Presentation #541.12 in the session “Computational Augmentation to Surveys and Science Programs”.
This study focuses on applying source separation methods to multi-band astronomical images from different datasets. We have used 5-band imaging obtained with the Hyper Suprime Cam (HSC) from the Subaru telescope along with single-band images from the Hubble Space Telescope (HST), Both datasets represent the same objects in the sky, but with different resolution and color composition. The study applied the deblending package SCARLET by Melchior et al. (2018) as a source separation tool to study the properties of the quasars host galaxies. As the images from the HSC survey come in 5 spectral bands, it provided us with the necessary information about the color of the host galaxies, in parallel, the mono-band high-resolution HST images were used to constrain the morphologies of the host and help in separating the host from the point source (i.e.quasar) in its center. We could see that our modeling technique has enhanced the detection of both point sources and especially extended faint sources by leveraging the features of high-resolution images and also color as a constraint to avoid overfitting of undetected sources. The study identified some issues like the effect of astrometry on shifting the detection of sources, which makes it hard to build an accurate model, but we think this can be overcome in the future with more advanced correction software for telescopes, especially for space telescopes. Our model could be better enhanced in terms of interpolating the PSF small size to a higher resolution model as well.