Effects of mesoscale eddies on intraseasonal variability of intermediate water east of Taiwan

The intraseasonal variability of the intermediate water (IW) and its characteristics east of Taiwan are studied utilizing 17 months of long-term, continuous and synchronous measurements of temperature, salinity and current from mooring sites deployed at 122° E/23° N from January 2016 to May 2017. Direct measurements revealed water masses east of Taiwan alternately show complete South China Sea Intermediate Water or North Pacific Intermediate Water (NPIW) characteristics, mostly in a mixed state, with NPIW dominating 70% of the time. For the first time, it is demonstrated that the variation of IW with periods of ~ 80 days is mainly related to mesoscale eddies. Anticyclonic (cyclonic) eddies corresponding to an increase (decrease) in temperature and salinity in the intermediate layer. Further mechanism analysis indicates the vertical motion of the water mass inside the eddies is one of the reasons for the thermohaline change in the intermedaiter layer. In addition, the anticyclonic eddies may increase the salinity concentration gradient across the Luzon Strait, and the enhanced advection is favorable to the outflow of water masses in the South China Sea. When the cyclonic eddies acts on the eastern part of Taiwan, the influence of the northward advection is weakened and the southward flow on its left side is more favorable to the transport of NPIW.

The Kuroshio is a strong western boundary current in the North Pacific Ocean that originates from the North Equatorial Current. It brings heat and salt from low latitudes to mid and high latitudes and has an important impact on air-sea interactions and climate change along its path [1][2][3] . Therefore, understanding the characteristics of the Kuroshio water masses is of great significance for studying its poleward volume, heat and salinity transport and for making future climate predictions.
In the Northwest Pacific Ocean, the circulation background is complex, water masses are mainly divided into subsurface water masses and intermediate water masses (Fig. 1b), and each has distinct characteristics. Intermediate water is widely distributed in the North Pacific and cold with relatively low salinity at depths of 400 ~ 800 m; it has a minimum salinity of approximately 34.10-34.20 psu and a potential density of 26.8 σ θ , which generally refers to North Pacific Intermediate Water (NPIW) 4,5 . The NPIW primarily originates in the Okhotsk Sea and is transported via the Oyashio current along Kuril Island and east of the Japanese island of Hokkaido [6][7][8][9][10] . Then, NPIW spreads throughout the North Pacific and can extend southward to ~ 15°N by the subtropical gyre; it has even been found in the Halmahera eddy (HE) 11 . At the same time, NPIW is transported through the Luzon Strait to the South China Sea where it forms South China Sea Intermediate Water (SCSIW) with a salinity minimum of approximately 34.4 psu at core depths of approximately 500 m [12][13][14] . In addition, many studies defined another water mass, referred to as Kuroshio Intermediate Water (KIW), along the western boundary from Luzon Island to an area east of Taiwan, according to plots of the average temperature versus salinity ( θ-S) of the Kuroshio 7,15-17 . However, these papers report different sources of KIW. Mensah et al. 7 claim that KIW is mostly found in the West Philippine Sea Basin along the Kuroshio, with saltier variations in NPIW. According to Chern and Wang 16 , Chen 15 , and Nakamura et al. 17 , KIW may contain both SCSIW and NPIW and is mostly found east of Taiwan.
