Chang et al. (1997) have proposed that the decadal variability in the tropical Atlantic is inherently due to coupling between the atmosphere and the ocean. There are two key processes that are occurring in their hypothesis. First, there is a local positive feedback between the atmosphere and the ocean that leads to growth of an initial disturbance and to the formation of the "dipole" pattern in SST and meridional displacement of the ITCZ that are so familiar from previous analyses. This local feedback happens on a short (subseasonal) time scale. Second, once the pattern gets established, the ocean dynamics (in particular, the exchange of mass between hemispheres) acts rather slowly to restore the system to its mean condition. We have been testing various key links in the Chang et al. hypothesis for the decadal variability in the tropical Atlantic. We dispay here just one of these tests.
We have applied idealized SST anomaly patterns to the T42 and T31 versions of the NCAR Community Climate Model (ver. 3; CCM3) and integrated the model to equilibrium. One such idealized SST anomaly is displayed in Fig. 1a. Note the lobes of SST anomaly are centered in the subtropics, with anomalously warm (cold) water in the northern (southern) subtropics. The response of the AGCM to this prescribed SST anomaly is indicated in Fig. 1b., where we show the surface wind stress anomaly from the T42 integration of the CCM3 with the SST anomalies prescribed in Fig. 1a. The response of the atmosphere is remarkably similar to the atmospheric anomalies that are observed in conjunction with the dipole in SST: there is a relaxation (enhancement) of the trades in the northern (southern) subtropics, with a strong southerly flow across the equator. Remarkably, there is a reduction in net surface energy flux out of (in to) the ocean (not shown) - precisely where the SST is anomalously positive (negative). Thus, we have reproduced the first of the aforementioned key processes in Chang et al.'s hypothesis.
We are presently sorting out why the GCM (and presumably the real atmosphere) responds the way it does. Are the anomalies due to a boundary layer (e.g., Lindzen/Nigam) response to SST anomalies? Or they better described as a wholesale rearrangement of the tropical Atlantic circulation in the troposphere (e.g., Gill/Zebiak)? Finally, we are exploring the possibility that the SST anomalies only catalyze the atmosphere, leading to land/atmosphere interactions over the Sahel (which are also simulated in the model in qualitative agreement with the observations) that are more responsible for the bulk anomalies in atmosphere and ocean in the oceanic regions.
We have performed a theoretical analysis and a modeling study that illuminates the basic effects of coupling the atmosphere to the oceans in midlatitudes (Barsugli and Battisti, in press). It is well known that variability in the atmosphere is responsible for a large part of the variability in the upper ocean, and that the ocean spectrum is "red" relative to the atmospheric forcing (e.g., Hasslemann and Frankignoul, 1977). In this study we have quantified how feedbacks between the atmosphere and the ocean change the spectrum of variability in the atmosphere. Our treatment of the problem allows us to better define the true source of "noise forcing" in atmosphere, and to better illuminate to relative contributions of noise forcing and feedback in the the net surface energy flux.
In Fig. 2, we show the spectra for the SST and atmospheric surface air temperature from a simple theoretical model under different coupling strategies. From this figure it is clear that coupling enhances the variability in the ocean and in the atmosphere, especially at low frequencies. Interesting, the surface fluxes are reduced in the coupled system compared to the uncoupled system, especially at low frequencies. The enhancement of variance and the reduction of surface fluxes has been described as a "reduced thermal damping" due to coupling.
An important result of this study is that it changes the interpretation of experiments in which ocean and atmosphere models are forced by prescribed forcing (e.g., prescribed SST anomalies under an atmospheric GCM, or prescribed surface heat fluxes or air temperature anomalies over an ocean GCM). For example, the results of the theoretical study indicate that in prescribing perfectly known SST anomalies under a perfect atmosphere model (the so-called MOGA and TOGA type experiments), the simulated atmospheric response will be underpredicted (by, typically, a factor of two), and the the simulated flux anomalies will be too large. We have confirmed these surprising theoretical predictions using a simplified atmosphere GCM. Our results help to explain an apparent deficiency in the atmospheric GCMs that has been reported in the literature over the last two decades: when compared to the observed atmosphere, the GCMs produce anemic anomalies when they are forced with the observed (prescribed) SST anomalies.