layer Experiment (OMLET) HIROTAKA OTOBE 1 *, KEISUKE TAIRA 2, SHOJI KITAGAWA 3, TOMIO ASAI 4 and KIMIO HANAWA 5 1 International Coastal Research Center, Ocean Research Institute, the University of Tokyo, Akahama, Otsuchi, Iwate , Japan 2 Japan Society for the Promotion of Science, Ichibancho, Chiyoda-ku, Tokyo , Japan 3 Ocean Research Institute, the University of Tokyo, Minamidai, Nakano-ku, Tokyo , Japan 4 Japan Science and Technology Corporation, Toranomon, Minato-ku, Tokyo , Japan 5 Department of Geophysics, Graduate School of Science, Tohoku University, Aoba, Sendai , Japan (Received 7 August 2002; in revised form 5 March 2003; accepted 5 March 2003) The heat balance of the surface layer in the vicinity of the former Ocean Weather Station Tango (OWS-T; 29 N, 135 E), where a large amount of heat is transported by the Kuroshio and transferred to the atmosphere, was studied by during Ocean Mixed Layer Experiment (OMLET) as an oceanographic component of the Japanese World Climate Research Program ( ). Temperature and velocity in the upper ocean measured using a surface moored buoy system deployed by the Ocean Research Institute, the University of Tokyo, in total 668 days of four time series namely the periods of April 1988 November 1988 (OMELET-88), August 1989 February 1990 (OMLET-89), April 1990 September 1990 (OMLET-901) and September 1990 January 1991 (OMLET-902). We have analyzed the moored buoy data of the upper 100 m for the latter three time series (OMLET-89, -901 and -902) and here we discuss the heat balance of the upper 100 m, in combination with surface heat flux and oceanographic data provided by the Japan Meteorological Agency. A large fluctuation of oceanic heat convergence/divergence of W/m 2 in amplitude with predominant period of days occurred in the first half of OMLET-89 period, which was just the early stage in the formation process of a large meander path of the Kuroshio. A large amount of heat convergence of 71 and 79 W/m 2 on average was detected in observation period of OMLET-89 and -901, respectively. During OMLET-902, relatively small heat convergence of 13 W/m 2 was obtained. It is suggested that these variations of oceanic heat convergence in this region were closely related to the fluctuation of the Kuroshio axis to the south of Japan. Keywords: Heat balance, heat budget, heat content, Kuroshio, mixed layer, OMLET, Shikoku Basin. 1. Introduction The subtropical recirculation area that includes the Kuroshio in the western North Pacific near Japan is one of the largest heat release areas as well as the western part of the North Atlantic in the world oceans (Weare et al., 1981; Hsiung, 1986; Otobe, 1989). Kurasawa et al. (1983) reported that the large heat release to the atmosphere is compensated by oceanic heat convergence by a horizontal process related to the Kuroshio. They based their conclusion on hydrographic data from the former * Corresponding author. otobe@wakame.ori.u-tokyo. ac.jp Copyright The Oceanographic Society of Japan. Ocean Weather Station Tango (OWS-T; 29 N, 135 E, 4800 m deep), where ship observations were continuously made from 1950 to 1953 by the Japan Meteorological Agency (JMA). However, there is no clarity about whether the effect of advection or diffusion is dominant in this area, nor about the details of the mechanism of heat convergence/divergence. In order to clarify the three-dimensional processes in the ocean that influence the development of the mixed layer and surface heat budget, the Ocean Mixed Layer Experiment (OMLET) at OWS-T was planned and performed in the northwestern North Pacific subtropical gyre. OMLET was a sub-program of the Japanese WCRP in its first phase from (Toba et al., 1991). One of OMLET s activities was a surface mooring system de- 619

