Fault scarps – topographic steps created at the Earth’s surface – are evidence of past displacement or aseismic motion along faults. We detect scarps and measure the height and relative ages of scarp-like landforms using a distributed template matching algorithm applied to large digital topographic datasets.
As a step towards plate boundary-scale scarp detection and dating, we have developed a distributed template matching framework suitable for deployment on a cloud computing platform. Testing shows performance improvements (figure above) in large-scale morphologic dating of fault scarps, enabling efficient processing of large, high-resolution topographic datasets.
Scarp-like landforms detected in this project are being validated against detailed field mapping of several San Francisco Bay Area and Northern California fault zones spanning a range of slip rates and documented seismic activity. These include the Olema Valley segment of the San Andreas Fault (SAF) and the North Coast segment of the SAF near Point Arena (figure above), where independent constraints are available from measurements of displacement during the 1906 earthquake and Holocene offsets estimated from paleoseismological trench studies. Less active and partially creeping fault zones, including the Rodgers Creek Fault and the Green Valley Fault, are being used to assess limitations of semi-automated fault mapping techniques in areas of complex initial topography and creeping fault segments.