How often is a Far-Infrared detection a blend of multiple galaxies?
The Far Infrared (FIR) wavelength is sensitive to hot dust in galaxies, and is particularly useful in detecting distant, highly star-forming galaxies at high redshift. However, FIR telescopes with a single mirror (both in space and on the ground) suffer from poor angular resolution, which makes it difficult to compare the FIR data to high resolution data at other wavelengths. Historically, each FIR-detected source has been assumed to be light coming from a single source, which implies very extreme star formation rates, currently hard to reproduce in models of galaxy formation.
To investigate the accuracy of assigning all the FIR flux to a single object, I use a Bayesian inference too, XID+, developed by Dr. Peter D. Hurley as part of the Herschel Extragalactic Legacy Project (HELP). HELP's main goal is to homogenize a large sample of multi-wavelength data products across all of the Herschel Legacy fields.
XID+ uses the positions of known sources from other wavelengths (in this work, 3.6 micron and 24 micron sources) to find the best distribution of the low-resolution Herschel data amongst these known sources. Typically, there are ~14 known sources beneath each 250 micron Herschel detection, but only 3–4 sources are found to be likely contributors to the FIR detection.
In investigating the brightest and second brightest contributors, we find that typically the brightest contributor contains between 40% and 60% of the total flux, with the brightest 250 micron sources divided more evenly between contributors. The second brightest contributor contains typically 20% of the available FIR flux. Overall, we find that in our sample, 95% of all FIR sources are combinations of at least two contributors, and not single, very bright sources.