The Authoritative Topographic-Cartographic Information System (ATKIS)

Abstract

Recent years have seen an added emphasis in Geographic Information Systems (GIS) on better conceptual data modelling and data transfer standards. A major achievement in this area is the United States Spatial Data Transfer Standard (SDTS). This standard represents one of many worldwide attempts to standardise the way in which geographic information is modelled and transferred. It is currently being reviewed for adoption in Australia. In a previous paper [Hesse and Williamson, 1993], the authors questioned the appropriateness of such an approach and suggested alternative approaches which should be considered to better reflect Australasian needs.

This paper describes an interesting and different alternative from Germany, the Authoritative Topographic-Cartographic Information System (ATKIS), together with a working implementation of the model using an object-oriented software system.

Introduction

A recent article by Hesse and Williamson [1993] provides background on the importance and range of applications of data standards. The paper highlighted the difficulties in using current data standards for GIS applications.

The paper suggested that at a more fundamental level, the conceptual data model used for a GIS can become a de-facto standard in laying the foundation for the means of data exchange between data supplier and data user. The data items (features) that form the conceptual data model have to be named and itemised typically with a unique feature identifier. Their valid attributes (if any) and attribute values (valid type or range) can be standardised. The standardisation of attribute values can be done even prior to the existence of any strategy for an overall conceptual data model or transfer mechanism.

In order to understand the systems and processes described in this paper, the data abstraction framework of Guptill [1991] is set out below:

Reality are the total phenomena as they actually exist.

Data reality is an abstraction of reality that includes only those entities thought to be relevant to anticipated needs. It is a definition of the scope of the data.

A data model specifies the sets of components and the relationships among the components pertaining to the specific phenomena defined by the data reality. A data model is independent of specific systems or data structures that organise and manage the data.

A data structure specifies the logical organisation of the components of a data model and the manner in which relationships among components are to be explicitly defined.

A file structure is a set of rules that specify the logical implementation of a data structure within computing system environments

Guptill further points out that

"...an important point to note is that many different data structures may be generated from one model ..."

The definition and standardisation of the conceptual model, its transition into data structures and the means of data exchange are very complicated issues which can take many years for development, publication, testing, revision and final approval; in some cases more than a decade. It has to be noted that many state agencies around the globe are currently considering or implementing their own spatial data model designs, Guptill lists the U.S. Bureau of Census, the U.S. Defence Mapping Agency, the Institut Géographique National in France, the Landesvermessungsamt in Germany and the United Kingdom Ordnance Survey.

A singular 'global' data model and transfer mechanism does not seem to be in sight.

Noting this, Hesse and Williamson questioned the Australian decision to adopt the U.S. based SDTS [ALIC 1990]. They stated that such adoption requires a careful examination of the specific local Australian needs for a spatial data model and transfer standard, and justifies the study of possible alternative approaches. They suggest that an appropriate alternative could use many of the concepts in ATKIS.

The objective of this paper is to describe ATKIS and to investigate its applicability for Australian conditions.

Background and Current Status of ATKIS

ATKIS - the "Authoritative Topographic - Cartographic Information System" is the result of a multi-stage research and development project of the German State Survey Working Committee (AdV) in the fields of state survey, cartography and automation. The Survey Authorities will use ATKIS to provide for the changing demands being made on the state survey. The future digital storage of a common nationwide topographic information system will assure that the data, which is captured and maintained with a significant amount of public money, will keep its significance for administration, the economy, the legal system and research.

The current status of ATKIS is summarised in a report, based on a translation [Hesse and Leahy 1990] of official AdV documents [AdV 1988, "Gelbes Heft"]. Detailed descriptions are available in separate documents [AdV Arbeitsgruppe ATKIS 1989].

ATKIS will contain a nationwide standardised digital topographic data base with a three-dimensional structure. The data base will be built in the public interest and transferred to public and private users as a state service. The data base will be authentic and current. ATKIS is based on a modern cartographic model theory, as shown in Figure 1.

The real world with its topographic objects and its thematic structure is viewed as the original. The topographic survey creates the primary model, a topographic landscape model. This model is transformed through cartographic work into a secondary model, the cartographic landscape model. During both these stages generalisation processes at various levels take place resulting in different degrees of detail. A tertiary model is built by adding user specific original data models or other derived models.

