Reanalysis of Stage-III cosmic shear surveys: A comprehensive study of shear diagnostic tests

TL;DR

This study systematically reanalyzed Stage-III shear catalogs from KiDS, DES, and HSC surveys using diagnostic tests to identify residual systematics, revealing specific failures and proposing strategies for future surveys.

Contribution

It provides a uniform reanalysis of shear catalogs across multiple surveys, identifying systematic test failures and offering recommendations for future weak lensing analyses.

Findings

KiDS-1000 fails the star ellipticity test

DES-Y3 and HSC-Y3 fail the B-mode test

Residual systematics have negligible impact on cosmological parameters

Abstract

In recent years, shear catalogs have been released by various Stage-III weak lensing surveys including the Kilo-Degree Survey, the Dark Energy Survey, and the Hyper Suprime-Cam Subaru Strategic Program. These shear catalogs have undergone rigorous validation tests to ensure that the residual shear systematic effects in the catalogs are subdominant relative to the statistical uncertainties, such that the resulting cosmological constraints are unbiased. While there exists a generic set of tests that are designed to probe certain systematic effects, the implementations differ slightly across the individual surveys, making it difficult to make direct comparisons. In this paper, we use the TXPipe package to conduct a series of predefined diagnostic tests across three public shear catalogs -- the 1,000 deg$^2$ KiDS-1000 shear catalog, the Year 3 DES-Y3 shear catalog, and the Year 3 HSC-Y3 shear catalog. We attempt to reproduce the published results when possible and perform key tests uniformly across the surveys. While all surveys pass most of the null tests in this study, we find two tests where some of the surveys fail. Namely, we find that when measuring the tangential ellipticity around bright and faint star samples, KiDS-1000 fails with a $\chi^2$/dof of 121.1/16 and 257.7/16 for bins 4 and 5 for faint, weighted stars. We also find that DES-Y3 and HSC-Y3 fail the $B$-mode test when estimated with the Hybrid-$E$/$B$ method, with a $\chi^2$/dof of 37.9/10 and 36.0/8 for the fourth and third autocorrelation bins. We assess the impacts on the $\Omega_{\rm m}$ - S$_{8}$ parameter space by comparing the posteriors of a simulated data vector with and without PSF contamination -- we find negligible effects in all cases. Finally, we propose strategies for performing these tests on future surveys such as the Vera C. Rubin Observatory's Legacy Survey of Space and Time.