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

Dynamics of seasonal and intraseasonal variability in the eastern equatorial Pacific

Michael J. McPhaden and Bruce A. Taft

NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington

Journal of Physical Oceanography, 18(11), 1713-1732 (1988)
Not subject to U.S. copyright. Published in 1988 by the American Meteorological Society.

6. Discussion and conclusions

Time series measurements from surface moored buoys in the eastern and central equatorial Pacific have been analyzed for the period 1983-86. The data consist of currents, temperatures, and winds on the equator at 110°, 124.5° and 140°W. The purpose has been to examine the dynamics of upper ocean variability on seasonal and intraseasonal time scales. We have worked primarily with the depth integrated zonal momentum equation to avoid problems with poorly known vertical eddy viscosities and coarse vertical resolution of currents.

We have found that in the mean, a balance exists between depth integrated zonal pressure gradient and zonal wind stress to within 20%, in approximate agreement with linear Sverdrup theory. Conversely, we found significant mean vertical and zonal advection of momentum on the equatorial plane. This advection in part is related to the rise of the EUC towards the east in the thermocline as indicated by the tendency for the two terms to cancel one another. However, the cancellation is incomplete and the net effect of nonlinearity is to reduce the surface westward flow via upward advection of eastward momentum from the EUC. This results in larger upper-ocean eastward transports along the equator than expected from linear theory, consistent with the transport calculations shown in Fig. 6. For comparison, linear theory (e.g., McPhaden 1981; McCreary 1981) would predict nearly equal and opposite transports in the SEC and EUC, with much smaller net transport than we observed.

We found that SST was warmer by about 1°C, the thermocline was deeper by 30-40 m and dynamic height was higher by about 10 dyn cm at 140°W in 1985-86 compared to 1983-84. Likewise an interannual increase in SST was noted at 110°W beginning in 1983. These observations are consistent with those of Wyrtki (1984) and Kessler and Taft (1987) who found that the eastern and central equatorial Pacific were unusually cool in the aftermath of the 1982-83 El Niño. The interannual changes in SST we have observed between 1983 and 1986 may be related to a doubling of the zonal wind stress and a concomitant doubling of the depth integrated zonal pressure gradient. Increased easterly winds, in more strongly tilting the thermocline down towards the west, would remove the cold water reservoir to great depths and thus might reduce the efficiency of upwelling to cool the surface. This effect seems to have dominated the tendency for enhanced cooling by stronger meridional Ekman divergence and stronger advection of cool water from the east that one would expect to be associated with intensified easterlies.

Near surface zonal currents and transport appear to propagate westward along the equator at the annual period between 110° and 140°W. Although from our data we cannot determine with accuracy the magnitude of the phase speed, the inferred direction of propagation is not unexpected. Both Meyers (1979) and Lukas and Firing (1985) show from climatological ship wind data that zonal wind stress propagates westward at 1 cpy in the eastern Pacific. Also, Lukas and Firing (1985) found evidence in central Pacific temperature data for an annual, westward propagating Rossby wave which presumably has a signature in surface currents at the equator. In contrast Hayes and Halpern (1984) inferred eastward propagation of transport between 110° and 91°W (Galápagos Islands) from 16 months of current and sea level data in 1980-1981. Their data were dominated by two large eventlike features which they interpreted as first baroclinic mode Kelvin waves generated by westerly winds in the western Pacific. These events, though they happened to occur in boreal spring of each year, were probably not part of the seasonal cycle. From longer sea level and wind records, Eriksen et al. (1983) found that similar events occur at other times of the year. The greater prominence of remotely forced, eastward propagating Kelvin waves to the east of 110°W may be related to a decrease in amplitude of the annual (i.e., 12 month) period in local zonal wind stress forcing east of 110°W compared to west of 110°W.

