The APNORMALIZE task simply divides the profile along the slit length by the number of pixels corresponding to the aperture width or the mean value of the central two pixels. The spatial profile along the slit length is retained. The spatial profile for median flat fielding contains intensity variation due to nonuniformity of the slit width. The spatial profile must be retained in the flat frame to calibrate spatial profile in the spectra of objects, sky, standard star, and comparison.
The APFLATTEN task uses `optimal extraction routine' for extracting the illumination profile to maximize signal-to-noise ratios (Marsh, 1989; Horne, 1986). The optimal extraction algorithm assumes that a spatial profile along the slit length is a smooth function of wavelength and can be modeled as a normalized probability distribution. While this method is ideal for spatially unresolved targets or short slit, it is not appropriate when spatial information along the slit is required or important as for objects with spatially extended emission line regions.
Figures 34 and 35 shows the flat frames made by the APNORMALIZE and APFLATTEN tasks. It is clear in these figures that the APFLATTEN task alters the spatial profiles, but the APNORMALIZE task does not. In addition, the APFLATTEN task takes the bad pixels or cosmic ray events into account, altering the intensity distribution around these pixels, but the APNORMALIZE task does not.
The large scale curved patterns, or ``tree-ring'' patterns, seen in Figure 34 and 35 show a characteristic flat-field variation of InSb detectors (McLean, 1997).
Figure 35 (a) shows a stripe pattern along the slit length for the entire aperture. This pattern may be due to the nonuniformity of the 
-band slit width.
Because we mainly concern spatially extended objects, we use the APNORMALIZE task to make a normalized flat frame.