Skilful decadal predictions of subpolar North Atlantic SSTs using CMIP model-analogues

Abstract Predicting regional climate variability is a key goal of initialised decadal predictions and the North Atlantic has been major focus due to its high level predictability potential impact on European climate. These often in sea surface temperatures (SSTs) subpolar gyre (NA SPG). In order understand value initialisation, justify costs such systems, are routinely measured against technologically simpler benchmarks. Here, we present new model-analogue benchmark that aims leverage latent information uninitialised model simulations make NA SPG SSTs. This system searches through more than one hundred thousand simulated years Coupled Model Intercomparison Project archives yields skilful target region comparable systems. Analysis underlying behaviour suggests origins this skill physically plausible. Such can provide useful for systems within also limits prediction have not yet reached.