Processing codes are intended as notes which are appended to subsets of the data. Suppose, for example, that an array of data has a gap of bad or missing data, and that the most frequent use of the data requires that the gap be filled by some form of interpolation. For efficiency, it is desirable to write the interpolated data into the file with the real data, so that the interpolation will not have to be done again and again, each time the dataset is read. On the other hand, it would be very dangerous to simply mix the interpolated and real data in the dataset indiscriminately.
Processing codes are meant to handle such situations. One can use Procgroup record groups to mark any part of the data to which the attention of the cautious user needs to be drawn.