Short-term fluctuations south of Japan and their relationship with the Kuroshio path: 8- to 36-day fluctuations
- 857 Downloads
To detect short-term fluctuations south of Japan, we applied wavelet analysis to ocean reanalysis data of the Japan Coastal Ocean Predictability Experiment 2 with a horizontal resolution of 1/36°. It was found that fluctuations of the 8- to 36-day period band appear as frontal waves in the Kuroshio Current. The amplitude of the fluctuations increases toward the downstream of Cape Shionomisaki. The fluctuations have a wavelength of about 300 km, and the signals propagate eastward. The fluctuations of the 8- to 36-day period band are stronger during the period of the nearshore non-large-meander Kuroshio path than during the period of the offshore non-large-meander Kuroshio path. We suggest that the 8- to 36-day fluctuation is a result of the instability of the accelerated velocity of the Kuroshio Current downstream of Cape Shionomisaki.
KeywordsKuroshio path Short-term fluctuation Instability Frontal wave Wavelet Ocean reanalysis
In addition to these path shifts, there are active short-term fluctuations off the southern coast of Japan. In this paper, short-term fluctuation means fluctuations of periods less than 80 days. Past studies have reported the existence of fluctuations of roughly 10- to 30-day period (Taft 1978; Taira and Teramoto 1981; Kimura and Sugimoto 1987, 1993; Kasai et al. 1993; Ramp et al. 2008). The fluctuations of this period appear as frontal waves of the Kuroshio. From hydrographic data, Taft (1978) found an eastward propagating wave having a wavelength of 330 km and a phase speed of 0.14 m/s. From velocity observations off Cape Shionomisaki, Kimura and Sugimoto (1993) estimated dominant periods of 5–8, 10–12, and 17–19 days with wavelengths of 100, 200, and 400 km, respectively. The frontal waves stimulate exchange between coastal waters off the southern coast of Japan and the waters of the Kuroshio. Because of the water exchange, the short-term fluctuations affect the supply of nutrients (Kimura et al. 1997), primary production (Kimura et al. 1997; Ramp et al. 2008), and fishery conditions (Kimura and Sugimoto 1987; Kasai et al. 2002; Waseda and Mitsudera 2002; Okazaki et al. 2003). A fluctuation of a period of around 50- to 70-day, which is another fluctuation, is related to the S-shaped meandering pattern of the Kuroshio around the Izu Ridge (Kasai et al. 1993; Mitsudera et al. 2006; Takahashi et al. 2011).
Some studies have shown that the periodicities of short-term fluctuations changed with the Kuroshio path. Takahashi et al. (2011) found that a period of less than 40 days became weak in Sagami Bay after the transition from a nearshore non-large-meander path to an offshore non-large-meander path around June 3, 2008. Instead, fluctuations of 50- to 70-day periods became dominant. This is consistent with the observation by Kasai et al. (1993) in that fluctuations of a 50-day period were dominant only when the Kuroshio took an offshore non-large-meander path.
Past observations of short-term fluctuations have been conducted in limited locations and periods. Therefore, descriptions of short-term fluctuations in the Kuroshio are intermittent. Satellite data of sea surface temperature (SST) and sea surface heights (SSH), which have good spatial coverage, have been used in studies on short-term fluctuations (Kasai et al. 1993; Kimura et al. 1997; Takahashi et al. 2011, 2012). Obtaining SST images from satellites, however, is often hampered by clouds. Since SSH observations by satellite altimeters usually have a 7- to 10-day time resolution, it is difficult to capture fluctuations of 10- to 30-day periods. On the other hand, ocean assimilation data can compensate for missing observational data in space and time with the help of a numerical model. The main objective of this study is to describe the short-term fluctuations of the Kuroshio using an ocean assimilation dataset, the Japan Coastal Ocean Predictability Experiment 2 (JCOPE2) reanalysis. We are particularly interested in how the short-term fluctuation responds to shifts in the Kuroshio path with time. Among the Kuroshio paths, we investigate the shifts between the nearshore and offshore non-large-meander paths.
We aim to determine not only the distribution of the fluctuations but also the time fluctuations of these distributions. Therefore, we used a wavelet analysis, which provides information regarding the amplitude of periodic signals from a time series and the fluctuation of the amplitude with time. Because past studies have identified two types of fluctuations, 10- to 30-day and 50 to 70-day fluctuations, we classified the fluctuations into bands of 8- to 36-days and 40- to 80-days. Section 2 describes how we selected these bands. In this paper, we particularly address the 8- to 36-day fluctuations.
This paper is organized as follows. In Section 2, we introduce the data used in this study, i.e., the JCOPE2 reanalysis dataset. In addition, we explain the wavelet analysis used to describe short-term fluctuations. In Section 3, we show the distribution of the fluctuations and its relationship with the Kuroshio path. In Section 4, we describe typical structures of the 8- to 36-day fluctuations. In Section 5, we show the vertical structure. In Section 6, we investigate the relationship of the fluctuations with the Kuroshio velocity. In Section 7, we summarize and discuss the results.
2 Data and methods
The ocean reanalysis data used in this study is a product of the JCOPE2 (Miyazawa et al. 2009). This dataset has a horizontal resolution of 1/36°, 46 vertical levels, and covers the area 28°N–36°N, 128°E–142°E. In this paper, we particularly discuss the region from around Cape Shionomisaki to around the Izu Ridge (135–140°E; Fig. 1). The lateral boundary conditions were determined from the standard JCOPE2 reanalysis (Miyazawa et al. 2009) using a one-way nesting method with a flow relaxation scheme (Oey and Chen 1992; Guo et al. 2003).
The dataset in this study is a revised version of that used in Miyama and Miyazawa (2013). While the model used in Miyama and Miyazawa (2013) was forced by the wind stress and heat flux fields of the 6-hourly National Centers for Environmental Prediction/National Center for Atmospheric Research reanalysis data (2.5 × 2.5° horizontal resolution; Kalnay et al. 1996), the reanalysis used in this study was forced by the 3-hourly data of the Japan Meteorological Agency Meso Scale Model analysis (MSM; 0.125 × 0.1° horizontal resolution; Saito et al. 2007). The available dataset in this study includes daily-mean data, spanning about five and a quarter years (February 23, 2006 to May 31, 2011), which is longer than the roughly four and half years (August 2003 to December 2007) of the dataset used by Miyama and Miyazawa (2013). Note that the dataset used in this study does not include the large-meander period from the second half of 2004 to the first half of 2005 (Miyazawa et al. 2008). Therefore, we discuss the short-term fluctuations during the periods of the nearshore non-large-meander and the offshore non-large-meander paths of the Kuroshio. Hereafter, we simply refer to these paths as the nearshore and the offshore paths, omitting the term “non-large-meander.” We can better analyze the short-term fluctuations in these nearshore and offshore paths from the longer data period (more than five years) than from the period in the data used by Miyama and Miyazawa (2013). We cannot use the whole period of the data in Miyama and Miyazawa (2013) continuously because of about 1 year of the large-meander path from the second half of 2004 to the first half of 2005, which was a state dynamically different form the rest of the period. The overlapping period between data used in this study and that in Miyama and Miyazawa (2013) shows qualitatively the same results (not shown).
2.2 Wavelet analysis
By averaging the wavelet spectrum over the whole time period (Section 5a of Torrence and Compo 1998), we can obtain the global wavelet power spectrum (solid line in Fig. 3). Figure 3 confirms that the short-term fluctuations are significant: the global power spectrum (solid line) is above the 95 % confidence level (dashed line). There are two peaks in the global spectrum: between 8 and 36 days and between 40 and 80 days. The existence of the two peaks is consistent with the past observations that there are two kinds of short-term fluctuations of periods of about 20 and 50 days (Kasai et al. 1993). From Fig. 3 and from similar results at other points (not shown), we classified the short-term fluctuations into 8- to 36-day fluctuations and 40- to 80-day fluctuations. However, we only discuss the 8- to 36-day fluctuations in this paper, except short descriptions on 40- to 80-day fluctuations in Appendix A.
