In the early days of modern ENSO studies (the early 1980s), it was common to find papers that used observed time-dependent fields of tropical surface fluxes of heat and momentum to drive an ocean model and ascertain that the resulting time-dependent fields of SST resembled the evolution of ENSO. It was also common to find papers that used observed time-dependent tropical fields of SST as boundary conditions for an atmospheric model to ascertain that the atmospheric response resembled both the local ENSO fields of sea-level pressure, precipitation and surface fluxes and the remote teleconnections of ENSO to the rest of the globe. Both types of studies are consistency checks on the models used for explaining ENSO but, of course, neither is, by itself, an explanation for ENSO. An explanation for ENSO must tell us where the atmospheric surface fluxes used to force the ocean and the oceanic SST used as a boundary condition for the atmosphere come from.

The only way to do this consistently is to couple an atmospheric model to an oceanic model and see if the coupled model exhibits both the atmospheric and oceanic aspects of ENSO. Even if a coupled model does exhibit ENSO behavior, it is not clear that the reasons for this behavior can be given: correct simulation does not guarantee correct understanding. It is for this reason that we begin with very simple, and therefore understandable, coupled models that exhibit interannual variability.