HÂþ»­

 

Nicolas Hosek

Ìý

B.Sc. (Honours) Thesis


(PDF - 2.1 Mb)

Measuring millimeter-scale changes in surface topography, in the field, is an important challenge for understanding numerous geomorphic processes, particularly for surfaces which evolve rapidly or for which quantitative prediction and/or measurement-directed land management decisions are necessary. Natural-tread recreational trails are an example of such a surface, as they evolve from factors that include trail design and construction, substrate, climate, and disturbance by users and natural processes on trail tread dynamics. Understanding trail dynamics is currently hindered by the measurement methods that are slow to collect, particularly in remote backcountry settings, are only 1 or 2 dimensional, and yield accuracies of only centimeters to decimeters. Our aim is to develop, test and apply structure-from-motion (SfM) photogrammetry as a new method of quantifying trail surface change in 3D. We sanipled·with photographs (254-400), pre-, during- and post- impact conditions, three sections (clay loam, grass and gravel surfaces) of a trail tread at York Redoubt Park, Nova Scotia in November, 2015, where a cyclocross race was taking place. Changes in sub-millimeter resolution DEMs of the clay loam surface, rendered in Agisoft Photoscan Professional, caused by 1186 passes by cyclists were analysed using DODs (Difference of DEMs ), transect mean elevation change, flow path vectors, sinuosity and height distribution functions. The grass and gravel surfaces did not reconstruct fully ( <20%) in Agisoft Photoscan Professional due to the small-scale repeating pattern of the grass and blurry images from water droplets on the camera lens, and were discarded from all subsequent analysis. Difference of DEMs and sign plots revealed that cyclists locally displace material outward from the trail axis as a cumulative effect of rutting. Transect mean elevation change and first standard deviation patterns can show that roughness increases as surface degradation increases from immediate impacts and becomes smoother as the surface relaxes. The total volume change ( +446 mm3 for pre to during, -790 mm3 for during to post, and -310mm3 for pre to post) was low given the study area (3.352m x 1.118m of trail), indicating material was moved within rather than lost from the trail. Flow paths (topographic gradients) showed rutting along the trail axis impeded·water flow perpendicular to and off the trail, but may amplify along-trail flows and rilling. Mean sinuosity of the entire surface changed little (1.09 for pre-, 1.13 for during- and 1.14 for post-), similar to earlier 2D measurements in Nova Scotia. Limitations to our results include difficulty using fixed control points on the imaged surfaces, because the surface is so dynamic and user-safety precludes fixed rigid markers. Our results show that sub-millimeter resolution DEMs, rendered from modern structure-from-motion photogrammetry software can be used to characterize trail surfaces and provides a new approach . to trail and land management.

Keywords:Ìýstructure-from-motion, erosion, trail, digital elevation model, geomorphology
Pages: 46
Supervisor: Lawrence Plug