7.4 Identification of Bad Pixels

  When we derived the best linear fit of each pixel's dark current signal to the q parameter (defined in §7.3 above), a byproduct that was easy to calculate was the dispersion of the measurements about the average trend. A vast majority of pixels had values clustered around a certain value, but some showed abnormally low or high values. We regard these pixels as defective and made an effort to catalog them so that their intensities would be disregarded in later analyses. Pixels that had a large dispersion were judged to be unstable or excessively noisy, while those with too low a dispersion were regarded as ``pegged'' at a certain value (usually 255 DN, because they are ``hot'' pixels giving a signal value that saturates at the top of the numerical range).

In addition to having some scattered, bad pixels, the CCD had two bad columns. These bad columns are best seen in Fig. 8. The total number of bad pixels and columns in the IMAPS CCD is about typical for the limited production runs of electron-bombardable CCD's that were available when the IMAPS camera was commissioned.