Background
The International Ultraviolet Explorer was launched in 1978 to observe the universe in the 1150 - 3200 Å region. The IUE mission was based on real time control of the satellite and instrument and real time acquisition of the scientific data. Expected to have a mission lifetime of three years, IUE functioned for over 18 years, generating greater than 107,000 ultraviolet spectra.
The original processing of IUE data was done with the IUESIPS (IUE Spectral Image Processing Software) software. Over the course of the mission several changes were made to processing algorithms and calibrations, which made comparisons between epochs difficult. Because of this, it was decided to re-create and homogenize the entire IUE archive using the new algorithms and calibrations of the NEWSIPS software system. The resulting archive is known as the IUE Final Archive.
Objective
IUESIPS and NEWSIPS data are archived at the Multimission Archive at STScI (MAST) and at the NASA Space Science Data Coordinated Archive (NSSDCA) at the NASA Goddard Space Flight Center. Nearly all IUE data (IUESIPS and NEWSIPS) reside safely on current media. As such, the NSSDCA is reviewing the possible release of approximately 2000 original IUE 9-track tapes.
This paper addresses the few IUESIPS file types on original tapes but not currently residing on recently created media held at MAST and NSSDCA. We examine the costs of recovering these files, and the perceived value of these file types. Our goal is to make sure that the NSSDCA carries the data sets which, together, hold all of the IUE science potential. However, we also want to be cognizant of any "rescue" effort whose potential benefit is outweighed by the cost of the effort.
Current holdings
The NSSDCA holds approximately 2,000 9-track tapes containing raw data and IUESIPS-processed spectroscopic image data. There are 1,423 NASA data tapes and 540 VILSPA data tapes. The oldest of the tapes are approximately 20 years old, the youngest approximately 3-4 years old (due to tape replacement). There are a corresponding number of backup tapes in addition to the original tapes. The original 9-track tapes have been read over the entire lifespan of the data set. The last use of each of the original 9-track tapes ranges between two and ten years ago.
Additionally, the MAST and NSSDCA archives hold the IUE data set in collections of Digital Linear Tapes (DLTs) and/or CDs. Both entities hold all raw data, all NEWSIPS data and most IUESIPS data on these media, but not the IUESIPS files unique to the 2000 9-track tapes.
Files in question and their scientific value
We have investigated the nature of the IUE files not included on IUESIPS DLT/CD (i.e. held only on the 2000 original tapes), and their importance in the scientific usability of the combined IUE data set. The IUESIPS-processed contents not included on either MAST or NSSDCA DLTs/CDs are 1) photometrically corrected image (PI) files generated prior to the last 3 - 5 years of IUE life, 2) most of the Fine Error Sensor (FES) image files, and 3) earlier versions of IUESIPS-processed data files.
Photometrically Corrected Image files. The photometrically corrected (PI) files are products of the first step in IUESIPS processing. Photometric corrections were applied to the raw images obtained from IUE spectral imaging cameras. These corrections were applied to offset non-uniformity and non-linearity of the detector response. Photometric correction involves conversion from DN (data numbers; raw pixel values) to FN (normalized flux numbers) using an intensity transfer function (ITF). The corresponding NEWSIPS-processed files are the linearized images (LI). However, the photometric correction procedure performed in NEWSIPS is fundamentally different from the correction performed in IUESIPS.
There are groups in the user community involved in the re-extraction of IUE spectra. Most users prefer to begin the re-extraction from the LI files (for NEWSIPS processing) or the PI files (for IUESIPS processing), rather than the raw data files, because of the time involved in generating ITFs and correcting for geometric distortions. Most groups are working with NEWSIPS data, for which the PI files are largely incompatible. The UK IUE Project is the primary group re-extracting spectra using PI files, which are required in the STARLINK IUE data reduction (IUEDR) program. The UK IUE project accesses its own PI files.
Well over 95% of IUE requests are for extracted data and not the first-level processing files (i.e. the PI or LI files).
