1. Introduction
The Mindanao Current (MC), the western boundary current of the North Pacific Ocean tropical gyre, splits into two parts at the entrance of the Indonesian Seas. One part flows westward into the Sulawesi Sea, supplying volume to the main stream of the Indonesian Throughflow (ITF) in the Makassar Strait, while some of this part retroflects back eastward to the Pacific Ocean through the Sangihe Islands. The other part retroflects back to the Pacific Ocean near the Talaud Islands and meets with another western boundary current from the Southern Hemisphere, the New Guinea Coastal Current/Undercurrent (NGCC/UC), forming the beginning of the North Equatorial Countercurrent (NECC) (Lukas et al. 1991; Gordon 2005; Kashino et al. 1996, 2013). Two quasi-permanent eddies called the Mindanao Eddy (ME) and the Halmahera Eddy (HE) are generated at the retroflection of the above two western boundary currents, which carry water masses from subtropical and subpolar gyres of both hemispheres to the far western tropical Pacific Ocean, forming the so-called water mass crossroads at the entrance of the ITF (Fine et al. 1994).
In the thermocline, two water masses from the North and South Pacific subtropical gyres, characterized by salinity maxima, called the North Pacific Tropical Water (NPTW) and the South Pacific Tropical Water (SPTW), respectively, meet at the entrance of the Indonesian seas (Lukas et al. 1991; Bingham and Lukas 1994; Fine et al. 1994). Below them, two intermediate water masses with salinity minima are transported to this area by the two western boundary currents from both hemispheres, which are called the North Pacific Intermediate Water (NPIW) and the Antarctic Intermediate Water (Lindstrom et al. 1987; Tsuchiya 1991; Talley 1993; Fine et al. 1994; Qu et al. 1999).
Existing studies indicate two major pathways of the ITF transports, i.e., the western and eastern routes, in the Indonesian seas (e.g., Gordon 2005). Several minor channels such as the Karimata Strait between the South China Sea and the Java Sea and the Sibutu Passage between the Sulu Sea and the Sulawesi Sea also supply volume to the ITF (Fang et al. 2010; Susanto et al. 2010; Gordon et al. 2012). Observations have shown that the major transport of the ITF, consisting of the thermocline waters from the North Pacific Ocean, enters the Makassar Strait, forming the western route of the ITF (Gordon et al. 1999, 2008; Susanto et al. 2012; Susanto and Song 2015; Gordon et al. 2019). Water transports through the Maluku Sea and the Halmahera Sea also contribute to the ITF, forming the eastern route (Gordon 2005; Yuan et al. 2018; Li et al. 2020). Compared to the long time series mooring observations in the Makassar Strait (Gordon et al. 2019), historical direct hydrography and current measurements in the Maluku Sea were rare, except for some sporadic cruise observations (Gordon and Fine 1996; Ilahude and Gordon 1996; Kashino et al. 2001), and a mooring with failed measurements above the 740-m depth (Luick and Cresswell 2001). A potential source of South Pacific salty waters to the Banda Sea through the Maluku Sea is suggested by the hydrography data (Gordon and Fine 1996).
Previous studies have shown that the ocean circulation near the Talaud Islands east of the Sulawesi Sea is very complicated. Kashino et al. (2013) indicated that the circulation differs significantly among different cruises with prominent seasonal and interannual variability. The strong variability of the cyclonic ME suggests complex patterns of the MC retroflection. Two moorings have been deployed in the channel between the Talaud and Morotai Islands in 1994/95 (Kashino et al. 1999), the results of which suggest that the South Pacific waters flow into the northern Maluku Sea during boreal summer in the subsurface. However, the current meters in the upper 350 m failed to return measurements.
Arruda and Nof (2003) pointed out that the beta effect and nonlinearity are essential for the generation of asymmetric double eddies by the confluence of the two western boundary currents. A western boundary current flowing by a gap experiences nonlinear bifurcation and hysteresis (Sheremet 2001; Kuehl and Sheremet 2009). The regime shifts of the current may occur if perturbed by mesoscale eddies or winds (Yuan and Li 2008; Wang et al. 2010; Yuan and Wang 2011). The intraseasonal variability in the Sulawesi Sea can be induced by eddy shedding of the MC (Qiu et al. 1999; Wang and Yuan 2012). Nonlinear collisions of two western boundary currents at a wide gap have multiple equilibria (penetration, choke, and eddy shedding), which are subject to regime shifts if perturbed by Rossby waves or eddies (Wang and Yuan 2012, 2014).
Since 2014, a major Western Pacific Ocean Circulation–Indonesian Throughflow (WPOC-ITF) mooring array has been constructed in the western Pacific and Indonesian seas by the Institute of Oceanology of the Chinese Academy of Sciences (IOCAS). Subsurface moorings have been deployed in the eastern Indonesian Seas under the cooperative agreement between IOCAS and the Research Center for Oceanography, Indonesian Institute of Sciences (RCO/LIPI). The results in the Maluku Channel have been reported recently by Yuan et al. (2018), showing strong current variability inconsistent with the local wind forcing. The forcing by the movement of the MC retroflection has been suggested, based on historical surface drifter trajectories and numerical modeling results. However, direct observations of the MC path shift have yet to be obtained with in situ measurements. Here we focus on the mooring measurements in the Talaud–Halmahera Channel (TH Channel hereinafter), where the MC retroflection in a cyclonic intrusion path shall enter to join the NECC. The movement of the MC retroflection is studied, using the nearly 2-yr time series of three subsurface moorings in the TH Channel from December 2016 to September 2018.
The rest of the paper is organized as follows. The data used in this study and their processing are described in section 2. Time series of the moored current measurements through the TH Channel are analyzed with transports estimated in section 3. The water mass properties in the TH section are analyzed using conductivity–temperature–depth (CTD) profiles of the cruises and the moored CTDs in section 4. The seasonal movement of the MC retroflection is discussed using satellite sea level and ocean color data. The mass budget in the upper northern Maluku Sea is diagnosed using a high-resolution ocean model simulation. Conclusions are summarized in section 5.
2. Data and methods
In this section, the configurations of the moorings used in this study and their data processing procedures are described. The method to estimate the transports is then introduced. The high-resolution ocean model simulations and several other datasets used are also described.
a. The moorings
Three moorings were deployed in a section between the Talaud and Halmahera Islands in early December 2016 and were rotated twice in early November 2017 and late September 2018 (Fig. 1). The positions of the moorings are 2°24.1′N, 127°30.1′E; 2°57.6′N, 127°16.0′E; and 3°30.0′N, 127°2.3′E for TH1, TH2, and TH3, respectively. The displacement of the moorings between the deployments is only a few tens of meters and is ignored in this study. The mooring deployment and rotation activities are conducted on board of the Indonesian research vessel Baruna Jaya VIII. Unfortunately, the TH3 mooring was stuck to a subsurface fishing net and was not fully recovered in November 2017. Only the upper part of the TH3 mooring with an upward-looking acoustic Doppler current profiler (ADCP) was recovered. This mooring was not redeployed in November 2017.
