Spurious Indo�Pacific connections to internal Atlantic multidecadal variability introduced by the global temperature residual method

The relative contributions of external forcing and internal processes to the observed spatial and temporal characteristics of "Atlantic Multidecadal Variability" (AMV) are still under debate. Here, the efficacy of the commonly-used "global temperature residual method" for isolating the internal component of AMV is investigated by means of model Large Ensembles where the truth is known a priori. In this method, local sea surface temperature variability associated with global-mean temperature (G) is removed via linear regression, and the residuals regressed upon the North Atlantic residual timeseries. We show that this method introduces spurious connections over the Indo-Pacific due to the fact that G in any single realization includes both external and internal components: the latter dominated by influences from Pacific Decadal Variability independent of AMV. This methodological shortcoming can be overcome by using the forced component of G in the residual method applied to individual model realizations and to observations.