Abstract. The mesosphere is one of the most difficult parts of the
atmosphere to sample; it is too high for balloon measurements and too low for
in situ satellites. Consequently, there is a reliance on remote sensing
(either from the ground or from space) to diagnose this region. Ground-based
radars have been used since the second half of the 20th century to probe the
dynamics of the mesosphere; medium-frequency (MF) radars provide estimates
of the horizontal wind fields and are still used to analyse tidal structures
and planetary waves that modulate the meridional and zonal winds. The
variance of the winds has traditionally been linked qualitatively to the
occurrence of gravity waves. In this paper, the method of wind retrieval
(full correlation analysis) employed by MF radars is considered with
reference to two systems in Antarctica at different latitude (Halley at
76∘ S and Rothera at 67∘ S). It is shown that the width
of the velocity distribution and occurrence of “outliers” is related to the
measured levels of anisotropy in the received signal pattern. The magnitude
of the error distribution, as represented by the wind variance, varies with
both insolation levels and geomagnetic activity. Thus, it is demonstrated
that for these two radars the influence of gravity waves may not be the
primary mechanism that controls the overall variance.