Mindanao Current and Undercurrent: Thermohaline Structure and Transport from Repeat Glider Observations

AUGUST 2017 S C H Ö NAU AND RUDNICK 2055

MARTHA C. SCHÖNAU AND DANIEL L. RUDNICK Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California

(Manuscript received 15 December 2016, in ﬁnal form 3 May 2017)

ABSTRACT

Autonomous underwater Spray gliders made repeat transects of the Mindanao Current (MC), a low- latitude western boundary current in the western tropical North Pacific Ocean, from September 2009 to 2 2 October 2013. In the thermocline (,26 kg m 3), the MC has a maximum velocity core of 20.95 m s 1, weakening with distance offshore until it intersects with the intermittent Mindanao Eddy (ME) at 129.258E. In 2 the subthermocline (.26 kg m 3), a persistent Mindanao Undercurrent (MUC), with a velocity core of 2 0.2 m s 1 and mean net transport, flows poleward. Mean transport and standard deviation integrated from the 2 coast to 1308Eis219 6 3.1 Sv (1 Sv [ 106 m3 s 1) in the thermocline and 23 6 12 Sv in the subthermocline. Subthermocline transport has an inverse linear relationship with the Niño-3.4 index and is the primary influence of total transport variability. Interannual anomalies during El Niño are greater than the annual cycle for sea surface salinity and thermocline depth. Water masses transported by the MC/MUC are identified by subsurface salinity extrema and are on isopycnals that have increased finescale salinity variance (spice variance) from eddy stirring. The MC/MUC spice variance is smaller in the thermocline and greater in the subthermocline when compared to the North Equatorial Current and its undercurrents.

1. Background and introduction Pacific and Indian Oceans has led to several initiatives in the last two decades to observe and model the MC/MUC The Mindanao Current (MC) is a low-latitude western system. Observations of the MC/MUC include sea level boundary current (LLWBC) in the western tropical gauges (Lukas 1988), hydrographic surveys (Hacker North Pacific, formed by the bifurcation of the North et al. 1989; Toole et al. 1990; Hu et al. 1991; Lukas et al. Equatorial Current (NEC) in the Philippine Sea (Fig. 1). 1991; Wijffels et al. 1995; Kashino et al. 1996; Qu et al. The MC flows equatorward along the coast of Mindanao 1998; Firing et al. 2005; Kashino et al. 2009, 2013), until splitting to contribute to the global overturning moorings (Kashino 2005; Zhang et al. 2014; Kashino circulation through the Indonesian Throughflow (ITF; et al. 2015; Hu et al. 2016), and Argo floats (Qiu et al. Gordon 1986) and to close the interior North Pacific 2015; Wang et al. 2015). However, MC/MUC water mass Sverdrup transport via the North Equatorial Counter transport, coastal structure, and variability are not well Current (NECC; Wyrtki 1961). The MC is the northern understood owing to strong currents, eddy activity, and source of subtropical water in the western Pacific warm large annual and interannual changes in surface forcing. pool (Fine et al. 1994; Lukas et al. 1996) and thus im- From hydrographic surveys, the MC current is narrow pacts the tropical heat and freshwater budgets and the at 108N, then widens, shoals, and increases in speed as it El Niño–Southern Oscillation (ENSO) phenomena flows equatorward until it separates into the ITF and (Gu and Philander 1997; Zhang et al. 1998). In the NECC between 78N and 68N(Lukas et al. 1991). Upper- subthermocline, the MC transports North Pacific In- ocean cyclonic recirculation in the Philippine Sea, such termediate Waters equatorward while a Mindanao as that associated with the quasi-permanent Mindanao Undercurrent (MUC; Hu et al. 1991) flows poleward. Eddy (ME), centered near 78N, 1308E, causes the MC The confluence of subtropical, tropical, and inter- transport to increase as it flows equatorward (Lukas mediate water masses of North and South Pacific origin et al. 1991). At 88N, the mean MC is in geostrophic and their role in salt and heat exchange of the tropical balance, extending approximately 200 km offshore, 2 with a subsurface velocity core of 0.8 to 1 m s 1 (Wijffels Corresponding author: Martha Schönau, [email protected] et al. 1995; Qu et al. 1998). Although the ME has been

DOI: 10.1175/JPO-D-16-0274.1 Ó 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses). Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2056 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 1. (a) Schematic of circulation in the tropical northwestern Pacific. Red is circulation from the surface to the bottom of the thermocline, typically 300-m depth. Main features are the NEC, MC, the Kuroshio, ITF, NECC, and ME. Subthermocline currents (blue) that oppose the direction of thermocline flow are the LUC, MUC, and NEUCs. (b) MDT from AVISO with contours at 0.015-m intervals. Glider observations are objectively mapped onto a mean line from 8.158N, 126.618E to 8.778N, 1308E (thick black line). The mean line is perpendicular to coastal bathymetry and MDT near the coast. observed by surface drifters (Lukas et al. 1991), its of the East Asian monsoon and the arrival of annual sporadic presence is difficult to observe in hydrographic Rossby waves from the central Pacific (Qiu and Lukas transects (Wijffels et al. 1995; Qu et al. 1998; Firing et al. 1996; Kim et al. 2004; Qiu and Chen 2010). Interannual 2005; Kashino et al. 2013). The velocity variability of the variability of the MC is forced by basinwide changes MC, its structure near the coast, and the variability of during ENSO. Numerical models suggest an increase in the ME remain inadequately observed. MC transport when the NEC bifurcates at higher latitude: In the subthermocline, the poleward MUC is roughly 50 annually in the fall and interannually during El Niño to 80 km offshore of the coast of Mindanao near 500-m (Qiu and Lukas 1996; Kim et al. 2004; Qiu and Chen 2 depth and potential density of 27.2 kg m 3 (Hu et al. 1991; 2010). Observations have both verified (Lukas 1988; Lukas et al. 1991; Qu et al. 1998), with a second velocity Kashino et al. 2009)andconflicted(Toole et al. 1990) core occasionally reported farther offshore (Hu et al. 1991; as to the relationship of transport with ENSO. Less is Qu et al. 1998). The MUC could be a quasi-permanent known about the transport variability of the MUC. recirculation or a poleward current transporting South Transport variability of the MC/MUC impacts the Pacific water masses; intermittent subthermocline eddies heat and salt balance in the tropical Pacific. Water that populate the region on intraseasonal time scales masses transported equatorward by the MC are the make previous observations inconclusive (Hu et al. 1991; North Pacific Tropical Water (NPTW), a subsurface Qu et al. 1998, 1999; Firing et al. 2005; Dutrieux 2009; salinity maximum in the thermocline, and the North Chiang and Qu 2013; Wang et al. 2014; Zhang et al. Pacific Intermediate Water (NPIW), a salinity minimum 2014; Chiang et al. 2015; Kashino et al. 2015; Schönau et al. in the subthermocline (Gordon 1986; Bingham and 2015; Hu et al. 2016). The forcing, persistence, and con- Lukas 1994; Fine et al. 1994; Kashino et al. 1996; nectivity of the MUC, and its relationship to the thermo- Schönau et al. 2015; Wang et al. 2015). Antarctic In- cline circulation are areas of ongoing research. termediate Water (AAIW), saltier than NPIW and of The transport of the MC/MUC forms a closed mass South Pacific origin, is also present in the Philippine Sea system with the NEC and Kuroshio Current (KC), and and may be carried northward by the MUC (Fine et al. their respective undercurrents, the North Equatorial 1994; Wijffels et al. 1995; Qu et al. 1999; Qu and Undercurrents (NEUCs) and the Luzon Undercurrent Lindstrom 2004; Schönau et al. 2015; Wang et al. 2015). (LUC) (Toole et al. 1990; Qiu and Lukas 1996; Qu et al. The volume transport and modification of these water 1998). The relative transports of the MC and KC depend masses as they transit the Philippine Sea are less well on the bifurcation latitude of the wind-driven NEC. The understood. Isopycnal stirring of large-scale salinity bifurcation latitude, dictated by Sverdrup theory as the gradients introduced by subsurface water masses creates zero line of zonally integrated wind stress curl, is affected finescale salinity variance that leads to isopycnal by local Ekman pumping and suction caused by shifts and diapycnal diffusion (Ferrari and Polzin 2005).

