SKAO HI Intensity Mapping: Blind Foreground Subtraction Challenge

Abstract

Neutral Hydrogen Intensity Mapping (H IM) surveys will be a powerful new probe of cosmology. However, strong astrophysical foregrounds contaminate the signal and their coupling with instrumental systematics further increases the data cleaning complexity. In this work, we simulate a realistic single-dish H IM survey of a 5000 deg 2 patch in the 950 − 1400 MHz range, with both the MID telescope of the SKA Observatory (SKAO) and MeerKAT, its precursor. We include a state-of-the-art H simulations and explore different foreground models and instrumental effects such as non-homogeneous thermal noise and beam side-lobes. We perform the first Blind Foreground Subtraction Challenge for H IM on these synthetic data-cubes, aiming to characterise the performance of available foreground cleaning methods with no prior knowledge of the sky components and noise level. Nine foreground cleaning pipelines joined the Challenge, based on statistical source separation algorithms, blind polynomial fitting, and an astrophysical-informed parametric fit to foregrounds. We devise metrics to compare the pipeline performances quantitatively. In general, they can recover the input maps' 2-point statistics within 20 per cent in the range of scales least affected by the telescope beam. However, spurious artefacts appear in the cleaned maps due to interactions between the foreground structure and the beam side-lobes. We conclude that it is fundamental to develop accurate beam deconvolution algorithms and test data post-processing steps carefully before cleaning. This study was performed as part of SKAO preparatory work by the H IM Focus Group of the SKA Cosmology Science Working Group.

Baryon acoustic oscillations from HI intensity mapping: the importance of cross-correlations in the monopole and quadrupole

“…Doing this in the radial P (k ) does not guarantee an avoidance of non-linear effects since even small k can still be affected by the non-linearities of physical small scales. Furthermore, isolating or modelling observational effects could potentially be more troublesome for the radial P (k ) method, as shown in Matshawule et al (2020) and Spinelli et al (2021). In any case, it is important for alternative approaches to be available for pursuing precision cosmology.…”

“…polarisation leakage or beam effects which we do not investigate in this work (see e.g. Carucci et al 2020;Spinelli et al 2021). It is expected that these issues should be avoidable either by modelling the systematics using information from the scanning strategy (McCallum et al 2021) or exquisite instrument calibration which should hopefully be achievable with a full SKAO survey.…”

“…where r(z) is the radial comoving distance to redshift z and θ FWHM is the beam angular resolution, which given SKA features, can be approximated as 0.8 (1 + z) deg. More sophisticated analyses, including beam sidelobes, can be found in Matshawule et al (2020) and Spinelli et al (2021).…”

Probing Ultra-light Axion Dark Matter from 21cm Tomography using Convolutional Neural Networks

“…The observed 21cm signal includes a type of contamination known as the foreground emission, mostly dominated by synchrotron emission from our own galaxy. This type of contamination is usually removed following different techniques (Spinelli et al 2021). As the foreground removal usually affects more the larger scale modes of the temperature maps (Asorey et al 2020;Shi et al 2020), we leave this for a future study and assume for this work a perfect foreground removal.…”

Measurements of the diffuse Galactic synchrotron spectral index and curvature from MeerKLASS pilot data

“…In this section we complement the results obtained in Section 4 by analysing the MeerKLASS pilot data maps under a different set of assumptions. As the cosmological 21 cm signal is orders of magnitude weaker than the astrophysical foregrounds in intensity mapping data, its recovery often relies on Blind Source Separation (BSS) methods for disentangling the different -astrophysical, cosmological, and spurious -components of the data cube (Spinelli et al 2021a). These methods make general statistical assumptions on the foregrounds, and require no prior knowledge on the 21 cm signal.…”