Foreground Implications in the Scientific Exploitation of CMB Data

The study of the Cosmic Microwave Background (CMB) represents one of the main sources of information on which the modern cosmology is based. The observables characterizing the CMB are: its own photon distribution function, the temperature anisotropies and its polarization. Any of these is affected by astrophysical foregrounds: spurious signals that contaminate the CMB observations. The level of contamination depends on the frequency of observation and on the region of the sky observed. Full-sky observation must face with high level of contamination, because of the strong Galactic emission in microwave bands. The first part of my thesis work focus on the CMB photon distribution function. In particular, I show the implementations and the updating phases characterizing a numerical integration code (KYPRIX) for the solution of the Kompaneets equation in cosmological context. The updating is mainly related to the formalism that must be used in order to include FORTRAN libraries necessary for the integral quantities computation and to the platform transfer of the code itself. Physical implementations were also performed: the introduction of the cosmological constant in the equations dedicated to compute the evolution of the primordial Universe; the choice of the primordial chemical abundances of H and He is now possible; the ionization fractions for the species involved in the physical processes were introduced; it was created an optional interface that links KYPRIX with codes, like RECFAST, in order to calculate a recombination history of the ionization fraction of H and He. A general gain in the performances of the code was obtained and some test on the accuracy of the code are shown. All of the physical implementations contributed to perform more realistic simulation of the spectral distortion of the CMB. Some of the highlighted case are related to the contribution of the cosmological constant for the generation of spectral distortions in late epochs (like reionization induced distortions) and to the more precise estimate of the spectral distortions that could occur just before the recombination. In the last case, a recombination history, instead of an instantaneous recombination, plays a fundamental role for precise spectral simulations. After a brief review on the past, on-going and future CMB dedicated experiments, the thesis focus on the importance of foregrounds in CMB experiment, paying attention on their description and the way they affect all the CMB osservables. In order to subtract the foregrounds from sky maps, their morphological characteristics and their spectral behavior must be know as better as possible. The Planck Sky Model (PSM) is a complete and versatile set of programs and data, to be used for the simulation or the prediction of sky emission in the frequency range of typical CMB experiments, and in particular of the upcoming Planck sky mission (the 10-1000 GHz range). It is being developed as part of the activities of Planck Working Group 2 (WG2) on component separation, and of the AD AMIS team at APC. During my second stage at APC I performed several tests on the PSM. The tests involved the latest two release of Galactic emission model, the Galactic foreground template derived by WMAP data and a clean CMB anisotropy map. For what concerns the PSM polarization prediction of the Galactic emission processes, the tests showed a clear improvement, at all the frequencies tested (the five WMAP bands), both for Q and U Stokes parameters. The last release of the PSM total intensity prediction of the Galactic processes showed results consistent with the previous ones for almost all the frequencies tested, while it still needs some tuning at 23 GHz, where synchrotron emission and free-free emission are more prominent. In order to recover the CMB signal, it is necessary to disentangle the mixed signals that a receiver collect looking at the sky. Component separation technics provide the tools to do it. In the thesis are exposed the main features of the component separation methods most used in CMB context, focusing on a particular tool: SMICA. I started using SMICA during my first stage at APC, in 2007. I used SMICA, and another filter (FFT filter) I developed, for a reprocessing of the IRIS mapset. Being already a reprocessing of the IRAS maps at 12, 25, 60 and 100 microm, the IRIS mapset still suffered from a non-optimal subtraction of the Zodical Light Emission (ZLE) residuals. The oscillations induced by these residuals affect any large scale statistical analysis of the mapset. Some preliminary operations had to be performed on the maps, like the subtraction of point sources and the masking Galactic plane and other relevant compact regions. The first phase of the reprocessing I performed consists in recovering the ZLE residuals pattern and power through SMICA, at all the IRIS frequencies. After several code runs and a continuos tuning of the maps and the code, the ZLE residuals pattern was clearly detected at 12,25 and 60 microm, while at 100 microm its contribution resulted to be totally negligible. The recovered component was then subtracted from the maps, but in some confined regions the subtraction wasn't optimal. A FFT filter was developed for this purpose and I used the filter only on localized regions, not on the full-sky maps. The dramatic improvements obtained on the IRIS maps are clearly visible just by eye. Anyway, a statistical analysis is in progress and preliminary results are showed in the thesis.