Mass Bias of Weak-lensing Shear-selected Galaxy Cluster Samples

We estimate the Eddington bias on weak-lensing mass measurements of shear-selected galaxy cluster samples. The mass bias is expected to be significant because constructions of cluster samples from peaks in weak-lensing mass maps and measurements of cluster masses from their tangential shear profiles share the same noise. We quantify this mass bias from large sets of mock cluster samples with analytical density profiles and realistic large-scale structure noise from ray-tracing simulations. We find that, even for peaks with signal-to-noise ratio larger than 4.0 in weak-lensing mass maps constructed in a deep survey with a high source galaxy number density of 30 arcmin(-2), derived weak-lensing masses for these shear-selected clusters are still biased high by similar to 55% on average. Such a large bias mainly originates from upscattered low-mass objects, which is an inevitable consequence of selecting clusters with a noisy observable directly linked to the mass measurement. We also investigate the dependence of the mass bias on different physical and observational parameters, finding that the mass bias strongly correlates with cluster redshifts, true halo masses, and selection signal-to-noise thresholds, but having moderate dependence on observed weak-lensing masses and survey depths. This bias, albeit considerable, can still be modeled accurately in statistical studies of shear-selected clusters, as the intrinsic scatter around the mean bias is found to be reasonable in size. We demonstrate that such a bias can explain the deviation in X-ray properties previously found on a shear-selected cluster sample. Our result will be useful for turning large samples of shear-selected clusters available in future surveys into potential probes of cosmology and cluster astrophysics.