Granger Causal Analysis between the NAO and SSTs in the North Atlantic basin

This paper uses Granger causality test techniques to investigate the causal order of the air-sea interactions between the anomalous NAO index and the NCAR/NCEP SST anomalies over the northern hemisphere. Compared to the simple lead/lagged correlation, the Granger causality analysis is generally more reliable since it ensures that any apparent oceanic influence upon the atmosphere is uniquely provided by the ocean and is not related to preceding anomalies in the atmosphere itself (and vice versa when looking at the atmospheric influence upon the ocean). Using this technique, it is found that on seasonal timescales, preceding SST anomalies around the Gulf Stream region show significant Granger causality with the wintertime NAO. In contrast, the preceding NAO anomalies’ influence on the wintertime SST field is rather restricted. Additional findings show that regions of the SST field can cause up to 20% of the variance in the winter NAO. Importantly the regions that show significant causality with the wintertime NAO are limited to the Gulf Stream extension. In contrast to simple correlation analysis, our analysis does not show significant causality related to SST anomalies in either the Greenland or subtropical regions. These results suggest that the Gulf Stream SSTs have an important influence in initiating disturbances of the atmospheric circulation over the wintertime North Atlantic. The dynamics related to this influence still need to be investigated further.