IRAF Newsletter -- Number 14 -- April 1998

[ Previous ] [ Next ] [ Table of Contents ] [ Search this issue ]

Iterative Correction of Spectra: TELLURIC and SKYTWEAK

Some types of calibration or correction operations on spectra require subtracting or dividing one spectrum by a second matching spectrum. Two examples of this are dividing an object spectrum by the spectrum of a "featureless" star spectrum to remove telluric absorption lines and subtracting a blank field spectrum from an object spectrum to remove telluric emission lines. Even though these are simple operations, despite the best efforts at data reduction, the match between the two spectra may not be sufficient to completely remove the telluric features. But the corrections can be improved by "tweaking" the spectra to improve the match and better remove the sky features. Because the features to be removed are mixed-in with features of the object spectrum this tweaking ultimately requires iteration, with the astronomer supplying the judgement as to what constitutes the best correction.

The two first order "tweaks" are small shifts in the wavelengths and finding the best intensity scale to match the spectra. One could use existing tasks to apply these tweaks, using image arithmetic tasks to perform the division or subtraction, and examining the result with something like splot . However this is very tedious if a lot of iterations or spectra need to be done. There are two new, related tasks in IRAF V2.11 to make this kind of iteration easier. These tasks are telluric and skytweak .

The telluric task intensity scales a calibration spectrum, normally that of a hot star, using a simple Beer's law to approximate the curve of growth of the telluric lines through the atmosphere. It also shifts the calibration spectrum in wavelength. The scaled and shifted calibration spectrum is divided into the object spectrum and three versions of the corrected spectrum are shown. The three versions use different values of shifting or scaling. The astronomer can then use the graphics cursor to select which of the three is best. Two new corrections with shifts or scales on either side of the selected correction are computed and graphed for the astronomer to again compare and select the best result. The program automatically or by user selection adjusts the steps between the possible shifts and scales to "home in" on the best result.

Caption: The two spectra shown in the figure are of a giant star in M92, with and without the removal of telluric H2O lines using the telluric task. The spectra were obtained from the NOAO WIYN Queue program. In addition to prominent stellar and interstellar components of the Na D lines, the original spectrum is cluttered with telluric water vapor lines, making the measurement of line strengths and velocities difficult. The telluric task was run automatically, without manual intervention to adjust the shift or strength of the telluric lines. The telluric Na D emission lines were also removed using the sky subtraction feature of dohydra, leaving spectra from which measurements can be made easily.

There are a number of additional features to help find and evaluate the correction. An initial shift may be determined by cross-correlating the object and calibration spectra over the whole spectrum or within certain regions. Regions which are fairly clear of object features can be defined interactively or by parameters and used to have the program automatically search for the scaling and shift that minimizes the RMS of the calibrated spectrum in those regions. The RMS is also shown during the interactive iterations as a guide to how well the correction is being done. The original object and calibration spectrum can also be graphed to help the user decide which features may or may not be removed.

The skytweak task is basically the same program except that the scaling is simple multiplicative scaling and the corrected result is the subtraction of the sky spectrum from the object spectrum. The same features of cross-correlations, pointing to the best result of three, and automatic RMS minimization in certain regions are available.

Learning to use these tasks takes some practice, but once you understand the features and how to approach the desired result they can be extremely powerful and quick to use to improve your telluric or night sky corrections.

Frank Valdes

[ Previous ] [Next ] [ Table of Contents ] [ Search this issue ]

IRAF Group, National Optical Astronomy Observatories, P.O. Box 26732, Tucson, AZ 85726, Phone: (520) 318-8160, FAX: (520) 318-8360, Email

Posted: 07May1998