Recent Atlantic Multidecadal Variability is Mostly Forced

How much of regional climate variability is due to anthropogenic forcing? It is a question of both time and space scales. On interannual timescales and local spatial scales, most of the variability is presumably internal; on global, centennial timescales, it is mostly forced. Then we are left with this large grey area of regional, decadal-to-multidecadal variability, where the relative magnitude of the internal and forced components is not known.

In this paper, we dive into the question of the relative contributions of internal ‘noise’ and externally forced ‘signal’ to Atlantic Multidecadal Variability (AMV). To do this we use a hierarchy of climate models including simple heuristic models, idealized dynamical models, comprehensive earth system models, and variants of earth system models with physical processes disabled. These models are put to the experimental test of simulating observations using large ensembles with historical and future forcing, which allows us to formally define the signal to noise ratio. We show that prior to 1950, the AMV is largely consistent with variability arising from coupling between the ocean and atmosphere internal to the climate system. However, after 1950, the AMV, its impacts, and the relationship between the two is largely forced.

We will further discuss the implications of these results more broadly in the context of interpreting ‘signal’ and ‘noise’ in large-scale modes of variability in both climate models and observations.