Subtropical Mode Water in a recent persisting Kuroshio large‐meander period: part I—formation and advection over the entire distribution region

Since August 2017, the Kuroshio has taken a large-meander (LM) path, which has forced the Kuroshio extension (KE) to be in its stable state against its wind-forced decadal variability. How such current conditions have impacted the formation and advection of North Pacific subtropical mode water (STMW) over its distribution region was examined using Argo float data during 2005–2020. Out of the whole STMW defined as a low-potential vorticity layer of 16–19.5 oC, a relatively cold variety of 16–18 oC, which was formed south of the KE and advected westward and southward, occupied more than 80% of the total volume. The formation rate of the 16–18 oC variety was low during 2006–2009 in an unstable-KE period and high during 2010–2015 in a stable-KE period, and then dropped drastically in 2016 despite the KE still being in the stable state. After a short unstable-KE period in 2016–2017, the LM-forced, stable-KE period began, but the formation rate of the 16–18 oC variety has not restored, possibly due to stronger background stratification propagated from the central North Pacific. In addition, the 16–18 oC variety has had to make a southern detour around the LM, and its westward advection from the formation region south of the KE to the region south of Japan has been significantly decreased, possibly because it is dissipated more strongly over a southern part of the Izu–Ogasawara Ridge. Due to such decline in the formation and advection, the volume of the 16–18 oC variety and hence that of the whole STMW have gradually decreased since 2016.


Introduction
The Kuroshio has taken a large-meander (LM) path south of Japan since August 2017 (Usui 2019). The LM path frequently occurred in 1975-1991(Kawabe 1995. Then, a non-large-meander (NLM) path (i.e., a nearshore NLM path and an offshore NLM path) had dominated for a quarter century, except for a 1-year LM period from July 2004 through August 2005.
The LM occurrence impacts currents and water masses in the surrounding area. Recently, it reset the decadal variability of the Kuroshio extension (KE) system , which had been controlled by the wind stress curl variability in the central North Pacific associated with the Pacific decadal oscillation (PDO; Mantua et al. 1997) and the subsequent westward propagation of sea surface height (and main thermocline depth) anomalies as baroclinic Rossby waves (Qiu and Chen 2005;Qiu et al. 2007). The KE transitioned from a stable state to an unstable state in mid-2016 when negative sea surface height anomalies generated in the warm PDO phase since 2014 started arriving in the area east of Japan, and was predicted to remain in the unstable state for several years (Qiu et al. 2014. However, the KE unexpectedly switched back to the stable state in late 2017, because when the Kuroshio takes an LM path, its passage over a northern channel of the Izu Ridge (Kawabe 1985) stabilizes the KE path in the downstream (Qiu and Chen 2005;Sugimoto and Hanawa 2012;Qiu et al. 2020). Thus, 1 3 the KE has been in the stable state since 2010, except for a short unstable period in 2016-2017. The Kuroshio LM also influences the North Pacific Subtropical Mode Water (STMW; Masuzawa 1969), which is formed as deep winter mixed layers south of the Kuroshio and the KE (Suga and Hanawa 1990) and advected southwestward by the mean flow in the southern part of the Kuroshio recirculation gyre (Suga and Hanawa 1995a) and by eddy activities (Uehara et al. 2003;Nishikawa et al. 2010;Xu et al. 2014Xu et al. , 2016Xu et al. , 2017aNakano et al. 2021). By analyzing climatological data and the repeat hydrographic section along the 137ºE meridian, earlier studies (Suga et al. 1989;Suga and Hanawa 1995a, b) demonstrated that the westward advection of STMW from the region south of the KE to the region south of Japan was interrupted during LM periods because the LM divided the Kuroshio recirculation gyre into two parts. On the other hand, Oka et al. (2019), who recently calculated the cross-sectional area of STMW in the same 137ºE section in 1967-2016, pointed out that the cross-sectional area dropped significantly during the longest observed LM period of 1975-1980, but did not fall in the other, shorter LM periods. There is, as such, a need to examine how these results can reconcile with each other.
STMW has also been modulated by the decadal variability of the KE system. For more than two decades since 1993 Rainville et al. 2014;Oka et al. 2015;Cerovečki and Giglio 2016), or possibly since late 1970s (Oka et al. 2019), the STMW formation south of the KE has decreased (increased) during unstable (stable) KE periods, presumably because enhanced (reduced) input of highpotential vorticity water north of the KE into the STMW formation region south of the KE hinders (facilitates) deepening of winter mixed layers there Oka et al. 2015). It is of interest to examine whether such decadal STMW formation variability is still valid in recent years when the Kuroshio LM has been stabilizing the KE in spite of the arrival of PDO-related negative sea surface height anomalies from the central North Pacific.
The current Kuroshio LM period since August 2017 has already exceeded three years, being the second longest observed period. In addition, it is different from the longest observed period of [1975][1976][1977][1978][1979][1980] in that the large-scale circulation of STMW is fully monitored by the Argo profiling float network (Roemmich et al. 2001;Oka 2009;Oka et al. 2015). Utilizing this opportunity, this paper examines the formation and advection of STMW over its entire distribution region in relation to a persisting Kuroshio LM and the resultant stable state of the KE. The accompanying paper (Nishikawa et al. in prep.) investigates how anomalously warm STMW, which is formed in the recirculation gyre off Shikoku when the Kuroshio takes an LM path or an offshore NLM path (Sugimoto and Hanawa 2014), has been formed and maintained in the current LM period.

Data and method
We used temperature (T) and salinity (S) data from Argo profiling floats in the North Pacific in 2005-2020, which were downloaded from the ftp site of the Argo Global Data Assembly Center (ftp:// usgod ae. org/ pub/ outgo ing/ argo, ftp:// ftp. ifrem er. fr/ ifrem er/ argos, https:// doi. org/ 10. 17882/ 42182) and edited as outlined in Oka et al. (2007). After discarding profiles shallower than 500 dbar and those in marginal seas, we vertically interpolated each profile onto a 1-dbar grid using the Akima spline (Akima 1970).
We also used T and S data obtained along the 137ºE section by the Japan Meteorological Agency (JMA) biannually in winter (latter half of January) since 1967 and in summer (around July) since 1972, up to 2020 (Oka et al. 2018). Gridded T and S data at every 20'-1º meridional and 1 dbar vertical interval are made using optimal interpolation by the JMA and are publicly available online (http:// www. data. jma. go. jp/ gmd/ kaiyou/ db/ mar_ env/ resul ts/ OI/ 137E_ OI_e. html). Note that the gridded data used in this study are different from those used in our previous studies (Oka et al. 2017(Oka et al. , 2018(Oka et al. , 2019, as the JMA revised the optimal interpolation procedures in 2018. This yields slightly different values of STMW properties from our previous studies. From the Argo float and 137ºE T/S data, we calculated potential temperature (θ), potential density (σ θ ), and potential vorticity (Q). Here, Q is defined as Q = gf ∂σ θ /∂p, neglecting relative vorticity , where g is the gravity acceleration, f the Coriolis parameter, and p is pressure.
Using these values, STMW in the 137ºE section in 1967-2020 was detected as non-outcropping areas with θ = 15.0-19.5ºC and Q < 2.0 × 10 -10 m −1 s −1 , as in Oka et al. (2019). STMW in Argo profiles in 2005-2020 is defined as layers with θ = 16.0-19.5 ºC and Q < 2.0 × 10 -10 m −1 s −1 , except for those existing only at depths less than 100 dbar. This definition differs from that used in our previous studies using Argo profiles, i.e., a layer where Q is lower than 2.0 × 10 -10 m −1 s −1 and θ at the Q minimum (called core θ hereafter) is between 16 and 19.5 °C (Oka 2009;Oka et al. 2011Oka et al. , 2015. In the new definition, we permit the existence of more than one STMW layer in each profile, which has been frequently observed in mode waters (Taneda et al. 2000;Oka et al. 2011Oka et al. , 2020Gao et al. 2016;Liu et al. 2017Liu et al. , 2019. It should also be noted that the θ range for the STMW definition is different between the 137ºE and Argo float data; the narrower and relatively warm θ range for the more contemporary Argo data reflects long-term warming of STMW (Sugimoto et al. 2017;Oka et al. 2019).

