Power spectrum estimation

We use the MASTER method [19] in order to measure the power spectrum on our maps. Within this framework, maps are obtained by coadding the filtered timelines on the sky. In the case of pure white noise, this leads to optimal maps. We filter our timelines keeping only frequencies between 1 and 45 Hz so that the resulting power spectrum is very close to be white. The pseudo-$C_\ell$ spectrum is obtained using anafast in Healpix package [20]. We then estimate the noise angular power spectrum on the coadded maps using a set of Monte-Carlo simulations. The effect of the filter on the underlying sky is also estimated via Monte-Carlo simulations. The mode mixing effect is deconvolved following [19]. All these corrections allow us to transform the pseudo-$C_\ell$ spectrum into a real angular power spectrum. We estimated the non optimality of our power spectrum (due to the non optimal maps) to be less than 30% at all scales. The power spectrum estimation is at the moment under test. The expected accuracy for 10 bolometers, obtained through full simulations from timelines to power spectrum with realistic noise (measured on the real timelines), is shown in top panel of Figure .

Figure: Estimated power spectrum accuracy for the 7h30 Kiruna scientific flight (top) and for the incoming 24 hours flights (bottom). Both were obtained for 10 bolometers from the average of one thousand realistic simulations of the Archeops timelines and noise structure.