Diagnosis of the zonal momentum equation at 1 cpy indicates that between 110° and 140°W, local acceleration is always negligible and that the zonal pressure gradient and wind stress tend to balance to within the errors of our analysis. Thus, the eastern equatorial Pacific appears to be responding in a succession of steady states to wind stress forcing at the annual period. In modeling the annual period, Philander and Pacanowski (1981) indicate that eastward current transport in the upper 250 m should be strongest when the zonal pressure gradient and wind stress are most negative, i.e., in boreal fall. The exact opposite is observed in our data: eastward current transport is largest in the spring when the pressure gradient and easterly winds are weakest. The discrepancy may be due to lateral eddy momentum fluxes associated with instability waves which were not evident in Philander and Pacanowski's (1981) simulations. These instabilities lead to an effective stress along the equator of O(-10 × 10 n m) (Hansen and Paul 1984; Bryden et al. 1986). Moreover, the amplitude of the instability is largest in boreal summer and fall and weakest in boreal spring (q.v. Fig. 3 and Philander et al. 1985). This suggests an annual variation in lateral eddy stress which should retard the eastward flow of the EUC more strongly in the summer and fall than in the spring.

No consistent patterns emerged from the analysis of 2 cpy variability because of the general weakness of the signal in our records. However, at higher frequencies we found evidence for eastward propagating waves in currents, temperature, and dynamic height with periods of 60-90 days. These waves have no signature in meridional velocity and are poorly correlated with local wind forcing. Velocity, temperature, and dynamic height variations are nearly in phase and tend to be vertically coherent in the upper 250 m. Eastward phase speeds are indistinguishable from those of first baroclinic mode Kelvin waves. Analysis of the momentum balance suggests that these waves are Kelvin-like, though they may be affected by nonlinearity. Their amplitudes are large enough at times to completely obscure the annual cycle as occurred, for example, in late 1984 and early 1985.

Enfield (1987) has documented 40-60 day fluctuations in sea level propagating eastward across the equatorial Pacific and poleward along the coasts of North and South America. He interprets these as first baroclinic mode Kelvin waves excited by atmospheric 40-50 day waves (e.g., Madden and Julian 1971) which in the Pacific are most energetic west of the dateline. The period of the 60-90 day Kelvin-like waves documented in this study is longer than observed in Enfield's study. However there is reason to believe that we are observing essentially the same wave phenomenon. The data in our study overlap with Enfield's, which encompasses the period 1979-85. For this time period Enfield finds that the wind forcing and the waves are nonstationary in time and in frequency. Shortly after then 1982-83 El Niño, for example, intraseasonal fluctuations of zonal winds in the western Pacific and sea level across the Pacific weakened (q.v. his Fig. 4). Subsequently, in the latter half of 1984, zonal winds and sea level fluctuations reestablished themselves. This same pattern can be seen in currents, temperature, dynamic height, and zonal pressure gradients computed from the mooring data (e.g. Figs. 3, 8b, 10b). In addition, sea level spectra presented in Enfield (his Fig. 6) suggest that after the 1982-83 El Niño, the period of the waves increased to about 70 days. To verify this, we calculated the periodogram and energy spectrum of Christmas Island sea level data from 1983-85 and found a large peak between 60 and 90 days with maximum periodogram estimate at a period of 73 days. Thus we conclude that the 60-90 day waves observed in the mooring data are essentially the same phenomenon that Enfield documented in sea level, but shifted to lower frequency during the time period of our analysis.

Acknowledgments. The authors would like to thank P. Freitag and M. McCarty for processing the mooring and XBT data used in this analysis. We are also indebted to A. Shepherd for his careful preparation and calibration of the VACMs and wind recorders for each deployment; and to D. Fenton of SeaMarTec, Inc., who oversaw mooring assembly, deployment, and recovery procedures. We would also like to thank W. Kessler, E. Johnson, D. Enfield, and S. Hayes for comments on an earlier version of this manuscript. K. Wyrtki kindly provided the sea level data from Christmas Island. This research was conducted as part of NOAA's Equatorial Pacific Ocean Climate Studies (EPOCS) Program.

Go back to previous section or forward to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage