Structure-from-Motion UAV Dataset at Dry Lake Valley, Central San Andreas Fault

Oct 15, 2020

OpenTopography is pleased to release a structure-from-motion small Uncrewed Aerial System (sUAS) dataset that covers 3 km2 along the creeping section of the central San Andreas fault. This dataset was collected in October 2017 by Michael Bunds and Nathan Toke from Utah Valley University, Chelsea Scott from Arizona State University, and several undergraduate students in Utah Valley University’s Geospatial Field Methods course. This dataset was collected to perform topographic differencing with the 2007 EarthScope Northern California airborne lidar dataset.

Bridge      Bridge

Perspective views of the RGB colored point cloud (left) and 3D topography (right) at Dry Lake Valley showing the Central San Andreas fault which forms a topographic step that stretches left-right across the image.  

The San Andreas Fault: The San Andreas fault in California forms the major boundary between the North American and Pacific tectonic plates. The central creeping section of the San Andreas fault is located between Los Angeles and San Francisco and extends 140 km from Parkfield to San Juan Bautista. While creeping faults can host large earthquakes, they typically slip slowly and relatively continuously. Measuring the tectonic creep rate is critical for constraining the seismic hazard along the creeping portion of the San Andreas fault.

UAV dataset collection: Optical imagery was collected by a fixed-wing Sensefly eBee Plus sUAS over two days in late October 2017. The 3533 photographs were processed into a point cloud using Agisoft Metashape with structure-from-motion methods. Georeferencing was conducted using two methods: (1) the eBee plus records 1Hz dGNSS positions calculated using the Post Processing Kinematic (PPK) method relative to a local reference station. (2) Students measured ground control points with a differential-GNSS throughout the study area. The point cloud georeferenced using the first method was shifted vertically to align with the ground control points. The resulting point cloud has 430 pts/m2. Additional details on the methods are provided in the metadata that accompanies the data set.

Bridge            Bridge

Topographic Differencing: Scott et al. (2020) calculated the horizontal and vertical surface displacements from fault creep at Dry Lake Valley between 2007 and 2017. The differencing reveals sharp right-lateral displacement across the San Andreas fault, signaling tectonic creep. The fault creep rate measured from topographic differencing is 2.5±0.2 cm/yr. The results show that the tectonic creep is localized or focused along the San Andreas fault.

The 2007 dataset used in the differencing was the EarthScope Northern California airborne lidar. The sUAS photogrammetry and airborne lidar “see” vegetation differently—the lidar signal samples through vegetation to the bear earth while the sUAV dataset is primarily sensitive to the tree tops. A ground classification removed vegetation from the sUAV point cloud, so that both datasets would see vegetation in a similar way.

The 3D differencing is performed with an Iterative Closest Point (ICP) algorithm. This algorithm was recently implemented into OpenTopography, as described here. Users can perform 3D differencing along the Central San Andreas Fault on OpenTopography here.


Topographic differencing at Dry Lake Valley along the Central San Andreas Fault from the 2007 EarthScope airborne lidar and the 2017 sUAS structure-from-motion dataset. Displacements show right-lateral offset revealing localized or focused motion along the fault. 

Reference: Scott, C., Bunds, M., Shirzaei, M., & Toke, N. (2020). Creep along the Central San Andreas Fault from Surface Fractures, Topographic Differencing, and InSAR. Journal of Geophysical Research: Solid Earth.

Funding: Bunds and Toke are grateful for an Engaged Learning grant, an award from the College of Science Scholarly Activities Committee at Utah Valley University, and Southern California Earthquake Center Awards 12050 and 13147. Scott was supported by National Science Foundation Postdoctoral Fellowship 1625221, the Southern California Earthquake Center through Award 19123, and by the School of Earth and Space Exploration at Arizona State University.