Bathymetry in the vicinity of the northern Maluku Sea based on ETOPO1 database. The three moorings TH1, TH2, and TH3 in the Talaud–Halmahera Channel used in this study are marked by black triangles. The thick black line across the channel represents the section for transport estimates. Talaud-Morotai North (TMN) and Talaud-Morotai South (TNS) are the mooring locations in Kashino et al. (1999) shown by yellow triangles. The three moorings of the Maluku Channel in Yuan et al. (2018) are shown by purple triangles. The locations of the CTD station and cruise tracks near the TH Channel during the mooring maintenance cruises are marked by dark gray dots and colored lines (orange, 2015; red, 2016; green 2017; blue 2018).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Bathymetry in the vicinity of the northern Maluku Sea based on ETOPO1 database. The three moorings TH1, TH2, and TH3 in the Talaud–Halmahera Channel used in this study are marked by black triangles. The thick black line across the channel represents the section for transport estimates. Talaud-Morotai North (TMN) and Talaud-Morotai South (TNS) are the mooring locations in Kashino et al. (1999) shown by yellow triangles. The three moorings of the Maluku Channel in Yuan et al. (2018) are shown by purple triangles. The locations of the CTD station and cruise tracks near the TH Channel during the mooring maintenance cruises are marked by dark gray dots and colored lines (orange, 2015; red, 2016; green 2017; blue 2018).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Bathymetry in the vicinity of the northern Maluku Sea based on ETOPO1 database. The three moorings TH1, TH2, and TH3 in the Talaud–Halmahera Channel used in this study are marked by black triangles. The thick black line across the channel represents the section for transport estimates. Talaud-Morotai North (TMN) and Talaud-Morotai South (TNS) are the mooring locations in Kashino et al. (1999) shown by yellow triangles. The three moorings of the Maluku Channel in Yuan et al. (2018) are shown by purple triangles. The locations of the CTD station and cruise tracks near the TH Channel during the mooring maintenance cruises are marked by dark gray dots and colored lines (orange, 2015; red, 2016; green 2017; blue 2018).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The moorings were deployed to monitor the exchange between the northern Maluku Sea and the open Pacific Ocean. In our earlier study, the upper currents in the Maluku Channel are found to be associated with the MC retroflection changes (Yuan et al. 2018) [see Fig. 1 for the mooring locations of Yuan et al. (2018) and this study]. Each mooring is equipped with a 75-kHz upward-looking ADCP manufactured by the U.S. Teledyne RD Instruments Inc. at a normal depth of 450 m. Due to inaccurate length of the rope, the instruments on the TH1 mooring during the first deployment were about 130 m shallower than designed. The depths of the ADCPs are 316, 442, and 441 m at TH1, TH2, and TH3 during the first deployment and 439 and 461 m at TH1 and TH2 during the second deployment, respectively. The 75-kHz ADCPs are configured to have sixty 8-m bins with the sampling ensemble time interval set at 1 h, which cover the full depth range from the ADCPs to the sea surface. In addition, MicroCATs 37SM CTDs made by the U.S. Sea-Bird Scientific Inc. were mounted at a nominal depth of 200 m to the moorings to measure the water mass properties in the thermocline. In addition, the 75-kHz shipboard ADCP (SADCP) and Sea-Bird 911 plus CTD system covering upper 300 m and at least 1000 m, respectively, were conducted during annual IOCAS-RCO/LIPI joint cruises since 2014. The shipboard CTD profiles supplement the moored CTDs in measuring the water mass properties in the channel, since the vertical resolution of moored temperature and salinity sensors is low. The temporal coverage of the shipboard measurements is listed in Table 1.
The time windows of the shipboard measurements in the Mindanao–Talaud–Halmahera section.
b. Mooring data processing
The raw data of the moored ADCPs were quality controlled according to a standard procedure following Cowley et al. (2008). Sporadic missing data were filled with linear interpolation in the vertical. The ADCP data were linearly interpolated onto a 1-m vertical grid and then averaged into a 10-m grid for analysis. The blowing down of the main floats due to strong tidal currents is usually several tens of meters within one day, and sometimes reaches as large as 200 m during spring–neap tides. The ADCP data in the top 50 m were contaminated by the reflection from the sea surface, i.e., the side-lobe effect, and were discarded. The hourly data were low-passed by a fourth-order Butterworth filter with a cutoff period of 3 days to suppress the tides and averaged into daily time series for further analysis. A low-pass Butterworth filter with a cutoff period of 120 days was then applied to eliminate the intraseasonal variability.
c. Transport estimate
The 1-yr volume transport of the upper ocean was estimated using the three moorings of the first deployment from December 2016 to November 2017. The along-channel velocity (ACV) was defined as 67° clockwise from due north perpendicular to a section from the coast of the Talaud Islands to the coast of the Halmahera Island (Fig. 1), through which the volume transport of the TH Channel was estimated. The highest probabilities of the current direction during the 2-yr measurements at TH2 vary from 40° to 80° from due north in the upper 400 m, suggesting the suitable choice of the 67°ACV for the transport estimate. The ACVs between the moorings were linearly interpolated. The ACVs north of TH3 and south of TH1 along the section were estimated using three kinds of boundary conditions at the coasts: assuming constant as the nearest mooring (freeslip scheme), linearly decreasing to zero at the coast (nonslip scheme), and extrapolation from the nearest two moorings (extrapolation scheme). As the cruise tracks do not cover the nearshore part of the section, the water depths between the coasts and the moorings were derived from the ETOPO1 Global Relief Model, obtained from the National Oceanic and Atmospheric Administration’s National Centers for Environmental Information.
d. Shipboard observations
The SADCP data were postprocessed using Common Ocean Data Access System (CODAS) software, including calibration and quality control. The barotropic tidal currents of four principal constituents (M2, S2, K1, O1) extracted from the TOPEX/Poseidon global tidal model (TPXO) version 9 (Egbert and Erofeeva 2002) were removed from the postprocessed SADCP velocities, which were then smoothed with a 2-h running mean filter to suppress the high-frequency variability. Note that the strong baroclinic tidal currents, with the amplitudes of over 25 cm s−1 in the upper ocean according to the mooring measurements (figure omitted), were not filtered out. The range of the SADCP measurements is only about 100 m during the 2015 cruise due to the low power of the transducers, which was replaced with a new set of SADCP after the cruise. The 911 CTD system is routinely calibrated every year by the producer, the Sea-Bird Scientific, Inc. The CTD data have accuracies of 0.001°C and 0.0003 S m−1 in temperature and conductivity, respectively, and were postprocessed with a standard procedure using the SBE Data Processing software and binned onto a 1-dbar vertical grid for further analysis.
e. MPIOM simulations
The high-resolution simulation of the Max Planck Institute of Meteorology Ocean Model (MPIOM) was downloaded and analyzed to show the mass budget of the northern Maluku Sea. The model employs a tripolar horizontal grid with a resolution of about 0.1° near the equator. Eighty uneven vertical layers from the sea surface to 6038.5 m are used in the vertical. The model was spun up for 25 years using the Ocean Model Intercomparison Project (OMIP) forcing derived from the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-15 data and then forced by the 6-hourly National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data from 1948 to present. The details of the MPIOM simulations have been described by von Storch et al. (2012, 2016). The monthly climatology during the period of 2001–14 was used for the analysis of the seasonal variability. The simulated circulation in the upper Maluku Channel has been verified to be reasonable with moored current meter data (Yuan et al. 2018).
f. Other datasets
The daily absolute dynamic topography (ADT) of the AVISO multisatellite merged gridded product were used to study the sea level variations associated with the MC movement. The 1/4° gridded data from 1993 to 2018 were averaged to form a monthly climatology. The 4-km monthly climatological chlorophyll-a data of the Ocean Color Climate Change Initiative program (OCCCI) version 4.2 (Sathyendranath et al. 2019) were also used to mark the seasonal movement of the surface MC retroflection. The World Ocean Atlas 2013 version 2 (WOA13v2) climatological density profiles were used to calculate the baroclinic modes at TH2 mooring site. CTD station profiles at 8°N, 127°E and 0.45°S, 133.5°E in the regions of the Pacific western boundary currents MC and NGCUC are used to represent the characters of the typical water masses from the North and South Pacific, respectively.
