Hydrogeomorphic edge detection and delineation of landscape functional units from lidar digital elevation models

A new method is introduced to delineate hydrogeomorphic elements from light detection and ranging (lidar) digital elevation models. Landscape segmentation is achieved using an edge-detection procedure to identify boundaries defining rapid changes in the tan α(d) index of landscape drainage potential. These boundaries define homogenous, functional landscape units that can be classified according to different topographically derived indices such as mean of the expected hydraulic gradient (approximated by tan αd), mean topographic wetness index, and mean ratio of flow path lengths to flow path gradients (L/G). Two case studies are presented in which the new method was applied (1) to map forested wetlands and nonwetland saturation-prone depressional areas and improve regression models of dissolved organic carbon source areas in the Muskoka-Haliburton region of south central Ontario and (2) to spatially characterize near-surface soil moisture patterns along a complex, upland hillslope catena in a small experimental watershed in northwestern Ontario. Both case studies point to the critical role of local drainage conditions and slope geometry in dictating spatial patterns of terrain wetness in complex Boreal Shield landscapes. The results cast uncertainty on the role of upslope contributing area as a first-order control on terrain wetness in this environment. Nevertheless, the results highlight the strong potential of digital terrain analysis to improve conceptualization of hydrological processes in the Boreal Shield, and they are conceptually consistent with the emerging paradigm of runoff generation in this region. We propose hydrogeomorphic edge detection and classification as a way to improve the characterization of landscape functional units in Boreal Shield watersheds for process-oriented and model-based ecohydrological research.