The creation of the primary model of the landscape takes place in the form of a three dimensional survey, based on topographic objects and object parts and their relationships. The objects are captured by shape, position and their topological relationship. Attributes are also assigned, coded and stored. Through this process of topographic modelling, a digital topographic landscape model is created - in ATKIS terms, a Digital Landscape Model (DLM). Depending on the data capture accuracy and degree of structure, DLM will differ in degree of information density.

The creation of a secondary model is based on the DLM as the primary model. The stored topographic objects are assigned with symbology according to cartographic generalisation rules. Through this process of cartographic modelling, a digital cartographic landscape model is created - in ATKIS terms, a Digital Cartographic Model (DKM). Depending on the symbology catalogue and its degree of generalisation chosen, a number of DKM versions for any map scale can be produced.

For the association of objects to object type, an object type catalogue (ATKIS-OK) is used. For the association of object type to its map symbology a symbology catalogue (ATKIS-SK) is used.

The overall concept of ATKIS and the various information flows between the original landscape, the DLM and DKM, the digital output and the user is shown in Figure 2:

The ATKIS data model structures the landscape by objects and object parts. Objects are built in a hierarchical structure and stored as object parts with geometric boundaries, attributes and relationships. The AdV intends to built the ATKIS data model independently, on the base of the Automated Cadastral Data Base (ALK) data model. The recommended data transfer standard is the 'standardised data base interface' (EDBS) of the ALK project. The possibility of integration with the European Transfer Format (ETF) standard is currently being investigated. ATKIS will be built in stages according to demand. Priorities will be given to the DLM over the DKM and more specifically to the DLM 25 (based on the content of the 1:25,000 Topographic Map Series), DLM 200 (based on the 1:200,000 Topographic Map Series) and the DLM 1000 (based on the 1:1,000,000 International World Map series) and in the order given.

There seem to be three distinct differences to the current SDTS proposal:

(1) Closer study of the concepts of the EDBS transfer standard used for ATKIS data, reveals a far more process-oriented approach to spatial data transfer, compared with the only 'data-oriented' SDTS proposal. ATKIS allows for incremental data updates in the form of user 'request posting' through the EDBS. For more details on the EDBS functionality see Stöppler [1989].

(2) ATKIS allows for cadastral links in its data structure (see Stöppler [1987]) and has a rich modelling capability in terms of object-hierarchies and complex objects. Its conceptual model is not fixed to any particular database implementation such as the relational model.

(3) ATKIS includes the modelling of cartographic objects based on (spatially accurate) digital landscape objects and provides for rule based generation of multiple cartographic representations.

During 1990, the State Survey Organisation of Hessen (HLVA) in Wiesbaden was evaluating two commercial systems for their ATKIS suitability: The Siemens-SICAD system, a conventional CAD system with external attribute data base, and Intergraph´s object-oriented TIGRIS GIS. Both systems were capturing topographic data sets for a true comparison of their performance and data model suitability. Emphasis was put on user interface aspects. The work station operators were changing between systems during the test phase in order to achieve an unbiased evaluation. This was done prior to conducting any bench-mark procedures. Only TIGRIS allowed the easy ‘natural’ extension of the conceptual object data structure into an object-oriented programming system (OOPS). The OOPS methodology should be of particular interest for the automatic derivation of DKM models (for example rule based generalisation problems).

Until July 1990, the State Survey Organisation of Hessen (HLVA) in Wiesbaden have experienced a data volume of approximately 25 Mb for one DSM(1)25, derived from up to 35 DGK5. DSM(1) means a selection of the main objects only and does not include terrain data. Some 200 TK25 cover the administrative area of Hessen which would result in a data volume of approximately 5 Gb. Hessen has an extensive road network and a population of 6 million people on 22,000 km2 (273 people/km2). These figures together with the data volume for the incomplete DLM give an indication of the overall data volume of the final ATKIS system. The creation of ATKIS will be a long and evolving process. Every participant realises that the creation of the ATKIS data base with several Digital Landscape Models (DLM) and the derivation of different Cartographic Models (DKM), will be a long term project with its final realisation in the 21st Century.

