Slow Supershear (sub-Eshelby) Earthquake Ruptures on Long Faults

Earthquake rupture speed influences near-field ground motion and radiated energy. Classical theory predicts that supershear ruptures propagate stably only at speeds faster than Eshelby's speed (√2 times S-wave speed). However, seismological observations have shown that steady sub-Eshelby supershear ruptures can occur in nature, such as in the 2018 Mw 7.5 Palu earthquake. How can we reconcile the dynamic theory of supershear rupture with these observations? Here we combine computational and analytical modeling of long ruptures in homogeneous 3D media to explain the slow supershear ruptures observed in nature. For purely-strike-slip faults, our analytical result shows that the steady energy release rate of supershear ruptures is a function of seismogenic width, rupture speed, stress drop and cohesive zone size, and predicts a steady-state supershear speed faster than Eshelby speed. But if faulting involves also a slight dip-slip component (the rake angle is not 0 degrees), our numerical simulations show that supershear ruptures can propagate steadily at sub-Eshelby speeds. Such unexpected rupture speeds apply to both the horizontal apparent speed and the real oblique speed. These slow supershear ruptures occur within a range of rake angles, which encompasses rake values constrained for the Palu earthquake. They result from the combined effects of seismogenic boundaries and rupture front geometry. Our new findings yield relations between steady-state supershear speed (both fast and slow), rake angle, and the fault frictional properties along the fault, which may provide new insight into earthquake physics and new physics-based constraints for kinematic source inversions. An alternative model of the slow supershear speed of the Palu earthquake relies on the hypothesized presence of a low-velocity fault damage zone. We will report on comparison between the two models, especially their implications for ground motions.