This documentation is intended both for QC scientists and SciOps astronomers (who may want to ignore the technical information displayed in grey).
|
|
|
|
|
bottom
plots
|
contact
|
|
NAME |
flat.prg
|
|
VERSION |
1.0 -- created: May 2003
1.1 -- have more detailed fibre plot (2003-11-24)
1.2 -- have efficiency rms and reference files (2003-12-22)
1.2.1- include ORDER and SETTING (2004-04-27)
1.2.2- P3 replaced by DISPLAY variable Y/N (2006-04-18)
1.2.3- check for mising fibres added (flat_missing.sh) (2007-01-12)
1.3 -- include scoreQC (2007-02-26)
1.3.1- includes qc_numsat (2009-02-17)
|
|
SYNTAX |
MIDAS
|
|
CALL |
processQC -a abname; or:
from $DFS_PRODUCT/FFLAT/$DATE
inmidas -P -j "@@ $DFO_PROC_DIR/flat.prg $date $primfile Y"
(the last parameter Y/N controls creation of graphics
on the screen)
|
|
INSTRUMENT |
GIRAFFE
|
|
RAWTYPE |
FFLAT
|
|
PURPOSE |
compares raw and product fibre FLAT files
for FLAMES_GIRAFFE QC assessment; the plots are later reviewed
within certifyProducts. The first plot (flat1.gif) is stored
in $DFO_PLT_DIR and archived. The fibre flats are taken at daytime
as robotic flats and, at least for the IFUs, suffer from insufficient illumination,
but are taken regularly and are used for instrument HealthChecks.
|
|
PROCINPUT |
first raw fibre FLAT; products analyzed:
FF_LOCWIDTH, FF_LOCCENTROID, FF_EXTSPECTRA, FF_EXTERROR, MASTER_FIBER_FLAT
where FF stands for fibre flat
|
|
QC1TABLE |
trending | table(s) in QC1 database:
giraffe_flat
|
|
TRENDPLOT |
trending | HealthCheck plot(s) associated to this procedure:
trend_report_FFLAMP_HC.html |
trend_report_FFLEVEL_HC.html |
trend_report_FFLOC_HC.html |
../FIBRE/CURRENT/Medusa1/trend_report_Medusa1_CURRENT.html
|
|
QC1PAGE |
trending | associated documentation:
lamp_qc1.html |
fibre_qc1.html
|
|
QC1PLOTS
top |
flat1.gif
box1: middle row in first raw (shows the fibre pattern);
box2: closeup of box1 (X950-1050); Gaussian shape of fibre signal (X direction = cross-dispersion);
box3: visualization of FF_LOCWIDTH (top) and FF_LOCCENTROID (bottom) over fibre index; these are the Y-averaged width and centroid positions of each found fibre. All width values should form a nice parabola, with small variation of the average width (broken line); outliers indicate a missing fibre or a bad localization solution!
All centroid points should form a nice straight line, with outliers indicating sky and simcal fibres; any other outlier indicates a missing fibre or a bad localization solution!
box3 is very sensitive to missing fibres (the ones not specified in the recipe call).
box4: histogram of 1st raw (black) and master (red); should drop to zero well below 65000 ADU (saturation check)
box5: closeup of box4 around the saturation limit; note that qc_numsat counts all pixels above 60,000 ADUs
box6: mean of extracted fibres: Y-collapsed FF_EXTSPECTRA, should show scatter around 1 and be higher than 1 only for SKY or SIMCAL fibres
box7: X-collapsed fibre efficiency for: fibre #1, #10, all fibres averaged (black)
top panel: name of analyzed product FF_EXTSPECTRA
bottom panel: file and QC1 parameters (exptime, nfib, mean_width, rms_width, mean_xrms, mean_xdiff, med_fib, num_out, mean_raw)
scoring: num_fib, num_out, mean_width, rms_width, flux, mean_raw, qc_numsat are scored
flat2.png
full display of raw frame (box: area displayed in flat3.png)
flat3.png
closeup of raw frame (box: area displayed in the graphical plot as box 2)
flat4.png
display of products (top to bottom: location width, fibre extraction, derivY1 where derivY1 is the derivative in Y direction with 1 px offset)
derivY1 is useful to see fringing effects. All three displays look smooth, without much structure, in case of proper tracing and extraction, and look wildly patterned if something went wrong.
flat5.png
fibre tracing plots (there are actually 2 for Medusa, and 4 for IFU/Argus).
They display the raw signal with the FF_LOCCENTROID results overplotted. They also read the fibre and ozpoz binary tables (attached as extensions to the raw files) to know about SIMCAL and SKY fibres, as well about allocated fibres. All these types are marked in these plots. They are useful as sensitive control plots for proper fibre extraction and location. Any plotted tracing is used for signal extraction downstream.
|
|
QC1PARAM |
QC1 parameters written into QC1 table:
QC1_db names: exptime | datancom | mean_raw | nfib | mean_width | rms_width | mean_xrms | mean_xdiff
| med_fib | sigma_fib | derivY | flux | lamp_effic
FITS key names: EXPTIME | DATANCOM | (flat.prg) | NPIX | (all others: flat.prg)
QC1 params created by pipeline: most
|
|
ALGORITHM |
Description of algorithms: see http://www.eso.org/observing/dfo/quality/GIRAFFE/qc/lamp_qc1.html
|
|
CERTIF |
Reasons for rejection:
- scoring alerts, in particular saturation
- unusual patterns in flat1.gif, box 3
- unusual patterns in flat4.png, deviations of tracing and signal in flat5.png
- these are almost always due to a missing (broken) fibre; then the input parameter set for that specific fibre set (slit) needs to be adapted; sometimes very steep flux gradients can prevent proper localization.
|
|
COMMENTS |
Scoring: num_fib, num_out, mean_width, rms_width, flux (this for HC settings only)
All pipeline-provided QC1 parameters are documented in the pipeline users manual (as of issue 2, 2007-12-28)
mean_effic has been added in 2008 (QC.MEAN.EFFIC) since it comes in opslog files; but not used in HC plot since otherwise HISTORY would be lost
Note that qc_numsat counts all pixels above 60,000 ADUs
|
top |
[2015-10-29T10:06:22] created by qcDocu v1.1.2, a tqs tool
|