ATKIS Implementation using TIGRIS

TIGRIS

The very existence of entirely object-oriented GIS (OOGIS) is denied by the followers of more traditional GIS implementations based on the relational map layer approach. For example, Morehouse [1990] finds

"... To my knowledge, there are no true object-oriented geographic information systems in existence, contrary to the claims of numerous marketing organisations. To be truly object-oriented, a system must support the development of a specific geographic data model from a generic one by means of inheritance and encapsulation ..."

However, the Topologically Integrated Geographic Resource Information System (TIGRIS) was the first commercially available GIS entirely based on object-oriented concepts with combined spatial and non-spatial data storage and interactive topology. It allows the easy creation of a specific (user-defined) data model from generic (vendor-supplied) components and makes use of inheritance and encapsulation.

The Administrator module allows the interactive graphic design of a data base schema consisting of the schema elements themes, composite features and base features. The schema is a graphical representation of the users view of their data (and it's attributes) and the relationships between the data. The design of the schema is entirely determined by the user to suit the needs of the GIS application. A generic sample schema is shown in a simplified way in Figure 3:

After defining the schema elements, the relationships between them, their graphic symbology and additional cartographic and geodetic information specific to the geographic location and extent of the project area (map projections and transformation parameters), the schema information is permanently stored in a data dictionary for later use. A description of the first TIGRIS installation in the Asia-Pacific region and first positive user impressions, especially in the data base administration area, can be found in [LINZnews 1989].

Within the schema hierarchy, themes represent the highest level of data abstraction (for example vegetation, transportation, hydrography). Composite features are the next level of data abstraction and represent a classification of a group of physical entities. These need not make up an actual physical entity on the map (for example railway, river, shopping complex, political boundary). Base features are the actual individual entities shown on maps and can be combined into composite features or themes. They are at the lowest level of data abstraction and can be of type point, line (oriented line) and area. All of the above schema elements (themes, composite and base features) can carry additional attributes (such as surface, width, name, water quality...) of different types (for example integer, real, character, coded, etc...). Coded attributes assure data integrity by only allowing the selection of certain user-defined values (for example the code list for the attribute surface_type of the the base feature highway_segment allows for "asphalt", "concrete", "gravel" and "other").

The establishment of connections between the various schema elements defines owner-component relationships. This approach goes beyond the value based approach of the relational model, where relationships are created only through the existence of common key (or link) attributes. These are described by Herring [1989] as 'object-oriented' in nature:

"... An "object-oriented system" is one which supports an abstract concept "object" ("entity", "feature") having existence independent of any attributes that entity may or may not have. The opposite is a "value-oriented system" which models only attributes (a classical RDB)."

The following example of a TIGRIS schema [Intergraph 1989] in figure 4, further illustrates the object-oriented principles of the approach.

Base features can belong to any number of themes and composite features but can be composed of only one type of topology which can be shared with other base features (for example a river can also be a political boundary).

Other utility schema elements are abstract attribute groupings and unions. For example, in figure 4 the schema elements "rr_bridge" and "hwy_bridge" could share the attributes "name", "bridge_type" and "weight_capacity" in form of the abstract attribute grouping "bridge". The attributes defined for this grouping are inherited by all schema elements with which the grouping is 'unioned'.

The sharing of schema components, the ownership relationships in the form of owner-to-component associations and the resultant inheritance of attributes, demonstrate the object-oriented design of TIGRIS and shows a much richer modelling capability than the conventional relational data base model. For a detailed discussion of extensions to the RDB standard query language (SQL) for spatial analysis in a topologically-structured GIS, within TIGRIS, see Herring et al. [1988].

ATKIS/TIGRIS Implementation

The implementation of the ATKIS data model at the Department of Surveying and Land Information, The University of Melbourne, using TIGRIS, was done without any conceptual modelling problems. The schema is based on the original German effort done by the State Survey Organisation in Hessen, however some features were 'Australianised' and some extensions were made.

The TIGRIS modelling constructs of theme, composite feature, base feature and abstract attribute groupings and unions [Intergraph 1991] are able to reflect the ATKIS object-part, object, object group and object theme constructs. The are very similar, yet both developments were completely independent.

When the ATKIS designers started to develop the first concepts for the system in 1980 they were not concerned with the detailed database implementation details, only with the creation of 'the right' conceptual model. It has taken them 10 years to get to the first prototype system with initial data capture for a pilot project and a careful selection of a suitable GIS system - there is more to the establishment of a successful system of this dimension than the purchase of hardware and software.

