The IRIS2-specific spectroscopy side of ORAC-DR has uncovered a
bug/feature in FIGARO/OPTEXTRACT that I'd like some clarification on.
When we do optimal extraction in ORAC-DR we use the VARIANCE component for
weighting. For IRIS2 they found that residuals from sky lines cause
spurious emission features to show up in extracted spectra if they keep
variance weighting on, so they turned it off. This produces a bug later on
in the pipeline when a signal-to-noise calculation is performed. It turns
out that for IRIS2 extracted spectra have no variance information, and
this is because they've turned off variance weighting in OPTEXTRACT.
The documentation for OPTEXTRACT says "the output spectrum has an error
array only if the input image has an error array." It does not say that if
you turn off variance weighting then the output spectrum will not have
variance information. Is this variance dropping an undocumented feature,
or is it a bug?
If it is a feature, is it possible to have OPTEXTRACT propogate variance
information even if variance weighting is turned off?
I do notice (running it at the commandline) that with variance weighting
turned off OPTEXTRACT reports "Warning: The data in the reference SPECT
have been re-shaped but the variance array was never updated. It will now
be deleted", so I guess it's a feature. Is it still possible to get the
variance propagated?
Stuart Ryder at AAO discovered this one (he maintains the IRIS2 ORAC-DR
code), and I've CC'ed him on this email. As I'm heading to Canada tomorrow
for a couple of weeks and away from email and thus won't be able to
forward on responses, can any responses please be CC'ed to him?
Cheers,
Brad.
|