Kuroshio Extension cold-core ring and wind drop-off observed in 2021–2022 winter

Abstract

Energetic cyclonic mesoscale eddies, which are called cold-core rings and are shed southward from the Kuroshio Extension jet and form closed streamlines, affect the atmosphere through the heat exchange across the sea surface. To investigate the effect of rings on the atmosphere, we performed atmosphere and ocean observations across a cold-core ring centered around 34.5° N, 150.0° E using a research vessel from November 2021 to January 2022 and a shallow-water profiling float from November 23 to 28, 2021. As heat is released from the sea surface, no significant spatial contrast in the sea surface and mixed layer temperatures was detected across the ring. Meanwhile, the sea surface wind was occasionally observed to be weak around the ring, possibly through the air–sea interactions. The wind drop-off maintained a turbulent heat flux small around the ring. The wind field associated with the wind drop-off was examined by the rotary empirical orthogonal function analysis of the satellite sea surface wind data. The minimum of the sea surface wind is found to shift northward relative to the ring center and to be more than approximately 5 m s$^{-1}$ lower than the surrounding region. The shallow-water profiling float deployed around the ring center observed a rapid freshening event in the mixed layer, which can be attributed to the water intrusion from the north of the Kuroshio Extension jet through the interaction with the jet. This suggests that the cold water from the north continually affects the atmosphere without leaving traces in the shipboard sea surface temperature observations.

1 Introduction

The Kuroshio and the Kuroshio Extension (KE), the western boundary current of the North Pacific subtropical gyre, transport a significant amount of heat from the tropics to subtropics (e.g., Bryden and Imawaki 2001; Nagano et al. 2009, 2010) and redistribute heat in the western subtropical North Pacific (e.g., Ando et al. 2021). Heat is released from the sea surface as turbulent (sensible and latent) heat fluxes in the KE region because of cold dry wind from the Eurasian Continent (e.g., Konda et al. 2010). Through the heat fluxes, the atmosphere is affected by the Kuroshio and KE. Furthermore, many energetic mesoscale eddies are shed from the KE jet through the development of the current meander and exist in the KE region. Such cyclonic and anticyclonic eddies are called cold-core rings (CCRs) and warm-core rings, respectively. Eddies in the KE region have kinetic energy equivalent to or larger than that of the climatological mean current of the Kuroshio and the KE (e.g., Wunsch 1981) and affect on the atmosphere (Yang et al. 2019). Therefore, cold- and warm-core rings are considered to play an important role in the climate.

The upward (downward) displacement of the main thermocline due to the baroclinic structure of rings occasionally exhibits low (high) temperatures near the sea surface. Such strong eddies, which swirl speed is higher than the propagation speed of the Rossby wave, can transport water holding it in the eddy interior. The strength of the capability of eddies to keep waters inside is measured by the local Rossby number, which is defined as $U/({\beta }R^2)$, where U represents the eddy current speed, $\beta$ represents the latitudinal variation of the Coriolis parameter, and R is the radius of the eddy (e.g., Chelton et al. 2011; Nagano et al. 2016). For example, Nagano et al. (2016) observed a CCR with a swirl speed exceeding 50 cm s$^{-1}$. The local Rossby number of the ring was significantly greater than 10; thus, it is in the strongly nonlinear regime. Cold, fresh waters from the north of the KE have been frequently observed within CCRs in the southern KE region.

The low (high) sea surface temperature (SST) around the center of cyclonic (anticyclonic) mesoscale eddies makes the temperature difference across air–sea interface low (high); thus, the turbulent heat flux is reduced (enhanced) in the central part (e.g., Kouketsu et al. 2012; Ma et al. 2015; Tomita et al. 2019). The SST and temperature in the mixed layer in rings vary through the air–sea heat exchange, which affects on the atmospheric boundary layer. In addition to the sea surface cooling, horizontal and vertical currents caused by various processes can change the temperature in the mixed layer. During the attenuation of a CCR after detachment from the KE, convergent flows toward the center of the ring occur around the edge of the ring through the downward displacement of the main thermocline (Nakajima et al. 2024). Different wind speeds relative to alternating currents due to a CCR can cause convergence of the Ekman current toward the eddy center regardless of the wind direction (McGillicuddy Jr 2016). Spatial variations in the heat release from the ocean into the atmosphere may induce downwelling around the rim of a CCR and then be associated with the convergence of surface currents toward the center, as Worthington (1972) hypothesized about active heat releases over the Kuroshio and other western boundary currents.

The atmosphere is affected by mesoscale eddies in the KE and other western boundary current regions. Sea surface wind speeds were found to be positively correlated to mesoscale SST structures by high-resolution satellite observations (e.g., Nonaka and Xie 2003; Chelton and Wentz 2005; Small et al. 2008; Chelton and Xie 2010; Ma et al. 2015). In other words, sea surface wind is weakened (strengthened) over the cold (warm) sea surface water within a cyclonic (anticyclonic) eddy. The wind weakening (strengthening) is attributable to the attenuated (enhanced) vertical mixing in the stable (unstable) atmospheric boundary layer (Wallace et al. 1989; Hayes et al. 1989) or the adjustment to a negative (positive) pressure anomaly (Lindzen and Nigam 1987) over a cyclonic (anticyclonic) eddy. Ma et al. (2015) suggested that the vertical mixing mechanism changes winds over mesoscale eddies in the KE region from the sea surface to a height of approximately 850 hPa.

From November 2021 to January 2022, we performed atmosphere and ocean observations using the research vessel (R/V) Mirai of JAMSTEC in the southern KE region, focusing on individual CCRs, to investigate the effects of rings on the atmosphere in winter. During these research cruises, we obtained atmosphere and ocean variables through the cruises and three whole sections and one half section of a CCR centered around 34.5° N, 150.0° E. In addition, a shallow-water profiling float was deployed around the center of the ring. We calculated the turbulent heat flux using a bulk flux algorithm (Fairall et al. 2003) and found that it was reduced around the ring due to the weakening of the overlying wind, the so-called wind drop-off, around the ring.

