Point spread function modelling and shear calibration for weak lensing surveys

Abstract

Gravitational lensing is a cosmological effect which causes the path of a light ray, emitted from a distant source such as galaxy, to be deflected as it travels towards an observer. Deflections occur because the gravitational field, caused by large concentrations of matter, acts like a lens. While gravitational lensing, is in itself an interesting cosmological effect to study, it is often used as a cosmological probe to further the understanding of the Universe and its evolution. When the deflections are small, changes to observed background galaxy shapes are very subtle, this regime is known as weak gravitational lensing. The effects of weak gravitational lensing cannot be directly measured from individual galaxies but rather through a statistical process on a population of galaxies. The benefit of weak gravitational lensing is that it can be used along any line of sight and therefore, is able to better constrain matter distributions on large cosmological scales. There are a number of ongoing weak lensing surveys such as the Kilo Degree Imaging Survey (KiDS) and the Dark Energy Survey (DES), which are currently producing large amounts of data. Larger datasets allow for higher precision on any inferred cosmological parameters. A crucial point, and the general motivation behind the work presented in this thesis, is that the improvement in precision must also be matched with a greater understanding of the systematics present in the measurement process. A suite of realistic image simulations based on KiDS data are presented. The simulations were used to derive an accurate correction function that was capable of constraining the systematic effects in the measurement process to the level of ~ 1%, as well as quantifying a number of improvements which were made to the shape measurement algorithm, lensfit. The subsequent work presented in this thesis is based on the upcoming Euclid mission, a space telescope which will produce the largest and most accurate weak lensing analysis of its time. For Euclid, the measurement errors caused by the systematics effects must be drastically reduced. The work presented is focused on the simulation and improvement of the measurement of the Point Spread Function (PSF). Uncertainties on the measurement of the PSF may have a major effect on the weak lensing analysis. A novel PSF modelling algorithm is presented and was specifically designed for use on Euclid. The algorithm can be configured to work for different types of observations which are expected to be taken across the mission lifetime. This includes calibration observations, observations where the focus of the telescope is varied as well as an algorithm to be used for the wide survey science observations. Additionally, a pipeline which is capable of simulating realistic Euclid-like broadband PSFs is also presented. A series of tests and results showed that the algorithm, in its many configurations, was generally able to recover a PSF model that satisfies Euclid's strict accuracy requirements. The thesis concludes with a proposed strategy for measuring Euclid's PSF from a number of observation types across the lifetime of the mission.