Orateur
Description
The fine measurement of the polarized emission of the CMB is intimately linked to the ability to separate it from the Galactic foregrounds. A specific difficulty for this goal is to take into account the spatial non-Gaussian structure of these foregrounds. In this work, we propose a new method that is able to extract the non-Gaussian features of the galactic dust emission from noisy observations, that we apply to a full-sky map from the SRoll2 HFI-Planck data at 353Ghz. This method introduces in particular the Cross Wavelet Scattering Transform (CWST). These statistics, which are able to characterize non-Gaussian statistical dependencies between different processes, play a crucial role in extracting the features of the dust emission from two half-missions with different noise realizations. This statistical separation of components allows to recover the statistics of the dust emission, including its non-Gaussian properties, until close to lmax~800, where the noise dominates in power by a factor 100 in power spectrum ( fsky=0.5), as well as to generate new realizations of the foregrounds with reproduce these statistical properties up to the same scale. Thanks to their realistic non-Gaussian features, we expect these new models of galactic dust emission to play a strategic role for future CMB analyses.