In this study, we describe the upper-ocean ($<\,1000$ dbar or m) hydrographic structure of the ring and its relationship with the variation in the turbulent heat flux along the sections observed during the cruises and discuss the impact on the atmosphere by using a multi-satellite wind vector product. Furthermore, we examined the mixed layer temperature and salinity variations obtained near the ring center using a shallow-water profiling float. The R/V Mirai observations and data are described in Sect. 2. In Sect. 3, we describe the hydrographic structure of the CCR, examine the turbulent heat flux, and reveal the impact of the ring on the wind by the shipboard data, and examined the characteristics of the wind variation by the rotary empirical orthogonal function (EOF) analysis of the satellite vector wind data, and discuss a possible oceanic process in the mixed layer affecting the atmosphere. Finally, we summarize the results in Sect. 4.

2 Observations and data

2.1 R/V Mirai observations

We performed the observations during the R/V Mirai cruise (MR21-06), which consists of two parts: Leg 1 from November 3 to November 27, 2021, and Leg 2 from December 18, 2021, to January 13, 2022. Detailed information about these cruises can be obtained from the report of the research cruises (Kitamura 2022). Referring to near-real-time altimetric sea surface height (SSH) distributions in late November 2021, we intensively observed a CCR centered around 34.5° N, 150.0° E (Fig. 1). We performed the observations using conductivity–temperature–depth (CTD) and expendable CTD (XCTD) sensors down to depths exceeding 1000 dbar or approximately 1000 m, respectively. Three whole sections and one half section of the CCR were obtained (Fig. 3): an XCTD section on November 23–24, 2021; an XCTD section on December 30, 2021; a CTD section on January 6–7, 2022; and a CTD section on January 8–9, 2022. In addition, we obtained atmosphere and ocean data near the sea surface across the ring centered around 34.5° N, 156.5° E.

Fig. 1

Map of SSH averaged from November 2021 to January 2022 and locations of hydrographic profiles collected by the R/V Mirai (open circles) and Argo floats (dots) from November 2021 to January 2022. The blue and red dots represent locations of typical spiciness profiles to the south (cluster 1) and north (cluster 2) of the KE, respectively. Green dots denote locations of profiles not classified into either cluster. Contour intervals of SSH are 10 cm. Regions of lower SSH than the current axis of the KE indicated by the 100 cm contour, following Nagano et al. (2016), are indicated by gray shades

We used an SBE911plus CTD system (Sea-Bird Electronics, Bellevue, WA) for the CTD observations and XCTD-1 and XCTD-1N probes with the MK-150N data logging system (Tsurumi-Seiki, Yokohama) for the XCTD observations. The nominal accuracies for the CTD temperature, electrical conductivity, and pressure were 0.001 °C, 0.0003 S m$^{-1}$, and 0.015% of the full scale range (10,500 dbar), i.e., 0.16 dbar, respectively. Using 12-l Niskin-X bottles (Sea-Bird Electronics) installed in the CTD frame, we collected water samples and measured salinity. Based on the root-mean-square (RMS) difference between the CTD sensor and sampling salinity values above 1000-dbar depths, the accuracy for the CTD salinity sensor is higher than 0.003 on the practical salinity scale. According to the RMS differences between the XCTD and CTD values calculated by Mizuno and Watanabe (1998), the accuracies for the XCTD temperature and salinity are 0.05 °C and 0.05, respectively. Note that the XCTD probe depth was estimated from the fall rate because no pressure sensor was installed; the depth error is approximately 5 m or 2%, whichever is larger; thus, the error can be ${\sim }$ 20 m. Vertical hydrographic profiles collected by CTD and XCTD probes are presented with pressure (dbar) and depth (m), respectively. A 1 dbar pressure increase is almost equivalent to a 1-m depth increase (e.g., Talley et al. 2011).

The temperature, electrical conductivity, and other oceanographic parameters of water taken in the R/V Mirai at a depth of approximately 4.5 m were measured at intervals of 1 min using an SBE45 (Sea-Bird Electronics) and other sensors. The nominal accuracies of temperature and salinity are 0.002 °C and 0.003, respectively. Meteorological sensors of R.M. Young (Traverse City, MI), Vaisala (Vantaa, Finland), Setra Systems (Boxborough, MA), R.M. Young, and Eppley Laboratory (Newport, RI) installed at heights of 25, 21, 13, 24, and 25 m on the mast of the vessel, respectively, observed the wind speed/direction, air temperature/relative humidity, barometric pressure, rainfall amount, and longwave/shortwave radiation at intervals of 6 s. Using the COARE3.0b bulk flux algorithm (Fairall et al. 2003), we calculated the sea surface latent and sensible heat fluxes from the 10-min averaged values of SST, air temperature, relative humidity, wind speed, air pressure, and longwave/shortwave radiation data.

2.2 Shallow-water profiling float observations

To observe temperature and salinity variations within the CCR on timescales of days and shorter timescales, a shallow-water profiling float, a multipurpose observation float (MOF) (Offshore Technologies, Yokohama), was deployed near the center of the CCR (34.75° N, 149.75° E) at 13:03 (UTC) on November 23, 2021. A CTD probe, JES10mini (Offshore Technologies), was installed on the float. The accuracies of the temperature and electrical conductivity sensors are 0.005 °C and 0.005 S m$^{-1}$, respectively. The accuracy of salinity is higher than 0.04. The float was set to drift at a parking depth of 130 dbar and to collect data every 12 h ascending from 100 dbar to the sea surface at intervals of 10 s. Data observed by the float were transmitted to land through the Iridium communication system. We obtained 10 profiles of water temperature and salinity from the deployment to the end of the observation (00:40 November 28). Data were processed with a 5-point (50 s) median filter to reduce noisy features and interpolated onto the vertical grids at intervals of 1 m down to a depth of 80 m using the Akima (1970) method.

