The determination of the dispersion relation for MXU data is done as follows: The recipe fors_calib identifies reference lines on MXU arc lamp exposures, and traces the spectral edges on the associated flat field exposures. With this information the spectral extraction mask to be applied in the scientific data reduction is determined. From the input flat field exposures a normalised flat field frame is also derived. The input arc lamp and flat field exposures are assumed to be obtained quasi-simultaneously, so that they would be described by exactly the same optical distortions. Wavelength calibration is performed row by row and produces a 2-D rebinned science frame. This has the effect of removing the instrumental line curvature and allows the user to extract her/his spectra and have them wavelength calibrated without any additional operation on the frame itself.
The global list of lines currently used by FORS2 pipeline can be found here.
Due to the high number of possible grism/slit combinations, the Pipeline currently uses a polynomial with a maximum degree of 5 (for 150I without filters) in the dispersion direction, while a 2nd  order polynomial is adopted in the spatial direction. The current version of the pipeline is known to have some problems for slitlets with extreme offsets, that result in arc spectra with few lines.

QC1 parameters

• average resolution (QC1 database table fors2_mxu_wave, column resol_avg)
  This column provides the mean resolution averaged over all spectral lines detected on the complete CCD. It may be misleading if there are small square slitlets, which are used just for centering the mask, but produce of course spectra with a very low resolution.
• mean rms of resolution (QC1 database table fors2_mxu_wave, column resol_meanrms)
  This column provides the mean RMS of the resolution per line averaged over the whole wavelength range.
• number of independent identified arc lines (QC1 database table fors2_mxu_wave, column nwave)
  This parameter gives the number of independent arc lines that were identified and then used to calculate the above listed parameters.
• ratio of line width in averaged spectrum vs. line width in central row (QC1 database table fors2_mxu_wave, column FWHM_ratio)
  A wavelength calibrated and rebinned arc lamp frame is averaged along the spatial axis. In addition the central row of the rebinned frame is extracted. In both 1-dimensional spectra the emission lines are identified and their FWHM is determined via a Gaussian fit. For the outermost lines of the central row spectrum the FWHM of the line in the averaged spectrum is divided by the FWHM of the corresponding line in the central row. The larger of these two ratios (blue and red end of spectrum) is then noted as FWHM_ratio. A FWHM_ratio above 2 is an indication that the spectral lines in the rebinned image may not be perfectly straight.