1 Introduction

Changes in the strength or water mass properties of the meridional circulation associated with the subtropical cells (STCs; McCreary and Lu 1994) are hypothesized to influence the Pacific decadal variability. Notably, the connections between the subtropics and the tropics through the western boundary route of the South Pacific are thought to play a major role in El Niño Southern Oscillation (ENSO) low frequency modulation (Luo and Yamagata 2001; Luo et al. 2003, 2005; Giese et al. 2002; Bratcher and Giese 2002; Chang et al. 2001; Holland and Raphael 2006; Nonaka and Sasaki 2007).

The South Pacific low-latitude western boundary currents (LLWBCs), carrying waters of subtropical origin, transit through the Solomon Sea before joining the Equatorial Pacific. This semienclosed basin is located north of the Coral Sea and is bounded on the west by Papua New Guinea, on the east by the Solomon Island chain, and on the north by the island of New Britain (NB; Fig. 1). A double system of LLWBCs (Melet et al. 2010a) connects subtropical water to the EUC as part of the STC circulation in the Solomon Sea region through three narrow straits (Vitiaz Strait, St. George’s Channel, and Solomon Strait, Fig. 1).

Fig. 1
figure 1

Bathymetry of the Solomon Sea (in m). The Solomon Sea is bounded on the west by the island of Papua New Guinea, on the north by the island of New Britain (NB), and on the east by the Solomon Islands chain. On the south, it is open to the Coral Sea. On the north, three straits connect the Solomon Sea to the equatorial Pacific: Vitiaz Strait, St Georges Channel, and Solomon Strait. GPC Gulf of Papua Current, NVJ North Vanuatu Jet, NGCU New Guinea Coastal Undercurrent, SEC South Equatorial Current, NB New Britain

Full size image

At thermocline level, the LLWBCs of the Solomon Sea represent a major source for the Equatorial Undercurrent (EUC), as shown both by modeling (e.g., Grenier et al. 2011) and observational studies (e.g., Tsuchiya et al. 1989). The upwelling of EUC water along its eastward route directly influences the equatorial Pacific sea surface temperature (SST), which in turn determines the location of deep atmospheric convection and influences the development of El Niño events (e.g., Hoerling and Zhong 1997; Kang and Kug 2002; Guilyardi et al. 2009).

In the surface layers, the LLWBCs of the Solomon Sea contribute to the variability of the warm water volume (WWV, defined as the volume of equatorial Pacific water above the 20 °C isotherm between 5° N and 5° S). This variability impacts the filling and draining of the warm pool at ENSO timescales (Kug et al. 2003; Ishida et al. 2008; Melet et al. 2010b; Lengaigne et al. 2012), which has long been suggested as a key ingredient in ENSO dynamics (e.g., Springer et al. 1990; Jin 1997; Meinen and McPhaden 2001).

Therefore, changes in the properties or transport of the LLWBCs of the Solomon Sea could impact the WWV and the EUC properties, with potential impacts on the Pacific climate and the low-frequency modulation of ENSO (Kleeman et al. 1999; Schneider 2004; Luo et al. 2005). Because of its potential importance in understanding climate variability, the Solomon Sea is being investigated through both modeling and in situ monitoring as part of the CLIVAR/Southwest Pacific Ocean and Climate Experiment (SPICE, Ganachaud et al. 2007) that frames this study.

Recently, the mean circulation in the Solomon Sea and its seasonal cycle have been described both in modeling (Melet et al. 2010a) and observational studies (Cravatte et al. 2011; Hristova and Kessler 2012). The complex bathymetry of the region is responsible for a double system of LLWBCs; the New Guinea Coastal Undercurrent (NGCU, Fig. 1) inflow is divided in different branches that join the equatorial Pacific through different pathways. At interannual timescales, altimetric data show that the Solomon Sea is a region of high variability in relation to ENSO (e.g., Melet et al. 2010b). Strong interannual sea level anomalies (SLAs) are generated by large wind stress curl anomalies that develop in the western and central south tropical Pacific a few months after ENSO’s peak. These anomalies are mainly related to displacements of the South Pacific Convergence Zone (SPCZ, see Vincent 1994 for a review): northeastward during El Niño events and southwestward during La Niña events (e.g., Folland et al. 2002). Figure 2 illustrates such positive wind stress curl anomalies during La Niña and how this signal in the southwest tropical Pacific is related to ENSO. The correlation between the corresponding wind stress curl and the Southern Oscillation Index [SOI, positive (negative) SOI is for La Niña (El Niño)] is 0.95 considering a 1–2 month lag. These large-scale wind stress changes control the large-scale response of the South Pacific LLWBCs to ENSO. In the Solomon Sea, a few months after the peak El Niño winds, the LLWBC transport increases, whereas the circulation weakens during La Niña events. More specifically, Ridgway et al. (1993) estimated an increase of 15 Sv of the Solomon Sea LLWBCs transport through the upper 150 m during the 1986–1987 El Niño event based on tide gauges and XBT data. Moreover, Melet et al. (2010b) estimated interannual variations of the transport of about 10 Sv using altimetric data. They showed that these LLWBCs transport anomalies counterbalance changes in the western Pacific WWV at ENSO timescale. This phasing of the South Pacific LLWBCs to ENSO characterized by LLWBC anomalies in phase opposition with the interior geostrophic flow was suspected in previous modeling studies (Kug et al. 2003; Lee and Fukumori 2003; Ishida et al. 2008; Lengaigne et al. 2012) and was recently confirmed by observational studies (Cravatte et al. 2011; Hristova and Kessler 2012; Davis et al. 2012). Therefore, if recent studies have highlighted the complexity of the circulation in the Solomon Sea at climatologic and seasonal timescales, our current state of knowledge at interannual timescales solely relates to the large-scale response of the South Pacific LLWBCs to ENSO.

