AbstractThis thesis investigates the application of fuzzy logic to geospatial problems. Fuzzy logic provides a means for dealing with vague information and sparse datasets inherent to many real world applications. A fuzzy site suitability analysis for prawn farming on remote Groote Eylandt demonstrates how fuzzy logic concepts can be incorporated into maps to facilitate site selection.
Another application presented relies on a published chemical plant operation dataset to illustrate how data driven modelling enables objective predictions on the basis of available information. In a third application, records of environmental variables associated with the foraging patterns of elephant seals are reinterpreted. The fuzzy rule-based data driven analysis of this small dataset of high dimensionality leads to an unambiguous conclusion more efficiently than other methods described in the literature.
Knowledge driven fuzzy rule-based modelling is illustrated by case studies aimed at improving the sustainability of the Timor Reef fishery. Firstly, the knowledge of experienced fishers is recorded to evaluate the fishing power of the Timor Reef fishery fleet to assess the need for recalibrating an existing map of productivity. Near constant fishing power across that fleet suggests that recalibration is not necessary. Secondly, a published fuzzy rule-based expert system is implemented. It can estimate fish susceptibility to fishing pressure from biological parameters only. The susceptibility to fishing pressure of target species of the Timor Reef fishery, estimated with this fuzzy expert system, differs from that estimated from globally averaged parameters. These discrepancies highlight the importance of local models for the development of sustainable fisheries.
Case studies in this thesis highlight the potential of fuzzy rule-based modelling in complementing statistical methods applied to spatial problems particularly when the uncertainty of the data is undefined. That potential is noteworthy in marine natural resource management.
|Date of Award
|Diane Pearson (Supervisor) & Stefan Maier (Supervisor)