The impact of stratospheric cooling on the tropospheric circulation in a dry primitive-equation AGCM with zonally symmetric boundary conditions and forcing is investigated. It is found that for sufficiently strong cooling of the stratospheric polar vortex, the tropospheric jet shifts poleward substantially and in a way that projects almost entirely and positively (by convention) on the AGCM's annular mode. Accompanying this substantial mean circulation response to stratospheric cooling is a substantial reduction in stratospheric eddy drag. The reduced eddy drag arises from a reduced upwelling wave-activity flux from the troposphere into the stratosphere and from an enhanced equatorward wave-activity flux from high latitudes to low latitudes.

To better understand the dynamics of the mean-flow and eddy-driving responses, the transient adjustment of the stratosphere-troposphere system is investigated using an ensemble of "switch-on" stratospheric cooling runs of the AGCM. The response to the switch-on stratospheric cooling descends from the upper stratosphere into the troposphere on a timescale of 150 to 200 days. The phase speed of descent of the signal appears to match downward-control theory estimates that the phase speed of descent scales as kH, where k is the thermal damping coefficient, and H is the density scale height.

The downward-control analysis is pursued using a zonally symmetric model in which the eddying AGCM's response can be decomposed into contributions from thermal and eddy-driving perturbations. The direct impact of the cooling perturbation on the zonally symmetric circulation is confined to the stratosphere. The direct impact of the the stratospheric eddy-driving perturbation extends into the mid-troposphere but does not account for the tropospheric jet shift. Thus, it is hypothesized that the stratospheric eddy-driving response initiates the tropospheric response, but that tropospheric circulation and eddy feedbacks are needed to yield the full response. This hypothesis is tested by imposing the stratospheric eddy-forcing response, extracted from the cooling run, in a run of the eddying AGCM in which the stratosphere is not cooled. This run yields the poleward shift of the tropospheric jet, in agreement with the hypothesis.