A radiation model, together with National Meteorological Center temperature observations, was used to compute daily net heating rates in the northern hemisphere (NH) for the Arctic late fall and winter periods of both 1988--1989 and 1991--1992 and in the southern hemisphere (SH) for the Antarctic fall and winters of 1987 and 1992. The heating rates were interpolated to potential temperature (&thgr;) surfaces between 400 K and 2000 K and averaged within the polar vortex, the boundary of which was determined by the maximum gradient in potential vorticity. The averaged heating rates were used in a one-dimensional vortex interior descent model to compute the change in potential temperature with time of air parcels initialized at various &thgr; values, as well as to compute the descent in log pressure coordinates. In the NH vortex, air parcels which were initialized at 18 km on November 1, descended about 6 km by March 21, while air initially at 25 km descended 9 km in the same time period. This represents an average descent rate in the lower stratosphere of 1.3 to 2 km per month. Air initialized at 50 km descended 27 km between November 1 and March 21.

In the SH vortex, parcels initialized at 18 km on March 1, descended 3 km, while air at 25 km descended 5--7 km by the end of October. This is equivalent to an average descent in the lower stratosphere of 0.4 to 0.9 km per month during this 8-month period. Air initialized at 52 km descenced 26--29 km between March 1 and October 31. In both the NH and the SH, computed descent rates increased markedly with height. The descent for the NH winter of 1992--1993 and the SH winter of 1992 computed with a three-dimensional trajectory model using the same radiation code was within 1 to 2 km of that calculated by the one-dimensional model, thus validating the vortex averaging procedure. The computed descent rates generally agree well with observations of long-lived tracers, thus validating the radiative transfer model.