Several details of the flux calibration process have been ignored so far; we have not removed spectral features in the flux calibrator star, we have not allowed for slit losses in the measurement of the flux calibrator, and we have assumed that terrestrial atmospheric absorption features were accurately removed during the flux calibration process. Each of these points can be addressed by applying a multiplicative wavelength-dependent correction to the object spectrum which we generically call a `flat spectrum star' correction.
The correction for spectral features in the flux calibrator is derived from a measurement of a star lacking these features. Generally, a late-type star is used to correct for features in an early type star, and vice-versa. The ratio of the continuum star spectrum to the flux calibrator spectrum is formed, a fitted continuum is multiplicatively removed, and all regions in the resultant spectrum except for the features of interest are set to unity. This minimises the noise added to the corrected spectrum, and prevents other spectral features in the continuum star being imprinted on the corrected spectrum.
Slit losses in the measurement of the flux calibrator must be corrected for in order to place the object observation on an absolute flux scale. This correction is formed by flux calibrating a wide slit measurement of the flux calibrator with the same narrow slit measurement used to flux calibrate the object spectrum. The overall shape of the calibrated wide slit spectrum will be severely affected by poor cancellation of terrestrial atmospheric absorption, but the level in clear regions defines the multiplicative slit correction constant to be applied to the object spectrum.
A correction for residual terrestrial atmospheric absorption is formed from a measurement of a `flat spectrum' star located near the object of interest. The `flat spectrum' star measurement is flux calibrated with the same calibrator measurement used for the object spectrum and any intrinsic spectral features removed. The absorption correction is again formed by multiplicatively removing a fitted continuum and setting featureless regions of the correction spectrum to unity.
The cspfltspec task is used to form the `flat spectrum star' correction to be applied. The parameters for this task are listed below. flatspec is the base name of the flat spectrum star. The correction can be based either on the flux calibrated object spectrum (.fos file extension) of this name or the unfluxed object spectrum (.osp file extension), and the correction spectrum is stored in a file named by appending .flt to this base name. The constant parameter defines the scaling constant by which the derived correction spectrum is divided. Since redgspec divides by the correction spectrum, the object spectrum is multiplied by this constant. This will normally be the value of the correction for slit losses, or unity if no slit loss correction is required.
I R A F
Image Reduction and Analysis Facility
PACKAGE = caspir
TASK = cspfltspec
flatspec= sp163 Flat spectrum star file to use
(constan= 1.0) Divisor constant
(verbose= yes) Verbose output?
(mode = q)
The `flat spectrum star' correction is formed by first fitting the continuum of the specified object spectrum interactively using the IRAF icfit task. The default sample region is the whole spectrum. This region can be deleted by typing z and new sample regions chosen to avoid spectral features by positioning the cursor at the start position and typing s and then repositioning the cursor at the end position and typing s again. Redo the fit by typing f, and replot the spectrum by typing r. The order of the fit is changed by typing :order 3. Type ? for help on the cursor key functions, and type q to quit icfit when a satisfactory continuum fit has been obtained.
The cspfltspec task then plots the ratioed spectrum in the graphics display using the IRAF splot task. Users should become familiar with the many features of this task listed below. Specifically for the moment, set regions away from the features of interest to unity by positioning the Y cursor at 1.00, the X cursor at the start of the region to set, and typing x. Move the cursor in X to the end of the region to set while keeping the Y cursor value at 1.00 and type x again. Repeat as necessary.
When a suitable correction spectrum has be constructed, remember to type i to save the correction spectrum to the output file from within splot. The cspfltspec task cannot do this for you. Type q to quit.
Once the `flat spectrum star' correction has been formed, object spectra can be divided by this correction using the redgspec task by setting the flatdiv flag and specifying the base name of the flat spectrum correction file in the flatspec parameter.