Inspection of Plates and the Digitized Sky Survey

Having generated a list of preliminary identifications using COSMOS, we then inspected the actual sky images at each radio position using either the survey films directly or, more conveniently, the Digitized Sky Survey (DSS; White et al., 1993). The DSS is a digital reproduction on CD-ROM of images of the whole sky from blue sensitive Palomar O and UK Schmidt IIIaJ plates; the pixel size is 17. It should be pointed out that COSMOS and DSS have complementary rôles in the identification process: the COSMOS parameters can not readily be generated from the DSS, and examination of DSS images is necessary for checking COSMOS classifications.

Visual inspection, usually via radio-DSS overlays, is an essential step in securing identifications for very extended sources. For such sources a simple position comparison is inadequate, because the host galaxy may be displaced significantly from any radio peak position, eg. a double-lobed source without a detectable core component. When such structures were recognised, the separate positions and flux densities were combined and listed as one source. The situation is further complicated in clusters where the morphologies of sources can often be highly distorted by drag forces arising from galaxy motion through the intra-cluster medium, from buoyancy forces, or from winds in the intra-cluster medium, all leading to significant displacements of the galaxy from the radio centroid.

Overlays of MOST contours on the corresponding optical images from the DSS were made using a series of programs in AIPS. In most cases it was possible to confirm or reject identifications by inspecting hard copies of these overlays, but occasionally it was necessary to inspect the original sky survey material. In a few cases it was not possible to make an identification with confidence, because the radio morphology was unusual or difficult to classify at the resolution of the MOST, and/or there was more than one plausible counterpart. An example is shown in Figure 3; sources such as this have been targeted for higher resolution observations with the Australia Telescope Compact Array (ATCA) to help clarify the identifications.

Experience showed that it was not possible to apply a rigid set of criteria but the following broad guidelines were followed in the inspection process, keeping in mind the results from the Monte Carlo tests described in Section 3.2 and Figure 2.

1. If a visual inspection confirmed a COSMOS galaxy classification and the source was unresolved or very slightly resolved, the identification was retained unless the 2% chance probability limit was exceeded.
2. If a visual inspection confirmed a stellar COSMOS classification, a possible QSO identification was accepted (subject to the 2% chance probability limit) if the automated process claimed it. No spectroscopic information was available for confirmation, and there is currently no colour information in readily accessible digital form.
3. If the object was classified as `faint' by COSMOS, and it could be clearly classified by visual inspection (only a small fraction), the above rules were applied. If a clear classification was impossible, the identification was retained if and the 2% chance probability limit was not exceeded. Such faint objects were kept separate from the galaxy and possible QSO identifications.
4. If the radio source was well resolved, or showed complex structure, it was assumed to be associated with a galaxy. The search for counterparts took into account what is commonly known about the structures of radio sources and how they can be distorted by the cluster environment. Galaxies between the lobes of double sources were considered, as were galaxies in the elbows of bent structures and at either end of elongated structures. In order to preserve reliability, only relatively bright galaxies (, see Figure 2(d)) were considered as counterparts to complex extended sources. Extended sources often had strong cores associated with bright galaxies, but this was not always the case.
5. There were 14 instances where two or more objects satisfied the above identification criteria. In those cases the following course was followed:
   • Galaxies were preferred over stellar objects, unless the galaxy offset was many times greater.
   • For two objects of the same type the closer one was normally preferred, unless the difference in separation was much less than the separations themselves in which case the brighter one was preferred.

In a few cases, a faint object or even a bright galaxy was not listed in the COSMOS catalogue and positions were measured directly from the DSS. The number of such objects is too small to significantly affect the identification statistics.

© Copyright Astronomical Society of Australia 1997