GSA Connects 2021 in Portland, Oregon

Paper No. 88-9
Presentation Time: 9:00 AM-1:00 PM


BUNDS, Michael, Department of Earth Science, Utah Valley University, 800 West University Parkway, Orem, UT 84058, SCOTT, Chelsea P., School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85281-1404, WHITNEY, Brigham, Department of Earth Science, Utah Valley University, 800 W. University Pkwy., Orem, UT 84058 and LEE, Vickie Jui-Chi, Department of Geosciences, Virginia Polytechnic Institute, Blacksburg, VA 24061

We produced a very high resolution topography dataset along ~40 km of strike length of the Southern San Andreas Fault, from Painted Canyon to south of Bombay Beach. The dataset comprises a point cloud (8x109 points), digital surface model (DSM; 10 cm pixels) and orthomosaic (5 cm pixels). It was made using photographs collected with a small uncrewed aerial system (sUAS or drone) and structure-from-motion processing. We are making the dataset publicly available on OpenTopography ( We anticipate a variety of applications including geomorphic and fault mapping, paleoseismic studies, topographic differencing with existing lidar datasets, and as pre-event data in the event of a future surface-rupturing earthquake. Field work was completed in less than 4 days by a 4 person team with a single sUAS. We see the effort as a test-run for rapid, accurate, high resolution topography generation following surface-rupturing earthquakes.

A Sensefly eBee Plus sUAS was used to collect 15773 photographs. Camera positions were determined using dual frequency on-board dGNSS processed against our geodetic quality reference stations (GCPs were not used for georeferencing). Field work consisted of daily set-up of a GNSS reference station, a total of 22 1-hour sUAS flights and measurement of bare-ground checkpoints for quality control. Nightly preliminary processing was completed on three GPU-equipped laptops. Final processing was performed on a cluster of five workstations with 12 GPUs in ~48 hours computer time in Agisoft Metashape and includes tie point generation on the ‘highest’ setting, inclusion of camera positions, iterative bundle adjustments and removal of large uncertainty tie points, and a ~10 cm vertical translation to account for systematic bias relative to 176 independently measured checkpoints. Horizontal and vertical RMS error of the point cloud relative to the checkpoints is 2.3 and 4.5 cm, respectively.

Our sUAS dataset is complimentary to the existing B4 lidar collected in 2005 given their different spatial resolution, coverage area, and error sources. Rapid production of high resolution topography (~10 km2/day by one sUAS and crew) has important potential for post-earthquake scientific response including repeat surveys to capture post-event surface deformation.

  • GSA_2021_SSAF_poster_v2b_GSA_upload.pdf (4.7 MB)