IARC-Given fit to single spectrum, fit all spectra in a 2-D arc

Description:

The manual arc line fitter, ARC, can be used to fit a single spectrum obtained by summing successive cross-sections taken from the center of a 2-D arc. IARC will then apply that fit to all the rows in the 2-D arc, starting from the center and working out in both directions.

Parameters:

IMAGE

EXTRACT can be used to sum cross-sections from the center of a 2-D arc - an image containing a set of arc spectra (a long-slit arc). ARC can then be used to fit them, producing a list of identified arc lines, channel number & wavelengths. IARC can then use this fit - in the file ARLINES.LIS - as a starting point for a fit to each of the cross-sections in the 2-D arc in the file whose name is given by IMAGE.

RSTART

Normally, IARC starts in the center - where the initial fit should be best - and works its way out. RSTART is the cross-section (or row) at which it starts, and should normally be the central cross-section of those summed to produce the arc spectrum used in the initial ARC fit.

RWIDTH

IARC can sum consecutive rows of a 2-D arc, instead of fitting each row individually. Increasing RWIDTH will speed up the fitting, since it reduced the number of spectra fitted, and may reduce the number of lines that get rejected, since the signal to noise ratio in the summed rows will be improved. However, it will degrade the smooth progression from one row to the next, and may make the lines broader.

FILE

The results of the 2-D fit are written to a file in a form suitable for use by the image scrunchers, ISCRUNCH and ISCRUNI. For ISCRUNCH the default name is usually satisfactory. For ISCRUNI you may want to exercise more control over the process and the files used.

RSIGMA

Normally, you should use the same sigma (the line width parameter) for IARC as was used for the original ARC fit, and this is always the default. However, IARC is a little more tolerant in searching for lines than is ARC, and there may sometimes be a case for using a slightly lower value of sigma, especially if the arc lines are very close together.

SPREAD

IARC normally searches for lines in a new cross-section in two steps, once with a larger sigma than that specified in order to make it less likely to loose lines that may be a little further away than expected, and then with the specified sigma in order to tighten up the centering. Setting SPREAD=NO will disable this feature and the search will be performed once, with the specified sigma.

GAP

If IARC fails to find a line, it will use the position it had previously for the line. If the line is missing a specified number of times, however, it is deleted from the line list. This maximum number of times is specified by GAP. The correct setting for GAP depends on the nature of the arc - an arc with high distortion should have a low GAP value to avoid lines becomming confused. If distortion is low but several cross-sections are missing or have poor data, GAP should be higher to let IARC 'jump' over them.

SIGMIN

SIGMIN is a measure of the strength of an arc line. When the program searches for lines to use as 'lock' lines, it evaluates their sigma - here the line height as a multiple of the square of the local continuum. Any line for which this value is less than SIGMIN will not be used.

XCORR

If XCORR is set, IARC will attempt to determine a linear shift between successive spectra using cross-correlation and will apply this to the arc lines found in the previous spectrum before looking for them in the new spectrum. This is particularly applicable to fibre data, where such linear shifts often occur. It is probably not useful for cases such as image tube distortion, where the spectra should change in a constant manner.

DETAIL

If DETAIL is set, IARC will list the details line position, wavelength, fitted wavelength, error) for each line fitted. This is a lot of information, and is not normally useful.

LOCK

If the original fit has regions where there are strongish lines that have not been identified, it may help to use their positions to lock down the fit in those regions. This does not improve the fit, since the wavelengths are simply assumed to be those given by the original fit, but the automatic fits will not be so free to wiggle in these regions. However, you have to be careful not to get a large number of weak lines that dominate the fit at the expense of stronger lines that were positively identified.

WEIGHT

Indicates whether the least-squares fit is to be weighted using the peak intensity of each line (so that more weight is given to strong lines).

Source comments:

 I A R C

 Performs a 2-D fit to an arc spectrum, given an initial fit
 to a single arc spectrum as a starting point.  Generally, the
 starting spectrum will have been extracted from the center of the
 2-D arc.  IARC then starts at a suitable cross-section in the
 2-D arc - usually the central one - and works out from there,
 fitting each cross-section individually, looking for the lines in
 the starting spectrum.  For each line, the program looks for a
 peak close to where it was found in the previous fit.  If a line
 does not show up in one cross-section, the previous position will
 be used, but if it fails to show up in the next cross-section, it
 will be dropped from the search list.  In many cases, there may be
 strong lines which were not indentified; these cannot be used to
 improve the fit, but they can be used to 'lock' it down in the
 regions where there are few or no identified lines.

 Command parameters -

 IMAGE     (Character) The name of the image containing the 2-D arc.
 RSTART    (Numeric) The starting cross-section to be used.
 RWIDTH    (Numeric) The number of cross-sections to be added
           together for each fit - if the arc is weak, this will
           need to be increased.
 RSIGMA    (Numeric) Normally, the sigma value from the arc line
           file is used, but this can be overidden by RSIGMA.
 GAP       (Numeric) Number of cross sections over which IARC can
           fail to find a line before deleting it from the lists.
 FILE      (Character) The name of the file to which the results
           are to be written.  If an extension is not given, .IAR
           will be used.
 SIGMIN    (Numeric) The minimum acceptable value for the sigma of
           an arc line found in the locking process.  Sigma here is
           the height of the line relative to the square root of the
           continuum.  Only used if LOCK set.

 Command keywords -

 LOCK      Indicates that a search is to be made for lines to
           'lock' the fit.
 SPREAD    Indicates that IARC looks for lines first with an
           increased sigma, then with the specified sigma in order
           to refine the fit.  If SPREAD is not set, the search
           is just with the specified sigma value.
 DETAIL    Indicates that full details of the fits are to be output
           (This is mainly a diagnostic tool.)
 XCORR     If set, IARC will attempt to determine a linear shift
           between successive spectra using cross-correlation. This
           is particularly applicable to fibre data, where such linear
           shifts may occur. It is probably not useful for cases such
           as image tube distortion, where the spectra should change
           in a constant manner.
 WEIGHT    Indicates whether the least-squares fit is to be weighted
           using the peak intensity of each line (so that more weight
           is given to strong lines).

 User variables used -

 (<) IARC_WMAX   (Numeric) Maximum wavelength for any of the spectra.
 (<) IARC_WMIN   (Numeric) Minimum     "       "   "  "   "    "
 (<) NOFITS      (Numeric) The number of rows that could not be fitted.
 (<) ORDER       (Numeric) The order of the original fit.
 (<) RMSMAX      (Numeric) Maximum RMS error from the fits.

 Input files -

 ARLINES.LIS   Contains the details of the starting fit.  For format
               details see comments for subroutine ARGETL, or the
               ARC command.

 Output files -

 As named by   Contains the details of the 2-D fit.  Format :
 the FILE      Name of image, (24X,A).
 parameter.    Dimensions, NX, NY (17X,I5,4X,I5).
               # rows not fitted properly, (42X,I5).
               Maximum RMS error, (20X,F10.2).
               Maximum order used, - ORDER - (33X,I3).
               Then, for each row, row number and ORDER+1 polynomial
               coefficients, the constant term being the last
               non-zero term.  (I14,10X,2D24.16,3(/3D24.16)).

                                KS / CIT  15th June 1984