How far can we push deconvolution? A SCUBA-2 test case

TL;DR

This paper investigates the limits of deconvolving confused far-infrared images using multi-wavelength data, demonstrating improved resolution and sensitivity through simulated SCUBA-2 data and its combination with TolTEC survey data.

Contribution

It introduces a method to deconvolve confused 850 micron SCUBA-2 data and explores the scientific benefits of combining it with TolTEC survey data.

Findings

Achieved 0.7 mJy RMS sensitivity at 850 microns after deconvolution.

Reconstructed SCUBA-2 data to the resolution of 1.1mm LMT.

Combined data constrains cold dust temperatures (<10K).

Abstract

How far can we use multi-wavelength cross-identifications to deconvolve far-infrared images? In this short research note I explore a test case of CLEAN deconvolutions of simulated confused 850 micron SCUBA-2 data, and explore the possible scientific applications of combining this data with ostensibly deeper TolTEC Large Scale Structure (LSS) survey 1.1mm-2mm data. I show that the SCUBA-2 can be reconstructed to the 1.1mm LMT resolution and achieve an 850 micron deconvolved sensitivity of 0.7 mJy RMS, an improvement of at least ~1:5x over naive point source filtered images. The TolTEC/SCUBA-2 combination can constrain cold (<10K) observed-frame colour temperatures, where TolTEC alone cannot.