Monitoring deformations of embankments in Arctic regions using unmanned aerial vehicle photogrammetry and terrestrial laser scanning

• Main Author: Stafford, Dylan
• Other Authors: Alfaro, Marolo (Civil Engineering), Arenson, Lukas (Civil Engineering) Paliwal, Jitendra (Biosystems Engineering)
• Format: Master Thesis
• Language: English
• Published: 2020
• Subjects: Embankment; Permafrost; Deformation; Unmanned aerial vehicle; Photogrammetry; Structure-from-Motion; Terrestrial Laser Scanning; Arctic; Faro; Inuvik; Tuktoyaktuk; Ice; permafrost
• Online Access: http://hdl.handle.net/1993/34593

Embankments in Arctic regions are typically constructed during winter with no cuts in the ground to preserve the permafrost foundation. These embankments are susceptible to deformations in the summer immediately following construction as ice within the embankment fill melts and in subsequent years as permafrost at the embankment toe thaws. Unmanned aerial vehicle (UAV) photogrammetry and terrestrial laser scanning (TLS) were used to monitor deformations of four high-fill embankment sections along the newly constructed Inuvik-Tuktoyaktuk Highway (ITH). Two UAVs (senseFly albris and DJI Phantom 4 Pro) and one laser scanner (FARO Focus3D X 330) were used. One of the high-fill sections was reinforced with wicking woven geotextiles to improve slope stability and instrumented to displacements within the embankment. UAV photogrammetry and TLS are both relatively new technologies being used to monitor deformations of structures. Significant effort was dedicated to learning about the technologies, developing best operating practices, calibrating the technologies to quantify their accuracies, and designing the on-site surveys. UAV and TLS surveys were conducted during summer in three consecutive years (2017–2019). UAV imagery and TLS data were processed using specialized software to generate point clouds of the high-fill sections. An RTK system was used to measure positions of checkerboard ground control points (GCP) for georeferencing point clouds. The accuracy of UAV and TLS point clouds was quantified based on GCP errors. Alignment of point clouds was required because of poor quality GCP measurements. Point clouds from each year were compared using multiscale model-to-model cloud comparison (M3C2) to determine deformations. A cross-section analysis was also performed for each high-fill section. High-fill sections along ITH showed deformations including toe subsidence and lateral spreading. Some of the high-fill sections showed positive change (e.g. heave, deposition) at the upper-slope and negative change (e.g. ...