Direct measurement of dark matter halo ellipticity from two-dimensional lensing shear maps of 25 massive clusters

We present new measurements of dark matter distributions in 25 X-ray luminous clusters by making a full use of the two-dimensional (2D) weak-lensing signals obtained from high-quality Subaru/Suprime-Cam imaging data. Our approach to directly compare the measured lensing shear pattern with elliptical model predictions allows us to extract new information on the mass distributions of individual clusters, such as the halo ellipticity and mass centroid. We find that these parameters on the cluster shape are little degenerate with cluster mass and concentration parameters. By combining the 2D fitting results for a subsample of 18 clusters, the elliptical shape of dark matter haloes is detected at 7 sigma significance level. The mean ellipticity is found to be <e > = <1 - b/a > = 0.46 +/- 0.04 (1 sigma), which is in excellent agreement with a theoretical prediction based on the standard collisionless cold dark matter model. The mass centroid can be constrained with a typical accuracy of similar to 20 arcsec (similar to 50 h(-1) kpc) in radius for each cluster. The mass centroid position fairly well matches the position of the brightest cluster galaxy, with some clusters showing significant offsets. Thus, the 2D shear fitting method enables us to assess one of the most important systematic errors inherent in the stacked cluster weak-lensing technique, the mass centroid uncertainty. In addition, the shape of the dark mass distribution is found to be only weakly correlated with that of the member galaxy distribution or the brightest cluster galaxy. We carefully examine possible sources of systematic errors in our measurements including the effect of substructures, the cosmic shear contamination, fitting regions and the dilution effect, and find none of them to be significant. Our results demonstrate the power of high-quality imaging data for exploring the detailed spatial distribution of dark matter, which should improve the ability of future surveys to conduct cluster cosmology experiments.