Evaluation of the ngVLA Revision D array configuration for stellar imaging

TL;DR

This study evaluates the ngVLA Revision D array configuration's effectiveness for stellar imaging, showing improvements over Revision C and highlighting the advantages of RML imaging methods for complex stellar structures.

Contribution

It compares the imaging capabilities of Rev D and Rev C configurations, demonstrating enhanced beam quality and the robustness of RML methods for stellar imaging.

Findings

Rev D improves the synthesized beam and image reconstruction quality.

RML methods perform comparably or better than CLEAN across configurations.

Rev D enhances imaging for complex stellar morphologies.

Abstract

A transformative science case for the proposed next-generation Very Large Array (ngVLA) is resolving the surfaces of nearby stars, both spatially and temporally, enabled by the combination of milliarcsecond-scale resolution and unprecedented sensitivity to thermal radio emission. In a previous study, we demonstrated the feasibility of stellar imaging with simulated observations of nearby stars, using both traditional CLEAN techniques and newly developed regularized maximum likelihood (RML) imaging methods for image reconstruction. In this memo, we present a continued study of stellar imaging with the ngVLA, evaluating the imaging capability of the Revision D (henceforth Rev D) Main Array configuration compared to the previous Revision C (henceforth Rev C) configuration. We find that the Rev D configuration, with more uniform coverage and better circular symmetry, improves the synthesized beam, resulting in better CLEAN reconstructions of simulated images of evolved stars with complex morphology, especially with robust weighting. However, the highly non-Gaussian nature of the synthesized beam still persists with both robust and natural weightings in the Rev D configuration and continues to limit the image fidelity of image reconstructions with non-uniform weighting. The RML methods show stable performance that is resilient to different array configurations with image quality comparable to or better than CLEAN methods in the presented simulation, consistent with our previous work. Our simulation results suggest that the Rev D configuration will provide a better deconvolution beam compared with the Rev C configuration, which would enhance the imaging capability for non-uniform weighting, and they continue to demonstrate that RML methods are an attractive choice, even for the improved array configuration.