The impact of spurious shear on cosmological parameter estimates from weak lensing observables

TL;DR

This study assesses how residual errors in shear measurements affect cosmological parameters from weak lensing, finding that biases vary across statistical methods and emphasizing the importance of correcting non-Gaussian noise effects.

Contribution

It quantifies biases from spurious shear on weak lensing statistics, highlighting differences among methods and the need for correction in large surveys.

Findings

Biases from power spectrum are smaller than previous estimates.

Morphological statistics are more biased than power spectrum and moments.

Non-Gaussian noise may significantly impact biases in large surveys.

Abstract

Residual errors in shear measurements, after corrections for instrument systematics and atmospheric effects, can impact cosmological parameters derived from weak lensing observations. Here we combine convergence maps from our suite of ray-tracing simulations with random realizations of spurious shear. This allows us to quantify the errors and biases of the triplet $(\Omega_m,w,\sigma_8)$ derived from the power spectrum (PS), as well as from three different sets of non-Gaussian statistics of the lensing convergence field: Minkowski functionals (MF), low--order moments (LM), and peak counts (PK). Our main results are: (i) We find an order of magnitude smaller biases from the PS than in previous work. (ii) The PS and LM yield biases much smaller than the morphological statistics (MF, PK). (iii) For strictly Gaussian spurious shear with integrated amplitude as low as its current estimate of $\sigma^2_{sys}\approx 10^{-7}$, biases from the PS and LM would be unimportant even for a survey with the statistical power of LSST. However, we find that for surveys larger than $\approx 100$ deg$^2$, non-Gaussianity in the noise (not included in our analysis) will likely be important and must be quantified to assess the biases. (iv) The morphological statistics (MF,PK) introduce important biases even for Gaussian noise, which must be corrected in large surveys. The biases are in different directions in $(\Omega_m,w,\sigma_8)$ parameter space, allowing self-calibration by combining multiple statistics. Our results warrant follow-up studies with more extensive lensing simulations and more accurate spurious shear estimates.