Links between the mountain torque and the Arctic Oscillation

The fact that the evolution of the Arctic Oscillation is in part affected by mountains is shown using various pieces of independent evidences. First, in the NCEP/NCAR reanalysis we show that, at periodicities around 30 days and below, the mountain torque (Tm) drives most of the changes in the Atmospheric Angular Momentum (AAM). At these periodicities, the variations in AAM are equally distributed between variations in wind AAM (Mw) and mass AAM (Mm). This equipartition is explained theoretically as the result of the geostrophic adjustment of a barotropic axisymmetric flow to the mountain torque.

As the Arctic Oscillation (AO) is associated with a redistribution of the atmospheric masses from the polar regions to the midlatitudes, it is associated with substantial mass AAM. Accordingly, we also found that in the NCEP data, the AO variations are very significantly related to mass AAM variations driven by Tm.

Second, these results are confirmed by using a 30-year simulation done with the Laboratoire de Météorologie Dynamique (LMDz), general circulation model. In this respect, the LMDz model has the great advantage of closing the AAM budget nearly exactly, which is not the case with the NCEP reanalysis data. In this model, we verify that the torque Tm and the AO are in lead-lag quadrature, the relationship between the two being associated with the variations in mass AAM driven by the montain torque.

Finally, as the Antarctic Oscillation (AAO) is also associated with a redistribution of mass from the polar regions to the midlatitudes, its contribution to the AAM budget is also presented. As there are much fewer mountains in the Southern Hemisphere, we show that in the model as well as in the reanalysis the changes in mass AAM during intraseasonal variations of the AAO are in good part equilibrated by changes of opposite sign in wind AAM.

The main interest of these results is that the mountain torque drives the changes in AAM, so it can sometimes participate actively in changes of the AO. It has a predictive value that is significant but small, around 10-15% for periodicities near and below one month, while a good fraction of the AO variability occurs at longer timescales.