USE OF 3D MODELS DERIVED FROM HANDHELD PHOTOGRAPHY IN PALEOSEISMOLOGY

Geological studies of the recent earthquake record require detailed investigations into the subsurface stratigraphy and deformation at carefully selected sites along active faults. This is most often accomplished through the excavation of trenches across fault surface traces that are recognized as likely to preserve buried evidence of earthquake ruptures. Traditionally, paleoseismologists document the walls of their trench exposures by either drafting a log by measuring points on contacts and faults relative to a reference grid or by taking photos of the trench wall, manually rectifying each photo, and pasting them together into a mosaic and using this mosaic as a base map for the trench log. These approaches can be very cumbersome and may leave out some of the rich textural and tonal information that aids in the interpretation of subsurface deposits. To improve paleoseismic methodologies both in the field and in the office, we have begun to utilize a structure-from-motion (SfM) program to assist in the production of photomosaics, establishing the 3D geometry of our trenches, and the 3D visualization of complex deformation. Photomosaics produced from this method are fully orthorectified based upon the 3D geometry of the trench wall, rather than ad-hoc rectification performed by manually warping reference gridlines in individual photos to original orthogonal shape. Although the output has the potential for comparable resolution to terrestrial laser scanning, structure-from-motion routines have the advantages of fast data collection, low equipment and software cost, low power consumption, and are highly portable. This makes collection of high resolution 3D data simple, efficient, and accessible to more users, facilitating research in a variety of settings, production of interactive displays for presentations and education, and enhanced archival preservation of the field sites for future examination.