Although some studies have detected fluctuations of around a 10-day period in addition to the fluctuations of 20- to 30-day period (Kimura and Sugimoto 1993; Hinata et al. 2008; Ramp et al. 2008), we did not treat these period bands separately because we found it difficult to differentiate their peaks in this analysis. For further discussion regarding this, see Section 7. Some studies (Kimura and Sugimoto 1993; Kimura et al. 1994; Ramp et al. 2008) found fluctuations of a period of a few days. Because analyzing periods less than 8 days is difficult when using daily data, we analyzed periods more than 8 days.
By averaging the wavelet power spectrum in Fig. 2b over a certain period band, we can obtain a time series of the averaged variance (square meter) (Section 5b of Torrence and Compo (1998)). The solid red lines in Fig. 4 is the time series of averaged variance over 8- to 36-day. In this paper, the period-averaged power spectrum is referred to as the “variance” of the corresponding period band.
It is also possible to reconstruct the original time series inversely from the wavelet coefficients (Section 3i of Torrence and Compo (1998)). By using the wavelet coefficients in a certain period band, we can reconstruct the time series of SSH fluctuations (in meters) in the corresponding period band. The solid black line in Fig. 4 is the reconstructed time series using the wavelet coefficients in the 8- to 36-day period band. This is equivalent to the 8- to 36-day band-pass-filtered time series of the original time series (Fig. 2a). In this paper, the band-pass-filtered fluctuation of the original time series using the wavelet method is referred to as the “signal” of the period band.
Throughout this paper, we use the method described above to obtain the variance and signal of the 8- to 36-day periods from the original time series in the region shown in Fig. 1. Among the available data in the JCOPE2 reanalysis, we start with the wavelet analysis for SSH, which reflects the ocean dynamics beneath the surface. Later in the discussion, we use velocity, temperature, and salinity (for density) data. Although we applied the wavelet analysis to the data of the whole time period (from February 23, 2006 to May 31, 2011), we only describe the data from April 15, 2006 to April 14, 2011 in order to minimize the influence of the data edges (indicated by the cross-hatched regions in Fig. 2b).
3 Horizontal distribution and its relation to the Kuroshio path
In order to study how short-term fluctuations respond to shifts in the Kuroshio path, we first define the Kuroshio path. Looking at the relationship between the surface velocity and SSH in the JCOPE2 reanalysis, we found that the SSH isoline of 0.1 m is located around the maximum of the surface velocity (not shown). Therefore, we define the SSH isoline of 0.1 m as the Kuroshio path. Similar definitions using SSH isolines were used by Qiu and Chen (2005) and Sugimoto and Hanawa (2011). Because we are interested in the response of the short-term fluctuations to the low-frequency change of the background flow, we used a low-pass-filtered SSH to define the Kuroshio path. We again used the wavelet method and obtained a 90-day low-pass-filtered signal of SSH by reconstructing the signal from the wavelet coefficients for periods longer than 90 days.
The variance of the 8- to 36-day fluctuations in the Enshu-nada Sea (around 137.5°E, 34°N) in Fig. 7b during OP is not low compared with the variance during NP in Fig. 7a, although the region is far from the Kuroshio path during OP as indicated by the thick black line in Fig. 7b. The variance of the 8- to 36-day fluctuations in the Enshu-nada Sea during the OP is also due to the effect of the S-shaped meander, which is indicated by the arrow in Fig. 7b. This is discussed in the next section.
4 Typical structures
Each signal in Figs. 8 and 9a propagates downstream (eastward) with intensified value. Corresponding to the signals, the snapshots of the Kuroshio path (thick solid line in Fig. 8), indicated by the unfiltered SSH isoline of 0.1 m on each day, reveal an undulating pattern relative to the mean Kuroshio path (thick dashed line in Fig. 8). The path of the Kuroshio becomes more undulating, as shown in Fig. 8b–h, corresponding to the increase in variance with time (Fig. 8a–g). Figure 9 clearly shows that the signals grow downstream of Cape Shionomisaki.
