Abstract: A recent progress report on European GIS standards states that comprehensive strategies are put forward for reaching at European standards. This international trend challenges the important national standardization efforts, e.g. the development and wide application of the DSFL exchange format in Denmark.

The Computer Graphics Metafile standard (ISO 8632) was adopted in 1987 and has achieved a market acceptance. This market acceptance may be largely outside the mapping community proper, but a substantial GIS market growth is expected to take place in this area. The CGM standard and other selected transfer formats are surveyed, and the relation between these standards is suggested to be closer investigated.

A reference frame is presented for the discussion on the standardization of geographical information. The reference frame, a model universe, including a theoretical model, is related to a recent research survey on the geographical information science. The possibilities of developing high-level tools for overcoming the barriers set by different GIS exchange formats is discussed, and ongoing development applying the Prolog language in this area is reported.

Keywords: Geographic information, transfer format, ISO 8211, ISO 8571, FTAM, ISO 8632, CGM, EDIFACT, theory, model, spatial reference frame, quality measures, visual language, Prolog.

1. Introduction

A section (2) on the motivation of this essay is followed by two larger parts: The first part (section 3) surveys some ISO standards which may 'compete' with the ISO 8211 which has been presented as a basis for a European Transfer Format. An awareness that the standardization effort may apply to several joints of a communication link from observer/data supplier to the end user triggered the development of a theoretically based reference frame for the standardization activities. This reference frame is presented in the second larger part: Section 4. Research on spatial or visual language is referred to, since this research may contribute to the development and implementation of standards for the transfer of geographical information.

2. Motivation

The Danish mapping community agreed on a national DSFL format for the transfer of digital map data in 1982 (Lauersen, 1988; Alexandersen, 1991). The diffusion in Danmark of this transfer format appears from a recent survey of the use of CAD-application, cf. Figure 1.

The main associations of the Danish construction industry issued a survey on the use and diffusion of CAD within this industry (CAD-undersøgelsen, 1991). The response rate amounted to 36%. The survey includes, among others, data on the exchange of drawings, etc. in digital format. The response on questions concerning the amount of data exchanged was inconclusive. The survey gives, however, an interesting account of the exchange formats used, c.f. Figure 1. The DXF-format is by far the most applied. The high number of exchanges, based on 'Internal format', is related to the benefit users of same system brand have, when system-specific database information are included with the transfer. Concerning the little use of IGES the report notes that almost all exchange is 2D-drawings which may be perfectly well transmitted using the other formats.

The diffusion of the DSFL-format is due, mainly, to the fact that the municipalities, the utility companies, and surveying companies in tenders for GIS demanded the availability of DSFL-generators and interpreters. The same mechanism is in operation in Norway where larger governmental agencies have demanded the use of the Norwegian SOSI transfer format from contractors and data suppliers (Østensen, 1991:5).

The advent of the 'Single Market' of the EC will eventually exclude this reference to national standards because the national standards hampers the free flow of goods and services. This trend is illustrated by the EC Directive 77/62 with later amendments which concerns procedures for the procurement by public agencies. According to the directive (as of 22. March 1988 88/295/EEC, section 7, subsection 2) technical specifications are to follow national standards, "which are reflections of the implementation of European standards.."(translated from Danish version). Several exceptions are mentioned, however. If the conformance with the mentioned standards would imply the acquisition of equipment which would be incompatible with existing equipment, or would be excessive expensive then exemptions may be allowed (section 7, subsection 3c).

The implications of the 'Single Market' and the mentioned Directive for the use of national standards is thus a matter of judgement. One attitude is to 'wait and see' expecting that the national standard still may have 5-10 years left (Alexandersen, 1991). Another approach, adopted here, is to analyze the standardization scenario, and to prepare for the 'mapping' of the national standard onto a European or international standard. By this approach the experiences of a national community may contribute to the international work, and the cooperation behind the national achievements may be continued. The alternative would be to let others determine the world view and communication practices which - explicitly or implicitly - have to be encoded into a standard. This is in fact a cultural submission.

The motivation to consider a specific international standard: the Computer Graphics Metafile (CGM, ISO/IEC 8632, 1987) is that this established standard allows for the integration of technologies. The GIS are for presentation and analysis of geographic information. However, many end users do not use the GIS personally. This is especially true for GIS in the public sector, where maps and diagrams have to be presented to politicians and the public in terms of figures in reports and folders, and on overhead foils etc. The CGM, among other capabilities, is an instrument to link the GIS results to word processing, and thus to integrate the two technologies. It seems likely, furthermore that the use of (low cost) GIS will increase substantially when the integration of GIS with word processing becomes more straightforward. Finally, this standard is mentioned in the national (Danish) prescriptions for the procurement of computers etc. which completes the mentioned EEC Directive. A survey of these prescripts is provided by the third of six EC brochures on the implementation of the 'Single market' (EKSF, 1992).

