Photometric Redshift with Bayesian Priors on Physical Properties of Galaxies

TL;DR

This paper introduces a Bayesian approach to photometric redshift estimation that incorporates physical galaxy properties as priors, improving accuracy and enabling simultaneous measurement of redshifts and galaxy characteristics.

Contribution

It presents a novel method combining Bayesian priors on physical properties with template fitting, enhancing photometric redshift accuracy and consistency in galaxy property estimation.

Findings

Priors reduce degeneracy in photometric redshift estimation.

Template error functions improve flux modeling accuracy.

Simultaneous measurement of redshift and physical properties is feasible.

Abstract

We present a proof-of-concept analysis of photometric redshifts with Bayesian priors on physical properties of galaxies. This concept is particularly suited for upcoming/on-going large imaging surveys, in which only several broad-band filters are available and it is hard to break some of the degeneracies in the multi-color space. We construct model templates of galaxies using a stellar population synthesis code and apply Bayesian priors on physical properties such as stellar mass and star formation rate. These priors are a function of redshift and they effectively evolve the templates with time in an observationally motivated way. We demonstrate that the priors help reduce the degeneracy and deliver significantly improved photometric redshifts. Furthermore, we show that a template error function, which corrects for systematic flux errors in the model templates as a function of rest-frame wavelength, delivers further improvements. One great advantage of our technique is that we simultaneously measure redshifts and physical properties of galaxies in a fully self-consistent manner, unlike the two-step measurements with different templates often performed in the literature. One may rightly worry that the physical priors bias the inferred galaxy properties, but we show that the bias is smaller than systematic uncertainties inherent in physical properties inferred from the SED fitting and hence is not a major issue. We will extensively test and tune the priors in the on-going Hyper Suprime-Cam survey and will make the code publicly available in the future.