Climate predictions on decadal scales have proven elusive. On these time-scales, the natural chaotic variability of the climate system dominates over the more predictable trends due to anthropogenic climate forcing. The differences among coupled general circulation models (CGCM) results are discouraging, and point to a lack of convergence in spatial resolution and to inadequate physical parametrizations. The difficulty is compounded by the short duration of observational records of the climate system---particularly those of the ocean and cryosphere---which hinders robust statistical analysis on decadal time scales.
A leading candidate for pockets of predictability in selected regions resides in the memory intrinsic in the slow, almost adiabatic evolution of the ocean below the mixed layer. The periodic reemergence of oceanic temperature anomalies in winter is a potential source of predictability on scales longer than one year. In order for this mechanism to provide robust interaction with the atmosphere, ocean temperature anomalies must survive the seasonal cycling and the eddy stirring on isopycnals. It is conjectured that temperature perturbations on scales larger than the eddy size survive isopycnal eddy stirring. The threshold amplitude and duration (or persistence) of perturbation that can withstand the seasonal cycling must be determined.
The focus of this proposal is on the Atlantic Ocean because in this region the natural time scales are decadal, through the wind-gyres dynamics, and multi-decadal, through the pole-to-pole overturning circulation. It is essential to work in the eddying regime, because mesoscale eddies determine the sub-thermocline oceanic stratification and thus the oceanic heat transport. Furthermore, at high resolution, damping is reduced allowing large regional gradients and longer duration of influence: both are necessary elements for robust atmospheric imprinting.
The approach is to use several ocean models spanning a range of complexities: coarse resolution forward and adjoint models both with parametrized baroclinic eddies, and a model at eddy-resolution.
The economical and social benefits of climate information for periods longer than a season and regions beyond the tropics cannot be overstated. The proposed work will identify the regions in the Atlantic Ocean that can enhance decadal predictability in the extra-tropics. The focus on adiabatic subsurface flow in the eddying regime will complement the traditional approach of studying diffusive, highly damped oceanic dynamics, and will identify the most robust oceanic physical processes for prediction at decadal scales.