2010MNRAS.402..207B

In this paper, we present a foreground analysis of the Wilkinson Microwave Anisotropy Probe (WMAP) 5-year data using the fastica algorithm, improving on the treatment of the WMAP 3-year data in Bottino et al.

We determine coupling coefficients between the WMAP data and templates commonly used to trace the dominant foreground emissions, and use them to study the spectral properties of the diffuse emissions and subsequently to clean the data. We confirm the dependence of the values of these scaling factors on the extension of the mask used in the analysis and we again demonstrate some anomalies when the Kp0 mask is adopted that remain unexplained.

We also revisit the nature of the free-free spectrum with the emphasis on attempting to confirm or otherwise the spectral feature claimed in Dobler et al. and explained in terms of spinning dust emission in the warm ionized medium. With the application of different Galactic cuts, the index is always flatter than the canonical value of 2.14 except for the Kp0 mask which is steeper. Irrespective of this, we cannot confirm the presence of any feature in the free-free spectrum.

We experiment with a more extensive approach to the cleaning of the data, introduced in connection with the iterative application of fastica. We confirm the presence of a residual foreground whose spatial distribution is concentrated along the Galactic plane, with pronounced emission near the Galactic Centre. This is consistent with the WMAP haze detected in Finkbeiner et al.

Finally, we attempted to perform the same analysis on full-sky (FS) maps. The code returns good results even for those regions where the cross-talk among the components is high. However, slightly better results in terms of the possibility of reconstructing an FS CMB map are achieved with a simultaneous analysis of both the five WMAP maps and foreground templates. None the less, some residuals are still present and detected in terms of an excess in the cosmic microwave background power spectrum, on small angular scales. Therefore, a minimal mask for the brightest regions of the plane is necessary, and has been defined.