On processes controlling seasonal North Atlantic sea surface temperature anomalies in ocean models

Mechanisms for the winter-to-winter persistence of North Atlantic sea surface temperature (SST) anomalies are studied in a one-dimensional Lagrangian upper ocean model. SST is most sensitive to surface heat flux perturbations. Strong end-of-winter sensitivity to heat flux from one year earlier exists in two anti-nodes of the SST tripole associated with the North Atlantic Oscillation: off the east coast of the United States, and in the sub-polar gyre. They are due to the reemergence of temperature anomalies during winter from the seasonal thermocline. The recurrence is weaker in the third tripole anti-node, west of North Africa, because upper ocean seasonality is weaker there. The effects of wind stress and freshwater flux anomalies on SST are less important than heat flux changes but can be of either sign. They depend on the season and the vertical temperature profile at the perturbation time. results confirm previous findings from a non-eddy-resolving North Atlantic ocean general circulation model (GCM). The implication of this agreement is that the key processes controlling North Atlantic SST over a seasonal cycle in the GCM are the local response to air–sea flux forcing, transport of anomalies through the seasonal thermocline, and subsequent anomaly reemergence into the model mixed layer. The residual SST anomaly left after computing the contribution from the surface flux anomaly in the previous 1–2 years leaves little scope for other dynamical mechanisms, such as ocean circulation changes or planetary waves in the GCM. An extension of the stochastic climate model due to C. Frankignoul and K. Hasselmann [1977. Stochastic climate models, part II: application to sea-surface temperature variability and thermocline variability. Tellus 29, 289–305] is proposed that accounts for these processes. The time evolution of SST anomalies predicted by this simple stochastic delay-differential equation agrees well with results from the Lagrangian upper ocean model and the GCM.