Mesoscale Eddies Observed by TOLEX-ADCP and TOPEX/POSEIDON Altimeter in the Kuroshio Recirculation Region South of Japan

Transcription

1 Journal of Oceanography, Vol. 56, pp. 43 to Mesoscale Eddies Observed by TOLEX-ADCP and TOPEX/POSEIDON Altimeter in the Kuroshio Recirculation Region South of Japan NAOTO EBUCHI 1 and KIMIO HANAWA 2 1 Center for Atmospheric and Oceanic Studies, Graduate School of Science, Tohoku University, Aoba, Sendai , Japan 2 Department of Geophysics, Graduate School of Science, Tohoku University, Aoba, Sendai , Japan (Received 12 February 1999; in revised form 28 April 1999; accepted 28 April 1999) Mesoscale eddies in the Kuroshio recirculation region south of Japan have been investigated by using surface current data measured by an Acoustic Doppler Current Profiler (ADCP) installed on a regular ferry shuttling between Tokyo and Chichijima, Bonin Islands, and sea surface height anomaly derived from the TOPEX/POSEIDON altimeter. Many cyclonic and anticyclonic eddies were observed in the region. Spatial and temporal scales of the eddies were determined by lag-correlation analyses in space and time. The eddies are circular in shape with a diameter of 500 km and a temporal scale of 80 days. Typical maximum surface velocity and sea surface height anomaly associated with the eddies are cm s 1 and 15 cm, respectively. The frequency of occurrence, temporal and spatial scales, and intensity are all nearly the same for the cyclonic and anticyclonic eddies, which are considered to be successive wave-like disturbances rather than solitary eddies. Phase speed of westward propagation of the eddies is estimated as 6.8 cm s 1, which is faster than a theoretical estimate based on the baroclinic first-mode Rossby wave with or without a mean current. The spatial distribution of sea surface height variations suggests that these eddies may be generated in the Kuroshio Extension region and propagate westward in the Kuroshio recirculation region, though further studies are needed to clarify the generation processes. Keywords: Mesoscale eddy, Acoustic Doppler Current Profiler (ADCP), satellite altimetry, Kuroshio recirculation. 1. Introduction Mesoscale eddies around the Kuroshio and Kuroshio Extension regions are thought to play important roles in transportation of mass, momentum, energy, heat and salinity in the subtropical gyre. There have been several studies of the generation, propagation and decay of these eddies using hydrographic observations and drifting buoys (e.g., Kawai, 1972; Kitano, 1975; Cheney et al., 1980; Mizuno and White, 1983). However, spatial coverage and temporal frequency of in-situ measurements such as hydrographic observations and drifting buoys are very limited. Infrared images from satellites have also helped trace these mesoscale eddies (e.g., Yasuda et al., 1992), but the images cannot be obtained under cloudy conditions. Corresponding author ebuchi@ocean.caos.tohoku.ac.jp Copyright The Oceanographic Society of Japan. Satellite altimetry is a powerful tool for studying oceanic variations, including mesoscale eddies, since it can provide sea surface dynamic topography over the global ocean with high spatial resolution and temporal frequency and can penetrate clouds. Altimeter data have been utilized to investigate the spatial scale, generation, propagation and kinematic energy distribution of mesoscale eddies in the Kuroshio Extension region (e.g., Tai and White, 1990; Qiu et al., 1991; Ichikawa and Imawaki, 1994; Aoki et al., 1995; Aoki and Imawaki, 1996). In the present paper, time series of surface current velocities from the Tokyo-Ogasawara Line Experiment (TOLEX) - Acoustic Doppler Current Profiler (ADCP) monitoring are combined with sea surface height data from the TOPEX/POSEIDON altimeter to investigate the mesoscale eddies in the Kuroshio recirculation region south of Japan. We investigate the detailed structure of the surface current velocity field associated with eddies using data of high temporal frequency from the TOLEX- 43

2 ADCP monitoring. The TOPEX/POSEIDON altimeter provides sea surface height data with wide spatial coverage and high temporal frequency to investigate the spatial scale and propagation of eddies and the distribution of eddy energy. TOLEX is a program operated by the Physical Oceanography Group, Tohoku University and the Japan Marine Science and Technology Center (JAMSTEC) to monitor the Kuroshio Current System south of Japan by using the ferry Ogasawara Maru, which regularly shuttles between Tokyo and the Ogasawara (Bonin) Islands. TOLEX-ADCP monitoring has been conducted since February 1991 using a 150 khz-type ADCP (Hanawa and Yoshikawa, 1993; Hanawa et al., 1996). By analyzing the preliminary data, Hanawa et al. (1996) reported observations of several cyclonic and anticyclonic eddies with scales of km in diameter and passage time of 2 4 months. Using the same ferry, XBT monitoring of the temperature profile has also been carried out every two months since 1988 (Yoshikawa et al., 1996). TOPEX/POSEIDON (hereafter T/P) is a joint U.S./ French space mission to study the world ocean circulation. The satellite was launched in August The mission carries two altimeters, the National Aeronautics and Space Administration (NASA) dual-frequency TOPEX altimeter, and the advanced, experimental solid-state POSEIDON altimeter, designed by Centre National d Etudes Spatiales (CNES). The satellite is taking exact repeat orbits of period 9.91 days. The ground track spacing is 315 km at the equator and about 200 km at mid latitude. Details of the mission, satellite, and instruments have been described by Fu et al. (1994). Ebuchi and Hanawa (1995) compared surface current velocity variations derived from the TOPEX altimeter data with those from the TOLEX-ADCP data, and reported that the two data sets agree well, with an rms difference of 9.5 cm s 1. The TOLEX-ADCP and T/P altimeter data are briefly described in Section 2. The nature of mesoscale eddies south of Kuroshio, including typical feature of the eddies, their spatial and temporal scales and propagating speed derived from the ADCP and altimeter data are described in Section 3. Generation and propagation of the eddies are discussed in Section 4, followed by a brief summary in Section Data 2.1 TOLEX-ADCP data TOLEX-ADCP monitoring was started in 1991 using the ferry Ogasawara Maru (110 m long, 3553 tons), operated by Ogasawara Kaiun Company (Hanawa and Yoshikawa, 1993; Hanawa et al., 1996). A 150 khz-type ADCP (RD Instruments, U.S.A.) was installed on the ferry. Figure 1 shows the ship track of the ferry together with the ground tracks of the T/P altimeter analyzed in the present study. This ferry regularly shuttles 58 round trips a year between Tokyo and Chichijima, which belongs to the Ogasawara (Bonin) Islands: that is, 116 cross sections are obtained in a year. By using the ADCP, a current velocity profile in vertical bins of 16 m from the surface to about 350 m depth was obtained at 48 spatial grid points located every ten minutes in latitude from 35 N to N along the ship track (Fig. 1). Detailed procedures for the ADCP data Fig. 1. Locations of the ship track of TOLEX-ADCP monitoring (solid circles) and ground tracks of the TOPEX/POSEIDON altimeter analyzed in the present study (dots). Thick outline and thin outline rectangles are explained in Subsection N. Ebuchi and K. Hanawa

3 processing including data averaging, ship velocity measurement by the Global Positioning System (GPS) and angle correction, have been described by Hanawa et al. (1996). In the present study, current velocity at 50 m depth was utilized as the most representative surface current, since the ADCP data are instantaneous velocities and include ageostrophic components, such as tidal current, internal waves and wind-driven surface current. Kizu et al. (1999) applied a harmonic analysis to remove the tidal current components from the ADCP data. The tide-free surface current vectors they provided were linearly interpolated in time to a regular 5-day interval to produce the data used in this study. The period of the data extends from April 1991 to November However, the total length of the data is about 5 years, since there are some intermittent gaps in the data due to problems with the ADCP observations and bad weather conditions. 2.2 TOPEX/POSEIDON altimeter data In the present study we analyzed TOPEX/ POSEIDON Sea Level Anomaly (SLA) altimeter products for a period of 5 years from October 2, 1992 to October 11, 1997 (cycles 2 186). These products are supplied by the Collecte Localisation Satellites (CLS) Space Oceanography Division, Toulouse, France (AVISO/ Fig. 2. An example of time series of surface current velocity vectors observed by the TOLEX-ADCP monitoring at a depth of 50 m. Mesoscale Eddies in the Kuroshio Recirculation Region 45

4 Altimetry, 1996). They are generated, using the conventional repeat-track analysis method, from the altimeter sea surface heights corrected for instrumental errors, environment perturbations (wet tropospheric, dry tropospheric and ionospheric effects), ocean wave influence (sea state bias), tide influence (ocean tide, earth tide and pole tide) and inverse barometric effect. The sea surface height anomaly is resampled every 7 km for a given track and for each cycle. Locations of the ground tracks analyzed in this study are shown in Fig. 1. Fig. 3. (a) Vector mean velocity, (b) mean zonal (circles) and meridional (triangles) velocity components, and (c) their standard deviations (thick line for zonal component, thin line for meridional component, and dashed line for root-mean-square of the zonal and meridional deviations) calculated for the whole data period. Fig. 4. Ensemble averages of rotary spectra of the surface current vectors calculated from the time series observed at latitudes from 27 to 30 N. Thick and thin lines represent clockwise (anticyclonic) and counter-clockwise (cyclonic) components, respectively. An error bar indicating the 95% confidence level is shown for a particular point at a peak. 46 N. Ebuchi and K. Hanawa

