South China Sea throughflow: A heat and freshwater conveyor

Authors


Abstract

[1] Analysis of surface flux data suggests that the South China Sea throughflow is a conveyor belt transferring up to 0.2 PW (1 PW = 1 × 1015W) of heat and 0.1 Sv (1 Sv = 1 × 106 m3 s−1) of freshwater from the South China Sea into the Indonesian maritime continent. As surface heat and freshwater fluxes display substantially different temporal variations with the South China Sea throughflow, we hypothesize that the South China Sea acts as a heat capacitor, storing heat in certain years and releasing it in others. Results from a high-resolution general circulation model confirm this hypothesis, implying that the South China Sea is likely to play a more active role than previously thought in regulating the sea surface temperature pattern in the Indonesian maritime continent and its adjoining western Pacific and eastern Indian Oceans.

1. Introduction

[2] The Indonesian maritime continent, situated at the confluence of the tropical Pacific and Indian Ocean, is an area of major climatic importance. The region, along with equatorial Africa and South America, is recognized as a site of vigorous atmospheric convection [Ramage, 1968], where interaction between the atmosphere and the ocean's mixed layer spawns the El Niño Southern Oscillation (ENSO) phenomena. The sea surface temperature (SST) in the region is variable on a range of spatial and temporal scales. Earlier studies have shown that small changes of SST in this region can result in significant variations in atmospheric convection and precipitation across the Indo-Pacific basin [e.g., Neale and Slingo, 2003; McBride et al., 2003]. The SST in the maritime continent is therefore important to study for both its regional and global significance in climate variability.

[3] A potentially important process that influences the SST in the maritime continent is the South China Sea (SCS) throughflow (SCSTF) (Figure 1), which involves the inflow through the Luzon Strait and the outflow through the Karimata, Mindoro, and Taiwan Straits [Qu et al., 2005; Fang et al., 2005; Wang et al., 2006; Song, 2006; Yu et al., 2006]. On the basin average, the SCS receives heat from the atmosphere at a rate that has widely ranged (10 to 50 W m−2) in the past. For example, estimates (∼49 W m−2) from OAFlux [Yu and Weller, 2006] are nearly twice as high as the estimates (∼23 W m−2) from the Comprehensive Ocean-Atmosphere Data Set [Oberhuber, 1988]. The NCEP re-analysis (Figure 2a) favors a value in the middle of the range (∼18 W m−2). Given a surface area of 3.7 × 1012 m2 in the SCS, this implies a net heat gain of 0.1–0.2 PW (1 PW = 1 × 1015 W) over the basin. A question that may arise immediately is where this heat goes. For the long-term mean it can only be balanced by horizontal advection, with water entering the SCS through the Luzon Strait significantly cooler than that leaving it through the Karimata and Mindoro Straits [Qu et al., 2004]. Here we hypothesize that the SCSTF is a heat conveyor, transferring heat from the SCS into the maritime continent.

Figure 1.

A schematic diagram showing the South China Sea throughflow adopted from Qu et al. [2005]. Water entering the South China Sea through Luzon Strait is lower in temperature (blue) than water leaving it through Karimata, Mindoro, and Taiwan Strait (red).

Figure 2.

Annual mean (a) surface heat flux (W m−2) and (b) fresh water flux (mm/day) from NCEP.

[4] Examination of CMAP (Climate Prediction Center Merged Analysis of Precipitation), GPCP (Global Precipitation Climatology Project), and TRMM (Tropical Rain Measuring Mission) precipitation data shows that the SCS is also a recipient of heavy rainfall, with an annual mean value of 0.2–0.3 Sv (1 Sv = 1 × 106 m3 s−1) over the basin. An accurate estimate of evaporation from observations is not available. The NCEP re-analysis product shows that precipitation in the SCS exceeds evaporation (P-E) by about 0.1 Sv (Figure 2b). To achieve salinity balance, the vapor gain from the atmosphere must be balanced by the import of salt via the SCSTF. As its upper-limb water through the Karimata and Mindoro Straits is fresher than its lower-limb water through the Luzon Strait, the SCSTF is also a freshwater conveyor transferring freshwater from the SCS into the maritime continent.