Fig. 2
figure 2

Top Wind stress curl anomalies (color, in 10−8 N m−3) and wind stress anomalies (vectors, N m−2) for La Niña composite state (see Section 2.2) in the Tropical Pacific Ocean as computed from the model forcing data set. Bottom Time series of the wind stress curl (black line, in 10−8 N/m3) averaged over the box defined in the top figure and of the SOI (centered on −2, red line). Data have been low pass filtered with a half power at 18 months

Full size image

The local response of the LLWBCs to ENSO in the Solomon Sea, beyond their large-scale strengthening/weakening, remains largely unknown. It could be regionally more complex for two main reasons. First, the intricate bathymetry of the semienclosed Solomon Sea induces different pathways for the subtropical waters to join the equatorial Pacific. Whereas Vitiaz Strait is the main pathway for the NGCU to exit the Solomon Sea, analysis of altimetric data showed that the highest transport anomalies are observed at Solomon Strait (Melet et al. 2010b). Therefore, these pathways are likely to be impacted differently by ENSO, with consequences on the routes followed by Solomon Sea water to join the equatorial Pacific. Second, the Solomon Sea is a region of important mixing that impacts the thermohaline characteristics of the water masses advected by the LLWBCs. Climatologically, the salinity maximum found in the upper thermocline and characteristic of the Subtropical Mode Water (STMW; Sokolov and Rintoul 2000) is eroded while a downward heat transfer occurs (Melet et al. 2011). The changes of circulation at ENSO timescales could impact these changes in water mass properties, with possible consequences on the heat and salt fluxes toward the equatorial region downstream of the Solomon Sea. ENSO-related changes of the different pathways and of their respective heat/salt fluxes could in return have different impacts on the WWV, the EUC, the equatorial Pacific SST, and ultimately on ENSO. Thus, the main question addressed in this study is: How does ENSO variability impact the circulation and water mass properties in the Solomon Sea?

This study is based on the analysis of the Solomon Sea interannual variability from a numerical simulation. The model and methodology are described in Section 2. ENSO-related changes in the Solomon Sea circulation are detailed in Section 3, and changes in water mass properties are analyzed in Section 4. A summary and conclusions are presented in Section 5.

2 Methodology

2.1 Model

2.1.1 Description

Model configurations are described in detail in Melet et al. (2010a, 2011); we only review their main characteristics in this section. The high-resolution (1/12°) primitive equations model of the Solomon Sea is interactively nested in a 1/4° model of the tropical southwest Pacific, which itself is embedded offline in a 1/4° OGCM through open boundaries. All three models are based on the oceanic component of the Nucleus for European Modelling of the Ocean (NEMO) system (Madec 2008). The vertical coordinate is geopotential, with 46 levels. The bathymetry of the 1/12° model has been carefully checked to ensure that it is realistic.

The diffusion of tracers is performed along isopycnal surfaces through a Laplacian operator, whereas the dissipation of momentum is achieved through a bi-Laplacian operator along geopotential surfaces. The vertical eddy viscosity and diffusivity coefficients are computed from the TKE turbulent closure scheme (Blanke and Delecluse 1993), with background diffusivity coefficients of 0.1 cm2 s−1 for tracers. In addition, tidal mixing is parameterized. Inside the Solomon Sea, we applied the formulation of Koch-Larrouy et al. 2007: Internal tides are assumed to entirely dissipate locally, and the vertical distribution of the dissipation reaches a maximum in the thermocline (Melet et al. 2011). Outside the Solomon Sea, we applied the formulation proposed by Jayne and St Laurent 2001 and introduced in NEMO by Bessières et al. 2008.

The regional and fine-grid models are forced with the same datasets as the global model. Atmospheric fluxes of momentum, heat, and freshwater are diagnosed through classical bulk formulae (Large and Yeager 2009). Realistic winds, atmospheric temperature and humidity are provided by the 40-year European Centre for Medium-Range Weather Forecasts Re-analysis (ERA-40). Precipitation and shortwave, longwave radiations are provided by the Common Ocean-Ice Reference Experiments dataset (Large and Yeager 2009). The regional models were initialized with the climatological mass field of Levitus et al. 1998 and were integrated over the 1984–2006 period, with daily outputs. An initial period of 2 years was used for the adjustment of the model. Thus, the 1986–2006 period was analyzed in this study.

2.1.2 Validation

The equatorial Pacific state of the global model has been validated in Grenier et al. (2011) and Lengaigne et al. (2012). These authors show that the global model realistically captures the mean state and annual to interannual variability of the EUC, WWV, and SST in the equatorial Pacific.

Our Solomon Sea model has already been carefully checked with the scarce available observations (Melet et al. 2010a, 2011). Recent in situ data analysis of Shipboard Acoustic Doppler Current Profiler data and of the Global Drifter Program data set confirm the mean circulation in the Solomon Sea inferred from our model and shown in Fig. 3 for both the surface (Hristova and Kessler 2012; see their Fig. 2) and the upper thermocline (Cravatte et al. 2011; see their Fig. 4b) waters (SW and TW, respectively, see Table 1). In both SW and TW layers, the signature of the LLWBCs is clearly seen. The NGCU is fed by both the North Vanuatu Jet (NVJ, Fig.1) and the Gulf of Papua Current (GPC, encompassing the Great Barrier Reef Undercurrent, the North Queensland Current, and the Hiri Current, see SPICE Community, CLIVAR letter 2012). It flows in the Solomon Sea and exits equatorward mainly through Vitiaz Strait. For SW, the SEC is maximum between 1° S and 4° S east of the Solomon Sea (Fig. 3). At these latitudes, part of the SEC flows into the Solomon Sea through Solomon Strait, heads southward in the central Solomon Sea, and recirculates in the NGCU. Therefore, the most striking difference of the Solomon Sea circulation between SW and TW is the strong southward inflow from the SEC entering the Solomon Sea at Solomon Strait in the surface layer, whereas waters flow equatorward through the straits from the LLWBCs at thermocline level.

Fig. 3
figure 3

Mean circulation vertically integrated in the surface (SW, top) and upper thermocline (UTW, bottom) waters as defined in Table 1. Units are in m2/s. Note the different scales for the colorbars

Full size image
Table 1 Definition of the water masses used in this study and in Melet et al. (2011)
Full size table

There are very few dataset to validate the simulated interannual variability in the Solomon Sea. Melet et al. 2010a showed good agreement of interannual anomalies of thermocline temperature in the Solomon Sea with observations provided by XBT data along the New Zealand–Japan ship track (their Fig. 3). The interannual variations of the simulated Solomon Sea SLA and surface transport through the south entrance of the basin are also in good agreement with altimetric data (not shown). Finally, the anomalies of velocities at the surface during El Niño presented on Fig. 6a compare well with those from Hristova and Kessler (2012, Fig. 6a) from drifter data.

