New paper online – discussion open
Our last paper titled “Distributed faulting following normal earthquakes: reassessment and updating of scaling relations” is now online and open for discussion on EGU Solid Earth journal!
Why distributed faulting?
Earthquake damage can be ascribed to 2 different mechanisms: surface faulting and the passage of seismic waves. Surface faulting occurs whenever a rupture reaches the surface; beside the primary (seismogenic) fault, ground breaks can occur along nearby structures: this is what we call distributed faulting (DF).
The probability of faulting decreases with distance from the primary fault and empirical relations have been proposed to model this phenomenon. Probabilistic fault displacement hazard assessment was firstly developed by Youngs et al. (2003) and this work still includes the only available attenuation relation for DF following normal faulting earthquakes.
In a previous study we analyzed earthquakes from the Italian Apennines and found that the empirical attenuation locally underestimates the hazard.
What we did?
We realized a follow-up study, where I could apply my passion for collections without looking too weird. We took the earthquakes already analyzed by Youngs et al. and supplemented them with additional case histories, for a total of 21 earthquakes occurred between 1887 and 2016, ranging in magnitude (Mw) from 6.0 to 7.5. We computed the probability of DF and looked at the role exerted by magnitude and year of occurrence of the event.
What we found?
Let me tell the truth: the work by Youngs et al. was extraordinary in capturing the overall pattern of DF attenuation. Simply, it’s now turning 18 and it was time to update the regressions.
We also propose two methodological approaches, one basically replicating the work by Youngs et al., and the other being more conservative. Indeed, averaging all the earthquakes together means mixing two end members: events which actually generated DF at a given distance and events which didn’t.
Why should we care?
Surface faulting is challenging from an engineering perspective, and the most obvious solution is avoidance – that is, “do not build on the fault”. Nevertheless, faults can be discovered when a plant is already built, or infrastructures like pipelines or roads cannot avoid to cross the fault. In these cases, probabilistic approaches are a viable solution for a proper risk management and land planning.
Our work provides an incremental advance and we hope that the updated regressions can be implemented into engineering practice. We still have a lot to understand and our future efforts will try to answer questions like:
- Why there is a peak in DF probability at 7-10 km in the hanging wall? Is it due to the structural architecture of normal fault systems?
- Can the introduction of deterministic constraints help in better estimate the probability of DF?