Covariances and linear predictability of the Atlantic Ocean

Abstract

The problem of understanding linear predictability of elements of the ocean circulation is explored in the Atlantic Ocean for two disparate elements: (1) sea surface temperature (SST) under the storm track in a small region east of the Grand Banks and, (2) the meridional overturning circulation north of 30.5°S. To be worthwhile, any nonlinear method would need to exhibit greater skill, and so a rough baseline from which to judge more complex methods is the goal. A 16-year ocean state estimate is used, under the assumption that internal oceanic variability is dominating externally imposed changes. No evidence exists of significant nonlinearity in the bulk of the system over this time span. Linear predictability is the story of time and space correlations, and some predictive skill exists for a few months in SST, with some minor capability extending to a few years. Sixteen years is, however, far too short for an evaluation for interannual, much less decadal, variability, although orders of magnitude are likely stably estimated. The meridional structure of the meridional overturning circulation (MOC), defined as the time-varying vertical integral to the maximum meridional volume transport at each latitude, shows nearly complete decorrelation in the variability across about 35°N—the Gulf Stream system. If a time-scale exists displaying coherence of the MOC between subpolar and subtropical gyres, it lies beyond the existing observation duration, and that has consequences for observing system strategies and the more general problem of detectability of change.