Evidence from observations and models indicates that tropical and mid-latitudes Atlantic sea surface temperature (SST) and the Atlantic meridional overturning circulation (MOC) exhibit variability on interannual and decadal timescales. Additionally, many model studies show potential predictability of the North Atlantic SST and MOC changes on decadal timescales.;In this thesis, the role of the ocean dynamics in Atlantic variability and in limiting the predictability of the tropical and North Atlantic climate is investigated using a hierarchy of models and theoretical tools. These problems are addressed by calculating and studying the optimal growth of initial anomalies in linearly stable regimes, in which a substantial growth of the anomalies can be achieved before their eventual decay. The main results and their implications for the variability and predictability of the Atlantic climate can be summarized as follows: (1) the MOC and SST are most sensitive to deep density perturbations. More specifically, tropical SST variability is optimally excited by deep ocean perturbations near the western boundary while the MOC and Northern Hemisphere SST are most efficiently excited by deep density perturbations in the Northern Hemisphere. (2) The mechanisms for the growth of SST and MOC anomalies are fully three-dimensional, including wave dynamics which are not captured by zonally averaged models. The tropical SST amplification is achieved by the excitation of Kelvin waves, while the amplifications of northern hemisphere SST and MOC anomalies involve the westward propagation of temperature anomalies as "thermal" Rossby waves where the mean potential vorticity gradient is provided by the mean meridional density gradient. (3) The ocean is found to participate actively in the amplification of the SST and MOC anomalies rather than to passively integrate over the imposed initial forcing. (4) MOC anomalies grow faster when deep ocean perturbations are allowed rather than when only the surface is perturbed. This result leads to the conclusion that predictability experiments in coupled models in which only the atmospheric state (equivalent to perturbing only the ocean surface) is perturbed may overestimate of the ocean predictability time.
展开▼