Measuring galaxy environments in large scale photometric surveys

TL;DR

This paper evaluates how redshift uncertainties in large photometric surveys affect galaxy environment measurements, comparing methods and quantifying the impact on detecting environment-related galaxy properties.

Contribution

It systematically studies the impact of photometric redshift errors on environment measurements and proposes optimal parameters for reliable analysis in large surveys.

Findings

Photometric environments have a smaller dynamic range than spectroscopic ones.

At DES-like redshift uncertainty of 0.1, correlation coefficient is 0.4.

Photometric samples need to be 6-16 times larger to match spectroscopic environment detection.

Abstract

The properties of galaxies in the local universe have been shown to depend upon their environment. Future large scale photometric surveys such as DES and Euclid will be vital to gain insight into the evolution of galaxy properties and the role of environment. Large samples come at the cost of redshift precision and this affects the measurement of environment. We study this by measuring environments using SDSS spectroscopic and photometric redshifts and also simulated photometric redshifts with a range of uncertainties. We consider the Nth nearest neighbour and fixed aperture methods and evaluate the impact of the aperture parameters and the redshift uncertainty. We find that photometric environments have a smaller dynamic range than spectroscopic measurements because uncertain redshifts scatter galaxies from dense environments into less dense environments. At the expected redshift uncertainty of DES, 0.1, there is Spearman rank correlation coefficient of 0.4 between the measurements using the optimal parameters. We examine the galaxy red fraction as a function of mass and environment using photometric redshifts and find that the bivariate dependence is still present in the SDSS photometric measurements. We show that photometric samples with a redshift uncertainty of 0.1 must be approximately 6-16 times larger than spectroscopic samples to detect environment correlations with equivalent fractional errors.