CHILES VII: Deep Imaging for the CHILES project, a SKA prototype

TL;DR

This paper discusses the deep imaging techniques and systematic error mitigation for the CHILES HI survey, a precursor to SKA, emphasizing methods to achieve high-fidelity images in deep radio surveys.

Contribution

It introduces novel approaches for creating high-quality image cubes from deep, wideband, multi-epoch radio data and addresses systematic errors relevant for future large-scale surveys.

Findings

Systematic errors can be effectively managed in the visibility domain.

High-fidelity images are achievable with current methods in deep HI surveys.

Lessons learned are applicable to SKA and ngVLA deep imaging projects.

Abstract

Radio Astronomy is undergoing a renaissance, as the next-generation of instruments provides a massive leap forward in collecting area and therefore raw sensitivity. However, to achieve this theoretical level of sensitivity in the science data products we need to address the much more pernicious systematic effects, which are the true limitation. These become all the more significant when we consider that much of the time used by survey instruments, such as the SKA, will be dedicated to deep surveys. CHILES is a deep HI survey of the COSMOS field, with 1,000 hours of VLA time. We present our approach for creating the image cubes from the first Epoch, with discussions of the methods and quantification of the data quality from 946 to 1420MHz -- a redshift range of 0.5 to 0. We layout the problems we had to solve and describe how we tackled them. These are of importance as CHILES is the first deep wideband multi-epoch HI survey and it has relevance for ongoing and future surveys. We focus on the accumulated systematic errors in the imaging, as the goal is to deliver a high-fidelity image that is only limited by the random thermal errors. To understand and correct these systematic effects we ideally manage them in the domain in which they arise, and that is predominately the visibility domain. CHILES is a perfect test bed for many of the issues we can expect for deep imaging with the SKA or ngVLA and we discuss the lessons we have learned.