How Well Can We Measure Galaxy Shapes with the LSST?: PSF Modeling for 3.2 Giga Pixels
Abstract
The focal plane of the LSST will be tiled with 189 4Kx4K CCDs, whose heights will vary 10 microns (peak-to-valley) in a complicated way relative to the nominal flat surface. Although this flatness deviation is small compared with other cameras, the small f-ratio of the LSST optics makes this focal plane flatness variation play a critical role in aberration-induced PSF elongation. In this poster, we present our simulation of the LSST PSF using the LSST optics model, the current specifications for the CCD assembly/fabrication, and the Cerro-Pachon atmosphere model. As expected, we observe sharp discontinuities in PSF pattern across chip gaps, where abrupt changes in focus can happen. However, when we perform chip-by-chip PSF interpolation with basis functions derived from principal component analysis, these discontinuities do not prevent reconstruction of high-fidelity PSF models across the entire LSST focal plane. The resulting residuals in galaxy shear errors are well below what is required for precision weak lensing for the LSST survey. We support this claim through image simulations with real galaxy images sampled from the HST/UDF data. Based on this work and related Subaru observations we make a projection of LSST cosmic shear systematic error and signal-to-noise ratio. With hundreds of dithered and rotated images of each sky patch in each optical band, the LSST focal plane non-flatness and atmosphere effects will produce residual shear correlation errors far below the cosmic shear signal over all angular scales of interest.
- Publication:
-
American Astronomical Society Meeting Abstracts #213
- Pub Date:
- January 2009
- Bibcode:
- 2009AAS...21346026J