Redshift distributions of galaxies in the DES Science Verification shear catalogue and implications for weak lensing

TL;DR

This study evaluates photometric redshift estimates for DES SV weak-lensing galaxies using four independent methods (ANNz2, SkyNet, TPZ, BPZ) against a large spectroscopic training set, simulations, and COSMOS data. It demonstrates a global redshift distribution mean of $z\approx0.72$ and tomographic-bin means of $z={0.45,0.67,1.00}$, with mean biases $\delta z\lesssim0.05$ per bin when using SkyNet as a fiducial reference. Propagating these photo-z uncertainties into cosmic-shear statistics shows modest shifts in $\sigma_8$ (typically $\lesssim$ a few percent) and small biases in the critical surface density, which can be mitigated by a Gaussian prior on photo-z bias with width $0.05$ per tomographic bin. The results establish a practical framework for photo-z calibration in DES SV and outline strategies to reduce systematic redshift biases for future, larger weak-lensing surveys.

Abstract

We present photometric redshift estimates for galaxies used in the weak lensing analysis of the Dark Energy Survey Science Verification (DES SV) data. Four model- or machine learning-based photometric redshift methods -- ANNZ2, BPZ calibrated against BCC-Ufig simulations, SkyNet, and TPZ -- are analysed. For training, calibration, and testing of these methods, we construct a catalogue of spectroscopically confirmed galaxies matched against DES SV data. The performance of the methods is evaluated against the matched spectroscopic catalogue, focusing on metrics relevant for weak lensing analyses, with additional validation against COSMOS photo-zs. From the galaxies in the DES SV shear catalogue, which have mean redshift $0.72\pm0.01$ over the range $0.3<z<1.3$, we construct three tomographic bins with means of $z=\{0.45, 0.67, 1.00\}$. These bins each have systematic uncertainties $δz \lesssim 0.05$ in the mean of the fiducial SkyNet photo-z $n(z)$. We propagate the errors in the redshift distributions through to their impact on cosmological parameters estimated with cosmic shear, and find that they cause shifts in the value of $σ_8$ of approx. 3%. This shift is within the one sigma statistical errors on $σ_8$ for the DES SV shear catalog. We further study the potential impact of systematic differences on the critical surface density, $Σ_{\mathrm{crit}}$, finding levels of bias safely less than the statistical power of DES SV data. We recommend a final Gaussian prior for the photo-z bias in the mean of $n(z)$ of width $0.05$ for each of the three tomographic bins, and show that this is a sufficient bias model for the corresponding cosmology analysis.

Figures (20)

• Figure 1: $i$-band magnitude histograms for various levels of cuts from the full Gold catalogue down to the final shear catalogue.
• Figure 2: Location of the six spectral fields and the main DES SV (SPT-East) field on the sky. The SN fields are the DES supernova fields while the other two have been observed with DECam outside of the DES survey.
• Figure 3: The normalised redshift distributions of the spectroscopic samples used in producing and testing the photometric redshift estimates. The solid line is the Kernel Density Estimate (KDE) ML estimate of the underlying density. Top panel: The combined training and validation samples. Middle panel: The independent sample (VVDS-F14). Bottom panel: The VVDS-Deep sample.
• Figure 4: The $10 \sigma$mag_auto detection limits of the matched spectroscopic sample (blue) compared to that of the weak lensing sample (red). The matched spectroscopic catalogue has a significantly larger detection limit due to the fact that many DES galaxies with spectra lie in the frequently observed DES supernova fields.
• Figure 5: The $i$-band magnitude distribution of the matched spectroscopic catalogue in shown in blue and the weak lensing sample is shown in red. The matched spectroscopic catalogue after weighting is shown as the grey histogram outline overlaying the weak lensing sample.