2.2 ENSO composites

To analyse how the LLWBCs respond to ENSO in the Solomon Sea, we computed composites of El Niño and La Niña states, based on the SOI. The El Niño composite corresponds to the November–February mean of years 1986–1987, 1991–1992, 1994–1995, 1997–1998, 2002–2003, and 2004–2005. The La Niña composite corresponds to the November–February mean of years 1988–1989, 1995–1996, 1998–1999, 1999–2000, 2000–2001, and 2005–2006. The neutral composite corresponds to the November–February mean of the remaining years.

2.3 Lagrangian diagnostics

Lagrangian diagnostics were performed offline using the ARIANE algorithm (available online at http://stockage.univ-brest.fr/~grima/Ariane), which is dedicated to the computation of 3D streamlines in a given velocity field and subsequent water masses analyses on the basis of multiple particle trajectories (Blanke and Raynaud 1997; Blanke et al. 1999). The daily averaged modeled velocity, temperature, and salinity fields were used to temporally compute the trajectories of the numerical particles. Following Blanke et al. (1999), each particle bears an infinitesimal transport (defined as the transport through its corresponding initial section grid cell) that is conserved along its trajectory because of the incompressibility of water in the model. Turbulent diffusion processes were not used explicitly to calculate the trajectories, as particles are only advected by the modeled velocity field. Instead, effects of turbulent mixing on the tracked water masses were provided implicitly by the along-trajectory changes in tracer properties (temperature and salinity, referred to as T and S in the following). A water mass can be defined by a set of particles that match for instance selected position and density criteria. We use the same definitions for the water masses of interest for this study as introduced in Melet et al. (2011), and restated in Table 1.

The computed trajectories allow us to describe the Eulerian stream functions between different sections defining the Solomon Sea and thus to trace the average origin (backward integration) or fate (forward integration) of selected water masses. For forward integrations, each grid cell of the initial section corresponding to the inflow to be tracked is seeded by a particle, everyday during the 1986–2002 period, while for backward integrations, each grid cell of the initial section corresponding to the outflow to be tracked is seeded by a particle, everyday during the 1990–2006 period. Particle trajectories were then integrated in time until they reach a section delimiting the Solomon Sea or up to 4 years.

Lagrangian diagnostics have already been used to study large-scale ENSO variability (Izumo et al. 2002; Fukumori et al. 2004, Holland and Mitchum 2005). Here, they are used to examine how the water masses characteristics and pathways differ during ENSO conditions as compared to neutral conditions in the Solomon Sea. Composites of the trajectories are computed to examine the dynamics of a typical ENSO event (rather than differences between ENSO events). To build the ENSO/neutral composites, only particles seeded during winter are considered for forward integrations, while only particles reaching a final section during winter are considered for backward integrations.

3 ENSO circulation anomalies

3.1 Vertical structure of the ENSO transport anomalies

The LLWBCs extend from the surface down to 1,000 m depth and have their core at thermocline level (Melet et al. 2010a; Murray et al. 1995; Davis et al. 2012). In our model, the depth-cumulative transport at the southern boundary of the Solomon Sea, mainly due to the NGCU, reaches 18 Sv at 800 m in neutral ENSO conditions and varies from 13 Sv during La Niña to 22 Sv during El Niño conditions (Fig. 4). These ENSO-related transport anomalies are mainly confined to the top 250 m of the water column where they represent a huge variation of 7.5 Sv in comparison with the 10 Sv mean transport. The upper 250 m of the water column encompasses both the surface water (SW) and the upper part of the thermocline water (UTW) (defined in Table 1). In the following, we will focus on these two water masses, which concentrate most of the interannual variability because of their importance in feeding the WWV and EUC.

Fig. 4
figure 4

Depth cumulated transport during the winter period (November–February) at the Solomon Sea southern entrance (9.8° S) in Sv for the mean state (black line), El Niño state (red line), and La Niña state (blue line). The difference between El Niño and La Niña is plotted as a black dashed line, and the two green points locate the depth of the surface (SW) and upper thermocline waters (UTW) as defined in Table 1

Full size image

3.2 Transport variations

The simulated low-pass filtered transports at interannual timescales through the southern entrance, Solomon and Vitiaz straits, are presented in Fig. 5 and listed in Table 2 for both surface and upper thermocline waters. Note that, in the following, the sign convention used for the transports is the classic one: positive, equatorward/eastward; negative, poleward/westward. Therefore, the transport corresponding to the inflow through the Solomon Sea southern section is positive, while the transport corresponding to the inflow through the surface layers of Solomon Strait is negative. The transports corresponding to the outflows through Vitiaz and Solomon straits are positive.

Fig. 5
figure 5

Time series of the low-pass filtered (half power at 18 months) transports in the surface (SW, top) and in the upper thermocline (UTW, bottom) layers as defined in Table 1 through the southern entrance (9.8° S, black line), Solomon St (red line), and Vitiaz St (green line). The SOI is plotted as the black dashed line, and it is centered on zero in a and the horizontal dotted line in b

Full size image
Table 2 Mean values of the transports (in Sv) and their interannual variability (rms of the transport, in bracket) at the Solomon Sea southern boundary, Vitiaz Strait, and Solomon Strait in the surface layers (SW) and the upper thermocline (UTW)
Full size table

The main Solomon Sea inflow for SW and UTW occurs through the southern entrance. It is mainly accounted for by UTW (6.4 Sv), with a strong correlation to the SOI (correlation factor of −0.8 when considering a 3-month lag. Note that the same correlation is found between the SOI and transport estimated from altimetric data [as in Melet et al. (2010b), not shown)]. SW and UTW transports increase during El Niño events and decrease during La Niña events. Whereas the main UTW inflow is mostly transmitted to Vitiaz Strait in the mean state, its interannual anomalies [root mean square (rms) of 1.1 Sv] seem to be equally transmitted between Solomon and Vitiaz straits (transport rms are 0.5 Sv, Table 2). The SW mean transport through the south entrance is quasi-null, but highly variable in time (rms of 1.4 Sv). As a result, a net outflow can occur in the surface layers through the southern boundary (Fig. 5a, black line). At the opposite of UTW, the interannual anomalies of SW transport are higher at Vitiaz Strait than at Solomon Strait (transport rms are 1. and 0.7 Sv, respectively).

The second main Solomon Sea inflow occurs in the surface layers at Solomon Strait, with a mean SW transport of −1.5 Sv that enters the Solomon Sea south-westward (Fig. 3a). The intensity of this SW transport decreases during El Niño events and increases during La Niña events (correlation factor to the SOI of −0.75).

The ENSO composites (defined in Section 2.2) illustrate the anomalies of the Solomon Sea circulation for both SW and UTW (Fig. 6). During El Niño, the changes in the circulation are mainly characterized by an increase in the equatorward flux entering the Solomon Sea through its southern boundary as a consequence of the large-scale wind stress forcing (see Section 1). Most of these circulation anomalies come from the open ocean (SEC/NVJ) for SW, in agreement with the analysis of the drifter data by Hristova and Kessler (2012), whereas UTW circulation anomalies mainly come from the Gulf of Papua Current and enter the Solomon Sea by getting around the Louisiade Archipelago (Fig. 1). The pathways to exit the Solomon Sea are different between the surface and upper thermocline layers. In the surface layer, most of the anomalies exit through Vitiaz Strait and increase the NGCU along the Papua New Guinea coast, whereas in the upper thermocline layer, most of the anomalies exit through Solomon Strait, contributing to a decrease in the SEC entering the Solomon Sea at these latitudes. Such a pattern is similar to the 1997 El Niño circulation anomalies described from altimetry in Melet et al. (2010b, their Fig. 7b).

Fig. 6
figure 6

El Niño (left) and La Niña (right) composites of circulation anomalies vertically integrated in the surface (top) and upper thermocline (bottom) layers. Units are in m2/s

Full size image
Fig. 7
figure 7

Partition of the two main Solomon Sea inflows between the different exits of the basin for each water mass defined in Table 1: a south inflow and b Solomon Strait inflow. For each water mass, the first bar represents the partition of the inflow for El Niño conditions, the second bar for neutral ENSO conditions, and the third one for La Niña conditions. Solomon Is refers to the different straits of the Solomon Islands (Fig. 1). Note the different vertical axis for each panel

Full size image

During La Niña, the changes in the circulation are mainly characterized by a decrease in the LLWBCs transiting the Solomon Sea (Fig. 6, right panel) as a consequence of the wind stress forcing anomalies (Fig. 2). Changes are particularly effective at Solomon Strait where the LLWBC decreases by up to 35 % for UTW. The SEC increase in the western equatorial Pacific intensifies the inflow entering at Solomon Strait and even reverses the equatorward current at Vitiaz Strait and offshore the PNG coast for SW (Fig. 5).

This description of the mass transports highlights some asymmetry in the Solomon Sea circulation between El Niño and La Niña conditions. During El Niño conditions, the main Solomon Sea inflow occurs through its southern boundary, and anomalies of the equatorward flow mainly exit the basin through the surface layer at Vitiaz Strait and the upper thermocline layer at Solomon Strait. During La Niña conditions, an additional inflow occurs through the surface layer at Solomon Strait, and anomalies of the LLWBC outflow mainly occurs through the upper thermocline layer at Vitiaz Strait. The Lagrangian approach presented below will detail this result.

3.3 Lagrangian diagnostics

In this section, Lagrangian experiments are used to diagnose the main Solomon Sea routes and determine the fate of the Solomon Sea main inflows (through forward integrations) and the origins of the Solomon Sea main equatorward outflows (through backward integrations) according to ENSO conditions.

3.3.1 Fate of the south inflow and of the Solomon Strait inflow

To analyze more precisely the partition and fate of the two main inflows of the Solomon Sea, two forward Lagrangian experiments were performed. The seeded section is the Solomon Sea southern entrance (along 9.8° S) in the first experiment and Solomon Strait in the second one. The partition of the two inflows between the different possible exits of the Solomon Sea is shown in Fig. 7 for SW and UTW and for El Niño, neutral, and La Niña conditions.

Particles seeded at the southern Solomon Sea entrance (Fig. 7a) mainly exit at Vitiaz Strait and also at Solomon Strait at thermocline level. A significant contribution of the recirculations (circulation that goes back to its original section) at the southern boundary can also be noted. They could be the signature of mesoscale activity in the Solomon Sea, as already mentioned in Melet et al. (2010b) and observed by Davis et al. (2012).

During El Niño conditions, a larger fraction of south inflow waters reaches Vitiaz Strait both in the surface and upper thermocline waters compared to neutral ENSO conditions, resulting in an increase in the mass transport through the strait. A significant increase in the fraction of the south inflow UTW reaching Solomon Strait can also be noted. In the surface layer, particles are also able to reach Solomon Strait, and a possible reverse from inflow to outflow can be observed in the Solomon Strait surface layer (see Fig. 5). In addition, the southern recirculations appear less intense during El Niño conditions than during neutral ENSO conditions. These results are in agreement with a net increase in the LLWBCs entering Solomon Sea through its southern boundary (total inflow minus recirculations).

On the contrary, during La Niña conditions, a smaller fraction of the south inflow SW and UTW reaches Vitiaz and Solomon straits, which contributes to the decrease in the straits’ outflows. However, the SW and UTW southern recirculations significantly increase during La Niña conditions. If recirculations at the southern boundary are effectively related to the mesoscale activity in the Solomon Sea, it means that the circulation in the Solomon Sea could be more unstable during La Niña conditions than during El Niño conditions.

Particles seeded at Solomon Strait (Fig. 7b) mainly recirculate, especially for the thermocline water. However, in the surface layer, some reach the southern boundary, while in the upper part of the thermocline, some reach Vitiaz Strait. During La Niña conditions, the large increase in Solomon Strait inflow occurring in the surface layer (Fig. 6) mainly reaches the Solomon Sea southern boundary. In fact, the transport increase along the Solomon Strait to the southern boundary pathway is the most striking difference in the fate of Solomon Strait waters between El Niño and La Niña conditions in the surface layer.

3.3.2 Origins of Vitiaz and Solomon straits outflowing water

A complementary analysis is performed in this section to determine the origins of the waters flowing equatorward through Vitiaz and Solomon straits during El Niño, neutral, and La Niña conditions. To do so, two backward Lagrangian experiments were performed, with the seeded sections being Vitiaz and Solomon straits.

Figure 8a shows that Vitiaz Strait outflowing waters mainly originate from the southern entrance of the Solomon Sea, with an almost exclusive contribution of this source at thermocline level. However, a small fraction of Vitiaz Strait surface waters come from Solomon Strait.

Fig. 8
figure 8

Origins of the water flowing equatorward through Vitiaz Strait (a) and Solomon Strait (b). For each water mass, the first bar represents the partition (in Sv) of the inflow for El Niño conditions, the second bar for neutral ENSO conditions, and the third bar for La Niña conditions. Solomon Is refers to the different straits of the Solomon Islands (Fig. 1). Note the different vertical axis for each panel

Full size image

During El Niño conditions, Vitiaz Strait outflow increases compared to neutral conditions: This increase is mainly due to an increase in the transport along the south inflow pathway. In the surface layer, the contribution of water originating in Solomon Strait increases during La Niña conditions, which are favorable to a significant inflow at Solomon Strait.

Waters outflowing through Solomon Strait mainly originate from the south inflow (Fig. 8b), especially for UTW. There is a net increase in this contribution during El Niño conditions, whereas no significant difference exists between neutral and La Niña conditions. A second important source of Solomon Strait outflowing waters is Solomon Strait itself. Recirculations are responsible for a large transport at Solomon Strait, notably in the surface layer. It is not the objective of this paper to go forward on the analysis of these recirculations, but it is notable that altimetric data show a strong eddy kinetic energy signal at the mouth of Solomon Strait (see Melet et al. 2010b, Fig. 4b).

3.3.3 Transit time of the particles in the Solomon Sea

The transit time of particles in the Solomon Sea varies greatly according to the different pathways followed and to ENSO conditions (Table 3). The fastest pathway is the direct route from the southern entrance to Vitiaz Strait via the LLWBCs. It takes about 1.5 months for the particles to transit along this pathway, and the transit time along this route only shows small variations in relation to ENSO. The second main route for the thermocline water to escape the Solomon Sea is the pathway from the southern entrance to Solomon Strait. Transit times along this route are longer and vary strongly according to ENSO conditions. It takes 4 months for the particles to transit along this pathway during neutral conditions, about 3.5 months during El Niño conditions, and up to 6 months during La Niña conditions. These variations of the transit time along the southern entrance to Solomon Strait pathway are related to changes in the strength of the LLWBCs and also partly to enhanced recirculations during La Niña conditions (Fig. 7a).

Table 3 Transit time (in days) of the particles in surface and upper thermocline waters for the different pathways in the Solomon Sea in different forward and backward Lagrangian analyses depending on the ENSO state
Full size table

For surface water, the pathway from Solomon Strait to the southern entrance (Fig. 7b) is particularly effective during La Niña conditions with a 4 months transit time. A minor pathway is the feeding of the outflow at Vitiaz Strait from Solomon Strait during La Niña conditions (Fig. 8a) with a 5-month transit time.

As for the recirculation at Solomon Strait, the transit times of particles are around 1 month in the surface layer; they are shorter during La Niña and doubled during El Niño. In the upper thermocline layer, it takes 3 months for Solomon Strait water to recirculate in the Solomon Sea during neutral and El Niño conditions and 1.5 months during La Niña conditions.

The Lagrangian experiments allowed us to diagnose the different main pathways followed by Solomon Sea waters according to different ENSO conditions. In the next section, we take advantage of the Lagrangian analysis ability to diagnose the partition of the south inflow between the different straits to test a possible cause of the asymmetry of the Solomon Sea circulation during different ENSO-type conditions.

3.3.4 Bathymetric control at Vitiaz Strait

Melet et al. (2010a) showed that the narrowness of Vitiaz Strait at thermocline level (~42 km) conditions the partition of the NGCU between a branch flowing through Vitiaz Strait and northwestward along the Papua New Guinea coast, and a branch flowing eastward along the south coast of New Britain before exiting through Solomon Strait (called New Britain Coastal Undercurrent in Melet et al. 2010a). We suspect that such a bathymetric control could play a role in the asymmetric response of the circulation to ENSO-type conditions, as described in Section 3.2. To test this assumption, a sensitivity simulation was run in which Vitiaz Strait was widened (by 60 %, from 42 to 67 km in the thermocline). The same Lagrangian experiments as in Section 3.3.1 were performed to analyze the response of the partition of the south inflow between the straits to the bathymetric constraint. This sensitivity experiment shows that when Vitiaz Strait is widened, the interannual variability of the transport through this strait increases while the one through Solomon Strait decreases (up to 60 % for UTW, see Table 2). Thus, the bathymetric control at Vitiaz Strait impacts the partition of the south inflow between the different equatorward straits at ENSO timescales. It plays a stronger role during El Niño, when the NGCU is strengthened (not shown). The increased effectiveness of this bathymetric control during El Niño than during La Niña conditions suggests that the higher variability of the thermocline transport observed through Solomon Strait than in Vitiaz Strait is not only due to the wind curl anomalies associated to El Niño conditions but also to the limitation of the transport through the narrow Vitiaz Strait.

To summarize, the circulation in the Solomon Sea presents a real asymmetry during El Niño and La Niña conditions. Solomon Sea inflows mainly come from the south entrance during El Niño, whereas a significant inflow is also provided from Solomon Strait during La Niña. The partition of the flow between Vitiaz and Solomon straits differs significantly according to ENSO-type conditions. The transit time of water in the Solomon Sea can double depending on ENSO-type conditions. Therefore, these differences in the Solomon Sea circulation between El Niño and La Niña conditions could have different impacts on how the equatorial Pacific is fed from the South Pacific LLWBCs and on how the water masses properties are modified in the Solomon Sea compared with neutral conditions.

4 ENSO T and S anomalies

In this section, the Solomon Sea water masses temperature and salinity are examined in relation to ENSO conditions. First, the large-scale response to ENSO conditions implies some modifications in the characteristics of the water masses entering the Solomon Sea. Second, the temperature and salinity of the water masses transiting through the Solomon Sea are modified along their routes (Melet et al. 2011). These changes in the T and S characteristics of the water masses could be affected by ENSO-related circulation variability as described in Section 3.

4.1 Large-scale anomalies

There is a large literature describing the ENSO-related variability of temperature and salinity in the southwest Pacific, in relation to the atmospheric variability (e.g., Delcroix and Henin 1989; Delcroix 1998; Kessler 1999; Gouriou and Delcroix 2002). At ENSO timescales, the anomalous wind stress curl (mainly occurring in the region shown by the box in Fig. 2) strongly pulls up (down) the thermocline during El Niño (La Niña), mainly through Rossby waves forced by the anomalous Ekman pumping (e.g., Capotondi et al. 2003; Li and Clarke 2007). As a result, notable changes in the vertical distribution of temperature are discernible in the southwest tropical Pacific Ocean. During El Niño conditions, the resulting thermocline shoaling modifies the temperature distribution over the whole water column all the way to the surface and is responsible for the SST cooling anomaly (−0.5 to −1 °C) detailed in Delcroix and Henin (1989). During an ENSO event, drastic sea surface salinity (SSS) changes occur below the mean position of the SPCZ, providing evidence that precipitation changes related to the migration of the SPCZ is the main mechanism responsible for the observed SSS changes (Gouriou and Delcroix 2002). To a lesser extent, geostrophic current anomalies can explain SSS changes by the advection of the southwestern tropical front that separates fresh, warm pool waters from salty subtropical waters (Kessler 1999). This salinity front moves westward (eastward) during an El Niño (La Niña) event, bringing saltier and colder waters in the southwestern tropical Pacific (Gouriou and Delcroix 2002).

As a result of these large-scale anomalies related to ENSO, the temperature and salinity of the southwestern tropical Pacific surface water presents large interannual variability that is well correlated to the SOI, with saltier and colder waters during El Niño than during La Niña events. The upper thermocline water undergoes the effect of forced Rossby waves in relation to the wind stress curl anomalies that develops at ENSO timescales in the area. During El Niño conditions, the resulting thermocline shoaling induces fresher and colder thermocline water than during neutral ENSO conditions. The opposite situation prevails during La Niña conditions, with saltier and warmer thermocline water. These large-scale modifications of the T and S properties of the water masses are brought to light when considering anomalies of temperature and salinity at the southern Solomon Sea entrance (Fig. 9). Such contrasts also exist when looking at Vitiaz or Solomon straits. At Solomon Strait, saltier waters are found during La Niña conditions not only at thermocline levels but also at the surface, with fresher water between them (Fig. 10). This can be explained by strong westward equatorial current anomalies (Fig. 6c) that develop during La Niña states and are responsible for the advection of central Pacific surface layer salty water that reaches Solomon Strait.

Fig. 9
figure 9

Top Anomalies of temperature (in °C) and bottom of salinity (in psu) at the southern boundary (9.8° S) of the Solomon Sea during El Niño conditions. The depth of the 23.3, 25.7, and 26.7 isopycnal surfaces, delimiting the surface, upper thermocline and lower thermocline layers are indicated (see Table 1)

Full size image
Fig. 10
figure 10

Anomalies of salinity (in psu, left) and meridional velocity (in cm/s, right) at Solomon Strait during La Niña (upper panel) and El Niño (lower panel) conditions. The depth of the 23.3, 25.7, and 26.7 isopycnal surfaces, delimiting the surface, upper thermocline and lower thermocline layers are indicated (see Table 1)

Full size image

4.2 Modifications in the Solomon Sea

During their equatorward transit through the Solomon Sea, water masses temperature and salinity are modified. In the climatological state, these modifications consist of a reduction of the vertical temperature and salinity gradients over the water column (Melet et al. 2011). More precisely, the high salinity found in the upper thermocline and characteristic of the STMW is eroded (by about −0.11 psu) and exported toward the surface and deeper layers, whereas a downward heat transfer occurs over the water column (with an associated cooling of 0.15–0.30 °C for the thermocline water from the Solomon Sea inflows to the equatorward outflows). These water mass modifications can be partly explained by strong diapycnal mixing in the Solomon Sea (Melet et al. 2011), notably sustained by tidal mixing.

In the following, we take a step forward from the climatological Solomon Sea water mass modification study of Melet et al. 2011 by analyzing the modifications of temperature and salinity from the Solomon Sea inflows to Vitiaz and Solomon straits, in relation to ENSO conditions. Similar Lagrangian diagnostics to the ones described in Section 3.3.2 are used.

To quantify the T and S modifications of the water masses between their entrance in the Solomon Sea and their exit through Vitiaz or Solomon straits, we computed transport-weighted temperature and salinity modifications (e.g., ∆T, ∆S) as follows:

$$ \begin{array}{*{20}{c}} {\Delta T = \frac{{\sum\limits_{{i = 1}}^{n} {\left( {{{T}_{{{\rm{out}}}}} - {{T}_{{{\rm{in}}}}}} \right) \times {\rm{T}}{{{\rm{r}}}_{i}}} }}{{\sum\limits_{{i = 1}}^{n} {{\rm{T}}{{{\rm{r}}}_{i}}} }};} & {\Delta S = \frac{{\sum\limits_{{i = 1}}^{n} {\left( {{{S}_{{{\rm{out}}}}} - {{S}_{{{\rm{in}}}}}} \right) \times {\rm{T}}{{{\rm{r}}}_{i}}} }}{{\sum\limits_{{i = 1}}^{n} {{\rm{T}}{{{\rm{r}}}_{i}}} }};} \\ \end{array} $$

where Tr i is the transport of particle i following the considered route, S out (T out) is its salinity (temperature) at the Solomon Sea outflow, and S in (T in) is its salinity (temperature) at the Solomon Sea inflow.

4.2.1 Modifications of Solomon Sea water outflowing through Vitiaz Strait

Figure 11 shows the temperature and salinity modifications of Solomon Sea waters flowing through Vitiaz Strait depending on their origin (south entrance or Solomon Strait entrance).

Fig. 11
figure 11

Transport weighted changes in temperature (left) and salinity (right) from the south inflow (top) and Solomon Strait (bottom) to Vitiaz Strait. Backward Lagrangian analysis including years 1990–2006. Note the different vertical axis for each panel

Full size image

Most of the Vitiaz Strait outflowing water comes from the south entrance (Fig. 8a). During their journey in the Solomon Sea, UTW is getting colder and fresher. It corresponds to a relatively low modulation of the modifications described for the mean state in Melet et al. (2011). SWs tend to become warmer (~0.18 °C) and fresher during El Niño conditions, whereas they tend to become colder (~0.3 °C) and saltier during La Niña conditions. From Solomon Strait, SWs (representing ~5 to 15 % of the outflowing waters through Vitiaz Strait, Fig. 8a) become colder and fresher, with stronger modifications during La Niña (~2 °C/~0.3 psu). During La Niña conditions, westward advection of warm and salty water from the SEC is particularly effective at Solomon Strait. These water properties strongly differ from the inherent Solomon Sea SW and isopycnal mixing between the different sources of surface waters could explain the T and S modifications along both pathways to Vitiaz Strait.

4.2.2 Modifications of Solomon Sea water outflowing through Solomon Strait

Figure 12 shows the temperature and salinity modifications of Solomon Sea waters flowing through Solomon Strait. As discussed in Section 3, most of the Solomon Strait outflowing water comes from the south entrance and from the Solomon Strait itself (recirculations). In the surface layers, recirculations are the dominant contribution, and SW become colder and fresher along these pathways. However, T and S modifications are stronger along the surface route coming from the south entrance. Along this route, SWs are getting warmer and fresher during El Niño conditions, and colder and saltier during La Niña conditions. Since similar modifications were noted along the southern entrance to the Vitiaz Strait route, similar processes might be acting to modify the water mass thermohaline characteristics.

Fig. 12
figure 12

Transport-weighted changes in temperature (left) and salinity (right) from the south inflow (top) and Solomon Strait (bottom) to Solomon Strait. Backward Lagrangian analysis including years 1990–2006. Note the different vertical axis for each panel

Full size image

As was the case for UTW flowing from the south entrance to Vitiaz Strait, UTW flowing to Solomon Strait are also getting colder and fresher, and diapycnal mixing might be the main contributor for this water mass modification (Melet et al. 2011).

To summarize, water mass T and S properties entering the Solomon Sea are highly influenced by the large-scale ENSO-type variability: lower (higher) temperature prevails on the water column during El Niño (La Niña) conditions compared with neutral conditions, whereas the surface layer is saltier and the subsurface layer is fresher during El Niño conditions. The Solomon Sea contributes to the transformation of temperature and salinity water masses of the LLWBCs, but its influence differs according to the ENSO-type conditions. During El Niño conditions, the entering T and S anomalies are notably enhanced in the Solomon Sea, whereas they are damped during La Niña conditions.

4.3 Heat and salt transport anomalies at Vitiaz and Solomon straits

The heat and salt fluxes likely to reach the equatorial Pacific from Vitiaz and Solomon straits combine large-scale constraints and local effects inside the Solomon Sea both in terms of mass transport and water masses T and S properties. In the previous sections, we intended to illustrate their respective variability at ENSO timescales. Here, the temporal variability of heat and salt fluxes (Fq and Fs, respectively) through Vitiaz and Solomon straits are estimated for SW and UTW based on the previous backward Lagrangian experiments and are computed as follows:

$$ \begin{array}{*{20}c} {\mathrm{Fq}={\rho_0}{C_p}\sum\limits_{i=1}^n {{T_i}\times \mathrm{T}{{\mathrm{r}}_i};} } & {\mathrm{Fs}=\sum\limits_{i=1}^n {{S_i}\times \mathrm{T}{{\mathrm{r}}_i}} } \\ \end{array} $$

where ρ 0 is the reference density of water (1,020 kg m−3), C P is the heat capacity of water under constant pressure, Tr i is the transport of the particle i, and T i its temperature, S i its salinity, and n the number of particles for the considered pathway.

The heat and salt fluxes, like the mass flux (e.g., Fig. 5), are well correlated with the SOI, and they increase with El Niño and decrease with La Niña (Fig. 13). Not surprisingly, the highest fluxes are found at Vitiaz Strait for UTW, and the lowest fluxes are found at Solomon Strait for SW. However, the highest variabilities are found at Vitiaz Strait for SW, and at Solomon Strait for UTW, where the fluxes are of similar amplitudes. It means that with regard to ENSO, SW at Vitiaz Strait could play a major role in the depletion/replenishment of the warm pool of the western Pacific, whereas UTW at Solomon Strait could have a significant impact on the characteristic of the EUC at ENSO timescales. During the 20-year period analyzed in this study, the strongest El Niño event is the 1997 one, whereas the strongest La Niña event is the 1996 one. The total (SW + UTW) heat and salt fluxes during the 1997 El Niño reach 1,240 TW and 450 psu Sv, which double the 1996 La Niña values of 600 TW and 260 psu Sv. The temperature/salinity contributions to the variability of the fluxes are marginal compared to the mass transport, but they are perceptible especially for the highest sources of variability, i.e., the SW at Vitiaz Strait and UTW at Solomon Strait. Changes in temperature of SW exiting through Vitiaz Strait during their transit through the Solomon Sea (Fig. 11) are of opposite sign to the ones induced by the large-scale ENSO-type variability (e.g., Fig. 9), with the effect of damping the heat flux, particularly during La Niña. For UTW at Solomon Strait, the temperature modifications inside the Solomon Sea (Fig. 12) induce a heat loss that enhances (damps) the El Niño (La Niña) temperature anomalies entering the Solomon Sea from the south. This results in a damping of the heat flux during El Niño conditions.

