When affected data are corrected, there is a marked improvement in the image. In figures 11 and 12, we show the central part of the image of the Cen A data. Each image has been smoothed with a 3 arcsec Gaussian. We do not expect that the background calculation will be significantly affected for normal observations.
We expect a second order effect on the number of sources detected (and possibly on the association between detections with different cell sizes). Whereas the positions will change only slightly or not at all, we expect additional source detections near threshold. For CenA we find 63 detections with a 12'' cell for the archival data but only 60 for the reprocessed run. Although this is opposite to our expectations, it might be caused by a 3% increase in livetime at HIbackground level 8 (most of the data had extremely low background levels). With the 24'' cell, the archival run reported 46 sources, but the reprocessed data overflowed with more than 200 detections.
In addition to having more sources, we expect the following:
|positions||minimal effects except for two close sources*|
|intensity||significant increase for point sources|
|s/n||significant increase for point sources|
|variability||second order effects|
* Further testing in 1999 March has convinced us that for some sequences, there are systematic offsets in addition to the general (random) smearing. Thus we would no longer be surprised to find a positional shift of one or two arcseconds in some instances.