Problems using ratios of galaxy shape moments in requirements for weak lensing surveys
1 Institute for Computational Cosmology, Durham University, South Road, Durham DH1 3LE, UK
2 Centre for Extragalactic Astronomy, Durham University, South Road, Durham DH1 3LE, UK
3 Universitätssternwarte München, Fakultät für Physik, Ludwig-Maximilians-Universität München, Scheinerstraße 1, 81679 München, Germany
4 Mullard Space Science Laboratory, University College London, Holmbury St Mary, Dorking, Surrey RH5 6NT, UK
Received: 26 February 2016
Accepted: 22 September 2016
Context. The shapes of galaxies are typically quantified by ratios of their quadrupole moments. Knowledge of these ratios (i.e. their measured standard deviation) is commonly used to assess the efficiency of weak gravitational lensing surveys. For faint galaxies, observational noise can make the denominator close to zero, so the ratios become ill-defined.
Aims. Since the requirements cannot be formally tested for faint galaxies, we explore two complementary mitigation strategies. In many weak lensing contexts, the most problematic sources can be removed by a cut in measured size. This first technique is applied frequently. As our second strategy, we propose requirements directly on the quadrupole moments rather than their ratio.
Methods. As an example of the first strategy, we have investigated how a size cut affects the required precision of the charge transfer inefficiency model for two shape measurement settings. For the second strategy, we analysed the joint likelihood distribution of the image quadrupole moments measured from simulated galaxies, and propagate their (correlated) uncertainties into ellipticities.
Results. Using a size cut, we find slightly wider tolerance margins for the charge transfer inefficiency parameters compared to the full size distribution. However, subtle biases in the data analysis chain may be introduced. These can be avoided using the second strategy. To optimally exploit a Stage-IV dark energy survey, we find that the mean and standard deviation of a population of galaxies’ quadrupole moments must to be known to better than 1.4 × 10-3 arcsec2, or the Stokes parameters to 1.9 × 10-3 arcsec2.
Conclusions. Cuts in measured size remove sources that otherwise make ellipticity statistics of weak lensing galaxy samples diverge. However, size cuts bias the source population non-trivially. Assessing weak lensing data quality directly on the quadrupole moments instead mitigates the need for size cuts. Such testable requirements can form the basis for performance validation of future instruments.
Key words: methods: data analysis / techniques: image processing / gravitational lensing: weak / large-scale structure of Universe
© ESO, 2017