Fig. 13
figure 13

Equatorward heat (top) and salt (bottom) fluxes at Vitiaz Strait (left) and Solomon Strait (right) for surface waters (SW, green curves) and upper thermocline waters (UTW, blue curves). The SOI is plotted as a black line (y-axis on the right), and El Niño events considered in this study are indicated in pink periods, while La Niña events are indicated in blue periods

Full size image

5 Conclusions

The South Pacific LLWBCs carry waters of subtropical origin through the Solomon Sea before joining the equatorial Pacific. The potential role of the South Pacific LLWBCs for the climate, by impacting the low-frequency modulation of ENSO, has already been highlighted and is one motivation of the SPICE/CLIVAR program. This study aims to supplement the underlying efforts made as part of the SPICE program to better understand the circulation in the Solomon Sea and its variability (Melet et al. 2010a, b, 2011; Cravatte et al. 2011; Hristova and Kessler 2012) by focusing on the interannual variability of the Solomon Sea circulation and water mass T and S modifications.

Because of the scarcity of in situ data, this study is based on numerical modeling. A high-resolution (1/12°) model of the Solomon Sea (already validated in Melet et al. 2010a, 2011) was implemented to resolve the complex circulation and eddy field structure. The model includes a parameterization of the vertical mixing induced by the dissipation of internal tides. Evaluation of our interannual simulation with the very few dataset available is satisfactory, which gives some confidence in the robustness of our results even though they remain model dependant.

Solomon Sea circulation exhibits a huge interannual variability related to the large-scale ENSO forcing, as shown by the model and altimetric data (Melet et al. 2010b). The LLWBC transports in the Solomon Sea are impacted by this large-scale ENSO forcing: They strengthen during El Niño conditions and weaken during La Niña conditions in order to counterbalance the interior geostrophic flow variability. Nevertheless, this simple large-scale picture is complicated because of the intricate bathymetry and high levels of turbulence characteristic of the Solomon Sea, which play an active role in constraining the circulation and water mass hydrologic properties. From a climatological viewpoint, modeling studies of the Solomon Sea have already documented (1) how the LLWBCs are divided into different branches, with consequences notably for how the LLWBCs connect to the equatorial Pacific and feed the EUC, and (2) how diapycnal mixing modifies water mass T and S properties. The objective of this paper is to make a step forward from the mean state by investigating some possible changes due to the ENSO-type variability that could affect the subtropical/equatorial connection at interannual timescales. This interannual variability mainly impacts the upper oceanic layers characteristics of both the surface waters and upper thermocline waters. Therefore, these two water masses, which are relevant both for the depletion/repletion of the warm pool and for the feeding of the EUC, are the main focus of this study.

In this modeling study, we show that the interannual variability of the transport in the Solomon Sea does not only correspond to a strengthening/weakening of the LLWBCs in relation to ENSO. Indeed, a strong asymmetry of the Solomon Sea circulation exists between El Niño and La Niña conditions. During El Niño conditions, the great majority of the waters entering the Solomon Sea do so through its southern boundary, whereas during La Niña conditions, a strong additional inflow occurs through Solomon Strait surface layers. This Solomon Strait inflow heads southwestward in the Solomon Sea and partly recirculates to join the equatorward flowing LLWBCs. Another notable difference of the Solomon Sea circulation between different ENSO-type conditions comes from where in the vertical the LLWBCs transport are impacted. During El Niño, the strengthening of the LLWBCs extends up to the surface, whereas during La Niña, changes in the LLWBCs mainly occur in the subsurface layers. These differences in terms of circulation between El Niño and La Niña conditions are mainly, but not completely, driven by the large-scale forcing. As already mentioned in Melet et al. (2010a), the narrowness of Vitiaz Strait imposes a bathymetric constraint that conditions the partition of the LLWBCs between Vitiaz and Solomon straits. Here, we show that this bathymetric constraint is more effective when the NGCU is strengthened. It means that the partition of the flow between the two straits changes according to El Niño and La Niña conditions, with a partition favorable to the Solomon Strait exit (in terms of percentage between the two straits) during El Niño conditions.

The temperature of the water masses is particularly important for the flow exiting the Solomon Sea to join the equatorial region because of possible impacts on the EUC heat content and equatorial Pacific SST. Water mass T and S properties entering the Solomon Sea are highly influenced by the large-scale ENSO-type variability. The surface layer is highly sensitive to the anomalous precipitation forcing, and the subsurface layer mainly responds to the anomalous wind stress curl forcing inducing an Ekman pumping response in the southwest Pacific mainly through the generation of Rossby waves. To summarize, lower (higher) temperature prevails on the water column during El Niño (La Niña) conditions compared with neutral conditions, whereas the surface layer is saltier and the subsurface layer is fresher during El Niño conditions. In addition, changes in the T and S characteristics of the water masses inside the Solomon Sea may enhance or reduce the interannual temperature and salinity anomalies entering the Solomon Sea. By performing Lagrangian experiments, we showed that the initial T and S anomalies entering the Solomon Sea are notably enhanced during El Niño conditions, whereas they tend to reduce during La Niña conditions. The combined effect of both the mass transport and the T and S modifications show that the Solomon Sea outflows are highly variable in the surface layer through Vitiaz Strait and in the upper thermocline through Solomon Strait. Their respective heat and salt fluxes are relatively similar and may vary by up to a factor of 2 depending on the ENSO state. They could impact differently the equatorial Pacific at ENSO timescale with the surface waters at Vitiaz Strait directly influencing the warm pool characteristic, whereas the thermocline waters at Solomon Strait could significantly impact the EUC properties at ENSO timescale.

A next step could be to follow these Solomon Sea outflows to the equatorial Pacific Ocean and to investigate their role in the variability of the EUC, but it requires a different model configuration. Another step could be to examine the high frequency variability in the Solomon Sea and its relation to the temperature and salt transports since altimetry (Melet et al. 2010b) and in situ glider data (Davis et al. 2012) have shown that mesoscale activity is strongly developed in the Solomon Sea. To this end, a high-resolution model of the Solomon Sea using a 1/36° spatial resolution is being implemented to resolve most of the fine mesoscale signature, and to analyze their variability and the processes generating them. Glider data and observations from the Pandora cruise hold in 2012 (that has deployed moorings in the different straits) will greatly help to assess the model results.