Search

The cosmological exploitation of modern photometric galaxy surveys requires both accurate (unbiased) and precise (narrow) redshift probability distributions derived from broadband photometry. Existing methodologies do not meet those requirements. Standard template fitting delivers interpretable models and errors, but lacks flexibility to learn inaccuracies in the observed photometry or the spectral templates. Machine learning addresses those issues, but requires representative training data, and the resulting models and uncertainties cannot be interpreted in the context of a physical model or outside of the training data. We present a hierarchical modeling approach simultaneously addressing the issues of flexibility, interpretability, and generalization. It combines template fitting with flexible (machine learning-like) models to correct the spectral templates, model their redshift distributions, and recalibrate the photometric observations. By optimizing the full posterior distribution of the model and solving for its (thousands of) parameters, one can perform a global statistical calibration of the data and the SED model. We apply this approach to the public Dark Energy Survey Science Verification data, and show that it provides more accurate and compact redshift posterior distributions than existing methods, as well as insights into residual photometric and SED systematics. The model is causal, makes predictions for future data (e.g., additional photometric bandpasses), and its internal parameters and components are interpretable. This approach does not formally require the training data to be complete or representative; in principle it can even work in regimes in which few or no spectroscopic redshifts are available., Comment: 24 pages, 11 figures

The Wide Field Infrared Survey Telescope (WFIRST) will conduct a galaxy redshift survey using the H$\alpha$ emission line primarily for spectroscopic redshift determination. Due to the modest spectroscopic resolution of the grism, the H$\alpha$ and the neighboring [NII] lines are blended, leading to a redshift bias that depends on the [NII]/H$\alpha$ ratio, which is correlated with a galaxy's metallicity, hence mass and ultimately environment. We investigate how this bias propagates into the galaxy clustering and cosmological parameters obtained from the WFIRST. Using simulation, we explore the effect of line blending on redshift-space distortion and baryon acoustic oscillation (BAO) measurements. We measure the BAO parameters $\alpha_{\parallel}$, $\alpha_{\perp}$, the logarithmic growth factor $f_{v}$, and calculate their errors based on the correlations between the line ratio and large-scale structure. We find $\Delta\alpha_{\parallel} = 0.31 \pm 0.23 \%$ ($0.26\pm0.17\%$), $\Delta\alpha_{\perp} = -0.10\pm0.10\%$ ($-0.12 \pm 0.11 \%$), and $\Delta f_{v} = 0.17\pm0.33\%$ ($-0.20 \pm 0.30\%$) for redshift 1.355--1.994 (0.700--1.345), which use approximately 18$\%$, 9$\%$, and 7$\%$ of the systematic error budget in a root-sum-square sense. These errors may already be tolerable but further mitigations are discussed. Biases due to the environment-independent redshift error can be mitigated by measuring the redshift error probability distribution function. High-spectral-resolution re-observation of a few thousand galaxies would be required (if by direct approach) to reduce them to below 25$\%$ of the error budget. Finally, we outline the next steps to improve the modeling of [NII]-induced blending biases and their interaction with other redshift error sources., Comment: 19 pages, 11 figures, minor changes, matched with MNRAS accepted version