Is the SST variability in midlatitudes consistent with an AR(1)-process?

A statistical analyses of monthly mean sea surface temperatures (SST) from observations and from a hierarchy of global coupled ocean-atmosphere models were carried out with the focus on the midlatitudes (25N-50N). The spectra of the simulated SSTs have been tested against the null hypothesis of Hasselmann's stochastic climate model, which assumes an AR(1)-process for the SST variability in its simplest version. It has been found that the spectra of the SST variability in the observations and in the CGCMs with fully dynamical ocean models differ significantly from AR(1)-processes, while the SST variability in an AGCM coupled to a slab ocean is consistent with an AR(1)-process. The deviations of the SST spectra from the fitted AR(1) spectra are not due to spectral peaks but are due to different shapes of the SST spectra from seasonal to decadal time scales. Parts of these differences can be attributed to the interaction between the mixed layer and the sub-mixed-layer ocean.

for the special session "On midlatitude air-sea interaction and its linkage to the tropics"