Next: Locating Virtual Observatory Resources With the Astrogrid Registry
Up: Algorithms & Classification
Previous: A Prototype Publishing Registry for the Virtual Observatory
Table of Contents -
Subject Index -
Author Index -
Search -
PS reprint -
PDF reprint
Dolensky, M. & Doug, D. 2003, in ASP Conf. Ser., Vol. 314 Astronomical Data
Analysis Software and Systems XIII, eds. F. Ochsenbein, M. Allen, & D. Egret (San Francisco: ASP), 338
Survey among Spectral Data Providers and Consumers
M. Dolensky
European Southern Observatory, Karl-Schwarzschild-Straße 2,
D-85748 Garching, Germany
D. Tody
National Radio Astronomy Observatory, 1003 Lopezville Road, Socorro,
NM 87801-0387
Abstract:
The IVOA Data Access Layer (DAL) working group is defining a Simple Spectral
Access (SSA) protocol for accessing 1D spectra and SEDs. To help plan this
effort a survey was conducted to provide use-cases to guide the design.
This poster summarizes the results from the point of view of both the
data provider and the data consumer.
The survey shows that a wide variety of spectroscopic data is readily
available on-line. Data structures and qualitative aspects, however,
vary substantially. A preliminary analysis shows that the definition of a
common format for spectra is a big challenge. Therefore, the first step
is to concentrate on the access protocol and then to work on the formal
representation in a standard format based on a spectral data model.
The survey consists of two parts. The first concerns spectral data
providers, and the second software tools which might use the SSA protocol
to remotely access spectra. Such software tools are referred to as data
consumers in this paper. Below is a summary of the information gathered
from ESA IDC & XMM archives, ESO/ST-ECF, HEASARC, Hyperleda, INTA/LAEFF,
NAOJ/SMOKA, NCSA, NOAO and CFLIB project, OHP and SDSS.
This survey did not aim to be complete, rather the goal was to obtain a
large enough sample to capture the major aspects to help design the SSA
protocol and spectral data model.
The following spectral data collections were examined:
- Elodie: 25000 extracted Echelle spectra, resolution 42000, 4000 -
6800 Å on the fly processing options available: wavelength resampling,
flux calibration etc., total volume 50 - 100 GB
- ESO/ST-ECF, 250000+ observations taken with 14 spectrographs in UV,
optical, near IR; archive is rapidly growing
- HEASARC: Variety of spectral data sources from X-ray and gamma-ray
missions, including both relatively high and low resolution spectral data
- Hyperleda: 10000 reference spectra of stars and galaxies, resolution
300 - 1000000, 1300 Å - 2.5 microns total volume 5 GB, collected from
literature and reformatted
- Indo-U.S. Library of Coude Feed Stellar Spectra (NOAO & CFLIB)
1300 spectra, 3465-9469 Å @ 0.44Å/pix
- ISO 1D Spectral Data: 10000 files, 2.4-197 micron, resolution
40-30000
- IUE Newly-Extracted Spectra (INES): 110000 spectra, 9500 objects,
1150-1980 Å & 1850-3350 Å resolution 1.676 & 2.669 Å/pix
- NAOJ/SMOKA: 40000+ observations with 7 spectrographs in optical
and IR including Echelle and multi-slit detectors, 150+ GB volume
- NCSA Astronomy Digital Image Library - NCSA BIMA Data Archive
- NOAO: various high resolution atlases and libraries of spectra
mainly in the optical and IR
- Sloan Digital Sky Survey (SDSS): Data Release 1, 186.000 objects,
3800-9200 Å , 1300 square degrees, resolution 1800
- XMM-Newton 1D Spectral Data: XMM/RGS, 20000 spectra (expected),
0.35-2.5 keV, resolution 200-800; XMM/EPIC, 1.5 mio. spectra (expected),
0.35-15 keV, resolution 20-50
The characteristics of spectral data present in the above collections
may be summarized as follows:
- number: 100 K - 1 mio. spectra/data provider; often only several 100s
in very specific catalogs
- size: few KBs - few MBs/spectrum
- wavelength/energy ranges: gamma, X-ray, UV, optical, IR, radio
- resolution: 20 - 1.000.000 (solar spectra)
- linear and non-linear/irregular sampling
- WCS information
- stellar parameters: e.g.: Teff, log(g), [Fe/H]
- noise
- masks (coded aperture mask data, INTEGRAL)
- observation dependent transfer matrix to convert counts to physical units
- spectral features (SLOAN): emission, absorption lines and parameters,
emission redshift params. cross-correlation parameters
- time resolved spectral information (XTE, BATSE, Swift, HETE)
Columns/data items:
- wavelength (better fravergy: frequency, wavelength, energy)
- flux (magnitude, flux, flux density, counts)
- flux error
- quality flags
- variance arrays
- photon events
- antenna temperature vs. frequency (e.g. SWAS spectra)
Currently there is no widely used spectral data format, and each
data collection tends to store spectra in a unique format. The range
of formats used include the following:
- ASCII tables (catalogs, often highly processed data)
- Database tables (e.g. Sloan)
- FITS binary tables
- FITS images (often used for raw data (CCD images) or stacks thereof;
radio spectral image cubes)
Yes, data are generally on-line, but proprietary periods and restricted
access depending on nationality may apply.
Information about the following tools was collected: aXe, MIDAS-MOS,
NOAO IRAF-Specplot, Pleinpot, Specview, ISO Survey Products Display Tool.
Some tools dealing with SEDs are: NED on-line service, AVO Prototype
SED utility.
The spectral analysis tools examined by the survey include the following
capabilities:
- reading various instrument dependent FITS and ASCII formats
- overplotting of multiple arrays
- various display options (labels, zoom, colors, log/linear scaling, ...)
- physical unit conversion
- model fitting capability
- classification against standards
- rebinning
- quality filters
Respondents were asked to comment on the desired characteristics of
spectral data to be read by an analysis tool. The following items were
mentioned:
- As a minimum, data should consist of (x,y) pairs of wavelength-flux
values expressed in wavelength/frequency/energy units, and spectral flux
density units respectively. The units information should be present as well,
in header keywords or table column descriptors or any other suitable form.
- Additional information that could be included with the above, (if
available) is a third value associated with each (x,y) pair representing
the measurement error in flux density.
- Uncalibrated data (such as in counts/s units) can also be ingested
but not much else can be done with it besides plain plotting.
- FITS WCS; useful for well sampled rasters
FITS (images, tables), XML, and plain ASCII are all supported by various
tools. Some tools (e.g., Specview) support multiple input formats.
It is fairly easy to deal with multiple input formats so long as they
implement a similar data model. For quicklook purposes also graphic
formats are used, like the ISO IDC poststamp service.
Acknowledgments
Many thanks to all who contributed to the survey. Special thanks to
Ivo Busko, Mike Fitzpatrick, Satoshi Honda, Stephen Kent, Tom McGlynn,
Pedro Osuna Alcalaya, Benoit Pirenne, Ray Plante, Phillipe Prugniel,
Enrique Solano, Alex Szalay, Francisco Valdes and Andreas Wicenec.
© Copyright 2004 Astronomical Society of the Pacific, 390 Ashton Avenue, San Francisco, California 94112, USA
Next: Locating Virtual Observatory Resources With the Astrogrid Registry
Up: Algorithms & Classification
Previous: A Prototype Publishing Registry for the Virtual Observatory
Table of Contents -
Subject Index -
Author Index -
Search -
PS reprint -
PDF reprint