Weak lensing mass modeling bias and the impact of miscentring

Abstract

Parametric modeling of galaxy cluster density profiles from weak lensing observations leads to a mass bias, whose detailed understanding is critical in deriving accurate mass-observable relations for constraining cosmological models. Drawing from existing methods, we develop a robust framework for calculating this mass bias in one-parameter fits to simulations of dark matter halos. We show that our approach has the advantage of being independent of the absolute noise level, so that only the number of halos in a given simulation and the representativeness of the simulated halos for real clusters limit the accuracy of the bias estimation. While we model the bias as a log-normal distribution and the halos with a Navarro-Frenk-White profile, our method can be generalized to any bias distribution and parametric model of the radial mass distribution. We find that the log-normal assumption is not strictly valid in the presence of miscentring of halos. We investigate the use of cluster centers derived from weak lensing in the context of mass bias, and tentatively find that such centroids can yield sensible mass estimates if the convergence peak has a signal-to-noise ratio approximately greater than four. In this context we also find that the standard approach to estimating the positional uncertainty of weak lensing mass peaks using bootstrapping severely underestimates the true positional uncertainty for peaks with low signal-to-noise ratios. Though we determine the mass and redshift dependence of the bias distribution for a few experimental setups, our focus remains providing a general approach to computing such distributions.

Mass calibration of distant SPT galaxy clusters through expanded weak-lensing follow-up observations with HST, VLT, & Gemini-South

“…We list the estimates for b∆,WL and the scatter σ(ln b ∆,WL ) including their statistical uncertainties for the different clusters incorporating miscentring in Tables 7, 8 and 9. For clusters already studied in S18, slight to moderate shifts can occur in the reported bias values for two reasons: First, we now account for a mass dependence of the bias (see also Sommer et al 2021). Second, our modifications in the source selection (especially for the clusters with new VLT data) changes the relative contributions of scales at different radii, thereby affecting the mass modelling bias.…”

“…Here we follow their conservative assumption, not only because of uncertainties in the assumed miscentring distributions, but also because our simulation analysis suggests that the assumption of a log-normal scatter is not strictly met when miscentring is included (see Sommer et al 2021). This constitutes the largest contribution to our systematic error budget for the analysis using SZ centres (see Sect.…”