Rob,
You asked how spectra that are not flux calibrated could be handled.
The VO spectrum model has fields
Spectrum.Char.FluxAxis.Accuracy.Calibration
as well as
Spectrum.Char.SpectralAxis.Accuracy.Calibration
with values ABSOLUTE, RELATIVE, UNCALIBRATED. These can be used to
indicate that the wavelengths are good but the fluxes are not
(or vice versa if appropriate!). It also handles the case
(RELATIVE) where things have the correct spectral shape but
(either because of slit width/position issues or lack of standard
star issues) everything can be off by a constant factor, and
distinguishes this from the case where it's basically raw counts
and the shape of the detector response is not calibrated out
(UNCALIBRATED).
This is a generalizable solution: if you have a data cube
with poistions, time, etc. you can independently specify
the calibration status of the astrometry, the clock, the
wavelengths, the fluxes - and thus give some sense of which
science the data is good for and which science it's not good
for.
(The other part of your discussion: Pedro and I are talking
about publishing things that were measured in mag originally
- possibly including AB-magnitude spectra I guess, but mostly
photometry - rather than back converting spectra to photometric
units. There certainly is an application for doing the latter
- what is the B mag that my spectrum implies? - but that should
be in a separate program and not built into the model/protocol.)
Jonathan Received on 2006-10-23Z17:14:14