Fine Error Sensor files. Two of the six IUE cameras were Fine Error Sensors. The Fine Error Sensor cameras were low-resolution image dissector devices, with an effective wavelength of 5200 Å and a response between 4000 - 7000 Å. The FES cameras were used to obtain a visual image of the star field to identify a target, and then to track the observation through offset guiding. The FES images were occasionally taken, especially in complicated fields, as a documentation of how the observations were actually done. There are about 1,750 FES images archived on the original 9-track tapes. These FES images were archived at the request of the Guest Observer, typically to save certain header file information during handovers from one tracking station to another. When archived, the FES images were copied to tape in raw form, and not processed (i.e. not calibrated). The low-resolution data returned from the FES cameras contain no spectral information. Due to improper archiving techniques, the VILSPA FES files have formatting problems. The FES images are not reprocessed as NEWSIPS data.
Earlier versions of reprocessed data
Recall that several modifications to IUESIPS software, over the lifetime of the IUE mission, resulted in several versions of reprocessed data for some observations. Reprocessing was done upon request, and for the purpose of having a set of uniformly processed data. Requesters of IUESIPS data are currently distributed (from MAST) the latest versions of re-processed IUESIPS data.
Scientific usability
The data files in question are all scientifically valid, however, representative members of the IUE community and MAST have indicated that the files' innate scientific usefulness is low to minimal. Approximately 99% of IUE data analyses use NEWSIPS extracted files. The main value of the PI files is their use in programs to re-extract spectra, which is of interest in determining a better way to extract spectra from IUESIPS data. The primary role of the FES files is that of an extra record of the matching observation.
MAST made a decision not to re-archive the data files in question because the effort to do this outweighed the level of demand for those files and the files' scientific potential. MAST had expressed an interest in keeping these files in the deep archive under the assumption of adequate funding and resources.
Recreatibility of PI and FES files
The PI files cannot be reproduced with available software. MAST continues to support IUE data analysis software, however IUESIPS pipeline processing software (software for creating PI files) is not archived at MAST. The intensity transfer functions (ITFs), used in the photometric corrections, were not archived either. In order to re-create PI files, intensity transfer functions would have to be re-generated. If the ITFs are re-generated, the resulting PI files would not be the exact same files as the originals.
FES files are raw data, and therefore cannot be re-created.
Estimated rescue costs
The 9-track tapes, on which the original IUESIPS data is currently held, are deteriorating. If the files in question are to be maintained, it is necessary to move the data on the 9-track tapes to newer media. The NSSDCA has examined various recovery efforts. The simplest effort is to copy the tapes in their entirety onto new media. Another effort is to extract all but the raw data, as the raw data is captured elsewhere. The FES files, which are raw data, would be captured as well. This has the benefit of reducing the physical size of the holdings by about fifty-percent.
The NSSDCA 9-track tapes are in IBM VBS format, and must go through processing software before they can be read onto any other system. There is a desire to move away from IBM-binary format to a vendor independent format such as standard IEEE-binary.
In order to retrieve all data from the 9-track tapes, both original and backup tapes may have to be read. This is the biggest time/cost factor. The files in question can be identified during the spinning of the tapes. The cost estimate to move the data is about $35,000, which includes media costs and programming costs. This estimate is regardless of how the formatting is done.
NSSDCA's present intent
Our research into the nature of the data files in question, and costs of data rescue, has led us to conclude (and MAST has agreed) that the data files in question cannot be recovered in their entirety from the original tapes at a cost justified by the potential value of the files. The $35,000 in question can be used more effectively elsewhere at NSSDCA. Thus, we intend to release the 2000 original tapes and their backups.
The community's feedback
This memorandum is intended to stimulate reasonable arguments in support or non-support of the NSSDCA's intent to release the original IUESIPS data residing on approximately 2,000 antiquated 9-track tapes. We solicit your concerns and/or recommendations prior to November 2000. Please direct comments to Dr. Ed Grayzeck, Head, NSSDCA at edwin.j.grayzeck@nasa.gov.
Questions or feedback about this page or astronomy and astrophysics data/services at the NSSDCA?
Please contact:
Dr. Ed Grayzeck, edwin.j.grayzeck@nasa.gov, +1-301-286-7355
Code 690.1, NASA Goddard Space Flight Center
Greenbelt, MD 20771, USA