A Taxonomy of Strategies for Developing Spatial Decision Support System

Abstract

Due to their extension and complexity, problems concerned with geographic space are usually classified as ill or unstructured. For these reasons decision-making processes (DMP) that conclude with the selection of optimal or satisfactory solutions require effective and efficient means of support. Spatial Decision Support Systems (SDSS) are computer systems developed to support DMP in which the problems have geographic dimensions and whose structure is complex or impossible to delineate. These systems are functionally composed of data and scientific models managed with the aim of providing maximised support to DMP. The component represented by spatial data is one of the main obstacles that have to be overcome for SDSS to give effective support. Geographic Information Systems (GIS) have been a paradigm in SDSS development strategies, fundamentally due to their capacity to collect, store, and handle spatial data. The scientific modelling component, represented by mathematical models of natural physical processes, usually is implemented in SDSS through specific software subsystems. Especially in the last seven years there has been great scientific interest in SDSS accompanied by a proliferation of adjacent technologies. Some authors have attempted to classify SDSS without, however, reaching a generally accepted proposal. This lack of clarity has made detailed analysis of existing systems and development of new projects more difficult. The aims of this research were: a) identify and analyse the main variables that determine current application software development methodologies in the field of SDSS; b) provide a taxonomy of these methodologies, with the objectives of being generic, general, and practical; c) identify and analyse the strategy that presents the greatest flexibility to develop effective SDSS. The research was based on a bibliographic survey and is part of the doctoral thesis of first author. The paper presents and criticises relevant issues about different development strategies and the respective systems produced. Three main variables were identified and guided the development of the taxonomy of the strategies. The paper proposes five taxonomic classes of coupling GIS technology and scientific modelling subsystems. These classes are defined and the paper argues that they are sufficient to categorise the main current methodologies as well as suggest places to expect new technologies. Each class is also exemplified with a number of SDSS applied to the watershed DMP domain. Within these classes, the research identified and analysed the one most likely to show the greatest flexibility to develop effective SDSS