3. Results
In this section, the mean structure and seasonal variability of the currents in the TH Channel are studied using the mooring measurements. The variability of the ACVs is further analyzed using the TH2 mooring measurements because of their good vertical and temporal coverage.
a. ACVs in the TH channel
The mean ACVs at the three mooring locations show similar vertical profiles with the speed decreasing with depth monotonously (Figs. 2d–f). The velocity is positive nearly throughout the ranges of the ADCP measurements, indicating northeastward mean flow from the northern Maluku Sea to the western Pacific Ocean. The mean ACVs in the TH channel are stronger in the northern channel than in the south, with the ACVs at 60 m reaching 50, 40, and 20 cm s−1, respectively, at the three mooring sites. The mean ACVs nearly vanish at 400 m at the bottom of the ADCP ranges at TH2 and TH3 (Figs. 2e,f). The mean vertical profiles of the ACVs during the two deployments are similar at TH1 and TH2 (Figs. 2d,e).
120-day low-passed along-channel velocity (cm s−1) of moored ADCP measurements at (a) TH1, (b) TH2, and (c) TH3 in the Talaud–Halmahera Channel. (d)–(f) The mean ACV profiles during each deployment (blue lines for 2017, red lines for 2018) and the whole period (black solid lines). One standard deviation of the daily ACVs is shown in black dashed lines.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
120-day low-passed along-channel velocity (cm s−1) of moored ADCP measurements at (a) TH1, (b) TH2, and (c) TH3 in the Talaud–Halmahera Channel. (d)–(f) The mean ACV profiles during each deployment (blue lines for 2017, red lines for 2018) and the whole period (black solid lines). One standard deviation of the daily ACVs is shown in black dashed lines.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
120-day low-passed along-channel velocity (cm s−1) of moored ADCP measurements at (a) TH1, (b) TH2, and (c) TH3 in the Talaud–Halmahera Channel. (d)–(f) The mean ACV profiles during each deployment (blue lines for 2017, red lines for 2018) and the whole period (black solid lines). One standard deviation of the daily ACVs is shown in black dashed lines.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The time series of the ACVs show vigorous seasonal variability of the currents (Figs. 2a–c). A strong annual variation in the upper ocean above 150 m is found at all three mooring sites. The current flows strongly northeastward toward the western Pacific Ocean during boreal winter with the velocity of more than 40 cm s−1, which is stronger and lasts longer in the northern part of the channel than in the south. It relaxes in boreal summer, or even reverses at TH1, suggesting that the waters from the Pacific enter the northern Maluku Sea above 100 m in the southern TH Channel. A prominent phenomenon is the southwestward flow below 150 m from April to July, especially in the northern part of the channel. Despite the interannual variations indicated by the year-to-year differences between the two deployments, the intrusion of the western Pacific waters in the subsurface layer annually is observed by the TH2 mooring measurements. Water mass analysis suggests its origin of South Pacific thermocline waters, which will be discussed in section 4a.
The annual period oscillations can be easily identified in the spectrum density of the unfiltered TH2 ACVs, with the core in the upper 150 m (Fig. 3a). The density of the spectrum is concentrated in the low-frequency band of over 150 days. The intraseasonal variability at periods of 50–60 days and 70–100 days can also be identified, which is at least one order of magnitude smaller than the seasonal variability. The first mode (62.1% of variance) of empirical orthogonal function (EOF), shows a vertically consistent structure in the upper ocean, with strong currents at the top decreasing with depth (Fig. 3b). The second mode (21.3% of variance) has a zero-crossing at about 150 m, suggesting reversed flows in the subsurface layer. The first two modes of EOFs are similar to the first two baroclinic modes based on the climatological density profile of WOA13v2, suggesting that the currents variability is dominated by the low baroclinic modes in this channel.
(a) Spectrum density and (b) empirical orthogonal functions (solid lines) of the along-channel velocity of the 2-yr measurements at TH2 mooring. The percentages in (b) stand for the explained variance. The dashed lines in (b) are the baroclinic mode functions based on the WOA13v2 climatological density profile at TH2.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(a) Spectrum density and (b) empirical orthogonal functions (solid lines) of the along-channel velocity of the 2-yr measurements at TH2 mooring. The percentages in (b) stand for the explained variance. The dashed lines in (b) are the baroclinic mode functions based on the WOA13v2 climatological density profile at TH2.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(a) Spectrum density and (b) empirical orthogonal functions (solid lines) of the along-channel velocity of the 2-yr measurements at TH2 mooring. The percentages in (b) stand for the explained variance. The dashed lines in (b) are the baroclinic mode functions based on the WOA13v2 climatological density profile at TH2.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Harmonic analyses are used to extract the annual and semiannual variability of the currents. The amplitudes of the annual cycle have two cores (Fig. 4a). One is over 25 cm s−1 at the depth of 80–100 m, with peaks in January and July. The other one is nearly 10 cm s−1 lying below 250 m, with the northeastward and southwestward flow reaching maximum in October and April, respectively. The semiannual amplitudes also have a core between 100 and 150 m (Fig. 4b), which is weaker than that of the annual cycle in the upper 150 m. The superposition of the annual and semiannual cycle in boreal winter leads to a strong northeastward velocity variation in the upper 150 m (Fig. 4c). In other seasons, the phases of the annual and semiannual variability differ, resulting in relatively weak but prolonged southwestward velocity variations in the surface layer. In the subsurface, the magnitudes of the annual and semiannual variability are comparable, the superposition of which results in a seasonal cycle with the peak southwestward variations from April to June. The summation of the annual and semiannual harmonics can reproduce the seasonal variability reasonably in general (Figs. 4c,d), explaining 50%–80% variance of the 120-day low-passed velocity variability in the upper ocean at TH1 and TH2.
(a) Annual and (b) semiannual harmonics of the ACV (cm s−1) at the TH2 mooring, and (c) their summation in comparison to (d) the 120-day low-passed ACV variations at TH2. The ACV variations in (d) are calculated by removing the 2-yr mean ACV from the time series of the ACV.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(a) Annual and (b) semiannual harmonics of the ACV (cm s−1) at the TH2 mooring, and (c) their summation in comparison to (d) the 120-day low-passed ACV variations at TH2. The ACV variations in (d) are calculated by removing the 2-yr mean ACV from the time series of the ACV.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(a) Annual and (b) semiannual harmonics of the ACV (cm s−1) at the TH2 mooring, and (c) their summation in comparison to (d) the 120-day low-passed ACV variations at TH2. The ACV variations in (d) are calculated by removing the 2-yr mean ACV from the time series of the ACV.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
b. Transport
The ACVs in the mooring section show strong northeastward currents especially in the northern part of the TH Channel based on SADCP measurements in the falls of 2015–18 (Fig. 5). The ACV is the maximum in the northern TH Channel, with northeastward velocity reaching more than 50 cm s−1 at the depth of about 100 m during the 2015–17 cruises. In the 2018 fall, the northeastward currents are very strong in the whole section in the upper 60 m, with the velocity larger than 80 cm s−1 at the surface. Below 150 m, southwestward currents are observed in the southern part of the section, with a maximum velocity of more than 20 cm s−1 in the 2016 and 2017 falls. The SADCP measurements suggest that the linear interpolation of velocity between the moorings in the transport estimates is appropriate generally. The cruises did not cover the gaps between the moorings and the coasts. The extrapolation in these gaps could lead to sizable uncertainties in transport estimates.
