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


Ocean Model Studies of Upper-Ocean Variability at 0°N, 160°W during the 1982–1983 ENSO: Local and Remotely Forced Response

D.E. Harrison

NOAA, Pacific Marine Environmental Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115

A.P. Craig

School of Oceanography, University of Washington, Seattle, WA 98195

Journal of Physical Oceanography, 23(3), 426-451 (1993)
Copyright ©1993 American Meteorological Society. Further electronic distribution is not allowed.

5. Dynamical balances in the experiments

A reasonably clear picture of the importance of forcing locally, from the east, and from the west of 160°W has emerged from the previous experiments. The results of these studies have been discussed primarily in terms of the zonal flow and temperature field variations. The numerical model, however, makes it possible to examine all of the terms in the zonal momentum and temperature equations and to determine how the balance of terms changes during the different phases of the ENSO event, or in different stages of the idealized forcing studies. The zonal momentum and heat equation balances have been scrutinized for each experiment. In the interest of brevity these results will not be presented for most of the experiments; the results of all experiments are available for the interested in Harrison and Craig (1992).

Although these balances are not essential to any of the previous results, a brief summary of their features may be of interest, and serves to illustrate some of the behavior that may have to be planned for when designing an equatorial local dynamics experiment. The idealized forcing experiments in an initially resting ocean typically have rather simple balances away from the forcing region. In WBex0.02 the balance is classic linear Kelvin physics; local zonal acceleration results from the zonal pressure gradient term and local temperature change results from vertical advection. In WBwp0.2 the same balances hold mostly, although zonal advection of zonal momentum can be important near the surface when the surface jet is present at the date line. Nonlinearity is present in WBwp0.2 both through the presence of the eastward surface jet and in the projection of the forcing onto the vertical modes; the forcing is strong enough to alter the vertical stratification in the forcing region so that the response is not that predicted in detail by linear theory (see Giese and Harrison 1990).

In WBcl0.2 the balances are more complex, because there is a preexisting circulation. In the climatological mean circulation the balances are rather involved, but will be touched on in the SADLER discussion below. The reader particularly interested in the mean and seasonal cycle balances under climatological forcing should consult Harrison and Hankin (1992) in which these balances are discussed in detail. The presence of instability waves means that the remote response is not so easily observed as in the resting state experiments, and the presence of the remote response causes a phase shift in the instability waves that would otherwise exist so that even subtracting the climatological flow does not greatly simplify looking at the balances. The first baroclinic mode response, with its vertically coherent zonal pressure gradient signal, can be located at the appropriate time, but subsequent modal zonal pressure gradient signals are difficult to discern. The local zonal flow acceleration is typically a small difference between several much larger terms and, with the instability wave signal also present and of comparable or larger magnitude to the remote response, the remote response is not easy to identify at 160°W. Even under these ideal model circumstances, it is difficult to isolate the dynamical signatures of the remote response. In the ocean, where instability wave amplitudes are substantial, it would be a very considerable observational challenge to isolate the remote response even with a very good notion of what one was looking for.

There is a recognizable dynamical signature of the remote response at 180°. The zonal acceleration shows the characteristic pattern of an eastward acceleration down through 250 m followed by thermocline westward acceleration and then weak eastward acceleration. Although there is a clear pressure gradient tendency of the opposite sign to the local acceleration, there is also a strong zonal advection component, particularly during the period of thermocline westward acceleration. There are also vertical advection changes of up to 40 cm s/mo between 80 m and the surface during and slightly after the thermocline westward acceleration period. Clearly, simple Kelvin processes are altered in the presence of a mean circulation when the forcing is as strong as in WBcl0.2.

Turning finally to the simulation experiments, let us first consider heat equation balances. Figures 17 and 18 present the heat equation terms for the SADLER and LOCAL experiments, respectively, with the same contour levels. The plotting convention is such that d/dt minus all the other terms equals zero, that is, zonal heat advection is -udT/dx, etc. Tendencies smaller than 2°C mo are not contoured, to facilitate examination of the dominant change processes. In the SADLER balances there is strong surface-to-thermocline warming in September 1982; there is strong thermocline cooling in December 1982 and January 1983. The July 1982 warming results almost entirely from zonal heat advection; in August 1982 there is an episode of near-surface cooling tendency from vertical advection that is balanced out by the other terms so that there is actually no substantial warming or cooling. The December 1982-January 1983 thermocline cooling results almost entirely from vertical heat advection in December, and is somewhat resisted by a warming tendency from meridional heat advection in January.