Although a relatively high variance is detected at 135°E, 32.5°N, the smooth connection toward the downstream region is hampered by the presence of Cape Shionomisaki when the Kuroshio path (thick black solid line) is located close to Cape Shionomisaki. The gap is indicated by the low variance near Cape Shionomisaki in Fig. 8a–c. Propagation of the signals from the upstream region is weak, as shown in Fig. 8b, d, and the first half of Fig. 9a. As the Kuroshio path shifts farther offshore from Cape Shionomisaki, the connection from the upstream to the downstream regions becomes smoother. The variance near Cape Shionomisaki is higher (Fig. 8e, g) than before. The propagation of a positive (red) signal of SSH passes Cape Shionomisaki in Fig. 8f, h, and the second half in Fig. 9a.
Velocity signals (the vectors in the right panels of Fig. 10) are consistent with the SSH signals in Fig. 8: anticyclonic (cyclonic) circulations around high (low) SSHs. Temperature signals (color shadings in the right panels of Fig. 10) also displayed a corresponding undulating pattern. The signals were strong on the inshore side of the Kuroshio path where the background temperature gradient was sharp. While the variance of the 8- to 36-day fluctuations increase with time (color shadings in Fig. 8a, c, e, g), the background velocity decreased (color shadings in Fig. 10a, c, e, g).
A unique feature during the offshore period in Figs. 11 and 12 is that there were westward-propagating signals around 34°N toward the Enshu-nada Sea. A typical high SSH and a corresponding high SST are tracked by the green arrows in Figs. 11 and 12, respectively. These westward signals happened because a westward-flowing branch of the Kuroshio intruded into the Enshu-nada Sea when the S-shaped meander of the Kuroshio around the Izu Ridge was enhanced (the thick solid lines in Figs. 11 and 12). To show how far the Kuroshio path intrudes into the S-shaped meander, a non-filtered −0.25-m isoline of SSH (red curve) is added to Figs. 11 and 12. Observations of the intrusion of water from the Kuroshio by the westward branch into the coastal areas of the Kumano-nada Sea and the Enshu-nada Sea with a dominant period of about 20 days have been reported (Kimura and Sugimoto 2000). It should be noted that the emergence of the S-shaped meander itself was not the result of the 8- to 36-day fluctuations. The enhanced S-shape (black solid contour relative to black dashed line) continued throughout Fig. 11, indicating that the emergence of the S-shaped meander was caused by lower frequency fluctuations than the 8- to 36-day fluctuations. The emergence of the S-shaped meander is the result of 40- to 80-day fluctuations (not shown), as observed by Kasai et al. (1993) and Takahashi et al. (2011).
Finally, we statistically confirm the typical undulating pattern described above. Figure 13a shows the horizontal distribution of the correlation (contour) and the regression (color shadings; meter/meter) of the SSH signal (meter) at each point to the SSH signal (meter) at point A (139.2°E, 33.5°N), which is close to the maximum variance shown in Fig. 7a, using the daily data during NP. As shown in the typical pattern described in the preceding subsection, undulating patterns of plus and minus values appear along the Kuroshio path (thick black solid line). The correlation toward the upstream of Cape Shionomisaki quickly decays. The distance between points A (139.2°E, 33.5°N) and A′ (136°E, 33.2°N) in Fig. 13a corresponds to a wavelength about 300 km (299 km). The lag correlation between the signals of points A and A′ becomes maximum (0.35) when A′ leads A by 21 days. Therefore, this statistical estimate shows that the wave period is 21 days, and the downstream phase speed is about 14 km/day (0.16 m/s).
Figure 13b shows the horizontal distribution of the correlation (contour) and the regression (color shadings; meter/meter) of the SSH signal (meter) at each point to the SSH signal at point B (139.5°E, 31.8°N), which is close to the maximum variance shown in Fig. 7b, using the daily data during OP. This figure also shows the undulating patterns of plus and minus values appearing along the Kuroshio path during OP. The continuation of the undulating pattern toward the upstream of Cape Shionomisaki is somewhat clearer than that during NP, but it is still small. The distance between points B (139.5°E, 31.8°N) and B′ (136.5°E, 33.05°N) in Fig. 13b corresponds to a wavelength of 314 km. Although the estimate of the wavelength during OP is slightly longer than that during NP, the significance of this difference is not certain because the current method is a rough estimate based on a pointwise correlation. The lag correlation between the signals of points B and B′ becomes maximum (0.19) when B′ leads B by 19 days. Therefore, this statistical estimate shows that the wave period is 19 days and that the downstream phase speed is about 17 km/day (0.20 m/s).
