EP43E-08:
Assessing the Ability of Pleiades Stereo Imagery to Determine Height Changes in Earthquakes: A Case Study for the El Mayor–Cucapah Epicentral Area

Thursday, 18 December 2014: 3:25 PM
Barry Parsons, Yu Zhou, John R Elliott, Ivana Barisin and Richard T Walker, University of Oxford, Oxford, United Kingdom
Abstract:
Very high-resolution (VHR) imagery can in principle measure surface topography with decimetric precision. The question is how well 3D mapping based on VHR stereo imagery is capable of capturing sub-metric height changes occurring in an earthquake, particularly those in remote areas? The aim of this study is to assess the capability of the Pleiades satellites to determine such height changes.

We acquired Pleiades stereo imagery with backward (B), near-nadir (N) and forward (F) orientations over a 40 km × 5 km area in Baja California, Mexico, part of the epicentral area of the 2010 Mw 7.2 El Mayor–Cucapah earthquake. Coverage was chosen to coincide with the post-seismic LiDAR digital elevation model (DEM) for the area. DEMs have been produced with four different combinations of incidence angles (B-N, N-F, B-F and B-N-F). The raw point cloud has the lowest average density of ~0.7 points/m2 for B-F, ~1.0 points/m2 for B-N and N-F, and the greatest average density of ~3 points/m2 for tri-stereo (B-N-F). Compared to the post-earthquake LiDAR DEM, the Pleiades DEMs have a mean error of 1 m to 4 m, with a standard deviation of 0.5 m to 0.7 m.

A series of topographic profiles from the Pleiades DEMs across the earthquake rupture were used to measure vertical offsets. These profiles show that the Pleiades DEM is capable of capturing vertical offsets of ~0.5 m. The measurements of vertical offsets from tri-stereo have smaller uncertainties than bi-stereo; this suggests that the tri-stereo DEM incorporating redundant observations is more robust and reliable. By differencing the 1-m tri-stereo Pleiades DEM and the pre-earthquake, lower-resolution (5-m) LiDAR DEM, we mapped metre and sub-metre offsets along the faults obtaining results comparable to a previous study that differenced the two LiDAR DEMs (Fig. 1). By extension, we expect it to be possible in similar settings to measure sub-metre vertical offsets occurring in earthquakes using pre- and post-event Pleiades stereo imagery.