Characterizing line-of-sight variability of polarized dust emission with future CMB experiments

TL;DR

This study investigates how superposition of multiple dust clouds along the line of sight affects the accuracy of foreground modeling in CMB experiments, highlighting potential biases and the limits of parameter recovery.

Contribution

It models the spectral energy distribution as an integral over dust cloud properties and assesses the biases in simple two-parameter MBB fits, proposing probabilistic approaches for better parameter estimation.

Findings

Simple MBB fits can unbiasedly estimate CMB polarization unless SED and polarization angle vary significantly.

Fitting broad dust property distributions biases temperature and spectral index estimates.

Only broad distributions' parameters can be reliably recovered from single-line-of-sight SED fitting.

Abstract

Galactic dust emission is often accounted for in cosmic microwave background (CMB) analyses by fitting a simple two-parameter modified blackbody (MBB) model in each pixel, which nominally accounts for the temperature and opacity of the dust. While this may be a good approximation for individual dust clouds, typically a number of such clouds are found along each line of sight and within each angular pixel, resulting in a superposition of their spectra. In this paper, we study the effects of this superposition on pixel-based foreground fitting strategies by modelling the spectral energy distribution (SED) in each pixel as the integral of individual MBB spectra over various physically-motivated statistical distributions of dust cloud properties. We show that fitting these SEDs with the simple two-parameter MBB model generally results in unbiased estimates of the CMB Stokes Q and U amplitudes in each pixel, unless there are significant changes in both the dust SED and polarization angle along the line of sight, in which case significant ($ > 10\sigma$) biases are observed in an illustrative model. We also find that the best-fit values of the dust temperature, $T_d$, and spectral index, $\beta_d$, are significantly biased away from the mean/median of the corresponding statistical distributions when the distributions are broad, suggesting that MBB model fits can give an unrepresentative picture of the physical properties of the dust at microwave wavelengths if not interpreted carefully. Using a Fisher matrix analysis, we also determine the experimental sensitivity required to recover the parameters of the $T_d$ and $\beta_d$ distributions themselves by fitting a probabilistic MBB model, finding that only the parameters of broad distributions can be measured by SED fitting on a single line of sight.