Astronomical Heritage, and Data At Risk

Elizabeth Griffin, Herzberg Institute of Astrophysics, Victoria, BC
A myriad of different topics nestles under the "Heritage" umbrella. The assortment ranges from things written (books, papers, literature), things created or constructed (fine arts, sculpture, architecture), the way people lived and thought (culture, history, philosophy), to the erudite things that they built and applied. Some historic observatories and sites are to become formally recognized in UNESCO's list of World Heritage Sites. However, what are considered worthy of distinction as "heritage" tend to be things that are concluded or finished, having "value" because they represent a past which has shaped our own present. It is relatively rare to come across "heritage" that is actually part of that present, but astronomy has good examples.
Accounts of how decisions were reached in astronomical policy, and their influences on developments and current achievements, are recorded in various writings. The most detailed are entombed in (auto)biographies, though there may be a tendency by the author to tread carefully and to step with diplomatic daintiness through areas muddied by differences of opinion. Some years ago the Heritage Committee initiated a project to video interviews of retirement-age Canadian astronomers who would have participated in influential decisions. The aim was to capture the "inside" view, permitting a proprietary period before the recording may be released. Although diffidence and diplomacy have continued to colour what could be saved in hard copy, even at this stage, our growing collection of tapes and videos is a unique library of personality portraits, and includes impressions and opinions that have not been formally voiced before.
Human memory is evanescent, and the Committee is striving to record what it can before it is too late. That sad but inevitable characteristic is shared with another aspect of astronomical heritage: historic observations. Observing has a rich past - in about 3 million pieces, clumped worldwide in observatory plate stores. Those observations are in a support hinterland, being neither artefacts of creative value, nor accessible to today's astronomers as complements to modern research. What the Heritage Committee supports is the policy to keep awareness of those data on the front burner, both internationally and particularly in Canada, and to recognize efforts to preserve their information digitally.
Observational Data at Risk
Astronomy's historic observations, some dating back over a century, constitute an irreplaceable resource of unique information on periodic, serendipitous and one-time events. Without data from the pre-digital era, many variability studies are seriously impoverished, or simply impossible. Canada (at the DAO) is particularly rich in historic spectra, but - in common with most other photographic archives - the observations are not digitized, and are thus not easily accessible for research.
The DAO has all the required expertise to set up an efficient scanning laboratory that can, within a few years, create digital versions of many of its heritage observations and place them in the public domain. The scientific case is very compelling, the requirements very modest. An expenditure of only tens of $K to upgrade two still serviceable purpose-built scanners, plus the appointment of two technical staff to do the scanning, seems a small price for realising the full potential of those heritage data. It is no argument that new equipment is more "important" because it can access new reaches of the cosmos. Heritage data extend backwards in time, a feat which today's equipment can never do. They can enable instant solutions regarding variabilities that have periods of decades, and they have captured events which may never recur.
However well the plates in any archive have been stored. time is beginning to run out at an accelerating rate. Plates degrade, physically and chemically. Those are monotonic processes; the emulsion becomes brittle with age, and lifts from its glass substrate. Incomplete rinsing of the chemicals during development accelerates the build-up of background fog, destroying the fidelity of the photometric information. Breakages can always occur. Maintaining a fairly low temperature and relative humidity are important; incorrect storage conditions speed deterioration. A growing problem is the declining availability of expertise concerning photographic plates. Replacing the photographic emulsion with the CCD as the workhorse detector created a polarity between born-digital and inherently non-digital observations; today's students are not required to know anything about photographic techniques, so expertise is not being replaced as key people retire. An even more insidious challenge is the hazard of human ignorance. When lack of access becomes misinterpreted as disinterest, the collections are at the mercy of edicts to throw them away. Here is one area where the Heritage Committee, CASCA, and the astronomical world at large can be proactive immediately.
New Science: Challenging the Digital Divide
Time-domain astronomy, now a burgeoning field, is badly constrained by inability to access heritage data. Short-term variability, such as can be covered adequately with data spanning just a few years, has become hot science, driving the need for major legacy surveys and for data-mining techniques to match. Innovation and capability are being revolutionized through the ingenious products of the Virtual Observatory. But how many (Canadian) astronomers can quickly access on-line data recorded more than ~ 25 years ago? Did variability science only commence with born-digital data? How much science is still waiting to be discovered?
What "new" science can "old" observations reveal? A long time-base can be essential for deriving accurate periods from radial-velocity or photometric variations, measuring precise proper motions, defining cluster membership, identifying relatively slow spectrum changes in advanced AGB stars due to actual evolution, or recognizing stellar multiplicity through period modulations. A time-base of 70-100 years is adequate for capturing the pre- and post-event states of the helium flash; to date a pathetic sample of only three stars (FG Sge, V605 Aql and Sakurai's object) have been serendipitously caught in the act, but who knows how much more cryptic evidence our heritage archives contain?
