0229-Masking versus removing point sources in CMB data: arxiv:1207.2315
Last updated
Last updated
Using the proper residual correction, we find that the CMB power spectrum obtained from the point source cleaned map without any source mask is fully consistent with the spectrum obtained from the masked map.
We see that, as expected since there is more information, the standard deviation of the estimated power spectrum when removing the point sources is smaller than when they are masked. We note, however, that when removing point sources one has to account for the bias introduced by the uncertainty in estimating the point source position and amplitude.
each source is masked with a rather large disk depending on the beam of the experiment and some data are lost. Another disadvantage with this approach is that analysis methods applying filters (for instance wavelets) to the map may contaminate an even larger part of the filtered map in the area of the point source holes
depends on modeling the source correctly
main idea: reconstruct fluctuations of CMB in contaminated region