Validation of PSF Models for HST and Other Space-Based Observations

TL;DR

This paper presents an empirical validation framework for PSF models used in space-based observations, demonstrating how to improve Tiny Tim PSFs for HST by fitting Zernike coefficients to match observed star images.

Contribution

It introduces a validation method for PSF models and shows how fitting Zernike coefficients enhances Tiny Tim's accuracy for HST observations.

Findings

Tiny Tim models differ significantly from observed PSFs in size.

Fitting Zernike coefficients reduces model-observation differences to negligible levels.

Updated Zernike coefficients improve model accuracy while maintaining most of the initial improvements.

Abstract

Forthcoming space-based observations will require high-quality point-spread function (PSF) models for weak gravitational lensing measurements. One approach to generating these models is using a wavefront model based on the known telescope optics. We present an empirical framework for validating such models to confirm that they match the actual PSF to within requirements by comparing the models to the observed light distributions of isolated stars. We apply this framework to Tiny Tim, the standard tool for generating model PSFs for the Hubble Space Telescope (HST), testing its models against images taken by HST's Advanced Camera for Surveys in the Wide Field Channel. We show that Tiny Tim's models, in the default configuration, differ significantly from the observed PSFs, most notably in their sizes. We find that the quality of Tiny Tim PSFs can be improved through fitting the full set of Zernike polynomial coefficients which characterise the optics, to the point where the practical significance of the difference between model and observed PSFs is negligible for most use cases, resulting in additive and multiplicative biases both of order approximately 4e-4. We also show that most of this improvement can be retained through using an updated set of Zernike coefficients, which we provide.