SCEC Award Number 15194 View PDF
Proposal Category Individual Proposal (Data Gathering and Products)
Proposal Title Collaborative Research Aerial2lidar3D: Development of a standard technique to measure 3D coseismic surface deformation for past and future large earthquakes that lack pre-event lidar data
Investigator(s)
Name Organization
James Dolan University of Southern California
Other Participants Christopher Milliner (USC), Sebastian LePrince (Caltech), Francois Ayoub (Caltech), James Hollingsworth (URAP Consultants)
SCEC Priorities 4a, 4b, 4c SCEC Groups Geodesy, FARM, Geology
Report Due Date 03/15/2016 Date Report Submitted 05/01/2016
Project Abstract
 Understanding surface deformation patterns of large magnitude earthquakes is of critical importance for the charac- terization of distributed deformation, providing constraints for finite slip inversions, comparisons of geologic- geodetic slip rate data, understanding the structural evolution of faults, and the dynamics and arrest of rupture propagation relative to fault structure. In the rare examples where both pre- and post-event high-resolution lidar data exist, differencing of these data sets to determine the full 3D, surface deformation field is the emerging gold standard. In most settings, however, pre-event lidar data do not exist, making it impossible to use lidar differencing to measure the 3D deformation field. The aim of our research is to develop an alternative and complementary methodology that utilizes differencing of post-event lidar data with pre-event digital elevation models (DEMs) generated from stereoscopic air photos, which are much more readily available than lidar data, inexpensive, and extend back many decades for many areas. Specifically, we aim to measure the full 3D surface deformation pattern of the 1999 Hector Mine earthquake, which has post-event lidar and pre-event optical imagery, but lacks pre-event lidar data. Our results so far have successfully generated a pre-earthquake point cloud from the stereo-air photos, matching these with a post-earthquake lidar dataset, detecting vertical motion along a 3.5 km stretch of the rupture. Our results show vertical motion ranging from 0.40 to 1.35 m, in good agreement with that measured in the field, however, in places of fault structural complexity we find significant differences, indicating the presence of off-fault deformation.
Intellectual Merit The project is a new approach to measuring co-seismic surface motion by differencing point clouds generated from stereo-pair air photos before an earthquake from lidar point cloud acquired after an event. This is significant, as lidar data are not yet ubiquitous and legacy lidar data are generally of poor quality. Thus, the successful implementation of our novel technique, combining aerial data with lidar, opens the door to measuring surface motion for future events that lack pre-event lidar data, as well as analysis of historic surface ruptures (e.g., the Hector Mine surface rupture as shown here). This technique also improves upon optical image correlation, which is limited to measuring the horizontal 2D pattern of deformation. In contrast, our approach can quantify the vertical component, and marks an improvement over traditional lidar differencing, which does not correct for horizontal motion and advection of topography, which can cause biases in ground motion (e.g., Oskin et al., 2012). Such data will hold fundamental importance to understand faulting kinematics, how slip varies along a rupture, how lateral motion is compensated (or not) by vertical motion, faulting mechanics to understand the magnitude of strain release, and vital data for finite-fault inversions to accurately constrain slip accommodated in the shallow crust (< 3 km depth).
Broader Impacts This project has led to training and professional development of two doctoral students at USC (Chris Milliner and Alexandra Hatem). Both have gained expertise in the use of highly sophisticated, state-of-the-art image correlation techniques, as well as more generally in the fields of active tectonics, fault-zone structure and surface rupture map- ping.
Exemplary Figure Figure 2. a) Point cloud result from pre-Hector Mine air photos produced using Agisoft Photoscan. Den- sity of point cloud is ~ 1 point/m2. b) Post-earthquake point cloud from lidar survey with higher density of 8.35 points/m2. c) Successful vertical component detected from point cloud matching using ICP algo- rithm, which clearly reveals the vertical fault motion along the northern end of the Hector Mine rupture.