Using galaxy two-point correlation functions to determine the redshift distributions of galaxies binned by photometric redshift

We investigate how well the redshift distributions of galaxies sorted into photometric redshift bins can be determined from the galaxy angular two-point correlation functions. We find that the uncertainty in the reconstructed redshift distributions depends critically on the number of parameters used in each redshift bin and the range of angular scales used, but not on the number of photometric redshift bins. Using six parameters for each photometric redshift bin, and restricting ourselves to angular scales over which the galaxy number counts are normally distributed, we find that errors in the reconstructed redshift distributions are large; i.e., they would be the dominant source of uncertainty in cosmological parameters estimated from otherwise ideal weak lensing or baryon acoustic oscillation data. However, either by reducing the number of free parameters in each redshift bin or by (unjustifiably) applying our Gaussian analysis into the non-Gaussian regime, we find that the correlation functions can be used to reconstruct the redshift distributions with moderate precision; e.g., with mean redshifts determined to similar to 0.01. We also find that dividing the galaxies into two spectral types, thereby doubling the number of redshift distribution parameters, can result in a reduction in the errors in the combined redshift distributions.