FINDSHIFT computes the relative shift between two spectra. Given two spectra (1 row images), it computes SHIFT, the number pixels that the second (shifted) spectrum would have to be shifted to match (in the least squared sense) the first (template) spectrum. The user can selected a spectral region for the chi-squared computation. The algorithm solves the interpolation problem, so the command provides accurate fractional shifts.
Since the algorithm minimizes the difference in counts, results are more accurate if the spectra are normalized to have the same number of counts within the spectral region of interest. The algorithm looses sense if the spectra are very different.
If two images are given (more than one row), a single row will be considered for the shift calculation, the template's starting row is the default row, but can be changed with the 'ROW=' keyword.