Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stecf.org/conferences/adass/adassVII/reprints/lahuisf.ps.gz Äàòà èçìåíåíèÿ: Mon Jun 12 18:51:47 2006 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 03:28:15 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: 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.
ISO--SWS Data Analysis
F. Lahuis 2 , E. Wieprecht 3 ,
ISO Science Operations Centre, Astrophysics Division, Space Science
Department of ESA, Villafranca, P.O.Box 50727, E­28080 Madrid,
Spain
O.H. Bauer 3 , D. Boxhoorn 2 , R. Huygen 4 , D. Kester 2 , K.J. Leech 1 , P.R.
Roelfsema 2 , E. Sturm 3 , N.J. Sym 2 and B. Vandenbussche 1,4
Abstract. We present aspects of the Data Analysis of the Short Wave­
length Spectrometer (SWS) on­board ESA's Infrared Space Observatory
(ISO). The general processing from the raw telemetry data of the instru­
ment, up to the final calibrated product is described. Intermediate steps,
instrument related aspects and data reduction techniques are highlighted.
1. The SWS Instrument
The Short­Wavelength Spectrometer (de Graauw et al. 1996) is one of the four
instruments on board ESA's Infrared Space Observatory (Kessler et al. 1996),
launched on November 17, 1995 and operational till approximately April 1998.
SWS covers the wavelength range from 2.38 to 45.2 µm with two nearly inde­
pendent grating spectrometers with a spectral resolution ranging from 1000 to
2000. In the wavelength range from 11.4 to 44.5 µm Fabry­Perot filters can be
inserted which enhance the resolution by a factor 10 to 20.
2. SWS Data Processing
The processing of the SWS data can roughly be split into three distinct parts.
The first part is the processing from the raw telemetry data to Standard Pro­
cessed Data (SPD). The second part starts with the SPD and results in a spec­
trum of wavelength versus flux, the Auto Analysis Result (AAR). The last stage
1 ISO Science Operations Centre, Astrophysics Division, Space Science Department of ESA,
Villafranca, P.O.Box 50727,E­28080 Madrid, Spain
2 Space Research Organization Netherlands, Postbus 800, NL­9700 AV, Groningen, The
Netherlands
3 Max Planck Institut f˜ur extraterrestrische Physik, Giessenbachstrasse, 85748 Garching,
Germany
4 Katholieke Universiteit Leuven, Instituut voor Sterrenkunde, Celestijnenlaan 200 B, B­3001
Heverlee, Belgium
224

ISO--SWS Data Analysis 225
involves post­processing and scientific analysis using the AAR. The first two
steps form part of the ISO Standard Product Generation `pipeline' and the
products mentioned (e.g., ERD, SPD and AAR) form part of the data sent to
the ISO observer. This processing is also part of the SWS Interactive Analysis
System IA 3 which is described in detail by Wieprecht et al. (this volume).
2.1. ERD to SPD
The raw data are stored in FITS files containing the Edited Raw Data (ERD) and
status and house­keeping parameters. Additional files containing information on
e.g., the pointing of the satellite and the status of the instrument and satellite
exist and can be used as well if required.
The ERD data consist of digital readouts at a rate of 24 per second for each
of the 52 SWS detectors (four blocks of 12 detectors for the grating and two
blocks of 2 detectors for the FP). The data is checked for saturation, corrected
for reset pulse after­e#ects, linearized (AC correction), cross­talk corrected and
converted to voltages. Next a slope is fitted for each detector and reset interval
and wavelengths are assigned for each of these points. The data are extracted
per reset interval and stored in a SPD.
2.2. SPD to AAR
For each reset interval the SPD contains a value for the slope in µV/sec and a
wavelength in µm plus ancillary information about timing, instrument settings,
data quality etc.
As a first step memory e#ects should be corrected. For the time being this is
still in an experimental phase and not included in the standard processing. Next
the dark currents are determined and subtracted. The relative spectral response
of the instrument (RSRF) is corrected and the absolute flux calibration is applied
(i.e., conversion from µV/sec to Jy). The wavelengths are corrected for the ISO
velocity towards target and finally the data is sorted by wavelength, extracted
and stored in an AAR.
2.3. AAR Processing
The processing using the AAR involves steps like aligning the signals of di#erent
detectors which cover the same spectral region, sigma clipping, rebinning, line
and continuum fitting etc. All this is possible within IA 3 and the ISO Spectral
Analysis Package (ISAP) (Sturm et al. this volume).
3. Edited Raw Data (ERD)
Figure 1 shows ten seconds of data from one detector. The data is taken with a
reset time of two seconds. This means that we have 48 data­points per reset per
detector. This plot illustrates the correction steps required at this stage of the
processing. The two main corrections are the correction of the ramp curvature
and the correction of glitches, caused by the impact of charged particles.
One source of curvature is the after­e#ect of the applied reset pulse (here
at 0,2,4,6, and 8 seconds). This e#ect can be fit with a decaying exponential and
can be determined from the data (this is required since the pulse­shapes appear

