As described above, the C21 correlator measures cross-correlations of all combinations of polarizations; 1420-MHz continuum polarimetry became available on the Synthesis Telescope in 1993 (see Smegal et al. 1997).
The antennas of the Synthesis Telescope are equatorially mounted, with feeds that cannot be rotated relative to the reflector, so the beam pattern always has a fixed orientation on the sky. While this has many advantages, it does complicate separation of spurious instrumental polarization from source polarization. Corrections for instrumental polarization must be determined by observing sources with little or no intrinsic polarization. The method by which these corrections are determined is discussed in detail by Smegal et al. (1997); put briefly, in the absence of an actual polarized signal, the cross-polarization correlations provide information about the orthogonality of the nominal LHCP and RHCP signals on each antenna, also known as "leakage''. The leakage terms measured for the Synthesis Telescope using 3C 147 and 3C 295 amount to between 1 and 5% on most antennas, although one antenna has about 10% leakage. If left uncorrected, these errors produce spurious instrumental polarization at the field centre amounting to several percent of the total flux. After correction, the residual instrumental polarization at the field centre is reduced by about an order of magnitude to 0.25% of the total intensity.
To measure polarization angle it is necessary to measure the phase
difference between RHCP and LHCP channels on one antenna, an arbitrarily
chosen reference. This is achieved by observing 3C 286, a strongly
polarized, compact source with a fractional polarization of 9.25% and
polarization angle of 33
at 1420 MHz. Polarimeter accuracy is
5
in polarization angle, and 10% in polarized intensity. The
precision of the polarization angle measurements is also limited by
ionospheric Faraday rotation, which amounts to about 3
in
polarization angle between day and night (at present no correction is
applied).
The off-axis polarization properties of the instrument are very good, and useful results can be obtained up to 90' from the field centre. The residual instrumental terms rise approximately quadratically from the field centre, reaching about 10% of total flux at this radius. Empirical corrections, determined by observing 3C 147 and 3C 295 at various positions in the field and interpolating to intermediate positions, are applied to remove this increase. These corrections have been measured three times over a three-year period; they are very stable.
At present, deviations from circularity of the feeds are not accounted for. Since the data are processed as if the polarizations were purely circular, a small amount of any incident linear polarization is actually ascribed to circular polarization (Stokes V). This has little effect on linear polarimetry, but severely limits the ability of the telescope to detect and measure circular polarization. This has minimal impact on the astrophysical questions addressed by the telescope. For details see Smegal et al. (1997).
An important aspect of polarization studies is determining rotation measure
to de-rotate observed polarization angles and recover the orientation of
the intrinsic magnetic field. The sensitivity of the Synthesis Telescope
to Faraday rotation is thus an important consideration. At 1420 MHz, a
rotation measure of 4 rad m-2 is required to produce a change in
observed polarization angle of 10,
which is the limit of detectability
with the present system.
In an image made from all 4 available bands, a rotation measure of
approximately 330 rad m-2 would cause 10% depolarization, with
complete depolarization occurring for a rotation measure of
rad m-2. In a single band these figures are over 5
times higher,
rad m-2 and
rad m-2, respectively.
Since the 4 bands are available for separate processing, it is possible to
use the frequency-dependent polarization angle to derive rotation measure.
However, the relatively small span of frequency available imposes fairly
high limits on the detectable rotation measures: 100 rad m-2 is
required to produce a change in angle of 10
between the centres of
bands A and D (see Fig. 2).
Copyright The European Southern Observatory (ESO)