Equatorward North Atlantic jet biases in CMIP models and implications for simulated regional atmosphere-ocean linkages

Abstract

<p>Climate model biases in the North Atlantic (NA) low-level tropospheric westerly jet are a major impediment to reliably representing variability of the NA climate system and its wider influence, in particular over western Europe. We highlight an early-winter equatorward jet bias in Coupled Model Inter-comparison Project (CMIP) models and assess whether this bias is reduced in the CMIP6 models in comparison to the CMIP5 models. Historical simulations from the CMIP5 and CMIP6  are further compared against reanalysis data over the period 1862-2005.  </p><p>The results show that an equatorward bias remains significant in CMIP6 models in early winter. Almost all CMIP5 and CMIP6 model realizations exhibit equatorward climatological jet latitude biases with ensemble mean biases of 3.0° (November) and 3.0° (December) for CMIP5 and 2.5° and 2.2° for CMIP6. This represents an approximately one-fifth reduction for CMIP6 compared to CMIP5. The equatorward jet latitude bias is mainly associated with a weaker-than-observed frequency of poleward daily-weekly excursions of the jet to its northern position. A potential explanation is provided.  Our results indicate a strong link between NA jet latitude bias and systematically too-weak model-simulated low-level baroclinicity over eastern North America in early-winter.  </p><p>Implications for model representation of NA atmosphere-ocean linkages will be presented. In particular CMIP models with larger equatorward jet biases tend to exhibit weaker correlations between temporal variability in jet speed and sea surface conditions over the NA sub-polar gyre (SPG). This has implications for the ability of climate models to represent key aspects of atmospheric variability and predictability that are associated with atmosphere-ocean interactions in the SPG region.  </p>