Weak gravitational lensing shear estimation with metacalibration for the Roman High-Latitude Imaging Survey

Abstract

ABSTRACT We investigate the performance of the metacalibration shear calibration framework using simulated imaging data for the Nancy Grace Roman Space Telescope (Roman) reference High-Latitude Imaging Survey (HLIS). The weak lensing programme of Roman requires the mean weak lensing shear estimate to be calibrated within about 0.03 per cent. To reach this goal, we can test our calibration process with various simulations and ultimately isolate the sources of residual shear biases in order to improve our methods. In this work, we build on the HLIS image simulation pipeline to incorporate several more realistic processing-pipeline updates. We show the first metacalibration results for 6 deg2 of the simulated reference HLIS and compare them to measurements on simpler, faster Roman-like image simulations. We neglect the impact of blending of objects. We find in the simplified simulations metacalibration can calibrate shapes to within m = (−0.01 ± 0.10) per cent. When applied to the current most-realistic version of the simulations, the precision is much lower, with estimates of m = (−0.76 ± 0.43) per cent for joint multiband multi-epoch measurements and m = (−1.13 ± 0.60) per cent for multiband coadd measurements. These results are all consistent with zero within 1–2σ, indicating we are currently limited by our simulated survey volume. Further work on testing the shear calibration methodology is necessary at the precision of the Roman requirements, in particular in the presence of blending. Current results demonstrate, however, that metacalibration can work on undersampled space-based Roman imaging data at levels comparable to requirements of current weak lensing surveys.