An improved model of charge transfer inefficiency and correction algorithm for the Hubble Space Telescope

Abstract

Charge-Coupled Device (CCD) detectors, widely used to obtain digital imaging, can be damaged by high energy radiation. Degraded images appear blurred, because of an effect known as Charge Transfer Inefficiency (CTI), which trails bright objects as the image is read out. It is often possible to correct most of the trailing during post-processing, by moving flux back to where it belongs. We compare several popular algorithms for this: quantifying the effect of their physical assumptions and tradeoffs between speed and accuracy. We combine their best elements to construct a more accurate model of damaged CCDs in the Hubble Space Telescope's Advanced Camera for Surveys/Wide Field Channel, and update it using data up to early 2013. Our algorithm now corrects 98% of CTI trailing in science exposures, a substantial improvement over previous work. Further progress will be fundamentally limited by the presence of read noise. Read noise is added after charge transfer so does not get trailed -but it is incorrectly untrailed during post-processing.

On scale-dependent cosmic shear systematic effects

“…We show this in Fig. 4 where we do the same test as in Massey et al (2014), but also supplement this with realizations that randomly sample from a Gaussian with a width chosen between σ ( ) = [0, 500] and a mean chosen between [100, 5000]. We find that indeed with a truncated functional expansion an integrated limit of Fig. shows the impact that different functional forms for the systematic power spectrum have on the ratio of bias/error for the dark energy parameter w 0 .…”

“…This is because in previous studies functional forms were chosen by randomly sampling from truncated functional expansions. For example 200 bins in were used in Massey et al (2014), and this is very unlikely to create an isolated spike in the reproduced power spectrum (a less than 1 : 10 100 chance of producing a single -mode peak at a height of 90 per cent of the maximum). We show this in Fig.…”

“…The green dots show the upper limit of the biases caused by realizations of a binned functional form in -mode (repeating the analysis of Massey et al 2013). The survey parameters are described in Section 2.3. σ 2 sys ≤ 10 −7 does ensure that dark energy biases are below bias/ error = 1 (or the revised requirement of 0.31 used in Massey et al 2014), but that a more thorough functional search, including power spectra with spike-like functional features can produce biases that are an order of magnitude larger.…”

“…Therefore, the expected residuals should be limited to a CCD chip scale and in the direction of the readout registers. To model the charge trailing due to CTI (see Massey et al 2014), we use a model where a maximum residual ellipticity, or size, after correction is assigned at the centre of a chip and linearly decreases to zero towards the chip edge. The amplitude of the maximum ellipticity, and the slope of the linear function can be changed on a chip-by-chip basis or over a field of view.…”

Challenges in photon-starved space astronomy in a harsh radiation environment using CCDs

“…In order to correct against the remaining charge loss and trailing caused by radiation damage in-orbit, it is possible to use post-processing, effectively "moving flux back to where it belongs" [27]. There has been extensive research into such corrections for the Hubble Space Telescope (HST), with the most recent improvements detailed in [27] and earlier work referenced there within.…”