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2057

Quantifying the relationship between stirring, which from March to August (Fig. 2b). The ENSO phase, as increases isopycnal salinity variance, and diffusion, measured by the Niño-3.4 index, was positive (El Niño) which decreases it, will give insight into heat and salt from September 2009 to April 2010 and negative (La transfer in the ocean interior and improve parameteri- Niña) from June 2010 to April 2011 and August 2011 to zations of mixing in models. March 2012. Repeat glider observations of the MC/MUC provide An example section from the return of mission mean water mass transport and variability during a 4-yr 11A03601 is shown in Fig. 3. Salinity with potential period. Spray gliders made 16 sections of the MC/MUC temperature contours are plotted on depth and density from September 2009 to October 2013 under the Origins surfaces as a function of along-track distance from the of Kuroshio and Mindanao Current (OKMC) initiative nearest position to the coast at 9.38N, 126.318Eto (Rudnick et al. 2015b). Gliders observed temperature, roughly 1308E. On depth surfaces, internal wave motion salinity, and depth-averaged velocity from the surface to causes waviness of isotherms (Fig. 3a; Rudnick and Cole 1000-m depth at horizontal length scales from 6 to 2011). To filter out this high-frequency noise, salinity, 2000 km during all seasons and one ENSO cycle. Ob- potential density, and potential temperature are objec- jective maps of geostrophic velocity, salinity, and po- tively mapped on depth surfaces (Fig. 3b). Alternatively, tential temperature resolve the mean structure and to maintain the horizontal resolution of 6 km, in- variability of the MC, the MUC, and the ME. Argo terpolating data to density surfaces also filters out in- climatology is compared to glider observations to assess ternal wave motion (Fig. 3c). annual and interannual variability. The thermohaline To make a mean section, glider observations are lin- structure is compared between the MC/MUC and the early projected and objectively mapped onto a mean NEC (Schönau and Rudnick 2015) to infer water mass line extending from the coast of Mindanao, from 8.158N, connectivity and modification. Gliders provide new ob- 126.618E to 8.778N, 1308E(Figs. 1b, 2a), hereinafter re- servations on the stability of the MC thermocline ferred to as the mean line. Only observations within transport and the relationship between the variable 80 km of the mean line are included in the map. The MC/MUC subthermocline transport and ENSO. The objective map is made using a Gaussian autocovariance paper proceeds with a description of data and analysis (Bretherton et al. 1976) with a length scale of 80 km. methods, the mean and variability of water masses, This length scale is based on the autocovariance calcu- geostrophic velocity and transport, and finescale ther- lated from observations of temperature and salinity and mohaline structure. is identical to that used in Schönau and Rudnick (2015). Geostrophic velocity, calculated from thermal wind equations, is referenced to the depth-averaged velocity 2. Data collection and analysis of the glider (Rudnick et al. 2004; Todd et al. 2011; To observe the MC/MUC, the autonomous underwater Pelland et al. 2013; Lien et al. 2014; Pietri et al. 2014). glider Spray (Sherman et al. 2001; Rudnick et al. 2004) The absolute geostrophic velocity from gliders com- made repeat transects from Palau to the coast of Mind- pares favorably to ADCP measurements from a moor- anao. The Spray glider is equipped with a Sea-Bird CTD, ing (Lien et al. 2014), satellite sea surface height Seapoint fluorometer, and is remotely controlled with (Rudnick et al. 2015a), and to Argo climatology refer- Iridium satellite. The glider completes a dive from the enced to a level of no motion at 2000 m in the tropical surface to 1000 m and back over a horizontal distance of Pacific (Schönau and Rudnick 2015). 6kminacycletimeof6h.Dataarecollectedduringas- The mean line extends from the 1000-m isobath ap- cent at a vertical resolution of about 1 m and binned into proximately 380 km offshore and was chosen at an angle 10-m vertical bins. The depth-averaged velocity of the 808E of north to be perpendicular to isobaths and con- glider is obtained from dead reckoning (Davis et al. 2008; tours of mean dynamic topography (MDT; Fig. 1b; Todd et al. 2011; Davis et al. 2012). AVISO 2014). The mean depth-averaged velocity is For each mission the glider profiled westward from perpendicular to the mean line near the coast where the Palau (7.58N, 134.58E) until reaching the 1000-m isobath current is strongest (Fig. 2a), supporting the choice of along the coast of Mindanao at a latitude between 78 and the line’s direction. Because of the curvature of the 108N and then profiled on the return to Palau (Fig. 2a). coast, 80 km is the cutoff for inclusion of profiles in There were 10 missions, with 16 sections of the MC/ the objective map (Fig. 2a). The inshore coordinates of MUC completed within the given latitude range the mean line were chosen to maximize the number (Fig. 2b). The temporal coverage of the glider was suf- profiles (968 profiles). Argo climatology (1/68; Scripps ficient to observe the MC at all times of the year, with 10 Institution of Oceanography and International Argo sections from September to February and 6 sections Program 2009; Roemmich and Gilson 2009) indicates

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2058 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 2. (a) Trajectories of the 10 gliders deployed from Palau to the 1000-m isobath along the coast of Mindanao. Arrows show the mean depth-averaged velocity from the objective map of glider profiles within 80 km of the mean line (dashed box). There are 968 profiles included. (b) Glider missions are coded by date and glider serial number. The first two digits are the abbreviated year, the third digit is the hexadecimal month, and the fourth and fifth digits are the glider serial number. The dashed box denotes a 220-km range where observations from each mission are linearly projected and objectively mapped onto the mean line (solid black) to assess variability between missions. Most missions included observations north and south of the mean line. Mission 11A03601 (thick magenta) is featured as an example in Figs. 3 and 10. (c) Annual coverage of glider missions colored by the Niño-3.4 index. Observations are at all times of the year and ENSO phases. that there is alongshore shoaling of the thermocline of line. On four of the missions, the glider trajectory exceeded 20 m within this 80-km range but that it is less than the 220 km from the mean line, and only the nearer transect annual thermocline displacement (Scripps Institution of was included in the map. For these single transects, Oceanography and International Argo Program 2009; alongshore thermocline depth variability was less than Roemmich and Gilson 2009). There were roughly an 10 m. Since strong currents and meanders influence the equal number of profiles north and south of the mean glider path, excluding observations because of distance line to accommodate for the spatial variability. Re- from the mean line would exclude temporal variability. moving a linear alongshore gradient to account for The standard deviation between glider missions thus re- shoaling does not affect the results of the objective map. flects spatial and temporal variability. However, compari- The objectively mapped mean is thus a spatial and sons to Argo climatology in section 4 andtomooring temporal average. results in the conclusion indicate that temporal variance The variability of the MC/MUC is assessed from the exceeds spatial variability within the region of standard deviation between the 10 glider missions observations. (Fig. 2b), each objectively mapped onto the mean line. An example of geostrophic velocity and salinity ob- Linearly projecting and mapping each glider mission to- jectively mapped on the mean line is shown for mission gether averages out the linear alongshore gradient, as 11A03601 in Fig. 3d. The MC flows equatorward near gliders tended to traverse north then south of the mean the coast (blue), and there are two subthermocline

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2059

FIG. 3. Example section from glider mission 11A03601 during its return from near the coast of Mindanao to Palau (see Fig. 2a). Along- track salinity (color) and temperature (contours) from 9.38N, 126.318E to 8.228N, 1308E: (a) vertically binned on depth surfaces, (b) objectively mapped onto depth surfaces, and (c) interpolated to potential density surfaces. Tick marks in (a) are at the location of each profile. (d) Geostrophic velocity referenced to the depth-averaged velocity (color) and salinity (contours) from an objective map of 2 mission 11A03601 onto the mean line. Dark gray lines denote 26 and 27 kg m 3, respectively. The range is extended to 9.158N, 1348Eto show the relatively weak velocities east of 1308E, denoted by the dashed line. Negative (positive) velocity is equatorward (poleward). (e) Objectively mapped depth-averaged velocity for mission 11A03601. poleward (red) cores offshore that may be expressions each deployment, the transform is performed on the of the MUC. The range is extended to 1348E to show the along-track isopycnal salinity that is detrended and small velocities east of the mean line, although these interpolated to a 1-km grid. Performing the wavelet may vary by section. The objectively mapped glider transform on the entire mission avoids cutting off low- missions are used to make composite potential wavenumber variance near the coast of Mindanao where temperature–salinity (T–S) diagrams to assess water the glider turns around. The resulting salinity variance is mass transport, and they are compared to monthly (18) projected for each of the 16 sections and averaged in 10-km and annual (1/28) Argo climatology (Scripps Institution bins along the mean line. The mean salinity variance is of Oceanography and International Argo Program an average over all transects. To separate length scales, 2009; Roemmich and Gilson 2009) and forcing from different ranges of wavelengths are included in the wind and precipitation to assess annual and interannual wavelet transform for finescales (10 to 80 km) and large variability. Monthly 10-m wind products are available scales (120 to 200 km). The finescale scale ranges from from NCEP–NCAR reanalysis (Kalnay et al. 1996; the nearest spacing of glider observations to the decor- NCEP 1996) and monthly precipitation data are avail- relation length scale, and the large scale begins at the e- able from CPC Merged Analysis of Precipitation folding envelope of the chosen Morlet wavelet. (CMAP; CPC 1997; Xie and Arkin 1997), each gridded at 2.58 and available from the Earth System Research Laboratory (ESRL). 3. Mean structure Isopycnal salinity variance, obtained using a wavelet a. Water masses transform, is used to examine thermohaline structure of the MC/MUC at different length scales and can be used Water masses are identified by their salinity extrema. to trace water masses (Todd et al. 2012; Schönau and The mean salinity has two haloclines (Fig. 4a): first, from Rudnick 2015). Following Todd et al. (2012), we use a the surface fresh layer that overlies the mean section to a Morlet wavelet c(x) 5 ei2pkxe2x2/2, with k 5 1. For subducted subtropical salinity maximum and second, in

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2060 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 4. Mean structure of the MC/MUC from an objective map of glider proﬁles within 80 km of the mean line: (a) salinity with potential temperature contours, (b) geostrophic velocity referenced to the depth-averaged velocity on depth surfaces with potential density contours (gray), and (c) geostrophic velocity referenced to the depth-averaged velocity interpolated to potential density surfaces with salinity contours. In (b) and (c), equatorward velocity is negative (blue), and poleward velocity is positive (red). Subsurface salinity extrema are nearest to the coast and advected equatorward in the MC.

