摘要

We assess the relative merits of weak-lensing surveys, using overlapping imaging data from the ground-based Subaru telescope and the Hubble Space Telescope (HST). Our tests complement similar studies undertaken with simulated data. From observations of 230,000 matched objects in the 2 deg(2) COSMOS field, we identify the limit at which faint galaxy shapes can be reliably measured from the ground using well-established shape-measurement techniques. Our ground-based shear catalog achieves subpercent calibration bias compared to high-resolution space-based data for galaxies brighter than i' similar or equal to 24.5 and with half-light radii larger than 1.8 ''. This selection corresponds to a surface density of 15 galaxies arcmin(-2) compared to similar to 71 arcmin(-2) from space. On the other hand, the survey speed of current ground-based facilities is much faster than that of HST, although this gain is mitigated by the increased depth of space-based imaging desirable for tomographic (3D) analyses. As an independent experiment, we also reconstruct the projected mass distribution in the COSMOS field using both data sets, and compare the derived cluster catalogs with those from X-ray observations. The ground-based catalog achieves a reasonable degree of completeness, with minimal contamination and no detected bias, formassive clusters at redshifts 0.2 < z < 0.5. The space-based data provide improved precision and a greater sensitivity to clusters of lower mass or at higher redshift.