Coupled processes for equatorial Pacific interannual variability

Diagnostics for coupled processes relevant to equatorial Pacific interannual variability are defined, using a simplified analytical model, and applied to general circulation model simulations. The analytical model is a simplification of the Neelin model (Neelin 1991). The atmospheric component is represented by a direct zonal thermal circulation. The oceanic dynamics take into account a shear layer and a Kelvin wave. Analytical calculations lead to an unstable slow-propagating coupled mode, referred to as the TK-mode. All the processes of this mode are explicitly identified, as are their contributions to the resulting total growth rate. These contributions are computed in forced oceanic (covering the 1979–93 period) and in coupled ocean/atmosphere general circulation model simulations. The results are sensitive to the parameters but allow relative comparisons between general circulation model simulations. The forced oceanic simulation suggests that the total instability is nearly constant all over the equatorial Pacific basin, but the central part is dominated by the zonal advective coupled process, while the eastern part is controlled by both zonal and vertical advective coupled processes. The thermocline coupled process appears largest in the eastern part of the basin, but its contribution is still weaker than the other two. The coupled simulation appears globally more unstable than the forced one. The total instability is larger in the eastern part of the Pacific basin. This result is consistent with the location and the frequent occurrence of the simulated interannual signals in the eastern part of the basin. The seasonal cycle of the forced simulation shows maximum instability between July and November. This could suggest an approach towards the understanding of the seasonal phase lock of the onset of warm El Nino/Southern Oscillation events. Possible interactions between the delayed oscillator mechanism or the westerly wind bursts with the TK-mode processes are discussed. In order to understand the real system, these diagnostics should be applied to the observations and not just models. Nonetheless, analysis of coupled processes allows objective comparisons between different general circulation model simulations, and the results presented in this paper should be considered as a first step in this direction.