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2061

FIG. 5. Potential T–S diagrams colored by (a) mean longitude and (b) net transport over the 10 glider missions, objectively mapped onto the mean line and binned by 0.28C and 0.01 psu. Water masses are the NPTSW, NPTW, NPIW, and AAIW. In (b) transport is summed in each bin then normalized by the bin size and number of sections. Equatorward transport is negative (blue) and poleward transport is positive (red). Water masses that had small net transport from recirculation or were measured infrequently appear as white. The MC has a net equatorward transport of NPTW and NPIW, and the MUC has a net poleward transport of water typical of AAIW. the transition to fresh intermediate water below the commonly identified by its higher oxygen (Qu and thermocline. The subsurface salinity extrema are Lindstrom 2004). The presence of AAIW and its path- greatest near the coast as displayed in a T–S diagram way into the Philippine Sea has been recorded by several (Fig. 5a). The surface fresh layer that overlies the section observational studies (Fine et al. 1994; Wijffels et al. is North Pacific Tropical Surface Water (NPTSW; 1995; Qu et al. 1999; Qu and Lindstrom 2004; Zenk et al. ,34.1 psu and .288C; Wyrtki and Kilonsky 1984), 2005), and a similar T–S curve for AAIW is observed formed by local precipitation and advection (Delcroix here. Lower oxygen intermediate waters that have and Hénin 1991). NPTSW is influenced by annual vari- mixed in the tropics, such as the North Pacific Tropical ability of the East Asian monsoon and interannual Intermediate Water (NPTIW), are found east of 1308E variability associated with ENSO (Delcroix and Hénin (Bingham and Lukas 1995). The poleward transport 1991; Li et al. 2013). NPTSW creates a large vertical of AAIW has been used to verify some connectivity salinity gradient within the upper 50 m that strongly between the MUC and southern North Equatorial impacts the western Pacific warm/fresh pool (Cronin Undercurrent (Schönau et al. 2015; Wang et al. 2015). and McPhaden 1998; Delcroix and McPhaden 2002). b. Geostrophic velocity Beneath NPTSW is the subsurface salinity maximum of NPTW (.34.95 psu) near the coast and centered at po- The mean absolute geostrophic velocity of the MC, 2 tential density 23.5 kg m 3. NPTW is formed in the referenced to the depth-averaged velocity, is equator- 2 subtropics due to excess evaporation (Tsuchiya 1968; ward, with a subsurface velocity maximum of 21.0 m s 1 2 Katsura et al. 2013) and advected westward in the NEC at 100-m depth (22 kg m 3) and 50 km offshore (1278E; and into the MC, freshening as it mixes with surface and Figs. 4b,c). Near the coast, where the MC is the stron- intermediate waters along the path of flow (Lukas et al. gest, there is advection of NPTW and NPIW (Fig. 4c). 1991; Fine et al. 1994; Qu et al. 1999; Li and Wang 2012; The MC decreases in strength with increasing distance Schönau and Rudnick 2015). In the subthermocline, offshore and intersects at the surface with what could the subsurface salinity minimum of NPIW (,34.4 psu), be a cyclonic ME roughly 250 km offshore (128.758E). formed in the Okhotsk Sea (Talley 1993), is at potential The core of the possible ME extends through the 2 density of 26.55 kg m 3 in the MC (Bingham and Lukas thermocline to 250-m depth, with similar poleward and 1994). Like NPTW, the extrema of NPIW is near the equatorward velocities and symmetric isohalines coast; however, its source may be either from the NEC (Fig. 4c). The mean depth-averaged velocity is cyclonic or more directly from the equatorward LUC beneath over this longitude range (Fig. 2a). Mean hydrographic the KC (Bingham and Lukas 1994; Fine et al. 1994; Qu sections from Wijffels et al. (1995) and Qu et al. (1998) et al. 1997; Schönau et al. 2015). AAIW, with potential have similar velocity structures as the glider-observed 2 density of 27.2 kg m 3 and salinity greater than 34.5 psu, mean MC, although neither of these studies found evi- is denser and saltier than NPIW (Fig. 5a) and is dence of a mean ME. The width of the possible ME is

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2062 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 6. (a) Total transport (black) summed along the mean line and separated into thermocline (blue; surface to 2 2 26 kg m 3) and subthermocline (red; 26 to 27.3 kg m 3) transports. Shading is the standard deviation between the 10 missions. (b) Transport per distance for the thermocline (blue), subthermocline (red), and total (black) as a function of longitude. (c) Transport per density along the mean line. For (b) and (c), shading is the standard error, and the legend gives the mean and standard deviation of each layer.

2 roughly 250 km, consistent with that found by surface to 26 kg m 3)of219 Sv and subthermocline transport 2 2 drifters at this location (Lukas et al. 1991). (26to27.3kgm 3)of23 Sv. Potential density 27.3 kg m 3 In the subthermocline, the equatorward MC extends is the densest isopycnal that is present in all sections. Po- 2 to 1000 m near the coast, decreasing in velocity with tential density 26 kg m 3, at the base of the thermocline, increasing depth and distance offshore (Fig. 4b). Off- divides the thermocline and subthermocline transport, shore of the equatorward subthermocline MC is a which encompasses the intermediate water and the MUC poleward MUC. The MUC has a maximum velocity of (Figs. 4b,c). This isopycnal has been used as a reasonable 2 0.17 m s 1 located 75 km offshore (127.758E) and 650-m separation of thermocline and subthermocline transports 2 depth (27.2 kg m 3). for the NEC (Schönau and Rudnick 2015) and NECC East of the MUC there is weak equatorward ﬂow until (Johnson et al. 2002; Hsin and Qiu 2012). The Sverdrup 128.58E; then mean poleward ﬂow exists through the transport from the zonally averaged wind stress curl offshore edge of the mean line. The coastal structure of was 228.3 Sv during this time period, and hydrographic the equatorward MC and mean location and width of the estimates of the mean transport from the surface to 1000 m MUC referenced to an absolute velocity are a funda- and coast to 1308E are around 227 Sv (Wijffels et al. 1995; mental improvement to previous observations, which Qu et al. 1998). These estimates exceed the glider trans- have either been time series at single locations or ref- port as they do not consider or resolve the poleward un- erenced to arbitrary levels of no motion. The mean dercurrent. To compare, glider transport summed from MC/MUC compares favorably with numerical results the coast approaches 230 Sv before decreasing offshore from Estimating the Circulation and Climate of the (Fig. 6a). The sum of equatorward transport from the coast 2 Ocean (ECCO) and the Parallel Ocean Program (POP) to 128.58E and surface to 27.3 kg m 3 is 230.4 Sv. In the simulation (Schönau et al. 2015). subthermocline, the poleward transport of the inshore MUC (127.258E) is 3.6 Sv, and the poleward sub- c. Transport thermocline transport across the mean line is 8.9 Sv. The transport of the MC/MUC across the mean As a function of longitude, mean transport is greatest 2 line and from the surface to 27.3 kg m 3 is 222.4 Sv in the thermocline at the location of the MC core and 2 (1 Sv 5 106 m3 s 1), with a thermocline transport (surface then decreases to zero at 128.58E(Fig. 6b). The possible