226 Lahuis et al.
0 2 4 6 8 10
time [second] wrt the start of the observation
1980
1990
2000
2010
2020
2030
2040
Signal
[Bits]
(16) Si:Ga 4 Figure 1. Sample of Edited Raw
Data (ERD) data illustrating the
curvature in the uncorrected slopes
and an example of a glitch (in the
fourth slope) as a result of a charged
particle impact.
to change from observation to observation). The pulse­shape is an additive term
to the slope and thus introduces an o#set to the calculated signal. Since this
a#ects all slopes in a similar manner it has no direct impact on the signal since
it will be corrected by the subtraction of the dark currents. The main reason for
correcting for the pulse­shape is to maximally linearize the slopes. This helps to
improve the detection and correction of glitches. The other source of curvature
can be found in the electronics in the form of a high­pass filter. The time constant
for this is known to a fairly high accuracy and experiments have shown that this
time constant appears to be fairly constant from observation to observation (for
most detectors less than 5­10%).
In the hostile environment of outer space, there is a constant impact of
charged particles. These show up in the data as a sudden increase in the raw
detector signal (as illustrated in figure 1). These are recognized, corrected and
flagged in the data. The flagging gives the user the ability to check the validity
of a#ected slopes in a later stage of the processing.
4. Relative Spectral Response
An important step in the processing from SPD to AAR is the application of
the Relative Spectral Response Function (RSRF). Part of the long wavelength
section of the SWS (from 12 µm to 29 µm) the spectrum is highly a#ected by
instrumental fringing as a result of Fabry Perot e#ects within the instrument.
Directly applying the RSRF to the data quite often gives unsatisfactory
results. The fringe residuals are significantly large and this limits the detection
and analysis of weak spectral features. The main reason for this is that the RSRF
has a limited accuracy, the fringes as they appear in the data are sometimes
shifted with respect to the fringes in the RSRF when the source is not exactly
in the center of the aperture or if the source is extended. Also the width of the
features (i.e., the resolution of the data) can change depending on the extent of
the source.
A number of methods have been developed within IA 3 to adapt, change or
correct the e#ects mentioned above. These can reduce the residuals down to a
level of 1--2%. The following techniques can be applied:
# shift correlation/correction between the RSRF and the data
# adapting the resolution of the RSRF
# fitting cosine functions to the RSRF corrected data
# Fourier filtering of the RSRF corrected data
# modeling the instrumental FP e#ects

ISO--SWS Data Analysis 227
13.4 13.6 13.8 14.0 14.2
1500
2000
2500
Signal
[lV/s]
13.4 13.6 13.8 14.0 14.2
600
700
800
900
Flux
[Jy]
13.4 13.6 13.8 14.0 14.2
Wavelength [microns]
660
680
700
720
740
760
780
Flux
[Jy]
C2H2
HCN
Figure 2. The impact of fringes in
part of the spectrum. In the top plot,
the signal from one of the twelve de­
tectors covering this spectral range is
shown. Overplotted with the dashed
curve is the scaled RSRF signal.
The second plot shows the result for
all twelve detectors as produced with
the standard processing: large fringe
residuals remain. In the last plot
the result of improved data reduc­
tion is shown. The fringe residuals
are low and allow detection of weak
spectral features. Here the detection
of the C 2 H 2 # 5 and the HCN # 2
vibration­rotation bands towards a
deeply embedded massive young star
are shown (Lahuis et al. 1998).
Figure 2 shows an example utilizing two of the techniques mentioned here. In
this case first a shift correction and enhancement of the RSRF was applied. After
this the residual fringes were removed by fitting cosine functions.
5. Future Developments
The data analysis of SWS has not reached its final stage. The knowledge about
the instrument is continuously increasing and consequently the software is evolv­
ing. After the satellite has ceased to function a post­operations phase starts in
which the improvement of the calibration and data­reduction of SWS will con­
tinue.
References
de Graauw, Th. et al. 1996, A&A 315, L49
Huygen, R & Vandenbussche, B., 1997, in ASP Conf. Ser., Vol. 125, Astronomical
Data Analysis Software and Systems VI, ed. Gareth Hunt & H. E. Payne
(San Francisco: ASP), 345
Kessler, M.F. et al., 1996, A&A 315, L27
Lahuis, F. et al., 1998, in First ISO Workshop on Analytical Spectroscopy with
SWS, LWS, PHT­S and CAM­CVF,ESA SP­419, European Space Agency
Leech, K. et al., SWS IDUM, SAI/95­221/Dc
Roelfsema, P.R. et al., 1993, in ASP Conf. Ser., Vol. 52, Astronomical Data
Analysis Software and Systems II, ed. R. J. Hanisch, R. J. V. Brissenden
& Jeannette Barnes (San Francisco: ASP), 254
Sturm, E. et al. this volume

228 Lahuis et al.
Wieprecht, E. et al. this volume