The location of the spectrum in the low-dispersion SI (in the cross-dispersion direction) is found by computing the wavelength-averaged flux-weighted centroid of the background-subtracted flux levels within a search region that is centered on the predicted spectrum location. For SWP images the region of the low-dispersion SI below 1233Å is excluded from this calculation so as to avoid contamination from geocoronal Ly. If insufficient signal exists from which to determine reliably a location for the spectrum, the predicted center position is used instead and a warning message to this effect appears in the processing history log.
If the spectrum centroid can be determined empirically, then the dispersion line that contains the maximum net flux (summed over wavelength) is also determined and is referred to as the spectrum peak location. The wavelength-averaged net flux of this dispersion line is computed and compared against the threshold value of 5 FN in order to determine whether the spectral signal is strong enough to attempt an empirical profile fit. For the SWP camera, the region below 1233Å is excluded from this computation so as to avoid contamination from geocoronal Ly. If the image is processed as a point source, then a warning message appears if the location of the spectrum peak is found to be more than one image line away from the spectrum centroid, since this may indicate that the spectrum is not characteristic of a point source but rather may be spatially extended.
If the average FN of the peak line of the spectrum is below the threshold value, then an empirical fit is not attempted. Instead, a default profile shape is used to compute the extracted fluxes. There are three default profiles for point sources, one for each camera. Each default profile has been derived from well-exposed observations of IUE standard stars, by averaging the profile fits for these observations over wavelength and renormalizing, thus producing a profile that is constant in wavelength. Since the spectral profile for IUE data is not repeatable from image to image, this average profile provides a rough weighting that is useful in very low signal images and provides a lower noise spectrum than attempting to fit the profile from the image. For extended and trailed sources, the default profile has a boxcar-like shape where all spatial image lines receive equal weights. Warning messages appear in the processing history log when a default profile is used.
A warning message is also issued if the location of the spectrum centroid is found to be more than two image lines away from the predicted spectrum center. This situation may arise if the location of the spectral format within the raw image space does not follow, for whatever reason, the normal behavior of the spectral format as a function of time (observation date) and camera temperature (THDA) as may be the case for objects poorly centered in the aperture. If this is the case, then the time and temperature-corrected dispersion constants that are computed for this particular image may be called into question. Note, however, that this warning only indicates that the location of the spectrum in the cross-dispersion (i.e., spatial) direction has been found to be unusual. We have no information as to whether or not a similar shift may have occurred in the dispersion direction of the image.
The search for the spectrum center and the determination of the average peak FN is skipped entirely for images classified as flat-field (i.e., nulls and other floodlamp only images), since they contain no spectral data. In this case the predicted spectrum center location and the default profile for extended/trailed sources are used to perform the computation of extracted fluxes.