During ongoing research on intermediate water east of Taiwan, many discussions concern the presence of SCSIW east of Taiwan. In previous work, SCSIW with a salinity of approximately 34.4 psu was found on the continental shelf northeast of Taiwan 16,18 . In a survey of the Luzon Strait and the Okinawa Trough, Chen 15 also For analysis of properties of IW in this study, the average characteristics of the water masses in two typical regions were selected to represent North Pacific Intermediate Water (in the red A1 box) and South China Sea Intermediate Water (in the green A2 box). The map at the lower right corner of (a) shows the zoomed-in topography east of Taiwan. (b) Schematic diagram of the water mass distribution and circulation east of Taiwan. www.nature.com/scientificreports/ found evidence of SCSIW spreading to southern Japan though water east of Taiwan. Nakamura et al. 17 used climatological data to describe IW east of Taiwan as a mixture of SCSIW and NPIW. Recently, Mensah et al. 6 used temperature data at 580 m to develop an empirical formula to estimate IW salinity,they found that IW east of Taiwan contains SCSIW and NPIW. Mensah et al. 6 also showed that IW salinity correlated with thickness of the Kuroshio. However, Chern and Wang 16 did not identify SCSIW in the water north of Green Island, according to survey data from several hydrographic sections in the water east of Taiwan, and they claimed that the ridge between Taiwan and Green Island (less than 500 m deep) blocks the northward flow of water from the northern part of the South China Sea, and thus, prevents SCSIW from being carried into the area east of Taiwan. These findings suggest that intermediate waters along the Kuroshio east of Taiwan are highly variable, and researchers speculate that SCSIW and NPIW may alternate below the core of the Kuroshio. Actually, most of the studies of IW in the area east of Taiwan use only a small amount of sectional data acquired during surveys and provide characteristics at a given point in time. As an area with a high incidence of mesoscale eddies propagating westward from the STCC, the Kuroshio east of Taiwan is affected by mesoscale eddies and experiences intraseasonal variability with periods of 80 ~ 100 d 3,[19][20][21][22][23][24][25][26][27] . Due to the lack of simultaneous long-term continuous observations of the temperature, salinity and currents in this area, knowledge of the properties of IW east of Taiwan is remain limited, and this knowledge is not sufficient to clearly reveal the distribution characteristics and variations of IW in this area. For the first time, it has been possible to study the variations and processes of IW east of Taiwan using 17 months of simultaneous and continuous measurements of temperature, salinity and current. A subsurface mooring was deployed at 23° N, 122° E (red triangle in Fig. 1) from January 2016 to May 2017 to monitor the temperature, salinity and current upper 800 m. Although Mensah et al. 6 hypothesized in his paper that the variation of IW are related to mesoscale eddies, whether IW in the intermediate layer is influenced by mesoscale eddies needs to be determined by direct measurement data, more than that the variation characteristics of IW and the relationship between IW and the Kuroshio are also not clearly.

Data
Mooring system data. The Institute of Oceanography, Chinese Academy of Sciences (IOCAS), conducted a large-scale survey of the Western Pacific in January 2016. During this survey, we deployed a subsurface mooring east of Taiwan at a water depth of 4900 m; in May 2017, we successfully recovered the mooring system using the research vessel R/V Science (location: 122° E, 23° N, Fig. 1). Figure S1 shows a simplified schematic diagram of the configuration and deployment of the subsurface mooring buoy system. The main floating ball integrated two up-looking and down-looking 75 kHz Acoustic Doppler Current Profilers (ADCPs) manufactured by Teledyne RD Instruments (TRDI), designed for use at depths of approximately 400 m, to measure currents at depths above 800 m according to the following parameters: the measurement interval was 1 h, the number of depth cells was 74, the bin size was 8 m, and the number of pings was 30 per measurement. At depths of 400-1000 m, we used conductivity-temperature-depth meters (CTDs, type: SBE37, manufactured by Sea Bird Instrument) at intervals of 100 m, and the sampling interval was 10 min. The current data from the ADCPs were controlled for quality, including a good threshold of 70% and a cutoff of 2 m/s for current speed. To filter out the influence of high-frequency signals, all current and CTD data were averaged daily for this study.
AVISO altimetry and salinity data. An Archiving, Validation, and Interpretation of Satellite Data in Oceanography (AVISO) altimetry dataset was used in this paper. The sea level anomaly (SLA) and geostrophic current data were obtained from the commercial AVISO Global ARMOR3D L4 Reprocessed dataset (http:// marine. coper nicus. eu/ servi ces-portf olio/ access-to-produ cts/). The SLA data had spatial resolution of 1/4° × 1/4°, and the dataset extended approximately 17 months, from January 2016 to June 2017.

Time series of intermediate water.
Because the main float of the mooring system was affected by the current, its depth changes greatly, which led to deviations of the entire mooring system. Therefore, all the instruments and equipment designed for use at a predetermined depth were basically in a fluctuating state. The daily average data in Fig. 2 shows large fluctuations for a maximum floating depth of 300 m. The results for the current show the velocity structure and variation characteristics of the Kuroshio (Fig. S2). The most significant variation in the Kuroshio east of Taiwan is expressed as intraseasonal variations with periods of ~ 85 d, and these variations are mainly modulated by mesoscale eddies propagating westward from the STCC, mainly including Kuroshio velocity, transport and main axis migration, et al. 3,21,26 . The relationship between significant intraseasonal variations in the Kuroshio and IW is one focus of this study. The minimum salinity ( S min ) at the core of the intermediate water, shown in Fig. 2c, is found mainly in the potential density range of 26.6-26.8 σ θ , where S min is approximately 34.15 psu and the depth is approximately 600 m, with corresponding IW temperature range from 7 to 8 °C showed in Fig. 2a. We also found that S min at the core showed discontinuous variability; for example, S min at the core was approximately 600 m during the period March-April 2016, and S min was approxi- www.nature.com/scientificreports/ mately 550 m by September 2016. There are 7 results for S min at the core in the observation period, at the same time, we found in Fig. 2 that there is some upward increase in isothermal and the isopycnal during these low-salt events, while there is a downward trend in isothermal and the isopycnal except during these low-salt events. According to the total measurement time, and an intraseasonal variation period of approximately 70-80 days was estimated. The mean salinity of each layer is shown in Fig. 2d, although the overall standard deviation was relatively small, the variance was larger at 500-550 m than in the other layers, indicating a relatively large variation in salinity in the middle layer. Of course, the standard deviation of temperature measurements decreases with increasing depth showed in Fig. 2b.

