Estimate of seismic fracture surface energy from pseudotachylyte-bearing faults

Earthquakes are the result of propagation at ∽km s^-1 of a rupture and associated slip at ∽m s^-1 along a fault. The total energy involved in a seismic event is unknown, but qualitatively most of it is dissipated by rock fracturing and frictional heat. Seismic fracture energy G (J m^-2) is the energy dissipated in the rupture propagation and can be estimated by the inversion of seismic waves. However, its physical significance remains elusive. G may include the contributions of both rock fracturing (energy to form new rock surfaces U_S, J m^-2) and fault frictional heating (Q, J m^-2) per unit fault area. Here we determine both U_S and Q in natural and experimental pseudotachylyte-bearing faults, following the approach used by Pittarello et al. (2008). In fact, in pseudotachylytes, or solidified frictional melts produced during seismic slip, (i) U_S is proportional to the surface of new fragments produced in both the slip zone and in the wall rocks, and (ii) Q is proportional to the volume of frictional melt.

The selected natural pseudotachylytes belong to the east-west-striking, dextral, strike-slip Gole Larghe Fault Zone (Adamello, Italian Alps). To estimate U_S we employed Electron Back-Scatter Electrons (EBSD), High Resolution Mid Angle Back-Scattered Electrons (HRMABSD) and Cathodoluminescence-Field Emission Scanning Electron Microscopy (CL-FESEM). In particular, CL-FESEM imaging reveals a microfracture network in the wall rocks that cannot be detected with the other techniques. In the pseudotachylyte-bearing fault, the microstructural analysis reveals (i) a high degree of fragmentation of the wall rock adjacent to the pseudotachylyte fault vein (formed along the slip surface), with clast size down to <90 nm in diameter, and (ii) a systematic difference in fracture density and orientation of the microfractures in the two opposite wall rock sides of the fault. In fact, in the northern wall rock the fracture density is low and the microfractures are oriented preferentially east-west, while in the southern wall rock the fracture density is high and oriented preferentially north-south. Instead, this asymmetric microfracture pattern is absent in the experimental pseudotachylytes produced by shearing pre-cut cylinders of tonalite (the rock that hosts natural pseudotachylytes) in the absence of a propagating seismic rupture. Thus, the formation of the asymmetric microfracture pattern is associated with the propagation of the seismic rupture and, therefore, can be used to estimate U_S.

In natural pseudotachylytes, fracture density decreases exponentially from the pseudotachylyte-wall rock contact towards the wall rock. The rock volumes with highest coseismic damage at the contact with the pseudotachylytes were assumed to represent the host-rock damage preceding frictional melting along the slip zone. Based on this assumption, U_S was estimated in the range 0.008-1.35 MJ m^-2, while Q was estimated from the thickness of the pseudotachylyte vein to be ∽32 MJ m^-2. In the case of the Gole Larghe Fault, numerical modelling of seismic rupture propagation yields fracture energies G in the range 8-67 MJ m^-2 suggesting that U_S is a subordinate component of G and that most of the seismological fracture energy is heat.