2.3 Argo float data

We used water temperature and salinity profile data collected by Argo floats in the region of 27–42° N, 142–165° E from November 2021 to January 2022. According to the Argo Science Team, the accuracies of the Argo temperature, salinity, and pressure data were reported to be 0.005 °C, 0.01 (psu), and 5 dbar, respectively (Hosoda et al. 2008). The profile data were interpolated onto the vertical grids at intervals of 10 dbar from 10 to 1000 dbar.

To classify hydrographic profiles based on the isopycnal water mass distribution, we adopt a parameter whose isopleths are orthogonal to those of the potential density in the potential temperature–salinity plane, called the spiciness parameter (Veronis 1972; Jackett and McDougall 1985), as a state variable of seawater. Using the polynomial prepared by Flament (2002), we calculated the spiciness parameter $\pi$ in kg m$^{-3}$ from the Argo float data and the data collected during the R/V Mirai cruises. As performed by Nagano et al. (2022a, 2022b, 2022c), we performed cluster analysis by applying Ward’s criteria (e.g., Ward 1963; Anderberg 1973) to spiciness parameter distributions in the isopycnal layer between 24 and 27 ${\sigma }_{\theta }$ with intervals of 0.01 kg m$^{-3}$ and classified all the hydrographic profiles into three clusters. As described in Sect. 3, typical spiciness profiles to the south and north of the KE were classified into clusters 1 and 2, respectively. Profiles that cannot be classified into these clusters were assigned to cluster 3.

2.4 Satellite SSH and SST data

As mentioned above, we used near-real-time daily altimetric absolute SSH maps to determine the locations of the observation lines across the CCR and the MOF deployment position. For the following analysis, we used daily delayed-time altimetric absolute SSH mapped onto $1/4^{\circ } \times 1/4^{\circ }$ grids in the region of 24–42° N, 135–176° E from November 2021 to January 2022. The data are the gridded delayed-time updated SSH (Level 4) anomaly data (Pujol and Mertz 2020) added to the mean SSH data provided by Rio et al. (2011). In addition, we used daily gridded satellite-observed SST data (NOAA OI SST V2 High-Resolution Data Set) from January 6 to January 10, 2022, in the region of 32–40° N, 143–157° E provided by Reynolds et al. (2007). The resolution of the SST dataset is $1/4^{\circ } \times 1/4^{\circ }$.

2.5 Satellite vector wind data

Furthermore, to examine the spatiotemporal variation in sea surface wind vector, we used 6-hourly $1/4^{\circ } \times 1/4^{\circ }$ gridded vector wind data from November 1, 2021, to January 31, 2022, in the region of 30–40° N, 140–160° E of the cross-calibrated multi-platform (CCMP) ocean vector wind analysis Version 3.1 (Mears et al. 2022a; Mears et al. 2022b). The CCMP V3.1 vector wind data are merged vector wind data at a height of 10 m above the sea surface based on measurements by multiple types of satellite microwave sensors and have the equivalent spatial resolution to the SSH and SST data described above.

3 Results and discussion

3.1 Characteristics of CCRs

Figure 1 shows the distribution of SSH averaged during the R/V Mirai observations. The KE jet is illustrated as a meandering path characterized by a sharp southward SSH elevation. CCRs, delineated by significantly low SSH compared with the surrounding values, were observed around 34.5° N, 150.0° E and 34.5° N, 156.5° E south to southeast of the KE jet meander troughs.

The locations of spiciness profiles classified into the three clusters by the cluster analysis based on the spiciness parameter $\pi$ are plotted in Fig. 1. Spiciness profiles classified into cluster 1 (blue dots) are mostly distributed in the subtropical region south of the KE. Although most of the profiles classified into cluster 2 (red dots) are mainly present to the north of the KE (i.e., the perturbed area), there are those trapped in the interior region of the CCRs centered around 34.5° N, 150.0° E and 34.5° N, 156.5° E, as described below. The salinity of cluster 1 profiles is higher than that of cluster 2 profiles in layers of density lighter than 26.0 ${\sigma }_{\theta }$ (Fig. 2). Thus, clusters 1 and 2 represent typical hydrographic profiles to the south and north of the KE, respectively. Other spiciness profiles (green dots) observed near the current axis are classified into cluster 3. They represent mixtures of water masses to the north and south of the KE in layers of density lighter than 26.0 ${\sigma }_{\theta }$ and intrusions of very fresh water into denser layers, as reported by Nagano et al. (2016).

Fig. 2

Potential temperature versus salinity diagram based on the data collected by the R/V Mirai and Argo floats from November 2021 to January 2022. The blue and red dots represent spiciness profiles classified into clusters 1 and 2, respectively. Green dots denote profiles not classified into either cluster. Potential density is indicated by solid contours at intervals of 0.5 kg m$^{-3}$

Notably, all the spiciness profiles near the center of the western CCR around 34.5° N, 150.0° E are classified into cluster 2. The water masses, which were derived from the north of the KE and are represented by cluster 2, are believed to be kept inside of the strong ring with a high local Rossby number, as described below and observed at a moored buoy station by Nagano et al. (2016). The SSH at the western CCR center is approximately 50 cm lower than the surrounding values (Fig. 3), and the swirl speed is estimated to be ${\sim }$ 50 cm s$^{-1}$ on average. The local Rossby number is greater than 10; thus, the CCR is sufficiently strong to hold water in the interior. Because of the substantial wind and intensive turbulent heat release during winter, there may be convergent flows (Worthington 1972), the wind-driven Ekman current (McGillicuddy Jr 2016), etc. Cluster 1 profiles, along with cluster 2 profiles, were observed near the ring center around 34.5° N, 156.5° E; therefore, the subtropical gyre water had been brought into the eastern ring. Because the western ring is more capable of holding water masses in the interior than the eastern ring, we examined the sections of the western ring in the following subsection.