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2063

ME, centered at 129.258E, has a cyclonic circulation of maximum (Fig. 7a), making heaving of isopycnals the roughly 2 Sv, with 0.5 Sv greater transport poleward than likely cause. Below the temperature mixed layer, equatorward. The subthermocline transport has an which extends to 80 m, isotherms tend to follow iso- equatorward maximum near the coast, consistent with pycnals, causing isopycnal displacement to appear as the subthermocline equatorward MC, and then becomes temperature variability. The thermocline depth (mea- 2 poleward between 1278 and 127.758E at the mean loca- sured as the depth of the layer 23–24 kg m 3)isata tion of the MUC. As a function of density, mean trans- minimum in March 2010, coinciding with El Niño, and port is greatest at the surface between potential density at a maximum in 2013 (Fig. 8a). Here, a positive 2 21 and 22 kg m 3 and uniform through the thermocline. anomaly is the shoaling of the thermocline, whereas a 2 Deeper than 27 kg m 3, the large standard error in- negative anomaly is the deepening of the thermocline. dicates that the direction of mean net transport is in- Annual changes in thermocline depth are forced by the conclusive (Fig. 6c). local wind and by Rossby waves generated by wind over Binning the volume transport of each objectively the central Pacific (Kim et al. 2004). During the winter, mapped mission by potential temperature and salinity, strong northeasterlies over the region create positive wind summing over all sections, and normalizing by bin size stress curl and Ekman suction that shoals the thermocline, and the number of sections creates a T–S diagram of leading to a cold layer at 100-m depth known as the water mass transport (Fig. 5b). The T–S transport in- Mindanao Dome (MD) that extends east to around 1508E tegrates to the average transport over the 10 missions between the MC, NEC, and NECC (Masumoto and from the surface to 1000-m depth. Both NPTW and Yamagata 1991). The MD decays in May as the East Asian NPIW have equatorward transport. In the sub- monsoon shifts to southwesterlies and the annual down- thermocline, there is a net poleward transport of water welling Rossby wave arrives from the central Pacific 2 with potential density 27 to 27.5 kg m 3 and salinity (Tozuka et al. 2002). Annual thermocline depth variability, greater than 34.5 psu, typical of AAIW. The poleward taken as the depth of the 218CisothermfromArgocli- 2 subthermocline transport shallower than 27.3 kg m 3 is matology at 8.58N is consistent with this forcing (Fig. 8b). 5 Sv, earning the MUC the designation as a current, as At 128.58E, the minimum depth is from January to April there is a mean net transport of a distinct water mass. and the maximum is from May to November. The annual Rossby wave arrives at this longitude in late summer. To assess annual and interannual forcing, glider ob- 4. Variability servations can be compared to Argo climatology and a Variability in the MC/MUC is forced by basinwide linear, 1.5-layer linear model that allows the propaga- wind stress, transient eddies, Rossby waves, annual tion of Rossby waves (Kessler 2006; Qiu and Chen changes in the East Asian monsoon, and interannual 2010). Following Kessler (2006), the depth anomaly of changes associated with ENSO. The forcing creates the thermocline, h (positive upward), is governed by water mass and geostrophic velocity variability between ›h ›h t glider missions. Variability is greatest in thermocline 1 c 1 Rh 5 = 3 . ›t r ›x f r depth, surface salinity, and subthermocline transport. The following sections assess the magnitude of vari- Here, R is a damping coefficient time scale of roughly ability and the responsible forcing. 9 months, f is the Coriolis parameter, t the wind stress, and r the density of seawater. The long Rossby wave a. Thermocline depth 2 speed is cr 52bc /f, where b is the meridional de- The standard deviation in potential temperature is rivative of f, and c is the internal gravity wave speed, 2 greatest in the thermocline, between depths of 100 to which is taken to be ;2.6–3 m s 1 across the Pacific basin 200 m, where the vertical temperature gradient is also a (Chelton et al. 1998). The solution is given by

ð x t 0 2 2 0 1 2(R/c )(x2x0) fx ,[t (x x )/cr]g 0 2(R/c )(x2x0) 0 h(x, t) 5 e r = 3 dx 1 e r h [t 2 (x 2 x )/c ], c f r E r r xE

where hE is the depth anomaly on the eastern boundary, in the central and eastern Paciﬁc, and Rossby wave ra- and h is the depth anomaly on the western boundary. diation from the eastern boundary from coastally trap- The thermocline depth thus depends on local wind, the ped Kelvin waves; however, the dominant term is the propagation of Rossby waves forced by wind stress curl propagation of Rossby waves generated in the central

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2064 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 7. Standard deviation of the objectively mapped glider missions (color) with mean contours (black) for each (a) potential temperature, (b) salinity, and (c) geostrophic velocity. 2 2 Potential density contours (white) are from (a) 22–27 kg m 3 at 1 kg m 3 increments and at 2 (c) 26 kg m 3. The standard deviation includes both temporal and spatial variability over the range of the glider missions in Fig. 2b.

Paciﬁc (Qiu and Chen 2010). At 128.58E, the model shows Argo climatology have the same mean thermocline depth variance consistent with annual forcing: deep in depth (Fig. 8c). Nearer to the coast, the glider better summer and fall and shallow during winter (Fig. 8c). resolves the coastal structure and has a deeper thermo- The model, Argo climatology, and glider results are cline than the climatology. The model and Argo clima- compared at 8.58N and 128.58E, where the glider and tology have roughly the same magnitude but differ in

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2065

FIG. 8. Observations and model of thermocline depth variability. For each, the depth anomaly is positive for 2 shoaling and negative for deepening. (a) Depth anomaly of potential density layer 23–24 kg m 3 for each glider mission. Shading on date axis indicates observations during El Niño (red) and La Niña (blue) according to the Niño- 3.4 index. (b) Annual cycle of 218C isotherm depth from Argo climatology (Roemmich and Gilson 2009) along 8.58N. (c) Depth anomaly of the 218C isotherm for each a linear, 1.5-layer, wind-forced model that allows Rossby wave propagation, monthly Argo climatology, and glider observations, all at 8.58N, 128.58E. The mean from June 2009 to October 2013 has been removed from each but not an annual cycle. The linear model and Argo climatology were smoothed with a 3-month running mean. The correlation coefﬁcient is 0.6 between the glider and Argo climatology, 0.59 between the glider and model, and 0.52 between the model and Argo climatology. phase. The model has a stronger annual cycle than Argo coincides with the low vertical salinity gradient between climatology (Fig. 8c) but weaker interannual variability. the two haloclines. The glider has the greatest extremes in shoaling and The surface salinity is saltiest during the 2010/11 El deepening as is expected of synoptic measurements Niño and freshest at the end of the 2010/11 La Niña, with when compared to gridded climatology. Both Argo cli- low variability in intermittent years (Fig. 9a). matology and gliders observed shoaling of the thermo- The annual cycle in surface salinity is less than cline during the El Niño in 2009/10 and deepening in 0.1 psu (Argo climatology), fresh from August to 2013. The shoaling in 2009/10 is consistent with forcing February and salty from July to August (Fig. 9b). As during El Niño, but the deepening of the thermocline in evaporation tends to be weaker than precipitation in 2013 is of unknown origin. this region (Cronin and McPhaden 1998), the surface salinity is compared to the regional precipitation b. Salinity (Fig. 9b). A positive (negative) salinity anomaly lags The standard deviation in salinity is greatest in the anomalously low (high) regional precipitation, consis- upper 100 m (.0.2 psu), with a secondary variance tent with the annual migration of the ITCZ. maximum in the subsurface halocline between the The relatively small magnitude of the annual cycle NPTW and NPIW (Fig. 7b). Where the vertical salinity suggests that glider-observed extremes are related to gradient is large, isopycnal heaving is sufﬁcient to ac- interannual precipitation forcing. Removing the annual count for the observed salinity variability, as calculated cycle, interannual anomalies of precipitation and sur- from the depth variance of isopycnals and their mean face salinity are each large following El Niño and La salinities. The vertical minimum in salinity variance Niña. The precipitation anomaly lags the Niño-3.4

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2066 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 9. Sea surface salinity and precipitation variability. (a) Salinity anomaly (0–50 m) from each glider mission. (b) Annual anomaly of precipitation and surface salinity (0–50 m) from Argo climatology (Roemmich and Gilson 2009), each with a 3-month running mean and averaged from 2004 to 2014. The Argo anomaly is at 8.58N, 127.58E, and precipitation is averaged over the region 3.758–13.758N and 126.258–136.258E, encompassing rainfall over the NEC, MC, and local recirculation from the NECC. (c) Niño-3.4 index and interannual anomalies of precipitation and glider surface salinity (0–50 m, 127.58E), where the glider mean and annual Argo anomaly are removed from glider observations. Precipitation is averaged over the same region as in (b) and ﬁltered with a 5-month running mean. The maximum, lagged correlation coefﬁcient between the Niño-3.4 index and the salinity anomaly is 0.8 and that between the Niño-3.4 index and precipitation anomaly is 20.75, with the Niño-3.4 index leading the salinity anomaly and precipitation by 1 month. The precipitation anomaly and salinity anomaly have a maximum correlation of 20.65 with a zero time lag. Results depend on the area over which precipitation is averaged. index: negative following the 2009/10 El Niño and pos- precipitation is small, and positive wind stress curl forces itive following the 2010/11 La Niña(Fig. 9c). The in- upwelling. In the summer, the ITCZ creates large pre- terannual anomaly from the glider (where the annual cipitation over the region, wind stress curl changes, and cycle from Argo climatology has been removed) is cor- the annual Rossby wave arrives to force downwelling respondingly positive during lack of rainfall and nega- (Fig. 8b). Interannually, low precipitation and upwelling tive with excessive rainfall. The magnitude of the each have extremes during the El Niño in 2009/10 interannual surface salinity anomaly is roughly 5 times (Figs. 8c, 9c). Thus, salty anomalies from ‘‘outcropping’’ greater than that of the annual anomaly. During ENSO of denser, saltier water occur when a fresh layer is ad- neutral conditions, when the Niño-3.4 index indicates vected away or from upwelling and mixing with un- neither an El Niño nor La Niña state, interannual derlying isopycnals. The relationship between surface anomalies of precipitation and surface salinity are small salinity, upwelling, and advection thus requires a more by comparison. complete salinity budget than presented here. Although these results assess the phase of surface c. Geostrophic velocity and transport salinity anomalies to precipitation, they are not entirely independent of mixing and advection caused by wind The MC/MUC geostrophic velocity and transport and wind stress curl. Annually, precipitation is in phase variability are assessed by the standard deviation of with wind forcing (Masumoto and Yamagata 1991). In geostrophic velocity (Fig. 7c) and transport as a function winter and early spring the ITCZ is south of the region, of longitude (Fig. 10) and time (Fig. 11). The MC core

