Reconstructing the 1947 Manix Earthquake (California) using Historical Aerial Imagery and Optical Image Correlation

Constraints on rupture geometry and fault slip distribution are typically lacking for historical earthquakes due to limited (or non-existent) seismic and geodetic data, potential lack of field surveys made soon after the earthquake, and the degradation of field evidence over time. However, archival aerial photography can aid in the retrospective analysis of surface deformation using Optical Image Correlation (OIC) for mid-20th century events when pre and post-earthquake imagery is available.  

We focus here on the 1947 Manix earthquake (M_L 6.2), which was one of the first known surface rupturing earthquakes documented in the Mojave Desert block of California, and which originally highlighted that this region was capable of hosting large surface rupturing earthquakes. Situated midway between the San Andreas and Garlock faults, this block currently accommodates ~25% (10-14 mm/yr) of the total right-lateral shear associated with the motion between the Pacific and North American Plates. More recent large earthquakes in the region include the 1992 Landers, 1999 Hector Mine, and 2019 Ridgecrest sequences. Nevertheless, the earlier Manix event remains poorly studied, or how this event may have promoted slip on these neighboring faults through stress redistribution. While previous studies describe two shallow sub-events rupturing a conjugate strike-slip fault system (including the Manix fault; Richter, 1947; Doser, 1990), confirmation of exactly which fault ruptured, and the extent of any surface displacement remains unclear.  

Here, we use OIC techniques to quantify the co-seismic displacement field using newly scanned pre- and post-event aerial photos from a variety of surveys. This technique enables us to recover a spatially dense 3D displacement field (with sub-pixel precision), revealing signals previously overlooked by field geologists following the earthquake. We first use Ames Stereo Pipeline to build an internally consistent camera network for each survey, from which we generate a coherent high resolution digital elevation model (DEM) and ortho-mosaic. These are then correlated using COSI-Corr to retrieve the 2D displacement field, while the vertical displacements are obtained from differencing the DEMs while accounting for the horizontal displacement. We provide new quantitative constraints on the surface.