Fig. 3

Locations of CTD (dots) and XCTD (crosses) profiles collected by the R/V Mirai on a November 23, 2021, b December 29–30, 2021, c January 6–7, 2022, and d January 8–9, 2022. Dots and crosses with profiles classified into clusters 1 and 2 are colored blue and red, respectively. Maps of SSH around the CCR observed in the R/V Mirai cruise on a November 23, 2021, b December 29, 2021, c January 6, 2022, and d January 8, 2022, are shown by color shades with contour intervals of 10 cm

3.2 Hydrographic sections of western CCR

Sections of potential density ${\sigma }_{\theta }$ across the western CCR are shown in Fig. 4. Because of the baroclinicity, the main pycnocline between approximately 25.4 and 27.0 ${\sigma }_{\theta }$ shoals toward the center of the ring. The mixed layer is identified as a layer of vertically uniform potential density. In this study, following the procedure described by Suga et al. (2004), we determined the mixed layer base as the pressure or depth at which the potential density difference from the sea surface value was 0.125 kg m$^{-3}$. The depth of the mixed layer base is indicated by thick black lines in Fig. 4. The depth increased from approximately 100 m in late November 2021 (Fig. 4a) to 200 dbar or more in early January 2022 (Fig. 4d). In particular, in early January (Fig. 4c and d), the thickness of the mixed layer was greater toward the rim of the ring with the depth of the main pycnocline, probably because the strong vertical gradient of potential density in the pycnocline prevents vertical mixing at the bottom of the mixed layer. The potential density in the mixed layer decreases toward the ring center in association with the low-salinity water in the interior, as described below.

Fig. 4

Sections of potential density, ${\sigma }_{\theta }$, (kg m$^{-3}$) across the CCR. These sections were plotted using the data obtained on a November 23, 2021, b December 29–30, 2021, c January 6–7, 2022, and d January 8–9, 2022, in the Mirai cruise. The intervals of thin contours for ${\sigma }_{\theta }$ are 0.2 kg m$^{-3}$. Vertical coordinates for the southwestern half section in c and the northeastern half section in d by CTD casts are pressure, and those for other sections by XCTD casts are depth. The 26.0 ${\sigma }_{\theta }$ contours are colored white. The thick black lines represent the mixed layer base. The inverted blue and red triangles at the top of the panels represent locations of hydrographic profiles classified into clusters 1 and 2, respectively

The fresh cold water derived from the north of the KE is represented by the negative spiciness anomaly ${\pi }_{a}$ at the stations within the CCR indicated by red inverted triangles in Fig. 5 and the saline warm subtropical water the positive spiciness anomaly (at the stations of blue inverted triangles). They are clearly separated from the lightest layer (or the mixed layer) to a layer of approximately 26.0 ${\sigma }_{\theta }$. Note that the water from the north of the KE exists only above a depth of approximately 200 m (or dbar) around the ring center (the depth of the 26.0 ${\sigma }_{\theta }$ isopycnal layer is denoted by white contours in Fig. 4) and mostly within the mixed layer (above the thick black line). Therefore, the mixed layer water derived from the north of the KE and trapped by the CCR is modified by the atmospheric thermal forcing, i.e., cooling. In layers deeper than an isopycnal layer of approximately 26.0 ${\sigma }_{\theta }$, the spatial variation in spiciness is small across the ring; therefore, the water from the north of the KE is indistinguishable from that of the subtropical water in the deep layer.

Fig. 5

Sections of spiciness anomaly, ${\pi }_{a}$, (kg m$^{-3}$) with respect to potential density ${\sigma }_{\theta }$ from the mean distributions. These sections were plotted using the data obtained on a November 23, 2021, b December 29–30, 2021, c January 6–7, 2022, and d January 8–9, 2022, in the Mirai cruise. Contour intervals of ${\pi }_{a}$ are 0.1 kg m$^{-3}$. No data are denoted by meshes. The blue and red inverted triangles represent hydrographic points classified into clusters 1 and 2, respectively

Although the salinity within the CCR (defined as the negative-${\pi }_{a}$ region here) is significantly lower than that in the surrounding regions (the positive-${\pi }_{a}$ region) (Fig. 7), there is no significant horizontal difference in the potential temperature of the mixed layer water across the ring, particularly in December 2021 (Fig. 6b) and January 2022 (Fig. 6c and d). The strong sea surface cooling due to the large difference between SST and atmospheric temperature outside the CCR (Kouketsu et al. 2012) mainly modified the mixed layer water temperature and substantially reduced the temperature contrast immediately after the detachment from the KE. Subsequently, the spatial temperature contrast disappeared. The uniform SST and mixed layer temperature are thought to be the result of the air–sea interactions.

Fig. 6

Same as Fig. 4, but for potential temperature (°C). Contour intervals are 1 °C

The seasonal changes in the potential temperature and salinity of the mixed layer in the CCR from autumn to winter were observed through the shipboard CTD/XCTD observations. The mixed layer potential temperature in the ring decreased by approximately 3 °C from November 23, 2021 (Fig. 6a), to December 29, 2021 (Fig. 6b), with an estimated cooling rate of ${\sim }$ 0.1 °C day$^{-1}$. The evaporation of freshwater from the sea surface and the entrainment of the deep saline water increased salinity in the mixed layer (Fig. 7). The salinity increase rate of the mixed layer near the ring center from November 23 (Fig. 7a) to December 29 (Fig. 7b) was ${\sim }$ 0.005 day$^{-1}$. The cooling and salinification rates after December 2021 have the same order of magnitude as or are less than the values from November 23 to December 29, 2021. The mixed layer potential temperature similarly changed regardless of whether it was inside or outside the CCR. The mixed layer water inside the ring is kept fresh even in winter, while that outside the ring becomes saline because of the entrainment of the deep water as in these observations and in the results of Nagano et al. (2014).

