GIS - standards in CEN and ISO

Introduction

Language is not only a subject for standardisation but also a means: standards are written down in documents, e.g. GIS object classification requires a well-defined semantic definition. Examples of standards can be found in ancient building and constructions, e.g. the Egyptians (1500 BC) paved the corridors in their houses by tiles with a fixed format and the Phoenicians (1000 BC) built ships of fixed sizes; in Europe Charles V (750) introduced a standardised handwriting (i.e. the Carolinian script, similar to the contemporary block letters). A main example is the definition of a useful unit of length which has been a topic of debate throughout centuries. The replacement of the many existing units of length in different European countries by the meter is one of the greatest successes of standardisation.

Fig. 3.1 Standardisation institutes.

The industrial revolution of the last centuries introduced an increase of scale requiring standardisation for technological and political developments. This standardisation was beyond applications and country boundaries. Around 1900 the technical standardisation became into development (e.g. BSI 1901) and nowadays almost all European countries have national standardisation institutes and eighteen of them are members of the CEN (Comité Européen de Normalization, Brussels in Belgium) i.e. Austria, Belgium, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, The Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and the United Kingdom. The EC council decided in 1986 (decision 87/95/EC) on the application of European standards at international level.

Globally there exists the ISO (International Standard Organisation, Geneva in Switzerland). For the standardisation for Electronics and Telecommunications there exist a separate network of standardisation institutes (see fig.1).

Within a country, standardisation is applied at three levels and for each level more detailed standards should be developed:

• general level (possibly national level);
• branch level;
• application level.

Standardisation, aim and definition

• quality improvement. Many data in GIS are of different homogeneity. In order to improve the intrinsic relative quality clear and well defined quality concepts should be defined;
• efficiency. Standards will lead to the improvement of the data transfer avoiding duplication of data capture and processing;
• to avoid information loss. Common standards will avoid the loss of information that occurs when transforming data from one system into another;
• knowledge transfer. Standards contribute to clarify the different aspects of all applications concerned. Users will understand their mutual requirements better when transforming from one to another system.

Though only standardisation and standards are used in English, in other languages norms and normalisation are frequently used as synonyms, although they may describe different concepts. Whenever both words occur in a language, the word standard has usually a wider meaning:

• norms (and normalisation, as the process to produce norms) is used for the official defined agreements that are applied, while
• standards (and standardisation, as the process to produce standards) is used for all type of agreements between users for mutual data transfer [4].So, the application of standards requires a complete acceptance and free agreement of the standard by both the supplier and the receiver of data in order the ease the transfer of information between them and to improve the quality and safety of the transfer.

Another problem in terms is the use of the words data exchange and data transfer. In general data exchange should be avoided: data goes from one user to another (transfer) and no data returns back from the recipient to the original sender. So usually nothing is exchanged but only transferred; in cases where exchange of data does really take place, i.e. data goes from sender to recipient and the recipient sends (even if it is the same) data in return one may also speak about a bi-directional data transfer.

It is difficult to determine the best moment introducing a standard. Obviously standards are required whenever user(s) have a need for it. But at the time that parties involved feel the need to introduce a standard in their application, they have undergone such a high level of development that the implementation of a new standard will create a fall-back in application or at least will be experienced as annoying to an experienced user, while a new user will not easily understand why standards should be developed let alone be applied. However, one should always realise that standards will create many new possibilities seen from a more general point of view in the arena of geo-information.

Besides, in the field of IT, standards have been applied very successfully, including graphics standards which are also of great importance to the field of GIS: many systems use PHIGS, DXF, IGES, etc.

In this presentation we will discuss the standardisation of data as opposed to the standardisation of the processes. However, in many cases the standardisation of processes will reflect quality control applying the ISO 9000 series of standards for defining requirements in the different sub-processes in order to certify the quality of the end products. One should realise that no standardisation of processes can occur without the standardisation of the data. The standardisation of data is the primarily goal of the standardisation proposals of the CEN/TC 287, while the ISO/TC 211 focuses more on standardisation of processes (see fig. 2).