3. Recent standardization efforts related to the transfer of graphics

Standardization of data exchange means that the use of geographic information becomes cheaper, the access to these data becomes easier, and the risk of misinterpretations is reduced (Cederholm, 1991b:5). Why, then, is standardization not commonplace?

Standardization is a difficult and resource consuming activity. You need much ingenuity to identify and solve the problems which are related to early application of new technology. To achieve a consensus at an international level through an open, public process is a tedious affair. The standardization process results in conventions, and related products, which are free for people to use. 'Free' means that you are more independent of the computer companies' proprietarian products. The companies' market interests in the standardization process are evident. Based on their market position they may decide to support one standardization effort, and to neglect another. To understand the standardization process you thus have to refer to social science concepts: Market, organization, roles, values, etc. And you need to understand the standardization process, because the 'investment' of one's scarce resources in the standardization effort has to be done with care.

The European mapping community is preparing an European Transfer Format. The contributions of the International Standardization Organization (ISO) were recognized by the Working Group V of the Comité European des Responsables de la Cartographie Officielle (CERCO) when they agreed to develop an interim transfer standard provided that such a standard would be compatible with ISO 8211 (Sowton, 1990).

However, one aspect of the relationships between the mapping actors and the ISO and the national standardization organizations appears from the statement of a Norwegian official: "The SOSI-exchange format is not within the realm of the Norwegian Standardization Organization (NSF), but is based on the governmental obligations and professional expertise of the Norwegian Mapping Agency. We have so fare considered this the most suitable arrangement". (Østensen, 1991:5/10; my translation). The statement may convey a general attitude. At least the proponents of the Computer Graphics Metafile note that the needs of cartography have not heretofore been expressed in the graphics standards community, and that the graphics standards are not widely used in those areas currently (Henderson & Mumford, 1990:341-342).

The focus of the standardization effort may relate to several joints of the communication chain. An example, referring to driver information systems, is given by Luc Heres of Phillips Consumer Electronics: One possibility is to standardize the finished article, ie the map data in the form which they are represented on the medium used by the system itself (e.g. a CD-ROM or a coded beacon message). Another possibility is to standardize the map data as they occur in a general purpose road database, or in the datafiles produced by data suppliers (Heres, 1990).

Within this framework the Danish DSFL format and the Norwegian SOSI format both focuses on (terrain) object definition and data collection, while the Finish EDI based format takes the transfer of data between computers by the means of an independent network into account (Cederholm, 1991). The CGM standard apply, preferably, to the last joint of the chain: the transfer of data to a printing device. It can be argued that the difficulties related to the establishing of a standard grows the more you focus on non-technical matters, i.e. moves the focus from printing devices, computers and networks towards terrain objects.

The following sections present a short survey of a selection of standards for transferring data, including the context of the standardization work, the main features of the standard, and the recognition of the standard by the mapping community.

B. ISO 8632 Computer Graphics Metafile

The CGM standard defines representations of graphics for storage and transfer between systems. The standard is fundamentally a file format and file transfer encoding standard (Arnold et al, 1991; Henderson & Mumford, 1990).

The standard evolved from a binary picture storage format devised by the US National Center for Atmospheric Research in the 1970s. Standardization efforts in the US (ACM Siggraph) merged with ISO activities related to the GKS standard (Graphics Kernel System, ISO 7042). General Motors' Manufacturing Automation Protocol (MAP) and Boing's Technical Office Protocol (TOP) initiatives were acknowledged in Europe by the mid-1980s. These protocols are based on the Open System Interconnection (OSI) standards. The CGM is included in these specifications (Henderson & Mumford, 1990:35).

Pictures are described in the CGM standard as a collection of elements of different kinds, representing, for example, primitives, attributes and control information. CGM is a multipart standard. Part 1 defines the available elements, the structure of the metafile and the order in which elements may appear. Parts 2-4 of the standard define three representation schemes, or encodings, for the abstract syntax of the elements, Part 2 is a character encoding aimed at compactness and transferability across networks. Part 3 is a binary encoding aimed at minimizing the processor effort involved in generating and/or interpreting the metafile. Part 4 is a clear text encoding aimed at a metafile that can be read and edited be people.

A CGM is structured as a series of levels:

• Metafile: A metafile consists of a metafile descriptor followed by a number of picture descriptions. The metafile descriptor contains information valid for the whole metafile, for example the precision of real and integer quantities in the metafile. Picture descriptions are self-contained.
• Picture: Pictures are bounded by picture delimiters. A picture starts with a picture descriptor that contains information about how this particular picture is stored. This is followed by the picture body, which contains the definition of this picture.
• Picture body: The elements at this level are essentially primitive and attribute elements describing the graphical content of the picture.

CGM only allows 2D pictures to be described and provides a total of 19 graphical primitive elements, including polyline, circular and elliptical arc, polymarker, text elements, filled-area elements, and cell array element.

