Changes in temperature of the entire antenna, differences in temperature
between components, and gradients in temperature across components will all
change the pointing characteristics of the antenna. We do currently account for
gross differences in temperature between the front and back legs, using an
empirically derived correction to the elevation pointing of
(6.5 0.6)
/
K.
This agree well with recent Finite Element Analyses (Simon Craig, Ian Pain).
However, we take no account of possible side-to-side gradients. According to
MT282, and by use of a coefficient of expansion of 1.2 
10
,
this may produce a ROLL of the elevation axis of 2.4
/
. The
continuous recording of such temperatures at JCMT indicates a typical diurnal
cycle of amplitude 0.5
K with a monotonic section between 6pm HST and
10am. This implies the possibility of a systematic 1.2
effect in ROLL,
converting to 1.2
sin(elevation) on the sky, ie a peak to peak
systematic error of 1.0
at intermediate elevations, and a contribution
to the rms error budget of 0.3
.
Insolation of the primary dish is also expected to induce pointing errors
of order 3
/
K of linear temperature gradient (MT282), but recent
measurements (FB) suggest that gradients effectively disappear without
direct insolation.
We estimate that uncertainties in the current thermal corrections
(and non-corrections) may add 0.5
rms to the error budget.