Ligue des Bibliothèques Europeénnes de Recherche, Groupe des Cartothécaires de LIBER

TRANSLATE ENGLISH to Chinese, French, German, Greek, Italian, Japanese, Korean, Portuguese, Russian, and Spanish!!!

From map catalogue to database with graphics in view

Barbara A. Morris, R.G. Healey and T.C. Waugh, Department of Geography, Edinburgh University

© LIBER and author
Published from: Bulletin Ligue des Bibliothèques Europeénnes de Recherche, 28(1986)

One of the most fundamental problems faced by map librarians is the need to store details of series maps as over 80 % of all mapping is published in sets or in series form. The coverage of series has traditionally been recorded on paper index maps. There is a real need in map libraries to have a flexible retrieval mechanism, based both on the attributes of individual sheets and on spatial location or window.
At present it is very difficult to hold details in machine-readable form of maps which are part of a series. The MARC format for exchanging machine-readable catalogue entries provides only the Level 1 entry which covers only the series heading although the British Library is working on modifications to record sheet and edition at Level 2 (Fig. 1).

No method of using catalogue entries to automatically update index diagrams has hitherto been made fully operational, although several organisations including ourselves have been investigating the problem.
The problem has three major components:
  1. The need for a highly efficient and flexible computer database system for storage and retrieval of map information.
  2. The need for a comprehensive computer mapping and graphics system to allow the automatic generation of index maps.
  3. The need for a well-designed software interface between the data retrieval and mapping programs to exploit the full capabilities of each, at the same time providing facilities for drawing index maps on both graphics terminals and graph plotters.
A solution to this problem has been devised in Edinburgh by linking two well-tried commercial software systems. We used ORACLE to create a cartographic information system. ORACLE is a very powerful fourth generation relational database system which is available on a number of different computers. We then linked ORACLE with GIMMS, the world's most widely used computer mapping system. A software interface has been written by Tom Waugh using these two packages so that when the reader or librarian addresses a query to the ORACLE database, GIMMS facilities are automatically called to draw an index map which fulfils that specific query. This system is currently running on the Department of Geography's VAX 11/750 computer in Edinburgh. Computer-generated graphic indexes of map coverage can be displayed interactively on a RAMTEK colour graphics terminal, a TEKTRONIX or BBC 'B' microcomputer or on our HEWLETT-PACKARD 7580A Plotter.
We have given several demonstrations of the system in operation. As a result we have been asked by the British Library to experiment with one of their existing paper indexes. Many libraries will have series indexes which have become congested and almost illegible as new sheets are acquired and additional information about EDITION and PRINTDATE added. These indexes cannot give a clear picture of the pattern of publishing for the series or a particular area. If an automated graphic index system such as the ORACLE/GIMMS System were used, the problems of stock control would be considerably eased, both in large national map collections and in university libraries.
The illustrations which follow are examples of the way in which stock of series mapping can be controlled by this system. The method is equally valid for locating the spatial distribution of individual maps.
Fig. 2 is copied from a British Library index of a 1:100 000 series of Germany.
The difficulty of reading the PRINTDATES and EDITION numbers for each sheet is obvious. It is not possible to scan this index to get an overall picture of the revision pattern. Nor can it be easily compared with any other series of maps of the same area.
Our digital outline contains several files. The first one gives an outline of the coast and international boundaries of Europe. Where more detail is required on the index background we can add rivers, country names, capitals and other towns. Catalogue information was loaded into the ORACLE Database System to include the UK MARC Level 1 series entry. At level 2 (sheet level) SHEET NUMBER, EDITION STATEMENT and PRINTDATE were recorded. For the purposes of graphic retrieval the location of the sheet corners was also recorded. A query to ORACLE to display all maps in this series which had been revised between 01-JAN-44 and 30- JUN-45 produced Fig. 3 on the screen or plotter within seconds. The concentration of revision activity along the Polish and Italian borders in 1944 and 1945 is very marked. This pattern would not have shown up on Fig. 2. The sheet numbers on the computer index have been temporarily altered to a numeric format - J 10 appears as 1010 (the 10th letter in the alphabet). A minor change has now been made to the graphics software to allow it to read alphabetic sheet numbers.

Our first experiments in Edinburgh to test the graphics facility were made on Ordnance Survey maps since we already had a digitised outline of the UK. Our stock control exercise was to assemble maps for field excursions. We wanted to locate the multiple copies of folded 1:50 000 topographic sheets, any large scale urban plans and all thematic maps, air photographs and colour slides covering the excursion areas. We are able to search the ORACLE Database Catalogue by KEYWORD or by any of our 70 entry points. In the query shown in Fig. 4 we listed the sheet number (SNO), name (SNAME) and number of COPIES of Ordnance Survey 1:50 000 topographic maps in our collection, where the number of COPIES was greater than five.
This search revealed that sheets 36 and 162 appeared twice so the query had to be widened to see if perhaps one group of maps was out of date.
A further query revealed the publication dates (Fig. 5). We had to decide at this point if the 43 sheets of GRANTOWN AND CAIRNGORM published in 1976 were too out of date in terms of roads or settlement to be used on the field excursion. We also had two separate collections of the GLOUCESTER AND FOREST OF DEAN sheet, both published in 1980. By further querying of ORACLE we revealed two different editions within that year (Fig. 6).
Minor changes had been made to the A edition, resulting in an A-edition. It rests with the map librarian and the academic staff concerned whether to update the holdings. This would probably only be necessary for the 1976 group.

As a result of the ORACLE/GIMMS software link we have been able to display on the screen or plotter all our maps which relate to field excursions. The 1:50,000 coverage related to excursions is shown in Fig. 7. This allows us to plan if we need to extend the coverage further for return visits to the same area.

In Edinburgh we started with the assumption that a smaller medium-sized map library might not have a complete set of paper indexes to cover its collection. Rather than digitise the corners of each sheet from a paper index diagram, we recorded the grid references of the bottom left and top right corners of each map sheet within a series. This allowed for window searching as well as for drawing the computer-generated indexes. Window searching is conducted by defining an area within a box of coordinates so that all maps of specified SCALE, SUBJECT or PRINTDATE, or whatever field required, are displayed. If you have good paper indexes it is faster to digitise the corners of the sheet lines. However if you have not it is necessary to record the bottom left and top right grid references of each map, or all four corners to allow for irregular polygon searches.

We allowed three SUBJECT headings for each map entry. The SUB 1 field is the same as the subject classification used in our department library for books and theses. This allows for cross-referencing between the material in the library and the Map Library. Many maps have more than one main subject so these can be accommodated in fields SUB 2 and SUB 3. Alternatively these fields can be used for a more detailed subject description. A glossary is kept of all subjects used.
Fig 8 shows the Soil Survey maps of Scotland at a scale of 1:250.000. They were traced in ORACLE under SUBJECT and SCALE. A selective search by date from the Soil Survey confined the plotted index to those sheets published after 1982. This was the same technique used for plotting the German sheets by PRINTDATE.

We also selected maps from different series in the same query (Fig. 10) to make a composite computer index (Fig. 11). Up to four colours can be used in the plot to distinguish between series. A window was defined by National Grid coordinates in which we wanted to extract all our holdings of 1:50,000 and 1:250,000 topographic maps. This was the area of the FOREST OF DEAN Excursion in Gloucestershire. We could have extended this query by joining the MAP, AIRPHOTO and COLOURSLIDE tables together to plot out the location of the photographs on the computer graphics index. The location of the photographs would be indicated by symbols. Computer generated indexes could be constructed in this way to superimpose new series sheet lines on old.

Additional search facilities for URBAN areas were built into the Edinburgh system so that the MAP table could be joined with other tables of cartographic information held in the department. By linking the GAZETTEER table with the map coverage of Germany we were able to call up a graphic index of maps between 500,000 and 900,000 (Fig. 12).

Further research and development on this project is planned. Anyone wishing to dial-up the system or to have a live demonstration should contact Barbara Morris, Department of Geography, University of Edinburgh, Drummond Street, Edinburgh EH8 9XP, Scotland. Telephone number 031-667- 6769 extension 4275.

Back to GdC Homepage