For each SCIENCE frame reduced in optimum extraction mode, the pipeline creates a result table of category "ORDER_TRACE" which contains detailed information about the position and fitted cross-dispersion width of the extracted signal. The width information is visualized in the QC1 plot "Signal width". There is one such plot per CCD (i.e. per reduced science spectrum).

The information is plotted in a matrix which has seven columns for the order-by-order plots, 1 column for aggregate plots, and a final column for summary information.

Find a complete sample plot for order position here.

There is a tutorial on position and FWHM plots which gives more information about the proper and problematic cases.

Order-by-order plots.
- The pipeline in optimum extraction mode fits a Gaussian to the cross-dispersion profile of the signal. For each order, the width of the fitted Gaussian is determined (with input from the order table) at about 50-100 positions. These widths are plotted as red dots (column :P_FWHM of ordertrace table).

- In the next step, a 2nd order polynomial is fitted to these values where all values with an error above a threshold are excluded. These points are taken from column :S_CH0 and plotted in green.

- In the last step, a kappa-sigma routine is used to clean the FWHM values from outliers with high weight factors (small errors). The final result is in column :S_CH1 and plotted in blue.

These plots are provided for each echelle order. Abscissa is X coordinate (dispersion direction) in pixels, ordinate is width in cross-dispersion direction in pixels. Inclination of the orders has been removed prior to the fitting procedure.

The fitting and clipping routine has been optimized to obtain a signal tracking and fitting solution stable against cosmic hits, bad columns, absorption or emission lines.

Aggregate plot (per 7orders). For each set of 7 orders constituting a row in the QC1 plot matrix, all :P_FWHM values (red dots) are plotted together in the 8th column, with their mean value indicated. This 'aggregate' plot allows, in cases of faint signal with large scatter of points, a better evaluation of the reliability of the results. (This example shows quite a large scatter.)

Aggregate plot (all orders). For the same reason, there is a similar plot (in column 9) with all :P_FWHM values (red dots) combined.

Aggregate plots (combined view). The aggregate plots in column 8, viewed together, permit a quick-look evaluation of certain typical extraction problems. E.g., they show whether the 'true' cross-dispersion profile deviates from a Gaussian, e.g. due to over-exposure or to a 2D structure of the object. Also exposures with 'too-high' S/N can be identified here easily (they would show a small-scale ripple structure around the mean value).

Summary information (mean values). Finally there is a plot with the mean FWHM values (orderwise). These values are also plotted in the main reduction QC1 plot and are stored in the FITS header (keywords HIERARCH ESO QC FWHMnn POS where nn is the order number). Here basically insufficient flux would become visible.