5 Vertical structure
While the features of the variance along 137°E continue downstream along 138°E (Fig. 14c, d), the variances along 138°E are more enhanced, and the local maxima of the offshore (southward) side of the background velocity maximum are less clear than along 137°E.
We also show that the composites in the Kuroshio path coordinate system in panels e and f of Fig. 14 corresponding the panels c and d of Fig. 14, respectively. The results in the Kuroshio path coordinate system is presented because the composites in the fixed geographical coordinate system might cause artificial diffusion when the latitude of the flow varies in time (Halkin and Rossby 1985; Howe et al. 2009). The results in the Kuroshio path coordinate system in Fig. 14f shows stronger background velocity (dashed contour) and higher velocity variance (color shading and thin contour; square meter/square second) than the composites in the geographical coordinate system in Fig. 14d during the OP. However, Fig. 14e, f still shows that the background velocity and the velocity variance are larger during NP than during OP.
6 Relation to the Kuroshio acceleration
In Section 4, we mentioned the Kuroshio acceleration at Cape Shionomisaki is a typical occurrence in the 8- to 36-day fluctuations during the nearshore path. Miyama and Miyazawa (2013) studied the sudden acceleration of the Kuroshio jet that often appears off Cape Shionomisaki when the Kuroshio flows near the cape. The velocity discontinuity is accompanied by cold water outcropping on the inshore side downstream of Cape Shionomisaki. Miyama and Miyazawa (2013) proposed that the dynamics of the Kuroshio acceleration are a manifestation of a hydraulic control at Cape Shionomisaki when the Kuroshio approaches the coast. Kawabe (1990) found that, in order to provide a theoretical explanation of their paths, the nearshore non-large-meander path requires a larger Kuroshio acceleration than the offshore non-large-meander path. Because the 8- to 36-day fluctuations also become active during the nearshore path, we discuss the relationship between the Kuroshio acceleration and the 8- to 36-day fluctuations.
As also reported in Miyama and Miyazawa (2013), the Kuroshio acceleration (red solid line in Fig. 15) shows a high correlation (0.80) with the latitude of the Kuroshio at 138.5°E (black line in Fig. 15; the same as the red line in Fig. 6), despite the absence of the Kuroshio large meander in the dataset and the different definition of the Kuroshio path compared with that used by Miyama and Miyazawa (2013). In Miyama and Miyazawa (2013), the latitude of the Kuroshio is defined by the location of maximum velocity between 136°E and 140°E, 30.5–35°E at 10-m depth. The high correlation between the Kuroshio acceleration and the Kuroshio latitude means that the Kuroshio acceleration (red solid line in Fig. 15) is stronger during the nearshore path (shaded yellow) than the offshore path (shaded lightgreen) as Kawabe (1990) has shown. Miyama and Miyazawa (2013) showed that the velocity on the downstream side of Cape Shionomisaki is controlled by the Kuroshio acceleration. In fact, the zonal velocity on the downstream side of Cape Shionomisaki at 136.2, 33.4°N (red dashed line in Fig. 15) shows a high correlation (0.86) with the Kuroshio acceleration (red solid line in Fig. 15).
Because the above results imply that the 8- to 36-day fluctuations are more active when the background velocity is stronger, it can be hypothesized that the fluctuations are the result of instability. To examine the degree of instability, we analyzed the local energetics in the same way as in Miyazawa et al. (2004).
Here, the wavelet 8- to 36-day band-pass-filtered signals are used for the prime quantities. The wavelet 90-day low-pass filter is used to obtain the overbar quantities.
Figure 17c, d show that high K and P values (color shadings) also appear in the composites during OP. There are still high values on the inshore side of the velocity maximum. However, the K and P values are much smaller than those during NP (Fig. 17a, b). This reflects a weaker velocity and accompanying temperature gradient during OP (contours in Fig. 17c, d) than during NP (contours in Fig. 17a, b).