Figure 17. The individual temperature equation terms for the SADLER experiment at (0°, 160°W). The convention is that local temperature change = all other terms, so,for example, the zonal advection tendency is -u*dt/dtx. Only warming and cooling tendencies greater than 2°C mo are contoured, to emphasize the major change periods. Positive (negative) values correspond to warming (cooling) tendencies. Contour interval is 2°C mo. (a) Zonal velocity at (0°, 160°W) from the B32 experiment, (b) temperature at (0°, 160°W) from the B32 experiment. The contour intervals are 5 cm s and 2°C.

Figure 18. The individual temperature equation terms at (0°, 160°W) for the LOCAL experiment, as in Fig. 17.

We have seen that remote forcing plays a significant role both in the July 1982 warming and the November-December 1982 thermocline cooling (uplift). How well can we relate simple Kelvin process ideas to these thermal balances? The simplest Kelvin balances involve heating or cooling entirely through vertical advection, but this balance assumes a horizontally uniform resting ocean. In the presence of zonal temperature gradients, zonal advection of heat from the Kelvin zonal flow can be significant. Vertical advection of heat from Kelvin upwelling or downwelling will be noticeable whenever there is significant vertical temperature gradient. In the SADLER experiment, the initial eastward surge advects warmer western Pacific water eastward and downwelling provides a little thermocline depression and warming. We saw above that zonal advection of heat dominates the SADLER balance during July; this is a rather extreme form of what might be expected from remotely forced Kelvin processes and apparently occurs because there is little upper-ocean stratification. The November-December thermocline cooling, which is dominated by vertical advection, is much more what would be expected from simple Kelvin-type processes.

In the LOCAL heat equation balances, the only significant warming on the scale of what occurred in SADLER takes place near the surface in late September-October 1982; there is no cooling of comparable magnitude in December-January. Nontrivial thermocline zonal advection warming tendency in July and August-September is offset by a vertical advection cooling tendency. Nontrivial cooling tendency from vertical advection in January-June 1983 is offset by a meridional heat advection warming tendency between January and May and by a zonal advection tendency in June 1983. Horizontal and vertical heat diffusion is not significant on these scales in the July 1982-March 1983 period. The LOCAL balances are very different from the SADLER balances, further confirming that very different processes were at work in the two experiments.

The SADLER momentum equation terms are shown in Fig. 19. To facilitate identifying the major features in these complex plots, periods of eastward and westward acceleration in excess of 100 cm s/mo are shaded light and dark, respectively. Before and after the large ENSO changes, say June 1982 and after March 1983, an eastward tendency from the zonal pressure gradient from the surface through the thermocline and an eastward tendency from vertical advection above the thermocline tend to be balanced by a westward tendency from vertical diffusion (which includes the surface wind stress) above the thermocline, and by zonal advection and horizontal diffusion below the thermocline. The mean and seasonal variations of the near-equatorial zonal momentum balances are discussed in much more detail in Harrison and Hankin (1992).

Figure 19. The individual zonal momentum equation terms at (0°, 160°W) for the SADLER experiment. (a) local zonal acceleration, (b) zonal advection, (c) vertical advection, (d) meridional advection, (e) surface zonal pressure gradient, (f) total zonal pressure gradient, (g) horizontal diffusion, and (h) vertical diffusion. The total pressure gradient term (f) is the sum of the baroclinic pressure gradient term (not shown) and the surface pressure gradient term (e) (see text); the vertical diffusion term includes the wind stress. The zonal momentum balance is (a) = (b) + (c) + (d) + (f) + (g) + (h).

Figure 19 (Continued). The contour interval is 20 cm s/mo for accelerations below 100 cm s/mo and 100 cm s/mo for accelerations above 100 cm s/mo. The zero contour line is not shown. Dark shading is westward acceleration greater than 100 cm s/mo, light shading is eastward acceleration greater than 100 cm s/mo.

