The observations were obtained from the Westerbork Synthesis Telescope (WSRT), using the 5120 channel digital backend. The main observational parameters are summarized in Table 1 (click here).
Table 1: Observational parameters of WSRT HI observations of Cas A
The noise in the observations is higher by a factor two than expected, since Cas A
(
K, aperture beam efficiency = 0.64) as well the HI background emission
contribute to the system temperature (in 1979 
K).
Because the HI absorbs the
continuum radiation in some channels partly or even almost completely, the above
effect is reduced and the noise in these maps is lower, i.e. the noise varies
from channel to channel. In the analysis we did not take these variations into account.
The data were calibrated using the standard procedures provided by WSRT staff in
Dwingeloo. The maps were not cleaned. Experiments showed that the effects of
the sidelobes of the synthesized beam were not large (below the noise), with
the exception of channels with are partly strongly absorbed. But due to the
"corrugation''-effect (Schwarz 1984), additional uncertainities are
introduced by cleaning, and some systematic effects for large optical depth
have to be taken into account, typically 
.
The Groningen GIPSY software package was used to analyze the
data and to prepare all figures.
The continuum radiation of the source is derived by averaging channels which
are free of line emission; the continuum image is displayed in the first panel of
Fig. 2 (click here).
Optical depths were computed according Eq. (1).
where 
is the spin temperature of HI and 
, 
, 
are the brightness of the line radiation, the
continuum and the galactic background (including HI-emission), respectively.
The continuum intensity has a peak value of 20.6 Jy/beam. Opacities were calculated
only for regions of the source with a
brightness above a cut-off of 
of 1.25 Jy/beam.
The effects of 
and 
are described in
Schwarz et al. 1995; since these
quantities are not known and are estimated to be much smaller compared than
(see Table 1 (click here)),
we neglect these terms. The errors in 
depend on the rms noise, 
, of the
observations and on 
. For small optical depths, the errors are 
.
This error is displayed in Fig. 2 (click here), in the second panel; in Table 1 (click here) we give
the value for the average continuum 
.
For large optical depths the errors are determined by noise and by the spectral dynamic range
(channel to channel calibration errors) of 1000. The maximum optical depth reliably
measured is 
.