5 Fig. 5. An example of the filtered time series of the surface current vectors (see text). 3. Results 3.1 Mesoscale eddies observed along the TOLEX monitoring line In this section, features of mesoscale eddies which passed across the TOLEX line are discussed by analyzing the time series of surface current velocity derived from the TOLEX-ADCP monitoring. Figure 2 shows an example of the time series of velocity vectors at 50 m depth. We can clearly recognize the position of the Kuroshio flowing eastward/northeastward within latitudes from 32 to 35 N. Though the Kuroshio vigorously and quickly changes its position over the ship track, it has two typical paths which are located around 34 and 32.5 N. These latitudes correspond to positions between Ohshima and Miyakejima, and south of Hachijojima (Hanawa et al., 1996). It is also seen that many cyclonic and anticyclonic eddies frequently passed across the ship track in a region south of the Kuroshio. These eddies are the main focus of the present study. Figure 3 shows vector mean velocity, mean zonal and meridional velocity components, and their standard deviations calculated for the whole period covered by the data. There exist two peaks of velocity located at 32.5 and 34 N which correspond to the two frequent paths of the Kuroshio axis as seen in Fig. 2. Reflecting large vari- Mesoscale Eddies in the Kuroshio Recirculation Region 47

6 ations of the Kuroshio axis, the standard deviations also show peaks around the location of the typical Kuroshio paths. Around 29 N, there exists a mean westward current with a maximum velocity of about 5 cm s 1. This westward current is considered to be the Kuroshio recirculation. In the profiles of standard deviations a broad peak is discernible around 29 N, which is located slightly north of the Kuroshio recirculation mentioned above. This peak might correspond to activities of the eddies south of the Kuroshio seen in Fig. 2. Figure 4 shows rotary spectra of the surface current vectors, which are calculated from the time series observed at latitudes from 27 to 30 N as ensemble averages of 18 raw spectra. Judging from Figs. 2 and 3, this region is unaffected by fluctuation of the Kuroshio axis. The sharp spikes at a period of about 15 days may be due to incompleteness of the tide removal procedures (Kizu et al., 1999). There are broad peaks at periods from 45 to 180 days in both the clockwise and counter-clockwise components. These peaks are considered to represent eddy activities south of the Kuroshio, implying that both cyclonic and anticyclonic eddies exist and have almost the same temporal scale. The kinematic energy in both the cyclonic and anticyclonic components shows a similar level, though the cyclonic energy in this band is about 30% greater than the anticyclonic energy in Fig. 4. It is Fig. 6. Example of an observed cyclonic eddy. (a) Surface current vectors observed by the ADCP and (b) a temperature profile observed by XBTs (unit C). The origins of the horizontal and vertical axis of the panel (a) are September 12, 1994 (UT) and 29.8 N. The XBT temperature profiles were observed on September 15 16, 1994 (UT). 48 N. Ebuchi and K. Hanawa

7 also shown that such eddy activities are the dominant oceanic variation in this region. In order to focus on the eddy activities, a recursivetype band-pass filter with cut-off periods of and 40.3 days (determined by the results in Fig. 4) and a cutoff gain of 45 db per octave is applied to the time series of the surface current vectors. An example of the filtered time series is shown in Fig. 5. The period is the same as in Fig. 2, though only the southern part from 27 to 32 N is shown. Several cyclonic (e.g., around day 376, 496, 596, 721, 881) and anticyclonic (e.g., around day 316, 426, 556, 646, 831) eddies are discernible. All centers of the eddies are located between 28 and 30 N. Figures 6 and 7 show examples of observed cyclonic and anticyclonic eddies, respectively, together with temperature profiles derived from the TOLEX-XBT monitoring (Yoshikawa et al., 1996). The cyclonic eddy shown in Fig. 6(a) has a diameter of a few hundred kilometers and a maximum velocity around 50 cm s 1, and stays for more than two months on the ship track. Location of the center of the eddy in the current field coincides with that in the vertical section of temperature (Fig. 6(b)). The vertical structure associated with the eddy extends from the subsurface layer to below 800 m, and the 15 C isotherm rises about 250 m at the center of the eddy (30 N) compared to the surrounding area. Fig. 7. Example of an observed anticyclonic eddy. (a) Surface current vectors observed by the ADCP and (b) a temperature profile observed by XBTs (unit C). The origins of the horizontal and vertical axis of the panel (a) are October 7, 1992 (UT) and 30.3 N. The XBT temperature profiles were observed on September 27 28, 1992 (UT). Mesoscale Eddies in the Kuroshio Recirculation Region 49

