The impact of high spatial frequency atmospheric distortions on weak lensing measurements

TL;DR

This study quantifies high spatial frequency atmospheric distortions in ground-based imaging, revealing their impact on weak lensing measurements and suggesting mitigation strategies for future surveys.

Contribution

It provides empirical analysis of atmospheric turbulence effects on PSF variations and proposes methods to mitigate their impact on weak lensing data.

Findings

High spatial frequency turbulence decreases with exposure time as t^{-1/2}.

Atmospheric contribution to PSF ellipticity is negligible for exposures longer than 180 seconds.

Correlating galaxy shear on exposures separated by over 50 seconds reduces turbulence effects.

Abstract

High precision cosmology with weak gravitational lensing requires a precise measure of the Point Spread Function across the imaging data where the accuracy to which high spatial frequency variation can be modelled is limited by the stellar number density across the field. We analyse dense stellar fields imaged at the Canada-France-Hawaii Telescope to quantify the degree of high spatial frequency variation in ground-based imaging Point Spread Functions and compare our results to models of atmospheric turbulence. The data shows an anisotropic turbulence pattern with an orientation independent of the wind direction and wind speed. We find the amplitude of the high spatial frequencies to decrease with increasing exposure time as $t^{-1/2}$, and find a negligibly small atmospheric contribution to the Point Spread Function ellipticity variation for exposure times $t>180$ seconds. For future surveys analysing shorter exposure data, this anisotropic turbulence will need to be taken into account as the amplitude of the correlated atmospheric distortions becomes comparable to a cosmological lensing signal on scales less than $\sim 10$ arcminutes. This effect could be mitigated, however, by correlating galaxy shear measured on exposures imaged with a time separation greater than 50 seconds, for which we find the spatial turbulence patterns to be uncorrelated.