Real or Interloper? The Redshift Likelihoods of z>8 Galaxies in the HUDF12

Abstract

In the absence of spectra, fitting template model spectra to observed photometric fluxes, known as Spectral Energy Distribution (SED) fitting, has become the workhorse for identifying high-z galaxies. In this paper, we present an analysis of the most recent and possibly most distant galaxies discovered in the Hubble Ultra Deep Field using a more robust method of redshift estimation based on Markov Chain Monte Carlo fitting (MCMC), rather than relying on the redshift of "best fit" models obtained using common chi^2 minimization techniques. The advantage of MCMC fitting is the ability to accurately estimate the probability density function of the redshift, as well as any other input model parameters, allowing us to derive accurate credible intervals by properly marginalizing over all other input model parameters. We apply our method to 13 recently identified sources and show that, despite claims based on chi^2 minimization, none of these sources can be securely ruled out as low redshift interlopers given the low signal-to-noise of currently available observations. We estimate that there is an average probability of 21% that these sources are low redshift interlopers.