Discussion
Multiple datasets and results derived from various inversion methods show that the 2021 Maduo earthquake ruptured simultaneously on the two branches of the fault with a branching angle of ~20°. However, numerical simulations suggest that simultaneous rupturing on both fault branches is usually difficult where the branching angle is narrow, due to strong stress interactions between faults [Aochi and Fukuyama , 2002; Kame et al. , 2003]. The dynamic stress interaction (or dynamic rupture weakening) plays an important role during earthquake ruptures. Given the stress shadow imposed by the rupture on the other fault branch, rupture through a fault bifurcation that has a narrow branching angle is most likely be simultaneous. Suppose that one of the fault branches ruptures earlier; then the Coulomb stress change on the other fault segment will prohibit its rupture. Only when the ruptures propagate simultaneously through the two fault branches (e.g., Fig.5c), back and forth slip of the fault segments could be avoided. Because the two sides of the wedge block have opposite slip directions (Fig.5c), the final dislocation of this block is smaller than that to the north and south of the block (e.g., Fig.1a inset). Numerical simulations also suggest that a simultaneous rupture is promoted when the friction coefficients are low [Aochi et al. , 2002]. However, we observe more high-frequency radiation from the bifurcated fault branches, which indicates rougher fault friction. More high-frequency energy released on the branched faults could also be an indication of more energy dissipation, which could eventually lead to the termination of the rupture. Also note, that the bifurcated fault branches are located within slightly higher mountain ranges, suggesting the rupture could have propagated from one geological unit to another [Ren et al. , 2022], which could also be a reason for different fault friction. The rupture speed of the Maduo earthquake was quite stable around 2.5 km/s, which is in contrast with supershear rupture speed derived from the other BP analysis [Yue et al. , 2022; Zhang et al. , 2022]. The primarily reason for this discrepancy could be that we applied travel time path calibrations (see more details in [Zeng et al. , 2022]) but the other BP studies did not. The rupture did not slow down at the bends and step-overs. This is because all bends/step-overs are releasing bends/step-overs, therefore normal stress decreased when rupture propagated through them, and facilitating the rupture pass-through. Fault bifurcations with narrow branching angle are quite common for strike-slip earthquakes (Fig.S16). The number of fault segments and fault segment length of the Maduo earthquake, including the bifurcated branches, fit well with statistics from [Klinger , 2010]. We therefore suggest that such simultaneous ruptures of the bifurcated fault branches could be common during strike-slip earthquakes. Dynamic simulations have to take into consideration complex fault geometry, specific stress conditions, and frictional heterogeneities to properly estimate the relationship between rupture through multiple fault segments.