Sunyaev-Zel'dovich cluster reconstruction in multiband bolometer camera surveys

¹ CEA – CE Saclay, DSM/DAPNIA, 91191 Gif-sur-Yvette Cedex, France e-mail: yvon@hep.saclay.cea.fr
² PACM, Princeton University, Princeton, NJ 08544, USA
³ California Institute of Technology, Mail Code 130-33, Pasadena, CA 91125, USA

Received: 12 July 2005
Accepted: 30 March 2006

Abstract

We present a new method for the reconstruction of Sunyaev-Zel'dovich (SZ) galaxy clusters in future SZ-survey experiments using multiband bolometer cameras such as Olimpo, APEX, or Planck. Our goal is to optimise SZ-Cluster extraction from our observed noisy maps. None of the algorithms used in the detection chain is tuned using prior knowledge of the SZ-Cluster signal, or other astrophysical sources (Optical Spectrum, Noise Covariance Matrix, or covariance of SZ Cluster wavelet coefficients). First, a blind separation of the different astrophysical components that contribute to the observations is conducted using an Independent Component Analysis (ICA) method. This is a new application of ICA to multichannel astrophysical data analysis. Then, a recent non linear filtering technique in the wavelet domain, based on multiscale entropy and the False Discovery Rate (FDR) method, is used to detect and reconstruct the galaxy clusters. We use the Source Extractor software to identify the detected clusters. The proposed method was applied on realistic simulations of observations that we produced as mixtures of synthetic maps of the four brightest light sources in the range 143 GHz to 600 GHz namely the Sunyaev-Zel'dovich effect, the Cosmic Microwave Background (CMB) anisotropies, the extragalactic InfraRed point sources and the Galactic Dust Emission. We also implemented a simple model of optics and noise to account for instrumental effects. Assuming nominal performance for the near future SZ-survey Olimpo, our detection chain recovers 25% of the cluster of mass larger than , with 90% purity. Our results are compared with those obtained with published algorithms. This new method has a high global detection efficiency in the high-purity/low completeness region, being however a blind algorithm (i.e. without using any prior assumptions on the data to be extracted).