8 Fig. 8. Composite surface velocity profiles for (a) cyclonic and (b) anticyclonic eddies. Number of cyclonic and anticyclonic eddies which passed across the TOLEX line during the whole period of the ADCP data lasting about 5 years and which are used to calculate the composite profiles are 15 and 17, respectively. Fig. 9. Results of lag-correlation analyses in time and space applied to the filtered velocity components observed at latitudes from 27 to 30 N. The auto-correlation functions with temporal lag for the (a) zonal and (b) meridional velocity components, and the cross correlation functions with spatial lag along the ship track for the zonal (c) and meridional (d) velocity components. Latitude difference in the horizontal axis of panels (c) and (d) is taken northward positive. 50 N. Ebuchi and K. Hanawa

9 Fig. 10. An ensemble average of power spectra of the altimeter-derived sea surface height anomaly calculated in an area between 27 and 30 N, and 135 and 150 E, which is indicated by a solid outline rectangle in Fig. 1. An error bar indicating the 95% confidence level is shown for a particular point at a peak. The anticyclonic eddy in Fig. 7(a) shows slightly larger scales in time and space than those of the cyclonic eddy in Fig. 6(a), though its maximum velocity is weaker. Collocation of the eddy centers in the current vector field and vertical temperature section (Fig. 7(b)) is also discernible. The vertical structure is less steep than that shown in Fig. 6(b), but it reaches below 800 m. Displacement of the 15 C isotherm reaches 200 m at the center of the eddy (30 N). Visual observation revealed 15 cyclonic and 17 anticyclonic eddies which passed across the TOLEX line during the whole period of the ADCP data, covering about 5 years. Composite profiles of these cyclonic and anticyclonic eddies are calculated and shown in Figs. 8(a) and (b), respectively. The centers of these eddies were determined by identifying the time and position where the velocity minimum exists together with maximum velocities of opposite directions south and north of the minimum. Then the centers were aligned to make up the composites. Mean latitudes of these centers were 29.7 and 29.3 N for the cyclonic and anti-cyclonic eddies, respectively. Both composite eddies show very similar and symmetrical features. The along-track spatial scales (the vertical axis) and temporal scales (the horizontal axis) are estimated to be a few hundred kilometers and about two months. The typical magnitude of the maximum velocity is a few tens of centimeters per second. It is suggested that the frequency of occurrence, temporal and spatial scales, and intensity of the eddies are almost identical for the cyclonic and anticyclonic eddies. This is consistent with the result from the rotary spectrum shown in Fig. 4. In regions south of the Kuroshio or Kuroshio Extension, most studies (e.g., Cheney et al., 1980; Mizuno and White, 1983; Ichikawa and Imawaki, 1994) have paid attention to the generation and propagation of cyclonic (cold) eddies. However, the results of this study show that both cyclonic and anticyclonic eddies are commonly observed in this region. In order to estimate spatial and temporal scales of the eddies, lag-correlation analysis in time and space is applied to the filtered velocity components observed at latitudes from 27 to 30 N. The results are shown in Fig. 9. Panels (a) and (b) show the auto-correlation functions with temporal lag for the zonal and meridional components, respectively. It is shown that the single-side temporal scale of the eddies is 40 days, where the correlation coefficient takes a minimum value. Thus, the typical temporal scale of the eddies is estimated as 80 days. By considering the time scale (80 days) and occurrence frequency of the eddies (3.4 eddies/year for each of the cyclonic and anticyclonic eddies), these eddies are almost always observed in this region. This suggests that the cyclonic and anti-cyclonic eddies are not solitary eddies but successive wave-like disturbances, though the temporal lag correlation functions shown in panels (a) and (b) do not show significant second peaks. This conclusion is supported by the fact that preceding and succeeding eddies with opposite polarity are discernible in the composite current velocity profiles shown in Fig. 8. Panels (c) and (d) show the cross correlation function with spatial lag along the ship track for the zonal and meridional components, respectively. The spatial scales of the eddies, which are defined by minima of the corre- Mesoscale Eddies in the Kuroshio Recirculation Region 51

