exoALMA IX: Regularized Maximum Likelihood Imaging of Non-Keplerian Features

Abstract

The planet-hunting ALMA large program exoALMA observed 15 protoplanetary disks at ~0.15" angular resolution and ~100 m/s spectral resolution, characterizing disk structures and kinematics in enough detail to detect non-Keplerian features (NKFs) in the gas emission. As these features are often small and low-contrast, robust imaging procedures are critical for identifying and characterizing NKFs, including determining which features may be signatures of young planets. The exoALMA collaboration employed two different imaging procedures to ensure the consistent detection of NKFs: CLEAN, the standard iterative deconvolution algorithm, and regularized maximum likelihood (RML) imaging. This paper presents the exoALMA RML images, obtained by maximizing the likelihood of the visibility data given a model image and subject to regularizer penalties. Crucially, in the context of exoALMA, RML images serve as an independent verification of marginal features seen in the fiducial CLEAN images. However, best practices for synthesizing RML images of multi-channeled (i.e. velocity-resolved) data remain undefined, as prior work on RML imaging for protoplanetary disk data has primarily addressed single-image cases. We used the open source Python package MPoL to explore RML image validation methods for multi-channeled data and synthesize RML images from the exoALMA observations of 7 protoplanetary disks with apparent NKFs in the 12CO J=3-2 CLEAN images. We find that RML imaging methods independently reproduce the NKFs seen in the CLEAN images of these sources, suggesting that the NKFs are robust features rather than artifacts from a specific imaging procedure.