Annu. Rev. Astron. Astrophys. 1992. 30: 653-703 Copyright © 1993 by Annual Reviews. All rights reserved

4.1 Confusion by Discrete Radio Sources

There have been a number of discussions in the literature of the effects on microwave background radiation observations of discrete radio sources (e.g. Hogan 1980, Fomalont et al 1984, Knoke et al 1984, Fomalont et al 1988, Readhead et al 1989, Myers 1990, Banday et al 1991, Myers et al 1992). The effects are most severe on small angular scales and low frequencies but they are also appreciable, and must be taken into account, on angular scales up to ~ 5° and at frequencies up to ~ 30 GHz, and possibly higher.

The most detailed of these analyses is that of Franceschini et al (1989), which supersedes the earlier paper from this group (Danese et al 1983). In the more recent paper the authors calculate the expected contribution to T/T due to discrete sources from 0'.1-300' and at frequencies from 5-100 GHz. This paper presents an excellent treatment of the problems of confusion, however, a fundamental difficulty arises in this kind of work due to the non-Gaussian distribution of observed intensities of these objects. There is a long, non-Gaussian tail in the distribution such that the presence or absence of a few bright sources has an appreciable affect on the predicted T/T. Franceschini et al deal with this by introducing a cutoff value in the flux density at the level of 3-5 . While this is a practical approach which enables one to calculate expected contributions to T/T to within a factor of a few, it obscures the basic problem introduced by the long positive tail in the discrete source distribution. An underlying assumption of Franceschini et al is, therefore, that sources brighter than the 3-5 level will be detected and identified as discrete sources, so that the contribution to T/T of these sources can be calculated on a case by case basis, and subtracted out. Franceschini et al further assume that the observations are limited by confusion. However, in many cases the noise level is set by thermal noise at a level considerably above the confusion level, so that the cutoff at 3-5 is really at too low a level for the individual identification and subtraction of sources. Thus, for example, the models of Franceschini et al underestimate the expected contribution to T/T of discrete sources for the RING (Myers 1990, Myers et al 1992) and NCP (Readhead et al 1989) programs. Nevertheless they do provide extremely useful estimates of the approximate levels of contamination and of the contributions to T/T of different classes of object.

The effects of discrete sources on microwave background radiation observations on arc second and arc minute scales have been discussed above in the sections devoted to these observations, where it was seen that this form of contamination is severe in observations such as those carried out at the VLA, at Tenerife, and at the Owens Valley Radio Observatory, at frequencies up to at least 20 GHz.

At the VLA some of the effects of the discrete sources can be eliminated internally, since the high resolution of the maps themselves make it possible to identify the brighter objects, but, as we have seen in Section 3.1.1, modelling is required to take account of the weaker objects. In lower resolution observations - be they either single dish or interferometric observations - it is not possible to identify and subtract the sources internally. Observations at higher resolution are required to identify the sources.

Although discrete source contamination is a severe problem, it is not insurmountable. One can monitor the flux densities of the discrete (often variable) sources on an instrument with significantly higher resolution than that being used for the microwave background radiation observations and apply the necessary corrections to the background observations. At the Owens Valley Radio Observatory we will adopt this strategy with the newly installed 5.5 meter telescope and the 40 meter telescope. At 30 GHz the 40 meter telescope is 15 times more sensitive to unresolved sources than the 5.5 meter telescope. The use of these two antennas in parallel allows us to monitor the strength of the discrete sources simultaneously with the microwave background radiation observations. With both telescopes observing in parallel, nine sources giving a 1 deflection in each 5.5 meter telescope field could be observed to a precision of 0.2 with the 40 meter telescope. The most effective strategy needs to be worked out, but it is clear that we can remove the effects of discrete sources on our 5.5 meter telescope observations to well below the temperature levels at which they plague the observations on the 40 meter telescope. It is likewise possible to monitor and make allowance for the confusing sources in the Tenerife observations.