Andrew J. Marr, Stephen MacDonell, and George L. Benwell
Spatial Information Research Centre,
Department of Computer and Information Science,
University of Otago
PO Box 56
Dunedin, New Zealand
Tel. +64 (3) 479 8301
Fax. +64 (3) 479 8311
WWW: http://divcom.otago.ac.nz:800/sirc/
E-mails: ajmarr@commerce.otago.ac.nz, stevemac@commerce.otago.ac.nz, gbenwell@gandalf.otago.ac.nz
 
Recent developments in GIS interoperability, particularly those emerging from the work of the Open GIS Consortium, are likely to result in significant improvements in the utilisation of spatial data resources. Current research into the development of OpenGIS interfaces which enable interoperability among heterogeneous spatial databases will offer fundamental improvements over traditional methods and techniques.
Parallel to the development of interoperability, is the increased demand for research into Spatial Process Modelling Systems (SPMS). SPMS represent the integration of GIS analysis with the functionality found in non-spatial simulation modelling tools (e.g. Stellar and ExtendOE). This is sometimes referred to as Geographical Modelling Systems (GMS). The integration of the two systems is difficult to achieve, and example applications to-date are typically too specific to the component subsystems and narrow in focus. A set of important emerging research themes in this area focus on the desire for generic and interactive visual construction tools that facilitate collaborative process model development. Further work is also required into modular hierarchical structures, and modelbase management systems that permit successful embedding of model structures in the form of composite model components.

For a flexible SPMS, there is an inherent requirement that the simulation of spatial modelling events occur using a variety of data sources, usually derived from a diversity of heterogeneous data formats. With interoperability, the restrictions caused by fixed proprietary vendor data formats are removed. In addition the capability is developed for querying of desired data subsets for retrieval. These attributes of interoperability highlight the virtual dependence of generic SPMS on such technology in a collaborative environment.

A fully integrated SPMS is a system that encompasses existing GIS principles and techniques but also includes extensions for additional spatial process modelling functionality. These additional simulation modelling requirements include consideration of feed back systems, support for temporal dynamic modes, equation generation, and model calibration and validation. In addition the complexity of some process model definitions may become too overwhelming. Some authors have asserted the need for the incorporation of AI and agents to guide users through the processing steps.

In this paper, a conceptual framework for the collaborative sharing of models describing spatial process structures is presented. The proposed functionally independent modules in the context of SPMS are constructed in the form of reusable services. These services are: graphical model design, model interpretation and implementation, data transformation, spatial operations, and visualisation. This set of services are viewed as the principle components required for basic system operation. It is assumed that other specialist components will also be included as required (e.g. DTM and network analysis). Each service is designed to be replaced as new technology isintroduced with minimal impact on the existing system.

The graphical model design service is composed of a graphical interface in which the user may visually construct the required spatial process model. The model is then written to a process modelling language design file. This generated file may then be exchanged and integrated as sub-components in other designs. Once a process model is complete it will be sent to the interpretation and implementation service. This service will then act on the design file and call supporting services such as data transformation and spatial operations as required. The current proposal for data transformation is to use the feature manipulation engine (FME) software which facilitates powerful interoperable functionality between diverse systems. While this software is suited to back-end mapping operations, there are limitations in the types of data that may be utilised and in the processing capabilities. When the objectives of the Open GIS project are realised, it is hoped that this service may be enhanced. To enable cross-platform portability, the intended system and graphical interfaces are being designed in Java.

The aim of the project is to develop a tool for integrated spatial process modelling, that is in-line with the with the identified research goals in this area. The proximity of objectives between Interoperable GIS and SPM are noted and seen as mutually supportive. The objectives of both concepts stress the desire to develop distributed `plug and play' tools that manipulate spatial data of variable formats. The incorporating of interoperability into SPMS is crucial if maximum flexibility and functionality of the system is to be achieved. In addition,the development of this system is also seen as providing methods for collaborative research.