Conclusions

This paper highlights the relevance of object-oriented design for general GIS data modelling and its impact on spatial data transfer standards. The U.S. Spatial Data Transfer Standard (SDTS) and its adoption for Australia should be critically reviewed under these aspects and within the Australasian LIS context, since the adoption of the SDTS will strongly influence the future directions of LIS/GIS in the region. The alternative German ATKIS data model and transfer standard and its object-oriented implementation with TIGRIS is presented and used to highlight some of the Australian spatial data modelling and data transfer requirements.

As stated in Hesse and Williamson [1993], there is a growing world-wide need for the definition of GIS data models and data transfer standards. While many different systems are currently proposed, there is little chance for a single 'world-wide standard'.

It appears Australia is committed to the SDTS, despite Australia's strongly developed digital cadastral 'culture' and SDTS's lack of it. At the same time there is the lack of a common vision in Australia between Federal, state and military topographic mapping standardisation efforts and minimal cooperation with cadastral mapping agencies towards a common standard. This leaves the user of spatial information in Australia with a number of different standards and problems.

The German ATKIS system represents some interesting alternative concepts to SDTS modelling efforts with its inclusion of cadastral data links and the reflection of digital landscape and cartographic objects. Its data transfer component is process oriented and the cartographic representation of the data is rule-based, allowing for multiple representations.

Object-oriented concepts inherent in the ATKIS allow the relatively easy implementation of complex conceptual data models (of a very complex reality), and appear to be a sound base for modern spatial information systems. This has been demonstrated with a straight forward TIGRIS implementation of the ATKIS model.

References

AdV 1988, Amtliches Topographisch - Kartographisches Informationssystem (ATKIS) - Das Vorhaben der Landesvermessungsverwaltungen zum Aufbau Digitaler Landschaftsmodelle und Digitaler Kartographischer Modelle , "Gelbes Heft", Arbeitsgemeinschaft der Vermessungsverwaltungen der Länder (AdV), Bonn

AdV Arbeitsgruppe ATKIS (Editor) 1989, Amtliches Topographisch - Kartographisches Informationssystem (ATKIS), Gesamtdokumentation, Arbeitsgemeinschaft der Vermessungsverwaltungen der Länder (AdV)

Australian Land Information Council (ALIC) (Editor) 1990, National Strategy on Land Information Management, Australian Land Information Council

Guptill S.C. (Editor) 1991, "An Enhanced Digital Line Graph Design", U.S. Geological Survey Circular 1048, U.S. Geological Survey

Herring, J.R. 1989, "A Fully Integrated Geographic Information System", Proceedings of AutoCarto 9, April 1989, 828-837, Baltimore, Maryland.

Herring, J.R., Larsen, R.C., Shivakumar, J. 1988, "Extensions to the SQL Query language to support Spatial Analysis in a Topological Data Base", GIS/LIS '88 Proceedings, 741 - 750, ASPRS, Falls Church, VA

Hesse, W., Leahy, F. 1990, "The Authoritative Topographic Cartographic Information System ATKIS", Consultancy Report for Intergraph Federal Systems

Hesse, W. Williamson, I.P. 1993, "Australasian Spatial data Transfer Standard Requirements", Submitted to The Australian Surveyor for publication.

Intergraph 1989, TIGRIS GIS Environment (TGE) Administrator’s Operator Training Guide, September 1989, Intergraph Corporation, Huntsville, Alabama

Intergraph 1991, TIGRIS/IGDS Database Administrator’s Guide, February 1991, Intergraph Corporation, Huntsville, Alabama

LINZNews 1989, "TIGRIS Arrives", LINZNews, No. 9, 6-7, DOSLI, Wellington, New Zealand.

Morehouse S. 1990, "The Role of Semantics in Geographic Data Modelling", Proceedings of the 4th International Symposium on Spatial Data Handling, 689-698, Department of Geography, University of Zürich

Stöppler, H.-W. 1987, "Die Automatisierte Liegenschaftskarte (ALK) - Überblick", Nachrichten aus dem öffentlichen Vermessungswesen, 64-88, Heft 2-3/1987

Stöppler, H.-W. 1989, "Die Einheitliche Datenbankschnittstelle (EDBS) im fachlichen Überblick", Nachrichten aus dem öffentlichen Vermessungswesen, 72-100, Heft 3-4/1989