7.1 General Considerations

  All of the data reduction was performed using programs written in IDL. Routines that did specialized manipulations of image data are those in the IDL library (version 3.0 for Windows); they will be explicitly identified in footnotes. All images were expressed in a format that oversampled the CCD pixels by a factor of two in both the vertical and horizontal directions, i.e., a representation that expressed each CCD pixel, as depicted in the raw picture encoded on tape, as a 2×2 block of identical numbers. This oversampling allowed for the interpolation of information that came out of transformations that yielded new coordinates that did not match the original CCD pixel boundaries. Henceforth, the term ``pixel'' will refer to a picture element in the oversampled representation where an image is represented as an array of dimensions 640×512 in x and y, respectively.

The sequence of operations to undo the undesirable effects marches backward through the parts of the IMAPS system. We start with artifacts produced by circuits that extract the signal from the CCD (§7.2), then to the CCD iself (§7.3 and 7.4), and then on to the photocathode to remove the variations in its response (§7.5) and correct for geometrical distortions in the image section (§7.6). Finally, we identify the magnitudes of the x and y motions from one exposure to the next (§7.7) that were caused by slippages in the echelle tilt mechanism (§6.1.4), so that differential offsets can be applied when the images are combined (§7.9). In various steps of combining multiple measures of intensities at a given wavelength, such as co-adding images (§7.9) or extracting the spectra (§7.10), we make sure that the relative weights reflect the respective exposure values that vary with location. That is, we do not simply add together corrected intensities, paying no attention to whether they are noisy or not.