Planck 2013 results. XXXI. Consistency of the Planck data

TL;DR

The paper assesses the internal consistency of Planck 2013 data across multiple frequency channels, confirming high agreement within calibration uncertainties and analyzing systematic errors to improve future measurements.

Contribution

It provides a detailed analysis of the internal consistency of Planck 2013 data across different frequency channels and discusses systematic errors and calibration procedures.

Findings

70/100GHz power spectra agree within 0.8%

143/100GHz power spectra agree within 0.4%

Likelihood analysis shows 0.35% consistency among channels

Abstract

The Planck design and scanning strategy provide many levels of redundancy that can be exploited to provide tests of internal consistency. One of the most important is the comparison of the 70GHz and 100GHz channels. Based on different instrument technologies, with feeds located differently in the focal plane, analysed independently by different teams using different software, and near the minimum of diffuse foreground emission, these channels are in effect two different experiments. The 143GHz channel has the lowest noise level on Planck, and is near the minimum of unresolved foreground emission. In this paper, we analyse the level of consistency achieved in the 2013 Planck data. We concentrate on comparisons between the 70/100/143GHz channel maps and power spectra, particularly over the angular scales of the first and second acoustic peaks, on maps masked for diffuse Galactic emission and for strong unresolved sources. Difference maps covering angular scales from 8deg-15arcmin are consistent with noise, and show no evidence of cosmic microwave background structure. Including small but important corrections for unresolved-source residuals, we demonstrate agreement between 70 and 100GHz power spectra averaged over 70<l<390 at the 0.8% level, and agreement between 143 and 100GHz power spectra of 0.4% over the same l range. These values are within and consistent with the overall uncertainties in calibration given in the Planck 2013 results. We also present results based on the 2013 likelihood analysis showing consistency at the 0.35% between the 100/143/217GHz power spectra. We analyse calibration procedures and beams to determine what fraction of these differences can be accounted for by known approximations or systematic errors that could be controlled even better in the future, reducing uncertainties still further. Several possible small improvements are described...(abridged)