Resumen de Higher resolution photometric redshifts for cosmological surveys
This PhD thesis is focused on the measurement of photometric redshifts in imaging galaxy surveys and its applications to extract cosmological information. In the first part of this thesis we forecast a galaxy survey with very precise redshift information, which can come either from spectroscopy or many narrow band images, using the Fisher matrix formalism. We use galaxy clustering, how galaxies group together in space, dividing a sample into two subsamples using other observable parameters. Using two overlapping subsamples reduces the sample variance in the observables, which improves the precision with which one can measure the expansion and growth history of the universe. In the second part of this thesis we measure highly precise photometric redshifts using the data from a novel imaging galaxy survey PAUS that contains a unique set of 40 narrow band filters. We develop two algorithms which use maximum likelihood or Bayesian evidence statistics to infer the redshift probability of each galaxy, which requires modeling both the continuum and emission line galaxy flux. The algorithm contains several corrections to account for systematic effects present in the data calibration which are tested in simulations developed for this purpose. The measurement of PAUS redshifts enables the science of the galaxy survey and can also be used to calibrate the redshift distribution of lensing surveys. The last part of this thesis implements for the first time a hierarchical Bayesian model in an N-body simulation to measure the redshift distribution of a lensing survey using both photometric and density information. Weak lensing is a very powerful tool to extract cosmological information, but it is very sensitive to any bias in the mean redshift of a sample of source galaxies. This method consistently combines all sources of information and merges the main techniques used in the literature to estimate redshift distributions.
