The very first step after correcting bias, dark, and overscan is to find the edges of the slitlets. This is done by the command LOCATE/MOS. This command locates the slitlets in an MOS flat-field frame by searching for the maximum (normalized) gradient in a trace perpendicular to the direction of dispersion. Position and width of the trace are given by (0). (0) gives the minimum normalized gradient that must be exceeded, after median filtering the scan with a median of width (0) and discarding scan values below (0). The result is written to the output table .tbl (which is used by most MOS commands) and the number of detected slitlets is written to (0). The programs allow at most 100 slitlets. If the algorithm does not find any slitlets the chosen threshold ( (0)) may either be too high (above the intensity of the flats in the center of the frame) or too low (below bias value). Also the width ( (0)) may be chosen to high or too small. Typical values are between 0.1 and 0.2 and 3 and 5, respectively.
It is also possible to define the slitlets interactively with DEFINE/SLIT. Here you first initialize the table .tbl (mos) and then enter the limits with the cursor on the displayed flat field frame. This comand also allows an easy definition of the table for long-slit data.
With LOCATE/MOS the offsets in dispersion direction between the slitlets will be read from the header of the flat-field frame for FORS data and stored in the table .tbl (mos) in column :xoffset. For other data or DEFINE/SLIT you will have to determine the offsets yourself using the command OFFSET/MOS on a wavelength calibration frame (see below).
As spectroscopic flat-fields normally exhibit the spectral characteristic of the lamp that was used to produce them you have to take out this characteristic in order to correct the CCD sensitivity variation and keep the original flux distribution. This is done with the command NORM/MOS. It takes an averaged flat frame and the slit limits stored in the table .tbl. There are two methods provided for the normalization ( (poly)): In case NORMMET=poly it averages separately for each slitlet the rows, fits a polynomial of chosen degree ( (3)) to the flux distribution obtained this way and divides each row in the slitlet by this polynomial. In case NORMMET=median it averages separately for each slitlet the rows, smooths with a median filter of pixels width and divides each row in the slitlet by the filtered average. You may also perform the flat correction at the same step using the command FLAT/MOS.