All galaxies were observed by the WSRT in full redundancy mode; the correlator sampled both parallel and crossed dipole configurations. The parallel dipole data (WSRT channels XX and YY) were used to produce high dynamic range total intensity maps with the aid of the NRAO AIPS package as follows; the WSRT data were loaded into AIPS and first processed with the special AIPS task REDUN. REDUN is a translation into AIPS of the corresponding Dwingeloo DWARF program. It computes telescope dependent amplitude and phase errors by analyzing the observed amplitudes and phases of the many redundant baselines in the Westerbork array. After this initial operation has been done, WSRT observations can be further processed into high dynamic range images using the AIPS programs MX and ASCAL as has been described by Perley (1986). The signals in the crossed dipole channels XY and YX may be combined with the difference in signal between XX and YY channels to yield maps of the Stokes polarization parameters Q and U. At 326 MHz electric vector position angles can be rotated from their intrinsic position angles by many tens of degrees (more than 100 degrees is not uncommon) because of Faraday rotation within the ionosphere. Therefore all observations were first corrected for ionospheric Faraday rotation using the method developed at Dwingeloo by Spoelstra (1981). The flux calibration scale is that of Baars et al. (1977). In Table 1 (click here) we have compiled relevant observational parameters for the WSRT maps.
The high-frequency observations have been carried out with the Effelsberg 100-m
telescope using the 2.7-GHz 1-horn, 3-channel receiver, the 4.8-GHz 2-horn,
3-channel, and the 10.6-GHz 4-horn, 8-channel receiver system, all installed
in the secondary focus of the telescope. The observations with the 10.6-GHz
system have been described in detail by Klein et al. (1994).
At 2.7 GHz and 4.8 GHz we had to use the single-beam mode (owing to limited
observing time allocation), which is more strongly affected by bad weather
conditions and terrestrial interference.
The maps were large enough to cover the source plus some emission-free
areas used for determination of zero levels and noise. Contrary to the
multi-beam technique, which requires scanning in the horizontal system, the
2.7-GHz and 4.8-GHz maps have been obtained by scanning alternately along, and
perpendicular to, the position angle of the source. The drive rates were
2/min at 2.7 GHz and 1
/min at 4.8 GHz, the scan interval 2
and 1
, respectively.
The individual maps were edited to diminish the influence of
weather or terrestrial interference before they were averaged to yield
final maps of Stokes I, Q, and U, employing
the Fourier filter technique of Emerson & Gräve (1988).
The 10.6-GHz maps shown by Klein et al. (1994) have been
CLEANed, applying the algorithm described by Klein & Mack
(1995). The 4.8-GHz maps have also been CLEANed, but the algorithm
is more complicated in the case of maps not observed in the horizontal
system so that some residual artifacts introduced by the antenna pattern
may be left.
Table 2 (click here) summarizes the relevant map parameters for each source.
The data have been calibrated applying the scale of Baars et al.
(1978). Because of the relatively large number of maps, where parts
of them had to be blanked because of weather or interference effects, the
noise level may vary significantly across the final maps. This is of special
importance for the calculation of the polarized intensity maps.
These have been produced as suggested by Wardle & Kronberg
(1974). Since the polarization information represents a
pseudovector where neither the amplitude nor the phase has a Gaussian
probability distribution one has to apply a correction term, especially in
the case of polarized low-brightness regions that we are concerned with. The
best estimate of the true polarized intensity can be calculated as
where Q, U is the intensity in
the Stokes Q- and U-map, respectively, and is the mean
value of the noise in the Q- and U-maps. The factor 1.2 has been found
empirically to be best suited to shift the peak of the (positive) noise
distribution function to zero.
In view of this correction it is clear that the determination of the proper noise value is very important to obtain the true polarized intensity. Therefore we have developed a routine which calculates the noise at each map pixel as a function of the number of individual maps (i.e. integration time) to be averaged at this pixel. The polarized intensity is thus calculated by accounting for the inhomogeneous distribution of the noise level.