Simulated modes of inter-decadal predictability in sea surface temperature

A methodology is proposed that allows an estimate of the contribution of decadal noise to the inter-decadal variability of climate variables. By removing this decadal noise, or unpredictable decadal variability, from the total inter-decadal variability, the residual variability is more likely to be associated with potentially predictable slow processes. This residual variability may be considered to be potentially predictable. We apply the method to sea surface temperature from a 1000 year (850-1850 AD) experiment with the CCSM4 model forced by solar variations, volcanic aerosols, orbital forcings, land use and greenhouse gas concentrations. Our analysis shows large potential SST inter-decadal predictability in the extra-tropical regions in both hemispheres, as well as in the subtropical/tropical Indian Ocean, the western Pacific and the Atlantic. In the tropical eastern Pacific, inter-decadal variability is dominated by unpredictable decadal noise. The two leading unpredictable modes of inter-decadal variability have features related to the two leading ENSO modes, and the PDO. The four leading potentially predictable modes of inter-decadal variability are shown to be related to the external forcing, the IPO, inter-hemispheric SST fluctuations and the AMO, and decadal/multi-decadal AO/NAO variability.