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2067

23 23 FIG. 10. Transport per distance for each glider mission summed from (a) the surface to 26 kg m and (b) 26 to 27.3 kg m , where 2 26 kg m 3 separates the thermocline and subthermocline transports. Note the magnitude of the color bars. Black bars in (b) denote glider mission 11A03601, used in (c)–(e). (c) T–S transport diagram summed along the mean line from the coast to 1288E to encompass the equatorward MC and first poleward MUC core; (d) as in (c), but from the coast to 1298E to encompass the equatorward MC, first poleward MUC core, and offshore equatorward flow; and (e) as in (c), but from the coast to 1308E. See Fig. 3 for a geostrophic cross section and depth-averaged velocity of mission 11A03601. See Fig. 5b for bin size and normalization of (c)–(e). velocity and transport are stable in the thermocline, typical of changes in location and breadth. The velocity whereas the subthermocline has a large transport range. variability in the subthermocline east of the MUC ex- Interannual transport variability is more pronounced ceeds the mean. than annual variability, and a relationship between the Velocity is integrated over density surfaces to yield subthermocline transport and ENSO emerges. thermocline and subthermocline transport for each The standard deviation of geostrophic velocity is glider mission (Fig. 10). In the thermocline, transport is large where the mean current velocity is strongest, greatest near the coast, tapering to zero between 127.58 approaching 50% of the mean velocity near the core of and 129.58E, and becomes poleward in roughly half the the MC and also large in the subthermocline by the coast sections (Fig. 10a). The poleward transport may be an (Fig. 7c). However, the standard deviation is less than expression of the ME as it is at the same location of large the mean equatorward MC in both the thermocline and variability near the ME core (Fig. 7c). The transport in subthermocline. Near the core of the ME (129.258E), the the subthermocline is variable: persistently equatorward variability is roughly equal to the mean, suggesting the near the coast with poleward transport in all sections ME was intermittently observed. There is a vertical (Fig. 10b). Roughly half the sections have multiple minimum in standard deviation near potential density poleward cores that previously have been reported as 2 26 kg m 3 that separates the thermocline and sub- double MUC cores (Hu et al. 1991). However, it is dif- thermocline transport, occurring even near the coast ficult to define multiple cores of the MUC, as it is unclear where the current is strong. In the subthermocline, the if each has net water mass transport or recirculates. For standard deviation is large on either side of the MUC, example, mission 11A03601 has symmetric isohalines in

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2068 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 11. Transport through the mean line for each glider mission from the surface to 2 27.3 kg m 3 (black) and divided into thermocline (blue) and subthermocline (red) transport by 2 potential density 26 kg m 3. Transports for each are shown as a function of (a) time, (b) day of year, and (c) Niño-3.4 index. The thermocline transport is relatively constant compared to the subthermocline transport, which has an inverse linear relationship with the Niño-3.4 index. the subthermocline between poleward and equatorward as conﬁrmed by cyclonic depth-averaged velocity cen- velocities, centered at 1298E, suggesting water mass re- tered at 1298E(Fig. 3e). Such recirculation, typical of circulation (Fig. 3d). T–S transport diagrams are used to eddies, occurs during other glider missions. However, all assess the net transport of water masses (Figs. 10c,d,e). glider missions had a net poleward subthermocline Subthermocline transport summed from the coast to transport of intermediate water that is saltier than NPIW 1288E has equatorward transport of NPIW and a pole- and typical of that found in the mean (Fig. 5b). The MUC ward transport of water typical of AAIW (.34.5 psu, is thus a persistent current by its net poleward transport 2 27.2 kg m 3; Fig. 10c). Summing transport from the coast of a distinct water mass even if at times it meanders, to 1298E(Fig. 10d) now includes saltier equatorward partially recirculates, or interacts with eddies. 2 subthermocline transport (;27 kg m 3, .34.5 psu). Integrating the velocity over the mean line for each However, this T–S range has net poleward transport when glider mission and for each of the thermocline and sub- integrating to 1308E(Fig. 10e). Thus, the subthermocline thermocline layers (Fig. 11) provides a transport time equatorward ﬂow between 1288 and 1298E(Fig. 10b)is series. The relative stability of the thermocline transport likely a partial cyclonic circulation from farther offshore, and variability of the subthermocline transport is

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2069

FIG. 12. Composite T–S transport diagrams for glider observations during ENSO (a) neutral, (b) positive, and (c) negative phases. The greatest transport difference between El Niño and La Niña states was in the subthermocline, where equatorward (poleward) transport dominated during El Niño (La Niña). See Fig. 5b for bin size and normalization. apparent in their ranges: about 10 Sv (224.0 to 213.6 Sv) The depth penetration of the equatorward MC ap- for the thermocline and 40 Sv (223.3 to 19.6 Sv) for the pears to be the source of the subthermocline transport subthermocline (4 times as large). The total transport difference between El Niño and La Niña. Both di- ranges from 2 to 237 Sv, fluctuating with the sub- rections of subthermocline flow were observed during thermocline (Fig. 11a). This is in accordance with pre- each event (Fig. 10b), but net equatorward transport was vious ranges of transport from hydrographic sections of 8 24 Sv greater on average during El Niño than during La to 240 Sv (Toole et al. 1990; Lukas et al. 1991). Unlike in Niña, compared to a 9-Sv difference in poleward trans- the NEC, where thermocline and subthermocline transport. Composite T–S transport diagrams during each port were highly correlated (Schönau and Rudnick 2015), ENSO phase show the distribution of water mass there does not appear to be a coherent relationship be- transport (Fig. 12). During neutral conditions (Fig. 12a) tween these transports in the MC/MUC. transport is similar to the mean (Fig. 5b). El Niño The MC/MUC transport may have annual and in- (Fig. 12b) and La Niña(Fig. 12c) have similar thermo- terannual variability (Lukas 1988; Toole et al. 1990; Qiu cline transports and notably different surface and sub- and Lukas 1996; Qu et al. 2008; Kashino et al. 2009). thermocline transports. During El Niño there is a lack of Plotting transport by day of year (Fig. 11b), the annual fresh NPTSW and greater equatorward than poleward cycle is not discernable. The observed range of transport transport in the subthermocline. Although there was is larger than a model-estimated annual transport range poleward subthermocline velocity during El Niño (June of 10 Sv (Qiu and Lukas 1996). Thus, the annual cycle and December 2009; Fig. 10b), the lack of net poleward may be masked by variability at other time scales. transport (Fig. 12b) suggests that there is recirculation Plotting by the Niño-3.4 index, an inverse linear re- or that poleward transport moved offshore of the mean lationship exists between the Niño-3.4 index and sub- line. During La Niña there was weak equatorward thermocline transport (Fig. 11c). Transport in the flow of NPIW and large poleward transport of saltier subthermocline is strongly poleward during La Niña and intermediate water across the mean line. The sub- equatorward during El Niño. During ENSO neutral thermocline transport was thus negative during El Niño conditions the direction of subthermocline transport because of net equatorward transport of NPIW and pos- tends to be equatorward and less than 8 Sv. The corre- itive during La Niña because of net poleward transport of lation coefficient between the Niño-3.4 index and the water typical of AAIW, a significant insight into the in- subthermocline transport is 20.92 and that between terannual transport variability of the MC/MUC system. the Niño-3.4 index and the total transport is 20.87. Assuming each glider mission is an independent degree 5. Isopycnal salinity variance of freedom, the correlation coefficient is significant within a 1% confidence interval. It should be noted The thermohaline structure of the MC/MUC at that only one ENSO cycle was observed, and these ob- finescales can be related to large-scale gradients by servations only describe circulation near the coast. comparing salinity variance on isopycnals, separated by However, during this time period subthermocline fluc- length scale with a wavelet transform. At finescales tuations were correlated with the Niño-3.4 index and (10 , 80 km; Fig. 13a) salinity variance per kilometer is 2 were the leading cause of total transport variability. large from the surface to 25 kg m 3, encompassing the

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2070 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

FIG. 13. (a) Finescale (10 , 80 km) and (b) large-scale (120 , 200 km) isopycnal salinity variance from the wavelet transform projected, binned, and averaged over glider missions. Black contours are mean salinity. Salinity variance is large in the subtropical and intermediate water masses, separated by a salinity variance minimum. (c) Average over the mean line for each the ﬁnescale and large-scale salinity variance and their ratio.

