Constraining Polarized Foregrounds for EoR Experiments I: 2D Power Spectra from the PAPER-32 Imaging Array

TL;DR

This paper analyzes polarized foregrounds in 2D power spectra from the PAPER-32 array to understand their impact on Epoch of Reionization experiments and assesses calibration effects on foreground leakage.

Contribution

It provides the first detailed analysis of polarized foregrounds in 2D power spectra from the PAPER-32 array, highlighting calibration effects and the polarization behavior in the EoR window.

Findings

Polarized foregrounds show a wedge shape similar to Stokes I.

Modest polarization calibration errors do not significantly leak power outside the wedge.

High signal-to-noise observations reveal systematic effects in polarization measurements.

Abstract

Current-generation low frequency interferometers constructed with the objective of detecting the high-redshift 21 cm background, aim to generate power spectra of the brightness-temperature contrast of neutral hydrogen in primordial intergalactic medium. Two-dimensional power spectra (power in Fourier modes parallel and perpendicular to the line of sight) formed from interferometric visibilities have been shown to delineate a boundary between spectrally-smooth foregrounds (known as the wedge) and spectrally-structured 21 cm background emission (the EoR-window). However, polarized foregrounds are known to possess spectral structure due to Faraday rotation, which can leak into the EoR window. In this work, we create and analyze 2D power spectra from the PAPER-32 imaging array in Stokes I, Q, U and V. These allow us to observe and diagnose systematic effects in our calibration at high signal-to-noise within the Fourier space most relevant to EoR experiments. We observe well-defined windows in the Stokes visibilities, with Stokes Q, U and V power spectra sharing a similar wedge shape to that seen in Stokes I. With modest polarization calibration, we see no evidence that polarization calibration errors move power outside the wedge in any Stokes visibility, to the noise levels attained. Deeper integrations will be required to confirm that this behavior persists to the depth required for EoR detection.