T-S characteristics of intermediate water.
To more clearly analyze the characteristics of IW east of Taiwan, we drew a T-S scatter plot of data obtained from the moored CTDs (Fig. 3a). For comparison, historical data from the Argo international project for average temperature and salinity east of Taiwan but away from the Kuroshio area (box A1 in Fig. 1a) and the South China Sea area (box A2 in Fig. 1a) represent NPIW and SCSIW, respectively. The S min values of NPIW and SCSIW are 34.18 psu and 34.39 psu, respectively. The main characteristics of IW in the water east of Taiwan are as follows: S min varies from 34.15 to 34.4 psu, corresponding to a temperature change in the range of 7 to 8 °C and a potential density variation of 26.6-26.8 σ θ for S min , respectively. The salinity distribution near the characteristic salinity value of NPIW is more concentration than SCSIW, also demonstrated form panel A in Fig. 3a (near the red curve), indicating that the overall characteristics of the water mass are closer to those of NPIW during the observations. Numerical results show only two moments in time when S min exceeded 34.39 psu in Fig. 3a, indicating that observations of IW with typical SCSIW characteristics are relatively infrequent. That is, most of the time, IW in the water east of Taiwan resembles a mixture of NPIW and SCSIW. Chern and Wang 16 proposed that the SCSIW could not flow directly to eastern Taiwan due to blocking by the Green Islands in southeastern Taiwan, because the ridge between Taiwan and Green Island is shallow than 500 m. Mensah et al. 6 report that NPIW and SCSIW could mix at the relatively southern location of the Luzon Strait, and the Kuroshio carried this water mass to the east of Taiwan. Therefore, we can obtain the typical average characteristics of this water mass, and the average θ -S curves are plotted in Fig. 3a (blue curve). S min and the corresponding temperature of this water mass were 34.28 psu and www.nature.com/scientificreports/ 7.5 °C, respectively. The core was located at 550 m, which corresponded to a depth between the S min core depths of 500 m for SCSIW and 600 m for NPIW. In this paper, a more widely used definition of KIW is adopted, which is considered to be mainly mixing between NPIW and SCSIW. To explore the ratio of the two water masses east of Taiwan, the concentration calculation equation was applied. First, salinity values of 34.19 psu and 34.39 psu were taken to characterize low-salt core of NPIW and SCSIW, respectively, according to the red and green curves in Fig. 3a, and the S min of each profile measured by the CTDs was taken to be the value of the two mixtures. Then, these three values were introduced into the concentration equation to calculate the mixing ratio. Figure 3b shows that the proportion of NPIW in KIW was relatively high, and 70% of the time, the NPIW mixing ratio exceeded 60%. There were approximately 7 moments in time when the proportion of NPIW reached 100%, which meant that there was basically only NPIW in the area east of Taiwan; these moments were evenly distributed during the observation period The results of previous studies basically on cruise sections data, but can only provide characteristics at certain moments.Therefore, SCSIW was found to exist east of Taiwan 17,18 , while others were not found in their study 16 . We used continuous salinity monitoring data to demonstrate that dynamic changes in salinity in the region may be responsible for the observed differences in salinity characteristics in these studies.  6 reported that the intraseasonal period of IW east of Taiwan was ~ 100 days. Compared with the result of Mensah et al. 6 , the intraseasonal signal of IW obtained from directly measured salinity data in this study may be more realistic reflection of the variation characteristics of the water mass. Of course, it is also possible that the difference in results is due to the timing of the two observations. To better understand the intraseasonal variability, the meridional velocity anomaly, temperature and salinity anomaly were calculated by subtracting the average value during the observation of each layer from the average daily mooring data displayed in Fig. 5a-c. The meridional velocity anomalies are basically consistent and banded in the 0-800 m range, and temperature and salinity anomalies also exhibit synchronization in the 400-800 m range. The alternating band structures of positive and negative shapes are clearly shown in the anomalies temperature and salinity graphs, also indicating an intraseasonal signal of approximately 3 months. The maximum negative and positive salinity anomalies were − 0.12 psu and 0.1 psu, and the maximum negative and positive temperature anomalies were − 1.5 °C and 2 °C, respectively. During the observation period, there were 6 negative salinity anomalies in 17 months, March-April, June, September-October, and November-December in 2016 and January-February and April-May in 2017; the positive anomalies occurred during the other observation times. Meanwhile, the temperature and salinity anomalies were consistent and showed synchronous changes. The distribution of the integrated current anomalies showed that at most moments, positive meridional velocity anomalies corresponded to positive anomalies of temperature and salinity, while negative meridional velocity anomalies corresponded to negative anomalies of temperature and salinity. The relatively consistent variation in current, temperature and salinity suggests that all three parameters may be influenced by the same factor.

Discussion
Intraseasonal variability of IW caused by mesoscale eddies. The wavelet power spectrum of the SLAs located at mooring site in Fig. 4a,b showed the strong intraseasonal variability with periods ~ 80 days throughout the observation time, indicating a possible relationship between mesoscale eddies and IW. Also, we checked the local wind stress has a period of ~ 15 days (figure not shown) is inconsistent with the ~ 80 days variability of the IW. This maybe indicates that the local wind is not a direct cause of the intraseasonal variability of the IW. To identify the relationship between mesoscale eddies, Kuroshio velocity and IW, we plotted the time series of the SLAs; the salinity averages were between 26.4 and 27.0 σ θ and the Kuroshio velocity averages between 0 and 400 m are shown in Fig. 6a. The results showed that they exhibited significantly consistent variations, with correlation coefficients of 0.63 and 0.52 between the SLA and Kuroshio velocity and between the SLA and salinity, respectively. Figure 6b,c show scatter plots of the measured potential temperature against salinity, composite with the SLAs and averaged V, respectively. Most of the fresher (saltier) water corresponded to negative (positive) SLAs displayed in Fig. 6b. Additionally, stronger currents carry saltier water, while only weak  (Fig. 7). Figure 7a-g show that salinity lagged behind SLAs from 0 to 60 days with an interval of 10 days. East of Taiwan, there was an area with a positive correlation coefficient (red) that gradually moved westward with www.nature.com/scientificreports/ time from the 60th day onward, and the correlation coefficient increased with its maximum increasing from 0.3 on the 60th day to 0.5 on the 0th day. There was a significant area with a positive correlation coefficient located at 125° E/22° N with a 60-day lag relative to the mooring site, while at the highest area located at 122° E/22.5° N with 0 lag time; as a result, the estimated westward propagation velocities of the mesoscale eddies were approximately 10 cm/s. As a comparison, Tsai et al. 27 used PIES data to conclude that the westward speed of an eddy near Taiwan was in the range of 11-24 cm/s. Also, the westward propagation speed of mesoscale eddies of first-mode baroclinic Rossby waves near 20° N was approximately 8 cm/s 2 .
Possible mechanism of IW variations. Measurements at the mooring site, a single point, are not enough to reflect the movement of salinity and the current-water mass relationship. Therefore, a wider range of Global Ocean Multi Observation Products (GOOP) data from global ocean observation was used to analyze this behavior. First, we examined the consistency between data from the moored CTDs and GOOP. The time series chart of GOOP salinity data is shown in Fig. S3 and compared to salinity data from the moored system in Fig. 2c. The salinity characteristics corresponded better at several moments, e.g., the low salinity characteristics in January, April, June, September, and November 2016 and January and May 2017, and the relatively high salinity characteristics in the period from July to August 2016, than at other times. Although the agreement was not complete, it is enough to indicate that the variation characteristics of salinity from GOOP are similar to those from the in situ measurements. Actually, the salinity at the intermediate layer was analyzed according to horizontal and vertical movement. First, we constructed the vertical structure of the anticyclonic and cyclonic eddy times to discuss IW movement in the vertical direction. Figure 8a-d show the anomalies temperature and salinity distributions from 0 to 1000 m along the center of cyclonic and anticyclonic eddies, respectively. In anticyclonic (cyclonic) eddies, isothermal and isosalinity lines showed obvious concave (convex) structures corresponding to vertical downward (upward) movement of the water mass at the center, corresponding the positive (negative) temperature anomalies. However, the structure was different for temperature and salinity because the water mass east of Taiwan was divided if the vertical motion is relatively small and mixing is weakening by weak anticyclonic eddies, even if the salinity increases in intermediate layer, but the water mass may still exhibit the characteristics of NPIW. Combined with the above analysis, the vertical motion of the water masses caused by the mesoscale eddies is one of the reasons for the temperature and salinity changes inside the eddies. However, it can also be considered that the salinity and temperature changes does not mean that the water mass will definitely change from NPIW to SCSIW, and it may also need to be considered from the perspective of horizontal transport mixing of different water masses caused by mesoscale eddies.   Fig. 9 show that there is a higher salinity gradient from west to east during the anticyclonic eddies period, also most of the Luzon Strait is occupied by high salinity SCS water. When cyclonic eddies impinge on the Kuroshio (Fig. 9a), the current is weakened by cyclonic eddies, and relatively fresher water surrounds the intermediate layer east of Taiwan. Additionally, Fig. 9a shows that little water is carried out of the SCS by the Kuroshio, which means that NPIW more easily dominates water east of Taiwan and maintains lower salt characteristics. In contrast, during anticyclonic interactions with the Kuroshio, the current is strengthened to the north, and the field of salinity is significantly higher in the intermediate layer along Luzon Island to the Taiwan coast (Fig. 9b).
In fact, we speculate that the increase in IW salinity may be associated with the path of Kuroshio intrusion into the SCS caused by the anticyclonic eddies. When the presence of the anticyclonic eddies in the east of Luzon Strait, forcing the Kuroshio to bend toward the SCS, will lead to the Kuroshio intrusion into the South China Sea to form a loop structure, can reach a depth of about 1000 m [28][29][30] . Therefore, the horizontal flow velocity and vertical influence depth of the Kuroshio will be enhanced, and may be more SCS water will be brought out from the SCS to the area east of Taiwan, thus increasing the salinity of the intermediate layer. During the cyclonic eddies, the northward transport capacity weakens with the Kuroshio weakens. Qu et al. 31 and Wang et al. 32 report that most of the intermediate water layer flows southward east of Luzon, even the intermediate water flow occasionally reverses to a southward course at southeast of Taiwan 6 , this also can be find in Fig. S2. Therefore, it is may be relative difficult for the SCS waters out of the Luzon Strait to be transported to the north by weaken Kuroshio, the southward flow on the west font of the cyclonic eddies restrains the northward transport and is able to carry the NPIW on the east side to the east of Taiwan.
Therefore, the results reveal that horizontal movement of IW carried by current is possible as follows. Anticyclonic eddies strengthened the Kuroshio and benefitted SCSIW flowing through the Luzon Strait, leading to an increase in salinity east of Taiwan. While cyclonic eddies weakened the Kuroshio and reduced SCSIW outflow from the Strait, this was conducive to relatively low-salt NPIW occupying the area east of Taiwan. Evaluating the influence of the IW east of Taiwan by the way and path of the Kuroshio intrusion into the SCS is the next step to be explored.

Summary
This study utilizes 17 months of long-term, continuous and synchronous measurements of temperature, salinity and current data from a mooring site located at 122° E/23° N east of Taiwan. The IW characteristics in eastern Taiwan were revealed as follows: S min varied from 34.15 to 34.4 psu, corresponding to a temperature variation in the range from 7 to 8 °C and a potential density variation of 26.6-26.8 σ θ for S min . We observed pronounced intraseasonal variations in IW with periods of ~ 80 days.
For the first time, we prove that intraseasonal variability (ISV) of IW are caused by mesoscale eddies propagating westward from the STCC. The mesoscale eddies not only influenced the Kuroshio but also brought intraseasonal signals to the water mass of the intermediate layer. The correlation coefficients between SLAs and current and between SLAs and salinity in the intermediate layer were 0.63 and 0.52, respectively. Results showed that positive (negative) SLAs, caused by the westward propagation of anticyclonic (cyclonic) eddies from the STCC, increased (decreased) the speed of the Kuroshio, also increase (decrease) the temperature and salinity in the www.nature.com/scientificreports/ 400-600 m in east of Taiwan. By using the map of correlations between SLAs and S min , the westward propagation speed of the mesoscale eddies was estimated to be approximately 10 cm/s. Combined the GOOP data, the movement of IW are discussed. The vertical movement of IW showed that the temperature and salinity increased (decreased) in the intermediate layer due to the vertical movement of water by anticyclonic (cyclonic) eddies. Meanwhile, during the horizontal movement of the water mass, anticyclonic eddies strengthened the Kuroshio and benefitted SCSIW flowing through the Luzon Strait, leading to an increase in salinity east of Taiwan. While the cyclonic eddies weakened the Kuroshio and reduced SCSIW outflow from the Strait, conditions were conducive to relatively low-salt NPIW in the area east of Taiwan.