We suggest that fault bends play a key role in explaining the scatter seen in maximum offset vs fault length relationships shown in existing fault scaling relationships. Primary detailed field measurements of the fault geometry and coseismic throw in the 2016-2017 central Italy earthquake sequence, together with observations of other large historical normal-faulting earthquakes within the literature, provide multiple examples where coseismic throw increases across bends in the strike of faults. We can quantify the expected change in throw across a bend by applying the “geometry-dependent throw-rate theory” (Faure Walker et al., 2015) to a single rupture, based on conservation of the strain-rate across the variable fault geometry.
We measured the geometry and kinematics of earthquake surface ruptures for the 24th August and 30th October 2016 earthquakes (Mw 6.0, Mw 6.5) in central Italy. Both datasets show that across an along-strike fault bend in the southern part of the Mt. Vettore fault, although the slip-vector azimuth and the coseismic heave vary by <10-20%, the coseismic Dmax increases by a factor of 2-3 where the strike of the host fault changes by ~30º and the dip increases by 20-25º. We observed large increase of throws on fault bends also for other historical large-offset normal faulting earthquakes, which ruptured across along-strike fault bends. We explained the large increase of throw for each example by calculations that relate it to strain-rate conservation across the varying geometry and kinematics of the host fault.
The largest offsets, observed across fault bends, for these earthquakes are always larger than maximum displacement predicted by Wells & Coppersmith, 1994, Dmax/fault length scaling relationship. We use these findings to suggest that the varying geometry of faults, expressed in variation of fault dip and plunge across along-strike fault bends, represent one of the possible causes of the scatter of values in Dmax/fault length scaling relationships. Hence, along-strike fault bends should be strongly considered when scaling relationships are used to infer stress drop variability for earthquakes or maximum magnitudes from vertical offsets in palaeoseismic datasets.