As described in arXiv:1710.00885, the HSC weak lensing group produced simulations to test the calibration of the weak lensing shear signals. The galaxy samples used for the simulations were defined using the HST COSMOS survey, which provides higher-resolution imaging that can be degraded to match the seeing in the HSC survey.

Section 3.3 of that paper describes a number of ways to make the simulations, and the parent galaxy sample used for the majority of the results in that paper is described in detail in section 3.3.4 (called `parent sample 4′ throughout the paper). As shown in plots and figures throughout the paper, this sample enables us to reproduce quite well the observed distribution of galaxy properties in the HSC survey, and includes nearby or neighboring structures around the central galaxy that is to be simulated. It can be used to simulate galaxies down to i~24.5; beyond that depth, the sample starts to be somewhat incomplete.

The data products provided here include all of the HST COSMOS images needed to make simulations of galaxies (in any survey with larger PSF than HSC) using parent sample 4. In particular, as described there, parent sample 4 actually consists of three sets of galaxies selected based on observations with PSF FWHM of 0.5, 0.7, and 1 arcsec, respectively. Hence the download consists of three tar files: parent_best_processed.tar.gz, parent_median_processed.tar.gz, and parent_worst_processed.tar.gz (respectively).

The tar files are designed such that they can be unzipped and then used directly with the open-source image simulation software GalSim to make simulations using the galsim.RealGalaxyCatalog() class (see online documentation here) and the galsim.RealGalaxy() class (see online documentation here). GalSim comes with extensive documentation and a series of demos; demo6 in python or yaml config script demonstrate how to use pre-packaged catalogs to simulate real galaxies using these two classes. Users who have unzipped the tar file for the ‘worst’ sample into the directory ./parent_worst_processed/ can then use the catalog with the following lines of python code:

import galsim
rgc = galsim.RealGalaxyCatalog(‘real_galaxy_catalog_worst.fits’, dir=’./parent_worst_processed’)

It is important to specify the catalog name and directory where the catalog is located, as shown in this example. After that point, RealGalaxy objects can be created from the `rgc’ catalog exactly as in the GalSim demos and documentation linked above.

Each of the three tar files for download contains the following:

1. The catalog itself, real_galaxy_catalog_X.fits where X=best, median, worst. The FITS catalog contains important reference information about the objects, such as their original RA/dec in the COSMOS field, the F814W magnitude of the nearest matching object in the COSMOS field, the file name and HDU containing the galaxy image, etc. Users do not tend to need to access any of this information directly; rather, it is used by GalSim when making the simulations.
2. A series of multi-HDU FITS files containing the galaxy and COSMOS PSF images, with names like real_galaxy_images_best_X_n#.fits and real_galaxy_PSF_images_best_X_n#.fits, where # is a number.
3. A file acs_I_unrot_sci_20_cf.fits that contains a representation of the COSMOS noise correlation function, which is also used internally by GalSim when making the simulations.

Use of these files should be acknowledged with a reference to the paper describing the samples and to Rowe et al. (2015) (for GalSim).