Astrophysical component separation of COBE-DMR 4-yr data with FastICA

Abstract

We present an application of the Fast Independent Component Analysis (FastICA) method to the Cosmic Background Explorer Differential Microwave Radiometer (COBE-DMR) 4-yr data. Although the signal-to-noise (S/N) ratio in the COBE-DMR data is typically ∼1, the approach is able to extract the cosmic microwave background (CMB) signal with high confidence from high galactic latitude regions. However, the foreground emission components have too low a S/N ratio to be reconstructed by this method (moreover, the number of components which can be reconstructed is directly limited by the number of input channels). The reconstructed CMB map shows the expected frequency scaling of the CMB and, fitting for the rms quadrupole normalization Qrms-PS and primordial spectral index n we find results in excellent agreement with those derived from the minimum-noise combination of the 90- and 53-GHz DMR channels without galactic emission correction. We extend the analysis by including additional channels (priors): the Haslam map of radio emission at 408 MHz and the Diffuse Infrared Background Experiment (DIRBE) 140-μm map of galactic infrared emission. The FastICA algorithm is now able to both detect galactic foreground emission and separate it from the dominant CMB signal. Fitting again for Qrms-PS and n we find good agreement with the results of Górski et al. where galactic emission has been taken into account by means of correlation analysis of the DMR signal. We investigate the ability of FastICA to evaluate the extent of foreground contamination in the COBE-DMR data further including an all-sky Hα survey to determine a reliable free-free. The derived frequency scalings of the recovered foregrounds are consistent with previous correlation studies. After subtraction of the thermal dust emission as in model 7 of Finkbeiner, Davis & Schlegel, we find a clear indication of an anomalous dust-correlated component which is the dominant foreground emission at 31.5 GHz and which is well fitted by a power-law spectral shape ν−β with β∼ 2.5 in agreement with Banday et al.