Parameterized reconstruction with random scales for radio synthesis imaging

¹ College of Big Data and Information Engineering, Guizhou University, Guiyang 550025, PR China
e-mail: lizhang.science@gmail.com
² Xinjiang Astronomical Observatory, Chinese Academy of Sciences, Urumqi 830011, PR China
³ Key Laboratory of Radio Astronomy, Chinese Academy of Sciences, Urumqi 830011, PR China
⁴ Key Laboratory of Solar Activity, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, PR China
⁵ Center for Astrophysics, Guangzhou University, Guangzhou 510006, PR China

Received: 27 August 2020
Accepted: 7 December 2020

Abstract

Context. In radio interferometry, incomplete sampling results in a dirty beam with side lobes, which obscures the celestial structures. Before the astrophysical analysis, the effects of the dirty beam need to be eliminated, which can be solved with various deconvolution methods.

Aims. Diffuse astronomical sources observed by modern high-sensitivity telescopes tend to be complex morphological structures, often accompanied by faint features, which are submerged under the side lobes of the dirty beam. We propose a new deconvolution algorithm called random multiscale estimator (RMS-Clean), which is mainly used to solve the difficult reconstruction of diffuse astronomical sources.

Methods. RMS-Clean models the sky brightness distribution as a linear combination of random multiscale basis functions whose scales are obtained by randomly perturbing a preset multiscale list. Random multiscale models are used to approximate the uncertain characteristics of the scales of complex astronomical sources.

Results. When the RMS-Clean method is applied to simulations of SKA observations with realistic diffuse structures, it can reconstruct diffuse structures well and provides a competitive result compared to the commonly used deconvolution algorithms.