10 Fig. 11. An example of the filtered time series of the T/P-derived sea surface height anomaly along Path 010 (see Fig. 1). Shaded and hatched areas correspond to the cyclonic and anticyclonic eddies, respectively (at latitude south of 32 N). Temporal interval of the repeat cycle of the T/P altimeter is 9.91 days and cycle 2 of Path 010 was performed on October 2, 1992, or day 641 in the time axis used in Figs. 2 and 5. lation coefficient, are 330 and 560 km for the zonal (westeast, u) and meridional (north-south, v) components, respectively. These temporal and spatial scales are consistent with the results of the rotary spectrum in Fig. 4 and the composite feature of the eddies in Fig. 8, and also coincide with those reported by Imawaki (1981). 3.2 Spatial and temporal scales and propagation of the mesoscale eddies observed by the TOPEX/ POSEIDON altimeter In this section, the sea surface height anomalies ob- served by T/P altimeter are analyzed to investigate further the spatial and temporal scales of the eddies and to explore propagation of the eddies. The power spectrum of the altimeter-derived sea surface height anomaly was calculated as an ensemble average in an area between 27 and 30 N, and 135 and 150 E, which is indicated by a solid outline rectangle in Fig. 1. The result is shown in Fig. 10. A peak can be seen at a period of one year, which represents the annual variation of the sea surface height including the steric change of the surface layer. There also exists a broad band of energy with several peaks at periods between 45 to 180 days. This period range coincides with the peaks in the rotary spectrum of the surface current shown in Fig. 4, and these are considered to represent eddy activities in this area. As in the previous subsection, a recursive-type bandpass filter with cut-off periods of and 38.4 days was applied to the time series of the sea surface height anomaly to extract the eddy activities. An example of the filtered time series along Path 010 (see Fig. 1) is shown in Fig. 11, where several cyclonic and anticyclonic eddies (shaded in the figure) are discernible. Larger peaks at latitudes north of 31 N are caused by fluctuations of the Kuroshio axis, and are not considered in this paper. By counting eddies which were located south of 31 N and whose surface height anomaly exceeded ±10 cm, the number of such cyclonic and anticyclonic eddies which passed across Path 010 during the whole 5-year period of the T/P data was determined to be 17 and 19, respectively. These numbers are close to the frequency of occurrence determined from the ADCP data in the previous subsection. Composite profiles of sea surface height anomaly are calculated from these cyclonic and anticyclonic eddies and shown in Figs. 12(a) and (b), respectively. Centers of the eddies were determined by choosing time and position where sea surface height anomaly showed a local maximum or minimum both in time and space. Then these centers were aligned to make up the composites. As for the composite profiles of surface current in Fig. 8, both types of eddy show very similar and symmetrical features. Typical sea surface height anomalies associated with the eddies are about 15 cm for both the cyclonic and anticyclonic eddies. The along-track spatial scales are estimated to be about 500 km. Typical magnitudes of cross-track velocity calculated from the gradient of the composite sea surface height are about cm s 1. The frequency of occurrence, along-track spatial scale, and intensity of the eddies are almost identical for the cyclonic and anticyclonic eddies. This is consistent with results from the surface current data presented in the previous subsection. In previous studies in the Kuroshio Extension and Gulf Steam regions (e.g., Richardson et al., 1979; Ring Group, 1981; Mizuno and White, 1983; Brown et al., 1986; Yasuda et al., 1992; 52 N. Ebuchi and K. Hanawa

11 Fig. 12. Composite profiles of sea surface height anomaly and cross-track velocity for (a) cyclonic and (b) anticyclonic eddies. Numbers of cyclonic and anticyclonic eddies which passed across Path 010 in the whole 5-year period of the T/P data and which are used to calculate the composite profiles, are 17 and 19, respectively. Thin lines in upper panels indicate standard deviations from the mean sea surface height anomalies. Ichikawa and Imawaki, 1994), most attention has been paid to the cyclonic (anticyclonic) or cold (warm) eddies south (north) of the jet. In the present study, however, it is found that both the cyclonic and anticyclonic eddies have almost the same frequency in the region south of the Kuroshio. As in the previous subsection, lag-correlation analysis in time and space was applied to the filtered sea surface height anomaly to estimate spatial and temporal scales of the eddies. The results are shown in Fig. 13. Panel (a) shows the auto-correlation functions with temporal lag for the filtered sea surface height anomaly along Path 010 between 27 and 30 N. It is shown that the single-side temporal scale of the eddies is 40 days, which coincides with the result in Fig. 9(a). Panel (b) shows the cross correlation function with spatial lag along Path 010 between 27 and 30 N. The along-track spatial scale of the eddies is estimated to be 500 km. Cross correlation coefficients with zonal spatial lag are calculated using data along the several T/P tracks in the region between 27 and 30 N (solid outline rectangle in Fig. 1). The zonal cross correlation function for Path 010 is shown in Fig. 13(c). There exist two clusters of data points at about ±250 km representing correlation coefficients with data along Paths 188 and 086, which are parallel to Path 010 (see Fig. 1). Small gaps of the correlation function in the right and left of the clusters are due to change of paths used to calculate the correlation coefficients. The zonal scale of the eddies is estimated to be 500 km. Since the alongtrack and zonal scales of the eddies coincide with each other, these eddies are considered to have a circular shape with radius of about 500 km, which also coincides with the results presented in the previous subsection. In order to exhibit the propagation of the eddies, a time-longitude diagram was composed from the filtered sea surface heights in a region between 28 to 29 N, which is indicated by a thin outline rectangle in Fig. 1. Average sea surface height anomalies in this latitude band along each track were calculated and are shown in Fig. 14. It is shown that several eddies with height anomalies greater than ±10 cm propagate westward. A discontinuous feature is also discernible around Paths 010 and 253, where bottom topography effects from the Izu Ridge are expected. The mean westward phase speed of the eddies is estimated to be 6.8 cm s 1, by calculating temporal lag correlation between the neighboring paths. This value of the phase speed is consistent with an estimate dividing the spatial scale of 500 km by the temporal scale of 80 days, obtained from the analyses of both the ADCP and T/P data. The propagating speed is also similar to that estimated by previous studies in this region (e.g., Imawaki, 1981; Aoki et al., 1995; Aoki and Imawaki, 1996). Further discussion on the phase speed of the eddies will be given in the following section. Spatial distribution of eddy activity, represented by root-mean-squared (rms) amplitude of the filtered sea Mesoscale Eddies in the Kuroshio Recirculation Region 53

