UAV Structure-from-Motion Monitoring of Landslides in Auckland, New Zealand

Abstract

Landslide monitoring is crucial for understanding the rates of slope displacement, evolution of slope morphology, and the factors associated with inducing landslide failure. Landslides are a global hazard, and regions like Auckland, New Zealand are readily affected by them. The undulating topography and regional sub-tropical climate, are influential factors when it comes to landslide failure events in the region. Recent advancements in remote sensing technology have implemented the use of Unmanned Aerial Vehicles (UAVs) for monitoring. Along with being cost-effective, UAVs allow for rapid and accurate collection of imagery for landslide hazard monitoring. This study aimed to utilize UAVs with combined Structure-from-Motion (SfM) and Multi-View Stereopsis (MVS) algorithms, to survey and monitor slow deforming slopes across the region over a fivemonth survey interval. Generated results produced high-resolution 3D point-clouds, photomosaics, Digital Surface Models (DSMs), and Digital Terrain Models (DTMs). Time-series differencing analysis was additionally conducted using UAV-SfM derived DSMs for surface change detection across the selected sites. Across the five-month survey interval (July-November), the results indicated no discernible change in slope movement. Limiting factors were likely to be the selected survey interval, and growth of surface site vegetation. The COSI-Corr feature tracking analysis was additionally trialled for this study to assess sub-pixel slope change. However, certain limitations including the survey interval did constrain the production of efficient results. Nevertheless, the UAV-SfM technique proved to represent a financially viable and rapid monitoring technique for this study, which produced highresolution results compared to more traditional techniques like airborne LiDAR.