2 Fig. 1. Location of the moored buoy and sketch of the Kuroshio axis, the warm eddy (WE) and the cold eddy (CE) appearing during the Kuroshio large meander (after Oceanographic Prompt Report by the Hydrographic Department, Japan Maritime Safety Agency (1990)). veloped by the Ocean Research Institute, the University of Tokyo (Taira et al., 1993) that was deployed during the OMLET period 10 miles south of the OWS-T, where the Ocean Data Buoy of JMA (ODB off Shikoku) has been maintained since 1972 replacing ship s observation (Fig. 1). Four time series of temperature and current velocity in the upper ocean were obtained in total 668 days, for the periods from April to November 1988 (OMLET-88), from August 1989 to February 1990 (OMLET-89), from April 1990 to September 1990 (OMLET-901) and from September 1990 to January 1991 (OMLET-902). The results of the first time series, OMLET-88, were analyzed and published by Toba et al. (1991) and Taira et al. (1993). Toba et al. (1991) described the occurrence of a warm water outbreak from the Kuroshio using satellite derived SST images and hydrographic data together with current data from the mooring system. Taira et al. (1993) analyzed the fluctuation of current fields and temperature changes after the passage of a typhoon in October However, thus far, no description has been given of the last three time series. Therefore, the present study describes the observational results, and closely examines the heat balance of the upper ocean during OMLET-89, -901 and We describe temperature and current velocity data of the three time series, OMLET-89, -901 and -902 obtained by a moored buoy in Section 2. The heat balance of the upper 100 m layer and their time variation are examined in Section 3. The variability of the heat balance of the upper ocean is discussed with special reference to Fig. 2. Surface mooring system used for the three observation periods of OMLET-89, OMLET-901 and OMLET-902. the fluctuation of the Kuroshio axis in Section 4. The conclusions are presented in Section Overview of Moored Buoy Data 2.1 Mooring and equipment The surface mooring systems for three observation periods are shown schematically in Fig. 2. These buoy systems were developed for the OMLET observations by the Ocean Research Institute, University of Tokyo (Taira et al., 1993). Eleven or thirteen thermometers (RMT, Rigosha Co.), two depth (pressure) meters (RMD, Rigosha Co.) and two electromagnetic current meters with a temperature sensor (Model ACM5000, Alec Electronics Co.) were attached from the surface to 200 m, as shown Fig. 2. Temperatures were recorded with an accuracy of ±0.05 C and pressures with an accuracy of ±0.5% at intervals of 30 minutes using solid memory devices. Current speeds were also recorded with an accuracy of ±0.5 cm/s and current directions with an accuracy of ±2 degrees every 60 minutes. 620 H. Otobe et al.

3 Fig. 3. Time series of daily mean water temperature and wind speed during OMLET-89 (a). Water temperature, pressure records at 110 m and 190 m and wind speed during OMLET- 901 (b), OMLET-902 (c). 2.2 Temperature Figure 3(a) shows the time series of daily mean water temperature and wind speed from ODB/JMA during OMLET-89. The mooring system was set up on 12 August 1989, but it started to drift southward at the end of December The line was retrieved on 26 February 1990 at N, 132 E. We discovered that the line was cut at 105 m depth and the deeper portion did not rise to the surface due to inadequate buoyancy. Figure 3(a) shows strong stratification in the summer season in the upper 100 m and development of a mixed layer in the cooling season. In August and September, the temperature difference in the upper 100 m was about 10 C and the depth of the surface mixed layer was shallower than 20 m. However, the thermocline was sometimes destroyed by wind stirring agitation when the wind speed was greater than 10 m/s (i.e., around 26 August, 6 and 21 September). Temperatures in the layer from 50 m to 100 m fluctuated with a period from 20 days to 30 days from the beginning of the record to October. The surface mixed layer of 50 m thickness was formed in November The temperatures in the layer deeper than 50 m fluctuated coherently with increasing amplitude toward shallower depth. In late December, the mixed layer deepened to 100 m, and temperature increased from late January It should be noted that the temperature data after the end of December were collected during the period when the buoy system was drifting southward, as mentioned earlier. Figure 3(b) shows the time series of temperature and wind speed during OMLET-901. This mooring system was deployed on 18 April 1990 and was retrieved intact on 8 September Depth records at 110 m and 190 m are also plotted in the middle panel of Fig. 3(b). Temperatures recorded by each thermometer at each time are plotted as the value at the predetermined depth of each thermometer by linear interpolation using the depth records. When the mooring system of OMLET-901 was deployed a weak seasonal thermocline had been found in the subsurface layer, and a significant stratification subsequently began to develop in May. The surface temperature increased rapidly from mid-june and reached a peak of about 30 C in late July due to heating by solar radiation. The surface temperature then gradually decreased and a relatively deep surface mixed layer was formed intermittently due to wind stirring from the passage of several typhoons in August Among them, Typhoon 9014 had a great effect on the temperature structure around August 1990, causing the mixed layer to deepen. Figure 3(c) shows the time series of temperature, depth and wind speed during OMLET-902. This mooring system was deployed on 9 September 1990 and was retrieved intact on 15 January Unfortunately, the excessive buoyancy at about 3000 m depth caused the upper 200 m portion of this buoy to surface when the wind speed was weak (see Fig. 2). Therefore, the temperature data deeper than 120 m during the periods from 22 October to 3 November and from 14 November to 20 November were eliminated from the figure, due to the sensors installed deepen than 120 m floating up to the shallower depth of 100 m during those periods. Figure 3(c) shows the formation process of the winter mixed layer of 100 m thickness. Immediately after the deployment of the mooring system in early September, the temperature at 20 m depth reached a peak of 28 C and fluctuated with a time interval of several days. Although the temperatures above 100 m during the cooling season (December to January) Variability of the Upper Ocean Heat Balance in the Kuroshio Region 621

4 Fig. 5. Time series of daily mean velocities at about 50 m (U - zonal component, V - meridional component) during OMLET-89 (a), OMLET-901 (b) and OMLET-902 (c). Fig. 4. Progressive vector diagrams from the current records at about 50 m during OMLET-89 (a), OMLET-901 (b) and OMLET-902 (c). during OMLET-902 fluctuated similarly to those of OMLET-89, the temperature fluctuation in the surface layer above 100 m remained 1 2 C higher than those measured during OMLET Current velocity The current velocity records were obtained only at about 50 m depth for three mooring systems due to malfunction of the current meters. The daily mean progressive vector diagrams from the current records during three observation periods are shown in Figs. 4(a), (b) and (c), respectively. Although the mean speeds were almost the same (about cm/s) in all three periods, the mean current direction varied between OMLET-89 and the other experiments. That is, the mean current direction was southward for OMLET-89 and westward for OMLET-901 and The flow field in the study area is often covered by the Kuroshio path (Taira et al., 1990). During OMLET- 89, the Kuroshio took the path of the transition stage from non-meander path to a large meander path, as described in detail in Section 4. The moored buoy was located on the eastern side of small warm eddy situated off Kyushu Island. Therefore, it is expected that the flow was southward during OMLET-89. In fact, observation by the Kobe Marine Observatory, JMA, showed that ADCP was southward (Kobe Marine Observatory, 1990). On the other hand, the moored buoy was located on the southern perimeter of the warm eddy during OMLET-901 and -902 when the Kuroshio took a large meander path. This suggests that the flow was westward, as shown in Fig. 1, which is in good agreement with our observations. 622 H. Otobe et al.

5 Table 1. Mean values of local time change of heat content ( H/ t), surface heat flux (Q) and divergence/convergence (F) over each observation period. Observation period H/ t Q (W/m 2 ) F OMLET OMLET OMLET Figure 5 shows the time series of the zonal (U) and meridional (V) components of current velocity during each observation period. The zonal component, U, during the first half of OMLET-89 (Fig. 5(a)) varied with a predominant period of days, as did temperature (Fig. 3(a)). The mean current speed over the same period was 19.4 cm/s in a south to southwestward direction (see Fig. 4(a)). A large change of the current speed and direction can also be found from April to June 1990 during OMLET 901 (Fig. 5(b)). Fluctuations with a period of days, similar to those of OMLET-89, were seen the westward flow prevailed through out the periods of OMLET-901 and -902, (Figs. 5(b) and (c)). 3. Estimates of Heat Balance in the Upper 100 m Column 3.1 Formulation The heat balance of an upper water column from the surface to depth D is given by the following equations, H = Q t + F, () 1 H = ρ C Tdz, p 0 ( 2) D Q= Q + Q + Q + Q, () 3 e h s b Fig. 6. Low-pass filtered (one month running mean) time series of time derivative of heat content ( H/ t), surface heat budget (Q) and oceanic heat convergence (F) for OMLET- 89 (a), OMLET-901 (b) and OMLET-902 (c). where H/ t is the local time change of heat content H integrated from the surface to a constant depth of D, T - temperature, ρ - sea water density, C p - specific heat at constant pressure, Q - the net surface heat flux, and F - oceanic heat convergence or divergence, consisting of three-dimensional advection, eddy heat flux and molecular diffusion terms. The heat content, H, is calculated using temperature profiles linearly interpolated to 10-m intervals from the raw data after depth correction. In the present study, D is set to 100 m. The net surface heat flux (Q) is the sum of latent heat flux (Q e ), sensible heat flux (Q h ), net shortwave radiation flux (Q s ) and net long-wave radiation flux (Q b ). These fluxes are calculated from the surface meteorological parameters including downward short-wave radiation obtained at intervals of 3 h by ODB/JMA, applying Kondo s aerodynamic bulk formulas (Kondo, 1975) for latent and sensible heat fluxes, and Kondo s formula (Kondo et al., 1991) for net long-wave radiation. is derived as a residual term in Eq. (1). 3.2 Results The mean values of H/ t, Q and F over each period of OMLET-89, -901 and -902 are summarized in Table 1. A large average heat convergence had occurred during both periods of OMLET-89 and On the other hand, the heat convergence was small during OMLET-902, and H/ t and Q balanced on average. Figure 6 shows the time series of H/ t, Q and F for the upper 100 m layer smoothed by a 31-day running mean for OMELET-89, -901 and A large fluctuation of the local time change of heat content with a range of several hundreds W/m 2, occurred during the period from Variability of the Upper Ocean Heat Balance in the Kuroshio Region 623

6 Fig. 7. Power spectra of the time series of oceanic heat convergence (F) for OMLET-89 (a), OMLET-901 (b) and OMLET-902 (c). August to September 1989 during OMLET-89 (Fig. 6(a)). In this period, the surface net heat flux, Q, was so small that the oceanic heat convergence or divergence (F) nearly balanced the values of H/ t. From late September to the end of December 1989, both the local time change of heat content H/ t and the surface heat flux Q decreased coherently with each other, and the oceanic heat convergence F persisted constantly. During OMLET-901, corresponding to the heating season, the local time change of heat content ( H/ t) varied with a predominant period of days and a range of about 200W/m 2. The large oceanic heat convergence F occurred during both periods from late April to early June and from late July to September (Fig. 6(b)). From mid-june to late July in OMLET-901, and during OMLET- 902, except late September, the surface heat flux, Q, and the local time change of heat content, H/ t, balanced each other on average, even though both of them had a relatively small fluctuation of about 100 W/m 2 (Figs. 6(b) and (c)). A spectral analysis was done to detect the dominant period of fluctuation of oceanic heat convergence/divergence F (Fig. 7). There are several peaks of spectral density in a period from 2 to 20 days during all three observation periods. However, dominant periods (with more than 90% confidence) are different for each OMLET period. In particular, a peak at 20-day period is noteworthy in OMLET-89 (Fig. 7(a)). This point will be discussed in the following section. 4. Discussion 4.1 Spatial scale of the large fluctuation of heat balance What is the origin of the large fluctuation for the local time change of heat content during OMLET-89 and -901, estimated in Section 3? In order to estimate the order of spatial scale for the large fluctuation of oceanic heat convergence, we examined the horizontal eddy heat flux derived from the current velocities and temperatures. We assume that the vertical advection and vertical eddy heat flux are negligibly small, compared to the horizontal value. The oceanic heat convergence term F in Eq. (1) can then be expressed as 0 F C U T V T TU TV = p D x + y + ( ) ρ x y + ( ) dz ( 4) where U and V are the mean east and north components of velocity, T - the mean temperature, U and V are deviations from U and V, T - deviation from T, D - the depth of water column, i.e., 100 m. Since current velocity data are not available, except for 50 m depth in an actual calculation described below, we used current velocity and temperature data at 50 m depth obtained by ADCP. In the subtropical gyre, temperature is the dominant component determining the water density, and geostrophic balance holds over long temporal scale and large spatial scales. It is therefore expected that isotherms run parallel to the streamlines. Based on this assumption, F values and horizontal eddy heat fluxes are almost of the same order, then the orders of horizontal spatial scale x and y can be approximately estimated by comparing the oceanic heat divergence and horizontal eddy fluxes in Eq. (4). In practice, we can obtain the spatial scale of orders of km dividing the oceanic heat convergence (Table 1) by the horizontal eddy fluxes (Table 2). Although this estimate is quite rough, it is interesting to note that 624 H. Otobe et al.

7 Fig. 8. Power spectra of the time series of ( UT ) (thick line) and ( VT ) (thin line) for OMLET-89 (a), OMLET-901 (b) and OMLET-901 (c). Table 2. Mean values of covariance for horizontal velocity component and temperature over each observation period. Observation period ( ) ρcp D( T V ) ρcp D T U ( 10 6 W/m) OMLET OMLET OMLET this scale approximately corresponds to the scales of mesoscale eddies in the Kuroshio recirculation area (Ebuchi and Hanawa, 2000, 2001, 2003). Moreover, we did the spectral analysis of (U T ) and ( VT ) in order to estimate the temporal scale of eddy fluxes. The results are shown in Fig. 8. The spectral peak periods are not exactly the same as between ( UT ) and ( VT ), but a peak around 20 days can be observed in OMLET-89 and -901 excluding the OMLET-902 data. This period of about 20 days is consistent with time series of temperature, current velocities and heat balance shown in Figs. 3, 5 and 6, as described above. The reason for the diversity of the spectrum peak periods mentioned above might be reflected by the rather turbulent nature of the eddy fields, including the warm water outbreak phenomenon described by Toba et al. (1991). These orders of temporal scale and horizontal scale lead us to hypothesize that the large fluctuation of H/ t or F may be due to the effect on the horizontal eddy heat fluxes caused by the Kuroshio meander, the axis of which is located between about km west or north from the mooring system. 4.2 Relationship between variability of the heat balance and fluctuation of the Kuroshio axis We now focus on the fluctuation of the Kuroshio axis south of Japan. It is generally known that the Kuroshio takes two different stable paths to the south of Japan: one is rather straight, being the no-meander path, and the other is the large meander path (e.g. Stommel and Yoshida, 1972). Sekine and Toba (1981) and Sekine (1990) investigated the formation of the large Kuroshio meander after performing a numerical experiment based on the observational evidence. He showed that the existence of a small meander with a cold cyclonic eddy in the initial stage to the south of Kyushu is necessary for the formation of a large meander path. He suggested that the small meander can move eastward to the Shikoku Basin with growing amplitude due to the effect on the topography of both continental slope and the flat bottom in the Shikoku Basin if the Kuroshio has a volume transport greater than 60 Sv (1 Sv = 10 6 m 3 /s), and would finally become a large meander off the Enshu-Nada, south of Honshu. Time series of the Kuroshio path around the south of Japan from August 1989 to January 1991 are displayed in Fig. 9. It seems that this transition is a typical case of large meander formation. A small meander appeared to the south of Kyushu in early August 1989 when the mooring system for OMLET-89 had just been deployed. The small meander then moved eastward downstream and arrived one month later at 135 E off Shikoku, just north of the mooring system. Late in November 1989, the meander had grown remarkably in amplitude and remained off Enshu-Nada. Thereafter, the Kuroshio took a large meander path and developed with time as shown in Fig. 9. The Ten-Day Marine Report (1990) by JMA showed negative surface temperature anomalies in the Kuroshio area of N, E with a periodic fluctua- Variability of the Upper Ocean Heat Balance in the Kuroshio Region 625

8 Fig. 10. Power spectrum of sea level difference between Kushimoto and Uragami during the period of OMLET-89. Fig. 9. Maps of the Kuroshio path around the south of Japan in each ten day period from August 1989 to January 1991 (Ten- Day Marine Report, JMA). tion with a period of about a month and an amplitude of about 1 C from early May to early September 1989 (not shown here). The negative SST anomalies persisted from mid-september 1989 to mid-april The Ten-Day Marine Report also indicated that there were several warm and cold mesoscale eddies around the moored buoy in this period. These warm and cold mesoscale eddies might migrate around the moored buoy and cause the large temporal change of oceanic heat content. On the other hand, it is widely known that the difference between daily mean sea level at Kushimoto and that at Uragami located at the tip of the Kii Peninsula indicates the distance from the Kuroshio axis to the coast, especially when the Kuroshio takes the large meander path (e.g. Kawabe, 1987). We thus examined the periodicity of this fluctuation. Figure 10 shows the power spectrum of the time series of the heat convergence, F, during OMLET-89. There is a clear peak at around a 20-day period. This supports that the large fluctuation of oceanic heat convergence with a period of about 20 days during OMLET-89 was affected by the variation of the Kuroshio axis. These data suggest the hypothesis that the large fluctuation of the heat content for the first half of the period of OMLET-89 had been caused by the moving of those eddies due the eastward propagation of the small meander, which occurred to the south of Kyushu and is the process by which the Kuroshio large meander path is formed. It also seems that the relatively stable heat convergence for the second half of the period of OMLET-89 was reflected by the formation process of the Off-Shikoku Warm Water Eddy, which usually occurs when the Kuroshio takes a large meander path. A large heat convergence appeared in the two periods from April to early June and from late July to early September during OMLET-901, and in a single period during only mid-september in OMLET-902, as shown in Fig. 6. In those periods, the Off-Shikoku Warm Water Eddy was already formed and no notable warm or cold mesoscale eddies appeared (JMA, 1990). Toba et al. (1991) studied a conspicuous outbreak of warm water from the large meander region of the Kuroshio by combining a series of NOAA-AVHRR infrared images with our moored buoy data obtained during OMLET-88, and with hydrographic data obtained by two research vessels. The warm water was about 100 m deep in the 137 E section along the edge of the Off-Shikoku Warm Water Eddy. It was estimated that about twenty outbreaks of this kind in a year could compensate a large heat loss to the atmosphere above this oceanic region. It was unfortunate that the NOAA-AVHRR SST image during those periods when large heat convergence occurred was not available due to cloud cover in the region. We believe that a large heat convergence might occur by several arrivals of such a warm water outbreak. The local time change of heat content during almost the whole period of OMLET-902 balanced approximately with surface heat flux when the meander of the Kuroshio axis was more stable. 5. Summary In this paper we have presented the temperature and current velocity data of the upper ocean in the Shikoku Basin from the observations by a moored buoy during OMLET-89, -901 and -902 ( ). The heat bal- 626 H. Otobe et al.

9 ance in the upper 100 m layer was examined using the moored buoy data and the marine meteorological data of OBD/JMA. The variability of the heat balance has also been discussed in relation to the fluctuation of the Kuroshio axis around the south of Japan. The main results of this study can be summarized as follows: 1. A large heat convergences of 71, 79 and 13 W/m 2 on average were estimated over the three OMLET periods. 2. Large fluctuation of oceanic heat convergence/ divergence in a range of W/m 2 and with a period of days, occurred during the first half of OMLET-89, closely corresponding to the early stage in the formation process of a large meander path of the Kuroshio. A fluctuation with a period of days was also found in the fluctuation of the Kuroshio axis around the south of Japan. On the other hand, fluctuation of heat convergence was more stable in OMLET-901 and Variation of oceanic heat convergence in the Shikoku Basin was closely related to the fluctuation of the Kuroshio axis. Acknowledgements The authors wish to express their thanks to Professors Y. Toba of Tohoku University, H. Ichikawa of Kagoshima University and R. Sugimoto of the University of Tokyo for their valuable discussion and comments. Thanks are also extended to the crew of the R/V Hakuho- Maru and the R/V Tansei-Maru, the University of Tokyo and the T/S Keiten-Maru, Kagoshima University, for their hearty cooperation in the deployment and recovery of the moored buoy systems. Two anonymous reviewers provided valuable and constructive comments. This work was performed as a part of the Ocean Mixed Layer Experiment (OMLET) supported by the former Ministry of Education, Science and Culture under the Japanese World Climate Research Program (WCRP). References Ebuchi, N. and K. Hanawa (2000): Mesoscale eddies observed by TOLEX-ADCP and TOPEX/POSEIDON altimeter in the Kuroshio recirculation region south of Japan. J. Oceanogr., 56(1), Ebuchi, N. and K. Hanawa (2001): Trajectory of mesoscale eddies in the Kuroshio recirculation region. J. Oceanogr., 57(4), Ebuchi, N. and K. Hanawa (2003): Influences of mesoscale eddies on variations of the Kuroshio path south of Japan. J. Oceanogr., 59(1), Hsiung, J. (1986): Mean surface energy fluxes over the global ocean. J. Geophys. Res., 91, Hydrographic Department, Japan Maritime Safety Agency (1990): Oceanographic Prompt Report, No. 3. Japan Meteorological Agency ( ): The Ten-Day Marine Report. Kawabe, M. (1987): Spectral properties of sea level and time series of Kuroshio path variations. J. Oceanogr. Soc. Japan, 43, Kobe Marine Observatory (1990): Oceanographic Prompt Report of Kobe Marine Observatory, No Kondo, J. (1975): Air-sea bulk transfer coefficients in diabatic conditions. Bound.-Layer Meteor., 9, Kondo, J., T. Nakamura and T. Yamazaki (1991): Estimation of the solar and downward atmospheric radiation. Tenki, 38, (in Japanese). Kurasawa, Y., K. Hanawa and Y. Toba (1983): Heat balance of the surface layer of the sea at Ocean Weather Station T. J. Oceanogr. Soc. Japan, 39, Otobe, H. (1989): Radiation balance and heat budget at the ocean/atmosphere interface in the western North Pacific. J. Oceanogr. Soc. Japan, 45, Sekine, Y. (1990): A numerical experiment on the path dynamics of the Kuroshio with reference to the formation of the large meander path south of Japan. Deep-Sea Res., 37(3), Sekine, Y. and Y. Toba (1981): Velocity variation of the Kuroshio during formation of the small meander south of Kyushu. J. Oceanogr. Soc. Japan, 37, Stommel, H. and K. Yoshida (1972): Kuroshio Its Physical Aspects. Univ. Tokyo Press, Tokyo, 517 pp. Taira, K., S. Kitagawa, K. Uehara, H. Hachiya and T. Teramoto (1990): Direct measurements of mid-depth circulation in Shikoku Basin by tracking SOFAR floats. J. Oceanogr. Soc. Japan. 46, Taira, K., S. Kitagawa, H. Otobe and T. Asai (1993): Observation of temperature and velocity from a surface buoy moored in the Shikoku Basin (OMLET-88). J. Oceanogr. Soc. Japan, 49, Toba, Y., H. Kawamura, K. Hanawa, H. Otobe and K. Taira (1991): Outbreak of warm water from the Kuroshio south of Japan, A combined analysis of satellite and OMLET oceanographic data. J. Oceanogr. Soc. Japan, 47, Weare, B. C., P. T. Strub and M. D. Samuel (1981): Annual mean surface heat fluxes in the tropical Pacific Ocean. J. Phys. Oceanogr., 11, Variability of the Upper Ocean Heat Balance in the Kuroshio Region 627