Results
The time series of cross-sectional area of STMW in the 137ºE section (Oka et al. 2019) was extended to the latest observation in summer 2020 ( Fig. 1). Since mid-1970s, the cross-sectional area had increased (decreased) during stable (unstable) KE periods with a time lag of about one year, as demonstrated by Oka et al. (2019). Since 2016, however, it has declined greatly in spite of the prolonged stable-KE period. Such decadal variations were commonly observed in the winter and summer 137ºE sections.
Decomposition of the cross-sectional area into θ bins reveals that the STMW in the winter 137ºE section consisted mainly of 16-17 ºC and 17-18 ºC varieties, the sum of which well explained the decadal variability of the total cross-sectional area (Fig. 2a). In addition, there are smaller contributions by the 15-16 ºC variety, mainly prior to mid-1990s, and the 18-19 ºC variety after mid-1990s, which also reflects the long-term warming of STMW. To examine in which region each variety of STMW was formed in the past 15 years, zonal distribution of core θ in spring was plotted over the formation region using Argo profiles (Fig. 3). As demonstrated by previous studies (Suga and Hanawa 1990;Oka and Suga 2003), colder STMW tended to be formed toward the east, but the zonal change of core θ was not continuous. In most years, relatively cold STMW of 16-18 ºC was formed east of 140ºE, while relatively warm STMW of 18-19.5 ºC was formed west of 150ºE and intensively west of 140ºE. There seems to be a gap of core θ near 140ºE, as demonstrated by Sugimoto and Hanawa (2014) for LM and offshore NLM periods of the Kuroshio during which the Kuroshio recirculation gyre split into two parts. Thus, STMW appearing in the winter 137ºE section consists mainly of that colder than 18 ºC formed south of the KE.
In the summer 137ºE section (Fig. 2b), the 16-17 ºC and 17-18 ºC varieties dominated the KE-related decadal periods of the KE during 1977-1992 based on a numerical hindcast (Qiu et al. 2014) and after 1992 based on satellite altimeter data . Shade denotes LM periods of the Kuroshio variability, as in the winter section. In addition, the 18-19 ºC variety originating near the 137ºE section occupied a comparable cross-sectional area to the 16-17 ºC and 17-18 ºC varieties, except in LM periods during which the 18-19 ºC variety almost disappeared. Argo float data supportively exhibit that in the late winters of 2005 and 2018-2020 during the LM period, the 19-19.5 ºC variety was formed south of Japan and the 16-17 ºC and 17-18 ºC varieties were formed south of the KE, while the 18-19 ºC variety was absent (Fig. 3).
Since 2016, not only the 18-19 ºC variety in the summer 137ºE section but also the 16-17 ºC and 17-18 ºC varieties in both the winter and summer sections have shrunk (Fig. 2). This suggests that the westward advection of STMW from the region south of the KE has been interrupted by the LM since summer 2017, as demonstrated by the previous studies (Suga et al. 1989;Suga and Hanawa 1995a,b). There is also a possibility that the STMW formation south of the KE has decreased, possibly in relation to the LM. Therefore, Argo float data will be analyzed next to examine the formation and advection of STMW over its entire distribution region in the past 16 years.
