3 Sensitivity, compact sources and confusion limit

 

All total intensity survey maps show a large number of compact sources, which are mainly of extragalactic origin. We have selected several small flat and empty fields from the survey maps and subtracted sources when necessary. The measured rms-noise is typically about 15 mK T_B.

No problem with confusing sources exists for polarized intensities, where a typical rms-noise value of 8 mK T_B has been found.

The measured rms-noise in total intensity is about a factor of three larger than that calculated from the system temperature of about 30 K. Source confusion is believed to be the limiting factor in sensitivity. Condon et al. (1989) have studied the effect of source confusion for the former 300-ft Green-Bank telescope, which is described by $\mathrm{\Delta} S_{\rm c} = 50\, \nu^{-2.7}$, where $\nu$ is in GHz and $\mathrm{\Delta} S_{\rm c}$ is the rms of intensity fluctuations in units of mJy/300-ft beam area. When calculating the expected confusion for the 100-m telescope from that approach we get about 16 mJy or 35 mK T_B at 1.4 GHz, which is significantly larger than the measured rms-noise of 15 mK T_B in the maps. We conclude that source confusion is significantly smaller than previously thought, but it is limiting the sensitivity for total intensities of the survey.

We have performed source counts from the survey maps based on a fitted two-dimensional Gaussian to each source. These counts have been compared with counts based on the recent 1.4 GHz VLA-survey (NVSS) by Condon et al. (1998), which is more sensitive to compact sources than our survey and suffers less from confusion due to its smaller beam size. We convolved a VLA map covering about $71.5~\ifmmode\hbox{\rlap{$\sqcap$}$\sqcup$}\else{\unskip\nobreak\hfil \penalty5... ...box{\rlap{$\sqcap$}$\sqcup$} \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi$ to the angular resolution of the Effelsberg map and applied the same source fitting procedure to both maps. In Fig. 1 we show cumulative source counts for a region in the Galactic anti-centre. We have also fitted the VLA data at their original angular resolution and found 910 sources in total. The corresponding cumulative source counts are also shown in Fig. 1 with a slope of about -1.4. This is close to the expected value of -1.5 for a uniform source density in the local universe. Source evolution and different source populations cause deviations from the -1.5 slope for sources weaker than about 100 mJy (see Condon et al. 1998 for details). Below about 40 mJy the fraction of sources which can not be fitted individually increases with decreasing flux density. For the Effelsberg data as well as for the convolved VLA data we obtain the same result over the entire flux density range. However, the deviation from the straight line starts already near 100 mJy. This effect is caused by the much larger confusion in the low resolution data: the total number of sources stronger than 10 mJy found in the original VLA data results in a mean source separation of about 22 VLA beam areas, but just about 1.8 beam areas of the 100-m telescope. Therefore, source counts for flux densities lower than about 100 mJy are more complete, when using the original VLA data. With decreasing flux density an increasing fraction of sources cannot be fitted individually due to increasing confusion effects.