The encoding of colour information can be an important component in determining the efficiency (or otherwise) of an implementation's encoding strategy. Two colour specification modes are provided in CGM: indexed, which is equivalent to the mechanism of GKS, and direct, in which the parameters of the colour-specification elements are RGB triples.

Similarly, different modes are specified for the definition of line width, marker size and (polygon) edge width. These later modes correspond to a specification of the coordinate system in which the dimensions of the primitive are to be interpreted.

The ISO standard 8632 Computer Graphics Metafile (ISO 8632) became adopted as an international standard in 1987. Today CGM drivers are generally available, also with 'desktop GIS' software package like "MapInfo" and "Atlas*GIS". A CGM generator has recently been developed for the Norwegian Pumatec system. As mentioned, the CGM standard is recognized by the national (Danish) prescripts on computer etc. procurement (Finansministeriet, 1991a). A German document processing system is based on the CGM and the related Standard Generalized Markup Language (SGML, ISO 8879). Experiences from a prototype are reported (Scheller, 1990).

The CGM was considered by Francois Salgé in a survey on existing solutions for exchange formats for geographical information (Salgé, 1989). The CGM was rejected because "CGA deals with graphics and has no possibility of adding geographical attributes" to lines etc.(:336). In a summary survey it is noted, furthermore, that raster data and topology is not covered by the standard (Salgé, 1989:344). Especially with the Addendum 1 to the CGM standard these assessments have to be reconsidered. The CGM element: 'External Elements' may be used for storing geographical attributes. (:276). The CGM element 'Cell Array' may transfer raster data (:254), and the concepts of 'closed figure' and 'segment' (:309-313) should allow for the rendering of topology.

The CGM standard was again presented to participants of the mainstream geographic information discussion in 1990 by Joachim Kammerer of Siemens Nixdorf (Kammerer, 1990), and was mentioned recently by Torbjørn Cederholm (Cederholm, 1991b).

C. ISO 8211 Data descriptive file for information interchange

ISO 8211 is a standard which specifies medium independent and system independent file and data record formats for the interchange of information between computer systems.

The standard originated in a tool developed for transferring information between several sites within the authority of the US Atomic Energy Commission. The data transferred could vary so much in form that it was necessary "to produce a means of transfer which was content independent (that is, which provided sufficient internal documentation of the data being transferred to enable it to be read with little or no external documentation)." (Ibbs & Sowton, 1990:43, cf Salgé, 1989). The mentioned article suggests ISO 8211 used as a basis for the British National Transfer Format (NTF).

The ISO 8211 transmits a description of the data being transferred with that data and provides for a wide variety of data types and structures within the data being transferred. Thus the structure of IS 8211 permits a range of automatic validation procedures to be carried out on the data, e.g. check of data type, or check of the occurrence and/or sequence of fields and sub-fields.

Also, it is argued, that the data descriptive content of ISO 8211 protects the user from changes in the NTF itself. There are several reasons why a transfer format changes over time. This may lead to problems where the version of the NTF transmitted is not the version expected by the recipient. A solution is suggested by writing "software which is shielded from all three form of change [not cited here. Est], so that a program which cannot understand some (notational) new form of NTF will still be able to extract meaningful data, without special action being taken by the person sending it, and so that a program expecting the new form will still be able to cope with data which does not contain all of the information it expects" (:43-44).

Mention is made of the fact that the ISO 8211 has been adopted by other exchange formats concerned with the exchange of spatially related data. The article gives no references to these exchange formats. Francois Salgé mentions that the US SDTS standard fits within the ISO 8211 standard (Salgé, 1989:340).

A rather thorough survey of scientific literature on the ISO 8211 standard provided three further references, two from the UK Association of Geographic Information 1990 conference in Brighton, and one, from 1984, to the further application of the 8211 standard in field of nuclear instruments (Tischler, 1984). File transfer formats or protocols were covered by several titles relating to the Open Systems Interconnection (OSI) family of standards, e.g. P. F. Linington: File transfer protocols (Linington, 1989), or to the EDIFACT family of standards. The Danish prescripts and guides for procurement by public agencies (Finansministeriet, 1991b) does not refer to the ISO 8211, but the ISO 8571 is recognized.

D. ISO 8571 File Transfer, Access, and Management (FTAM)

A file transfer protocol is a protocol which allows for the specification of a file and its properties, as a prelude to the communication of a part or all of the file's content.

The ISO 8571 standard is defined within the Open Systems Interconnection (OSI) standardization activity, which set out from 1978. Among others, a Norwegian networking project, Uninett, developed a file management protocol based on the OSI structure about 1980. General Motors' Manufacturing Automation Protocol (MAP) and Boing's Technical Office Protocol (TOP) initiatives provided the context in which the ISO FTAM standard was eventually formally adopted at the start of 1988 (Linington, 1989).

Information about the file of interest is exchanged by reference to an explicit model of the filestore which is specified as part of the protocol. The model defined in FTAM is known as the virtual filestore. It is based on the definition of a set of file attributes which are pieces of information capturing the essential properties of the file. Attributes includes properties of the file content such as the type of information stored.

"Kermit" is an example of a transfer utility. This utility requires a substantial awareness of the nature of the target system. Such limitations are dissolved by the ISO 8571 which, furthermore, seems to include benefits of the ISO 8211 mentioned above.

ISO 8571 recognizes the very many types of file and file structure, varying from the simple text file to the complex, indexed, record-oriented file. The FTAM standard uses a hierarchial model, a tree: The file is considered to be a tree with subtrees: file access data units (FADU's) which can be addressed independently, and nodes with identifiers and associated file data units (DU's). This structure is the structure on which the protocol operates, its access structure.

Provision is made for an even richer structure of references within the file to support the meaning of the information stored. The protocol, the dynamic aspect of the file transfer, handles a variety of file types. The file types, however, does not form part of the file transfer protocol. The values of abstract data types, and information on encoding and interpretation of data are defined by a document type. The meaning to be associated with the transferred data values can thus be transferred within the format.

The FTAM standard itself includes a few common document types to satisfy basic requirements for text or binary file transfer. Further document types are to be defined and registered by standards bodies, user organizations, or suppliers.

Østensen considers the FTAM standard among future Norwegian standards, while EDIFACT is considered the future standard at the subsequent, higher level (Østensen, 1991). This combination of standards is in accord with the recommendations of the European standards organizations who are developing 'functional standards' or 'functional profiles', e.g. 'F-profiles' for the transfer of text and graphics, among others (Finansministeriet, 1991b:14).

E. The relations between selected ISO standards

The main stated benefit of the ISO 8211 was that the standard specifies a file format which transmits a description of the data being transferred with that data. This seems possible within the ISO 8571 standard, also. The functionality of the 'document type' specification and the ISO 8211 ought to be compared to clarify their relative advantages. The larger complexity of the FTAM standard, the efforts put into the preparation of the standards, and the market adoption of the two standards has to be taken into account, among others. As mentioned above the ISO 8211 is adopted by other exchange formats concerned with the exchange of spatially related data.

The EDIFACT standard provided the basis for the Finnish VHS 1040 and VHS 1041 format. The EDIFACT standard originated within the auspices of the United Nations, but its syntax and coding is now issued as ISO standards, including ISO 9735. The prospect of the standard is to reduce the time lag and cost of re-keying data, e.g. mercantile data or CAD data, and to allow for electronic post facilities. EDI formatted messages can be transferred by several services, including the FTAM-based services (Finansministeriet, 1991:32).

EDI, i.e. Electronic Data Interchange means that electronic transfer of data takes place through a network which is often operated by a third party who may then provide 'value-adding services'. The alliances which may be created through the introduction of this new technology means that the introduction of the technology may be considered a competition parameter in its own right. These non-technical issues are mentioned to enhance the awareness that the selection of one standard in preference to another may influence the market in an unintended way.

This EDIFACT standard is not described here; reference is made to Berge: The EDIFACT Standards (Berge,1991). The ISO 8879 Standard Generalized Markup Language (SGML) 1986, and the ISO 8613 Office Document Architecture (ODA) are bypassed, also. A survey of these standards may be found in a DIN Technical Report on Computer Integrated Manufacturing (DIN, 1988).

Concluding this part of the paper, the evidence provided so far justifies a review of the interrelations of the ISO standards to clarify the technical and other criteria for the selection of a basis for the European Transfer Format.

4. Geographic Information: The object of the standardization effort

A. Introduction: A reference frame for the discussion

The previous part of the paper described the standardization effort relative to certain phases of the communication joints from observer to end user. In this part of the paper an attempt is made to structure the 'geographic information' scene. The standards organizations and their committees, e.g. the new CEN/TC 287 Geographic Information, structures their problem area when they set up working groups with specific tasks. The research community performs a similar task when researchers present a survey of the 'research frontier', and when governmental research councils allocate grants to a comprehensive research programme.

The structuring principle of this part will be a model universe, emanating from the field of systems control (Nørgaard Nielsen, 1986). This framework will provide a reference for discussing a research review of geographical information science (Goodchild, 1992) and for discussing selected contributions from the efforts to reach at an European Transfer Format.

The model universe consists of a set of models which emphasize the static aspects of a field of interest. This field of interest is described by a set of models, each of which is expressed by a specific 'language'. The set of models make a sequence where the next model of the sequence expresses not more of the field of interest than the previous. The 'language' used, however, is more 'closed' than the previous, which means that preconditions and valid domains can be stated explicitly. The objective of applying this set of models is that an 'open' field of interest, varied and full of ambiguities, may be analyzed with the rigor of mathematics.

The model universe is made up of the following: The field of interest is considered a System, defined and delimited from the system's environment by the system author's System Model. The System Model is mapped onto a Theoretical Model, which is again mapped onto a Mathematical Model. While the entities and relations, etc. of the System Model may be expressed in natural language terms, the Theoretical Model may be expressed only by theoretical terms, and the same apply to the Mathematical Model which has to be expressed in mathematical terms.

Model building is no new idea. Operations research, systems analysis, location-allocation analysis, etc. provided the basis for much social betterment, since the arrival of General Systems Theory in the 1950s. A summary of the systems analysis approach is the following:

1. The system to be investigated is explicitly distinguished from its environment.
2. The internal elements of the system are explicitly stated.
3. There are relationships between the system and its environment that are explicitly stated.
4. Where these relationships involve deductions the canons of logical or of mathematical reasoning are employed.
5. Assertions concerning the relationships between the system and the real world are confirmed according to the canons of the scientific method (Kaplan, 1968, as cited by Coffey, 1981:18)

Item 5: the canons of scientific method, may be interpreted as providing evidence through observations or measurements, and a subsequent statistical analysis. This is the usual procedure followed, e.g. for calibrating traffic models. Here item 5 is interpreted as the Theoretical Model of the model universe. It is likely that this is a new approach.

B. Topics within the System Model

The lack of common terminology and glossary, data classification and feature coding, and common conceptual data models are frequently counted as impediments for the use of GIS on a European scale. These topics and belong to the 'system model' of the model universe. Nick Land (Land, 1989) has referred to dictionaries published by specialist disciplines, eg 'A dictionary of the natural environment' by Monkhouse and Small, and to feature codes (taxonomies of terrain objects) applied by governmental agencies or adopted by codes of professional practice. In a survey of geographical information science Michael F Goodchild points to the fact that the integration of GIS and spatial analysis is hampered by the lack of standards for data models (Goodchild, 1992:39). He considers the possibility of developing a common language, whose primitive elements wold represent the fundamental operations of spatial analysis. Based on experiences from attempting at defining such a language he suggest to begin with the conceptual framework provided by a comprehensive data model.

The efforts needed to reach at a comprehensive and generally adopted data model may be illustrated by the variety of approaches presented in the textbooks on GIS which have become available recently. Also, some of the scarce resources necessary to reach at a general solution may be allocated to more specific tasks. The establishment of a monitoring and auditing system to support the Common Agricultural Policy (CAP) of the European Community (cf. KOM(91) 533 of 13. December 1991) certainly is in a need to clarify the national concepts involved. Another example is the CADDIA programme, aiming at improving data compatibility within the custom and agriculture areas (cf. SEK (92) 1598 Final 2. September 1992, p9).

To reduce the task at hand it has been suggested that a single common model ought to be defined at a very low level, and that classifications and standards ought to be application specific, while overlap encountered between application areas were resolved with the use of shared catalogs and techniques of 'mapping' between them (MVA Systematica, 1990: Summary of discussions, p11)

The concept of the 'Systems Model' may assist in resolving certain ambiguities by emphasizing questions of what is the 'System' to be modelled by the GIS? For topographic surveys the system may be the countryside, the terrain objects: streams, buildings, hedges, etc. But it might be, alternatively, be the analogue map which is to be digitized. The different agencies of public administration have each their 'world view', largely determined by current legislation. The same section of the surface of the Earth is thus conceived as several 'Systems' by the different departments, who each define their specific 'System Model'. The variety of data models from a scientific or textbook perspective thus has to be amplified by the variety of public and private 'worlds views' to reach at an estimate of the task ahead.

One important aspect of the different 'world views' mentioned is the spatial resolution of the 'System Model' (cf. Maes, 1990: Section 3.2). A system model (e.g. in the cadastral field) may include several area entities: lot, property, township, municipality, and region, and the relationships between these entities and entities from 'adjacent' departments: building permits department, environmental control, etc. have to be clearly defined.

The issues mentioned may be considered by the WG 2: Models and descriptive languages for geographic information (Secretariat: France) of the CEN/TC 287 Geographic Information, and of the AG1 'Klassificering af geografisk information' (Classification of geographical information) of the Swedish 'STANdardisering av LanndskapsInformation' project (a national standardization activity).

C. Topics within the Theoretical Model

The topics which belong to theoretical model may be separated from other topics by asking: what is specific for this knowledge area, and does not belong to the System Model or the Mathematical Model? Concepts like point, line, area belong to the field of geometry which is a part of mathematics. House, brook, boundary mark, etc. belong to the system and system model. The coordinate system as such belongs to the mathematical sphere. The geodesist and the land surveyor is, however, concerned not only with the cartesian coordinate system, but also with its application to real world phenomenons: the planet earth, a part of the surface of this planet, and the rendering of the relative position of objects of this surface on a 2-dimensional plane. The lucid presentation by Jeff Maes of GIS experiences of the CORINE programme illustrates the importance of different map projections when data are compared at an European scale (Maes, 1990: Section 3.2).

