Wednesday

For ground-based low surface brightness science, the study of objects hundreds to thousands of times fainter than the night sky, accurate sky subtraction is paramount. While observational techniques can mitigate the night sky's impact on LSB science, removing its flux from images still relies on often temperamental modeling, particularly in photometric bands in which the sky shows complex structure. Additionally, the accurate construction of such models requires large fractions of every image to be free of astrophysical sources, which is simply not possible for some targets. In the worst-case scenarios, inaccurate sky modeling can result in systematic flux loss or gain, skewing results in this already quite sensitive regime of astronomical imaging.

We thus have begun to develop and test a nearly model-independent sky subtraction routine. Briefly, we align, flux-scale, and subtract a preliminary, traditionally sky-subtracted image coadd from each exposure to remove astrophysical objects, which produces noisy but tailored images of the night sky that can be processed into viable sky models of individual exposures. At this early stage of development, tests with idealized model galaxies injected into images prior to sky subtraction show negligible total flux loss due to this technique, with post-sky subtraction surface brightness profiles stable to within 10% down to nearly 33 magnitudes/arcsec^2, far below even the most optimistic depth estimates for surveys like LSST. We conducted some of these tests on a field rife with Galactic cirrus, indicating that the technique is robust as well to foreground contamination. While it does have some short-comings, we are optimistic that this technique can be put to use in large-scale surveys, where night-sky conditions are often not optimal for low surface brightness observations.