Speaker
Description
Weak lensing science relies crucially on high fidelity PSF estimation (using unresolved field stars) in order to estimate underlying shears in large numbers of faint background galaxies. The response of pixel boundary distortions induced (during image integration) - by the building accumulation of conversions at the channel - is referred to as the brighter-fatter effect (BFE) and its mitigation is a key step of instrument signature removal (ISR).
Recent findings (Broughton et al.) show that even with large volumes of high quality flat pair data to characterize the BFE, the implemented pixel signal redistribution prescription appears to correct only ~90% of the effect, leaving an undesirable, uncorrected shape systematic in spot images. These reveal the rows/columns orientation of the sensor. This is a generic result that spans at least two sensor vendors.
In this presentation we describe a path forward for correcting the evasive, systematic structure that is the residual 10%. We discuss the collective influence of two flavors of confinement barriers present near the channel and explain why their presence, in real sensors, breaks down the curl-free symmetry assumption centrally built into the Coulton et al. (2018) algorithm. This algorithm is employed to invert pixel signal correlation maps (computed with flat pairs) into electrostatic pixel potential solutions which are then used in the ISR signal displacement correction steps. We suggest that any sensors exhibiting anisotropic pixel correlation patterns over 2D lag space share these properties. We provide placeholder lookup tables to reintroduce nonzero curl terms in the influence function, eliminated from the correction by definition, when these unmodified algorithms are utilized.
Keywords for your contribution subject matter (this will assist SOC in accurately characterizing your contribution)
brighter-fatter effect mitigation; PSF retrieval; electrostatic drift modeling
| contribution subject matter | point spread function fidelity |
|---|
