The predictability of decadal changes in the North Pacific is investigated
with an ocean general circulation model forced by simplified and realistic
atmospheric conditions. First, the model is forced by a spatially fixed win
d stress anomaly pattern characteristic for decadal North Pacific climate v
ariations. The time evolution of the wind stress anomaly is chosen to be si
nusoidal, with a period of 20 years. In this experiment different physical
processes are found to be important for the decadal variations: baroclinic
Rossby waves dominate the response. They move westward and lead to an adjus
tment of the subtropical and subpolar gyre circulations in such a way that
anomalous temperatures in the central North Pacific develop as a delayed re
sponse to the preceding wind stress anomalies. This delayed response provid
es not only a negative feedback but also bears the potential for long-term
predictions of upper ocean temperature changes in the central North Pacific
. It is shown by additional experiments that once these Rossby waves have b
een excited, decadal changes of the upper ocean temperatures in the central
North Pacific evolve without any further anomalous atmospheric forcing. In
the second part, the model is forced by surface heat flux and wind stress
observations for the period 1949-1993. It is shown that the same physical p
rocesses which were found to be important in the simplified experiments als
o govern the evolution of the upper ocean in this more realistic simulation
. The 1976/ 77 cooling can be mainly attributed to anomalously strong horiz
ontal advection due to the delayed response to persistent wind stress curl
anomalies in the early 1970s rather than local anomalous atmospheric forcin
g. This decadal change could have been predicted some years in advance. The
subsequent warming in the late 1980s, however, cannot be mainly explained
by advection. In this case, local anomalous atmospheric forcing needs to be
considered.