A knowledge-based, spatial, multi-criteria decision analysis methodology for pond site evaluation

Abstract

The use of knowledge-based systems (KBS) has been increasingly recognized as a way to combine scientific understanding of the processes of the natural world with the heuristic rules developed by managers through observation and experience, even when the processes behind those observations may not be fully understood. KBS combine applied knowledge acquired from local experts and published literature with a model of the human reasoning process or "ontology," presenting the accumulated knowledge in a synthesized manner. KBS can provide greater accountability, transparency, and consistency in these decision-making processes, which can be represented as a series of rules, and can support regional-scale natural resources planning and management. The use of geographic information systems (GIS) can be an invaluable tool in the evaluation and analysis of natural resources problems, which often involve spatial relationships. Because natural resources problems are frequently complex and multifaceted, a multi-criteria decision analysis (MCDA) approach may be necessary to address these issues in a synthesized and integrated manner. As demonstrated in this research, the weighted average aggregation algorithm for an MCDA is mathematically similar to the use of certainty factors in a forward-chaining, data-driven KBS. Using a weighted average MCDA as the "reasoning" or "inference" engine within a KBS provides an increased level of flexibility in a rule-based system. This research integrates a weighted average MCDA within a forward-chaining KBS to enable a natural resources management evaluation and comparison of potential pond development sites based primarily on local knowledge that has been captured in knowledge and rule bases. The KBS has been programmed in an Excel database with links to an ArcView GIS spatial database and spatial analysis tools in a prototype model called the Waterfowl and Augmentation Pond Site Assessment Model. The prototype model application was for the evaluation of potential pond sites to be developed for waterfowl habitat and managed groundwater recharge for streamflow augmentation under Colorado's prior appropriation water law.