U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications

The circulation of the eastern tropical Pacific: A review

W. S. Kessler

NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington

Prog. Oceanogr., 69, 181–217, 2006.
Copyright ©2006 Elsevier Science Ltd. Further electronic distribution is not allowed.

4.3. Eastern equatorial Pacific: mean and annual cycle

Relatively much less is known about the circulation in the far eastern than in the central equatorial Pacific, and a full picture of the currents and their interconnections is yet to be accomplished. In the central Pacific, where the winds are nearly zonal, the circulation is dominated by a westward surface SEC, with an eastward EUC in the thermocline (Fig. 6, top); Ekman divergence transports surface water away from the equator in both hemispheres, and consequently produces strong equatorial upwelling. In the east several of these elements are modified by weaker trade winds that also have a strong southerly component (Fig. 1). In addition, the zonal thermocline slope that balances equatorial easterlies (McPhaden and Taft, 1988) means that the thermocline and EUC rise much closer to the surface towards the east, trapping a complex mix of processes within a thin layer (Fig. 6). Upwelling increasingly drains water and momentum from upper layers of the EUC as it flows east.

Drifter tracks (Fig. 4; and see Reverdin et al., 1994) suggest that the near-surface SEC is weak east of about 85°W, and rapidly gains speed as it flows west. This was described by Wyrtki (1966), who correctly interpreted the increase as a sign that the SEC was gaining mass from the decelerating EUC below (as noted in Section 2 the role of the vertical circulation was not understood until later). Although there are no data to allow an estimate of the transport of the SEC east of 95°W, where it appears to be quite small (Fig. 6, bottom, and Fig. 4), the Johnson ADCP data (Appendix C.1) resolves SEC transport on the 15° longitude spacing of the TAO lines, beginning at 95°W. SEC transport above 200 m within 6°S–6°N at 95°W is about 12 Sv, increasing to about 35 Sv by 140°W, consistent with this interpretation.

Both the drifters (Fig. 4) and the ADCP data (Fig. 6) show that the SEC is split into two lobes, and is thin and weak on the equator, where the EUC surfaces or nearly so. Just south of the equator from the Galapagos to about 110°Wthe mean surface zonal flow is near zero, and a strong southward divergence from the equator is seen (Fig. 4). Although the drifter population is smaller in this region (because the drifters diverge rapidly from the equator), sampling problems do not appear to explain this unexpected feature. The same weak surface current appears in the Johnson et al. (2002) ADCP data and is consistent with the very small mean surface zonal currents measured by the TAO mooring at 0°W, 110°W (Kessler et al., 1998). The splitting of the SEC has been attributed to upwelling of eastward momentum from the EUC, and interpreted as an important nonlinear effect of easterly equatorial winds (Philander and Pacanowski, 1980; McPhaden, 1981), but weaker easterlies over the cold tongue, caused by atmospheric boundary layer effects of the cold SST (Chelton et al., 2001), may also contribute.

The EUC surfaces during boreal spring, changing the near-surface flow at 0°W, 110°W to eastward at more than 20 cm s during the average April and May (Kessler et al., 1998; Yu and McPhaden, 1999a,b); this reversal reduces the annual mean zonal current at 110°W to near zero. While the eastern EUC shoaling occurs when local easterly winds are weakest, this is part of a basinwide phenomenon that begins in the far east in March, and reaches the dateline by about August (Reverdin et al., 1994; Yu and McPhaden, 1999b; Johnson et al., 2002); it is forced by the westward-propagating annual cycle of zonal wind along the equator producing a mixture of equatorial Kelvin and long Rossby wave modes (Yu and McPhaden, 1999b; Wang et al., 2000). Wind-forced ocean models show this signal clearly (e.g., Yu et al., 1997; Kessler et al., 1998). Because the mean EUC is much shallower in the east, it is only here that an actual reversal of the surface westward flow is seen: the surface current along the equator is eastward from about 140°W to at least 95°W during boreal spring (Johnson et al., 2002). It is worth noting that the seasonal cycle has an unexplained low-frequency modulation and has been observed to change substantially in amplitude in different decades (Gu et al., 1998).

The fact that the weakest SEC is found just south of the equator (Figs. 4 and 6, bottom) is consistent with the meridional circulation induced by southerly cross-equatorial winds (Fig. 1); note that the coldest SST is also south of the equator (Fiedler and Talley, 2006). These winds drive a vertical circulation cell that is downwind (northward) at the surface and upwind in the thermocline, with enhanced upwelling south of the equator (Philander and Delecluse, 1983; Mitchell and Wallace, 1992; Kessler et al., 1998). This cell therefore advects westward wind-input momentum northward across the equator, and upwells eastward momentum from the EUC at about 1–2°S, contributing to the asymmetric pattern seen in Fig. 4. (Although the northward surface velocity predicted by this theory is not apparent in Fig. 4, it is seen in the TAO moored velocity at 110°W.)

A comprehensive examination of hydrographic data by Lukas (1986) showed that high-salinity, high-oxygen water flowing eastward in the EUC can be detected all the way to the coast of Ecuador and Peru (Fiedler and Talley, 2006), indicating that at least the deeper layers of the EUC continue further east than the Galapagos. Some model results suggest that the physical blocking of the EUC by the Galapagos Islands leads to a surge of upwelling west of the archipelago, abruptly draining off its upper layers (Eden and Timmermann, 2004), but the salinity and oxygen properties imply that the lower EUC feeds the Peru Undercurrent and becomes a source of the Peru upwelling (Brink et al., 1983). Toggweiler et al. (1991) confirmed Lukas' result using C measurements from coral heads. They suggested that low C water originates in the subantarctic region of the southwest Pacific, flows north to enter the lower reaches of the EUC, and populates the east Pacific thermostad to eventually upwell off Peru. Note that Peruvian coastal upwelling is strongly seasonal, and appears to reach far into the thermocline and below (Fig. 9, right), consistent with this being the destination of thermostad water. Steger et al. (1998) found evidence of the EUC splitting around the Galapagos with the main branch flowing to its south. On the other hand, the only published measurement of the EUC in the far east was made using an underway ADCP profiler on WOCE cruise P19 along 86°W in March 1993 (Tsuchiya and Talley, 1998); this showed what appears to be a strong EUC (70 cm s) centered just north of the equator at 70 m depth. However, this period was during a weak El Niño that produced several Kelvin wave pulses of eastward equatorial currents that were observed further west around this time (Kessler and McPhaden, 1995b), so the single P19 section may very well not represent the mean. In addition to Lukas' water property evidence, it should be suspected that the EUC would be found south of the equator because of the meridional circulation cell mentioned above in connection with the SEC.

Dynamically, it is not clear why there should be an EUC east of the Galapagos, because the usual understanding of the undercurrent suggests that it is driven by the zonal pressure gradient, which itself is due to the mean easterly winds characteristic of the central Pacific. East of about 90°W, the zonal winds are very weak or westerly (Fig. 1), so according to linear dynamical ideas there should be no pressure gradient or EUC. (Note that the weak SEC in this region (Fig. 4) is consistent with the weak zonal winds). In fact, the integrated zonal pressure gradient calculated from the AOML XBT data relative to 450 m is reversed in the far east. This difficulty was recognized by Roden as early as 1962. A possible piece of the answer was proposed by Kessler et al. (2003), who used an ocean GCM to show that eastward advection of cyclonic relative vorticity on the flanks of the EUC strengthened both the EUC and SEC branches well east of where an eastward undercurrent would be expected from linear dynamics, which is consistent with observations of these currents at 140°W and 110°W. While they had no data to compare with the far eastern Pacific mode results, this nonlinear mechanism would act to produce an EUC in the far east even without a pressure gradient to drive it.

Although most of the EUC appears to flow all the way to the Peru coast in the thermocline, there is also substantial downstream loss of EUC water to the surface layer by upwelling, and the destination of this water remains in question. Presumably the surface flow combines a diverse and hard-to-quantify set of competing processes acting within the thin upper layer, including poleward Ekman divergence, westward and northward directly wind-driven currents, and the surfacing eastward EUC. An inverse calculation using the hydrographic data compilation of Johnson et al. (2002) suggested that about 10 Sv of equatorial water above the thermocline (including water upwelled from the upper layers of the EUC) is transported northward in the surface layer (Sloyan et al., 2003). They argued that this was one of the pathways by which intermediate water entering the South Pacific was transported to the northern hemisphere (and eventually to the Indonesian Throughflow). By contrast, Blanke and Raynaud (1997) diagnosed Lagrangian trajectories in an ocean GCM and came to the opposite conclusion: that most of the water upwelled from the EUC in the east exited towards the south. Clearly this issue remains unsettled.

The role of the distinctive equatorial circulation in producing the vertical and horizontal exchanges that maintain the east Pacific cold tongue and the sharp SST front along about 2°N has been the subject of a great deal of research over 50 years (e.g., Cromwell, 1953). Most of this work has concerned the central Pacific upwelling zone, but many of these processes are relevant to at least the part of the eastern equatorial Pacific west of the Galapagos, and the major points will be reviewed briefly here. The temperature balance is generally thought to be among equatorial upwelling and vertical mixing, meridional mixing across the SST front, and surface heating. Perhaps surprisingly, zonal advection may be less important. The continuous band of cool SST connecting the coast of Peru with the equatorial cold tongue suggests advection of coastal water out along the equator. However, the southwestward surface currents west of Peru (Fig. 4), consistent with the Ekman drift implied by the mean winds (Fig. 1), do not support such a hypothesis, nor is there evidence of continuous flow connecting the coastal and offshore regions (Fig. 4). Some model experiments also contradict this idea. Kessler et al. (1998) put a wall extending 700 km westward from the Ecuadorian coast at 4°S in an ocean GCM and found little difference in SST along the equator west of the Galapagos, either in the mean or in the annual cycle. Note that coolest SST along the equator occurs less than a month after that at the coast (Fig. 9), probably too fast to be due to advection. Instead, the thermal structure changes in both regions are consistent with local wind-driven upwelling; SST cooling is associated with thermocline uplift that occurs in phase with increases of the local winds. A contrasting view was given by Swenson and Hansen (1999), who used drifter data to estimate that advection had an non-negligible influence on the heat budget on the equator, and hypothesized that this was partly due to transport from the coast. It is also thought that during El Niño events, zonal advection from the west can be a significant contribution to SST warming, at least in the central Pacific (Picaut et al., 1996).

