MSc candidate, 2011 - present
Meaning-based landscape classification procedure using landscape and local scale terrain attributes
Using a 25m resolution digital elevation model (DEM), a preliminary procedure for the automated classification of landscape features was developed through the implementation of expert-knowledge and fuzzy logic sets for the NTS 92G map section of British Columbia. Local morphological terrain attributes such as plan curvature, profile curvature, total curvature, and slope were used to identify concave, convex, and depression landscape characteristics.
Additionally, a recursive-multiple-flow-routing algorithm using the SAGA Modified Catchment Area was used to calculate a relative hydrological slope position index which served as the landscape scale model input. Fuzzy attribute rules and weights were derived from the semantic meaning of each landform class using ArcGIS and ArcSIE. This procedure resulted in the identification of 10 unique landforms defined by flow characteristics and slope position. Model results, in the form of hardened maps, were semantically valid within the range of rules used to define each landform, and illustrated the potential for relatively simple conceptual models to identify landscape features of interest to soil mappers in a complex landscape, but the model did not appear so useful in flat landscapes. Future works in this project should include the assessment of neighbourhood size in calculating local terrain derivatives; the refinement of landscape classification rules for low relief terrains; and the linking of landscape classes with soil attributes.The proposed method can aid in digital soil mapping and ecological mapping, as well as contribute to the fields of hydrology, geomorphology, and geomorphometry.