When the ropp_pp_occ_tool fails to process an occultation, for whatever reason, QC output should be given in the NetCDF file. For example, the option of stat.opt. with 'regular' search and fit, should write out the fitting coefficients. Same principle, but different in detail, for the option 'convoluted'.

A profile should always be processed to the end if at all possible (sometimes it may not be possible because of very erroneous data, but then QC parameters indicating that should be written out). The output should be written to parameters similar to the exsiting SO_badness and L2_badness scores.