2004A&A...414....7G

The analysis of the recent WMAP source catalog shows that the vast majority of bright foreground extragalactic sources detected in CMB maps are Blazars. In this paper we calculate the contamination of CMB anisotropy maps by this type of flat-spectrum, strongly variable and polarized extragalactic radio sources using up-to-date results from recent deep multi-frequency surveys. From a careful archive search and from multi-frequency catalog cross-correlations we found that more than 50 known Blazars (or Blazar candidates) expected to be above the sensitivity limit of the BOOMERANG experiment are included in the 90/150GHz BOOMERANG anisotropy maps, a factor >15 larger than previously reported. Using a recent derivation of the Blazar radio logN-logS we show that Blazars, whose counts continue to grow steeply down to faint fluxes, can sensitively affect CMB fluctuation maps with even moderate resolution and sensitivity. We calculate specifically that these sources induce an average sky brightness of 0.4Jy/deg², corresponding to an average temperature of ≃4-8µK. Moreover, we find that the associated level of fluctuations is of the order of C_l,Blazar=2x10^–2µK² sr at 41 GHz and C_l,Blazar=1x10^–3µK² sr at 94GHz. Taking into account both Blazar variability, causing the detection of a number of weak sources that rise above the detector sensitivity during flares in long-exposure satellite experiments, and the many steep-spectrum radio sources that flatten at high frequencies, as well as the contribution of radio-galaxies, we find that the level of residual fluctuation due to discrete extragalactic foreground sources could be factor of ∼2-3 higher than the above estimate. We show that the Blazar induced fluctuations contaminate the CMB spectrum at the level of ≃30-90% at l=500 and 80-100% at l=800, and thus they cannot be neglected in the derivation of the primordial CMB fluctuation spectrum. Since Blazars are bright sources because they point their jets towards the observer, at faint fluxes a more abundant population of less-aligned sources is likely to increase the contamination of future high-resolution high-sensitivity CMB maps. Careful cleaning for Blazar contamination of CMB maps is therefore necessary before firm conclusions about weak features, like secondary high-l peaks of the CMB power spectrum or very weak signals like CMB polarization measurements, can be achieved.