USING TERRESTRIAL LIDAR AND PHOTOGRAMMETRY TO MEASURE DISPLACEMENT OF A WESTERN CASCADE COMPLEX LANDSLIDE

Landslides pose a serious hazard across the globe. Utilizing high resolution monitoring techniques at a sub-annual temporal scale can help researchers better understand the mechanics of mass wasting processes and possibly lead to better mitigation of their danger. This research uses terrestrial lidar scanning (TLS) and photogrammetry to monitor displacement of a ~4km long by ~400m wide landslide complex in the western Cascade Mountain Range near Mill City, Oregon. This site experienced failure of an internal scarp in July 2014 which reactivated the lower portion of the slide. TLS scans were collected quarterly over one year at five locations along the margin of the landslide beginning fall of 2015 and ending fall of 2016. These produced 3D point cloud models of the landslide surface with an average resolution of ~175 points per meter squared. Additionally, to fill in data gaps due to shadowing effects in the TLS surveys, very low altitude aerial images were collected and used to produce supplementary point cloud data via Structure-from-Motion (SfM). This SfM point cloud was of a more consistent and higher point density than the TLS, >500 points per meter squared on average. Preparatory processing of the scans included aligning them to north and converting the coordinate system to UTM. Relative alignment of the scans was accomplished by first using the iterative closest point algorithm to align stable, off-slide terrain, and then applying the same rigid body translation to the remaining area. This was repeated for each scan at each location. Landmarks, such as tree trunks, were then manually selected at each location and their coordinates were recorded from the initial scan and each successive scan. These coordinates were converted to displacement vectors. Average standard deviation of the vectors of features on stable areas was about ±4cm. Displacement in the section of the study area at the base of the failed scarp was largest at 0.89 meters over the year. Displacement totals decreased with increasing distance down from the internal scarp failure to a low of just a few centimeters near the toe. Displacement velocities for all locations on the slide also decreased with each successive scan. This implies that the measured displacement represents the landslide responding to the undrained load applied by the internal scarp failure.