The ACVs (cm s−1) between the Talaud and Halmahera Islands measured by the shipboard ADCP current meter in the (a) 2015, (b) 2016, (c) 2017, and (d) 2018 cruises, respectively. Inset in (a) shows the shipboard ADCP sections during the 2015–18 cruises, which are also shown in Fig. 1. The black triangles at the top of the panels mark the mooring locations of TH1, TH2, and TH3.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The ACVs (cm s−1) between the Talaud and Halmahera Islands measured by the shipboard ADCP current meter in the (a) 2015, (b) 2016, (c) 2017, and (d) 2018 cruises, respectively. Inset in (a) shows the shipboard ADCP sections during the 2015–18 cruises, which are also shown in Fig. 1. The black triangles at the top of the panels mark the mooring locations of TH1, TH2, and TH3.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The ACVs (cm s−1) between the Talaud and Halmahera Islands measured by the shipboard ADCP current meter in the (a) 2015, (b) 2016, (c) 2017, and (d) 2018 cruises, respectively. Inset in (a) shows the shipboard ADCP sections during the 2015–18 cruises, which are also shown in Fig. 1. The black triangles at the top of the panels mark the mooring locations of TH1, TH2, and TH3.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The 1-yr volume transport between 60 and 300 m through the TH Channel is estimated using measurements of the three moorings to be 13.1, 10.1, and 13.2 Sv (1 Sv ≡ 106 m3 s−1) from the northern Maluku Sea to the western Pacific Ocean based on the freeslip, nonslip, and linear extrapolation boundary conditions at the coasts (see section 2c), respectively (Fig. 6). Due to lack of observations near the coasts, it is difficult to judge which extrapolation scheme leads to the most accurate transport estimate. The standard deviations of the daily subtidal velocity are large (Table 2), but the mean transports are large enough to be statistically significant. The standard errors of the mean are also large, with a similar magnitude as the mean transports, implying large uncertainties of the mean transports estimated based on the 1-yr time series.
Volume transport (Sv) between 60- and 300-m depths in the Talaud–Halmahera Channel during December 2016–November 2017. Three extrapolation schemes described in section 2 are used. The daily transports are smoothed by a 15-day running average window. Numbers are the mean transports.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Volume transport (Sv) between 60- and 300-m depths in the Talaud–Halmahera Channel during December 2016–November 2017. Three extrapolation schemes described in section 2 are used. The daily transports are smoothed by a 15-day running average window. Numbers are the mean transports.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Volume transport (Sv) between 60- and 300-m depths in the Talaud–Halmahera Channel during December 2016–November 2017. Three extrapolation schemes described in section 2 are used. The daily transports are smoothed by a 15-day running average window. Numbers are the mean transports.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Mean transports (Sv) estimated using three concurrent moorings in the Talaud–Halmahera Channel during 2016–17 and their statistics. Standard errors are calculated at a 95% significance level.
Considering the northward mean transport of the Maluku Channel is 1.0–1.3 Sv (Yuan et al. 2018) between 60 and 315 m, which is only about one-tenth of the TH Channel transport, the flow (at least 8.8 Sv transport) through the TH Channel is dominated by the MC retroflection between the Sangihe and Talaud Islands or through the Sangihe Island chains (Fig. 1). The latter pathway is regarded as a longer route of the MC retroflection in the Sulawesi Sea, if the transport through the Sibutu Passage can be ignored or contributes only to the ITF in the Makassar Strait. The results suggest that the mean transport through TH Channel mainly consists of the MC retroflection of the North Pacific origin. There have been no direct current measurements in the Sangihe–Talaud Channel and in between the Sangihe Islands in history. The mass budget of the northern Maluku Sea will be discussed in section 4c using a high-resolution model simulation.
The northeastward flow reaches a maximum in boreal winter exceeding 25 Sv assuming constant velocity between the moorings and the coasts (freeslip scheme). This strong outflow in winter from the northern Maluku Sea includes both the MC retroflection and the Indonesian sea outflow from the Maluku Channel (Yuan et al. 2018). The transport decreases sharply in March and reverses in May, reaching a maximum inflow of ~5 Sv from the Pacific into the Indonesian seas. It becomes northeastward again in June, but with weaker magnitudes in fall than in winter.
4. Discussions
a. Water masses
The water mass properties are found to vary significantly along the Mindanao–Talaud–Halmahera section based on shipboard CTD measurements during the cruises (Fig. 7). Generally speaking, the salinity extreme characteristics of the water masses from the Northern Hemisphere are evident at subsurface depths in the northern part of the section, whereas the waters tend to be well mixed in the southern part, showing an isohaline structure similar to the Indonesian sea waters. The NPTW, characterized by its high salinity over 35.0 psu near 24.0σθ isopycnals, usually exists north of the Talaud Islands, except during the 2017 cruise, when the salinity maximum of the NPTW is much weaker. In the lower thermocline, the NPIW from the midlatitude North Pacific Ocean with a salinity minimum on 26.5–26.8σθ isopycnals (Talley 1993) also spread into the northern channel, with a similar coverage as the NPTW. The horizontal boundary between the North Pacific waters with two salinity extrema and the well-mixed Indonesian sea waters is easily identified in the northern part of the TH Channel. The CTD profiles in 2016, 2017, and 2018 were collected at the time, when northeastward currents occupied the TH3 site (Fig. 2; Table 1), the T–S relation of which suggests the MC mainstream retroflection through the northern part of the TH Channel.
(left) Potential temperature–salinity relation and (right) salinity sections based on shipboard CTD profiles during (a),(b) 2014; (c),(d) 2015; (e),(f) 2016; (g),(h) 2017; and (i),(j) 2018 cruises in the TH Channel, respectively. The station maps of the cruises are shown by inlets of the T–S plots. The color in the T–S plots stands for the latitudes of the CTD stations. Gray points in the T–S plots stand for the CTD data other than along the Mindanao–Talaud–Halmahera section.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(left) Potential temperature–salinity relation and (right) salinity sections based on shipboard CTD profiles during (a),(b) 2014; (c),(d) 2015; (e),(f) 2016; (g),(h) 2017; and (i),(j) 2018 cruises in the TH Channel, respectively. The station maps of the cruises are shown by inlets of the T–S plots. The color in the T–S plots stands for the latitudes of the CTD stations. Gray points in the T–S plots stand for the CTD data other than along the Mindanao–Talaud–Halmahera section.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
(left) Potential temperature–salinity relation and (right) salinity sections based on shipboard CTD profiles during (a),(b) 2014; (c),(d) 2015; (e),(f) 2016; (g),(h) 2017; and (i),(j) 2018 cruises in the TH Channel, respectively. The station maps of the cruises are shown by inlets of the T–S plots. The color in the T–S plots stands for the latitudes of the CTD stations. Gray points in the T–S plots stand for the CTD data other than along the Mindanao–Talaud–Halmahera section.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The North Pacific water intrusion into the northern Maluku Sea is remarkable during the 2015 cruise, when the high-salinity NPTW over 35 psu and the low-salinity NPIW are found in the middle of the TH Channel near 2.5°N (Figs. 7c,d). All of the cruises were conducted roughly in the same season in November except the one in September 2018 (Table 1). The differences among these snapshots of water mass distributions are likely due to interannual variations of the MC retroflection in 2015, which is evidenced by the low sea level anomalies at the entrance of the ITF (see section 4b).
The shallow CTD of TH1 mooring during the first deployment provides a glance of the water property variations above the 150-m depth. The instrument suffered strong blowdown due to the strong currents, with the depth varying between 60 and 180 m. In spite of the dispersive T–S plot, the water properties suggest that the water masses in the upper thermocline differ significantly from those in the western boundary currents from either hemisphere (Fig. 8a). The low salinity in the upper ocean suggests that the water masses come from the Indonesian seas, where large precipitation combined with enhanced vertical mixing results in a fresh upper thermocline (Ffield and Gordon 1996; Gordon 2005). Based on hydrographic data along 130°E and the Mindanao–New Guinea sections, a relatively freshwater mass in the upper thermocline is suggested to be advected out of the Indonesian seas through the TH Channel to supply the NECC as the North Pacific tropical subsurface water (Li et al. 2018), which is characterized by a meridional salinity minimum sandwiched between the salty tropical water from both hemispheres and a vertical salinity maximum at 22.5–25.5σθ isopycnals compared to the fresh surface and the intermediate waters in the western Pacific Ocean (Wang et al. 2013). The mooring measurements evidence that the water masses from the Indonesian seas contribute to the strong northeastward flow in the upper thermocline of the TH Channel, which is a source of the eastward NECC across the Pacific Ocean. The source of these Indonesian sea waters in the upper ocean can be traced further to the southern Maluku Sea or the Sulawesi Sea. The mooring measurements in the Maluku Channel suggest a mean transport of 1.0–1.3 Sv northward between 60 and 315 m. The surface currents flowing eastward through the gaps of the Sangihe Ridge out of the Sulawesi Sea have also been observed by the surface drifters (Li et al. 2018; Yuan et al. 2018). These routes of the Indonesian seas waters into the northern Maluku Sea are the source of the outflow in the upper thermocline of the TH Channel.