Fig. 7

Same as Fig. 4, but for salinity. Contour intervals are 0.1

3.3 Sea surface turbulent heat flux

We calculated the latent and sensible heat fluxes along the cruise tracks of Legs 1 and 2. The mean values of the latent heat flux (172 W m$^{-2}$ for Leg 1 and 290 W m$^{-2}$ for Leg 2) were much larger than those of the sensible heat flux (20 W m$^{-2}$ for Leg 1 and 82 W m$^{-2}$ for Leg 2). The standard deviations of the latent heat flux (104 W m$^{-2}$ for Leg 1 and 120 W m$^{-2}$ for Leg 2) were also much larger than those of the sensible heat flux (19 W m$^{-2}$ for Leg 1 and 68 W m$^{-2}$ for Leg 2). The sea surface turbulent (sensible plus latent) heat fluxes along the tracks of the R/V Mirai cruises are shown in Fig. 8. Following the procedure described by Nagano and Ando (2020) and Nagano et al. (2022a,b), we decomposed the variation in latent heat flux $F_{\mathrm{L}}$, which is much greater than the sensible heat flux variation, into components based on the variations in SST $T_{\mathrm{S}}$, air temperature $T_{\mathrm{A}}$, specific humidity q, and wind speed U to investigate the contributions of the individual variables to the heat flux variation, as follows:

$$\begin{aligned} {\Delta }F_{\mathrm{L}} \, {\approx } \, \left( \frac{\partial F_{\mathrm{L}}}{\partial T_{\mathrm{S}}} \right) {\Delta }T_{\mathrm{S}} + \left( \frac{\partial F_{\mathrm{L}}}{\partial T_{\mathrm{A}}} \right) {\Delta }T_{\mathrm{A}} + \left( \frac{\partial F_{\mathrm{L}}}{\partial q} \right) {\Delta } q + \left( \frac{\partial F_{\mathrm{L}}}{\partial U} \right) {\Delta }U, \end{aligned}$$

(1)

where ${\Delta }T_{\mathrm{S}}$, ${\Delta }T_{\mathrm{A}}$, ${\Delta } q$, and ${\Delta }U$ denote the variations in SST, air temperature, specific humidity, and wind speed, respectively. Because the function of the bulk flux algorithm is very complex, the rate of change of each term on the right-hand side of Eq. (1) was computed by setting the other three variables to the mean values during each cruise period. The units of all the terms of Eq. (1) are the same as that of the turbulent heat flux, i.e., W m$^{-2}$.

Fig. 8

Sea surface turbulent (latent plus sensible) heat flux in W m$^{-2}$ on tracks of the R/V Mirai a from November 3 to 26, 2021, and b from December 18, 2021, to January 10, 2022. The positive values represent fluxes from the ocean to the atmosphere. Maps of SSH during the periods are shown with contours. Contour intervals of SSH are 10 cm, and low SSH (< 100 cm) regions are indicated by gray shades

As described above, the sea surface turbulent heat flux is positive during the observation periods (Fig. 8), principally due to dry cold wind (Figs. 9b–d and 10b–d) from the Eurasian Continent, which appears to be enhanced in the KE region and is described in the following rotary EOF analysis. This enhancement of the wind is primarily responsible for the larger heat flux near the KE than those in the surrounding regions. Meanwhile, the contribution of SST to the heat flux has no remarkable peak in the KE current but increases toward the south because of the southward increase in SST. This is different from the case of the Kuroshio south of Japan studied by Nagano and Ando (2020). Because the Kuroshio south of Japan exhibits a warm core near the current axis of the Kuroshio (e.g., Taft 1978), the SST is a dominant factor and has a maximal SST contribution to the turbulent heat flux along the current axis of the Kuroshio south of Japan (Nagano and Ando 2020).

Fig. 9

Components of latent heat flux variation due to the variations in a SST $({\partial F_{\mathrm{L}}}/{\partial T_{\mathrm{S}}}) {\Delta }T_{\mathrm{S}}$ in Eq. (1), b air temperature $({\partial F_{\mathrm{L}}}/{\partial T_{\mathrm{A}}}) {\Delta }T_{\mathrm{A}}$, c specific humidity $({\partial F_{\mathrm{L}}}/{\partial q}) {\Delta } q$, and d wind speed $({\partial F_{\mathrm{L}}}/{\partial U}) {\Delta }U$ on the R/V Mirai track from November 3 to 26, 2021. The positive values represent fluxes from the ocean to the atmosphere. SSH map during the period is shown on each panel with contour intervals of 10 cm, and low SSH (< 100 cm) regions are indicated by gray shades

Fig. 10

Same as Fig. 9, but on the R/V Mirai track from December 18, 2021, to January 10, 2022

The R/V Mirai crossed the western and eastern CCRs in the weak wind during the first 4 days from November 3 to 6, 2021, of Leg 1 (Fig. 11b). A sharp wind drop-off was observed near the western ring on November 23, immediately after the eastward passage of the strong wind disturbance. The wind drop-off can be attributed to the wind reduction caused by the KE meander trough around 35.5°, 147.5° E (not shown). On December 29, 2021 (Fig. 12a), the R/V Mirai crossed the meander trough (Fig. 8b). The anomalously low SST (the dashed arrow in Fig. 12b) associated with the KE meander negatively contributed to the latent heat flux (Fig. 10a). At this time, the wind speed was also significantly decreased (Fig. 12b). The weak wind was observed over the meander trough at 06:00 on December 29 in the CCMP satellite data (Fig. 13a). Tomita et al. (2013) reported a similar response of the atmosphere to a meandering KE front associated with a cloud hole and sea surface wind divergence over sea surface warm water south of the KE. Although the R/V Mirai passed by the ring on December 30, we found no ocean imprint because of the passage of an atmospheric disturbance accompanied by a strong wind.

