Item

Abstract

Aspects of low-frequency variability of the atmospheric circulation in the Euro-Atlantic domain, such as the frequency of wintertime blocking and the North Atlantic Oscillation, have been found to exhibit significant predictability in large-ensemble seasonal and decadal forecasts. Part of this predictability arguably stems from the realistic initialization of the North Atlantic ocean, which influences the atmospheric circulation through air–sea interaction. Yet, various model deficiencies, such as systematic model biases and the misrepresentation of coupled processes may be limiting the emergence of the associated predictable signals. We present results from three recent studies indicating that increasing model resolution (typically moving to 0.25° in the ocean and 0.5° in the atmosphere) is crucial for mitigating various such deficiencies. Specifically, we show that: (i) increasing model resolution in a multi-model ensemble of coupled historical HighResMIP simulations leads to mitigating certain long-standing extratropical SST biases that contribute to the under-representation of European blocking, (ii) increasing the atmospheric model resolution in a multi-model set of historical HighResMIP simulations forced with observed sea surface temperatures allows for the realistic representation of the atmospheric circulation response to interannual Gulf Stream variability, expectedly important for predictability, and (iii) increasing model resolution in a multi-model ensemble of coupled HighResMIP simulations improves the realism of extratropical North Atlantic atmosphere–ocean variability through mitigating oceanic biases that control the variability of the deep ocean circulation at multi-annual and longer timescales. Current work focuses on evaluating how such model deficiencies affect seasonal and decadal predictability.