12 Fig. 13. Results of lag-correlation analyses in time and space applied to the filtered sea surface height anomalies along Path 010 at latitudes from 27 to 30 N. (a) The auto-correlation functions with temporal lag, (b) the cross correlation function with spatial lag, and (c) cross correlation coefficients with zonal spatial lag (see text). Distance in the horizontal axis of panels (b) and (c) is taken northward and eastward positive, respectively. surface height anomaly, is shown in Fig. 15. The rms amplitude was calculated in each of the data grids along all the paths in Fig. 1 and was averaged along each path over 1 in latitude. The contour plot in Fig. 15 was drawn by using theses points. High eddy activity is discernible along the Kuroshio and Kuroshio Extension. A region of local maximum extends from the Kuroshio Extension (around 31.5 N, 150 E) to the Kuroshio recirculation region (29.5 N, 144 E) and south of Shikoku (28.5 N, 135 E). This might be the propagation route of the cyclonic and anticyclonic eddies investigated in this study. From this result, it could be speculated that these eddies are generated in the Kuroshio Extension region, which is the most energetic region in the North Pacific (Tai and White, 1990; Qiu et al., 1991; Aoki and Imawaki, 1996), and then propagate westward along the Kuroshio recirculation to south of Japan. South of Kyushu, the eddies might disappear by merging into the Kuroshio. It can be expected that the propagation route of the eddies is affected by bottom topography effects and the location of the Kuroshio recirculation. Further discussion of the generation of the eddies will be given in the following section. 4. Discussion In the previous section, the phase speed of westward propagation of the cyclonic and anticyclonic eddies in the Kuroshio recirculation region south of Japan is estimated to be 6.8 cm s 1. The phase speed of the westward propagation of mesoscale disturbances in the Gulf Stream and Kuroshio Extension regions have been discussed by comparison with the phase speed of the baroclinic first- 54 N. Ebuchi and K. Hanawa

13 Fig. 14. Time-longitude diagram of the filtered sea surface heights in a region between 28 and 29 N, which is indicated by a thin outline rectangle in Fig. 1. Horizontal axis represents locations of the T/P paths along the latitude band. Fig. 15. Spatial distribution of root-mean-squared (rms) amplitudes of the filtered sea surface height anomaly (unit cm). Mesoscale Eddies in the Kuroshio Recirculation Region 55

