Title: Constraining the masses of high-redshift clusters with weak lensing: Revised shape calibration testing for the impact of stronger shears and increased blending

Abstract: WL measurements have well-known shear estimation biases, which can be partially corrected for with the use of image simulations. We present an analysis of simulated images that mimic HST/ACS observations of high-redshift galaxy clusters, including cluster specific issues such as non-weak shear and increased blending. Our synthetic galaxies have been generated to match the observed properties of the background-selected samples in the real images. First, we used simulations with galaxies on a grid to determine a revised signal-to-noise-dependent correction for multiplicative shear measurement bias, and to quantify the sensitivity of our bias calibration to mismatches of galaxy or PSF properties between the real data and the simulations. We studied the impact of increased blending and light contamination from cluster and foreground galaxies, finding it negligible for $z>0.7$ clusters, whereas there is an effect at the $\sim 1\%$ level for lower redshift clusters. Finally, we studied the impact of fainter neighbours and selection bias mimicking the positions and magnitudes of galaxies in CANDELS data. The initial SExtractor object detection causes a selection bias of $-0.028 \pm 0.002$, reduced to $-0.016 \pm 0.002$ by further cuts. We compared our CANDELS-based estimate to a grid-based analysis, with added clustered galaxies reaching even fainter magnitudes, yielding a refined estimate of $\sim -0.013$. Our pipeline is calibrated to an accuracy of $\sim 0.015$, which is fully sufficient for current and near-future weak lensing studies of high-redshift clusters. As an application, we used it for a refined analysis of three highly relaxed clusters from the SPT-SZ survey, including measurements down to $r>200$ kpc. Compared to previously employed scales, this tightens the cluster mass constraints by a factor 1.38 on average.