NPTW, a minimum at the base of the thermocline from (Fig. 14a). Similarly, the salinity minimum of the NPIW 2 25.2 to 25.5 kg m 3, and large in intermediate water becomes saltier and shifts to a denser isopycnal. The 2 deeper than 26 kg m 3. The large-scale salinity variance downward flux of salt and density between the warm, (120 , 200 km) has a similar vertical structure (Fig. 13b). salty NPTW and cool, fresh NPIW is typical of double At all scales, salinity variance is greatest near the coast, diffusion (Schmitt 1981). However, the well-mixed layer where the subsurface salinity extremes of the NPTW between the two (shaded) has a density ratio such that and the NPIW create large-scale horizontal salinity turbulence caused by internal wave shear is likely a gradients. dominant mechanism of diapycnal mixing. The ratio of finescale to large-scale salinity variance, a Changes in finescale salinity variance can be used to type of horizontal Cox number (Osborn and Cox 1972), track diffusion of water masses. Finescale salinity vari- is relatively constant through the water column, with the ance decreases along the isopycnals of NPTW between absolute value dependent on the range of wavelengths the NEC and MC (Fig. 14b). In the subthermocline, included in the transform (Fig. 13c). A vertically con- salinity variance is greater in the MC/MUC than in the stant horizontal Cox number was also observed across NEC/NEUCs. This corresponds to finescale salinity the NEC (Schönau and Rudnick 2015). Both results gradients that are introduced by the stirring of the fresh confirm a direct relationship between the finescale and NPIW with saltier intermediate water. 2 large-scale thermohaline gradients, where the stirring of A salinity variance minimum near 25.5 kg m 3 between large-scale gradients causes finescale salinity variance. subtropical and intermediate water masses is a notable feature in both the NEC and MC (Fig. 14b). The salinity variance minimum decreases along the path of flow, 6. Water mass modification 2 similar to NPTW, but shifts from 25.6 to 25.4 kg m 3, Water masses are modified as the NEC advects them opposite the direction of flux by the salinity extrema, into the MC and the MUC advects them into the which shifts to denser isopycnals (Fig. 14a). It is unclear if southern NEUC (Schönau et al. 2015; Wang et al. 2015). this is a remnant of sampling or would apply in other Changes in salinity are caused by horizontal and dia- regions. The cause of the variance minimum is unknown. pycnal diffusion. Salinity decreases and density in- It is within an isopycnal range that has a low horizontal creases for the subsurface salinity maximum of NPTW salinity gradient across much of the North Pacific, as

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2071

FIG. 14. (a) Mean salinity averaged on isopycnal surfaces for the MC (Fig. 4) and the mean NEC (Schönau and Rudnick 2015). Salinity extremes of NPTW and NPIW are smaller in the MC than in the NEC. Gray shaded area indicates a density ratio 5 , Rr , 10, where Rr is defined as auz/bSz, and a and b are the thermohaline expansion and saline contraction coefficients, respectively. (b) Finescale (10 , 80 km) isopycnal salinity variance for the MC and NEC. Variance is smaller (larger) for the thermocline (subthermocline) transport in the MC than the NEC. For both (a) and (b), the average for the MC is across the mean line and that for the NEC is from 98 to 158N along 134.38E. observed by Argo climatology, and where interannual influences the total transport variability. During the climatic variability at the surface has been observed to ENSO cycle observed, subthermocline transport along cause temporal spice variance (Yeager and Large 2007; the mean line had an inverse linear relationship with the Sasaki et al. 2010). A step in salinity of 0.08 psu was ob- Niño-3.4 index. There was no evident relationship be- served in this isopycnal range in 2012 but was likely ob- tween the thermocline and subthermocline transport and servable only because of the small spatial variance along no discernable annual cycle for total transport, although it this isopycnal. The persistence of the variance minimum, may have been dwarfed by strong interannual variability. and its shift to a less dense isopycnal, would make it an The subsurface salinity extrema of NPTW and NPIW interesting layer to study mixing. decrease in magnitude between the NEC and MC along the path of flow, as does the salinity variance in the thermocline. The salinity variance of subthermocline 7. Conclusions and discussion water in the MC/MUC is greater than that in the NEC/ Repeat glider observations resolve the mean ther- NEUCs, as fresh NPIW is stirred with saltier in- mohaline structure and transport of the MC/MUC and termediate water in the tropical Pacific. The relationship their variability during a 4-yr period. The results provide between isopycnal salinity at large and small scales is new insight into the structure of the currents near the consistent with stirring and diffusion of salinity extrema. coast, water mass transport, variability, and finescale Glider-observed mean velocity and standard de- structure. The MC is a strong current with a persistent viation compare favorably to observations by a single transport of subtropical water masses, extending to a mooring with an ADCP, located at 88N, 127.38E from depth of 1000 m near the coast and decreasing in velocity 2010 to 2012 (Zhang et al. 2014). The mooring was off- with distance offshore. In the subthermocline, the MUC shore of the MC core and had a mean velocity and 2 is a persistent poleward flow that meanders offshore of standard deviation of 0.7 6 0.2 m s 1 at 100-m depth, the MC, with mean transport of a distinct water mass. nearly identical to that in the glider (Figs. 4b, 7c). The The range of subthermocline transport is 4 times inclusion of spatial variability between glider missions greater than that of the thermocline and strongly thus had minimal impact on the mean velocity and

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2072 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47 velocity variance. The gliders passed the mooring only tropics. These observations are useful in a western a few times, making a direct velocity comparison un- boundary current where deep, strong currents near the reasonable; however, velocity comparisons between coast pose a challenge to moorings, floats, and satellite gliders and a mooring array near the Luzon Strait show altimetry. Combining such glider observations with good agreement between glider-derived absolute geo- Argo and satellite observations, and incorporating these strophic velocity and mooring ADCP (Lien et al. 2014). observations into numerical models, would be a pow- Single moorings in the MC/MUC have observed erful tool to examine large-scale water mass transport intraseasonal velocity variability on time scales of 50– and heat budgets in the tropical Pacific. 80 days (Kashino 2005; Zhang et al. 2014; Kashino et al. 2015). In the subthermocline, mooring observations and Acknowledgments. Glider observations and analysis model results suggest the MUC velocity varies on even were funded by the Office of Naval Research as part of shorter time scales and is also affected by pressure gra- the Origins of Kuroshio and Mindanao Current dients from Rossby waves and ENSO (Hu et al. 2016). (OKMC) project through Grants N00014-10-1-0273 and Subthermocline eddies that have been observed and N00014-11-1-0429, and through Flow Encountering modeled in the region at these time scales (Firing et al. Abrupt Topography (FLEAT) project Grant N0014- 2005; Dutrieux 2009; Chiang and Qu 2013; Wang et al. 15-1-2488. Glider observations are available upon re- 2014; Chiang et al. 2015) were observed by the glider quest ([email protected]). We thank the Instrument (Fig. 10). However, the relationship between intra- Development Group at Scripps Institution of Ocean- seasonal velocity variability and transport variability ography for all facets of the operations of Spray un- cannot be determined from the temporal resolution of derwater gliders. We thank Pat and Lori Colin at the glider observations. Coral Reef Research Foundation for their support The relationship between the interannual sub- in Palau. thermocline transport variability and ENSO is difficult to assess as only one ENSO cycle was observed. The REFERENCES increase in equatorward subthermocline transport during the El Niño may be related to the vertical structure AVISO, 2014: Combined mean dynamic topography. CLS Space of NEC bifurcation, the strength of the NEC, re- Oceanography Division/CNES AVISO, accessed 14 May 2014. [Available online at https://www.aviso.altimetry.fr/en/ circulation of the NECC around the MD, or forcing data/products/auxiliary-products/mdt.html.] from the equator. The bifurcation latitude of the NEC Bingham, F. M., and R. Lukas, 1994: The southward intrusion increases with increasing depth, leading to a more of North Pacific Intermediate Water along the Mindanao northerly origin of the MC subthermocline water (Qu Coast. J. Phys. Oceanogr., 24, 141–154, doi:10.1175/ and Lukas 2003). As the bifurcation latitude shifts, the 1520-0485(1994)024,0141:TSIONP.2.0.CO;2. vertical structure of the bifurcation also changes, with ——, and ——, 1995: The distribution of intermediate water in the western equatorial Pacific during January–February greater changes in the subthermocline. When the bi- 1986. Deep-Sea Res. I, 42, 1545–1573, doi:10.1016/ furcation occurs at the highest latitude, the sub- 0967-0637(95)00064-D. thermocline bifurcates at higher latitude than at the Bretherton, F. P., R. E. Davis, and C. B. Fandry, 1976: A technique surface (Qu and Lukas 2003; Kim et al. 2004), possibly for objective analysis and design of oceanographic experi- causing poleward transport to increase more in the ments applied to MODE-73. Deep-Sea Res. Oceanogr. Abstr., subthermocline than in the thermocline. Preliminary 23, 559–582, doi:10.1016/0011-7471(76)90001-2. Chelton, D. B., R. A. Deszoeke, and M. G. Schlax, 1998: Geo- investigations with Argo climatology verify that al- graphical variability of the first baroclinic Rossby radius of though this occurs during strong ENSO events, there deformation. J. Phys. Oceanogr., 28, 433–460, doi:10.1175/ was only a small northward shift of bifurcation latitude 1520-0485(1998)028,0433:GVOTFB.2.0.CO;2. in the subthermocline and no change in the thermocline Chiang, T.-L., and T. Qu, 2013: Subthermocline eddies in the western during the 2009/10 El Niño. The vertical structure of the equatorial Pacific as shown by an eddy-resolving OGCM. J. Phys. Oceanogr., 43, 1241–1253, doi:10.1175/JPO-D-12-0187.1. MC/MUC transport and its relationship to ENSO has ——, C.-R. Wu, T. Qu, and Y.-C. Hsin, 2015: Activities of 50-80 day implications for the ITF transport, which is at a mini- subthermocline eddies near the Philippine coast. J. Geophys. mum during El Niño and maximum during La Niña Res. Oceans, 120, 3606–3623, doi:10.1002/2013JC009626. (Gordon et al. 1999; Hu et al. 2015). Regional transport CPC, 1997: Merged analysis of precipitation (CMAP). NOAA/ changes over a longer temporal range could be investi- OAR/ESRL PSD, accessed 5 August 2016. [Available online gated with Argo but would be limited near the coasts. at http://www.cpc.ncep.noaa.gov/products/global_precip/html/ wpage.cmap.html.] The high horizontal resolution and absolute reference Cronin, M. F., and M. J. McPhaden, 1998: Upper ocean salinity velocity from repeat gliders’ observations have provided balance in the western equatorial Pacific. J. Geophys. Res., accurate water mass transport and exchange in the 103, 27 567–27 587, doi:10.1029/98JC02605.

