1709.00992
Dark Energy Survey Year 1 Results: Cross-Correlation Redshifts - Methods and systematics characterization
Gatti, et al
Use numerical simulations to characterize the performance of a clustering-based method to calibrate photo-z biases. In particular, X-correlate WL source galaxies from DES Y1 sample with redMaGiC galaxies (luminous red galaxies with secure photo-z) to estimate the z distribution of the former sample. The recovered redshift distributions are used to calibrate the photo-z bias of standard photo-z methods applied to the same source galaxy sample. Apply the method to 3 photo-z codes run in the simulated data: BPZ, Directional Neighborhood Fitting (DNF), and Random Forest-based photo-z (RF). Characterize the systematic uncertainties of the calibration procedure, and find that these systematic uncertainties dominate the error budget. The dominant systematics are due to the assumption of involving bias and clustering across each z bin, ad to differences between the shapes of the z distributions derived by clustering vs photo-z's. The systematic uncertainty in the mean redshift bias of the source galaxy sample is Delta z <~ 0.02, though the precise value depends on the z bin under consideration. Discuss possible ways to mitigate the impact of the dominant systematics in future analysis.
No comments:
Post a Comment