Robust Optical Richness Estimation with Reduced Scatter

TL;DR

This paper presents a robust method for estimating optical richness of galaxy clusters that reduces scatter and is adaptable to various data conditions, improving cluster mass estimates for cosmology.

Contribution

We introduce a modified richness estimator that minimizes scatter, is data-calibrated, and robust against background uncertainties and filter choices.

Findings

Deeper luminosity cuts reduce scatter in X-ray luminosity at fixed richness.

Including blue galaxies increases the scatter in richness estimates.

The estimator is robust to background uncertainties and filter variations.

Abstract

Reducing the scatter between cluster mass and optical richness is a key goal for cluster cosmology from photometric catalogs. We consider various modifications to the red-sequence matched filter richness estimator of Rozo et al. (2009), and evaluate their impact on the scatter in X-ray luminosity at fixed richness. Most significantly, we find that deeper luminosity cuts can reduce the recovered scatter, finding that sigma_lnLX|lambda=0.63+/-0.02 for clusters with M_500c >~ 1.6e14 h_70^-1 M_sun. The corresponding scatter in mass at fixed richness is sigma_lnM|lambda ~ 0.2-0.3 depending on the richness, comparable to that for total X-ray luminosity. We find that including blue galaxies in the richness estimate increases the scatter, as does weighting galaxies by their optical luminosity. We further demonstrate that our richness estimator is very robust. Specifically, the filter employed when estimating richness can be calibrated directly from the data, without requiring a-priori calibrations of the red-sequence. We also demonstrate that the recovered richness is robust to up to 50% uncertainties in the galaxy background, as well as to the choice of photometric filter employed, so long as the filters span the 4000 A break of red-sequence galaxies. Consequently, our richness estimator can be used to compare richness estimates of different clusters, even if they do not share the same photometric data. Appendix 1 includes "easy-bake" instructions for implementing our optimal richness estimator, and we are releasing an implementation of the code that works with SDSS data, as well as an augmented maxBCG catalog with the lambda richness measured for each cluster.