Radio Weak Lensing Shear Measurement in the Visibility Domain - II. Source Extraction

TL;DR

This paper develops and tests a method for measuring galaxy ellipticities directly in the radio visibility domain, demonstrating its effectiveness for weak lensing surveys with SKA1-MID and analyzing bias characteristics.

Contribution

It introduces a detailed source extraction algorithm in the visibility domain and evaluates its performance for realistic galaxy populations in radio weak lensing.

Findings

Multiplicative bias is only 1.5 times worse than for individual sources at SNR >= 10.

Additive bias is larger than in optical surveys but depends on uv coverage shape.

A uv-plane weighting scheme could reduce additive bias.

Abstract

This paper extends the method introduced in Rivi et al. (2016b) to measure galaxy ellipticities in the visibility domain for radio weak lensing surveys. In that paper we focused on the development and testing of the method for the simple case of individual galaxies located at the phase centre, and proposed to extend it to the realistic case of many sources in the field of view by isolating visibilities of each source with a faceting technique. In this second paper we present a detailed algorithm for source extraction in the visibility domain and show its effectiveness as a function of the source number density by running simulations of SKA1-MID observations in the band 950-1150 MHz and comparing original and measured values of galaxies' ellipticities. Shear measurements from a realistic population of 10^4 galaxies randomly located in a field of view of 1 deg^2 (i.e. the source density expected for the current radio weak lensing survey proposal with SKA1) are also performed. At SNR >= 10, the multiplicative bias is only a factor 1.5 worse than what found when analysing individual sources, and is still comparable to the bias values reported for similar measurement methods at optical wavelengths. The additive bias is unchanged from the case of individual sources, but is significantly larger than typically found in optical surveys. This bias depends on the shape of the uv coverage and we suggest that a uv-plane weighting scheme to produce a more isotropic shape could reduce and control additive bias.