Characterizing unknown systematics in large scale structure surveys

TL;DR

This paper introduces a method to estimate the magnitude of unknown systematics in large scale structure surveys by analyzing cross-correlations, helping to improve cosmological measurements despite calibration uncertainties.

Contribution

It develops a novel approach to characterize unknown systematics using cross-correlations, enabling better data quality control in large scale structure surveys.

Findings

The method estimates contamination levels from unknown systematics.

Discarding contaminated bins improves bias estimates.

Application to SDSS data demonstrates practical effectiveness.

Abstract

Photometric large scale structure (LSS) surveys probe the largest volumes in the Universe, but are inevitably limited by systematic uncertainties. Imperfect photometric calibration leads to biases in our measurements of the density fields of LSS tracers such as galaxies and quasars, and as a result in cosmological parameter estimation. Earlier studies have proposed using cross-correlations between different redshift slices or cross-correlations between different surveys to reduce the effects of such systematics. In this paper we develop a method to characterize unknown systematics. We demonstrate that while we do not have sufficient information to correct for unknown systematics in the data, we can obtain an estimate of their magnitude. We define a parameter to estimate contamination from unknown systematics using cross-correlations between different redshift slices and propose discarding bins in the angular power spectrum that lie outside a certain contamination tolerance level. We show that this method improves estimates of the bias using simulated data and further apply it to photometric luminous red galaxies in the Sloan Digital Sky Survey as a case study.