The STMW thickness from individual Argo float profiles (if there are more than one STMW layer in a profile, their thickness was summed up) was horizontally interpolated and mapped for each month in the same manner as Oka et al. (2015). Due to the change in the STMW definition (Sect. 2), the total volume of STMW in the study region of 15-40ºN, 120ºE − 170ºW during 2005-2014 (black curve in Fig. 4) is larger than that presented by Oka et al. (2015;their Fig. 7a) by 9% on average, while the two time series exhibit similar seasonal and decadal variability (correlation coefficient = 0.98), including a decrease in 2006 − 2009 in the unstable KE period and an increase after 2010 in the stable KE period. Since 2016, the total volume has gradually declined in spite of the extended stable-KE period, exhibiting the same tendency as observed in the cross-sectional area in the 137ºE section (Fig. 1). Oka et al. (2015) divided the STMW volume into two regions north and south of 28ºN where STMW is ventilated and unventilated, respectively (Oka and Suga 2003;Oka 2009). We further divide these regions zonally at 140ºE where the gap of core θ existed (Fig. 3), defining four regions: the NE region at 28-40ºN, 140ºE-170ºW, the NW region at 28-40ºN, 120-140ºE, the SE region at 15-28ºN, 140ºE-170ºW, and the SW region at 15-28ºN, 120-140ºE (Fig. 4). In the NE region, the STMW volume exhibited a distinct seasonal variation with a maximum in February-April and a minimum in November-December. The volume in the NW region showed a less clear seasonal variation with a maximum around February-May and a minimum around September-December. Seasonal variations were much smaller in the SE and SW regions, indicating that STMW observed in these regions were formed in the NE and NW regions and advected southward. As the formation location within the NE and NW regions critically depends on the θ of STMW (Fig. 3), it is instructive to also decompose the STMW volume into θ bins.
The volume of the 16-18 ºC variety (Fig. 5a) accounted for the great majority of the whole STMW layer (16-19.5 ºC; Fig. 4), as in the 137ºE section (Fig. 2), and exhibited a similar variation to the whole STMW layer in each region, except in the NW region where the 16-18 ºC variety lacked seasonality. Seasonal variations were seen only in the NE region where the volume reached a maximum in March-April and a minimum in November-December. The 16-18 ºC variety is formed in this region and advected westward and southward to the other three regions.  (Fig. 5b). It tended to be smaller (larger) than the erosion volume during the former (latter) period, causing a decadal variation of volume in this region with a minimum in 2009 and a maximum in 2014 (Fig. 5a). This decadal volume variation spread to the other regions through advection; the Fig. 6 a Distribution of mixed layer depth in March where θ at 10-dbar depth is 16-18 ºC, based on Argo float data. Mixed layer depth, which is defined as the shallower value of the depth at which σ θ increases by 0.125 kg m −3 from 10-dbar depth and that at which θ changes by 0.5 ºC from 10-dbar depth (e.g., Levitus 1982;Suga et al. 2004), was computed for each float profile and then mapped in the same manner as the STMW thickness. b Monthly time series of mixed layer volume where θ at 10-dbar depth is 16-18 ºC in the region of 15-40ºN, 120ºE-170ºW, based on Argo float data annual-mean volume in the NW, SE, and SW regions during 2005-2015 was highly correlated (coefficient > 0.9) with that in the NE region when lagged by 1 year.
In 2016, the formation volume dropped to 0.23 × 10 15 m 3 despite the KE still being in the stable state (Fig. 5b). Late winter mixed layers in the NE region of this year were anomalously shallow (Fig. 6) and warm (Fig. 3), and became relatively thin STMW over an unventilated, colder STMW layer formed in the previous year. (Such insufficient development of mixed layers in the NE region in the late winter of 2016 will be an interesting theme for future studies.) The formation volume increased to 0.35 × 10 15 m 3 in 2017 in the short unstable-KE period, but did not recover any further in the following LM-forced stable-KE period (Fig. 5b) (Fig. 5a). At the beginning of the stable-KE period since 2010, the annualmean volume increased by 68% from 2010 to 2012 in the NE region and by 32, 181, and 185% from 2011 to 2013 in the three other regions. After the current LM began, however, the annual-mean volume in the NE region increased by 6% from 2017 to 2019, while that in the SE, NW, and SW regions dropped by 22, 36, and 39%, respectively, from 2018 to 2020. The unanticipated volume decline in the downstream regions, particularly the larger decrease in the NW and SW regions west of 140ºE, suggests that the westward advection of STMW was interrupted by the LM, as demonstrated by the previous studies. When we look at the thickness distributions of the 16-18 ºC variety in recent years (Fig. 7), the 16-18 ºC variety thicker than 100 dbar had been advected westward from the region south of the KE to the region south of Japan prior to mid-2017. Since the current LM began, the 16-18 ºC variety has had to make a southern detour around the LM to enter the region south of Japan, which likely led to the decreasing advection with time. Thus, the reduced formation in the NE region since 2016 and the decreasing advection to the downstream regions since 2018 have contributed to the decline of annual-mean total volume of the 16-18 ºC variety from 0.74 × 10 15 m 3 in 2015 to 0.55 × 10 15 m 3 in 2020 (Fig. 5a).
The 18-19 ºC variety occupied a much smaller volume than the 16-18 ºC variety (Fig. 8). Its volume in the NE region exhibited a seasonal variation with a maximum in January-February (Fig. 8a), which is 2 months earlier than the peak of the 16-18 ºC variety in this region (Fig. 5a). This implies that some portion of the 18-19 ºC variety observed in January-February was further cooled to become the 16-18 ºC variety in March-April. Consistently, the 18-19 ºC variety widely spread over the NE region in January-February (not shown), but was found mostly west of 150ºE in April (Fig. 9).
The 18-19 ºC variety was also formed in the NW region and, to the less extent, in the SE and SW regions south of 28ºN, where its volume was at a peak in March-April of particular years such as , 2008, 2011-2015. On the other hand, its formation in the NW region almost ceased in 2005 and 2018-2020 during the LM period, as observed in the 137ºE section (Fig. 2). After spring, the 18-19 ºC variety shrank before being substantially advected southward (Fig. 9). Its volume in the NW and SW regions tended to be larger in July than in the following January (Fig. 8b, d), which is consistent with its crosssectional area in the 137ºE section being larger in summer than in winter (Fig. 2).
The 19-19.5 ºC variety also occupied a small volume in the whole STMW layer (Fig. 10). In the NE region, its volume reached a maximum mostly in January and was small after spring (Fig. 10a). Such a seasonal variation probably reflected the formation process of colder varieties of STMW and not that of the 19-19.5 ºC variety. In the NW region, the 19-19.5 ºC variety had the largest volume in March-May (in other words, it was formed) in 2005, 2009, 2018. Interestingly, these were the years in which the 18-19 ºC variety was not formed in this region (Fig. 8b). The formation of the 19-19.5 ºC variety in 2018 and 2020 in the current LM period supports Sugimoto and Hanawa (2014), who analyzed Argo float data in 2005-2011 to demonstrate that STMW warmer than 19 ºC was formed in the recirculation gyre south of Japan when the Kuroshio took an LM or offshore NLM path.
Like the 18-19 ºC variety, the 19-19.5 ºC variety shrank after spring and almost disappeared by fall, but exceptionally survived for more than one year south of Japan in 2018-2019 (Figs. 10b, 11). Its formation and temporal evolution in relation to colder varieties of STMW in the current LM period will be detailed in our accompanying paper (Nishikawa et al. in prep.).

Discussion
Replotting the volume of the 16-18 ºC, 18-19 ºC, and 19-19.5 ºC varieties together for each region (Fig. 12) confirms the dominance of the 16-18 ºC variety in the NE, SE, and SW regions. In the NE region where the 16-18 ºC variety is formed, it had an average volume of 0.31 × 10 15 m 3 during 2005-2020, occupying 84% of the whole STMW layer (Fig. 12b). This ratio increased up to 91% when only values in April-December without formation months were considered. In the SE and SW regions south of 28ºN where the 16-18 ºC variety is not ventilated, its volume amounted to 0.24 × 10 15 m 3 on average, occupying 88% (Fig. 12d, e).
