Weak lensing shear calibration with simulations of the HSC survey

TL;DR

This study develops a extensive GalSim-based simulation framework to calibrate weak-lensing shear biases in the first-year Hyper Suprime-Cam (HSC) survey by embedding real galaxy morphologies from the COSMOS/HST dataset into HSC-like observing conditions. It demonstrates that including nearby galaxies and blends is crucial to reproducing observed size and magnitude distributions, and it carefully models multiplicative and additive biases as well as selection effects due to weights and cuts. Through a detailed analysis pipeline that mirrors the HSC data processing, the work achieves ~1% level control of multiplicative biases and characterizes additive biases and selection biases, validating the calibration approach under various perturbations to the simulated galaxy population. The results emphasize the importance of realistic blending and morphology in simulations and discuss paths forward, including metacalibration, for future improvements in shear calibration for larger, more precise surveys.

Abstract

We present results from a set of simulations designed to constrain the weak lensing shear calibration for the Hyper Suprime-Cam (HSC) survey. These simulations include HSC observing conditions and galaxy images from the Hubble Space Telescope (HST), with fully realistic galaxy morphologies and the impact of nearby galaxies included. We find that the inclusion of nearby galaxies in the images is critical to reproducing the observed distributions of galaxy sizes and magnitudes, due to the non-negligible fraction of unrecognized blends in ground-based data, even with the excellent typical seeing of the HSC survey (0.58" in the $i$-band). Using these simulations, we detect and remove the impact of selection biases due to the correlation of weights and the quantities used to define the sample (S/N and apparent size) with the lensing shear. We quantify and remove galaxy property-dependent multiplicative and additive shear biases that are intrinsic to our shear estimation method, including a $\sim 10$ per cent-level multiplicative bias due to the impact of nearby galaxies and unrecognized blends. Finally, we check the sensitivity of our shear calibration estimates to other cuts made on the simulated samples, and find that the changes in shear calibration are well within the requirements for HSC weak lensing analysis. Overall, the simulations suggest that the weak lensing multiplicative biases in the first-year HSC shear catalog are controlled at the 1 per cent level.

Figures (18)

• Figure 1: The noise correlation function as a function of distance, averaged over all the blank sky regions. The color scale is truncated to increase the dynamic range; the correlation is defined such that the pixel at $(0,0)$ has a value of 1.
• Figure 2: The logarithmic color-scale shows a 2D histogram of $\Delta$mag (the difference in observed magnitude in the worst-seeing coadds vs. the best-seeing coadds for objects detected in both) as a function of the average of the observed magnitudes in the worst- and best-seeing coadds. The five red lines show the $(5, 16, 50, 84, 95)$ percentile values of $\Delta$mag, while the dashed black line shows the ideal value of 0. The coherent skew towards negative values indicates that a subset of the galaxies in the worst-seeing coadd appear consistently brighter than the same galaxies in the best-seeing coadd, due to unrecognized blending effects.
• Figure 3: Example simulated images for simulations with the four parent samples as labeled on each plot. All images have the same zero points, and the symlog color scales are the same on each panel. For the purpose of illustration we have chosen images with the same PSF model. Dashed red lines show the artificial boundaries between individual postage stamps; we have shown a $5\times 5$ postage stamp region ($320\times 320$ pixels) out of the simulated image composed of $100\times 100$ postage stamps. Note that the color-scale is saturated in some places for sample 4 simulations, because some of the stamps include bright stars that happen to lie close enough to our central objects that they are included in the simulations. The sample 3 and sample 4 images look quite different from each other, despite the choice to not mask nearby objects in either sample, because sample 4 images were created from larger postage stamps (sufficiently large that they include some irrelevant structures that would never be blended with the central objects). We demonstrate later in this work that the measured properties of the central galaxies in the postage stamps and their shear calibration biases are very similar in samples 3 and 4 (Fig. \ref{['fig:parentsamp_dist']}).
• Figure 4: Distribution of the noise variance, for each field and for simulations vs. data (S16A), after reweighting the simulations to match the imaging conditions in the data. The top two rows are for each field individually, while the bottom left is for the average over all fields; all panels show reasonable consistency between simulations and data. Finally, the bottom middle panel shows the average over all fields, without reweighting to account for differences in imaging conditions.
• Figure 5: Distribution of the PSF FWHM for each field in HSC and for simulations vs. data (S16A), after reweighting the simulations to match the imaging conditions in the data. The top two rows are for each field individually, while the bottom left is for the average over all fields; all panels show reasonable consistency between simulations and data. Finally, the bottom middle panel shows the average over all fields, without reweighting to account for differences in imaging conditions. The weighted mean PSF FWHM values are given on the plot and indicated with vertical lines.