Peter Weilbacher This recipe detects arc emission lines and fits a wavelength solution
to each slice of the instrument.
The wavelength calibration table contains polynomials defining the
wavelength solution of the slices on the CCD.
Processing trims the raw data and records the overscan statistics,
subtracts the bias (taking account of the overscan, if --overscan is not
"none") and converts them from adu to count. Optionally, the dark can be
subtracted and the data can be divided by the flat-field, but this is
not recommended. The data is then combined using input parameters,
into separate images for each lamp.
To compute the wavelength solution, arc lines are detected at the center
of each slice (using threshold detection on a S/N image) and
subsequently assigned wavelengths, using pattern matching to identify
lines from the input line catalog. Each line is then traced to the edges
of the slice, using Gaussian centering in each CCD column. The Gaussians
not only yield center, but also centering error, and line properties
(e.g. FWHM). Deviant fits are detected using polynomial fits to each arc
line (using the xorder parameter) and rejected.
These analysis and measuring steps are
carried out separately on images exposed by the different arc lamps,
reducing the amount of blending, that can otherwise influence line
identification and Gaussian centering.
The final two-dimensional fit uses all positions (of all lamps), their
wavelengths, and the given polynomial orders to compute the final
wavelength solution for each slice, iteratively rejecting outliers. This
final fit can be either unweighted (fitweighting="uniform", for fastest
processing) or weighted (other values of fitweighting, for higher
accuracy).