The ratio was also high (82% on the annual average and 86% for April-December) in the whole study region (Fig. 12a).
The 16-18 ºC variety is also the only variety that had fluctuated decadally in all regions in relation to the decadal KE variability until 2015. What made its formation  (Fig. 5b)? As the eddy activity in the KE region was low in both periods, the southward transport of high-Q and fresher water north of the KE to the STMW formation region, which presumably hinders deepening of winter mixed layers there Oka et al. 2015), was expected to have stayed small. Another factor that can affect the STMW formation is the PDO-related sea surface height anomalies propagating from the central North Pacific, which had controlled the decadal KE variability until the current, LM-forced stable-KE period began. As negative sea surface height anomalies have been arriving since mid-2016, the associated stronger background stratification above the middle of the main thermocline Sugimoto and Hanawa 2010) is expected to have hindered the STMW formation.
How is the westward advection of the 16-18 ºC variety from the NE region to the SW and SE regions interrupted by the LM? In addition to the longer advection route associated with the southern detour, enhanced diapycnal mixing over the Izu-Ogasawara Ridge along ~ 140ºE (Hibiya et al. 2007;Whalen et al. 2012;Xu et al. 2017b) might have an influence. Xu et al. (2017a) analyzed Argo float data in 2004-2016 to demonstrate that thick STMW, possibly trapped in anticyclonic eddies, preferentially passed through a bathymetric gap of the ridge near 30ºN. When the 16-18 ºC variety makes a southern detour and passes over the ridge south of the gap (Fig. 7), it might be dissipated more strongly there. Fig. 8 Monthly time series of volume of STMW with θ = 18-19 ºC in the whole study region (a; black curve) and the NE (a; red curve), NW (b), SE (c), and SW (d) regions, otherwise following Fig. 4. Note that the scale on the ordinate is half that of Figs. 4, 5

Summary
Argo float data in 2005-2020 have been analyzed to examine the formation and advection of STMW over its entire distribution region in relation to the recent, persisting LM of the Kuroshio and the LM-forced stable state of the KE. Out of the whole STMW defined as a low-Q layer of 16-19.5 ºC, a relatively cold variety of 16-18 ºC, which was formed east of 140ºE and north of 28ºN in late winter and then advected westward and southward (Fig. 13a), occupied more than 80% of the total volume. The formation volume of the 16-18 ºC variety was small during 2006-2009 in an unstable-KE period and large during 2010-2015 in a stable-KE period, and then dropped In addition, since the current LM began in summer 2017, the 16-18 ºC variety has had to make a southern detour around the LM, and its westward advection from the formation region south of the KE to the region south of Japan has been significantly decreased (Fig. 13b), possibly because it is dissipated more strongly over a southern part of the Izu-Ogasawara Ridge. Due to such decline in the formation and advection, the volume of the 16-18 ºC variety and hence that of the whole STMW have gradually decreased since 2016. The cross-sectional area of STMW in the 137ºE repeat hydrographic section has also dropped rapidly since 2016, as previously seen in the longest observed LM period of 1975-1980.
The 18-19 ºC variety of STMW was formed south of Japan and south of the KE west of 150ºE in late winter in some years, while the 19-19.5 ºC variety was formed south of Japan in the other years. After spring, both varieties rapidly shrank before being substantially advected southward and almost disappeared by fall. Since the current LM began, the 19-19.5 ºC variety was formed in the late winter of 2018, survived for more than a year, and was renewed in the late winter of 2020. These events will be described in detail in our accompanying paper (Nishikawa et al. in prep).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.

Fig. 13
Schematic illustrating the formation and advection of STMW with θ = 16-18 ºC in a non-LM periods and b LM periods of the Kuroshio. Blue shading denotes the formation region. Light blue shading with thin black arrows represents advection from the formation region. Thick black arrow indicates the path of the Kuroshio and the KE