[Paper Review] Photometric redshifts and clustering of emission line galaxies selected jointly by DES and eBOSS

S. Jouvel, Timothée Delubac+66 more|arXiv (Cornell University)|Sep 23, 2015

Galaxies: Formation, Evolution, Phenomena4 citations

TL;DR

This study presents photometric redshifts and clustering measurements for emission-line galaxies (ELGs) selected from the Dark Energy Survey (DES) and observed via the eBOSS spectroscopic survey. Using 9.2 deg² of test plate data, it achieves a 71% success rate in the 0.6 < z < 1.2 redshift window, measures galaxy bias values of 1.72 ± 0.1 and 1.78 ± 0.12 for bright and faint samples, and demonstrates that combining template fitting and random forest algorithms reduces photometric redshift outlier fractions while maintaining high completeness.

ABSTRACT

We present the results of the first test plates of the extended Baryon Oscillation Spectroscopic Survey. This paper focuses on the emission line galaxies (ELG) population targetted from the Dark Energy Survey (DES) photometry. We analyse the success rate, efficiency, redshift distribution, and clustering properties of the targets. From the 9000 spectroscopic redshifts targetted, 4600 have been selected from the DES photometry. The total success rate for redshifts between 0.6 and 1.2 is 71\% and 68\% respectively for a bright and faint, on average more distant, samples including redshifts measured from a single strong emission line. We find a mean redshift of 0.8 and 0.87, with 15 and 13\% of unknown redshifts respectively for the bright and faint samples. In the redshift range 0.6

Motivation & Objective

• To evaluate the performance of DES photometry in selecting emission-line galaxies (ELGs) for the eBOSS spectroscopic survey.
• To measure the redshift distribution, success rate, and stellar contamination of ELG target selections based on DES photometry.
• To assess the clustering properties and galaxy bias of DES-selected ELG samples in the redshift range 0.6 < z < 1.2.
• To investigate photometric redshift accuracy and reduce outlier fractions using hybrid techniques combining template fitting and machine learning.
• To validate the consistency of clustering measurements using both spectroscopic and photometric redshifts for large-scale structure analysis.

Proposed method

• Target selection of ELGs using DES griz photometry from SVA1 data, with color-based cuts to isolate emission-line galaxies.
• Spectroscopic redshifts obtained from eBOSS test plates covering 9.2 deg², enabling direct comparison with photometric redshifts.
• Photometric redshifts computed using template fitting (LePhare) and machine learning (random forest, TPZ), with outlier fraction assessed via comparison between methods.
• Galaxy clustering analyzed via 3D two-point correlation function monopole ξ(s) and projected real-space correlation function w(rp), used to compute large-scale galaxy bias.
• Systematic effects (dust, airmass, depth, stellar contamination) were quantified to ensure their impact on power spectrum measurements remains below 15%.
• Outlier reduction strategies applied by removing color regions with >10% outlier fractions or by excluding 15% of the sample based on template vs. machine learning redshift discrepancies.

Experimental results

Research questions

• RQ1 What is the success rate and redshift distribution of ELG target selection using DES photometry for the eBOSS survey?
• RQ2 How does the galaxy bias of DES-selected ELGs compare to theoretical expectations and previous surveys like VIPERS?
• RQ3 To what extent can photometric redshift outlier fractions be reduced using hybrid methods combining template fitting and machine learning?
• RQ4 How do clustering measurements based on spectroscopic versus photometric redshifts compare in the 0.6 < z < 1.2 window?
• RQ5 What are the dominant systematic effects in the target selection process, and do they remain within acceptable thresholds for cosmological power spectrum analysis?

Key findings

• The bright DES ELG sample achieved a 71% success rate in the redshift range 0.6 < z < 1.2, with a mean redshift of 0.8 and 15% of redshifts unknown.
• The faint DES ELG sample achieved a 68% success rate in the same redshift window, with a mean redshift of 0.87 and 13% of redshifts unknown.
• The galaxy bias was measured as 1.72 ± 0.1 for the bright sample and 1.78 ± 0.12 for the faint sample, consistent with theoretical expectations for M_B - 5 log h < -21.0 galaxies.
• Stellar contamination was found to be less than 2% in both samples, indicating high purity of the ELG selection.
• Using a random forest algorithm (TPZ) and removing color branches with >10% outlier fraction reduced the outlier fraction to 10% while maintaining 71% completeness.
• The clustering analysis showed no significant difference between correlation functions computed using spectroscopic versus photometric redshifts, validating the use of photometric redshifts for large-scale structure studies.