http://arxiv.org/abs/2301.03274
The brighter-fatter effect affects all CCD sensors to various degrees. Deep-depleted thick sensors are seriously affected and the measurement of galaxy shapes for cosmic shear measurements requires an accurate correction of the effect in science images. We describe the whole correction chain we have implemented for the CCDs of the Hyper Suprime-Cam imager on the Subaru Telescope. We derive non linearity corrections from a new sequence of flat field images, and measure their statistics, namely their two-pixel function. We constrain an electrostatic model from flat field statistics that we use to correct science images. We find evidence that some fraction of the observed variance and some covariances is not due to the combination of Poisson statistics and electrostatics — and the cause remains elusive. We then have to ignore some measurements when deriving the electrostatic model. Over a wide range of image qualities and in the 5 bands of the imager, stars in corrected science images exhibit size variations with flux small enough to predict the point spread function for faint objects to an accuracy better than $10^{-3}$ for the trace of second moments — and even better for the ellipticity and the fourth radial moment. This performance is sufficient for upcoming large-scale cosmic shear surveys such as Rubin/LSST.
P. Astier and N. Regnault
Tue, 10 Jan 23
50/93
Comments: 18 pages, 24 figures, accepted for publication in A&A