[5] The freshwater transport via the SCSTF further influences the heat budget in the maritime continent and its adjoining western Pacific and eastern Indian Oceans. Earlier studies have shown that the intrusion of freshwater from the SCS effectively inhibits the Makassar Strait surface water from freely flowing southward and as a consequence, the Indonesian throughflow (ITF) heat transport is significantly reduced during the northeast monsoon season [Gordon et al., 2003]. Similarly, when the intrusion of freshwater from the SCS reaches its maximum strength during the mature phase of El Niño, it enhances the northward-directed pressure gradient in the Makassar Strait, thus generating an anomalous northward flow in the surface layer of the strait [Qu et al., 2005].

[6] For the long-term mean the SCSTF conveyor is primarily balancing the surface heat and freshwater fluxes, so as to establish a stable thermocline and halocline in the SCS. This delicate balance, however, cannot always be maintained at interannual time scales. With substantially different temporal variations in surface fluxes and SCSTF, the SCS heat content varies greatly from year to year. We thereby hypothesize that the SCS acts as a heat capacitor, storing heat in certain years and releasing it in others. This implies that the SCS is likely to play a more active role than previously thought in regulating the SST pattern and modulating conditions in the maritime continent and its adjoining western Pacific and eastern Indian Oceans.

2. Data and Methods of Analysis

[7] To test the hypotheses described above, we analyze results from the high-resolution Ocean General Circulation Model (OGCM) for the Earth Simulator (OFES). Here we briefly describe the model configuration. See Sasaki et al. [2004, 2007] for more details. The OFES was based on the Modular Ocean Model (MOM3). Its domain covers a near-global region extending from 75°S to 75°N, with a horizontal resolution of 0.1 degree. The vertical resolution varies from 5 m near the surface to 330 m near the bottom, with a total of 54 levels and a maximum depth of 6065 m. The model topography was constructed from the 1/30° bathymetry dataset created at the Southampton Oceanography Center.

[8] A 50-year climatological spin-up was first executed from annual mean temperature and salinity fields of the World Ocean Atlas 1998 (WOA98) with no motion. Then, a hindcast integration from 1950 to 2003 was conducted. The surface fluxes were specified from NCEP re-analysis, with monthly mean data for the spin-up run and daily mean data for the hindcast run, in addition to a surface salinity restoring to the climatological value of WOA98. To suppress grid-scale noises, a scale-selective damping of bi-harmonic operator was adopted for horizontal mixing, and the K-Profile Parameterization (KPP) scheme was employed for the vertical mixing. Results from the 54-year hindcast run are presented in the following sections.

3. Results

[9] With its high resolution and realistic topography, the OFES was able to reproduce most, if not all, of the detailed phenomena observed in the global ocean [Masumoto et al., 2004]. Particular attention has been given to the SCS and the Indonesian throughflow region, suggesting that the circulation from OFES is one of the best available in a model [Du et al., 2005]. The SCSTF from OFES has an annual mean value of 3.8 Sv, being allocated almost equally to the Karimata, Mindoro, and Taiwan Straits. Its interannual variation shows essentially the same pattern as described by Qu et al. [2005], though the time series used in the present study is about three tomes longer. On this time scale, variation in the Taiwan Strait is negligibly small, and nearly all the SCSTF variation occurs in the Karimata and Mindoro Straits (Figure 3a). In general, the SCSTF gets stronger during El Niño years and weaker during La Niña years. Its temporal correlation with the Southern Oscillation Index (SOI) reaches −0.42, with the maximum transport leading the mature phase of El Niño by about 6 months. This variation is well out of phase with the ITF, and has been shown to be closely related to the bifurcation of the North Equatorial Current in the Pacific [Qu et al., 2005].

Figure 3.

(a) Volume (Sv) and (b) heat (1014W) and freshwater (Sv) transport of the SCSTF about their long-term mean values of 3.8 Sv, 0.5 × 1014W, and 0.08 Sv, respectively, compared with SOI and transport through Karimata and Mindoro Straits. The 13-month mean filter has been applied twice to remove the mean seasonal cycle. The correlation between the SCSTF volume transport and SOI is −0.42, with the former leading by about 6 months, and the instantaneous correlation between the SCSTF volume and heat (freshwater) transport is 0.98 (0.83).

3.1. Heat Conveyor

[10] The SCSTF heat transport is calculated as Cpρequation imagedA, where Cpρ is the specific heat capacity per unit volume, and L represents the lateral boundary of the SCS, consisting of Luzon, Karimata, Mindoro, and Taiwan Straits. dA is a two dimensional element of area, T is temperature, and vn is the velocity normal to the lateral boundary, with positive values indicating outward flow. The overbar represents the ensemble monthly mean values, simultaneously ignoring small-scale (<1 month) eddy heat flux and the sub-grid scale mixing.

[11] The flow through the Karimata, Mindoro, and Taiwan Straits is shallower, and on the annual average its transport-weighted temperature is warmer by about 1.8°C than that through the Luzon Strait. With a mean volume transport of 3.8 Sv, this temperature difference implies a cooling advection of 0.05 PW, balancing a large fraction (∼65%) of the surface heat flux from the atmosphere (0.08 PW). Note that, averaged over the entire South China Sea, the NCEP re-analysis heat flux (18 W m−2) used to force the OFES is smaller by a factor 2.7 than the newly released OAFlux product [Yu and Weller, 2006]. For this concern, the SCSTF heat conveyor from the OFES might have been underestimated.

[12] The heat conveyor varies on interannual time scales (Figure 3b). Its correlation with SOI reaches −0.48, with the maximum heat transport occurring about 4 months prior to the mature phase of El Niño. A comparison between the SCSTF volume and heat transport indicates that they are nearly perfectly correlated (r = 0.98). This seems to suggest that, despite variations in sea surface temperature associated with ENSO [e.g., Wang et al., 2000], the heat conveyor is primarily controlled by the SCSTF.

3.2. Freshwater Conveyor

[13] The SCSTF freshwater transport is calculated in the same way as that for the heat transport. In the model, the annual mean transport-weighted salinity is 34.6 psu in the Luzon Strait, significantly higher than that in the Karimata (33.3 psu), Mindoro (34.4 psu), and Taiwan Straits (34.3 psu). With a volume transport of 3.8 Sv, this salinity difference implies a salt import of 2.8 × 106 kg s−1, or equivalently, a freshwater export of 0.08 Sv, based on a standard sea water salinity of 35 psu. The freshwater conveyor, on the annual average, balances most (>90%) of the excess of precipitation over evaporation (0.09 Sv) at the surface. This small amount of freshwater flux drives a circulation (i.e., the SCSTF) that is about fifty times stronger than the forcing itself.

[14] The freshwater conveyor also varies on interannual time scales (Figure 3b), and its correlation with SOI reaches −0.41, with the maximum freshwater transport occurring about 4 months prior to the mature phase of El Niño. Most of this interannual variation is due to the circulation associated with SCSTF. The temporal correlation between the SCSTF volume and freshwater transport exceeds 0.83 (Figure 3), implying an essentially important role of the conveyor in the SCS freshwater budget.

[15] Note that the surface freshwater flux and salinity distribution in the model may contain large uncertainties. For one reason, the river runoff is not included in the model simply because the data are not ordinarily available. For another, as in most existing ocean models, the freshwater flux in OFES is applied as a salt flux, based on the Boussineq assumptions. This boundary condition, together with the relaxation of sea surface salinity, may introduce systematic errors in the salinity budget [Huang, 1993], and to overcome this difficulty some non-Boussineq approaches need to be considered [Song and Hou, 2006]. Given these facts of the model, we focus our following analysis on the heat budget. We leave the freshwater budget analysis for future studies.

3.3. Upper Layer Heat Budget

[16] Averaged over the entire SCS, the NCEP surface heat flux shows a good correspondence (r = −0.67) with SOI (Figure 4a). During El Niño years, the SCS receives more heat from the atmosphere than in other years, with its maximum leading the mature phase of El Niño by about 2 months. The newly released OAFlux product [Yu and Weller, 2006] shows essentially the same phase for the period from 1984 through 2002, except for a significant difference in 1993–94.

Figure 4.

(a) OFES surface heat flux compared with OAFlux surface heat flux and Southern Oscillation Index (SOI), and (b) upper layer (0–432 m) heat content change (HCC) compared with heat advection and the sum of surface heat flux and heat advection averaged over the South China Sea (SCS). Here, heat advection represents heat convergence in the SCS and has an opposite sign to the SCSTF heat transport shown in Figure 3b. The 13-month mean filter has been applied twice to remove the mean seasonal cycle. Unit is 1014 W. The instantaneous correlation between HCC and surface heat flux is 0.33 and between HCC and the SCSTF heat transport is −0.70.

[17] The heat conveyor is well in phase (r = 0.60) with surface heat flux (Figure 4b). But, differences still exists between them, and as a consequence, the upper (0–432 m) SCS heat content varies from year to year. Most of the upper SCS heat content change is due to heat export of the conveyor, and the temporal correlation between the two reaches −0.7 (Figure 4b). During El Niño years, the heat export of the conveyor approaches its maximum strength, and the SCS loses more heat than it receives, despite the enhanced surface heat flux (Figure 4a). There are intriguing exceptions to this trend, though. During the two super El Niños of the last century, the 1982–83 and the 1997–98, the surface heat flux into the SCS was stronger than the heat export of the conveyor, increasing the upper SCS heat content. As such, the model has clearly demonstrated that the SCS acts as a heat capacitor, storing heat in certain years and releasing it in others.

4. Discussion

[18] Results from the OFES confirm the hypothesis that the SCSTF is a heat and freshwater conveyor. A large part of the conveyor appears to be driven by the deepwater overflow through Luzon Strait. Recent studies have shown that below about 1500 m there is a persistent baroclinic pressure gradient driving flow from the Pacific into the SCS [Qu et al., 2006]. The low-temperature, high-salinity Pacific water sinks after crossing the Luzon Strait, and to achieve a mass balance, upwelling must occur somewhere else. Here, we emphasize that a tidal mixing as large as 10−3 m2 s−1 is a key process to maintain the heat and freshwater conveyor. Because of the intense tidal mixing, water of the Pacific origin upwells and leaves the SCS through a number of shallow passages to form the upper limb of the conveyor.

[19] Results from the OFES also confirm the hypothesis that the SCS acts as a heat capacitor. In most cases, the upper SCS receives an excess of heat during La Niña years and releases it during El Niño years, primarily as a result of heat export by the heat conveyor. Intriguing exceptions exist, and need to be investigated further by research. With a total heat transport of up to 0.2 PW, the heat conveyor is likely to play a more active role than previously thought in regulating the SST pattern and modulating conditions in the Indonesian maritime continent and its adjoining western Pacific and eastern Indian Oceans. Further experiments with atmospheric and oceanic models are apparently needed to understand its full implication for climate variability.

Acknowledgments

[20] This research was supported by the National Aeronautics and Space Administration through grant NAG5-12756 and by Japan Agency for Marine-Earth Science and Technology (JAMSTEC) through its sponsorship of the International Pacific Research Center (IPRC). Support is also from NSF-China through grants 40406006 and 40576013. The OFES simulation was conducted at the Earth Simulator under the support of JAMSTEC. Thanks are extended to T. Jensen, T. Song, D. X. Wang, and members of the IPRC South China Sea working group for many useful discussions, and to G. E. Speidel and two anonymous reviewers for thoughtful comments on the earlier manuscript. School of Ocean and Earth Science and Technology contribution 7014, and IPRC contribution IPRC-423.

Ancillary

Advertisement