The Danish DSFL format has been in use since 1982, and the development of this transfer format has been user driven (Alexandersen, 1991). It is, therefore, interesting to note the way spatial reference frames are reflected in this format. From the outset an appropriate coding discerned between different map projections, eg. a Danish projection called System 34, and the UTM, which, by the way, has a datum which differs from System 34. Later, the need of the utility companies resulted in the concept of 'reference points', 'reference lines' and 'reference areas'. This means that you can render the topology of a network through the DSFL format. Recently, an update of the DSFL format allows you to transfer data without referring (terrain) object codes to coordinates. Instead the localization is provided by reference to the national coding of streets (postal address coding), which has been in effect since 1977.

In a recent review of Danish experiences concerning the development of national LIS' a rather comprehensive survey has been made of spatial reference frames, cf Figure 2. Important is that spatial reference frames need not to be coordinate based. A spatial reference frame is a mathematical construct related to a physical body (Stubkjær, 1991b:217), and there are mathematical constructs which may replace the continuous Euclidean space.

Andrew U. Frank and David M. Mark note that the science of natural processes deals more with continuous space while everyday human experiences sees a world of objects (Frank & Mark, 1991, as cited by Goodchild, 1992:37), and Goodchild concieves "an empty space littered with possibly overlapping objects" as an alternative to a world (i.e. a System Model) of continuous variables (Goodchild, 1992:37).

Spatial reference frames are treated by the WG 4: Positional reference models for Geographic Information (Secretariat: Germany) of the CEN/TC 287 Geographic Information. Reference systems are explicitly addressed, also, within the Swedish STANLI project (Cederholm, 1991a:15).

The spatial reference frame constitute one theoretical topic, a generalization of the coordinate system. Another theoretical topic concerns the quality of the data. Michael F Goodchild calls for "generic models of uncertainty, analogous to the role played by Gaussian distribution in the theory of measurement error" (Goodchild, 1992:41, cf 35+). A linguistic model of communication has been suggested as a reference frame for a rather comprehensive set of quality measures (Stubkjær, 1990). Standardization efforts in Sweden focus at quality measures: AG5 Kvalitetsmärkning (Cederholm, 1991a:12). Specific accuracy groups may be stated within the current Danish DSFL format (Alexandersen, 1991). Maes mentions the problem of consistency among different data sets (Maes, 1990: Section 3.2). This measure of quality is related to more commonplace quality measures, like positional accuracy, through the suggested communication framework.

Summarizing the review of GIS research, M F Goodchild mentions that "the difficult problems of data modelling, error modelling, integration of spatial analysis, and institutional and managerial issues remain"(Goodchild, 1992:42). The three former issues have been addressed. The development of GIS' at a national scale, cadastral development, must relay, mainly, on an understanding of these organizational issues. Some attempts in this area has been made (Stubkjær, 1991b; 1991c)

A spatial or visual language?

Recent research in GIS and 'Visual languages' has addressed the issue of developing a specific language for spatial statements. Spatial reasoning, a common language for GIS data models, etc. is based on such language (Goodchild, 1992:39.43; IEEE VL89, 1989). From the linguistic point of departure the spatial dimension is addressed by investigations into the natural language used for expressing spatial relationships, e.g. route descriptions. A Danish masters thesis: Semantic analysis of Danish spatial statements (Kaas, 1989) may serve as one example of this approach. The 'message' concept of the communication model which was mentioned in the previous section may be developed, alphabets of 'sign's and 'supersign's specified, and grammars developed which recognize whether the 'sign's are literal or pictorial (cf. Stubkjær, 1991a).

A grammar can discern the literal or pictoral kind of a sign, if the sign is accompanied by this information. The requirement for a European Format that a description shall be transferred with the data (cf. Ibbs & Sowton, 1990) is quite analogue to the requirement that a sign shall be accompanied by information on its kind. A third analogy is the suggestion that a message may be accompanied by the code of that message (Stubkjær, 1990).

The mentioned requirement has been aimed at by the Norwegian SOSI format with an impressive outcome (Statens Kartverk, 1992). The SOSI format is written on paper. However, its structure is not far from the declarative part of a Prolog program. If the SOSI format was written as a Prolog program it would be executable, i.e. be able to analyze if a file was in accordance with the SOSI format. This, again, would assist a programmer developing a generator or interpreter for the SOSI format.

The mentioned Prolog program may be specified with a view to the Danish DSFL format and other formats in the mapping field and thus provide an interface to an European Transfer Format. By such activity an 'application profile' (Mumford, 1989) may be within reach.

The research effort to reach at these capabilities have to be carefully assessed. It is interesting to note the achievements of the project: Foundations for Intelligent Graphical Interfaces. A system has been developed which allows for the interactive imposing of meaning to drawings in computer graphics (Klein & Pineda, 1990). The system, GRAFLOG, is implemented in Prolog. Experiences with the system in the field of CAD (architecture) are reported.

D. The mathematical model

Analytical geometry, matrix algebra, and the calculus of probabilities are well known potential components of this model. Relational algebra which forms the conceptual basis for relational databases is another potential component. Goodchild has surveyed the research on spatial statistics and identified new mathematical techniques pertinent for the explicit accounting of uncertainty (Goodchild, 1992:35-36). Graph theory, especially when emphasis is on algorithms, counts among the important topics, as does topology which deals with spatial relations as inclusion, adjacency, and holes in surfaces, among others.

Several textbooks on 'Discrete mathematics for computer science' are available for the teaching of computer science students. Similar textbooks ought to be prepared for geoscience students, who have largely been confined to a continuous conception of space.

This survey of mathematical concepts and techniques is far from being complete, but it does serve to illustrate a field of knowledge which, together with application specific theory, has always provided the basis of scientific reasoning, including spatial enquiries.

5. Conclusion

Standards for the transfer of digital geodata are presented, not only as technical prescripts but also as a plaything of societal forces. The standardization effort is thus considered a 'game' with certain behavioral patterns. Among these patterns the following are mentioned: Defending national interests, creating new alliances, or ignoring such attempts, and estimating the market acceptance of a standard.

It is stated that available scientific and administrative texts frequently refer to the Open Systems Interconnection (OSI) family of standards, including FTAM, and to the EDIFACT standards, while the ISO 8211 standard is discussed mainly by those who participates in the efforts to create an European Transfer Format (ETF) for digital map data. It is argumented that a rejection, in 1989, of the Computer Graphics Metafile (CGM) standard shall be reconsidered.

A model universe, including a theoretical model, is presented and related to a survey of recent research on geographical information. The model universe is applied as a reference frame for the structuring of the standardization efforts.

A criteria for the ETF is that data description shall be transferred with the geodata. This criteria is related to an interpretation of a linguistic communication model, and to recent research which attempt at unifying meaning with drawings in computer graphics systems. A study of these research results are suggested with a view to assist the computer programmer who is developing a generator or interpreter for transfer formats in the geodata area.

Acknowledgements

Thanks to the staff of the Aalborg University Library who enthusiastically lead me to most of the non-GIS titles which are cited below.

