Cosmic Reionization On Computers. Mean and Fluctuating Redshifted 21 cm Signal

TL;DR

This paper uses numerical simulations to analyze the mean and fluctuating 21 cm signals during cosmic reionization, highlighting the importance of substructure and challenging previous analytical models.

Contribution

It provides new simulation-based insights into the 21 cm signal's behavior and emphasizes the impact of substructure on power spectrum estimates during reionization.

Findings

Mean 21 cm signal varies between ±20 mK.

Pre-reionization dip extends only to about -20 mK, requiring higher sensitivity.

Neglecting substructure can lead to 25-50% errors in power spectrum estimates.

Abstract

We explore the mean and fluctuating redshifted 21 cm signal in numerical simulations of cosmic reionization from the Cosmic Reionization On Computers (CROC) project. We find that the mean signal varies between about $\pm20\rm{mK}$. Most significantly, we find that the negative pre-reionization dip at $z\sim10-15$ only extends to $\langle\Delta T_B\rangle\sim-20\rm{mK}$, in agreement with prior simulation results and in significant contrast to Pritchard & Loeb analytical model, requiring substantially higher sensitivity from global signal experiments that operate in this redshift range (EDGES-II, LEDA, SCI-HI, and DARE). We also explore the role of dense substructure (filaments and embedded galaxies) in the formation of 21 cm power spectrum. We find that by neglecting the semi-neutral substructure inside ionized bubbles, the power spectrum can be mis-estimated by 25-50\% at scales $k\sim 0.1-1h\rm{Mpc}^{-1}$. This scale range is of a particular interest, because the upcoming 21 cm experiments (MWA, PAPER, HERA) are expected to be most sensitive within it.