Note: This part only lists the published researches during my Ph.D. study. On-going works will be uploaded once accepted.
Under the traditional view, earthquake ruptures were typically considered to propagate away from their hypocenters, a process called forward propagation. However, in some cases rupture takes a backward turn after propagating certain distances along the forward path, a phenomenon called back-propagating rupture (BPR). The lack of a comprehensive understanding of BPR motivates us to explore why and how BPR could occur, by combining the therotical anlysis, numerical simlations, and experimental or natural observations.
In this work, we find that: (1) BPR is an intrinsic feature of dynamic rupture; but its signal is easily masked by the interference effect behind the main rupture during a relatively stable rupture process. (2) Introducing perturbations during a relatively stable rupture process can enhance the observability of BPR. Possible perturbation mechanisms include rupture “reflection” on the free surface, collision of multiple rupture fronts, subshear-supershear rupture transition (Figure 3), and fault geometry complexity or medium heterogeneity. (3) There are two modes of BPR—fault interface waves or high-order re-ruptures, depending on whether the fault can heal after the main rupture front and whether additional stress release can be generated.
With the popularization of dense stations and near-fault records, we can be expected to record more and more BPR and others counter-intuitive features. This study deepens the understanding of source physics and is of great significance for exploring the evolution and distribution of fault properties and accurately assessing earthquake hazards.
Paper: Ding, X., Xu, S., Fukuyama, E., & Yamashita, F. (2024). Back‐propagating rupture: Nature, excitation, and implications. Journal of Geophysical Research: Solid Earth, 129, e2024JB029629. https://doi.org/10.1029/2024JB029629 [Editor's highlight] [pdf]
Editor's highlight: https://eos.org/editor-highlights/rewinding-the-fault-stress-perturbations-promote-back-propagating-ruptures
Figure 3. Excitation process of Back-propagating rupture (BPR) when a subshear rupture transites to a supershear rupture. A clear BPR phase can be found behind the forward propagating phase from the change of shear stress (right panel) and the fault-vertical particle velocity (left panel).
One of predictions from classical fault-mechanics theory, such as Anderson’s faulting theory or Mohr-Coulomb failure criterion, is that conjugate faults should form or become reactivated at ~60° under dry, brittle (with typical friction coefficient of 0.6) and small-strain conditions. However, in the last decades an increasing number of natural observations have shown that near-orthogonal conjugate faults can be activated during earthquakes, such as the 2012 Sumatra earthquake and 2019 Ridgecrest earthquake. Although several special mechanisms, such as high pore fluid pressure, low intrinsic rock friction and ductile deformation, are later proposed to explain the near-orthogonal configuration of conjugate faults, these mechanisms still do not provide a satisfactory explanation for the detailed process of fault activation during earthquakes.
Focusing on the 2023 Mw 7.6 Türkiye earthquake, finite-fault inversion and aftershock pattern indicate a complex rupture process, featuring multi-segment rupture and in particular near-orthonognal conjugate faulting in the west. In this work, we combine 2D dynamic rupture simulations with off-fault plastic yielding to show this apparent discrepancy can be resolved by dynamic loading: near-orthogonal conjugate faulting can be realized during earthquakes even under typical upper-crust conditions. The simulation results show that: (1) a range of high-angle conjugate faults, including those with acute, right, and obtuse angles, can be dynamically activated during earthquakes; (2) considering the fault geometry of the Mw 7.6 earthquake, rupture termination by a barrier near the end of the first fault can facilitate the activation of the second conjugate fault (Figure 2); (3) “continuation” of rupture along a third fault (parallel to the first one) beyond the fault junction can impede the activation of the second conjugate fault.
These results provide important insights into the rupture process and aftershock pattern of the 2023 Mw 7.6 Türkiye earthquake. Especially, the fact that the near-orthogonal conjugate fault is triggered mainly on the extensional side of the first fault suggests a vital role of dynamic normal stress change during triggering, without the need to invoke ductile deformation, low fault friction, or high pore fluid pressure.
Paper: Dynamic activation of near-orthogonal conjugate faults during earthquakes: Insights from the 2023 Türkiye Mw 7.6 earthquake (in Chinese). (2024). Chinese Science Bulletin, 69: 1501–1516, doi: 10.1360/TB-2023-0894. [pdf]
Figure 2. Coulomb failure stress change (ΔCFS) for models with variable fault continuity settings (First row: a small-scale barrier in the triple junction point; Second row: a large-scale barrier in the triple junction point; Third row: continous and uniform sets in the triple junction point). A large-scale barrier will obviously add the dynamic triggering possiblity by increasing ΔCFS.
Multiple lines of evidence indicate that the 2023 Mw 7.8 Kahramanmaraş (Türkiye) earthquake started on a splay fault, then branched bilaterally onto the nearby East Anatolian Fault (EAF). This rupture pattern includes one feature previously deemed implausible, called backward rupture branching (like a boomerang): rupture propagating from the splay fault onto the SW EAF segment through a sharp corner (with an acute angle between the two faults).
To understand this feature, we perform 2.5-D dynamic rupture simulations considering a large set of possible scenarios. We find that both subshear and supershear ruptures on the splay fault can trigger bilateral ruptures on the EAF, which themselves can be either subshear, supershear, or a mixture of the two. In most cases, rupture on the SW segment of the EAF starts after rupture onset on its NE segment: the SW rupture is triggered by the NE rupture (Figure 1(i)). Only when the EAF has initial stresses very close to failure can its SW segment be directly triggered by the initial splay-fault rupture, earlier than the activation of the NE segment (Figure 1(ii)). These results advance our understanding of the mechanisms of multi-segment rupture and the complexity of rupture processes, paving the way for a more accurate assessment of earthquake hazards.
Paper: The sharp turn: Backward rupture branching during the 2023 Mw 7.8 Kahramanmaraş (Türkiye) earthquake. (2023). Seismica, 69: 1501–1516, doi: 10.1360/TB-2023-0894. [pdf]
Figure 1. Two possible triggering process of backward branching (branching from Narli fault to the SW-side of EAF) in the 2023 Mw7.8 Türkiye earthquake. (i) Cascading process: the backward rupture need the participate of NE-side rupture; (ii) Direct triggering process: the dilational lobe from the Nali fault rupture can directly trigger the failure in the SW-side.