Logarithmic wavelength axis: problems with splot?

pweilbacher wrote on Aug 22, 2011

Dear all,

is IRAF splot supposed to be able to read log-resampled spectra properly? I created some with my own software and found that splot displays what I think are wrong wavelengths. It could be that I understood the FITS WCS papers (especially "Representations of spectral coordinates in FITS") wrongly but I'm not sure...

An example spectrum is at http://www.aip.de/People/PWeilbacher/spec_log.fits, the relevant header keywords are

NAXIS   =                    1
NAXIS1  =                 1036
CRPIX1  =                   1.
CD1_1   =              0.78125
CTYPE1  = 'AWAV-LOG'
CUNIT1  = 'Angstrom'
OBJECT  = 'Test (DATA)'
CRVAL1  =                5530.

Any hints?
Peter.

---

pweilbacher wrote on Aug 22, 2011

OK, so I found http://adsabs.harvard.edu/full/1993ASPC...52..467V ("The IRAF/NOAO Spectral World Coordinate Systems") in the meantime and this is apparently still the way the spectral WCS is handled in IRAF. So while my file may be correct, and is actually handled correctly e.g. in SPLAT (and as 2D spectrum also in DS9), it probably cannot be recognized correctly by IRAF tasks until they are rewritten to take into account current WCS standards.

---

Francisco Valdes wrote on Aug 22, 2011

Hello,

As you surmised, the IRAF spectroscopy system predates the FITS work on spectral coordinate system by decades and updates to the IRAF coordinate systems is, unfortunately, pretty low priority in such a stable system. Logarithm coordinates are supported. In your example you need to

DC-FLAG =                    1
CRVAL1  =      3.7427251313047   (log10(5530.))
CD1_1   =   6.1350560486062E-5 (log10(5530.+0.78125)

This could certainly be automated easily in IRAF if there is a lot of data.

Yours,
Frank Valdes

---

Last post on Aug 22, 2011