Analysis of Landslide Kinematics Using Multi-temporal Unmanned Aerial Vehicle Imagery, La Honda, California

Abstract

Abstract The combination of unmanned aerial vehicle (UAV) photography with structure-from-motion (SfM) digital photogrammetry provides a quickly deployable and cost-effective method for monitoring geomorphic change, particularly for hazards such as landslides. The Scenic Drive landslide is a deep-seated slope failure in La Honda, CA, with episodic activity in 1998 and 2005–06. Heavy rainfall during 2016–17 initiated movement of a new and separate landslide directly upslope of the existing Scenic Drive landslide, damaging three residences. We acquired imagery of the Upper Scenic Drive landslide beginning 2 days after initial motion using a global positioning system–enabled UAV. We used this imagery to generate seven digital elevation models (DEMs) between January and May 2017, with spatial resolutions of ∼3–10 cm/pixel. We compared these DEMs with each other and with available light detection and ranging (LiDAR) data to assess landslide kinematics, including horizontal displacement vectors, rates of motion, and total mass redistribution, and to test the accuracy and applicability of UAV/SfM-derived measurements. We estimated the maximum horizontal displacement of the slide was at least 5 m during the monitoring period and calculated that ∼3,000 m3 of material was displaced by the landslide. Comparing the UAV-derived topography with synchronous terrestrial LiDAR scanning showed that accuracies of the two techniques are comparable, generally within 0.05 m horizontally and within 0.20 m vertically in unvegetated areas. This study demonstrates the capability of combining UAV and SfM to map and monitor active geomorphic processes in emergent situations where high-resolution digital topography is needed in near-real-time.