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


Variability in the Eastern Equatorial Pacific Ocean During 1986-1988

Michael J. McPhaden and Stanley P. Hayes

NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington

Journal of Geophysical Research, 95(C8), 13,195-13,208 (1990)
Not subject to U.S. copyright. Published in 1990 by the American Geophysical Union.

4. Temporal Variations at 110W and 140W

4.1 Winds

Figure 2 shows time series of winds, SST, dynamic height 0/250 dbar, and 20C isotherm depth at 110W and 140W. The zonal winds at both locations tend to follow the normal seasonal cycle of strong southeasterlies in the latter half of the year and weak southeasterlies in boreal spring. However, easterly winds were weaker by about 1 m s at 140W in 1987 and stronger by a similar amount at 110W from January to July 1987. Meridional winds also tend to be stronger from the south in 1986-1987 during the ENSO and weaker from the south during 1988. The variations in local meridional wind component may be due in part to enhanced (reduced) vertical mixing in the atmospheric boundary layer during periods of warm (cold) equatorial SST anomalies. As Wallace et al. [1989] have shown, northward wind speed just 100 m above sea level in the eastern equatorial Pacific are more than twice as strong as at the surface. Wallace et al. and Hayes et al. [1989] have argued that SST variations near the equator affect the stability of the atmospheric boundary layer and therefore the efficiency with which momentum aloft is mixed down to the surface.

4.2 Sea Surface Temperature

SST at both locations was near normal for the first 4 months of 1986, followed by a weak anomalous warming during May-July. Persistent anomalies of >1C then developed in September 1986 at both 110W and 140W and lasted until January-February 1988. The warmest SST anomalies occurred at 110W in September 1987 (+3.4C) and at 140W in June 1987 (+3.0C). Cold anomalies developed abruptly in early 1988 following the ENSO. From March to May 1988, SST dropped nearly 8C at 110W, and in April 1988, SST dropped 4C at 140W. Subsequently, 2-3C cold anomalies persisted throughout 1988, with the coldest anomalies at 110W in December 1988 (-3.7C) and at 140W in November 1988 (-4.0C).

The development of these anomalies is not obviously related to the local zonal winds. Coincident with the warm SST anomalies in 1986 and 1987, the zonal wind component was either close to or weaker than normal at 140W, whereas it was close to or stronger than normal at 110W. A consistent argument in terms of anomalous Ekman divergence and local upwelling is not possible for both locations. Also, the precipitous drop in SST in the spring of 1988 occurred at a time when the local zonal winds were near normal, again ruling out anomalous local Ekman divergence as the causative mechanism.

4.3 Dynamic Height and 20C Isotherm Depths

Surface dynamic height and the depth of the 20C isotherm (which is an index for the depth of the thermocline) tend to be mirror images of one another at 110W and 140W, similar to what Rebert et al. [1985] found throughout the equatorial Pacific. The mean seasonal cycle in both variables is dominated by the annual and semiannual harmonics with an annual range in dynamic height of 5 dyn. cm (110W) and 13 dyn. cm (140W) and an annual range in 20C isotherm depth of 19 m (110W) and 42 m (140W). In contrast, we note that during 1986-1988, variability is dominated by an interannual rise in the dynamic height from 1986 to 1987 and subsequent drop into 1988, and an intraseasonal variation with periods of 2-3 months. The amplitude range of both these time scales is larger than the annual range. The 20C isotherm depth, for example, shoaled by over 100 m at 110W from January 1987 to June 1988 when it broke the surface; similarly, the 20C depth ranged between 180 m and 0 m at 140W from 1987 to 1988. These changes and similar ones in dynamic height are large enough to obscure the mean seasonal cycle in these variables during 1986-1988.

On interannual time scales there is a tendency for warm SST anomalies to correspond to high surface height and depressed thermocline, and vice versa. This is most evident at 110W, where the mooring climatologies (and therefore anomalies) are better defined. For example, persistent warm anomalies, elevated dynamic heights, and depressed 20C isotherm depths develop in September 1986 and last until late 1987 to early 1988. Similarly, cold SSTs in March-May 1988 set in with a 40-m rise in the depth of the 20C isotherm and a 10 dyn. cm drop in dynamic height. As with SST, the interannual changes in dynamic height and 20C isotherm depths cannot be simply related to local zonal wind variations.

Much of the intraseasonal variability with periods of 2-3 months in Figure 2 is due to the excitation of first baroclinic mode Kelvin waves. McPhaden and Taft [1988] discussed these wave time series spanning 110-140W for the period November 1983 to June 1986, the last 5 months of which overlaps with the present study. They found a nearly in-phase relationship of zonal velocity, thermocline temperature, and dynamic height which propagated eastward at approximately 210 cm s. The eastward phase propagation can be seen visually in Figure 3a for dynamic height 0/250 dbar and in Figure 3b for dynamic height 80/250 dbar (in the thermocline near the mean position of the Equatorial Undercurrent core). Well-defined peaks and troughs at 140W lead those at 110W by about 2 weeks, implying phase speeds in approximate agreement with those reported by McPhaden and Taft [1988]. One can also see the expected Kelvin wave relationship between zonal currents and 20C isotherm depths by comparing Figures 2a and 4. The 2- to 3-month periodicity is most apparent in zonal currents in the thermocline at depths below the Undercurrent core where the seasonal cycle and the 20- to 30-day instability waves are of diminished amplitude. By overlaying the 20C isotherm depth time series in Figure 2a and the zonal currents at, say, 120 m in Figure 4, one would find that on intraseasonal time scales a depression of the thermocline corresponds to enhanced eastward velocity and vice versa. This is also evident in the decadal time series shown in Plate 2. There is no obvious evidence in the local winds for the energetic 2- to 3-month periods observed in the oceanic data sets (Figure 2), implying that the variability must be nonlocally forced as previously inferred by Enfield [1987] and McPhaden and Taft [1988].

Figure 3. Time series of dynamic height (a) at the surface relative to 250 dbar and (b) at 80 dbar relative to 250 dbar. The 0, 125W time series is indicated by a dashed line for clarity. Also shown are the dynamic height differences over 110-140W superimposed on the mean seasonal cycle of dynamic height difference. Wind stress at 0, 165E (solid line) and 2, 165E (dashed line) appear in Figure 3a superimposed on the Wyrtki and Meyers [1975] stress climatology. Daily data have been smoothed with an 11-day Hanning filter.

Figure 4. Time series of zonal velocity in the upper 250 m superimposed on estimates of the monthly mean seasonal cycle. Daily data have been smoothed with an 11-day Hanning filter. Shown at night are the depths of the velocity measurements relative to the 5-year mean profile of zonal velocity (solid line) and temperature (dashed line). Depths on the ordinates are for January 1986 to October 1987 (left) and October 1987 to October 1988 (right).

Figure 3 also shows the difference between dynamic height at 110W and 140W, which is proportional to the zonally averaged zonal pressure gradient across these longitudes. The mean seasonal cycle indicates dynamic height differences at the surface (0/250 dbar) and in the thermocline (80/250 dbar) that are weakest in the spring and strongest in the fall, in approximate balance with the mean seasonal cycle of zonal wind stress [McPhaden and Taft, 1988]. During 1986-1988, dynamic height differences are modulated by intraseasonal wave activity with peak-to-peak amplitudes larger than the annual range. There are in fact a few occasions of 1 month duration when the height difference across 110W-140W nearly vanishes because of these waves. Height differences vary interannually as well across 110W-140W: magnitudes are larger than normal in 1986 and smaller than normal in 1988, during which near-zero values are observed for several consecutive months in the thermocline in the early part of the year.

Dynamic height differences on a smaller zonal scale can be inferred from additional data at 0, 125W during 1987-1988. Differences are generally smaller between 110W and 125W than between 125W and 140W, and there are periods of 1 month when the 0-/250-dbar and 80-/250-dbar surfaces are 1-2 dyn. cm higher at 110W than at 125W. This is consistent with the zonal variations in thermocline slope along the equator in 1988 as seen in Plate 4 and implies a weaker driving force for the Equatorial Undercurrent at 110W as compared to 140W.

4.4 Undercurrent Variations

Figure 4 shows time series of zonal velocity in the upper 200-250 m at 110W. The location of the current meters relative to the mean current and temperature structure at 110W is shown at night. There is a significant amount of 20- to 30-day instability wave energy in these time series which tends to obscure lower-frequency fluctuations. However, two periods of pronounced Equatorial Undercurrent weakening are evident. The first is February-March 1987 in the upper 250 m when flow is on average 25-50 cm s weaker than normal; the second is December 1987 to December 1988 at 80-120 m (in and below the mean depth of the Undercurrent core) when flow is 20-25 cm s weaker than normal. In part these reductions in eastward flow at fixed levels are due to vertical excursions of the thermocline in which the Undercurrent is embedded (compare the upper and lower panels of Plate 2). Depression of the thermocline in early 1987 leads to anomalously westward flow above the Undercurrent core, whereas elevation of the thermocline in 1988 leads to anomalously westward flow below the Undercurrent core. However, these westward flow anomalies are not entirely compensated by eastward flow anomalies below the Undercurrent core in early 1987 or above the Undercurrent core in 1988, as would be expected if vertical displacements were the only influence on the velocity records. Thus there is a large net reduction in transport per unit width during February-March 1987 and during 1988 relative to the normal seasonal cycle at 110W (Figure 5a). The prolonged reduction in 1988 moreover is consistent with diminished isothermal spreading in the thermocline near the equator (Plate 3), and diminished zonal pressure gradient at the surface and in the thermocline (Figure 3). At 140W in contrast, the transport per unit width is near normal during most of 1988 (Figure 5b), possibly due to the presence of stronger zonal pressure gradients in the thermocline between 140W and 125W vis--vis between 125W and 110W, as suggested by Plate 4 and Figure 3b.

Figure 5. Zonal transport per unit width (10-200 m) at (a) 0, 110W and (b) 0, 140W superimposed on estimates of the monthlv mean seasonal cycle. Daily data have been smoothed with an 11-day Hanning filter.

The reduction in Equatorial Undercurrent speed and eastward transport at 110W in February-March 1987 is mirrored in an even more dramatic collapse of eastward transport in the upper 200 m at 140W about 1 month earlier (Figure 5b). This collapse roughly coincides with the dynamic height trough associated with the strongest cycle of intraseasonal wave variability during 1986-1988. Figures 6 and 7 show the vertical profiles of velocity and temperature for February 1987, which is the period bracketed by the minimum transports at 110W and 140W. Also plotted are February profiles for 1983-1988 at 110W and 1984-1988 at 140W. It is clear that February 1987 zonal currents are the weakest of the 5 to 6-year period shown in Figures 6 and 7, with the exception of February 1983 at 110W during the previous ENSO when the Undercurrent disappeared [Halpern, 1987a]. The Undercurrent does not disappear during the 1986-1987 ENSO, but in February 1987 it is significantly reduced in speed especially at 140W, where its core velocity is only 45 cm s . It is also evident that this reduction is not due to a meridional advection of the Undercurrent off the equator. Average meridional velocities in February 1987 are near zero in the thermocline and are not significantly different from the meridional velocities during other years.

Figure 6. Profiles of monthly mean zonal velocity, meridional velocity, and temperature in the upper 250 m for each February from 1983 to 1988 at 0, 110W.

Figure 7. Profiles of monthly mean zonal velocity, meridional velocity, and temperature in the upper 250 m for each February from 1984 to 1988 at 0, 140W.

4.5 Relationship of Zonal Currents to Surface Temperature

Typically at 110W, flow at 10 m is eastward from March to June (the springtime reversal of the South Equatorial Current) and westward the remainder of the year ( Figure 4). However, in 1986 the springtime reversal in March-June is 31 cm s stronger to the east, and the South Equatorial Current in September-November is 32 cm s weaker to the west. (See Figure 10 for low-pass versions of 10-m zonal currents and SST at 110W.) Both these periods correspond to periods of enhanced eastward transport as well ( Figure 5a).

The stronger than normal springtime reversal occurs when the magnitude of the zonal SST gradient is typically weakest along the equator (e.g., -T/x ~ 0.2-0.3 10 C m during boreal spring between 110W and 140W based on Reynolds' [1988] climatology), so that one might expect anomalous advective warming during this time of <1C. Consistent with this, the observed SST anomaly was 0.5C during March-June 1986 at 110W; allowing for a 1-month lag in response to changing zonal currents, the anomaly was 0.7C during April-July 1986. These are of questionable significance given the uncertainties in estimating anomalies as discussed in section 2 and Appendix B. Conversely, during the boreal fall, SST gradients along the equator between 110W and 140W are typically strongest (i.e., -T/x ~ 0.8-1.0 10 C m). At 32 cm s zonal current anomaly acting on a gradient of this magnitude for 3 months would produce a warm SST anomaly of approximately 2C, comparable to the average 1.4C (1.3C) anomaly actually observed in September-November (October-December) 1986. These observations are consistent with the model results of Harrison and Schopf [1984], who argued that advective warming in the eastem Pacific by anomalous currents should be more pronounced in boreal fall than in boreal spring because of the relative strength of the zonal SST gradients during these seasons. Note also that in boreal fall 1986, anomalously weak westward flow was accompanied by a thermocline depression of 20-40 m at 110W (Figure 2a). Hence local upwelling, which would continue unabated because of near-normal local easterlies, would bring warmer water to the surface and contribute to the development of SST anomalies. A similar sequence of events occurred at 140W in 1986 (not shown), with weaker than normal westward flow in the South Equatorial Current and a deeper than normal thermocline associated with the development of SST anomalies of >1C.

Zonal current anomalies at 10 m are less obviously related to SST anomalies at other times during 1986-1988. In particular, rapid cooling that occurs during the termination of the warm event in early 1988 coincides with a near-normal eastward springtime reversal at 110W and 140W (Figures 4 and 10). At 110W the reversal starts late but finishes strong such that the eastward flow from March to June 1988 is within 1 cm s of the mean seasonal average. Thus SST cooling of nearly 8C from March to May at 110W cannot be due to surface zonal advection, which would tend to produce weak warming. It is more likely that uplift and entrainment of thermocline waters into the surface layer as suggested by the rapid coincidental shoaling 20C isotherm in Figure 2 is the mechanism generating anomalously cold SSTs in early 1988. This upwelling may in part be related to the anomalous zonal convergence in eastward mass transport that develops in early 1988 between 110W and 140W (Figure 5). In the mean, the equatorial SST minimum is maintained by eastward transport of cold thermocline water which is continuously upwelled and transported poleward in the surface layer [Bryden and Brady, 1985]. Mass continuity requires that anomalous zonal convergence be balanced by enhanced divergence on the meridional plane, which could intensify this upwelling circulation and lead to additional surface cooling. Consistent with this hypothesis, the coldest SST anomalies in 1988 were found between 110W and 140W as, for example, can be seen in the June 1988 chart shown in Plate 1.


Go back to previous section or forward to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage