[Paper Review] The HI intensity mapping power spectrum: insights from recent measurements

Hamsa Padmanabhan, Roy Maartens+2 more|arXiv (Cornell University)|May 16, 2023

Galaxies: Formation, Evolution, Phenomena4 citations

TL;DR

This paper develops a redshift-space halo model for the HI intensity mapping power spectrum, incorporating 1-halo, shot noise, and finger-of-god effects to model nonlinear scales. It fits the model to recent MeerKAT data, finding good agreement at z=0.32 but evidence for a higher-than-expected power amplitude at z=0.44, possibly due to enhanced bias or baryonic effects.

ABSTRACT

The first direct measurements of the HI intensity mapping power spectrum were recently made using the MeerKAT telescope. These measurements are on nonlinear scales, at redshifts 0.32 and 0.44. We develop a formalism for modelling small-scale power in redshift space, within the context of the mass-weighted HI halo model framework. This model is consistent with the latest findings from surveys on the HI-halo mass relation. In order to model nonlinear scales, we include the 1-halo, shot-noise and finger-of-god effects. Then we apply the model to the MeerKAT auto-correlation data, finding that the model provides a good fit to the data at redshift 0.32, but the data may indicate some evidence for an adjustment at $z \sim 0.44$. Such an adjustment can be achieved by an increase in the HI abundance or halo model bias.

Motivation & Objective

• To develop a theoretical framework for modeling the small-scale HI intensity mapping power spectrum in redshift space, accounting for nonlinear structure and redshift-space distortions.
• To incorporate the 1-halo term, shot noise, and finger-of-god effects within a mass-weighted halo model for HI, consistent with recent HI-halo mass relation constraints.
• To test the model against the first direct MeerKAT measurements of the HI auto-correlation power spectrum on nonlinear scales at z=0.32 and z=0.44.
• To investigate discrepancies between theoretical predictions and data at z=0.44, exploring possible causes such as higher-order bias or complex baryonic physics.
• To predict two-dimensional redshift-space power spectra for future comparison with high-resolution intensity mapping surveys.

Proposed method

• Adapts the halo model formalism to HI intensity mapping, treating HI as a mass-weighted tracer of dark matter haloes.
• Incorporates redshift-space distortions via the finger-of-god effect using a weighted average damping parameter, following Zhang et al. (2020).
• Derives analytical expressions for the monopole of the redshift-space power spectrum, including contributions from 1-halo, 2-halo, and shot noise terms.
• Applies the model to MeerKAT auto-correlation data using a likelihood fitting procedure to constrain parameters such as HI bias and normalization.
• Uses the halo model parameters calibrated from stacking HI galaxy surveys (Bera et al., 2022) to inform the model.
• Predicts two-dimensional power spectra in (k⊥, k∥) space, with a conservative horizon limit imposed to avoid foreground contamination.

Experimental results

Research questions

• RQ1 Does the halo model with redshift-space distortions provide a good fit to the first direct MeerKAT measurements of the HI intensity mapping power spectrum on nonlinear scales?
• RQ2 Why does the model underpredict the power spectrum amplitude at z=0.44 compared to the data, despite good agreement at z=0.32?
• RQ3 What physical mechanisms could explain the observed excess power at z=0.44, such as higher-order bias or baryonic effects?
• RQ4 How do the predicted two-dimensional power spectra in redshift space compare to the monopole measurements, and what insights do they offer?
• RQ5 To what extent do vorticity corrections or non-standard velocity correlations between HI and dark matter affect the power spectrum on small scales?

Key findings

• The model provides a good fit to the MeerKAT data at z=0.32, successfully capturing the small-scale structure of the power spectrum including finger-of-god suppression.
• At z=0.44, the data show a higher-than-predicted power amplitude, suggesting a possible need for an increased HI halo model bias or additional physical effects.
• The fitting procedure favors the standard shot noise term over the finger-of-god modulated version at small scales, indicating that the suppression of power is well described by velocity dispersion alone.
• The predicted two-dimensional power spectra in redshift space show distinct anisotropies, with the horizon limit (k∥ ≈ 0.26k⊥) marking the conservative boundary for foreground contamination.
• Vorticity corrections to the power spectrum are found to be too small to explain the observed discrepancy at z=0.44, though they remain a possible contributor in future work.
• The model is consistent with HI-halo mass relation constraints from stacking surveys and supports the use of mass-weighted tracers in intensity mapping cosmology.