The shear testing programme 2: Factors affecting high-precision weak-lensing analyses

Richard Massey, Catherine Heymans, Joel Bergé, Gary Bernstein, Sarah Bridle, Douglas Clowe, Håkon Dahle, Richard Ellis, Thomas Erben, Marco Hetterscheidt, F. William High, Christopher Hirata, Henk Hoekstra, Patrick Hudelot, Mike Jarvis, David Johnston, Konrad Kuijken, Vera Margoniner, Rachel Mandelbaum, Yannick MellierReiko Nakajima, Stephane Paulin-Henriksson, Molly Peeples, Chris Roat, Alexandre Refregier, Jason Rhodes, Tim Schrabback, Mischa Schirmer, Uroš Seljak, Elisabetta Semboloni, Ludovic Van Waerbeke

    Research output: Contribution to journalArticlepeer-review


    The Shear Testing Programme (STEP) is a collaborative project to improve the accuracy and reliability of weak-lensing measurement, in preparation for the next generation of wide-field surveys. We review 16 current and emerging shear-measurement methods in a common language, and assess their performance by running them (blindly) on simulated images that contain a known shear signal. We determine the common features of algorithms that most successfully recover the input parameters. A desirable goal would be the combination of their best elements into one ultimate shear-measurement method. In this analysis, we achieve previously unattained discriminatory precision via a combination of more extensive simulations and pairs of galaxy images that have been rotated with respect to each other. That removes the otherwise overwhelming noise from their intrinsic ellipticities. Finally, the robustness of our simulation approach is confirmed by testing the relative calibration of methods on real data. Weak-lensing measurements have improved since the first STEP paper. Several methods now consistently achieve better than 2 per cent precision, and are still being developed. However, we can now distinguish all methods from perfect performance. Our main concern continues to be the potential for a multiplicative shear calibration bias: not least because this cannot be internally calibrated with real data. We determine which galaxy populations are responsible for bias and, by adjusting the simulated observing conditions, we also investigate the effects of instrumental and atmospheric parameters. The simulated point spread functions are not allowed to vary spatially, to avoid additional confusion from interpolation errors. We have isolated several previously unrecognized aspects of galaxy shape measurement, in which focused development could provide further progress towards the sub-per cent level of precision desired for future surveys. These areas include the suitable treatment of image pixellization and galaxy morphology evolution. Ignoring the former effect affects the measurement of shear in different directions, leading to an overall underestimation of shear and hence the amplitude of the matter power spectrum. Ignoring the second effect could affect the calibration of shear estimators as a function of galaxy redshift, and the evolution of the lensing signal, which will be vital to measure parameters including the dark energy equation of state. © 2007 RAS.
    Original languageEnglish
    Pages (from-to)13-38
    Number of pages25
    JournalMonthly Notices of the Royal Astronomical Society
    Issue number1
    Publication statusPublished - Mar 2007


    • Cosmology: observations
    • Gravitational lensing
    • Methods: data analysis


    Dive into the research topics of 'The shear testing programme 2: Factors affecting high-precision weak-lensing analyses'. Together they form a unique fingerprint.

    Cite this