Reconstruction of weak lensing mass maps for non-Gaussian studies in the celestial sphere

TL;DR

This paper introduces an improved spherical weak lensing mass reconstruction method that enhances accuracy and reduces systematic errors, enabling better non-Gaussian analyses of cosmic structures in large sky surveys.

Contribution

The paper presents the SKS+ method, an advanced Kaiser-Squires approach on the sphere that outperforms previous flat-sky and spherical methods in accuracy and bias reduction.

Findings

SKS+ reduces correlation errors by 10 times.

Peak signal-to-noise ratio improves by 20-30%.

Third- and fourth-order moments errors decrease by factors of 2 and 4.

Abstract

We present a novel method for reconstructing weak lensing mass or convergence maps as a probe to study non-Gaussianities in the cosmic density field. While previous surveys have relied on a flat-sky approximation, the forthcoming stage IV survey will cover such large areas with a large field of view (FOV) to motivate mass reconstruction on the sphere. Here, we present an improved Kaiser-Squires (KS+) mass inversion method using a HEALPix pixelisation of the sphere while controlling systematic effects. As in the KS+ methodology, the convergence maps were reconstructed without noise regularisation to preserve the Information content and allow for non-Gaussain studies. The results of this new method were compared with those of the Kaiser-Squires (KS) estimator implemented on the curved sky using high-resolution realistic N-body simulations. The quality of the method was evaluated by estimating the two-point correlation functions, third- and fourth-order moments, and peak counts of the reconstructed convergence maps. The effects of masking, sampling, and noise were tested. We also examined the systematic errors introduced by the flat-sky approximation. We show that the improved Kaiser-Squires on the sphere (SKS+) method systematically improves inferred correlation errors by 10 times and provide on average a 20-30 % better maximum signal-to-noise peak estimation compared to Kaiser-Squires on the sphere (SKS). We also show that the SKS+ method is nearly unbiased and reduces errors by a factor of about 2 and 4 in the third and fourth-order moments, respectively. Finally, we show how the reconstruction of the convergence field directly on the celestial sphere eliminates the projection effects and allows the exclusion or consideration of a specific region of the sphere in the processing.