Title:The Impact of $\texttt{CLEAN}$ing on Strong Gravitational Lens Modelling

Abstract:We present a comparison of image and uv-plane galaxy-galaxy strong lensing modelling results for simulated ALMA observations with different antenna configurations and on-source integration times. Image-plane modelling is carried out via use of the $\texttt{CLEAN}$ algorithm, and we explore the effects of different visibility weighting schemes on the inferred lens model parameters. We find that direct modelling of the visibility data consistently outperforms image-plane modelling for both the naturally and Briggs-weighted images. We also find that the modelling of images created with Briggs weighting generally produces more accurate results than those obtained by modelling images constructed with natural weighting. We explain this by quantifying the suppression of information due to $\texttt{CLEAN}$ing on scales at which the modelling is sensitive, and how this differs between Briggs and natural weighting. At higher resolutions, the differences between the lens modelling techniques are much less pronounced and overall, modelling errors are significantly reduced. We also find that time-binning the visibilities by up to a factor of three makes no significant difference to the inferred lens parameters when directly modelling in the uv-plane. This work provides some guidance on navigating the many choices faced when modelling strong lens interferometric imaging data.