Interferometric HI intensity mapping: perturbation theory predictions and foreground removal effects

TL;DR

This paper uses perturbation theory within the Effective Field Theory framework to predict the HI intensity mapping power spectrum, assesses foreground removal impacts, and provides guidance for cosmological parameter estimation with interferometric surveys.

Contribution

It introduces perturbation theory predictions for HI intensity mapping including foreground effects, and evaluates their impact on cosmological parameter constraints.

Findings

Unbiased parameter constraints with <3% errors in ideal conditions.

Foreground removal causes strong biases in key parameters.

Scale cuts are necessary to mitigate biases in parameter estimation.

Abstract

We provide perturbation theory predictions for the HI intensity mapping power spectrum multipoles using the Effective Field Theory of Large Scale Structure (EFTofLSS), which should allow us to constrain cosmological parameters exploiting mildly nonlinear scales. Assuming survey specifications typical of proposed interferometric HI intensity mapping experiments like CHORD and PUMA, and realistic ranges of validity for the perturbation theory modelling, we run mock full shape MCMC analyses at a redshift bin centred at $z=0.5$, and compare with Stage-IV optical galaxy surveys. We include the impact of 21cm foreground removal using simulations-based prescriptions, and quantify the effects on the precision and accuracy of the parameter estimation. We vary 11 parameters in total: 3 cosmological parameters, 7 bias and counterterms parameters, and the HI brightness temperature. Amongst them, the 4 parameters of interest are: the cold dark matter density, $\omega_{\rm c}$, the Hubble parameter, $h$, the primordial amplitude of the power spectrum, $A_{\rm s}$, and the linear HI bias, $b_1$. For the best case scenario, we obtain unbiased constraints on all parameters with $<3\%$ errors at $68\%$ confidence level. When we include the foreground removal effects, the parameter estimation becomes strongly biased for $\omega_{\rm c}, h$, and $b_1$, while $A_{\rm s}$ is less biased ($< 2\sigma$). We find that scale cuts $k_{\rm min} \geq 0.03 \, h/{\mathrm{Mpc}}$ are required to return accurate estimates for $\omega_{\rm c}$ and $h$, at the price of a decrease in the precision, while $b_1$ remains strongly biased. We comment on the implications of these results for real data analyses.