A global view of air–sea thermal coupling and related non-Gaussian SST variability

Two recent studies of observed sea surface temperature (SST) variations have provided striking evidence that their non-Gaussian statistics can be understood within the context of linear stochastically forced models in which the amplitude of the noise forcing depends linearly on the SST. One study is based on a general 1-d Langevin model, whereas the other considers an explicitly coupled 2-d model of SST and local surface air temperature, and emphasizes the crucial role of thermal air–sea coupling in generating the non-Gaussian statistics. It is shown here that these studies are not necessarily mutually inconsistent. In particular, it is shown that the 1-d model can be derived from the 2-d model as a special case, in which the air temperature is approximated as comprising a part linearly dependent on SST plus a pure noise part. This version of the 1-d model, however, can only predict positive SST skew given the observed thermal damping and coupling constants over the globe. The fact that the more general 1-d model can also capture the essence of the local SST dynamics in regions of negative SST skew therefore implies that other physical mechanisms not included in simple local thermally coupled models, such as stochasticity in oceanic heat transports, may also be important in those regions.