The GFDL primitive equation model in the form used by Philander and Seigel (1985) does not permit explicit evaluation of the surface pressure gradient term in the horizontal momentum equation unless every time step is available, so this term has to be estimated from the residual in the zonal momentum equation balance and then added to the zonal pressure gradient resulting from vertically integrating the horizontal gradient of the density field. We find that with our 3-day sampling interval in time, the best zonal momentum equation balances are obtained if the "surface" pressure gradient term is evaluated from the momentum equation imbalance well below the thermocline. In this calculation the 21st vertical level, with a nominal depth of 490 m, is used. This surface pressure gradient term is shown in the zonal momentum equation term balances for reference, but is not a separate zonal momentum equation tendency term. In our discussion in the text, references to "the zonal pressure gradient tendency" always correspond to the "U due to total pressure" panel.

The ENSO period changes in the zonal momentum terms in SADLER begin with the July 1982 strong thermocline deceleration accompanied by near-surface acceleration. There is a sharp brief reduction in the zonal pressure gradient in the near surface and a reversal in the thermocline, the vertical advection term briefly goes to zero, horizontal diffusion drops almost to zero, zonal advection is eastward near the surface and westward in the thermocline, and meridional advection is westward in the upper thermocline. In short, everything changes in the zonal momentum balance. It was the complexity of the signal in these terms that led us initially to try more restricted experiments in the hope of simplifying the response and learning more about the dynamical signatures of different types of simple responses. While our simpler studies indicate that the observed zonal current and thermal field changes are consistent with remote Kelvin forcing, it would be difficult to try to argue this position on the basis of the July 1982 zonal momentum equation balances.

The next major zonal flow change is the November eastward acceleration of the surface jet, and its rapid penetration into the thermocline. We have seen that purely local winds also produce eastward acceleration at this time, but that the resulting vertical penetration is less and the amplitude is reduced from the full wind stress experiment. The three terms with near-surface eastward tendency in November are 1) vertical diffusion down to about 30 m (recall that this term includes the wind stress through the vertical boundary condition), 2) zonal advection down to about 120 m, and 3) vertical advection between about 80 m and 180 m and meridional advection between about 40 m and 120 m. The retarding terms are horizontal diffusion and the zonal pressure gradient, which vanished in early October and remains reversed from normal until April 1983.

It is instructive to compare these terms with those of the LOCAL experiment (Fig. 20). LOCAL shows eastward acceleration in November from the surface to about 100 m. The eastward tendency terms are zonal advection from the surface to 60 m, vertical and meridional advection (weakly) from about 20 m to 60 m, and vertical diffusion from the surface to 100 m. The primary westward term is horizontal diffusion (the zonal pressure gradient is very weakly westward). Thus, the balance of terms during this period is not very different from that in SADLER, even though we have established that remote forcing is significant in SADLER.

Figure 20. The individual zonal momentum equation terms at 0$#176;, 160$#176;W for the LOCAL experiment as in Fig. 19.

Figure 20. (Continued)

Now consider the deceleration phase of the eastward surface jet during December 1982 and January 1983. The eastward terms in SADLER are zonal advection (near the surface until mid-December), vertical advection (with a complex vertical structure, and until mid-December), vertical diffusion (very near the surface), and meridional advection (40 m to 80 m). The largest westward terms are the zonal pressure gradient and meridional advection very near the surface; smaller contributions come from horizontal diffusion, zonal advection (below the near-surface eastward tendency in December and then at most depths into January), and vertical advection. The pattern is complex, but the tendency toward Kelvin-type dynamical patterns is present in the fact that the zonal pressure gradient is a major factor over the entire period and depth range of the deceleration. However, again, it would be rather bold to infer solely from these patterns that remote forcing was a major element of this deceleration phase, given the magnitude of the other terms and their changes.

The zonal momentum equation terms in LOCAL during this period are quite different from those in SADLER. There is no sharp westward deceleration phase in December, to begin with, and the only term to exceed 100 cm s/mo tendency is the eastward vertical diffusion of momentum. Zonal advection, zonal pressure gradient, meridional advection (very near the surface), and horizontal diffusion are all weakly westward over the depth range of interest.

From our perspective, examination of the zonal momentum equation balance of terms is interesting in the complexity it reveals. Only with the benefit of the various simpler cases of 1982-83 winds and of the idealized westerly remote forcing studies are we able to make reasonably confident assertions about the relative importance of local and remote forcing, and particularly about the role of Kelvin-type physics in the response to westerly remote forcing. In the presence of a strongly nonlinear background circulation undergoing large changes like those of this region during this period, the zonal momentum equation balance of terms reveals great complexity in space and time.


Return to previous section or go to the next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage