of the weakly
ongoing deceleration of the EUC in August seems to be related to forcing to
the east of the region and Rossby wave propagation.U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications
A primitive equation general ocean circulation model has been used to explore the processes that may have been responsible for the dramatic temperature and current changes observed on the equator at 159°W in 1982-83. Although the SADLER hindcast experiment using the wind stress analysis of J. Sadler at the University of Hawaii does not perfectly reproduce the observations, it offers as good a point comparison between model results and data as has been published for the tropical Pacific. All of the major observed features have counterparts in the hindcast, with comparable amplitudes. For this reason we judged it defensible to study the model results in order to try to make inferences about the processes responsible for the behavior of the ocean during this time.
A series of experiments was conducted to explore the importance of local and remote forcing in the SADLER hindcast experiment. Neither local nor remote wind stress changes alone suffice to account for the observed behavior. LOCAL forcing alone can account qualitatively for the character of the variability in the zonal current records but does not reproduce the thermal field evolution nor the quantitative current changes well. Remote forcing to the east of 160°W seems to account for only a very small portion of the remote forcing response, and meridional wind stress at and east of the date line has little effect. Almost all of the remotely forced response at 160°W is propagated into the region from the west.
The idealized remote forcing experiments illustrate the importance of vertical modes higher than the second nonlinearity in the response through alteration of the stratification under the forcing event as the forcing proceeds, zonal advective processes and mean flow interaction. The more realistic the forcing event and the background field on which it acts, the more the elements of the response resemble those seen in SADLER during the periods of EUC deceleration, easterly surface jet formation and decay, and thermocline uplift and intensification. The key response element is that of higher vertical mode forcing, although the various elements of nonlinearity considerably shape the details of the response. It has become rather common to think of the basic remote response of the equatorial upper ocean east of a westerly wind event as consisting primarily of a first vertical mode response with eastward acceleration from the surface to at least the thermocline and a deepening of the thermocline. The idealized numerical model experiments indicate clearly that this commonplace scenario is not correct within at least 30° of longitude east of the forcing region. Westward acceleration in the thermocline can equal or exceed eastward near surface acceleration, and thermocline shallowing and intensification can figure as strongly as deepening.
In previous numerical studies that have concentrated on the eastern equatorial Pacific response to remote westerly event forcing (e.g., Harrison and Giese 1988), attention has been focused typically on the first and second baroclinic Kelvin modes. Observational support for the remote forcing of Kelvin response in the equatorial Pacific has been based largely on sea level variability, in which only the first baroclinic mode typically has enough amplitude to be visible. The first and second modes seem to provide most of the eastern Pacific SST variability in the model studies, both because there is substantial surface velocity response in them and because the numerical model mixing physics (Pacanowski and Philander 1981) seems to damp out third and higher modes from reaching the eastern Pacific. However, the third, fourth, and fifth vertical modes are strongly forced by wind events over stratifications typical of the numerical model ocean and of the western Pacific according to linear theory (Giese and Harrison 1990), and their subsurface zonal velocity and vertical velocity amplitudes can be substantial near to the forcing region, as we have shown. For central Pacific response it is clear that even with the strong Richardson number-dependent mixing of the numerical model, third and higher Kelvin modes can play a major role in the model response.
The dramatic deceleration and flow reversal of the EUC from July 1982 to November
1982 at 160°W is the result of changes in the local winds, zonal wind variability
west of the region, and to a very much more limited degree, meridional wind
variability east of the region. Kelvin-type response to remote westerly forcing
plays a major role in the July deceleration, even though its signature in the
zonal momentum equation balance of terms is not clearly the passage of an eastward-propagating
Kelvin pulse. Our studies do not enable us to sort out the relative importance
of discrete westerly event forcing (e.g., June 1982 near the date line), compared
with the eastward-propagating zonal stress pattern west of 180° from April through
June 1982, but we note that each of the primary elements of the observed July
response is reproduced in our most realistic westerly event forcing experiment.
The resonant Kelvin-type response mechanism discussed by McCreary
and Lukas (1986) is not essential to account for the observed behavior.
LOCAL zonal wind changes also tend to decelerate the EUC, although very much
more slowly than observed; this occurs over about 3 months rather than 1 month.
No more than about 20 cm s of the weakly
ongoing deceleration of the EUC in August seems to be related to forcing to
the east of the region and Rossby wave propagation.
The formation and eventual degradation of the surface-trapped jet in November and December 1982 also result from both local and remote forcing. The local westerly winds that began in late October 1982 accelerate the surface flow eastward and remote forcing from the westerly event in November from near the date line out to at least 150°E provides zonal advection and thermocline vertical advection that give it its full vertical penetration. The jet accelerates eastward its entire duration against the zonal pressure gradient. Its deceleration results from horizontal diffusion, vertical diffusion, and the zonal pressure gradient, and so is consistent even with rather simple Kelvin forcing dynamics.
Changes in the thermal structure in the SADLER hindcast are large when compared with the LOCAL or LOCAL + EAST experiments, indicating that most of the temperature variability in the SADLER hindcast comes from remote forcing west of the region. The deepening thermocline and appearance of warm surface water in July 1982 results from forcing to the west and corresponds to the Kelvin response that rapidly decelerates the EUC in July 1982. We found that the heating is caused primarily by zonal advection rather than vertical advection, apparently because the zonal temperature gradient in the thermocline is large in this region, while the vertical temperature gradient near the surface is modest. The cooling and sharpening of the thermocline in November and December 1982 also result from remote forcing, apparently the November westerly event followed by continuing weakening of the westerlies in December.
By and large we find that meridional wind stress variation has little impact on the equatorial zonal flow and thermal fields at 160°W; zeroing the meridional stress locally or east of 160°W has little impact on our results. We also find that little response is forced by zonal wind stress variations east of 160°W. This finding is not inconsistent with the recent idealized coupled-model results that suggest a delayed oscillator process involving both Kelvin and Rossby modes of response as a model for the ENSO cycle, because in at least some of these studies, the Rossby-Kelvin mode forcing and propagation/reflection happen primarily between the central Pacific and the western Pacific. The eastern waveguide does not play a fundamental role. In our studies, in which the initial condition is climatological January, there is no preexisting Rossby or Kelvin energy in the waveguide left over from a previous ENSO; all of our response is forced by 1982-83 wind stress changes. This study was not intended to explore the plausibility of the delayed oscillator mechanism and our results, while not inconsistent with these mechanisms, also do not provide any specific support for them.
These studies illustrate how challenging it is to clearly identify the effects of even strong remote forcing when there is local forcing variability and when nonlinearity is important. A number of experiments were necessary to understand the relative importance of local versus remote forcing, zonal versus meridional wind forcing, and linear versus nonlinear Kelvin wave response. In the end, a clearer and more detailed picture of the processes responsible for the large variability of the central equatorial Pacific during the heart of the 1982-1983 ENSO event was obtained. Future observational efforts to follow the waveguide response to westerly wind changes will clearly require detailed knowledge of the forcing across the waveguide as well as of the density structure in the forcing region if they are to have much hope of exploring the extent to which even simple Kelvin response ideas pertain to the actual response.
Acknowledgments. This work was supported by NOAA/PMEL, by the NOAA EPOCS project, and by NASA UPN-578-22-13-02. Computing was carried out on the NIST CYBER 205 at Gaithersburg, Maryland. The assistance of the TMAP group (Steve Hankin, Kevin O'Brien, and Jerry Davison) and discussions with Dr. Ben Giese are also appreciated.
Return to previous section or go to the next section