Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stecf.org/conferences/adass/adassVII/reprints/morganteg.ps.gz Äàòà èçìåíåíèÿ: Mon Jun 12 18:51:47 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 04:05:31 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

Astronomical Data Analysis Software and Systems VII
ASP Conference Series, Vol. 145, 1998
R. Albrecht, R. N. Hook and H. A. Bushouse, e
Ö Copyright 1998 Astronomical Society of the Pacific. All rights reserved.
ds.
Pipeline Processing and Quality Control for Echelle Data
G. Morgante and F. Pasian
Osservatorio Astronomico di Trieste, 34131 Trieste, Italy
Email: pasian@ts.astro.it
P. Ballester
European Southern Observatory, D­85748 Garching, Germany
Abstract. In the framework of a collaboration between ESO and OAT
concerning the development of a data reduction pipeline and the related
quality control subsystems software for UVES/VLT, a detailed feasibil­
ity study is being performed on the basis of data extracted from the
EMMI/NTT archive.
The pipeline reduction is based on an accurate set of ``pre­calibrated
solutions'' and will also result in a powerful tool for the release and the
distribution of the scientific data, from a simple quick look at the data
just collected to a complete echelle reduction with the highest possible
accuracy.
1. Overview
It is ESO policy to perform data reduction for VLT instruments in ``pipeline''
mode, either at the observing premises or at headquarters. The result of the
pipeline will be a ``standard quality'' data product, certified by the Observatory.
In order to verify if a pipeline was feasible for echelle data, in particular
for UVES 1 (Ultraviolet­Visible Echelle Spectrograph), a test on EMMI 2 (ESO
Multi­Mode Instrument) data acquired in echelle mode and extracted from the
NTT archive was performed:
. a preliminary processing pipeline was defined using the MIDAS echelle
package;
. a set of pre­calibrated solutions was available, and other have been de­
fined, to optimize the processing speed while achieving results within an
acceptable level of accuracy;
. archive data were processed in pipeline mode;
1 http://www.eso.org/observing/vlt/instruments/uves/
2 http://www.ls.eso.org/lasilla/Telescopes/NEWNTT/NTT­MAIN.html
337

338 Morgante, Pasian and Ballester
. quality control on the data was performed, with the purpose of determining
the instrumental stability.
It is important to note that the approach followed in the development of the
work has been to minimize assumptions, or a priori knowledge of the EMMI
instrument, of its data and parameters; the decision was made to rely uniquely
on the contents of the archive and of the keywords of the files stored therein. As
a consequence, an evaluation of the completeness of the archive of EMMI echelle
data was also performed.
2. Key Items
The key points of the work were identified to be the following:
. test of the performance of the preliminary version of the automatic reduc­
tion chain compared with the interactive calibration;
. adaptability of the pipeline procedures on data collected with the R4 grat­
ing;
. verification of instrument stability over time with particular reference to
geometrical and photometric stability;
. analysis of the accuracies expected from wavelength and flux calibrations
for the eventual definition of new calibration strategies;
. evaluation of di#erent e#ects, such as atmospheric variations, di#erential
refraction, the impact of mechanical deformations on optical performance
and instrument maintenance;
. definition of procedures for quality control tests for EMMI data;
. estimation of CPU and hardware requirements to run the final pipeline;
. the completeness of the information stored in the archive, and the corre­
sponding requirements to be imposed to operations to ensure such com­
pleteness;
. requirements to be imposed to data processing to ensure that the inevitable
instrument instabilities are correctly recovered by the data reduction soft­
ware, both in pipeline and in interactive modes.
3. Pipeline Processing and Pre­calibrated Solutions
The MIDAS environment, and in particular its Echelle Package, include proper
tools for building pipeline DRS for generic echelle instruments, and in the future
for UVES, with some minor improvements. In the pipeline built for EMMI, pre­
calibrated solutions (i.e., tables containing the results of interactive processing
on selected ``reference'' spectra) were used for order detection and wavelength
calibration as starting points.

