Motivation

The plate archive of the Hamburg Observatory has a stock of about 45,000 photographic plates taken from 1^st December 1911 until the end of the last century. So far this plate archive was not really usable for science due to the lack of digitally searchable meta-data.

The digitization of photographic plates taken before 1984 have never after the installation of a PDS G1010 Microdensitometer and all plates of the Hamburg-Quasar-Survey (HQS) and the Hamburg-ESO-Survey /HES) were digitized for both projects.

Nearly all the plates taken before 1984 have never been digitized or digitally analyzed. At that time a complete digitization of all the plates seemed to be impossible because of the huge amount of data (about 30TB) and the seemingly endless time required to scan such a vast collection.

Photographic plates are in so far the most long-lasting storage media (more than 100 years but with diminishing quality) compared to the round about 10 years life time of the modern reading devices.

Nevertheless, there are various good reasons to initiate a digitization just now (- worsening, + improving):

1. - The oldest plates display aging effects of the emulsion at higher densities. Due to inappropriate storage peeling of the emulsion and mold infestation were discovered and emulsions are sensitive to scratches and dust.

2. - The number of people who use photographic plates is decreasing rapidly. This means that the scientific interest in the plates will decrease, although each plate is a unique observation which is especially interesting if variability in time or position in the sky are observable.

3. + Costs for scanners and storage media are decreasing permanently.

4. + The long term storage of the digital data seems feasible at acceptable costs.

5. + The provision of data in the Internet becomes possible due to increasing bandwiths.

There are 3 groups of people with scientific interests for the data.

Astronomical research: Series of plates from the same field offer the possibility for a computer based search. Many plates have been taken just to analyze one object. Most other objects are discarded. Drastical changes in stellar objects such as outbreaks or the change of path of objects in the solar system will immediately increase the interests in former observations.

History of Science: The collection of plates together with the written information in logbooks or plate covers discloses a century of astronomical research, telling the story of observations, their problems, and the many new ideas to improve their quality. The aim of the Hamburg Observatory to become UNESCO World Heritage will increase the public interests in accessible data.

Amateur-Astronomy: The free accessible data will be a treasure trove for thousands of amateur astronomers who can now observe the same objects with smaller telescopes but powerful CCDs.

The combination of a data base and a web server which displays all available information as high- and low-resolution-scan plus plate cover, observer notes and logbook pages will enable the search for and the use of plates and objects.

Method and Catalog

Within the framework of a pilot project more then 3500 plates were scanned and the automatic creation of the web-pages has been developed.

The Epson Expression 10000 XL with transmission unit was chosen as scanner. It can measure densities up to 3.8. The pixel size of 10.6 micron gives a reasonable scan resolution of 2400 dpi which is comparable to PDS scans. The somewhat smaller density range compared to the PDS seems to be sufficient for the historical plates. The only disadvantage is the positional inaccuracy in scan-direction which can be overcome by an additional second scan after rotating the plate by 90 degrees. Such a second scan is only necessary if the scans should be used for astrometric purposes. Further advantages of the scanner are autofocusing and automatic calibration. The most important plus factor is the short scan time which allows to scan a length of 10 inch in about 7 minutes. The PDS will need more than 9 hours.

The total time for the scanning of plates, logbooks, and plate envelopes was estimated to be around 15 years for one person. By using several scanners and more manpower the duration may be decreased to a couple of years.

One problem in cataloging the plates is the large number of telescopes, cameras and plate sizes. The following directory structure is used:

Scans / direct photography / telescope / plate-size / [ jpegs | x-scans | y-scans ]

Scans / spectra / telescope / plate-size / [ jpegs | x-scans / y-scans ]

The names of the files are composed of 2 characters for the telescope initials (except Spiegelteleskop, S) plus 5 further digits.

Low-resolution-scans (.jpeg) are made in color if there are markings on the emulsion free side of the plate. Otherwise low-resolution JPEGs are calculated from the 16 bit gray-scale data of the high-resolution-scans. The high-resolution-scans are stored in FITS-format which is the standard data format used in astronomy.

A scan-PC stores the data temporarily on a local hard-disk. After customizing the data they are collected on a RAID6-system. Additionally, the data are stored on a distant tape robot at the university to ensure long term storage. The storage costs will drop with the time since every few years the double amount of storage capacity is available for the same price.

The meta-data are created from the information out of logbooks, observer notes and plate envelopes. They will be used for the search function of the data base. Beside the plate number further information such as coordinates of the plate center, object name, date and time of observation, exposure time, sky conditions, filter or objective prism are documented in table form.

The web pages of the catalog are first classified according to the telescopes as the quality of stellar images differs strongly between telescopes (e.g. plates taken with the 1 meter telescope are only usable in the central region due to large coma effects of the mirror).

Following the telescope-link tables will be shown which display the meta-data.

The plate number links to a web-page where all scans of this plate can be found.

The JPEG image of the low-resolution-scan enables a visual quality check. If desired high-resolution-data can be downloaded manually. The size of the FITS-data may be quite large, for example at 13cmx18cm 400 MB, at 24cmx24cm 1 GB and at 30cmx30cm 1,5 GB. For this reason the size is displayed in the download window.

User friendly search facilities using meta data are available in the archive section. The additive effect of search parameters offers the possibility to limit the search to small groups of plates with similar characteristics such as selecting for objective prism plates or direct plates with a special emulsion and/or specific filter types.

In addition it is possible to study the observer notes or the logbooks with a reader. Once an interesting entry is found it is possible to jump directly to the web page of the corresponding plates.

Linkage and Identification

In most cases the plate number is used for identifying plates and envelopes. Nevertheless, there are exceptions as for example with the plates of the Great-Refractor. In this case several series exist with plate numbers starting with "1". These plates are sorted by date and small letters are assigned replacing the leading zero(s). Another issue is that not all observer notes are available for a telescope, consequently only information given on the plate or the envelope can be used. In some rare cases even the date or the object is missing.

Even more complicated is the identification of spectral plates from the 1m-Mirror-Telescope. Here the plate numbers start with letters such as 'DS' or 'KS'. In the latter case 'KS' is already used as telescope code for 'Kleiner Schmidt-Spiegel'. These IDs are transformed as follows: KS --> SK or DS --> SD. This is not an elegant solution but a necessary approach since each identification number must be unique.

In cases where pages or envelopes do not carry any information, they are omitted.

The logbooks are a register for the plates, comparable to the present-day meta data. They were handwritten, mostly not by the observer. In cases where preprinted tables were used it was possible to clip the appropriate row automatically and to have it linked to the respective web page in addition. In all other cases, wherever possible only the whole pages are linked.

General comments on photographic plates

Photographic plates are non-digital media and are even after digitizing not linear in respective to the density values compared to CCD-frames. At the time they were exposed special plates were exposed with spots or stripes (in the case of spectra) with defined brightness scale, partly already on the plate itself. Without such a scale a brightness calibration would be impossible. Nowadays it is possible to calibrate the plates in retrospect due to on-line available catalogs with millions of stars with known brightness and color gradient. Such a calibration may also be possible for objective prism plates. However, single spectra cannot be calibrated without available calibration spots.

Errors

Inspite of great care some meta-data entries may be faulty. Through the online publishing of all handwritten information together with the meta-data it is always possible to recheck their correctness. In case any discrepancy or incorrectness is detected, please inform the project leader by e-mail (This email address is being protected from spambots. You need JavaScript enabled to view it.).