Investigating air-sea interaction in the extratropical North Pacific Ocean: comparing observations to a coupled GCM

Previous work has shown that a null hypothesis of extratropical air-sea interaction can be well characterized by a local, coupled stochastically-forced model consisting only of air (TAIR) and sea surface temperature (SST) anomalies. We build such a model empirically using observed weekly averaged anomalies of SST and TAIR over the North Pacific Ocean. Firstly, the model successfully reproduces the observed lagged covariance statistics with lead times of up to 90 days, confirming that it is an appropriate tool for diagnosing seasonal air-sea interaction. Surprisingly, the model does just as well within the Kuroshio-Oyashio region as it does farther east where ocean dynamics are relatively less important. In fact, we show that the only difference between the two regions is the amount of stochastic atmospheric and SST forcing, which is an order of magnitude larger in the vicinity of the western boundary current.

Secondly, the validity of the model allows for a detailed inspection into the role of coupling. Over 80% of monthly averaged SST anomalies disappear when the model is integrated in an uncoupled mode. A notable exception is within the Kuroshio region where 50% of SST anomalies are retained, highlighting the importance of intrinsic oceanic processes (e.g. eddies). Finally, we repeat our analysis with a coupled GCM that is run separately at high (0.5° atmosphere, 0.1° ocean) and low (1° atmosphere and ocean) resolutions. The difference in coupling is striking: the low-resolution simulation significantly underestimates the amount of intrinsic SST variability, convincingly showing the important role of ocean eddies that are captured by the high-resolution model. One implication is that reproducing the observed air-sea coupling strength depends more strongly on the resolution of the ocean model than that of the atmosphere.