Mechanisms underpinning skillful decadal prediction in the North Atlantic

A number of studies have shown that initialized, coupled 20^th century simulations can produce skillful predictions of sea surface temperature (SST) variability in the subpolar North Atlantic at multi-year to decadal lead times. This skill is generally associated with AMOC initialization, which is assumed to confer at least a damped-persistence skill in predicting the anomalous oceanic meridional heat transport (MHT) that drives high latitude SST change. However, the details of the internal ocean and coupled air-sea processes that contribute to or detract from decadal prediction (DP) skill in the subpolar Atlantic remain rather poorly understood. We analyze and compare two DP experiments run with the Community Earth System Model (CESM) in order to elucidate the mechanisms underpinning prediction skill in the subpolar gyre. In both experiments, 10-member coupled ensembles are initialized in each year from 1955 to 2014 and integrated forward 10 years. The two experiments differ as follows: (i) they use different atmospheric models (CAM4 vs CAM5, both at 1^o resolution), (ii) they are initialized from slightly different ocean states in the Tropics, and (iii) they are initialized at different times of year (January 1^st vs November 1^st). Both experiments will be brought to bear on such questions as: What are the relative roles of gyre and overturning heat transports, and which is more predictable? Is heat transport skill associated primarily with anomalous circulation (V’<T>) or anomalous heat content (<V>T’)? Is there any skill in predicting surface water mass formation or wind stress curl, or does circulation skill derive solely from the initialization of density anomalies? What is the role of abyssal flow interaction with topography in determining prediction skill?