Point source detection using the Spherical Mexican Hat Wavelet on simulated all-skyPlanckmaps

Abstract

We present an estimation of the point source (PS) catalogue that could be extracted from the forthcoming ESA Planck mission data. We have applied the Spherical Mexican Hat Wavelet (SMHW) to simulated all-sky maps that include cosmic microwave background (CMB), Galactic emission (thermal dust, free–free and synchrotron), thermal Sunyaev–Zel'dovich effect and PS emission, as well as instrumental white noise. This work is an extension of the one presented in Vielva et al. We have developed an algorithm focused on a fast local optimal scale determination, that is crucial to achieve a PS catalogue with a large number of detections and a low flux limit. An important effort has been also done to reduce the CPU time processor for spherical harmonic transformation, in order to perform the PS detection in a reasonable time. The presented algorithm is able to provide a PS catalogue above fluxes: 0.48 Jy (857 GHz), 0.49 Jy (545 GHz), 0.18 Jy (353 GHz), 0.12 Jy (217 GHz), 0.13 Jy (143 GHz), 0.16 Jy (100 GHz HFI), 0.19 Jy (100 GHz LFI), 0.24 Jy (70 GHz), 0.25 Jy (44 GHz) and 0.23 Jy (30 GHz). We detect around 27 700 PS at the highest frequency Planck channel and 2900 at the 30-GHz one. The completeness level are: 70 per cent (857 GHz), 75 per cent (545 GHz), 70 per cent (353 GHz), 80 per cent (217 GHz), 90 per cent (143 GHz), 85 per cent (100 GHz HFI), 80 per cent (100 GHz LFI), 80 per cent (70 GHz), 85 per cent (44 GHz) and 80 per cent (30 GHz). In addition, we can find several PS at different channels, allowing the study of the spectral behaviour and the physical processes acting on them. We also present the basic procedure to apply the method in maps convolved with asymmetric beams. The algorithm takes ∼72 h for the most CPU time-demanding channel (857 GHz) in a Compaq HPC320 (Alpha EV68 1-GHz processor) and requires 4 GB of RAM memory; the CPU time goes as N3/2pix log(Npix)], where Npix is the number of pixels in the map and is the number of optimal scales needed.