Dark Energy Survey: Galaxy Sample for the Baryonic Acoustic Oscillation Measurement from the Final Dataset

Abstract

In this paper we present and validate the galaxy sample used for the analysis of the baryon acoustic oscillation (BAO) signal in the Dark Energy Survey (DES) Y6 data. The definition is based on a color and redshift-dependent magnitude cut optimized to select galaxies at redshifts higher than 0.6, while ensuring a high-quality photo-$z$ determination. The optimization is performed using a Fisher forecast algorithm, finding the optimal $i$-magnitude cut to be given by $i$<19.64+2.894$z_{\rm ph}$. For the optimal sample, we forecast an increase in precision in the BAO measurement of $\sim$25% with respect to the Y3 analysis. Our BAO sample has a total of 15,937,556 galaxies in the redshift range 0.6<$z_{\rm ph}$<1.2, and its angular mask covers 4,273.42 deg${}^2$ to a depth of $i$=22.5. We validate its redshift distributions with three different methods: directional neighborhood fitting algorithm (DNF), which is our primary photo-$z$ estimation; direct calibration with spectroscopic redshifts from VIPERS; and clustering redshift using SDSS galaxies. The fiducial redshift distribution is a combination of these three techniques performed by modifying the mean and width of the DNF distributions to match those of VIPERS and clustering redshift. In this paper we also describe the methodology used to mitigate the effect of observational systematics, which is analogous to the one used in the Y3 analysis. This paper is one of the two dedicated to the analysis of the BAO signal in DES Y6. In its companion paper, we present the angular diameter distance constraints obtained through the fitting to the BAO scale.

Figures (10)

• Figure 1: Footprint for the DES Y6 data. Each pixel is colored as a function of its depth in the $i$ band. The total area of the footprint considering the detection fraction of each pixel is 4,374.20 deg${}^2$.
• Figure 2: Area of the Y6 footprint mask as a function of the $i$-magnitude limit (blue solid line). The orange dashed line indicates the limit for the Y3 analysis ($i_{\rm max}=22.3$), whereas the green dashed line represents the same but for the Y6 ($i_{\rm max}=22.5$). We find that we barely lose any area after applying the $i=22.5$ cut (we still have 4,357.01 deg${}^2$ from the total of 4,374.20 deg${}^2$).
• Figure 3: Heat-map of $\sigma_{\rm BAO}$ obtained for samples selected with different values of $a$ and $b$ following \ref{['eq:i_selection_ab']}. The white star represents the sample with the lowest $\sigma_{\rm BAO}$, whereas the red points correspond to the next 20 samples with lower values for this variable. The optimal sample has $a=19.64$ and $b=2.894$ and a value of $\sigma_{\rm BAO}=0.0162$.
• Figure 4: Redshift distributions of the Y6 BAO analysis. In the case of DNF, we show both DNF_ZN and the stacking of DNF PDF (blue histograms and green lines, respectively). We also include the distributions of VIPERS Z_SPEC (orange points with error-bars), WZ (black points with error-bars) and the fiducial choice (red lines), which corresponds to the redshift distributions of DNF PDF but shifted and stretched with respect to WZ in the first 4 redshift bins and with respect to VIPERS Z_SPEC in the last 2, following the methodology described in \ref{['sec:shift_stretch_algorithm']}. The mean and width of all these redshift distributions (computed using the expressions given in \ref{['app:photoz']}) are displayed in \ref{['tab:z_average_w_68_all']} and plotted in \ref{['fig:mean_z_W_68_all']}.
• Figure 5: Top panel: average redshift of the different redshift distributions of the Y6 analysis. For visualization purposes, we subtracted the middle redshift, which is given by the average of the limits for each redshift bin (0.65, 0.75, 0.85, 0.95, 1.05 and 1.15, respectively). Bottom panel: width of the different redshift distributions of the Y6 analysis. Cases included in this plot: DNF_ZN (blue), DNF PDF (green), Z_SPEC (orange), WZ (black) and fiducial choice (red).