Previous: Flux calibration
Up: Reducing your data
Previous Page: Flux calibration
Next Page: LEXT manual
Once the data in the image frame has been preprocessed, and the required subsidiary data are available, the multiple spectra can be extracted and calibrated. The main routine for this purpose is called EXTRACT. This routine has a large number of options allowing most of the sophisticated reduction processing to be preformed with the one command, provided the necessary LISTs are present. The normal requirements would be a BOXES list, a COEFS list and an FSPEC list.
The minimum requirement is the availability of an appropriate BOXES list to define the expected positions of the spectra and objects. This list should have been produced by running PREBOX on the main MASK list for the field. The BOXES list itself can be updated in the light of the actual data using the UPDBOXES command. This allows the user to interactively identify the spectrum and object ranges from a summed spatial profile.
The syntax of the EXTRACT command is to have the box numbers specified on the command line:
LEXT -- IMAGE 1 > EXTRACT 2-5 6-10 15or, if all spectra should be processed:
LEXT -- IMAGE 1 > EXTRACT ALLPrior to running EXTRACT the user must have set up a name for the field with SET OBJECT objectname. The extracted spectra are then written to 1D FIGARO spectrum files with file names formed by objectname_boxnumber. There are many options which can be specified with EXTRACT and full details of all of these are given in the on-line HELP. Here we consider the most pertinent ones.
By default EXTRACT subtracts the sky from behind the object and then makes a gaussian weighted sum of the counts in the object rows. If a COEFS list is present in the LIST stack, which is produced by ARC, then it is picked up automatically and used to resample the data on a linear wavelength scale. If an FSPEC fluxing function is available, and either the /FLUX or /DFLUX options specified, then the output spectrum will also be flux calibrated. The FSPEC list itself must be obtained from an LDSS observation of a standard star, whose spectrum is known from some other source.
The level behind the object is estimated by fitting to the rows of the spectrum which contain just sky. The user will probably want to specify the /INTER option, at least the first time through the data, so as to have some control over this. In fact polynomial fits are used, one to the blocked spatial profile and another to the residuals in each column after normalizing the blocked spatial profile. The idea is that the blocked profile gives good signal to noise for gauging the slit profile and gross features in the background, whilst the extra column fit can account for column to column deviations which might occur, particularly close to bright night sky lines. The user is prompted for the polynomial orders for both of these fits. Obviously if either of these is given as zero, then the fit becomes effectively either pure block fitting or pure column fitting.
The gaussian weighting scheme used in the extraction has been found to give fairly good results, although it can be over-ridden by the /OBJECT qualifier. Alternatively the HORNE routine can be used to provide more sophisticated optimal extraction, although this requires an already sky-subtracted image, such as can be produced by EXTRACT/ALTER.