Redshift distribution uncertainty in weaklensing cosmology

Abstract

Cosmological information from weak lensing surveys is maximised by dividing source galaxies into tomographic sub-samples for which the redshift distributions are estimated. Uncertainties on these redshift distributions must be correctly propagated into the cosmological results to fully account for statistical and systematic errors on the estimations of these distributions. In this thesis I present a new method for marginalising over redshift distributions in gravitational weak lensing and clustering cosmological analyses, called Hyperrank, which allows discrete samples from the space of possible redshift distributions to be used, meaning the full uncertainty can be explored. In Hyperrank the set of proposed redshift distributions is ranked according to a small (~1) number of summary values, which are then sampled as hyper-parameters along with other nuisance parameters and cosmological parameters in the Monte Carlo chain used for inference. This work focuses on the case of weak lensing cosmic shear analyses and demonstrate our method using simulations made for the Dark Energy Survey. Hyperrank is compared to the common mean-shifting method of marginalising over redshift uncertainty, its numerical performance assessed and the resulting confidence contours used to validate its use in the DES Year 3 cosmology results. I also introduce the process to estimate and calibrate the distribution of source galaxy redshifts for the Dark Energy Survey Y3 analysis, including details of the SOMPZ scheme, a machine-learning algorithm to leverage deep field photometry to constrain the color-redshift relation of the wide survey galaxies. We describe the process to estimate the uncertainty associated to the different systematic effects involved in the estimation of the source redshift distributions, with an emphasis on the effect of sample variance and the stochastic nature of the SOMPZ training phase.

Date of Award	1 Aug 2021
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Ian Harrison (Supervisor) & Sarah Bridle (Supervisor)