Seismic hazard assessment for large continental strike-slip faults (vertical faults allowing for tectonic plates to move horizontally past each others) such as the San Andreas fault in USA, or the Dead Sea fault in Middle East, involves combining past and present seismic activity along that fault, rate of tectonic loading and the geometry of the fault. These data are used to build synthetic catalogues over long periods that are then used to consider and weight different scenarios in hazard assessment probabilistic models.
In this framework, beside uncertainties related to measurements, both the seismic activity and the tectonic loading rate are solid observation with only limited ground for discussion. Conversely, the fault geometry, and more specifically the way strike-slip faults are laterally structured, is still a matter of fierce debates. Following different approaches that would consider either continuous faults or segmented faults, associated to a set of rules defining how a seismic rupture propagates from one segment to the next one, seismic scenario might vary drastically with strong implication for hazard models.
A research group from the IPGP, CY Cergy Paris Université and Sorbone Université, has focused on the fault segmentation issue, aiming at testing if one could evidence intrinsic parameters controlling lateral strike-slip fault geometry. Thus, they have conducted a series of analogue sand-box experiments to reproduce shear at the base of the brittle crust. One of the properties of the sand is that it is allowing for localization of deformation, behaving as a brittle material while it accommodates deformation, although it does not store elastic energy. During experiments, photos have been acquired at regular time steps to image the top surface of the experiments. Then, using the image correlation tool MicMac, it has been possible to measure details of the surface deformation, including the formation of fault segments.
Several parameters have been systematically tested during these experiments, such as the internal friction of the material (which type of sand and how sand is initially packed in the box), basal friction (how does the sand interact with material at the bottom of the experimental setting), and thickness of the sand pack. This last parameter appears to be critical in controlling the size of the segments, with a ratio between the thickness of the sand and the length of the successive segments, close to 1. The other parameters have only second order impact on the segment geometry.
The comparison between experimental results and data derived from actual observations of surface ruptures for large continental earthquakes shows that a similar value of ~1 can be derived from the ratio between the length of rupture segments, as observed in the field, and the thickness of the brittle crust. Hence, it suggests that both in the experimental and in nature the physical processes involved are the same.
Thus, this work shows that for continental strike-slip faults, fault segmentation exists that spatial scale is controlled by the thickness of the brittle crust, in a similar fashion to the effect of the sand thickness in the experiments. Because the thickness of the brittle crust does not vary significantly in continents, it explains why the length of rupture segments, as observed during large earthquakes, is always the same at first order in any continental setting. This result allows reconsidering the concept of fault segmentation, which was mostly limited to a descriptive perspective so far, into the framework of physical processes involved during earthquake ruptures. It should lead to a better integration of the fault geometry in the seismic hazard models, with an emphasis on specific spatial scales that are relevant for earthquake rupture propagation.
More information :
M. Lefevre, P. Souloumiac, N. Cubas, Y. Klinger; Experimental evidence for crustal control over seismic fault segmentation. (2020) Geology ; 48 (8): 844–848. doi: 10.1130/G47115.1