An analysis of North Pacific SST anomalies by means of a linear thermodynamic stochastic two-dimensional model

The evolution of sea surface temperature anomalies (SSTA) in the North Pacific is studied with a 2-D linear heat transport equation with seasonally varying feedback and constant diffusion. The extension of the model to include an average cyclo-stationary SSTA advection field, estimated from seasonal ship-drift data of surface currents, does not improve the fit to the data. The model parameters are determined by an optimal fit of the observed SSTA lag-time auto-correlation function and the optimal model turns out to be statistically satisfactory (at the 95% confidence level) in most (83%) of the analysed region. The model is also able to explain the lag-time SSTA auto-covariances. Additionally, by applying the model operator to the observed monthly SSTA time series we obtain the actual stochastic forcing. This forcing is consistent with the usual concept of midlatitude atmospheric behaviour. In particular, the basic hypothesis of the stochastic approach, that the forcing is uncorrelated in the SSTA time scale, is verified. Both the success of the model hierarchy fitting procedure and the comparison of the best-fit value of the feedback parameter with mixed-layer depth data suggest that the feedback can be seen as an essentially local process closely related to the thermal inertia of the ocean mixed-layer.