Updating the WFC3/UVIS CTE model and Mitigation Strategies

The pixel-based charge transfer efficiency (CTE) correction was last updated for WFC3/UVIS in 2016. Since the strength of CTE generally increases linearly with time, as does the population of warm pixels, the effect is almost twice as strong after five additional years of HST operations, and there are twice as many warm pixels available to use in deriving the correction. Unfortunately, the new model confirms that charge-transfer losses have continued to increase steadily since WFC3/UVIS was installed. Now, even with a background of 20 electrons, CTE losses are almost 50% for sources near the chip-gap (i.e., sources that undergo the maximum number of parallel transfers), compared to about 30% in 2016. The pixel-based algorithm works well only when the correction is small enough to be considered a perturbation on the signal received. If the correction is too large, then the algorithm tends to amplify noise, which causes more harm to the image than benefit. For this reason, the default pipeline setting is now designed to suppress the correction for faint sources in order to avoid noise amplification. In general, for observers with relatively bright targets (S/N > ~30) on image backgrounds of at least 20 e- per pixel, the new pixel-based CTE correction works well, correcting targets to within 5%. Observers with fainter targets, which the new algorithm treats minimally in order to avoid noise amplification, will need to make additional adjustments. We provide advice to users about how to plan observations and how to carry out reductions in order to get the most out of WFC3/UVIS observations. A future ISR will provide more detailed prescriptions on how to correct sources bright and faint.