Stochastic Forcing of the North Atlantic Wind-Driven Ocean Circulation. Part II: An Analysis of the Dynamical Ocean Response Using Generalized Stability Theory

As discussed in Part I of this study, the magnitude of the stochastic component of wind stress forcing is comparable to that of the seasonal cycle and thus will likely have a significant influence on the ocean circulation. By forcing a quasigeostrophic model of the North Atlantic Ocean circulation with stochastic wind stress curl data from the NCAR CCM3, it was found in Part I that much of the stochastically induced variability in the ocean circulation is confined to the western boundary region and some major topographic features even though the stochastic forcing is basinwide. This can be attributed to effects of bathymetry and vorticity gradients in the basic state on the system eigenmodes. Using generalized stability theory (GST), it was found in Part I that transient growth due to the linear interference of nonnormal eigenmodes enhances the stochastically induced variance. In the present study, the GST analysis of Part I is extended and it is found that the patterns of wind stress curl that are most effective for inducing variability in the model have their largest projection on the most nonnormal eigenmodes of the system. These eigenmodes are confined primarily to the western boundary region and are composed of long Rossby wave packets that are Doppler shifted by the Gulf Stream to have eastward group velocity. Linear interference of these eigenmodes yields transient growth of stochastically induced perturbations, and it is this process that maintains the variance of the stochastically induced circulations. Analysis of the large-scale circulation also reveals that the system possesses a large number of degrees of freedom, which has significant implications for ocean prediction. Sensitivity studies show that the results and conclusions of this study are insensitive and robust to variations in model parameters and model configuration.