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


Variability of the sea surface temperature in the eastern equatorial Pacific during 1986-88

S.P. Hayes

Pacific Marine Environmental Laboratory, National Oceanic and Atmospheric Administration, Seattle, Washington

Ping Chang

Joint Institute for the Study of the Atmosphere and Ocean, Department of Atmospheric Sciences, University of Washington, Seattle, Washington

M.J. McPhaden

Pacific Marine Environmental Laboratory, National Oceanic and Atmospheric Administration, Seattle, Washington

Journal of Geophysical Research, 96(C6), 10,533-10,566 (1991)
Copyright ©1991 by the American Geophysical Union. Further electronic distribution is not allowed.

5. Discussion and Conclusions

Time series of surface wind and upper ocean temperature and velocity have been used to describe the variations of the mixed layer temperature for the period January 1986 to June 1988. This period included the development of the 1987 ENSO warm event and the subsequent cooling. Mixed layer heating was compared to estimates of the surface heat flux using a combination of the moored measurements and climatology. Oceanic heat transports were parameterized and estimated from the moored measurements. Given the uncertainties in all the estimates and the approximations involved in the parameterizations, the agreement between the changes in mixed layer temperature and the net heating of this layer is remarkably good (Figures 6e and 11). The major features of the temperature changes in 1986 and through the first half of 1987 are reasonably well represented. However, in late 1987 and in spring 1988 temperature change and heat flux were out of phase. In particular, the cooling which followed the 1987 warm event was not well described. During this period the mixed layer was very shallow and the assumptions used in the calculations were not valid.

Although no single term in the temperature (1) dominated the mixed layer heating, the most important terms in the mean balance were the net incoming surface heat flux and the vertical flux out the bottom of the mixed layer. Both the penetrative solar radiation (Q) and the vertical turbulent flux (Q) contributed to the latter. The mean net heat flux at the surface was 165 W m and the sum of Q and Q was 180 W m; thus the mean heat input at the ocean surface nearly balances the vertical flux through the mixed layer.

Fluctuations in Q were also quite important in the variability of the mixed layer heat content. A decrease in the vertical turbulent flux out of the mixed layer contributed about half of the heat flux change which led to the warming in September-October 1986, and an increase in Q could account for nearly all of the cooling in May 1987. Overall the vertical turbulent heat flux variation was well correlated with the mixed layer heating (r = 0.5), and the rms amplitude of the Q and Q fluctuations were comparable.

Vertical entrainment of cool water into the mixed layer also plays an important role in the variability of the mixed layer temperature (Figure 6c). This term is well correlated with Q (0.64) and inclusion of Q in Q increases the correlation coefficient between Q and Q from 0.5 to 0.7. Our estimate of the entrainment velocity is quite crude and fails particularly when the mixed layer is very shallow, such as in spring 1988. From March to May 1988 the 20C isotherm rose 50 m and in the daily averaged time series actually broke the surface (Figure 2). This upwelling likely contributed substantially to the SST cooling observed at that time. However, with the data available there is no way to accurately estimate this vertical advective cooling. Figure 6e suggests that about -60 W m would be required to bring Q and Q into balance during this period. The mean upwelling of the thermocline from March to May was about 1 m d (1.1 10 m s), and the temperature change from the surface to 5 m (our minimum mixed layer depth) was 1C. Thus, from (8) the vertical entrainment cooling could easily contribute -50 W m.

Zonal advection was a significant contribution to the heat flux variability. The seasonal cycle of the surface current [Halpern, 1987; McPhaden and Taft, 1988] leads to advective warming in boreal spring and cooling in the fall. The total change between the two seasons was about 75 W m in 1986. However, the relation of the advective heat flux to the mixed layer temperature change is not simple. Maximum SST and eastward surface current were nearly in phase; hence the springtime warming of the mixed layer occurred before the surface current reversed. Thus on seasonal time scales, the advective heat flux tended to be out of phase with the warming. During the onset of the 1986-87 ENSO warm event in the eastern Pacific in late 1986, reduced westward surface current contributed to the warming as suggested in MH. However, the seasonal cycle persisted through the 1987 warm event, and the increasing mixed layer temperature in January-March 1987 was opposed by advective cooling. The large heating in January 1987 coincided with the deepening of the thermocline associated with the arrival of a Kelvin pulse (Figures 8b and 9). This signal, however, had little surface zonal velocity expression and hence was not apparent in the zonal heat advection.

Meridional advection and diffusion were examined during December to May of the 3 years considered. Since meridional SST gradients are generally weakest during boreal spring, the overall importance of the meridional terms may be underestimated in our results. On the other hand, the eddy coefficient used was derived by Hansen and Paul [1982] based on measurements in boreal summer and fall when the tropical instability waves are largest. Hence this coefficient is likely to be too large.

The meridional advective heat flux was generally small, perhaps because the site is directly on the equator. This term can, however, contribute to the intraseasonal fluctuations. For example, meridional advection supplied nearly 100 W m in January 1987 during the passage of the Kelvin wave event. As noted above, this contribution from the meridional velocity is, presumably, indicative of the effects which the Kelvin signal has on the instability waves. In the model studies [Geise and Harrison, 1990] the magnitude of the meridional term depends critically on the amplitude and phase of the instability waves at the time of the event.

Meridional diffusion is highly parameterized in (10). However, if the diffusive heat flux is proportional to the meridional curvature of the temperature field, then the phase of this term is reasonably well represented in our analysis. Meridional diffusion appears to be important in the seasonal heating. In December-January this term contributes 40-50 W m at the time when the mixed layer is most rapidly warming. In 1986 and 1988 this meridional term was crucial in order to obtain a near balance in the January heat budget. In February-March of each year the equatorial cold tongue nearly vanished at 110W and the temperature curvature was a minimum. At this time the instability waves disappeared and the diffusive heat flux was near zero. In April-May the equatorial cold tongue returned and meridional diffusion tended to warm the equator. This warming is counter to the general cooling trend in the mixed layer and slows the development of the cold tongue.

In summary, the picture which emerges from this analysis of the eastern equatorial Pacific heat budget is one in which many heat flux terms are important. The local change in mixed layer temperature cannot be ascribed to any single process for the duration of the record. The relatively simple correlations between SST changes and local wind on seasonal and interannual time scales (Figure 3) is the result of a complicated interaction of variations of surface fluxes and oceanic processes. Our conclusion is similar to that of Weingartner and Weisberg [1991], who found that the seasonal upper ocean heat budget in the equatorial Atlantic was also the result of the interplay of several terms. Our analyses relied heavily on parameterizations of both the atmospheric flux terms and the oceanic processes. A more quantitative study of the mixed layer heat budget will require more accurate measurements, and some improvement of the parameterizations, of these terms.

Acknowledgments. We would like to thank C. Deser for providing the COADS data set and her analysis of climatological surface parameters. We are grateful for the assistance of L. Mangum, N. Soreide, and L. Stratton in the analysis and the computations. This work was supported, in part, by the Equatorial Pacific Ocean Climate Studies (EPOCS) Project of ERL/NOAA through grants to PMEL (S.P.H., M.J.M.) and JISAO, University of Washington (P.C.). NOAA Pacific Marine Environmental Laboratory contribution 1160.


Go back to previous section or forward to next section

Go back to Abstract

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage