Mechanisms of Atlantic variability and sensitivity of the Atlantic to surface initial conditions

Abstract

The climate system is a complex system, in the sense that it has many components that interact with each other and in the difficulty of modelling all the interactions within the system. The nonlinear nature of these interactions raise questions on the ability of the system to be predicted. Climate predictability is concerned not only with skill of the prediction itself, but address questions that could lead to a better understanding of climate sensitivity or processes of variability in the climate. The goal of this thesis is contributing to this understanding focusing on the Atlantic and Weddell Sea, key regions of the climate system.

Decadal predictability is limited by errors in the initial conditions and bound- ary conditions (model parametrizations, forcing, etc.). Understanding the impact of initial errors in surface fields (Sea Surface Temperature, SST, or Sea Surface Salinity, SST) has not been developed fully in a non-linear framework, previous approaches have provided optimal perturbations in a linear framework. The impact of errors in the SSS field are explored with a set of experiments perturbing the SSS field with Gaussian perturbations centred in the Irminger Sea parametrized by its horizontal extension and intensity (magnitude of the maxima of the Gaussian distribution). Focusing on the impact of these errors on the Atlantic Meridional Overturning Circulation (AMOC), main driver of the Atlantic climate in various time-scales, we obtained a response to these initial errors ranging between 0.001 to 0.08 Sv psu−1 for lower and higher extensions respectively. Defining statistically a linear regime allowed us to compare this approach to previous linear approaches and, to understand the differences between the predicted linear responses and the model non-linear response. The different sign of the initial perturbation, positive or negative, have an enhancing or reducing response when compared to the linear prediction. The departure from the linear prediction is associated with different parameters of the perturbation: magnitude and extension respectively.

In addition to initial condition errors, we focused on a mechanisms of North Atlantic variability obtained in Ocean PArallelise (OPA) under Mixed Boundary Conditions (MBC). The variability, characterized by oscillations with a time scale of 42-years, is driven by the westward propagation of Thermal Rossby waves changing the zonal density gradient promoting or reducing the meridional transport. We found a strong link between AMOC and Atlantic Multidecadal Oscillation (AMO) oscillations, which are on phase. The mechanism of propagation, through geostrophic self-advection, contributes to changes in other components of the sub- polar region modifying convection and horizontal transport in the region.

Under the same model configuration, variability driven by the Weddell Sea causes a peak in the North Atlantic power spectrum on a centennial time scale was investigated. The multi-centennial time scale rises from a combination of convective and non-convective periods in the Weddell Sea. These two periods are associated with changes in the Atlantic Heat content and variations in the overturning rate. Changes between the two periods are driven by changes in the subsurface heat storage and ice cover in a region west of the Weddell Sea. A similar signature of ice variability is been observed coinciding with regions where previous polynyas have been observed.

More information

Identifiers

Catalogue record

Export record