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2073

Davis, R. E., M. D. Ohman, D. L. Rudnick, and J. T. Sherman, Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Re- 2008: Glider surveillance of physics and biology in the south- analysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, ern California Current System. Limnol. Oceanogr., 53, 2151– doi:10.1175/1520-0477(1996)077,0437:TNYRP.2.0.CO;2. 2168, doi:10.4319/lo.2008.53.5_part_2.2151. Kashino, Y., 2005: Variability of the Mindanao Current: Mooring ——, W. S. Kessler, and J. T. Sherman, 2012: Gliders measure observation results. Geophys. Res. Lett., 32, L18611, western boundary current transport from the South Pacific to doi:10.1029/2005GL023880. the equator. J. Phys. Oceanogr., 42, 2001–2013, doi:10.1175/ ——, M. Aoyama, T. Kawano, N. Hendiarti, Y. Anantasena, JPO-D-12-022.1. K. Muneyama, and H. Watanabe, 1996: The water masses Delcroix, T., and C. Hénin, 1991: Seasonal and interannual varia- between Mindanao and New Guinea. J. Geophys. Res., 101, tions of sea surface salinity in the tropical Pacific Ocean. 12 391–12 400, doi:10.1029/95JC03797. J. Geophys. Res., 96, 22 135–22 150, doi:10.1029/91JC02124. ——, N. España, F. Syamsudin, K. J. Richards, T. Jensen, ——, and M. McPhaden, 2002: Interannual sea surface salinity P. Dutrieux, and A. Ishida, 2009: Observations of the North and temperature changes in the western Pacific warm pool Equatorial Current, Mindanao Current, and Kuroshio Cur- during 1992–2000. J. Geophys. Res., 107, 8002, doi:10.1029/ rent System during the 2006/07 El Niño and 2007/08 La Niña. 2001JC000862. J. Oceanogr., 65, 325–333, doi:10.1007/s10872-009-0030-z. Dutrieux, P., 2009: Tropical western Pacific currents and the origin ——, A. Atmadipoera, Y. Kuroda, and Lukijanto, 2013: Observed of intraseasonal variability below the thermocline. Ph.D. features of the Halmahera and Mindanao eddies. J. Geophys. thesis, University of Hawai‘i at Manoa, 278 pp. Res. Oceans, 118, 6543–6560, doi:10.1002/2013JC009207. Ferrari, R., and K. L. Polzin, 2005: Finescale structure of the T–S ——, I. Ueki, and H. Sasaki, 2015: Ocean variability east of relation in the eastern North Atlantic. J. Phys. Oceanogr., 35, Mindanao: Mooring observations at 78N, revisited. J. 1437–1454, doi:10.1175/JPO2763.1. Geophys. Res. Oceans, 120, 2540–2554, doi:10.1002/ Fine, R. A., R. Lukas, F. M. Bingham, M. J. Warner, and R. H. 2015JC010703. Gammon, 1994: The western equatorial Pacific: A water mass Katsura, S., E. Oka, B. Qiu, and N. Schneider, 2013: Formation crossroads. J. Geophys. Res., 99, 25 063–25 080, doi:10.1029/ and subduction of North Pacific tropical water and their in- 94JC02277. terannual variability. J. Phys. Oceanogr., 43, 2400–2415, Firing, E., Y. Kashino, and P. Hacker, 2005: Energetic sub- doi:10.1175/JPO-D-13-031.1. thermocline currents observed east of Mindanao. Deep-Sea Kessler, W. S., 2006: The circulation of the eastern tropical Res. II, 52, 605–613, doi:10.1016/j.dsr2.2004.12.007. Pacific: A review. Prog. Oceanogr., 69, 181–217, doi:10.1016/ Gordon, A. L., 1986: Inter-ocean exchange of thermocline water. j.pocean.2006.03.009. J. Geophys. Res., 91, 5037–5046, doi:10.1029/JC091iC04p05037. Kim, Y. Y., T. Qu, T. Jensen, and T. Miyama, 2004: Seasonal and ——, R. D. Susanto, and A. Ffield, 1999: Throughflow interannual variations of the North Equatorial Current bi- within Makassar Strait. Geophys. Res. Lett., 26, 3325–3328, furcation in a high-resolution OGCM. J. Geophys. Res., 109, doi:10.1029/1999GL002340. C03040, doi:10.1029/2003JC002013. Gu, D., and S. G. Philander, 1997: Interdecadal climate fluctu- Li, Y., and F. Wang, 2012: Spreading and salinity change of North ations that depend on exchanges between the tropics Pacific tropical water in the Philippine Sea. J. Oceanogr., 68, and extratropics. Science, 275, 805–807, doi:10.1126/ 439–452, doi:10.1007/s10872-012-0110-3. science.275.5301.805. ——, ——, and W. Han, 2013: Interannual sea surface salinity Hacker, P., E. Firing, R. Lukas, P. L. Richardson, and C. A. Collins, variations observed in the tropical North Pacific Ocean. 1989: Observations of the low-latitude western boundary cir- Geophys. Res. Lett., 40, 2194–2199, doi:10.1002/grl.50429. culation in the Pacific during WEPOCS III. Proc. Western Lien, R.-C., B. Ma, Y.-H. Cheng, C.-R. Ho, B. Qiu, C. M. Lee, and Pacific Int. Meeting and Workshop on TOGA COARE, M.-H. Chang, 2014: Modulation of Kuroshio transport by Nouméa, New Caledonia, Institute Français de Recherche mesoscale eddies at the Luzon Strait entrance. J. Geophys. Scientifique pour le Développement en Coopération/Centre Res. Oceans, 119, 2129–2142, doi:10.1002/2013JC009548. ORSTOM, 135–143. Lukas, R., 1988: Interannual fluctuations of the Mindanao Current Hsin, Y.-C., and B. Qiu, 2012: The impact of eastern-Pacific versus inferred from sea level. J. Geophys. Res., 93, 6744–6748, central-Pacific El Niños on the North Equatorial Counter- doi:10.1029/JC093iC06p06744. current in the Pacific Ocean. J. Geophys. Res., 117, C11017, ——,E.Firing,P.Hacker,P.L.Richardson,C.A.Collins, doi:10.1029/2012JC008362. R. Fine, and R. Gammon, 1991: Observations of the Mind- Hu, D. X., M. C. Cui, T. D. Qu, and Y. X. Li, 1991: A subsurface anao Current during the western equatorial Pacific Ocean northward current off Mindanao identified by dynamic cal- circulation study. J. Geophys. Res., 96, 7089–7104, culation. Oceanography of Asian Marginal Seas, K. Takano, doi:10.1029/91JC00062. Ed., Elsevier Oceanography Series, Vol. 54, Elsevier, 359–365. ——, T. Yamagata, and J. P. McCreary, 1996: Pacific low-latitude ——, and Coauthors, 2015: Pacific western boundary currents and western boundary currents and the Indonesian Through- their roles in climate. Nature, 522, 299–308, doi:10.1038/ flow. J. Geophys. Res., 101, 12 209–12 216, doi:10.1029/ nature14504. 96JC01204. Hu, S., D. Hu, C. Guan, F. Wang, L. Zhang, F. Wang, and Q. Wang, Masumoto, Y., and T. Yamagata, 1991: Response of the western 2016: Interannual variability of the Mindanao Current/ tropical Pacific to the Asian winter monsoon: The generation Undercurrent in direct observations and numerical simulations. of the Mindanao Dome. J. Phys. Oceanogr., 21, 1386–1398, J. Phys. Oceanogr., 46, 483–499, doi:10.1175/JPO-D-15-0092.1. doi:10.1175/1520-0485(1991)021,1386:ROTWTP.2.0.CO;2. Johnson, G. C., B. M. Sloyan, W. S. Kessler, and K. E. McTaggart, NCEP, 1996: NCEP/NCAR Reanalysis monthly means and 2002: Direct measurements of upper ocean currents and water other derived variables, accessed 19 March 2014. [Available properties across the tropical Pacific during the 1990s. Prog. online at https://www.esrl.noaa.gov/psd/data/gridded/data.ncep. Oceanogr., 52, 31–61, doi:10.1016/S0079-6611(02)00021-6. reanalysis.derived.surfaceflux.html.]