References

Arnold, D B, S T Liapakis, G J Reynolds and N P Vezirgiannis (1991) Practical considerations in transporting Computer Graphics Metafiles Computer-aided Design 23(4) (May 1991) 306-
Alexandersen, Sören Riff (1991) Exchange of data. Proceedings, AM/FM-GIS Nordic Region Conference, 28-30 October 1991, Mariehamn, Finland.
Berge, John (1991) The EDIFACT Standards. NCC Blackwell, 1991.
CAD-undersøgelsen '91 (1991) (An investigation of the use of CAD in Danish construction industry). BPS-centret, DK-2970 Hørsholm, May 1992. 34pp
Cederholm, Torbjörn (1991a) ULI föder STANLI. EDI för geografi. Lantmäteritidskriften 83(1) 12-18.
Cederholm, Torbjörn (1991b) International standardization & GIS. Proceedings, AM/FM-GIS Nordic Region Conference, 28-30 October 1991, Mariehamn, Finland.
Coffey, William J (1981) Geography - Towards a general spatial systems approach. Methuen, London.
DIN (1988) The standardization of interfaces for computer integrated manufacturing (CIM). DIN Technical Report No 15. Beuth Verlag, Berlin.
EKSF (1992) Gennemførelse af det indre marked - Bd 3: Rammerne om virksomhedssamarbejde. Et åbent marked for offentlige indkøb. EF-Kommissionen, Bruxelles. Heri bl a: Offentlige indkøb, p 71+.
Finansministeriet (1991a) EF regler om anvendelse af standarder for informationsteknologi ved offentlige indkøb. Administrations- og Personaledepartementet, København.
Finansministeriet (1991b) Standarder for datakommunikation i den offentlige forvaltning. Administrations- og Personaledepartementet, København.
Goodchild, Michael F (1992) Geographical information science International Journal of Geographical Information Systems 6(1) (January-February 1992) 31-45.
Henderson, L R & A M Mumford (1990) The Computer Graphics Metafile. Butterworth, London. 409 pp.
Heres, Luc (1990) Data needs of driver information systems and standardization achieved so far; in: MVA Systematica, 1990.
IEEE VL89 (1989) IEEE Workshop on Visual Languages, October 4-6 1989, Rome. Proceedings.
ISO 8211 (1987) Information Processing - Specification for a data descriptive file for information interchange. International Organization for Standardization, Geneva.
ISO 8632 (1987) Information processing systems - Computer graphics - Metafile for the storage and transfer of picture description information (CGM). International Organization for Standardization, Geneva.
ISO 8632 (1990a) Information Processing Systems - Computer Graphics - Metafile for the storage and transfer of picture description information - Amendment 1. International Organization for Standardization, Geneva.
ISO 8632 (1990a) Information Processing Systems - Computer Graphics - Metafile for the storage and transfer of picture description information - Amendment 3 (Draft Amendment Standard). International Organization for Standardization, Geneva.
Kaas, Ruth (1989) Semantisk analyse af danske rumlige udsagn (Semantic analysis of Danish spatial statements). Unpublished masters thesis. Københavns Universitet, Institut for Nordisk Filologi. 133p + annexes.
Kammerer, Joachim (1990) Evolving concepts for the future of a geographical services market; in MVA Systematica, 1990.
Kaplan, M A (1968) International systems; in: International Encyclopedia of the Social Sciences. MacMillan & Free Press, New York.
Klein, Ewan & Louis A Pineda (1990) Semantics and graphical information, in: Diaper D et al (Eds): Human-Computer Interaction - INTERACT '90. Elesvier Science Publishers (North-Holland) 485-491.
Land, Nick (1989) Transfer standards - Towards a national standard for feature definition and classification Mapping Awareness 3(1) (March/April 1989) 42-43
Lauersen, Vagn W (1988) The Danish geographic exchange format - A format for the exchange of topographic, cadstral, and utility information. Proceedings, XVI International Congress of Photogrammetry and Remote Sensing, Kyoto, July 1988. Commission 4.
Linington, P F (1989) File transfer protocols. IEEE Journal on selected areas in communication 7(7) (September 1989) 1052-1059
Maes, J F (1990) CEC CORINE Programme - Results and GIS experiences. in: MVA Systematica 1990.
Mumford, A M (1989) Application profiles for computer graphics standards - a touch of realism? Computer and Control Abstracts 24(277) Entry 33505, p 2348.
Mumford, A M (1991) Computer Graphics Metafile standard - an update Computer-Aided Design 23(4) (May 1991) 303-305.
MVA Systematica (1990) Report on Workshop on GIS. Brighton, 22 October 1990; Prepared for the Commission of the European Communities, DG XIII. Authors Christopher Nailer, Ian Johnson.
Nørgaard Nielsen, J (1986) Modeller og modellering (Models and modelling); in: Algreen-Ussing H, N O Fruensgaard, J Nørgaard Nielsen og E Stubkjær: Metode i projektarbejdet ved teknins-naturvidenskabelige uddannelser. Aalborg Universitetsforlag, Aalborg, Denmark. 290 p. (pp 72-149).
Salgé, Francois Geographic, cartographic, located data exchange formats: an overview of the existing solutions over the world. Proceedings, 13th Urban Data Management Symposium, Lisbon, 29. May - 2. June 1989. Vol 1, 331-344.
Scheller, A (1990) Usage of the SGML within the German Research Network DFN. Inf.ormations Tech.nologie _IT (West Germany) 32(3) (June 1990) 186-193.
Sowton, Michael (1990) Aspects of European Data Exchange; in: MVA Systematica, 1990 (no page numbers)
Statens Kartverk (1992) Samordnet opplegg for stedfestet informasjon - SOSI Et standardformat for digitale geodata. Versjon 2.0. 89p + bilag.
Stubkjær, Erik (1990) Communication Theory - A basis for data exchange and for measures of the quality of data. Proceedings, Third Scandinavian Research Conference on Geographical Information Systems, November 14-16, 1990 Helsingør, Denmark. Hermansen B, Balstrøm T & Frederiksen P (Eds) Danmarks Geologiske Undersøgelser, Miljøministeriet. ISBN 87-88640-66-3. Vol 2, 213-229
Stubkjær, Erik (1991a) Cartographic communication, graphic sign systems, and meaning. Proceedings, 2nd Polish-Dutch Symposium on Geodesy: Data collection for GIS (LIS) and their application, 2-5 April 1991, Delft. Organized by Faculty of Geodesy, Delft University of Technology, Delft, Netherlands and Faculty of Geodesy and Rural Land Development, Agricultural-Technological University of Olsztyn, Poland. Delft, April 1991. p 155-166.
Stubkjær, Erik (1991b) The development of national, multi-purpose spatial information systems - Danish experiences in a theoretical context. International Geographical Union's Conference: Geographic Information Systems - Multiple representation and multiple uses, April 22 - 25, 1991, Brno, Czechoslovakia. Computers, Environment and Urban Systems 16(2) (May/June 1992) 209-217.
Stubkjær, Erik (1991c) Organizational issues related to spatial information systems - a research review. Skriftserie No 72. Department of Development and Planning. University of Aalborg. September 1991. 11 p.
Tischler, J Z (1984) Nuclear Instruments & Methods in Physics Research. Section 8. 222(1-2) (1984) 339-40.
Østensen, Olaf M. (1991) Data standardisering og dataudveksling (Standardization of data and of exchange formats) Proceedings, AM/FM-GIS Nordic Region Conference, 28-30 October 1991, Mariehamn, Finland.