Enhancing the detectability of ionized Regions during the Epoch of Reionization

TL;DR

The study addresses the challenge of detecting and characterizing large ionized regions during the Epoch of Reionization in 21 cm data by extending a Bayesian matched-filter approach to use an eight-parameter spheroidal filter, improving over prior spherical filters. The method is validated on mock SKA-low visibilities across configurations AA2, AA*, and AA4, showing higher SNRs and robust orientation information for non-spherical bubbles, with representative results at $ar{x}_{\mathrm{HII}} = 0.4$ and $z=7$ yielding around a 10σ detection in about 1 hour for suitable layouts. Realistic reionization scenarios demonstrate significant SNR gains and morphology insights (via orientation parameters $\alpha_X$ and $\gamma_Z$), enabling targeted deep follow-ups and better constraints on the ionizing source populations. The work emphasizes the practical utility of combined detection-imaging strategies for the EoR and discusses extensions to Cosmic Dawn and synergies with JWST-era infrared observations, while noting optimistic assumptions about perfect calibration and foreground removal.

Abstract

We present an improved matched filter method for detecting large ionized regions in 21 cm observations of the Epoch of Reionization. In addition to detection, the method constrains the properties of these regions, offering insights into the underlying source populations. Extending a previously developed Bayesian framework, we replace the spherical filter with an eight-parameter spheroidal filter, enabling a more flexible characterization of ionized bubbles. This enhancement significantly improves both detectability and recovery of bubble orientations. For a representative reionization scenario with mean ionization fraction $0.4$ at $z=7$, we find that a $10σ$ detection of the largest ionized region can be achieved with $\sim 1$ h of observations using the SKA-low AA4 and AA$^{\star}$ layouts. Our method can help identify regions in the observed field that host large ionized bubbles, making them prime targets for deeper follow-up observations.

Figures (13)

• Figure 1: Layouts of the three stages in the construction of the SKA-low telescope. The panels from left to right represent AA2, AA$^{\star}$, and AA4 SKA-low configurations with $68$, $307$, and $512$ antennas, respectively.
• Figure 2: UV coverage at frequency $\nu=177$ MHz for the three SKA-low array layouts at different construction stages. The panels from left to right represent AA2, AA$^{\star}$, and AA4 SKA-low configurations, respectively. These correspond to mock observations at a part of the sky with right ascension $0^{\circ}$ and declination $-30^{\circ}$ for $4$ h daily.
• Figure 3: This displays the relation of the number density of antenna pairs having baseline $U$ at frequency 177 MHz for different configurations of SKA-low.
• Figure 4: Two-dimensional slices of the simulated $\delta T_{\rm b}$ light cones. The left-to-right panels correspond to considering the Galaxy-noQSO, Galaxy-QSO10Myr and Galaxy-QSO30Myr scenarios, respectively, at $z=7$. The red dot points show the location of the rare quasar formed in the most massive dark matter halo in the simulation box.
• Figure 5: Two-dimensional slices of the brightness temperature from the fiducial simulated light-cone considering the Galaxy-QSO10Myr scenario and an observation time of 20 h, selected around the most massive dark matter halo. From left to right, the panels show the expected image (in the absence of noise) at $z=7$ using the SKA-low AA4, AA$^{\star}$, and AA2 configurations. The original simulated map resolution is $1.075$ arcmin, corresponding to a physical scale of $2.76$ Mpc.