Dynamical response of midlatitude ocean to the associated wind variation act as negative feedback which determines the time scale (Latif and Barnett, 1994). It is very plausible that the observed interdecadal variability in the Pacific is not explained by only one theory. Therefore it is important to separate variabilities into multi atmosphere-ocean coupled modes and to explore their spatial structures and timescales. In this paper we focus on the two interdecadal modes in the Pacific that are concurrently found in one coupled general circulation model (GCM) (Yukimoto et al. 1996, 1998).
MODEL
We used a full coupled atmosphere-ocean GCM (AOGCM). Atmospheric part of this model is a grid atmospheric GCM (AGCM) with 5°× 4°horizontal resolution and 15 vertical levels. Oceanic part is a global ocean GCM (OGCM). Its horizontal resolution is 2°× 2.5°except near the equator where it has finer meridional resolution up to 0.5°. The two components are coupled with exchanging fluxes of heat, freshwater and momentum (wind stress). A flux adjustment for heat and freshwater is employed to keep realistic sea surface climatology. We made a 150 year intergraon of AOGCM.
For comparison, we used another coupled model (refered to as SGCM), which has the same atmospheric part (AGCM) with the AOGCM but coupled to a slab mixed layer ocean with uniform 50m thickness (Kitoh et al., 1998). The slab is driven by atmospheric surface heat flux and heat flux adjustment which includes effects of climatological oceanic advection. Comparison between AOGCM and SGCM clarifies effects of ocean dynanilcal processes on the variability. We made a 100 year integration of SGCM.
SST VARIABILITY IN THE PACIFIC
We examined spatial and temporal structures of long-term variability for simulated SST, and compared these with observed one. Figure 1a shows leading mode empirical or thogonal functions (EOFs) of low-pass (＞8 years) filtered SST in the Pacific region (120°E-70°W, 34°S-60°N). We used 92 year record (1903-1994) of GISST (Parker et al., 1994) for observational SST. The spatial pattern of the observed SST shows a wedge shaped pattern with a positive signal in the central tropical Pacific through the midlatitude eastern coast of Americas and negative midlatitude signals in the North Pacific and in the west-central South Pacific. The AOGCM shows similar spatial pattern (Fig.1b) with observed one. The SGCM also shows tropical wedge shaped pattern, but it is zonally elongated and extremely concentrated on the equator.
Spectra of the temporal confficients of the EOFs are examined (Fig. 2). The leading EOF of observed SST reveals large variability in the decadal to interdecadal timescale of 10 years to 30 years and has spectral peaks at 13 years and 23 years periods. The observed SST reveals red spectral property. for the second EOF and has dominant period at 15 years for the third EOF. The AOGCM also reveals large variance in the period range of 10-30 years, and spectral peaks at 21 years for the first EOF. The second and third modes are also consistent with those for the observed SST. The SGCM shows quite different spectral characteristic from those for observation and AOGCM.