Numerous attempts to measure upwelling velocity have been based on finding the divergence of measured horizontal currents; e.g., Weisberg and Qiao (2000) used moored current meters, Johnson et al. (2001) used repeated shipboard ADCP sections, and Hansen and Paul (1987) used surface drifter currents. Indirect methods have been based on divergence of geostrophic and Ekman transports around large boxes spanning the cold tongue (Wyrtki, 1981; Bryden and Brady, 1985; Meinen et al., 2001). All these studies have estimated upwelling velocities on the order of a few m day, with total vertical transport of 30–50 Sv over the east-central Pacific, though the spatial pattern of upwelling (whether broad and relatively slow, or narrow and correspondingly fast) remains unknown. There is also little agreement on the depth to which upwelling reaches as it works against the stratification of the upper thermocline. Cross-isopycnal transport requires either heating from above (for example through penetrating radiation) or turbulent mixing, which is just beginning to be understood (see, e.g., Lien et al., 1995). While upwelling transport is due to the local Ekman divergence, its effect on SST and nutrients in the upper layer depends very much on the background stratification, which is modulated by basinscale conditions. Despite the many uncertainties, there is no doubt that upwelling is an order(1) term in the heat and mass balance of the equatorial Pacific, both in the mean and in its seasonal and interannual variability. Since upwelling is so important but at the same time impossible to measure directly, attention has been given to consistency checks that compare the heat fluxes due to upwelling to other elements of the heat balance (Wang and McPhaden, 1999, 2000).

Mixing across the SST front north of the equator opposes the cooling due to upwelling. This mixing primarily takes the form of tropical instability waves (TIW), which are due to the shears between the eastward EUC and NECC and the westward SEC (Fig. 6), (Philander, 1978; Willett et al., 2006). It was first thought that the NECC-SEC shear was the key element, but some investigators now argue that the EUC-SEC shear is important as well (Yu et al., 1995; Chelton et al., 2003), and this question remains controversial. The TIW distort the front into cusp-like shapes that are easily seen in satellite imagery (Legeckis, 1977; Chelton et al., 2000; Chelton et al., 2001), and result in a substantial equatorward heat transport by mixing across the front, as tongues of warm water move south and cool water move north. Hansen and Paul (1987) and Bryden and Brady (1989) estimated this heat flux as comparable to that due to upwelling, and other observational balances and model experiments agree.

The tendency to compensation between upwelling cooling and TIW warming occurs in the annual cycle as well. As winds strengthen in the second half of the year (Fig. 9), the SEC and NECC both strengthen (see Section 4.2 and Fig. 7), as does equatorial upwelling. These tendencies cool the equator and warm the region north of the SST front, increasing the SST gradient as well as strengthening the shear that drives the TIW and thereby produces stronger mixing. Therefore, during June–December, opposing SST tendencies are generated and the net effect of these ocean fluctuations on SST is smaller than might be assumed from the upwelling increase alone (Enfield, 1986; Kessler et al., 1998; Swenson and Hansen, 1999). During El Niño events, by contrast, surface and local vertical fluxes are dominant (see next section).

Although the sun crosses the equator twice a year, in March and September, only an annual cycle is observed in SST, which is coldest in September (Mitchell and Wallace, 1992; Kessler et al., 1998; Fiedler and Talley, 2006). A major factor constraining SST to a simple annual cycle is the cooling due to equatorial upwelling. The entire thermocline rises about 20 m during June–September (Fig. 9, middle), bringing cool water within reach of mixing by the stronger winds. Another is the variations of the extensive stratus decks that cover the cool-water part of the eastern tropical Pacific, south of the SST front (Klein and Hartmann, 1993), that are at least partly driven by annual variations of the subtropical high pressure zones. The interaction of stratus clouds and cool SST comprises a positive feedback, since cool SST encourages the formation of stratus that reduces the September solar cycle maximum and therefore reinforces the SST anomalies. A similar situation prevails in the eastern Atlantic. Disentangling the diverse ocean-atmosphere influences on the eastern tropical oceans remains a central climate problem of the region that is just beginning to be attacked (Cronin et al., 2002).

Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage