DESI Complete Calibration of the Color-Redshift Relation (DC3R2): Results from early DESI data

TL;DR

This paper presents initial results from DESI's DC3R2 survey, calibrating the color-redshift relation using 230k galaxies with multi-band photometry to improve photometric redshift estimates for weak lensing surveys.

Contribution

It introduces a calibration method for the color-redshift relation using a large spectroscopic sample, addressing systematic uncertainties and photometric noise impacts.

Findings

Spectroscopic coverage spans 56% of Euclid and LSST color space.

Photometric noise causes apparent magnitude dependence of redshift at fixed color.

Systematic uncertainties from different photometry sources are analyzed.

Abstract

We present initial results from the Dark Energy Spectroscopic Instrument (DESI) Complete Calibration of the Color-Redshift Relation (DC3R2) secondary target survey. Our analysis uses 230k galaxies that overlap with KiDS-VIKING $ugriZYJHK_s$ photometry to calibrate the color-redshift relation and to inform photometric redshift (photo-z) inference methods of future weak lensing surveys. Together with Emission Line Galaxies (ELGs), Luminous Red Galaxies (LRGs), and the Bright Galaxy Survey (BGS) that provide samples of complementary color, the DC3R2 targets help DESI to span 56% of the color space visible to Euclid and LSST with high confidence spectroscopic redshifts. The effects of spectroscopic completeness and quality are explored, as well as systematic uncertainties introduced with the use of common Self Organizing Maps trained on different photometry than the analysis sample. We further examine the dependence of redshift on magnitude at fixed color, important for the use of bright galaxy spectra to calibrate redshifts in a fainter photometric galaxy sample. We find that noise in the KiDS-VIKING photometry introduces a dominant, apparent magnitude dependence of redshift at fixed color, which indicates a need for carefully chosen deep drilling fields, and survey simulation to model this effect for future weak lensing surveys.