Bias and Uncertainty of the Relationship between AO and Winter Synoptic Temperature Variability over the Northern Hemisphere under Present and Future Climate

Abstract

Abstract In this study, the relationship between AO and winter synoptic temperature variability (STV) over the Northern Hemisphere is examined in 34 CMIP5/CMIP6 model outputs. With significant model bias around the North Pacific and North Atlantic, most models fail to capture the correct AO–STV pattern in historical simulations compared to observations. To investigate the bias of AO–STV relationship simulations, AO-related processes for the connection between AO and winter STV are examined in high pattern score (HPS) models and low pattern score (LPS) models, respectively. Furthermore, the bias of AO impact can be traced back to AO pattern simulations. On the one hand, compared to observations, HPS models can overall capture the intensity in the North Pacific and North Atlantic center of AO. On the other hand, LPS models tend to overestimate the North Pacific center and underestimate the North Atlantic center. In addition, similar to historical simulations, a robust AO–STV relationship can still be found over the Northern Hemisphere in future projections based on HPS models. Meanwhile, the uncertainty of the projected AO–STV relationship in the multimodel ensemble is confined mainly to the North Pacific, consistent with the large diversity of intensity over the North Pacific center of AO, which is related to the uncertainty of the relationship between AO and regional mode variability.