Fig. 2 Mutual relation between the standardisation of data and processes.

First some theoretical background required for the process of standardisation will be discussed, the technical aspects that are involved when developing and using standards and the political and organisational impact of standardisation within a country in relation to continental and global developments.

Being member of CEN TC/287 (Technical Committee 287 of Européen Comité de Norma-lisation on "Geographical Information") the developments within the (sub-) working groups will be presented.

Theoretical background

• object type descriptions. Reality is supposed to be a set of mutual depending objects which are modelled by the description of the objects into entities, attributes and the mutual relations. These descriptions are made by a future user and as such are user-subjective;

• application dependency. Because of the application specification in defining the nominal ground only the objects of interest for a specific user will be considered, avoiding all other objects (which may be of interest for other users. But who is paying for the capture and updating of those objects? ).

Because of the specific geographical character of geographic objects their attributes should include:

• semantic attributes, representing the meaning of all objects by human interpretation;
• geometric attributes. They are the essential characteristics that distinguishes geographical objects from other database objects (though they may appear as well in raster as in vector format). Because of the k-dimensionality of geographic objects special geometrical structures have to be designed in order to allow to store in existing 1-dimensional databases. Geometric attributes indicate:
  • position, in a pre-defined geodetic co-ordinate-reference-system;
  • topology, to relate mutually the neighbourhood between objects;
  • shape, in mathematical interpolation terms;
• graphics, the store the representation of objects by line-type and -width, point- and line-symbolic, area-patterning, colour, text-size and-style, etc.;
• descriptions by text, photographs, drawings, video and other type of pictures. These attributes are usually stored in other databases because of their different nature. The link is made by unique keys in both databases.

• quantitative attributes, measured and expressed in a pre-defined unit as length, temperature, time, etc., or
• qualitative attributes, where each attribute can have only one values from a pre-defined set of values as names, addresses, colours, etc. A discrete value is nominal when there is no ordering possible e.g. forest, house, red, etc. or ordinal when a (mathematical) ordering is possible e.g. house numbers in addresses or alphabetical ordering of names.

• specification of the quality aspects;
• specification of the required quality.

Occurrences of entities in the database can be attributed accor-ding to their properties however often also cartographic represen-tation attributes are included. Graphical definitions such as line styles, point- and line symbols, line thickness, area patterns, colour, texts are often found either as attributes or as separate mm occurrences in the data base (see fig. 4). They are mixed with the actual topographical information in order to enable to make maps out of the data base but in fact, with these cartographic symbology it is only possible to make one certain type of map (only the one that is defined in the data) from this data base. If the data base serves a multitude of applications a clear distinction should be made between the nominal ground (i.e. the ideal set of topographical information of the terrain for all intended applications) and the graphical presentation of the nominal ground using a separate symbology catalogue for every map. The transfer standard should enable to transfer both.

GIS standards

• terms used in the standards and the definition of objects (entities) and relations and their attributes;
• reference systems for position and semantics Metric position can be expressed in vector mode by geodetical point co-ordinates or in raster mode by grids, pixels, voxels through their row- and column number. From theoretical point of view the elementary differences between positional vector- and raster model refers to the implicit and explicit definitions of the entities' characteristics (see table 2);
• conceptual models, including the general level-1 IRDS-models and the sub-models for semantic definitions, geometry, quality and metadata (including quality), last three being special for GIS.

• · functions and predicates for updating and querying the database;
• · data transfer format and mechanism for transferring data from one system to another. Special transfer formats require the transfer of files and records, while the logical model uses coverage, sub-coverages, levels, modules, entities, relations, attributes, etc. The general transfer mechanism contains definitions how to convert types of a logical model into a physical transfer format;
• · application schemes; each application may consists of several datasets for which the schemes should be designed in respect of metadata, semantic, geometry, quality and position reference systems;
• · overview, indicating the function and the use of the different GIS standards.
• · often digital geographic data is used for displaying in maps either on screen or on paper. In both circumstances standards for representation of objects is required forcing a standard symbology.