The effect of imperfect corrections of PSF anisotropy on cosmic shear measurements

Current measurements of the weak lensing signal induced by large scale structure provide useful constraints on a range of cosmological parameters. However, the ultimate succes of this technique depends on the accuracy with which one can correct for the effect of the Point Spread Function (PSF), in particular the correction for the PSF anisotropy. With upcoming large weak lensing surveys a proper understanding of residual systematics is necessary. In this paper we examine the accuracy of the PSF anisotropy correction using images of fields with a large number of stars. A randomly selected subset of stars is used to characterize the PSF and to correct the shapes of the remaining stars. The ellipticity correlation function of the residuals is studied to quantify the effect of imperfect corrections for PSF anisotropy on cosmic shear studies. These imperfections occur on the chip scale and consequently the systematic signal decreases rapidly with increasing angular scale. Separation of the signal into “E” (curl-free) and “B” (curl) components can help to identify the presence of residual systematics, but in general, the amplitude of the “B”-mode is different from that of the “E”-mode. The study of fields with many stars can be beneficial in finding a proper description of the variation of PSF anisotropy, and consequently help to significantly improve the accuracy with which the cosmic shear signal can be measured. We show that with such an approach it is feasible that the accuracy of future cosmic shear studies is limited by the statistical noise introduced by the intrinsic shapes of the sources. In particular, the prospects for accurate measurements of the cosmic shear signal on scales larger than ∼ 10 arcminutes are excellent.