Multimodel Future Projections of Wintertime North Atlantic and North Pacific Tropospheric Jets: A Bayesian Analysis

Abstract

The impact of future greenhouse gas forcing on the North Atlantic and North Pacific tropospheric jets remains uncertain. Opposing changes in the latitudinal temperature gradient-forced by amplified lower-atmospheric Arctic warming versus upper-atmospheric tropical warming-make robust predictions a challenge. Despite some models simulating more realistic jets than others, it remains the prevailing approach to treat each model as equally probable (i.e., democratic weighting). This study compares democratically weighted projections to an alternative Bayesian-weighting method based on the ability of models to simulate historical wintertime jet climatology. The novel Bayesian technique is developed to be broadly applicable to high-dimensional fields. Results show the Bayesian weighting can reduce systematic bias and suggest the wintertime jet response to greenhouse gas forcing is largely independent of this historical bias (i.e., not state dependent). A future strengthening and narrowing is seen in both winter jets, particularly at the upper levels. The widely reported poleward shift at the level of the eddy-driven jet does not appear statistically robust, particularly over the North Atlantic, indicating sensitivity to current model deficiencies.