Pipeline Processing and Quality Control for Echelle Data 339
The UPDATE/ORDER MIDAS command was used to re­define the positions of
the orders starting from the pre­calibrated order table.
The wavelength calibration has been performed using IDENT/ECHELLE with
the GUESS method (i.e., searching for the optimal solution starting from a pre­
defined position of calibration lines). The result was accepted only after checking
the plot of residuals and the ``percentage of identification among the half brighter
lines'', that should be the highest, and in any case above 50%.
The analysis of applicability for the pre­calibrated solutions has verified
that, for a stable instrumental configuration and performance, the current pro­
cedures are quite adequate. Minor software upgrades could be made to take care
of extreme situations; e.g., for EMMI a non­rigid shift in order positions due to a
changed field lens was corrected with a custom modification of the UPDATE/ORDER
command. A new set of pre­calibrated solutions shall however be computed by
sta# whenever a manual intervention on the instrument is made; such interven­
tions shall be appropriately reported in the archive. New pre­calibrated solutions
should be computed during every test period, to minimize a­posteriori software
recovery.
4. Quality Control
For an e#cient quality control to take place, the setup of proper operational
procedures is necessary:
. at the telescope:
-- test frames must be regularly acquired (once during every test pe­
riod):
# bias and long dark frames are useful to monitor the detector
parameters and features;
# flat fields for di#erent instrumental configurations provide addi­
tional information on the instrument stability;
-- scientific data sets must contain all calibration frames necessary for
re­using the data and assessing their quality:
# the optical performance of the instrument can be monitored by
the analysis of the flat field stability and standard stars fluxes;
# a list of standard stars of a very early spectral type or with a
featureless spectrum should be selected in order to evaluate and
subtract the atmospheric contribution;
# the ThAr spectra allow the geometrical stability and the disper­
sion relation accuracy over time to be checked;
-- information on manual interventions must be regularly logged and
archived;
-- instruments should always be in a stable and documented state.
. at headquarters:
-- whenever a modification to an instrument occurs, the related process­
ing script (or parameters) should be updated;

340 Morgante, Pasian and Ballester
-- previous versions of processing scripts should be archived;
-- the test data should be promptly processed to produce a proper trend
analysis on instrument behaviour.
5. Archive Completeness
In this work, it was decided to rely uniquely on the contents of the archive
and of the keywords of the files, because this is the standard approach an ESO
user would take in accessing data he/she would have not acquired directly. In
accordance with the results of P’eron et al. (1994), some incompleteness in the
archive of EMMI echelle data has been identified. In the future, for EMMI but
especially for UVES, the following are key items for archive completeness:
. routine acquisition (e.g., in service mode) of all frames necessary for proper
reduction of the scientific data;
. indication in the archive or in the header of the scientific frame, or in
a dedicated Observation Summary Table (OST) of the calibration files
considered by ESO as the ones to be used for an ``observatory standard''
reduction of data;
. creation of data sets (one per scientific frame), including science data,
calibration frames (acquired with the science data, or extracted from the
archive), and the OST;
. correctness of the keywords in the frame headers.
Acknowledgments. The work has been carried out under ESO Contract
No. 47980/DMD/96/7513 to OAT. Useful information and valuable suggestions
were provided during meetings at ESO by H.Dekker, S.d'Odorico, H.Hernsberge,
R.Gilmozzi, P.GrosbÜl, L.Kaper, B.Leibundgut, P.Molaro, P.Nielsen, and L.Pa­
squini. Useful discussions with P.Bonifacio and S.Monai of OAT are gratefully
acknowledged.
References
Ballester, P. 1992, in Proc. 4 th ESO/ ST­ECF Data Analysis Workshop, ESO
Conf. and Workshop Proc. No. 41, 177
Ballester, P., Rosa, M. R. 1997, Astr. Astroph., in press, ESO Scientific Preprint
No. 1220
Morgante, G., Pasian, F. 1997, Feasibility Study of Pipeline and Quality Control
Software for Echelle Data ­ Final Report, OAT Technical Report
P’eron, M., Albrecht, M. A., Grosbol, P. 1994, in Proc. ESO/OAT Workshop
Handling and Archiving Data from Ground­based Telescopes, M.A.Al­
brecht & F.Pasian eds., ESO Conf. and Workshop Proc. No. 50, 57
Verschueren, W., Hernsberge, H. 1990, in Proc. 2 nd ESO/ST­ECF Data Anal­
ysis Workshop, ESO Conf. and Workshop Proc. No. 34, 143