Making Maps from Planck LFI 30GHz Data with Asymmetric Beams and Cooler Noise

TL;DR

This paper evaluates the impact of realistic instrument systematics on Planck 30 GHz maps, comparing different mapmaking codes and analyzing the effects of asymmetric beams, cooler noise, and pointing errors.

Contribution

It provides a detailed simulation-based analysis of systematics effects on Planck LFI 30 GHz data and compares the performance of various mapmaking algorithms.

Findings

Beam mismatch causes detectable TE spectrum bias.

Cooler noise and pointing errors are minor concerns.

Optimal mapmaking codes yield consistent results, while destripers depend on baseline length.

Abstract

The Planck satellite will observe the full sky at nine frequencies from 30 to 857 GHz. The goal of this paper is to examine the effects of four realistic instrument systematics in the 30 GHz frequency maps: non-axially-symmetric beams, sample integration, sorption cooler noise, and pointing errors. We simulated one year long observations of four 30 GHz detectors. The simulated timestreams contained CMB, foreground components (both galactic and extra-galactic), instrument noise (correlated and white), and the four instrument systematic effects. We made maps from the timelines and examined the magnitudes of the systematics effects in the maps and their angular power spectra. We also compared the maps of different mapmaking codes to see how they performed. We used five mapmaking codes (two destripers and three optimal codes). None of our mapmaking codes makes an attempt to deconvolve the beam from its output map. Therefore all our maps had similar smoothing due to beams and sample integration. Temperature to polarization cross-coupling due to beam mismatch causes a detectable bias in the TE spectrum of the CMB map. The effects of cooler noise and pointing errors did not appear to be major concerns for the 30 GHz channel. The only essential difference found so far between mapmaking codes that affects accuracy (in terms of residual RMS) is baseline length. All optimal codes give essentially indistinguishable results. A destriper gives the same result as the optimal codes when the baseline is set short enough. For longer baselines destripers require less computing resources but deliver a noisier map.