[News] QC Photometry kick-off email
Ronald Vermeij
ronald at astro.rug.nl
Fri Jul 9 12:15:00 CEST 2004
Dear all,
I hereby resend the email about the QC Photometry that was send earlier to Mark and
Roberto.
Regards,
Ronald Vermeij
-----------------------------------------------------------------------------------------------
Dear Roberto and Mark,
As was agreed in our telecom of 1 July, I send you a small list of QC measures that are
in place/defined in the photometric software as it stands today. Also included are some
suggestions for any other QC measures. Not all QC measures are visible to the user or
have a QC flag associated with them.
Note that color terms are not mentioned here, because deriving these falls outside the
scope of the automatic photometric pipeline. Also note that because the photometric
pipeline runs on images that have been (pre-)processed by the image pipeline, that
quality issues regarding the flatfield/image pipeline also play a role.
I) Catalog creation:
1) Correctly identifying standard stars:
Depends on : proper astrometric calibration, quality of catalogued coordinates
of standard stars.
Currently, the robustness of identification is improved by removing artifacts
from the catalog before association with the standard star catalog (hot pixels).
Whenever possible the catalogued coordinates of the standard stars are also
improved.
2) Removing sources that are `unfit for purpose':
Before association, saturated sources, blended sources, and sources with clipped
or otherwise de-formed apertures are removed from the catalog. This way, `rotten'
standard stars never enter the photometric pipeline. This measure also helps to
improve the robustness of the association process.
Suggestions for improvements/points of concern:
How to monitor and influence the quality of the aperture photometry ? Can any
meaningful QC scheme be defined for this ? Curves-of-growth ? It will involve
heavy manipulation of SExtractor. Suggestions are welcome.
II) Atmospheric extinction:
1) In daily routine, the atmospheric extinction will be dealt with by the monitoring
requirement of the OmegaCAM pipeline (req 562). This requirement produces a report
containing a measurement of the extinction for every image of the polar field that
goes into making it (at least three). This measurement is obtained by fitting a
standard extinction curve to the data that has been simultaneously observed in
every one of the four key-bands (composite filter).
The QC measures/flags in place are:
- the measured zeropoints for all the stars observed in one kwadrant of the
composite filter are clipped for outliers before a single value for the
extinction is determined from these. This clipping will also remove any
sources from the pipeline that have mistakenly been identified as a standard
star. After determining this value, the result is checked against a standard
extinction curve. Raising a QC flag at this point is only a first warning
that something fishy might be going on with the atmosphere. This QC warning
is only send to the log.
- when for all the four kwadrants of the composite filter an extinction has
been derived, an extinction curve is fit to the four points to derive the
ultimate extinction for that particular moment in the night. The fit
parameters are send to the log for perusal.
- after having processed all the polar images in the same way, an assesment
is made of the quality of the night (QC = SHIFT_OF_CURVE_VARIED), of the
stability of the overall transmission properties of the atmosphere
(QC = SHAPE_OF_CURVE_VARIED), and of the validity of the assumed shape of
the standard extinction curve (QC = INVALID_ASSUMPTION_FOR_SHAPE_OF_CURVE).
These QC-flags are stored in the DB together with the separate values for
the extinction through the night, and the fitted shifts of the extinction
curve.
Suggestions for improvements/points of concern:
Although QC-flags have been defined, the software that checks on the anomalies
must be written. A set of criteria on the basis of which the QC-flags are
raised has not fully solidified yet.
III) Zeropoint:
1) The zeropoint for a given chip/filter combination is derived by combining the
extinction results from req 562 with the raw zeropoints contained in the catalog
for that particular chip/filter combination. Note that these catalogs are derived
from equatorial standard field data. The error bar on the resulting zeropoint serves
as a QC parameter; this number should be within the specs as given in the requirements.
For further QC, a comparison can be made with the zeropoint derived the day before.
(Remember that any color terms that might be needed have already been taken care of
in the catalog creation stage.)
Suggestions for improvements/points of concern:
Clipping of the input data should also be done here. It might be interesting
to make a residual plot for validation. Systematic off-sets from the zero-line
could invalidate the photometric calibration. A residual slope is a sign that
the color-terms determined independently in req 565 need revision. The scatter
is an indicator for the quality of the illumination correction and the
flatfielding.
This is it for now. Hope to hear from you.
Regards,
Ronald Vermeij
Astro-Wise
Kapteyn Astronomical Institute
Groningen, The Netherlands
More information about the News
mailing list