Weak gravitational lensing: reducing the contamination by intrinsic alignments

Abstract

Intrinsic alignments of galaxies can mimic to an extent the effects of shear caused by weak gravitational lensing. Previous studies have shown that for shallow surveys with median redshifts z_m = 0.1, the intrinsic alignment dominates the lensing signal. For deep surveys with z_m = 1, intrinsic alignments are believed to be a significant contaminant of the lensing signal, preventing high-precision measurements of the matter power spectrum. In this paper we show how distance information, either spectroscopic or photometric redshifts, can be used to down-weight nearby pairs in an optimised way, to reduce the errors in the shear signal arising from intrinsic alignments. Provided a conservatively large intrinsic alignment is assumed, the optimised weights will essentially remove all traces of contamination. For the Sloan spectroscopic galaxy sample, residual shot noise continues to render it unsuitable for weak lensing studies. However, a dramatic improvement for the slightly deeper Sloan photometric survey is found, whereby the intrinsic contribution, at angular scales greater than 1 arcminute, is reduced from about 80 times the lensing signal to a 10% effect. For deeper surveys such as the COMBO-17 survey with z_m = 0.6, the optimisation reduces the error from a largely systematic 220% error at small angular scales to a much smaller and largely statistical error of only 17% of the expected lensing signal. We therefore propose that future weak lensing surveys be accompanied by the acquisition of photometric redshifts, in order to remove fully the unknown intrinsic alignment errors from weak lensing detections.