Fig. 11

Time series of a SSH (cm) and b latent heat flux components (W m$^{-2}$) due to the variations in SST $({\partial F_{\mathrm{L}}}/{\partial T_{\mathrm{S}}}) {\Delta }T_{\mathrm{S}}$ in Eq. (1) (magenta lines), air temperature $({\partial F_{\mathrm{L}}}/{\partial T_{\mathrm{A}}}) {\Delta }T_{\mathrm{A}}$ (green lines), specific humidity $({\partial F_{\mathrm{L}}}/{\partial q}) {\Delta } q$ (cyan lines), and wind speed $({\partial F_{\mathrm{L}}}/{\partial U}) {\Delta }U$ (orange lines) along the R/V Mirai track from November 3 to 26, 2021. In panel (a), SSH depressions less than approximately 100 cm indicate KE meander troughs and cold-core rings

Fig. 12

Same as Fig. 11, but for the period from December 18, 2021, to January 10, 2022. The dashed and solid arrows indicate wind drop-offs over the KE meander trough and the CCR, respectively

Fig. 13

Satellite-observed a wind speed (m s$^{-1}$, color shades and white contours) and b wind vector (m s$^{{-1}}$, arrows) and divergence (${10}^{{-5}}$ s$^{{-1}}$, color shades) near the sea level at 06:00 on December 29, 2021, on the absolute SSH map (cm, black contours). Intervals of white contours for wind speed and black contours for SSH are 1 m s$^{-1}$ and 10 cm, respectively. A reference wind vector with a magnitude of 10 m s$^{-1}$ is shown on the lower right corner of panel (b)

Despite the horizontally uniform SST and mixed layer temperature observed by the R/V Mirai, the turbulent heat flux was suppressed near the western CCR centered around 34.5° N, 150.0° E (Fig. 8b) in January 2022. Kouketsu et al. (2012) and Ma et al. (2015) reported that mesoscale cyclonic eddies in the KE region affect on the latent heat flux through the low SST in the interior, possibly because the past studies did not treat KE meanders and detached eddies differently. In this study, the low latent heat flux events around the detached ring were mostly attributed to the wind speed drops on January 6, 9, and 10, 2022 (the solid arrows in Fig. 12b). For the wind drop-offs on January 6 (Fig. 14a), 9 (Fig. 15a), and 10 (Fig. 16a), we identified wind speed minima near the CCR from the satellite-observed CCMP data. The winds appear to be convergent over the ring (Figs. 14b, 15b, and 16b). However, as examined in the following rotary EOF analysis, the convergent wind on January 6 is revealed not to be attributed to the wind drop-off.

Fig. 14

Same as Fig. 13, but at 18:00 on January 6, 2022

Fig. 15

Same as Fig. 13, but at 00:00 on January 9, 2022

Fig. 16

Same as Fig. 13, but at 00:00 on January 10, 2022

Decomposition of the wind vector variation by the rotary EOF modes is useful to reveal the characteristics of the variation (Denbo and Allen 1984; Thomson and Emery 2014). To examine the characteristics of the variation in the sea surface wind vector, we tried to compute the rotary EOF modes which express the wind drops near the CCR. Because the atmospheric disturbances have complicated and diverse forms as shown in Figs. 13, 14, 15, and 16, there would be no versatile modes representing all the disturbances. Nevertheless, we obtained mode patterns with wind depressions around the CCR by taking the calculation region to 31–39° N, 144–155° E around the ring center. The rotary EOF modes may represent common features of the atmospheric disturbances with the wind drops observed in the Mirai cruise. The first rotary EOF mode, which accounts 68% of the total variance, represents the variation with a uniform eastward phase increase (Fig. 17b) and a slight amplification over the meandering KE front (Fig. 17a). The divergence of the wind by the first rotary EOF mode variation is not substantial, despite that the wind direction greatly varies (Fig. 17c–d) mostly from southward to eatward, as indicated by the distribution of the time coefficient phase of ${\sim }$ 180°–270° (Fig. 20a).

Fig. 17

a Amplitude and b phase of the first rotary EOF mode of the 10-m height wind vector variation from November 1, 2021, to January 31, 2022. White contour intervals for amplitude and phase are 0.002 and 30°, respectively. Wind vectors (arrows) and their divergences (color shades) at time coefficient phases 0°, 90°, 180°, and 270° are shown in panels (c–f), respectively; for the calculations, the amplitude of the time coefficient was set to 300. The SSH map averaged during the observation is shown by black contours at intervals of 10 cm

The wind drop-off around the CCR observed by the R/V Mirai can be represented by the second and third rotary EOF modes, which exhibit the minima in amplitudes immediately to the northwest and northeast of the mean center of the ring (Figs. 18a and 19a), accounting 12 and 7% of the total variance, respectively. The phase of the second (third) rotary EOF mode increases counterclockwisely (clockwisely) around the amplitude minimum (Figs. 18b and 19b). Because the phases of their time coefficients mostly distribute between 120° and 360° (Fig. 20b and c), the wind vector variations of both rotary EOF modes can be convergent and divergent (Figs. 18c–f and 19c–f). The leading three rotary EOF modes explain approximately 87%. The percentages of the higher rotary EOF modes are smaller than 4%.

Fig. 18

Same as Fig. 17, but for the second rotary EOF mode of the 10-m height wind vector variation

Fig. 19

Same as Fig. 17, but for the third rotary EOF mode of the 10-m height wind vector variation

Fig. 20

Histograms of the phase of the time coefficient of the a first, b second, and c third rotary EOF modes

The wind speeds and vectors on January 6, 9, and 10, 2022, by the second and third rotary EOF modes are shown in Fig. 21. The wind drop-offs around the CCR observed by the R/V Mirai are presented by the second and third rotary EOF modes as decreases in wind speed by more than approximately 5 m s$^{-1}$ near the ring. The attenuation of the wind is greater than $<3$ m s$^{-1}$ and $<0.3$ m s$^{-1}$ reported by Nonaka and Xie (2003) and Ma et al. (2015), respectively. The discrepancies would be mainly attributed to the time filtering (Nonaka and Xie 2003) and time averaging (Ma et al. 2015).

Note that the wind field in association with the wind drop-off is not always convergent; for example, the divergence is positive on January 6 (Fig. 21b) and almost vanishes on January 9 (Fig. 21d). If the pressure adjustment mechanism was effective on the wind reduction, the wind field would have always been convergent. Thus, the wind drop-off around the CCR is considered to be caused not by the pressure adjustment but the vertical mixing reduction due to the stabilization of the atmosphere around the ring. As the atmospheric boundary layer become more stable toward the north due to the northward cooler sea surface water in the KE region (Tokinaga et al. 2006), the northward shifts of the wind drops (i.e., the minima of the amplitudes of the rotary second and third EOF modes) relative to the ring center may be related to the more stable background atmospheric condition toward the north. The vertical mixing reduction near the CCR would disconnect the atmospheric boundary layer from the upper atmosphere. Therefore, the wind near the ring is considered to be weak in speed but quite variable in direction.

Fig. 21

Wind speeds (left), and wind vectors and divergences (right) by the second and third rotary EOF modes; a and b at 18:00 on January 6, c and d at 00:00 on January 9, and e and f at 00:00 on January 10, 2022. White contour intervals for wind speeds are 1 m s$^{-1}$. The SSH map averaged during the observation is shown with black contours at intervals of 10 cm

Even if the CCR contained cold water in the interior immediately after the separation from the KE jet (late June 2021), the influence of the low SST is not considered to have remained persistent in the atmosphere for approximately 6 months. Satellite SST observations may detect the variation associated with the reduction in the turbulent heat flux around the CCR. Figure 22 shows satellite-observed SSTs on the occasions with wind drop-offs observed by the R/V Mirai. Although there is no SST trace on January 6 and 9, 2022 (Fig. 22a and b), in association with the CCR, a cold tongue was observed to extend toward the ring on January 10 (Fig. 22c). Therefore, the wind drop-offs over the ring may have been forced by intermittent cold water intrusions into the ring. However, spatiotemporal scales of such temperature fluctuations may be too small to be fully resolved by both satellite and shipboard observations.

Fig. 22

Satellite-observed SST in °C on January a 6, b 9, and b 10, 2022. Contour intervals are 0.5 °C

3.4 Short-term hydrographic variations within western CCR

After deployment on November 23, 2021, the MOF moved cyclonically along with the eddy current within a radius of approximately 30 km from the center (Fig. 23). When the float was deployed, the measured potential temperature and salinity were 22.0–22.1 °C and 34.25–34.40, respectively, which are consistent with the XCTD observations on November 23 (Figs. 6a and 7a). We obtained a reasonable time series of the potential temperature and salinity near the ring center down to a depth of 80 m (Fig. 24a and b), below which the mixed layer base was present (Fig. 4a).

Fig. 23

Map of SSH (cm, top color bar) on November 23, 2021, and locations of hydrographic profiles (bottom color bar) collected by a multipurpose observation float (MOF). Contour intervals of SSH are 10 cm. Measurements were performed at 00:00 and 12:00 (UTC) except for the first measurement (November 23 14:29 UTC). Profiles were sequentially numbered from the first

Diurnal intrusions of warm and saline water, which were probably due to the entrainment of water from below the mixed layer, were observed around 12:00 on November 23, 25, and 26, 2021 (Fig. 24a and b). The rates of changes in the potential temperature and salinity due to the entrainment reached maxima of 0.1 °C day$^{-1}$ (Fig. 25a) and 0.4 day$^{-1}$ (Fig. 25b) around the deepest layer of the observation (${\sim }$ 80-m depth), respectively. The fluctuations in the potential temperature were vertically attenuated, whereas those in the salinity extended to near the sea surface. Thus, in addition to the sea surface forcing, the entrainment from the deep layer was observed to modify the mixed layer water.

Because of their accumulated effects, the mixed layer water became cooler and more saline toward winter, as observed by the shipboard CTD/XCTD observations (Figs. 6 and 7). The potential temperature of the mixed layer water smoothly decreased at a rate of approximately 0.1 °C day$^{-1}$ through the MOF observation of 4 days and a half (Fig. 25a), being rapidly adjusted to the sea surface cooling and equivalent to the seasonal change obtained by the shipboard observation. Remarkably, the salinity decreased at rates up to 0.4 day$^{-1}$ (Fig. 25b), which is opposite to and two orders of magnitude larger than the seasonal change (${\sim }$ 0.005 day$^{-1}$). Because the effects of the potential temperature and salinity changes on density almost cancel each other, there is no significant tendency in potential density (Fig. 25c). The water to the north of the KE is colder and fresher than that to the south (Fig. 2; Kawai 1972; Nagano et al. 2014). Thus, the MOF-observed opposite salinity change, i.e., freshening, of the mixed layer water, suggests that cold water intrudes from the north of the KE on a time scale no longer than days and imprints the ring signal on the atmosphere.