14 mode Rossby wave (e.g., Richardson, 1980; Mizuno and White, 1983; Brown et al., 1986; Qiu et al., 1991; Jacobs et al., 1993; Aoki et al., 1995). Theoretically, the phase speed of the baroclinic first-mode long Rossby wave is expressed as, Cp = βa 2, () 1 where β is the meridional gradient of the Coriolis parameter, a is the first-mode internal Rossby deformation radius. By substituting a = 45 km from Emery et al. (1984), the phase speed, Cp, is estimated to be 4.0 cm s 1, which is only 60% of the result reported in the previous section. Several previous studies have pointed out that the theory of the linear baroclinic first-mode long Rossby wave underestimates the phase velocity of the mesoscale disturbances (e.g., Richardson, 1980; Brown et al., 1986; Olson and Evans, 1986; Aoki et al., 1995; Aoki and Imawaki, 1996). Aoki and Imawaki (1996) considered a simple reduced-gravity model which includes effects of finite wavelengths and Doppler shift by mean current in the upper layer. They used a formula proposed by Kessler (1990), in which the zonal phase velocity of the baroclinic Rossby wave in the quasi-geostrophic system at mid-latitude is expressed as, 2 2 ( ) ( ) k l Cpx = β U ( a + k + l ) a + k + l ( 2) where k and l are the zonal and meridional wavenumbers, respectively, and U is the mean eastward current velocity. The second term on the right-hand side represents the Doppler shift by the mean current. By using Eq. (2), let us try to estimate the phase velocity of the eddies investigated in this study. The wavenumbers k and l are given by the spatial scale of 500 km found in the previous section. the eastward mean current U is estimated as 5 cm s 1 from the result shown in Fig. 3. By substituting these values into Eq. (2), the phase velocity is estimated as 4.4 cm s 1, which is still slower than the result shown in the previous section, though the effects of finite wavelengths and Doppler shift by mean current may explain some portion of the underestimation by Eq. (1). Aoki and Imawaki (1996) also reported that their observed phase speed of mesoscale disturbances was faster than the theoretical estimates from Eq. (2). It might be expected that a nonlinear effect of finite amplitude, which is not taken into account by linear theory of the baroclinic Rossby wave, may affect the phase velocity. Further studies exploring such an effect are necessary to explain the underestimation of the phase speed., For the generation of the cyclonic and anticyclonic eddies, the result shown in Fig. 15 suggests that these eddies are generated in the Kuroshio Extension region and then propagate westward along the Kuroshio recirculation to the region south of Japan. Studies analyzing sea surface height observed by altimeters have reported that the maximum of eddy kinetic energy is located in the Kuroshio Extension region (e.g., Tai and White, 1990; Aoki and Imawaki, 1996). Several studies have been conducted on the generation of cold and warm rings in the region (e.g., Kawai, 1972; Cheney, 1977; Mizuno and White, 1983; Yasuda et al., 1992; Ichikawa and Imawaki, 1994). By using data from the Geosat altimeter, Ichikawa and Imawaki (1994) demonstrated detachment of a cyclonic eddy from the Kuroshio Extension. A southward meander pinched off to form the cyclonic eddy. As they showed, it can readily be assumed that the cyclonic eddies observed in the present study are also generated by pinch-off of meanders in the Kuroshio Extension region. However, this process cannot explain the generation of anticyclonic eddies, which are observed as frequently and significantly as the cyclonic eddies in this study. Further studies concerning the generation area and processes of these eddies are necessary to solve this problem. As demonstrated by the previous studies and also by the present study, the spatial distribution of the sea surface heights observed by satellite altimeters is an essential tool for such studies. 5. Summary In the present study, surface current velocity observed by an ADCP installed on a regular ferry and sea surface height anomaly derived from the T/P altimeter were analyzed to investigate mesoscale eddies in the Kuroshio recirculation region south of Japan (Fig. 1). The time series of surface currents from the TOLEX-ADCP monitoring (Fig. 2) were utilized to present detailed features of the vector surface current field associated with the eddies. The T/P data were utilized to demonstrate the spatial distribution and westward propagation of eddies. Power spectrum analyses of both of the surface current and sea surface height anomaly (Figs. 4 and 10) showed that there exists a broad peak in a range of periods from 45 to 180 days, which represents eddy activity in this region. According to this result, band-pass filters were designed and applied to both of the time series in order to extract variations associated with the eddies. Spatial and temporal scales of the eddies were determined from the filtered data (Figs. 5 and 11) by lag-correlation analyses in space and time (Figs. 9 and 13). It is found that the eddies are circular in shape with a typical diameter of 500 km and their temporal scale is 80 days. Composite features of the cyclonic and anticyclonic eddies were produced (Figs. 8 and 12). Typical maximum 56 N. Ebuchi and K. Hanawa

