ABSTRACT
Forested ecosystems in mountainous terrain show significant heterogeneity in atmospheric and land-surface interactions. Forest ecosystem simulation modelling requires management of geographically based information detailing these complex interactions. We present a knowledge-based approach for a geographically based information system designed to infer and simulate objects using simple attributes. Higher order objects are recursively defined using query models and aggregation. They are organized and accessible as distinct, identifiable landscape units such as hillslopes and forest stands. Object-orientated and knowledge-based tools simulate ecosystem processes on the basis of higher order geographic features. Ecological simulations are organized around a model of terrain based on a hierarchy of nested watersheds, hillslopes and streams. We organize spatial relations into a graph of topological links around which is built a graphical query interface. The interface translates user requests into database queries using a query model that maps graphical objects to database objects. We show how the query system integrates structural, logical and procedural knowledge to provide answers to complex queries based on simulation results. We demonstrate that this system insulates the user from implementation details of the simulation system, and that it can manage the ecological I spatial objects in a transparent manner throughout simulation experiments.