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC 2074 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 47

Osborn, T. R., and C. S. Cox, 1972: Oceanic fine structure. Geo- Schmitt, R. W., 1981: Form of the temperature-salinity rela- phys. Fluid Dyn., 3, 321–345, doi:10.1080/03091927208236085. tionship in the central water: Evidence for double-diffusive Pelland, N. A., C. C. Eriksen, and C. M. Lee, 2013: Subthermocline mixing. J. Phys. Oceanogr., 11, 1015–1026, doi:10.1175/ eddies over the Washington continental slope as observed by 1520-0485(1981)011,1015:FOTTSR.2.0.CO;2. Seagliders, 2003–09. J. Phys. Oceanogr., 43, 2025–2053, Schönau, M. C., and D. L. Rudnick, 2015: Glider observations doi:10.1175/JPO-D-12-086.1. of the North Equatorial Current in the western tropical Pietri, A., V. Echevin, P. Testor, A. Chaigneau, L. Mortier, Pacific. J. Geophys. Res. Oceans, 120, 3586–3605, doi:10.1002/ C. Grados, and A. Albert, 2014: Impact of a coastal-trapped 2014JC010595. wave on the near-coastal circulation of the Peru upwelling ——, ——, I. Cerovecki, G. Gopalakrishnan, B. Cornuelle, system from glider data. J. Geophys. Res. Oceans, 119, 2109– J. McClean, and B. Qiu, 2015: The Mindanao Current: Mean 2120, doi:10.1002/2013JC009270. structure and connectivity. Oceanography, 28, 34–45, Qiu, B., and R. Lukas, 1996: Seasonal and interannual variability of doi:10.5670/oceanog.2015.79. the North Equatorial Current, the Mindanao Current, and the Scripps Institution of Oceanography and International Argo Pro- Kuroshio along the Pacific western boundary. J. Geophys. gram, 2009: Roemmich-Gilson Argo Climatology, accessed Res., 101, 12 315–12 330, doi:10.1029/95JC03204. 20 October 2016. [Available online at http://sio-argo.ucsd.edu/ ——, and S. Chen, 2010: Interannual-to-decadal variability in RG_Climatology.html.] the bifurcation of the North Equatorial Current off the Philippines. Sherman, J., R. E. Davis, W. B. Owens, and J. Valdes, 2001: The J. Phys. Oceanogr., 40, 2525–2538, doi:10.1175/2010JPO4462.1. autonomous underwater glider ‘‘Spray.’’ IEEE J. Oceanic ——, ——, D. L. Rudnick, and Y. Kashino, 2015: A new paradigm Eng., 26, 437–446, doi:10.1109/48.972076. for the North Pacific subthermocline low-latitude western Talley, L. D., 1993: Distribution and formation of North Pacific boundary current system. J. Phys. Oceanogr., 45, 2407–2423, Intermediate Water. J. Phys. Oceanogr., 23, 517–537, doi:10.1175/JPO-D-15-0035.1. doi:10.1175/1520-0485(1993)023,0517:DAFONP.2.0.CO;2. Qu, T., and R. Lukas, 2003: The bifurcation of the North Equa- Todd, R. E., D. L. Rudnick, M. R. Mazloff, R. E. Davis, and B. D. torial Current in the Pacific. J. Phys. Oceanogr., 33, 5–18, Cornuelle, 2011: Poleward flows in the southern California doi:10.1175/1520-0485(2003)033,0005:TBOTNE.2.0.CO;2. Current System: Glider observations and numerical simu- ——, and E. J. Lindstrom, 2004: Northward intrusion of Antarctic lation. J. Geophys. Res., 116, C02026, doi:10.1029/ Intermediate Water in the western Pacific. J. Phys. Oceanogr., 2010JC006536. 34, 2104–2118, doi:10.1175/1520-0485(2004)034,2104: ——, ——, ——, B. D. Cornuelle, and R. E. Davis, 2012: Ther- NIOAIW.2.0.CO;2. mohaline structure in the California Current System: Obser- ——, T. Kagimoto, and T. Yamagata, 1997: A subsurface coun- vations and modeling of spice variance. J. Geophys. Res., 117, tercurrent along the east coast of Luzon. Deep-Sea Res. I, 44, C02008, doi:10.1029/2011JC007589. 413–423, doi:10.1016/S0967-0637(96)00121-5. Toole, J. M., R. C. Millard, Z. Wang, and S. Pu, 1990: Observations ——, H. Mitsudera, and T. Yamagata, 1998: On the western of the Pacific North Equatorial Current bifurcation at the boundary currents in the Philippine Sea. J. Geophys. Res., 103, Philippine coast. J. Phys. Oceanogr., 20, 307–318, doi:10.1175/ 7537–7548, doi:10.1029/98JC00263. 1520-0485(1990)020,0307:OOTPNE.2.0.CO;2. ——, ——, and ——, 1999: A climatology of the circulation and Tozuka, T., T. Kagimoto, and Y. Masumoto, 2002: Simulated water mass distribution near the Philippine coast. J. Phys. multiscale variations in the western tropical Pacific: The Oceanogr., 29, 1488–1505, doi:10.1175/1520-0485(1999)029,1488: Mindanao Dome revisited. J. Phys. Oceanogr., 32,1338– ACOTCA.2.0.CO;2. 1359, doi:10.1175/1520-0485(2002)032,1338: ——, J. Gan, A. Ishida, Y. Kashino, and T. Tozuka, 2008: Semi- SMVITW.2.0.CO;2. annual variation in the western tropical Pacific Ocean. Geo- Tsuchiya, M., 1968: Upper Waters of the Intertropical Pacific Ocean. phys. Res. Lett., 35, L16602, doi:10.1029/2008GL035058. Johns Hopkins Press, 50 pp. Roemmich, D., and J. Gilson, 2009: The 2004–2008 mean and an- Wang, F., N. Zang, Y. Li, and D. Hu, 2015: On the subsurface nual cycle of temperature, salinity, and steric height in the countercurrents in the Philippine Sea. J. Geophys. Res. global ocean from the Argo program. Prog. Oceanogr., 82, Oceans, 120, 131–144, doi:10.1002/2013JC009690. 81–100, doi:10.1016/j.pocean.2009.03.004. Wang, Q., F. Zhai, F. Wang, and D. Hu, 2014: Intraseasonal vari- Rudnick, D. L., and S. T. Cole, 2011: On sampling the ocean ability of the subthermocline current east of Mindanao. using underwater gliders. J. Geophys. Res., 116, C08010, J. Geophys. Res. Oceans, 119, 8552–8566, doi:10.1002/ doi:10.1029/2010JC006849. 2014JC010343. ——, R. E. Davis, C. C. Eriksen, D. M. Fratantoni, and M. J. Perry, Wijffels, S., E. Firing, and J. Toole, 1995: The mean structure and 2004: Underwater gliders for ocean research. Mar. Technol. variability of the Mindanao Current at 88N. J. Geophys. Res., Soc. J., 38, 73–84, doi:10.4031/002533204787522703. 100, 18 421–18 435, doi:10.1029/95JC01347. ——, G. Gopalakrishnan, and B. D. Cornuelle, 2015a: Cyclonic Wyrtki, K., 1961: Physical Oceanography of the Southeast Asian Waters. eddies in the Gulf of Mexico: Observations by underwater NAGA Rep., Vol. 2, Scripps Institute of Oceanography, 195 pp. gliders and simulations by numerical model. J. Phys. Ocean- ——, and B. Kilonsky, 1984: Mean water and current structure ogr., 45, 313–326, doi:10.1175/JPO-D-14-0138.1. during the Hawaii-to-Tahiti Shuttle Experiment. J. Phys. ——, S. Jan, and C. Lee, 2015b: A new look at circulation in the Oceanogr., 14, 242–254, doi:10.1175/1520-0485(1984)014,0242: western North Pacific: Introduction to the special issue. MWACSD.2.0.CO;2. Oceanography, 28, 16–23, doi:10.5670/oceanog.2015.77. Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year Sasaki, Y. N., N. Schneider, N. Maximenko, and K. Lebedev, 2010: monthly analysis based on gauge observations, satellite es- Observational evidence for propagation of decadal spiciness timates, and numerical model outputs. Bull. Amer. Meteor. anomalies in the North Pacific. Geophys. Res. Lett., 37, Soc., 78, 2539–2558, doi:10.1175/1520-0477(1997)078,2539: L07708, doi:10.1029/2010GL042716. GPAYMA.2.0.CO;2.

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC AUGUST 2017 S C H Ö NAU AND RUDNICK 2075

Yeager, S. G., and W. G. Large, 2007: Observational evidence of Zhang, L., D. Hu, S. Hu, F. Wang, F. Wang, and D. Yuan, 2014: winter spice injection. J. Phys. Oceanogr., 37, 2895–2919, Mindanao Current/Undercurrent measured by a subsurface doi:10.1175/2007JPO3629.1. mooring. J. Geophys. Res. Oceans, 119, 3617–3628, Zenk, W., G. Siedler, A. Ishida, J. Holfort, Y. Kashino, Y. Kuroda, doi:10.1002/2013JC009693. T. Miyama, and T. J. Müller, 2005: Pathways and variability of Zhang, R.-H., L. M. Rothstein, and A. J. Busalacchi, 1998: Origin the Antarctic Intermediate Water in the western equatorial of upper-ocean warming and El Niño change on decadal Paciﬁc Ocean. Prog. Oceanogr., 67, 245–281, doi:10.1016/j. scales in the tropical Paciﬁc Ocean. Nature, 391, 879–883, pocean.2005.05.003. doi:10.1038/36081.

Unauthenticated | Downloaded 10/07/21 09:18 AM UTC