Fig. 24

Time–depth plots of a potential temperature in °C, b salinity, and c potential density ${\sigma }_{\theta }$ in kg m$^{-3}$ observed by the MOF. Contour intervals for potential temperature, salinity, and potential density are 0.05 $^{\circ }$C, 0.05, and 0.05 kg m$^{-3}$, respectively

Fig. 25

Time–depth plots of the variation rates in a potential temperature (°C day$^{{-1}}$), b salinity (day$^{{-1}}$), and c potential density (kg m$^{{-3}}$ day$^{{-1}}$) observed by the MOF. Contour intervals for potential temperature, salinity, and potential density variation rates are 0.05 °C day$^{{-1}}$, 0.1 day$^{{-1}}$, and 0.1 kg m$^{{-3}}$ day$^{{-1}}$, respectively

Note that, in contrast with the westward propagating character of the Rossby wave, the ring remained in the region east or southeast of the trough of the KE meander for more than 3 months during the observations, possibly interacting with the eastward flowing KE jet. The horizontal entrainment of water from the north of the KE can be caused by the interaction with the KE jet through processes associated with such as filament-like water intrusions with a smaller spatial scale, i.e., submesoscale, which was barely observed by the satellite SST observation on January 10, 2021 (Fig. 22c). In other words, the mixed layer water within the CCR would be constantly affected by the advection of water from the north of the KE.

The unidentified contrast in the SST and mixed layer potential temperature between inside and outside the ring indicates that a rapid convective adjustment had occurred. Perhaps, there had been more intensive convective adjustments in the region along the outer periphery of the CCR, where the KE water warmer than that north of the KE was advected, than in the central part of the ring. The computation of the turbulent heat flux using the bulk algorithm uses SST, air temperature, humidity, and wind speed. Of these variables, SST is the only parameter that accounts for changes in the ocean state. If the convective adjustment in the mixed layer is too rapid for the variation to be fully detected using sparse shipboard and satellite temperature measurements, the computation based on the bulk algorithm can be underestimated. For more exact measurements of the turbulent heat flux, additional appropriate methods such as the eddy correlation flux measurement technique are required in the intensive heat release regions such as the KE region.

4 Summary

Strong cyclonic mesoscale eddies, called CCRs with a high local Rossby number greater than unity, are frequently generated in the southern KE region through the development of the jet meanders. Such strong mesoscale eddies play critical roles in modifying turbulent heat flux in the KE region. To examine the impact of CCRs on the atmosphere, we performed atmosphere and ocean observations using the R/V Mirai (MR21-06) in the southern KE region from November 3 to 27, 2021 (Leg 1), and from December 18, 2021, to January 13, 2022 (Leg 2). Referring to satellite SSH data, we detected sufficiently strong CCRs centered around 34.5° N, 150.0° E and 34.5° N, 156.5° E. During the cruises, we obtained three and a half hydrographic sections of the western ring, concurrent with the collection of atmosphere and ocean data along the ship track. Furthermore, we deployed a shallow-water profiling float (MOF) around the center of the ring.

We performed cluster analysis for the spiciness parameter calculated using the temperature and salinity profiles collected by Argo floats and the R/V Mirai and classified them into three clusters. The classification was consistent with the distributions of hydrographic profiles to the north and south of the KE. The interior of the western CCR was occupied with a fresh (negative spiciness anomaly) water mass from the north of the KE down to a depth of approximately 26.0 ${\sigma }_{\theta }$ (above a depth of approximately 200 dbar around the ring center). Unexpectedly, the SST and mixed layer water temperature were found to be uniform across the CCR. The SST signal within the ring could not be fully monitored by the satellite observation. Accordingly, we treated the impact of the detached CCR on the atmosphere differently from that of the KE meander with the SST signal. The turbulent heat flux estimated using the bulk flux algorithm was suppressed by wind drop-offs around the ring.

Furthermore, using the CCMP satellite sea surface vector wind data, we examined the characteristics of the wind drop-off by the rotary EOF analysis of the wind vector variation. The first rotary EOF mode (68%) represents the wind variation associated with a slight amplification over the KE current. The wind drop-off around the CCR was found to be represented by the second (12%) and third (7%) rotary EOF modes with the minima in the amplitudes immediately to the northwest and northeast of the CCR. The reduction in the wind speed relative to the surrounding region exceeded approximately 5 m s$^{-1}$, which is notably larger than values reported in preceding studies. Systematic convergent wind field was not associated with the wind drop-off. Thus, the reduction in wind speed is considered to be caused by the attenuation of the vertical mixing due to the stabilization of the atmospheric boundary layer. The northward shift of the minima of the wind drop-off may be attributed to the northward increase in the background stability of the atmospheric boundary layer in the KE region.

The MOF-observed potential temperature and salinity variations due to the diurnal entrainment of deep water from beneath the mixed layer base around the center of the ring. The mixed layer potential temperature slowly decreased at a rate equivalent to the seasonal cooling (${\sim }$ 0.1 °C day$^{-1}$). A freshening event with a magnitude of up to 0.4 day$^{-1}$, which is two orders of magnitude greater than and opposite to the seasonal variation in salinity, was observed by the MOF. This suggests that fresh cold water had been intruded from the north of the KE.

We expect that the mixed layer water convectively adjusts to the intensive heat loss at the sea surface in the KE region, particularly intensively outside of the CCR, and that the spatial contrast in SST and mixed layer temperature is diminished. The convective adjustment of the mixed layer water is too rapid for the variation to be fully detected through conventional temperature observations using research vessels, satellites, etc. Because SST is the only parameter that accounts for the variation in the ocean state in the estimation of the turbulent heat fluxes using the bulk flux algorithm, the calculated variations in turbulent heat fluxes are considered underestimated. Therefore, additional appropriate methods such as the eddy correlation flux measurement technique are required to investigate the air–sea interactions in the KE region.