Foredrag på GIS-seminar om "Data og analysemetoder" i regi av Universitetet i Odense.

Foranderlige Landskaber - Center for strategiske studier i Kulturmiljø, Natur og Landskabshistorie på Hollufgård, Odense, Danmark 18. mai 1999.

Av Stein Tage Domaas

Høgskulen i Sogn og Fjordane, Avdeling for Naturfag, Seksjon for Landskapsøkologi

GIS and historical maps

(GIS og historiske kart åpner dørene til ny kunnskap)

The technical aspects are based on the following articles:

Domaas, S. T. 1999 (in prep.). Historical land reallocation maps and GIS.

Hauge, L., Domaas, S. T. og Austad, I. 1999 (in print). Ormelid. HSF rapport nr. xx-99. Høgskulen i Sogn og Fjordane.

INTRODUCTION

The opportunity to analyse an area in a chronological perspective can be an important criterion for the selection of areas for various types of biological, ecological and historical studies. In Norway the most important and extensive historical material is provided by land reallocation maps from the nineteenth century.

This project forms part of an interdisciplinary research project entitled "The traditional western Norwegian farm as a biological and cultural system". It is financed by the Research Council of Norway and involves collaboration between Sogn og Fjordane College, the Havråtun Museum, the Norwegian Agricultural Museum, and the county office of cultural affairs in Sogn og Fjordane. Professor Ingvild Austad at Sogn og Fjordane College has been the leader of the project.

EQUIPMENT

A main objective for the project has always been to work only with PC's due to the fact that this is the only platform where our institution can offer any support.

For this study, we used a PC equipped with a 100 MHz Pentium processor, 32 Mb RAM and a Summagraphics Microgrid Ultra digitizer. The software ArcInfo 3.4.2/3.5.1 for PC/Windows was used for digitizing and processing map files, and ArcView 2.1b/3.0a were used for displaying and analysing map files. The platforms used were MS-DOS and MS-Windows 3.1 for digitizing and processing data, and WindowsNT 3.51/4.0 for displaying the analyses.

Today the platforms for the systems are mainly WindowsNT4.0-SP4. The programs used are the latest versions of PCArcInfo (3.5.2) and PCArcView (3.1.1) from ESRI. The processors have become faster and we use 64-128 Mb RAM.

MAPS

Modern maps

Maps are often constructed so that sloping areas appear smaller on the map than they really are, because they are orthogonal projections of the surface. Areas measured on the map will therefore differ in length and size from the same lines and areas measured on the ground.

In this project the "economic" map series 1:5000 (ØK) were used (NGO 1948). The first maps in this series covering Sogn og Fjordane were produced in the 1960s, and cover only areas of economic interest up to an altitude of 600 - 800 m.

Historical maps

The availability of detailed historical maps in Norway is more limited than in most other European countries. Military maps from the eighteenth century cover some areas, but the most important and extensive material is provided by land reallocation maps from the nineteenth century.

Historical maps are generally large-scale: most land reallocation maps were drawn at a scale of 1:2000. They contain large amounts of information on structures, areas and elements of the landscape. The surveyors were skilled craftsmen, and were thus able to verify the accurate position of landscape features.

This suggests that it ought to be a relatively simple matter to fit together or combine information from historical and modern maps. However, this is not the case, as each land reallocation map is constructed using its own, local coordinate system. They also contain other systematic errors, for instance inaccuracies in the direction of true north, and non-systematic errors.

Digitized land reallocation maps used in the project are:

· Grinde 1874 (1:2000)

· Skjestølen-Seljevollen 1898 (1:2000)

· Engeseter 1910 (1:2000)

· Utmark 1958 (1:5000)

Other maps used are:

· Grinde 1948 (1:2000) land reallocation

· Holding 4/2 Grinde 1996 survey

Digitization

All maps used for this part of the project had to be digitized manually in vector format.

In order to retain the information on point, line and area features given by the land reallocation documents, the map objects were digitized in vector format. This laborious process required precision to achieve the best possible correspondence between the original analogue map and the digital map.

In order to code the digitized points and lines accurately, it was essential to decide on certain principles before the process was started.

Accuracy

In the project we evaluated the accuracy of a digitized and transformed historical map on the basis of the standard set by the Norwegian Mapping Authority for modern maps.

The standard used by the Norwegian Mapping Authority, SOSI v. 3.0 - part 4, includes a description of the accuracy which is acceptable for maps at different scales. Referring to point mean error (pme) the level of accuracy is acceptable when more than 68% of the map elements have an error of less than the pme multiplied by a factor of 1, fewer than 27% of the elements have an error of the pme multiplied by a factor between 1 and 2, and fewer than 5% of the elements have an error of the pme multiplied by a factor of more than 2.

In this case we must consider pme for the scales 1:2000 (0.58 m) for the land reallocation maps and 1:5000 (2.0 m) for the Norwegian economic map sheets (ØK). A survey of holding no 2/4 is considered to be accurate to within 10 cm (pme 0.1 m). When two or more maps are combined, pme increases even further.

In our case this gives a new pme of 2.08 m, when the historical map and the economic map sheet are combined, and a pme of 2.09 m if the survey of holding 2/4 is included as well.

If the discrepancies between the maps used for the study do not exceed the accuracy acceptable for a pme of 2.09 m, the accuracy will be as good as that obtained using only modern maps.

However, accuracy may also be defined on the basis of other considerations, since one of the maps is an historical map based on an old surveying method, which introduces both systematic and non-systematic errors.

An appropriate goal could be to ensure that no line or area element can be mistaken for another element.

CONCLUSIONS ON ACCURACY OF LAND REALLOCATION MAPS

Analysis of the material demonstrates clearly that line and area elements on the maps coincide very well and that they can appropriately be linked and used for overlay analysis. Elements with a total length of 1415 m were controlled against buffers representing the mean error groups.

COMBINING HISTORICAL MAP DATA AND RECORDS

Early in the process we realised that we could not foresee every detail we would approach in the written records. (There were 1500 hand-written posts in the 1874 land reallocation map of Grinde). Using traditional database technology/thinking could therefore limit our use of this type of material.

We therefore decided on using spreadsheet technology in the recording process. This allowed us to split the information into pieces, which were suitable for our needs, as we went along.

Even if we later realised a need for more detailed splitting of the information, this could easily be achieved, since the records were linked, not embedded.

TIME LAPSE STUDIES

GIS is an excellent tool when doing interdisciplinary research if each discipline at an early stage can agree on doing field work and data collection in such a way that it is compatible with a common GIS-structure.

Combining historical maps, modern surveys, written historical information and information from informants, makes it possible to do reliable geographic related landuse studies from today and several hundred years back.

This procedure allows for direct matching and comparison of point, line and area features from various dates. Subsequent analysis enables us to illustrate gradual changes of structures in a cultural landscape, for instance the loss of archaeological and architectural remains, changes in land use, the spread of built-up areas, etc.

We may even be able to demonstrate whether cultural landscape structures that are generally perceived as old or of special interest were in fact created by long-lasting sustainable processes or are more recent developments. In this way, up-to-date information can be used to improve our understanding of land-use patterns and historical processes.