Potential temperature–salinity relation measured by the top CTD of the moorings (a),(d) TH1; (b),(e) TH2; and (c) TH3 during two deployments [(top) 2016–20 and (bottom) 2017–18]. All the CTDs are designed to sit at 200 m. The CTD on the TH1 mooring during the first deployment is shallower than designed by 130 m (see text for details). The color stands for the time in a year of the CTD measurements. CTD station profiles at 8°N, 127°E (black solid lines) and 0.45°S, 133.5°E (red solid lines) in the regions of the Pacific western boundary currents MC and NGCUC are overlapped to represent the typical water masses from the North and South Pacific, respectively.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Potential temperature–salinity relation measured by the top CTD of the moorings (a),(d) TH1; (b),(e) TH2; and (c) TH3 during two deployments [(top) 2016–20 and (bottom) 2017–18]. All the CTDs are designed to sit at 200 m. The CTD on the TH1 mooring during the first deployment is shallower than designed by 130 m (see text for details). The color stands for the time in a year of the CTD measurements. CTD station profiles at 8°N, 127°E (black solid lines) and 0.45°S, 133.5°E (red solid lines) in the regions of the Pacific western boundary currents MC and NGCUC are overlapped to represent the typical water masses from the North and South Pacific, respectively.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Potential temperature–salinity relation measured by the top CTD of the moorings (a),(d) TH1; (b),(e) TH2; and (c) TH3 during two deployments [(top) 2016–20 and (bottom) 2017–18]. All the CTDs are designed to sit at 200 m. The CTD on the TH1 mooring during the first deployment is shallower than designed by 130 m (see text for details). The color stands for the time in a year of the CTD measurements. CTD station profiles at 8°N, 127°E (black solid lines) and 0.45°S, 133.5°E (red solid lines) in the regions of the Pacific western boundary currents MC and NGCUC are overlapped to represent the typical water masses from the North and South Pacific, respectively.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Other CTDs near 200 m measured the properties of the lower-thermocline waters (Figs. 8b–d). Near the isopycnals 25.2–26.4σθ, the salinity of the northern and southern Pacific waters is quite different, with the latter much saltier than the former. Most of the time, the water properties in the channel are close to the North Pacific waters except during April to July, when the subsurface currents flow into the Indonesian Seas. The subsurface salinity as high as 35 psu suggests the intrusion of the SPTW. The results underline the importance of the Maluku Sea as a route of the ITF for the South Pacific waters to enter the Indonesian seas. In boreal spring–summer, the intrusion of high-salinity SPTW can be identified at all three mooring sites. This is consistent with the observations of Kashino et al. (1999) showing southwestward currents carrying South Pacific waters entering the Maluku Sea from the southern mooring location Talaud-Morotai South (TMS) during 1993/94 (Fig. 1). The South Pacific waters are suggested to spread over the entire TH Channel from the TMS channel. A sill of about 300 m deep between TMS and Talaud-Morotai North (TMN) probably has prevented the South Pacific waters from entering the northern channel TMN (Fig. 1).
b. The seasonal movement of the MC retroflection
The MC is the western boundary current of the cyclonic North Pacific tropical gyre, which retroflects at the entrance of the Indonesian seas to form the beginning of the NECC. The low ADT inside the ME, surrounded by the North Equatorial Current (NEC), MC, and NECC, can be used to identify the movement of the MC retroflection (Fig. 9a). The ME is the strongest and largest in boreal winter, with a tongue of low sea level extending into the northern Maluku Sea, suggesting an intrusion of the MC into the Maluku Sea. The ME becomes smaller and weaker during summer to fall than in winter, whereas the HE gradually expands and occupies the area northeast of the Halmahera Island at this time. In August–September, when the ME is the smallest north of the Talaud, the MC retroflection is suggested to retreat to a northernmost path. The seasonal variations of the ADT suggest an annual cycle movement of the MC retroflection, which is consistent with the mooring measurements above 150 m in the TH Channel (Fig. 2) and with historical surface drifter trajectories (Yuan et al. 2018).
Monthly climatological absolute dynamical topography (ADT) of AVISO data (cm).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Monthly climatological absolute dynamical topography (ADT) of AVISO data (cm).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Monthly climatological absolute dynamical topography (ADT) of AVISO data (cm).
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The ocean color representing the chlorophyll-a concentration of the surface waters suggests rich chlorophyll in the marginal seas and low chlorophyll in the open ocean. The MC retroflection produces a chlorophyll front between the tropical gyre and the Indonesian sea waters near the Talaud Islands (Fig. 10). The 1 × 10−4 g m−3 contour of chlorophyll-a concentration lies between the Talaud and Halmahera Islands in boreal winter but moves northward across the Talaud Islands in April. The low-chlorophyll water occupies the northern Maluku Sea again in November. The movement of this front suggests the same annual cycle of the MC retroflection movement as that of the mooring and satellite sea level observations.
Monthly climatological chlorophyll-a concentrations (10−4 g m−3) of OCCCI v4.2 data. The original data with the 4-km resolution have been smoothed by a 13 × 13 grid running-mean window.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Monthly climatological chlorophyll-a concentrations (10−4 g m−3) of OCCCI v4.2 data. The original data with the 4-km resolution have been smoothed by a 13 × 13 grid running-mean window.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Monthly climatological chlorophyll-a concentrations (10−4 g m−3) of OCCCI v4.2 data. The original data with the 4-km resolution have been smoothed by a 13 × 13 grid running-mean window.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
In this region, the seasonal surface wind is dominated by the Asian–Australian monsoon. However, the seasonal currents in the TH Channel are inconsistent with the monsoonal wind forcing and the surface Ekman transports, suggesting remote forcing through the strong inertia of the two western boundary currents. The MC has an annual cycle in the upper 150 m with a peak southward transport in April (Ren et al. 2018). The NGCC/UC has the maximum northwestward transport in July and the minimum in winter when the NGCC reverses directions to flow to the south in boreal winter (Ueki et al. 2003; Zhang et al. 2020). Since the transport of the MC is much larger than that of the NGCC/UC, the retroflection of the former should penetrate more deeply than the latter (Wang and Yuan 2014), which is the situation in boreal winter. As the NGCC/UC strengthens in spring and the MC peaks in April, a regime shift occurs so that the MC retroflection changes from the penetrating to the choking states of the nonlinear collision system. Both western boundary currents are observed to retroflect outside the entrance of the Indonesian seas in late spring and summer, which is consistent with the choking regime of the collision system. The existence of the Talaud Islands at the entrance of the Indonesian seas facilitates the MC intrusion in fall–winter (Mei et al. 2019). Eddies and the interocean ITF with significant seasonal variability could also impact on the movement of the MC path (Yuan et al. 2019; Song et al. 2019). These facts suggest that the dynamics of the collision of the MC and NGCC/UC are quite complicated, deserving a separate study in the future.
The satellite altimeter sea level provides a glance of the interannual variations of the MC movement. In 2015, the ME with negative sea level anomalies was sitting at the entrance of the ITF, throughout the year (Fig. 9b), indicating that the seasonal variability of the ME and HE in the far western Pacific was overwhelmed by the 2015/16 super El Niño. Kashino et al. (2013) pointed out that the anomalous sea level during El Niño is similar to that in winter in the far western Pacific, with a strong ME and a weak HE, based on composite analysis in this region. The North Pacific water intrusion in 2015 was particularly strong as observed by the CTD measurements (Figs. 7c,d), suggesting the strong influence of the 2015/16 super El Niño on the penetration of the MC retroflection into the northern Maluku Sea. The extraordinarily penetrating MC retroflection and the associated strong ME anomalies laid the ground for the NECC interannual shift and the large-amplitude rebound of sea level near Palau induced by downwelling Rossby waves in the aftermath of El Niño (Qiu et al. 2019). The interannual MC retroflection during warm and cold phases of ENSO and their interactions with the seasonal variations require a separate study in the future as longer time series of direct current observations are accumulated in the TH Channel.
c. Mass budget in the northern Maluku Sea
The MPIOM reproduces well the seasonal variations of the volume transport between 50 and 150 m of the TH Channel in the mooring observations, except for a simulated low transport in April, about one month earlier than the observed. The mass balance in the northern Maluku Sea is controlled by four channels, the Sangihe–Talaud Channel, Sangihe–Sulawesi Channel, Maluku Channel, and TH Channel (Fig. 11). In the Maluku Channel, the simulation shows northward/southward transports in boreal winter/summer with an annual cycle variation of nearly 2 Sv, consistent with the multiyear mooring observations (Yuan et al. 2018). The transport through the Sangihe–Sulawesi Channel does not show a remarkable seasonal variation in this layer with a relatively stable eastward transport of about 3 Sv, suggesting the longer branch of the MC retroflection via the Sulawesi Sea. In the Sangihe–Talaud Channel, where the MC retroflection enters the northern Maluku Sea, the southward transport reaches a maximum of nearly 6 Sv in boreal winter and relaxes to less than 4 Sv in April. The maximum inflows to the northern Maluku Sea occur during boreal winter both from the Sangihe–Talaud Channel and Maluku Channel, leading to the maximum outflow in the TH Channel. The MC retroflection in the Sangihe–Talaud Channel contributes about half of the outflow transport in the TH Channel. The minimum outflow through the TH Channel occurs in April and lasts till June, because of the different timing of the channel transports. From April to September, the Maluku Channel transport is southward as evidenced by the mooring observations (Yuan et al. 2018; Hu et al. 2019), suggesting that the southern Maluku Sea waters can only contribute to the outflow transport in the TH Channel in this layer in boreal winter.
Climatological surface circulation in (a) January and (b) June in MPIOM simulation. The color shadings represent the sea surface height (SSH; cm). The volume transport (Sv) along the sections of the northern Maluku Sea in (c) 50–150- and (d) 0–50-m layers. The locations of the sections are plotted in (a) and (b). The transports in section 1, 2, and 3 are positive for inflows to the northern Maluku Sea. The outflow in sections 4 and 5 is positive to the western Pacific.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Climatological surface circulation in (a) January and (b) June in MPIOM simulation. The color shadings represent the sea surface height (SSH; cm). The volume transport (Sv) along the sections of the northern Maluku Sea in (c) 50–150- and (d) 0–50-m layers. The locations of the sections are plotted in (a) and (b). The transports in section 1, 2, and 3 are positive for inflows to the northern Maluku Sea. The outflow in sections 4 and 5 is positive to the western Pacific.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
Climatological surface circulation in (a) January and (b) June in MPIOM simulation. The color shadings represent the sea surface height (SSH; cm). The volume transport (Sv) along the sections of the northern Maluku Sea in (c) 50–150- and (d) 0–50-m layers. The locations of the sections are plotted in (a) and (b). The transports in section 1, 2, and 3 are positive for inflows to the northern Maluku Sea. The outflow in sections 4 and 5 is positive to the western Pacific.
Citation: Journal of Physical Oceanography 51, 12; 10.1175/JPO-D-21-0048.1
The successful simulation of the currents between 50 and 150 m gives confidence in analyzing the circulation in the northern Maluku Sea in the upper 50 m, where the moored ADCP data are polluted by the sea surface reflection. The MC retroflection through the Sangihe–Talaud Channel has a similar but weak seasonal variability as that between 50 and 150 m. The inflow from the Sangihe–Sulawesi Channel has an annual cycle with the maximum in boreal winter. In the Maluku Channel, the annual change of the surface current is somewhat opposite to that between 50 and 150 m, with the southward/northward flow in boreal winter/summer forced by the Asian–Australian monsoon. The outflow above 50 m in the TH Channel has a similar annual cycle as that between 50 and 150 m, except for a late minimum by two months. The seasonal movement of the MC retroflection at the surface is reproduced well by the MPIOM simulations, as represented by the varying ME low sea level (Figs. 11a,b). The vertical transport is only of an order of 0.1 Sv within the box and is ignored in the mass balance analyses. It is worth mentioning that the simulated currents below about 150 m in the TH Channel are not supported by the observations (figure omitted), the diagnosis of which is not discussed here.
5. Conclusions
In this paper, the currents and water mass properties in the upper ocean of the Indonesian sea entrance are studied using measurements from three subsurface moorings deployed in the TH channel during December 2016–September 2018. The 2-yr mean currents across the mooring section are toward the Pacific Ocean in the upper 400 m, with decreasing amplitudes at depths. The mean current is found stronger in the northern channel than in the southern, with the velocity at 60 m reaching 50, 40, and 20 cm s−1, respectively, at the three mooring sites, showing the retroflection of the MC mainstream through the northern TH Channel. A mean volume transport is estimated based on the concurrent three moorings during December 2016–November 2017 to be 10.1–13.2 Sv between 60 and 300 m, flowing from the Maluku Sea to the Pacific Ocean, depending on the extrapolation schemes between the moorings and the coasts. At least 8.8 Sv of the MC retroflection enter the TH channel, consisting of the majority of the transport, with about only one-tenth of the transport from the Maluku Sea in the upper 300 m or so (Yuan et al. 2018).
The variance of the subtidal ACVs is dominated by low-frequency variability with periods longer than 150 days. The annual harmonics of the ACVs have the maximum velocity of more than 25 cm s−1 at the depth of 80–100 m with northeastward (southwestward) peaks in January (July), whereas the semiannual maximum velocity of 15 cm s−1 in the upper 150 m is weaker with the northeastward peaks in October and April. The superposition of the annual and semiannual harmonics results in stronger and shorter northeastward variations in boreal winter and weaker and longer southwestward variations in boreal summer.
The Mindanao–Talaud–Halmahera CTD sectional measurements during the mooring maintenance cruises show clearly the water masses of the North Pacific origin, characterized by the salinity maximum of the NPTW in the thermocline and the salinity minimum of the NPIW at the intermediate depths, occupying the northern channel. The presence of the boundary in the TH channel suggests that both the MC retroflection and the Indonesian sea waters supply transports to the origin of the NECC. The northeastward flow in the upper thermocline of the TH Channel carrying waters of the Indonesian seas with salinity lower than tropical waters from either hemisphere is believed to be a source of the North Pacific tropical subsurface water. Southwestward flow is found in the subsurface layer from April to July, which transports the SPTW into the northern Maluku Sea.
The seasonal variability of the upper currents in the TH Channel is suggested to be associated with the movement of the MC retroflection. The seasonal path changes of the MC can be identified in the satellite sea level and ocean color data, in which the ME with low sea level and chlorophyll-a concentrations is found to migrate into the northern Maluku Sea in boreal winter and back to north of the Talaud Islands in boreal summer. The shifting of the MC retroflection is consistent with the nonlinear collision of two western boundary currents (Wang and Yuan 2014).
A mass balance in the northern Maluku Sea is diagnosed based on an eddy-resolving MPIOM simulation, the seasonal transports of which are validated with the mooring data in the TH Channel and in the Maluku Channel by Yuan et al. (2018). The simulation suggests that the seasonality of the TH Channel transport between 50 and 150 m is controlled by the movement of the MC retroflection. It also suggests that waters from the southern Maluku Sea contribute to the transport through the TH Channel to join the NECC of the Pacific Ocean in boreal winter.
The three moorings in the TH channel have provided solid evidence, based on direct current measurements, about the path shifting of the MC, suggesting strongly nonlinear dynamics controlling the circulation at the entrance of the ITF. The results are of paramount importance for the study of Rossby wave reflection at the western boundary, which have traditionally been understood in the framework of linear dynamics, and for the interbasin exchange of the Pacific and Indian Oceans. The western boundary processes are known to play an important role in the cycling of ENSO and in the formation of the global ocean conveyer belt. The results of this study shall lay a solid ground for further investigations of the nonlinear dynamics of the ocean circulation at the Pacific entrance of the ITF.
Acknowledgments
This study is supported by NSFC (41720104008, 91858204), the National Key Research and Development Program of China (2020YFA0608800), and by NSFC (41421005 and 41876025), CAS (XDB42000000), QMSNL (2018SDKJ0104-02), and the Shandong Provincial projects (U1606402). D. Yuan is also supported by “Taishan Scholars program” of Shandong Province the “Kunpeng project” of the Ministry of Natural Resources of China. We thank the crews of R/V Baruna Jaya VIII and the technicians from RCO-LIPI for the successful cruises, especially the shipboard instrument technicians Mr. Muhadjirin and Mr. Nur Atmodjo. We also thank AVISO project, JPL, and NOAA for sharing their datasets. The sea level data are downloaded at the Copernicus Marine and Environment Monitoring Service (CMEMS) (http://www.marine.copernicus.eu/). The World Ocean Atlas 2013 version 2 (WOA13v2) climatological density profiles are downloaded at https://www.nodc.noaa.gov/OC5/woa13. The STORM simulation outputs are downloaded at https://verc.enes.org/storm. The mooring data used in this study can be accessed at the web page http://itf.qdio.ac.cn/xzlxz/.
REFERENCES
Arruda, W. Z., and D. Nof, 2003: The Mindanao and Halmahera eddies—Twin eddies induced by nonlinearities. J. Phys. Oceanogr., 33, 2815–2830, https://doi.org/10.1175/1520-0485(2003)033<2815:TMAHEE>2.0.CO;2.
Bingham, F. M., and R. Lukas, 1994: The southward intrusion of North Pacific intermediate water along the Mindanao coast. J. Phys. Oceanogr., 24, 141–154, https://doi.org/10.1175/1520-0485(1994)024<0141:TSIONP>2.0.CO;2.
Cowley, R., B. Heaney, S. Wijffels, L. Pender, J. Sprintall, S. Kawamoto, and R. Molcard, 2008: INSTANT Sunda data report description and quality control. CSIRO, 288 pp., http://www.marine.csiro.au/~cow074/INSTANTdataQC_v4.pdf.
Egbert, G. D., and S. Y. Erofeeva, 2002: Efficient inverse modeling of barotropic ocean tides. J. Atmos. Oceanic Technol., 19, 183–204, https://doi.org/10.1175/1520-0426(2002)019<0183:EIMOBO>2.0.CO;2.
Fang, G., and Coauthors, 2010: Volume, heat, and freshwater transports from the South China Sea to Indonesian seas in the boreal winter of 2007–2008. J. Geophys. Res., 115, C12020, https://doi.org/10.1029/2010JC006225.
Ffield, A., and A. L. Gordon, 1996: Tide mixing signatures in the Indonesian Seas. J. Phys. Oceanogr., 26, 1924–1937, https://doi.org/10.1175/1520-0485(1996)026<1924:TMSITI>2.0.CO;2.
Fine, R., R. Lukas, F. M. Bingham, M. J. Warner, and R. H. Gammon, 1994: The western equatorial Pacific: A water mass crossroads. J. Geophys. Res., 99, 25 063–25 080, https://doi.org/10.1029/94JC02277.
Gordon, A. L., 2005: Oceanography of the Indonesian seas and their throughflow. Oceanography, 18, 14–27, https://doi.org/10.5670/oceanog.2005.01.
Gordon, A. L., and R. Fine, 1996: Pathways of water between the Pacific and Indian Oceans in the Indonesian seas. Nature, 379, 146–149, https://doi.org/10.1038/379146a0.
Gordon, A. L., R. D. Susanto, and A. Ffield, 1999: Throughflow within Makassar Strait. Geophys. Res. Lett., 26, 3325–3328, https://doi.org/10.1029/1999GL002340.
Gordon, A. L., R. D. Susanto, A. Ffield, B. A. Huber, W. Pranowo, and S. Wirasantosa, 2008: Makassar Strait throughflow, 2004 to 2006. Geophys. Res. Lett., 35, L24605, https://doi.org/10.1029/2008GL036372.
Gordon, A. L., B. A. Huber, E. J. Metzger, R. D. Susanto, H. E. Hurlbut, and T. R. Adi, 2012: South China Sea throughflow impact on the Indonesian throughflow. Geophys. Res. Lett., 39, L11602, https://doi.org/10.1029/2012GL052021.
Gordon, A. L., and Coauthors, 2019: Makassar Strait through flow seasonal and interannual variability: An overview. J. Geophys. Res. Oceans, 124, 3724–3736, https://doi.org/10.1029/2018JC014502.
Hu, X., and Coauthors, 2019: Interannual variability of the Sulawesi Sea circulation forced by Indo-Pacific planetary waves. J. Geophys. Res. Oceans, 124, 1616–1633, https://doi.org/10.1029/2018JC014356.
Ilahude, A. G., and A. L. Gordon, 1996: Thermohaline stratification within the Indonesian Seas. J. Geophys. Res., 101, 12 401–12 409, https://doi.org/10.1029/95JC03798.
Kashino, Y., M. Aoyoma, T. Kawano, N. Hendiarti, Syaefudin, Y. Anantasena, K. Muneyama, and H. Watanabe, 1996: The water masses between Mindanao and New Guinea. J. Geophys. Res., 101, 12 391–12 400, https://doi.org/10.1029/95JC03797.
Kashino, Y., H. Watanabe, B. Herunadi, M. Aoyama, and D. Hartoyo, 1999: Current variability at the Pacific entrance of the Indonesian throughflow. J. Geophys. Res., 104, 11 021–11 035, https://doi.org/10.1029/1999JC900033.
Kashino, Y., E. Firing, P. Hacker, A. Sulaiman, and Lukiyanto, 2001: Currents in the Celebes and Maluku Seas, February 1999. Geophys. Res. Lett., 28, 1263–1266, https://doi.org/10.1029/2000GL011630.
Kashino, Y., A. Atmadipoera, Y. Kuroda, and Lukijiano, 2013: Observed features of the Halmahera and Mindanao eddies. J. Geophys. Res. Oceans, 118, 6543–6560, https://doi.org/10.1002/2013JC009207.
Kuehl, J., and V. A. Sheremet, 2009: Identification of a cusp catastrophe in a gap-leaping western boundary current. J. Mar. Res., 67, 25–42, https://doi.org/10.1357/002224009788597908.
Li, B., D. Yuan, and H. Zhou, 2018: Water masses in the far western equatorial Pacific during the winters of 2010 and 2012. J. Oceanol. Limnol., 36, 1459–1474, https://doi.org/10.1007/s00343-018-6068-2.
