In this case, the decadal oscillation remained, but it slowly decayed in time; however, when the coupling coefiicient between SST and wind anomalies for SVD3, φ,was increased somewhat, the solution developed a self-sustained oscillation almost identical to the one in the main run. These results suggest that it is ocean-atmosphere interactions within the subtropics that cause the main run's decadal signal. To confirm this conclusion, we obtained the “inverse” test solution in which coupling was deleted from 10°S to 35°N: No decadal oscillation ever developed regardless of the amplitude of φ.
The interannual oscillation is also absent when SVD1 is removed from the model atmosphere. This result is not surprising since SVD1 is so clearly related to the equatorial feedback processes thought to be crucial to Pacific interannual (ENSO) variability. Conversely, omission of SVD3 eliminates the decadal signal, while still allowing the interannual one. This property provides additional support for the idea that it is subtropical ocean-atmosphere interactions that generate the solution's decadal signal, since most of the subtropical wind-stress variability in the model atmosphere is associated with SVD3 (Figure 1).
Finally, in a test solution without wind-speed feedback on the latent heat flux, the decadal oscillation is absent but the interannual one remains. These properties are consistent with existing ideas about ENSO dynamics and Pacific decaclal variability, that is, that SST anomalies are caused by wind variations in the former but involve surface heat fluxes in the latter (Gu and Philander, 1997).
5 SUMMARY AND CONCLUSIONS
In this paper, we report solutions to an intermediate coupled model confined to the Pacific region. For reasonable parameter choices, solutions develop both interannual and decadal oscillations, and variability at both time scales is present in the eastern, equatorial ocean.