15 velocity and sea surface height anomaly associated with the eddies are found to be cm s 1 and 15 cm, respectively. It is also pointed out that the frequency of occurrence, temporal and spatial scales, and intensity are all nearly the same for the cyclonic and anticyclonic eddies. These eddies are considered to be successive wavelike disturbances rather than solitary eddies. From the time-longitude diagram of the filtered sea surface height anomaly (Fig. 14), the phase velocity of westward propagation of the eddies is estimated as 6.8 cm s 1, which is faster than the theoretical estimation based on the baroclinic first-mode Rossby wave. The spatial distribution of sea surface height variations (Fig. 15) suggests that these eddies might be generated in the Kuroshio Extension region and propagate westward in the Kuroshio recirculation, though further studies are needed to clarify the generation processes. Acknowledgements We are grateful to the captain and crew of the ferry Ogasawara Maru and Ogasawara Kaiun Co. for their cooperation to the TOLEX monitoring. The SLA altimeter products utilized in the present study are supplied by the CLS Space Oceanography Division, Toulouse, France. The present study is partially supported by Grants-in-Aid for Scientific Research (project No ) and for an International Cooperative Research Project (project No ), and the NEAR-GOOS (North East Asian Regional-Global Ocean Observing System) Project, from the Ministry of Education, Science and Culture, Japan. References Aoki, S. and S. Imawaki (1996): Eddy activities of the surface layer in the western North Pacific detected by satellite altimeter and radiometer. J. Oceanogr., 52, Aoki, S., S. Imawaki and K. Ichikawa (1995): Baroclinic disturbances propagating westward in the Kuroshio Extension region as seen by a satellite altimeter and radiometers. J. Geophys. Res., 100, AVISO/Altimetry (1996): AVISO User Handbook for Sea Level Anomaly Altimeter Products (edition 2.0), AVI-NT CN, CLS Space Oceanography Division, Toulouse, France. Brown, O. B., P. C. Comillon, S. R. Emmerson and H. M. Carle (1986): Gulf Stream warm rings: A statistical study of their behavior. Deep-Sea Res., 33, Cheney, R. E. (1977): Synoptic observations of the oceanic frontal system east of Japan. J. Geophys. Res., 82, Cheney, R. E., P. L. Richardson and K. Nagasaka (1980): Tracking a Kuroshio cold ring with a free-drifting surface buoy. Deep-Sea Res., 27, Ebuchi, N. and K. Hanawa (1995): Comparison of surface current variations observed by TOPEX altimeter with TOLEX- ADCP data. J. Oceanogr., 51, Emery, W. J., W. G. Lee and L. Magaard (1984): Geographic and seasonal distributions of Brunt-Väisälä-frequency and Rossby radii in the North Pacific and North Atlantic. J. Phys. Oceanogr., 14, Fu, L. L., E. J. Christensen, C. A. Yamarone, M. Lefebvre, Y. Menard, M. Dorrer and P. Escudier (1994): TOPEX/ POSEIDON mission overview. J. Geophys. Res., 99, 24,369 24,381. Hanawa, K. and Y. Yoshikawa (1993): TOLEX-XBT/ADCP monitoring. Ext. Abst. Int. WCRP Symp. Clouds and Ocean in Climate, Nagoya, Japan, Hanawa, K., Y. Yoshikawa and T. Taneda (1996): TOLEX- ADCP monitoring. Geophys. Res. Lett., 23, Ichikawa, K. and S. Imawaki (1994): Life history of a cyclonic ring detached from the Kuroshio Extension as seen by the Geosat altimeter. J. Geophys. Res., 99, 15,953 15,966. Imawaki, S. (1981): Vertical structure and horizontal scale of the mesoscale baroclinic variability in the western North Pacific. La mer, 19, Jacobs, G. A., W. J. Emery and G. H. Born (1993): Rossby waves in the Pacific Ocean extracted from Geosat altimeter data. J. Phys. Oceanogr., 23, Kawai, H. (1972): Hydrography of the Kuroshio Extension. p In Kuroshio; Its Physical Aspects, ed. by H. Stommel and K. Yoshida, Univ. Tokyo. Press, Tokyo, Japan. Kessler, W. B. (1990): Observation of long Rossby waves in the northern tropical Pacific. J. Geophys. Res., 95, Kitano, K. (1975): Some properties of the warm eddies generated in the confluence zone of the Kuroshio and Oyashio Currents. J. Phys. Oceanogr., 5, Kizu, S., T. Taneda and K. Hanawa (1999): Estimation of tidal constituents from the TOLEX-ADCP data. J. Oceanogr. (submitted). Mizuno, K. and W. B. White (1983): Annual and interannual variability in the Kuroshio current system. J. Phys. Oceanogr., 13, Olson, D. B. and R. H. Evans (1986): Rings of the Agulhas Current. Deep-Sea Res., 33, Qiu, B., K. A. Kelly and T. M. Joyce (1991): Mean flow and variability in the Kuroshio Extension from Geosat altimetry data. J. Geophys. Res., 96, 18,491 18,507. Richardson, P. L. (1980): Gulf Stream ring trajectories. J. Phys. Oceanogr., 10, Richardson, P. L., C. Maillard and T. B. Stanford (1979): The physical structure and life history of cyclonic Gulf Stream ring Allen. J. Geophys. Res., 84, Ring Group (1981): Gulf Stream cold-core rings: Their physics, chemistry and biology. Science, 212, Tai, C. K. and W. B. White (1990): Eddy variability in the Kuroshio Extension as revealed by Geosat altimetry; Energy propagation away form the jet, Reynolds stress, and seasonal cycle. J. Phys. Oceanogr., 20, Yasuda, I., K. Okuda and M. Hirai (1992): Evolution of a Kuroshio warm-core ring: Variability of the hydrographic structure. Deep-Sea Res., 39, suppl. 1, S131 S161. Yoshikawa, Y., K. Ando, H. Mitsudera, K. Muneyama and K. Hanawa (1996): Data report: Tokyo-Ogasawara Line Experiment XBT monitoring, JAMSTEC Data Report, JAMSTEC, Yokosuka, Japan, 102 pp. Mesoscale Eddies in the Kuroshio Recirculation Region 57