Lensing Without Borders. I. A Blind Comparison of the Amplitude of Galaxy-Galaxy Lensing Between Independent Imaging Surveys

TL;DR

This study performs a blind comparison of galaxy-galaxy lensing signals across multiple surveys, finding good overall agreement but identifying potential systematic effects at higher redshifts, and demonstrating the value of cross-survey validation.

Contribution

It introduces a comprehensive cross-survey comparison method for galaxy-galaxy lensing, highlighting systematic uncertainties and their impact on lensing amplitude measurements.

Findings

Good agreement within 2.3σ of systematic errors across surveys

Detection of a 3-4σ correlation between lensing amplitude and survey depth at high redshift

Systematic errors below 15-25% in most lens bins, insufficient to explain the lensing discrepancy

Abstract

Lensing Without Borders is a cross-survey collaboration created to assess the consistency of galaxy-galaxy lensing signals ($\Delta\Sigma$) across different data-sets and to carry out end-to-end tests of systematic errors. We perform a blind comparison of the amplitude of $\Delta\Sigma$ using lens samples from BOSS and six independent lensing surveys. We find good agreement between empirically estimated and reported systematic errors which agree to better than 2.3$\sigma$ in four lens bins and three radial ranges. For lenses with $z_{\rm L}>0.43$ and considering statistical errors, we detect a 3-4$\sigma$ correlation between lensing amplitude and survey depth. This correlation could arise from the increasing impact at higher redshift of unrecognised galaxy blends on shear calibration and imperfections in photometric redshift calibration. At $z_{\rm L}>0.54$ amplitudes may additionally correlate with foreground stellar density. The amplitude of these trends is within survey-defined systematic error budgets which are designed to include known shear and redshift calibration uncertainty. Using a fully empirical and conservative method, we do not find evidence for large unknown systematics. Systematic errors greater than 15% (25%) ruled out in three lens bins at 68% (95%) confidence at $z<0.54$. Differences with respect to predictions based on clustering are observed to be at the 20-30% level. Our results therefore suggest that lensing systematics alone are unlikely to fully explain the "lensing is low" effect at $z<0.54$. This analysis demonstrates the power of cross-survey comparisons and provides a promising path for identifying and reducing systematics in future lensing analyses.