Filtering in CMB data analysis with application to ACT DR4 and Planck observations

TL;DR

This paper examines how Fourier-space filtering and masking affect CMB data analysis, revealing artifacts that can bias cross-spectra, but these effects do not fully account for discrepancies between ACT DR4 and Planck data.

Contribution

It identifies and characterizes the artifacts caused by filtering and masking in CMB maps, and assesses their impact on data comparison between ACT and Planck.

Findings

Filtering creates artifacts that extend beyond masks.

Artifacts can bias or increase variance in cross-spectra.

Systematic effects are insufficient to explain ACT-Planck discrepancies.

Abstract

Motivated by observed discrepancies between the Atacama Cosmology Telescope Data Release 4 (ACT DR4) and Planck 2018 cosmic microwave background (CMB) anisotropy power spectra, particularly in the cross-correlation of temperature and E-mode polarization, we investigate challenges that may be encountered in the comparison of satellite and ground-based CMB data. In particular, we focus on the effects of Fourier-space filtering and masking involving bright point sources. We show that the filtering operation generates bright cross-shaped artifacts in the map, which stretch far outside typical point-source masks. If not corrected, these artifacts can add bias or additional variance to cross-spectra, skewing results. However we find that the effect of this systematic is not large enough to explain the ACT-Planck differences presented with ACT DR4.