SKA-Low simulations for a cosmic dawn/epoch of reionisation deep field

TL;DR

This work presents a realistic end-to-end SKA-Low CD/EoR simulation designed to benchmark foreground-mitigation methods. It combines a detailed sky model (CD/EoR, Galactic diffuse emission, and extragalactic foregrounds) with a full telescope model (512 SKA-Low stations in a Vogel layout) and a comprehensive error model including partial de-mixing, ionospheric DD effects, and DI/thermal noise, producing visibilities, images, and EoR power spectra. The dataset serves as a benchmark for the SKA Science Data Challenge 3a and demonstrates the complex interaction between sky components and instrumental systematics, guiding the development of multi-stage foreground-removal pipelines. While acknowledging simplifications (e.g., a single track and station layout), the authors provide open access to data and code to enable ongoing testing and improvement of EoR signal recovery in realistic observing conditions.

Abstract

We present a realistic simulation of an SKA-Low cosmic dawn/epoch of reionisation (CD/EoR) observation, which can be used to further the development of foreground-mitigation approaches. The simulation corresponds to a deep (1000 h) integration pointing over the 106 MHz-196 MHz frequency range. The sky components include the CD/EoR signal, extragalactic foreground emission featuring strong (over 5 Jy at 150 MHz) out-of-field sources and in-field sources down to 1 microJy at 150 MHz, and Galactic emission from the GSM2016 model complemented with small-scales structure beyond its native $\sim 1$ deg resolution from a magneto-hydrodynamic simulation of the interstellar medium. Modeled errors include a partial de-mixing of the out-of-field sources, direction-dependent calibration errors leading to residual ionospheric effects, and direction-independent gain calibration errors, on top of thermal noise. Simulated observations are delivered as visibilities as well as imaging products with natural weighting. The true, uncorrupted, CD/EoR signal is also delivered, to allow an assessment of the efficacy of foreground-mitigation approaches. The codes used to generate these simulations are also delivered, so that new simulated datasets can be produced. This simulation has been the basis for the SKA Science Data Challenge 3a (SDC3a), which addressed foreground removal.

Figures (8)

• Figure 1: $u,v$ coverage corresponding to the simulated observation.
• Figure 2: Differential source counts for the extragalactic sources inserted in the simulation. We also indicated the flux limits used for the out-of-field sources, the sources included as a source list, and those included as a coarse and fine grid. For the T-RECS sources (triangles) the counts are also shown separately for AGN (dashed line) and SFGs (dash-dotted line). In total we included over 15 million sources.
• Figure 3: 21cmFAST parameters used for the generation of the EoR signal and corresponding mean (red) and standard deviation (blue) of the brightness temperature as a function of frequency. The vertical dotted lines represent the first and last frequency channel in the simulation.
• Figure 4: Lightcone in physical quantities corresponding to the chosen reionisation scenario over the redshift range of the simulation.
• Figure 5: Primary beam response at 106 MHz, which shows the effective side-lobe level used to represent successful de-mixing. The vertical line shows the edge of the 8$\times$8 deg field.