Li, X., and Coauthors, 2020: Moored observations of transport and variability of Halmahera Sea currents. J. Phys. Oceanogr., 50, 471–488, https://doi.org/10.1175/JPO-D-19-0109.1.
Lindstrom, E., R. Lukas, R. Fine, E. Firing, S. J. Godfrey, G. Meyers, and M. Tsuchiya, 1987: The western equatorial Pacific Ocean circulation study. Nature, 330, 533–537, https://doi.org/10.1038/330533a0.
Luick, J. L., and G. R. Cresswell, 2001: Current measurements in the Maluku Sea. J. Geophys. Res., 106, 13 953–13 958, https://doi.org/10.1029/2000JC000694.
Lukas, R., E. Firing, P. Hacker, P. L. Richardson, C. A. Collins, R. Fine, and R. Gammon, 1991: Observations of the Mindanao current during the western equatorial Pacific Ocean circulation study. J. Geophys. Res., 96, 7089–7104, https://doi.org/10.1029/91JC00062.
Mei, H., Y. Qi, B. Qiu, X. Cheng, and X. Wu, 2019: Influence of an island on hysteresis of a western boundary current flowing across a gap. J. Phys. Oceanogr., 49, 1353–1366, https://doi.org/10.1175/JPO-D-18-0116.1.
Qiu, B., M. Mao, and Y. Kashino, 1999: Intraseasonal variability in the Indo-Pacific throughflow and the regions surrounding the Indonesian seas. J. Phys. Oceanogr., 29, 1599–1618, https://doi.org/10.1175/1520-0485(1999)029<1599:IVITIP>2.0.CO;2.
Qiu, B., S. Chen, B. Powell, P. Colin, D. Rudnick, and M. Schönau, 2019: Nonlinear short-term upper ocean circulation variability. Oceanography, 32, 22–31, https://doi.org/10.5670/oceanog.2019.408.
Qu, T., H. Mitsudera, and T. Yamagata, 1999: A climatology of the circulation and water mass distribution near the Philippine coast. J. Phys. Oceanogr., 29, 1488–1505, https://doi.org/10.1175/1520-0485(1999)029<1488:ACOTCA>2.0.CO;2.
Ren, Q., Y. Li, F. Wang, L. Song, C. Liu, and F. Zhai, 2018: Seasonality of the Mindanao Current/Undercurrent system. J. Geophys. Res. Oceans, 123, 1105–1122, https://doi.org/10.1002/2017JC013474.
Sathyendranath, S., and Coauthors, 2019: An ocean-colour time series for use in climate studies: The experience of the Ocean-Colour Climate Change Initiative (OC-CCI). Sensors, 19, 4285, https://doi.org/10.3390/s19194285.
Sheremet, V. A., 2001: Hysteresis of a western boundary current leaping across a gap. J. Phys. Oceanogr., 31, 1247–1259, https://doi.org/10.1175/1520-0485(2001)031<1247:HOAWBC>2.0.CO;2.
Song, X., D. Yuan, and Z. Wang, 2019: Hysteresis of a periodic or leaking western boundary current flowing by a gap. Acta Oceanol. Sin., 38, 90–96, https://doi.org/10.1007/s13131-018-1251-z.
Susanto, R. D., and Y. T. Song, 2015: Indonesian throughflow proxy from satellite altimeters and gravimeters. J. Geophys. Res. Oceans, 120, 2844–2855, https://doi.org/10.1002/2014JC010382.
Susanto, R. D., G. Fang, I. Soesilo, Q. Zheng, F. Qiao, Z. Wei, and B. Sulistyo, 2010: New surveys of a branch of the Indonesian throughflow. Eos, Trans. Amer. Geophys. Union, 91, 261–263, https://doi.org/10.1029/2010EO300002.
Susanto, R. D., A. Ffield, A. Gordon, and R. Adi, 2012: Variability of Indonesian throughflow within Makassar Strait, 2004–2009. J. Geophys. Res., 117, C09013, https://doi.org/10.1029/2012JC008096.
Talley, L., 1993: Distribution and formation of North Pacific Intermediate Water. J. Phys. Oceanogr., 23, 517–537, https://doi.org/10.1175/1520-0485(1993)023<0517:DAFONP>2.0.CO;2.
Tsuchiya, M., 1991: Flow path of Antarctic intermediate water in the western equatorial South Pacific Ocean. Deep-Sea Res. I, 38A, S273–S279, https://doi.org/10.1016/S0198-0149(12)80013-6.
Ueki, I., Y. Kashino, and Y. Kuroda, 2003: Observation of current variations off the New Guinea coast including the 1997–1998 El Niño period and their relationship with Sverdrup transport. J. Geophys. Res., 108, 3243, https://doi.org/10.1029/2002JC001611.
Von Storch, J.-S., C. Eden, I. Fast, H. Haak, D. Hernández-Deckers, E. Maier-Reimer, J. Marotzke, and D. Stammer, 2012: An estimate of the Lorenz energy cycle for the world ocean based on the STORM/NCEP simulation. J. Phys. Oceanogr., 42, 2185–2205, https://doi.org/10.1175/JPO-D-12-079.1.
Von Storch, J.-S., H. Haak, E. Hertwig, and I. Fast, 2016: Vertical heat and salt fluxes due to resolved and parameterized meso-scale eddies. Ocean Modell., 108, 1–19, https://doi.org/10.1016/j.ocemod.2016.10.001.
Wang, Z., and D. Yuan, 2012: Nonlinear dynamics of two western boundary currents colliding at a gap. J. Phys. Oceanogr., 42, 2030–2040, https://doi.org/10.1175/JPO-D-12-05.1.
Wang, Z., and D. Yuan, 2014: Multiple equilibria and hysteresis of two unequal-transport western boundary currents colliding at a gap. J. Phys. Oceanogr., 44, 1873–1885, https://doi.org/10.1175/JPO-D-13-0234.1.
Wang, Z., D. Yuan, and Y. Hou, 2010: Effect of meridional wind on gap-leaping western boundary current. Chin. J. Oceanol. Limnol., 28, 354–358, https://doi.org/10.1007/s00343-010-9281-1.
Wang, F., Y. Li, Y. Zhang, and D. Hu, 2013: The subsurface water in the North Pacific tropical gyre. Deep-Sea Res., 75, 78–92, https://doi.org/10.1016/j.dsr.2013.01.002.
Yuan, D., and R. Li, 2008: Dynamics of eddy-induced Kuroshio variability in Luzon Strait (in Chinese). J. Trop. Oceanogr., 27, 1–9.
Yuan, D., and Z. Wang, 2011: Hysteresis and dynamics of a western boundary current flowing by a gap forced by impingement of mesoscale eddies. J. Phys. Oceanogr., 41, 878–888, https://doi.org/10.1175/2010JPO4489.1.
Yuan, D., and Coauthors, 2018: Observed transport variations in the Maluku Channel of the Indonesian Seas associated with western boundary current changes. J. Phys. Oceanogr., 48, 1803–1813, https://doi.org/10.1175/JPO-D-17-0120.1.
Yuan, D., X. Song, Y. Yang, and W. Dewar, 2019: Dynamics of mesoscale eddies interacting with a western boundary current flowing by a gap. J. Geophys. Res. Oceans, 124, 4117–4132, https://doi.org/10.1029/2019JC014949.
Zhang, L., J. Wu, F. Wang, S. Hu, Q. Wang, F. Jia, F. Wang, and D. Hu, 2020: Seasonal and interannual variability of the currents off the New Guinea coast from mooring measurements. J. Geophys. Res. Oceans, 125, e2020JC016242, https://doi.org/10.1029/2020JC016242.