Figure 18c, d during OP corresponds to Fig. 18a, b during NP. The energy conversion terms in Fig. 18c, d are small compared with those in Fig. 18a, b because the Kuroshio acceleration is small during OP. However, the values of K and P downstream of Cape Shionomisaki all along the Kuroshio path are still relatively higher than those of the surrounding regions. This explains the downstream enhancement of the 8- to 36-day fluctuations during OP, albeit weaker than NP.
Figure 18 also shows that the energy conversion terms are negative on the upstream side of Cape Shionomisaki, indicating that the eddy energy is absorbed into the mean energy. This explains why the variances near Cape Shionomisaki decrease in Fig. 5.
7 Summary and discussion
This study describes 8- to 36-day fluctuations south of Japan using the JCOPE2 reanalysis data with a horizontal resolution of 1/36°. To detect the short-term fluctuations, we applied wavelet analysis. The amplitude of the 8-to 36-day fluctuations increases eastward from Cape Shionomisaki toward downstream around the Izu Ridge. The fluctuations of the 8- to 36-day period band are more active during the period of the nearshore non-large-meander path of the Kuroshio than during the period of the offshore non-large-meander path. The fluctuations of the 8- to 36-day period band appear as frontal waves on the Kuroshio Current. The waves have a wavelength of about 300 km, and the signals propagate eastward.
We found a relationship between the Kuroshio acceleration at Cape Shionomisaki and the amplitude of the 8- to 36-day period band. An analysis of energy conversion terms shows that both barotropic and baroclinic instabilities are important for generating eddies corresponding to the 8- to 36-day fluctuations. If the Kuroshio acceleration is a super critical flow resulting from hydraulic control, as proposed by Miyama and Miyazawa (2013), generation of a disturbance can be interpreted as a kind of hydraulic jump. A hydraulic jump is an abrupt transition process, which is often accompanied by turbulence and undulating patterns, from an unstable supercritical flow to a subcritical flow (Pratt and Whitehead 2007).
Because the Kuroshio acceleration produces a strong velocity shear and a sharp temperature gradient on the inshore side of the velocity maximum, the variance of the velocity is strong on the inshore side of the velocity maximum. As the velocity core shifts southward with depth, the high variance of the velocity moves southward with depth. The variance of the SSH, which reflects the underlying physics is high around the Kuroshio path. Because the Kuroshio acceleration is strong during the nearshore path, the variance of the 8- to 36-day fluctuations is higher during the nearshore path than during the offshore path.
The signals upstream of Cape Shionomisaki do not show a high correlation with the signals downstream. From the local minimum around Cape Shionomisaki, the variance is enhanced toward the downstream region around the Izu Ridge. The reason for the small upstream influence is partly because the condition at Cape Shionomisaki is adverse to eddy growth (negative values of the conversion terms in Fig. 18), and partly because the local growth of disturbances downstream diminishes the upstream influences. Because these factors are relatively small along the path during the offshore path, the correlation between upstream and downstream is slightly higher than during the nearshore path. In the Gulf Stream, Savidge (2004) found that high-frequency (3- to 8-day) fluctuations decay almost completely downstream of Cape Hatteras, with growth in the longer period (30- to 120-day) fluctuations. Comparing similarities and differences between the Kuroshio and the Gulf Stream is an interesting topic.
Because of the local enhancement of the disturbance, we focused on the region between Cape Shionomisaki and the Izu Ridge in this study and did not discuss fluctuations upstream of Cape Shionomisaki. A small upstream influence in the variance does not exclude the possibility that signals from upstream sometimes flow downstream of Cape Shionomisaki and/or that small activities upstream can be a trigger of instability on the downstream side. The influence of the fluctuations upstream of Cape Shionomisaki will be further investigated in a future study.
Another interesting subject for a future study would be to determine how the short-term fluctuations in this study affect fluctuations further downstream. Just as the Kuroshio path affects the fluctuations in this study, Sugimoto and Hanawa (2011) showed that the Kuroshio path also affects short-term fluctuations in the Kuroshio Extension. Unfortunately, we cannot discuss the relationship between the short-term fluctuations in this study and eddy activities downstream because the data used in this study do not extend past the eastern boundary at 142°E.
An interesting question is why disturbances of periods of around 8- to 36-days are generated. Frontal waves of similar frequencies have been reported in other coastal areas in Japan: in the East China Sea (Sugimoto et al. 1988; Qiu et al. 1990; James et al. 1999), in the Tokara Strait (Qiu et al. 1990; Maeda et al. 1993; Feng et al. 2000), south of Shikoku (Awaji et al. 1991), near the separation point of the Kuroshio to the Kuroshio Extension (Itoh and Sugimoto 2008), and in the Kuroshio Extension (Tracey et al. 2012). Frontal wave disturbances have been also found in the Gulf Stream (Lee and Atkinson 1983; Tracey and Watts 1986; Oey 1988; Savidge 2004). Itoh and Sugimoto (2008) successfully applied the analytical model of baroclinic instability by Pedlosky (1987) to the variability of the Kuroshio near the separation point from the western boundary, and they also discussed similar fluctuations in other regions of the Kuroshio. However, to explain the variability in this paper, the theory of pure baroclinic instability is insufficient because the analysis of the energy conversion terms shows that barotropic instability is also important. Using an instability analysis and a theoretical model, we will seek factors to determine the time and spatial scale of frontal waves in future studies.
Discussion of the 40- to 80-day fluctuations will be addressed in another paper. In contrast to the 8- to 36-day fluctuations, the peak variance of the 40- to 80-day fluctuations during OP is higher than that during NP (Appendix A). Because the peak region of 40- to 80-day fluctuations overlaps with the peak region of 8- to 36-day fluctuations on the downstream side of the Izu Ridge, the interaction between them might be important here. A full understanding of the short-term fluctuations on the downstream side of the Izu Ridge is beyond the scope of this study and needs further investigation.
We identified the properties of short-term fluctuations in the form of frontal waves and determined their relationship with the background velocity and the Kuroshio path. This knowledge will be beneficial for predicting when short-term fluctuations become active. As mentioned in the Section 1, the frontal waves of the Kuroshio Current affect biological activities and fisheries. Therefore, a better understanding of these short-term fluctuations will be important for the management of ecological environments.
This work is a part of the Japan Coastal Ocean Predictability Experiment (JCOPE) promoted by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC). Tomohiko Tsunoda conducted the calculation of the JCOPE2 with 1/36 resolution. We would like to thank Sergey M. Varlamov, Takuji Waseda, Humio Mitsudera, and Sourav Sil for their helpful discussions. The authors would like to thank Enago (www.enago.jp) and Textcheck (www. textcheck.com) for the English language review. The authors would like to thank the editor of this paper, Leo Oey, and two anonymous reviewers for their suggestions, which helped to improve the manuscript.
- Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C, Wang J, Leetmaa A, Reynolds R, Jenne R, Joseph D (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77(3):437–471. doi: 10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2 CrossRefGoogle Scholar
- Kimura S, Sugimoto T (1987) Short period fluctuations in oceanographic and fishing conditions in the coastal area of Kumano-nada Sea. Nippon Suisan Gakkaishi 53(4):585–593, http://agriknowledge.affrc.go.jp/RN/2010350534.pdf CrossRefGoogle Scholar
- Taft AB (1978) Structure of the Kuroshio south of Japan. J Mar Res 16:77–117Google Scholar
- Takahashi D, Morimoto A, Nakamura T, Hosaka T, Mino Y, Saino T (2011) Flow variability with periods of 50–70 days in Sagami Bay, Japan during the offshore non-large-meander path of the Kuroshio (in Japanese with English abstract and legends). Oceanogr Jpn 20(3,4):59–83Google Scholar
- Takahashi D, Morimoto A, Nakamura T, Hosaka T, Mino Y, Dang VH, Saino T (2012) Short-term flow and water temperature fluctuations in Sagami Bay, Japan, associated with variations of the Kuroshio during the non-large-meander path. Prog Oceanogr 105:47–60. doi: 10.1016/j.pocean.2012.04.012 CrossRefGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.