资源环境科技发展态势分析平台

Atlantic Multidecadal Variability (AMV) is a multivariate phenomenon. Here for the first time we obtain a multivariate AMV index and associated patterns using Multivariate Empirical Orthogonal Function (ME OF) analysis to explore the multivariate nature of AMV. Coherent multidecadal variability that is unique to the Atlantic is found in the observed MEOF-extracted AMV, various AMV-related indices, and an Atlantic Meridional Overturning Circulation fingerprint. When the signal associated with global mean sea surface temperature is removed from both observations and Coupled Model Intercomparison Project Phase 5 (CMIP5) simulations, the residual CMIP5 forced basin-wide sea surface temperature-based AMV index disagrees strongly with the observed residual. Only the observed residual basin-wide sea surface temperature-based AMV index retains a strong AMV signal. The MEOF approach still extracts a residual CMIP5 forced AMV signal that is unique to the Atlantic, although very different from observations. Our findings suggest that the observed AMV is not dominated by external forcing.

Plain Language Summary Observed Atlantic Multidecadal Variability is a multivariate phenomenon and has various important climate impacts at global and regional scales. Understanding the underlying physical mechanism is crucial for successful future prediction of Atlantic Multidecadal Variability and associated climate impacts. From a multivariate perspective, this study shows that the simulated externally forced Atlantic Multidecadal Variability in coupled climate model simulations disagrees strongly with that observed. The results suggest that the observed Atlantic Multidecadal Variability, after removal of the signal associated with global mean sea surface temperature, is not dominated by external forcing. This observed variability is unique to the Atlantic and is linked to multidecadal variations of the Atlantic Meridional Overturning Circulation.