Low-frequency pycnocline variability in the northeast Pacific

The output from an ocean general circulation model (OGCM) driven by observed surface forcing is used in conjunction with simpler dynamical models to examine the physical mechanisms responsible for interannual to interdecadal pycnocline variability in the northeast Pacific Ocean during 1958-97, a period that includes the 1976-77 climate shift. After 1977 the pycnocline deepened in a broad band along the coast and shoaled in the central part of the Gulf of Alaska. The changes in pycnocline depth diagnosed from the model are in agreement with the pycnocline depth changes observed at two ocean stations in different areas of the Gulf of Alaska. A simple Ekman pumping model with linear damping explains a large fraction of pycnocline variability in the OGCM. The fit of the simple model to the OGCM is maximized in the central part of the Gulf of Alaska, where the pycnocline variability produced by the simple model can account for similar to 70%-90% of the pycnocline depth variance in the OGCM. Evidence of westward-propagating Rossby waves is found in the OGCM, but they are not the dominant signal. On the contrary, large-scale pycnocline depth anomalies have primarily a standing character, thus explaining the success of the local Ekman pumping model. The agreement between the Ekman pumping model and OGCM deteriorates in a large band along the coast, where propagating disturbances within the pycnocline, due to either mean flow advection or boundary waves, appear to play an important role in pycnocline variability. Coastal propagation of pycnocline depth anomalies is especially relevant in the western part of the Gulf of Alaska, where local Ekman pumping-induced changes are anticorrelated with the OGCM pycnocline depth variations. The pycnocline depth changes associated with the 1976-77 climate regime shift do not seem to be consistent with Sverdrup dynamics, raising questions about the nature of the adjustment of the Alaska Gyre to low-frequency wind stress variability.