Over 70 publications can be cited in which major or subordinate use has been made of photographic plates in research since 2000. Most involve direct plates, because very few spectra are available on-line, but they demonstrate how use is made eagerly of whatever on-line resources are accessible. The recent scanning of a few hundred plates of M44 from the Harvard collection (about 1% of its total) revealed hundreds of new variables with periods of the order of 20 years and amplitudes ~ 0.5-1 mag. A few of the spectra also indicate quite bizarre objects, hitherto simply unrecognized as such. No-one knew that those stars were variable, and no-one yet knows why they vary as they do. Such information is foundational to astrophysics, and can only be extracted from appropriately long time-domain data.
Historic observations have also been re-analyzed for astronomical and transdisciplinary studies which are far removed from the purposes of the original observers: extracting information about the Earth's ozone at a time before dedicated ozone spectrographs were widely in use, or examining evidence of other atmospheric constituents and pollutants, are only two examples of what has been attempted. A programme to rescue and preserve digitally our historic resources can thus benefit other science disciplines.
No-one knows for sure all that is waiting to be discovered, and without on-line inventories of the archives we cannot even hazard a guess. Only when some very pressing need arises (e.g. to study the progenitor of a supernova, refine the orbit of a suspected near-Earth or other solar-system body, or search for optical counterparts of bizarre objects) do researchers turn to the almost forgotten hoards of plates in the basement - only to realize that the necessary digitizing instrumentation is no longer in service (or no longer exists) and there is no-one around who can help. As with the memories of those once central to key political decisions, human knowledge about plates is a shrinking commodity that needs to be tapped before it is too late.
While photographic collections are by far the largest category of astronomy's data at risk, records on magnetic tape or in private collections, often in non-standard formats and lacking meta-data, are similarly vulnerable. Hand-written log-books - often the only source of the meta-data which are essential for variability research - are likewise at risk from deterioration, fire or water damage, as well as possible mid-handling. Those, plus card-index lists, constitute the only inventories of our heritage which we presently have, and need to be converted into searchable catalogues so that any astronomer (including the archive owners!) can discover what plates are actually there.
A Future for our Past?
Science-driven initiatives are starting up around the world to push for preserving and digitizing selections of astronomy's photographic heritage, though the scanning giants like the APM (IoA), SuperCOS (ROE) and StarScan (USNO) have been closed down through lack of institutional support. In Canada the PDS scanner at the DAO is the only machine of its calibre presently available in the whole of North America for digitizing photographic spectra. Add to those losses the ageing of people with the necessary skills, and the challenge takes on a growing urgency.
To prepare a digital database of actual spectra requires the skilled use of an instrument such as a PDS (or similar microphotometer) that was specifically designed for the job. Desk-top scanners are usually not appropriate substitutes for quantitative scanning, regardless of what is claimed for the top end of the range. Tests have shown that positional fidelity can vary, increasingly so off-axis, while photometric fidelity cannot in principle be fully reliable on account of scattered light, the principal effect of which is to reduce the dynamic range. Positional accuracy is of greater concern when digitizing direct plates for astrometry, but good photometric accuracy is a sine qua non for spectroscopy whenever line-profiles are to be measured.
The DAO Collections
The world's stores of astronomical plates totals about 3 million; some 2 million are images and objective-prism spectra, the rest are slit spectra. The latter are roughly divided bimodally, peaking at high-dispersion for research-quality studies of individual stars, and low-dispersion for classification purposes. Canada's largest (and the world's second largest) archive of photographic slit spectra is at the DAO. It contains ~ 16,000 coudé plates commencing in 1962, and > 93,000 Cassegrain spectra dating from 1918. Individual spectra can be hunted down through a fairly convoluted paper-chase, but very little meta-data or actual spectra are yet available on-line (though they can now be ordered under a contractual arrangement). The Dominion Observatory's collection of 17,000 low-dispersion spectra from 1905-1937 is also housed at the DAO, but unfortunately the log-books prior to 1915 cannot be found. As a research resource the DAO's heritage observations are impotent, and will remain so until they have been accurately converted into modern digital formats.
The 1.8-m Plaskett Telescope has been in use, since its inauguration in 1918, by staff members, Canadians and visiting researchers. Prevailing protocol required that plates taken away for analysis be returned eventually, so the collections are reasonably complete. On both that telescope and the newer, dedicated-coudé 1.2-m one, variability studies have been pursued relentlessly over the decades by countless observers, resulting in long time-series of the spectra of many objects. At the time, most of those spectra will have been measured manually for just one parameter (the stellar radial velocity), leaving possibly unique astrophysical information contained in the strengths and profiles of the spectroscopic lines largely untouched. Between them, those plates constitute a veritable gold-mine of spectra of variable stars in particular, and of many other objects besides.

