Architecture of an information system in model-based environmental problem solving

Environmental problems are often complex as they typically deal with several disciplines and affect large regions. Tackling such problems requires integration of computational tools. Solving environmental problems requires cooperation between the groups that develop the tools and the groups that use the tools and transfer of information and technology. The process is becoming more and more complex as the problems we try to solve become larger. We are faced with the task of designing an architecture of an information system (IS) for managing the problem solving process. Attempts to overcome difficulties in solving complex real-world problems, such as environmental problems, include advances in the theory of complex systems, development of information system frameworks, and development of open standards. The theory of dynamic networks gives us tools to understand and possibly design better and more robust and resilient systems (Watts 2003). The idea of an information system framework for environmental modeling is an answer to the general need of linking a model to a database, to another model, to a GIS, or to other tools. In the field of geoinformatics there are several projects that develop open standards, while in the field of environmental modeling there are fewer such projects. In spite of these efforts there is still a need for new approaches, which link and help to link work carried out in different communities. In this paper we discuss the linkages between simulation models, environmental data and end users, and make an attempt to define the design processes of an IS architecture in model-based environmental problem solving. We present key points of the development of a DSS for the CLIME (Climate and Lake Impacts in Europe) project. The general IS development process is often divided into five phases: problem analysis, architectural design, development, implementation, and evaluation. In the analysis phase the problem in question is analyzed and an understanding is developed. In the design phase the architecture of the IS is prepared. The development phase deals with the actual construction of the IS. In the implementation phase the constructed IS is transferred to the work environment in which it will be used. The evaluation is the real-world test of the implemented system. From evaluation the process goes back to analysis. The environmental problem, which the CLIME project studies, is climate change and its effects on lakes and the impacts of the changes in the lakes to the society. Besides the environmental and socioeconomic domains, there are also other domains as the future DSS users and the scientists within the project. The CLIME project has a designated, but heterogeneous, end-user community, which is the primary target user group of the DSS. The architecture of the CLIME DSS is at the highest level a network of four nodes in chain. The offline part of the system produces data, which is stored in a database. The user has access to the information in the database through the online part of the system. The information exchange is two-way in each connection, but, because of the structure, the user does not have a direct access to the database or to the offline system. This architecture is a, very generic, solution to the knowledge transfer problem. Besides providing a platform for causal Bayesian networks, the design of the CLIME DSS includes visualization capabilities. The visualizations are prepared in the offline part of the DSS, stored in the database, and used in the online part. Our conclusion is that the problem analysis and architectural design phases are not given enough thought considering how important they are. One reason for this may be that the methodological tools are not yet fully matured, which is understandable as the fields of problem analysis and software architecture are still rather new and developing. New innovations in the above fields could provide a promising avenue for integrating methods used by modelers, software developers, and decision makers in environmental problem solving.