Interpreting the HI 21-cm cosmology maps through Largest Cluster Statistics. Part II. Impact of the realistic foreground and instrumental noise on synthetic SKA1-Low observations

Abstract

The Largest Cluster Statistics\,(LCS) analysis of the redshifted 21\,cm maps has been demonstrated to be an efficient and robust method for following the time evolution of the largest ionized regions\,(LIRs) during the Epoch of Reionization\,(EoR). The LCS can, in principle, constrain the reionization model and history by quantifying the morphology of neutral hydrogen\,(\HI) distribution during the different stages of the EoR. Specifically, the percolation transition of ionized regions, quantified and constrained via LCS, provides a crucial insight about the underlying reionization model. The previous LCS analysis of EoR 21\,cm maps demonstrates that the convolution of the synthesized beam of the radio interferometric arrays, e.g. SKA1-Low with the target signal, shifts the apparent percolation transition of ionized regions towards the lower redshifts. In this study, we present an optimal thresholding strategy to reduce this bias in the recovered percolation transition. We assess the robustness of LCS analysis of the 21\,cm maps, considering the effects of antenna-based gain calibration errors and instrumental noise for SKA1-Low. This analysis is performed using synthetic observations simulated by the \textsc{21cmE2E} pipeline, considering SKA1-Low AA4 configuration within a radius of 2\,km from the array centre. Our findings suggest that a minimum of $2000$\,hours of observation (SNR $\gtrapprox 3$) are required for the LCS analysis to credibly suppress the confusion introduced by thermal noise. Further, we also demonstrate that for a maximum antenna-based calibration error tolerance of $\sim 0.02\%$ (post calibration), the reionization history can be recovered in a robust and relatively unbiased manner using the LCS.

Figures (10)

• Figure 1: Schematic diagram of the 21cmE2E-pipeline based on OSKAR and CASA software. This pipeline is used to estimate LCS from 21 cm observation results.
• Figure 2: Left-hand panel: Telescope layout of the SKA1-Low array assembly 4 (AA4) configuration within a radius of 2 km from the array centre. Right-hand panel: Baseline coverage in the UV plane for an observation time of 2 hours ($\pm 1$ HA). The U, V and -U, -V are plotted here using different colours for visual clarity.
• Figure 3: The visual representation of one such slice of the image cube at $\Bar{x}_{\text{HI}}\approx 0.55$. The observed Hi$21$ cm brightness temperature map: without any corruption (Top Left), with $\sim 0.02\%$ residual calibration errors (Top Middle), and $\sim 0.1\%$ residual calibration errors (Top Right). Bottom: The recovered Hii regions after applying the optimum thresholding method. The white (black) regions in these maps represent the recovered neutral (ionized) regions.
• Figure 4: The pictorial representation of Hi maps at $\Bar{x}_{\rm HI}=0.2$ after performing multiscale cleaning with the natural weighting scheme through the 21cmE2E-pipeline. Left: The observed Hi field without adding noise. Middle: Instrumental noise added to the Hi$21$ cm field for an observation time of $2000$ hours when observed with SKA1-Low with $296$ stations. Right: Identified ionized region from the $21$ cm field after applying optimum thresholding algorithm. The white (black) regions in these maps represent the recovered neutral (ionized) regions.
• Figure 5: Comparison of the obtained LCS from 21 cm observation maps against neutral fraction for different thresholding methods. The original image is a hypothetical scenario without telescope effect and noise, and the corresponding threshold for LCS is set at zero. The red dash-dot and green dashed curves illustrate the obtained LCS based on the threshold set by the optimum thresholding and gradient-descent methods. The optimum thresholding method demonstrates superior performance compared to the gradient-descent method.