By way of illustration, Fig. 1 shows a digitized version of the first spectrogram exposed at the 1.8-m telescope in 1918; Fig. 2 shows a digitized spectrum of DI Her recorded with the same instrument 30 years later. Fig. 3 is a superposition of two exposures of e Aur made with the 1.2-m McKellar telescope, illustrating changes between inside and outside total eclipse. The spectra have all been intensity calbrated.

Fig. 1 shows a digitized version of the first spectrogram exposed at the 1.8-m telescope in 1918
Fig. 2 shows a digitized spectrum of DI Her recorded with the same instrument 30 years later
Fig. 3 is a superposition of two exposures of e Aur made with the 1.2-m McKellar telescope, illustrating changes between inside and outside total eclipse. The spectra have all been intensity calbrated.

 

In common with almost every other astronomical plate store, the DAO has no digital inventory of its plates, so we cannot easily discover which objects have been observed, when, or at what resolutions or wavelengths. We do have all the log-books, so in principle the crucial meta-data can be retrieved, albeit in only as much detail as the observers entered.
A Spectroscpic Plate-Scanning Laboratory
The DAO is presently the only institution in North America to possess a working PDS capable of scanning spectra accurately, and the only institution in the world actually supporting such a programme. Its PDS is receiving only slender institutional support, and its present performance is passable but flakey. It is currently operated by a volunteer, partly as a service for outside users. The situation is clearly unsustainable, and is no basis for the wholesale scanning of plate archives which time-domain astronomy requires.
One machine alone is inadequate to the task of digitizing the entire DAO plate collection within a reasonable number of years, and needs two other PDS instruments to be upgraded and operated in parallel. The necessary upgrading tasks are well understood, not expensive, and will take only a few months under available supervision to implement. The laboratory will also need part-time technical support. `Pipelines' have already been created for calibrating and converting the digital information rapidly and efficiently into 1-D spectra.
Preparing a searchable public inventory of the DAO plate archive is the first step to undertake. Transcribing the entries can be carried out by (e.g.) students or retirees, under supervision. Then, once the scanning machines are all working correctly and up to speed, the photographic spectra in the DAO archives can be digitized, reduced and placed in a public domain within a few years, an activity needing a couple of full-time research assistants. The finished data sets can be made available through, and managed by, the CADC.
It is important, both for our scientific history and for time-domain astronomy, that the new decade come to grips with this challenge before it is too late and the raw information becomes lost through natural ageing or destructive actions. In the whole of North America only the DAO has the equipment, the expertise and the skills to carry out an appropriate digitizing programme of stellar spectra. By placing the digitized spectra in the public domain, the modest amount of resources absorbed will pay dividends of unforeseen value across astronomy. It will also establish Canada as pioneer and leader (which it already is, but unofficially) in this exciting new field. Given the potential to set a valuable precedent, it will in fact be to our public discredit if we fail to carry out this very modest but much needed project during the next 5 years.


File translated from TEX by TTH, version 3.40.
On 08 Mar 2010, 22:59.