Society for Ecological Restoration - Northwest Chapter Conference

Presented by Lance W. Vail, Pacific Northwest National Laboratory

Abstract
Fuzzy logic has been utilized to develop an improved approach to estimate habitat diversity indices for water and land use management alternatives. The need for an improved approach was identified during the Multi-Species Framework Project. Fuzzy methods provide a compact and flexible representation that accommodates the pervasive imprecision of process understanding.

Natural resource managers are constantly attempting to balance multiple conflicting, incommensurate objectives in an environment characterized by high uncertainty, varying data quality and availability, and competing models and approaches. The reliability of natural resources management policy depends on the integration and interaction of measurements, response models, process models, and policy models across the variety of temporal and spatial scales each represents. The potential exists to dramatically improve the linkage of physical and biological assessment models, an imperative to supporting a science-based approach to managing natural resources.

Innovative approaches are needed to overcome the structural differences that typically exist between physical models and biological models. Biological models used in natural resource management are structurally different from physical models. Where physical models rely on formulas defined over a continuous range, biological models are based on categorical statements about the habitat. Fundamental scale and structural differences between physical and biological models significantly complicate their linkage. The spatial and temporal scales of biological models are typically considerably different from the scales associated with most physical processes. For instance, a forest succession model will typically involve time scales of decades, whereas the hydrology and climate parameters required as input to the forest model will vary on the scale of hours.

Fuzzy logic is an extension of multi-valued logic based on the concept of fuzzy sets. Fuzzy logic has provided a methodology for computing that is currently used for many applications. Fuzzy set theory relates to classes of objects with unsharp boundaries where membership in the set is a matter of degree. For instance, the boundary between 'optimal habitat' and 'good habitat' is generally fuzzy. Fuzziness describes the degree to which an event occurs, not whether it occurs. The uncertainty of an event occurring is an issue for probabilistic methods. Fuzzy methods do not replace probabilistic methods but complement them.

The feasibility of an advanced tool for bridging physical and biological models which incorporates fuzzy logic has been established. Fuzzy representations allow consideration of the multivaluedness associated with spatial upscaling and the vagueness associated with rules used to categorize physical and biological aspects of habitat. Neural networks will be used to combine sparse site-specific knowledge and general process knowledge to provide fuzzy estimates of habitat parameters.

^ top