The Herschel Deconfusion Project: Constraining the Cosmic Evolution of Dust in Galaxies Using the Deepest Available Data
Abstract
With the end of the Herschel mission and no immediate successor at far-infrared wavelengths, it is imperative to extract as much information as possible from the existing data. The difference between the theoretical noise limit and the confusion limit suggests that significant improvements can be made with a more sophisticated treatment of source confusion. This is possible because we have a lot of information about the Herschel deep fields from other wavelengths. The project will use existing already-reduced data from Herschel's deepest observations, which targeted the CANDELS. These data have a wealth of observations from Hubble, Spitzer, Chandra and many other telescopes. The main work will be to develop and employ a new Bayesian technique that incorporates spectral-energy-distribution priors to constrain the range of likely far-infrared fluxes for each source that is detected by Hubble. The far-IR images are then segmented and the regions which are likely to suffer the most confusion are simultaneously fit, using the (broad) constraints on the likely farIR fluxes as a Bayesian prior. The first pass of photometry will yield reliable photometry for sources at least a factor of two fainter than existing catalogs. Subsequent passes can yield full probability distributions for the ensemble Far-IR SEDs of much fainter sources (overcoming some of the limitations of stacking in image space). We will used the improved and deeper FIR photometry to address two "crises" in reconciling galaxy evolution models with high-z galaxy observations: (1) the surprisingly young ages of most bright Lyman-break galaxies at redshift z=3 and (2) the surprisingly high star-formation rates and dust masses high-redshift sub-mm and FIR-selected galaxies. The former could potentially be explained if many of the descendants of UVbright galaxies at z=4 have too much dust by z=3 to be included in Lyman-break samples. The latter problem could be resolved if the fluxes of many FIR and sub-mm selected galaxies are affected by blending. The project will employ state-of-the art semi-analytical models for galaxy evolution, both for guidance in developing flexible Bayesian priors, and for guidance on the interpretation of the results. As part of the work we plan to further test and improve the treatment of dust in these models.
- Publication:
-
NASA ADAP Proposal
- Pub Date:
- 2014
- Bibcode:
- 2014adap.prop..234F