Fuzzy-logic modeling of land suitability for hybrid poplar across the Prairie Provinces of Canada

Determining the feasibility of a large-scale afforestation program is one approach being investigated by the Government of Canada to increase Canada’s potential to sequester carbon from the atmosphere. Large-scale afforestation, however, requires knowledge of where it is suitable to establish and grow trees. Spatial models based on Boolean logic and/or statistical models within a geographic information system may be used for this purpose, but empirical environmental data are often lacking, and the association of these data to land suitability is most often a subjective process. As a solution to this problem, this paper presents a fuzzy-logic modeling approach to assess land suitability for afforestation of hybrid poplar (Populus spp.) over the Prairie Provinces of Canada. Expert knowledge regarding the selection and magnitudes of environmental variables were integrated into fuzzy rule sets from which estimates of land suitability were generated and presented in map form. The environmental variables selected included growing season precipitation, climate moisture index, growing degree days, and Canada Land Inventory capability for agriculture and elevation. Approximately 150,000 km2, or 28% of the eligible land base within the Prairie Provinces was found to be suitable for afforestation. Accuracy assessments conducted with fuzzy accuracy methods provided a more descriptive assessment of the resulting land suitability map than figures generated from a more conventional Boolean-based accuracy measure. Modeling, mapping and accuracy assessment issues were identified for future extension of this work to map hybrid poplar land suitability over Canada.