EXAMINING THE SOURCES OF VARIABILITY IN STRIKE-SLIP FAULT SLIP DISTRIBUTIONS WITH LANDSCAPE EVOLUTION MODELS AND RECENT RUPTURES (Invited Presentation)

Geomorphic markers provide evidence of lateral displacement on strike-slip faults from historic and paleoseismic earthquakes, and recent earthquakes show that slip distributions are highly variable along strike. Long-wavelength variability is often due to observable parameters such as distributed deformation or complex fault geometry, but the causes of short-wavelength variability are debated. There is not yet consensus on whether short-wavelength variability is fundamental to the rupture process (e.g., from variations in fault roughness) or results from incomplete characterization of tectonic slip (e.g., measurement and interpretation error), or a combination of both. Tectonic variability is directly correlated to seismic hazard, but variability related to measurement error is not. Thus, to improve the accuracy of earthquake history studies and improve probabilistic seismic hazard analysis, tighter constraints on the relative contributions of tectonic, geomorphic, and epistemic (measurement) sources of variability are needed. Here, we use numerical landscape evolution models that simulate strike-slip faulting to show that limits in a geologist’s ability to interpret the faulted landscape and reconstruct pre-earthquake morphology of laterally offset markers is a significant contribution to short-wavelength variability in slip distributions. Image correlation techniques outperform manual measurements of the imposed displacement on DEMs output from the landscape evolution simulations. When offsets are measured a long time after an earthquake, landscape evolution and channel capture also contribute to short-wavelength variability, but the extent of that influence is likely related to the rate of geomorphic degradation relative to fault slip rate. A comparison of slip distributions from recent earthquakes measured with image correlation and by hand also reveals less variability from the automated techniques. These results suggest that studies that interpret the amount of slip in past earthquakes from peaks in cumulative offset probability distribution curves constructed from dense datasets of offset markers measured by hand may contain much greater uncertainty than has been previously recognized.