Research Fields and Current Research Interests

Rupture scenarios derived from interseismic locking models

The interseismic locking models derived from geodetic observations have been utilized in earthquake risk assessment as they reveal the slip deficit and stress accumulation on the fault. We derive rupture scenarios based on interseismic locking models in Nicoya. We find that the final rupture extent, magnitude, and ground motions are highly dependent on the hypocenter locations (Yang et al., 2019, EPSL).

Figure 5 in Yang et al. (2019, EPSL). Final slip and ground motions in two ruptures derived from a locking model but with different hypocenter locations (red stars).

We derive rupture scenarios with the same hypocenter location with the 2012 Nicoya Mw 7.6 earthquake. The scenarios show great coherency with the 2012 Mw 7.6 event in rupture extent, magnitude, and moment rate function (Yang et al., 2019, JGR).

Figure 9 in Yang et al. (2019, JGR). Final slip and moment rate functions in a scenario derived from a locking model and kinematic models for the 2012 Nicoya Mw 7.6 earthquake.

The slip rate on the fault in the rupture scenario derived from the locking model in Nicoya

Rupture dynamics and earthquake physics

Seismic hazard assessment demands understanding on earthquake rupture process. We conduct numerical simulations to investigate the factors that impact earthquake rupture development using general models. For instance, we find that the barrier on the fault may slow and stop ruptures but may also induce supershear ruptures (Weng et al., 2015); we find that damage fault zones can promote rupture extent and increase earthquake potency (Weng et al., 2016); seismogenic width controls aspect ratios of earthquake ruptures (Weng and Yang. 2017).

Figure 2 in Weng & Yang (2017). Final slip in models with different seismogenic width and nucleation size.

Dynamic source parameters are of great significance to reveal the mystery of earthquake physics, yet difficult to be constrained from direct observation. Using numerical simulations, we investigate the estimation results of slip weakening distance (Dc) from near-fault displacement (Dc''). We found seismogenic width may cause a linear scaling of Dc'' even with the constant real Dc; existence of low-velocity fault zones may lead to huge overestimation.

Figure 2 in Chen & Yang (2020). Illustration of determining Dc, Dc' and Dc'''

Despite general models, we conduct dynamic rupture simulations for large earthquakes including the 2012 Nicoya Mw 7.6 earthquake and 2015 Nepal Mw 7.8 earthquake. Near-field observations are utilized to constrain the rupture models and the frictional properties on faults. Our dynamic rupture models can excellently fit the near-field observations and reveal the dynamic source parameters on seismogenic faults.

Figure 1 in Weng & Yang (2018). Kinematic and dynamic models for the 2015 Nepal Mw 7.8 earthquake.

Figure 1 in Yao & Yang (2020). 3-D map for the 2012 Nicoya Mw 7.6 earthquake

Subduction zone dynamics and megathrust earthquakes

Subduction zone megathrusts host Earth’s largest earthquakes, however, the Mariana subduction zone is characterized as an “aseismic” end-member due to the absence of megathrust earthquakes with magnitudes greater than 8. To get a better understanding of the faulting behavior of the megathrust, we use multiple seismic techniques to investigate the seismicity and seismic structures through multiple Ocean Bottom Seismic experiments near the Challenger Deep in the southern Mariana subduction zone (Figure 3-1). Both seismicity characteristics of megathrust faults and factors contributing to megathrust spatial extent and variability are central to understand the “aseismic” behavior and geodynamic processes of the Mariana Subduction Zone. It also contributes to seismic and tsunami hazard assessment around subduction zones.

Top: The 3D view of Mariana subduction zone and deployed Ocean Bottom Seismomter. Bottom: the 2D view of seismicity projected on the P wave velocity model.

3-D plate bending model

The lithosphere bends at subduction zone and induces outer-rise normal faulting earthquakes which may trigger large tsunami and cause significant loss of life and damage. We developed the 3-D plate bending model and the particle swarm optimization (PSO) inversion model programs to study the 3-D bending deformation of subducted plate (Figure 3-2). We focus on the relationship between plate bending deformation, in-plate stress and the distribution of outer-rise earthquakes at the southern Mariana Subduction. Our investigations show that the change of plate deformation along the trench strike can produce variation of bending shear stress, which may cause the uneven distribution of outer-rise earthquakes.

3-D plate bending deformation under boundary loading

High-resolution imaging of crustal fault zones and subsurface structure

Beside source physics, crustal structure is another critical issue in seismology. Crustal fault zones (FZ) host earthquakes, which may produce permanent damage and generate a low‐velocity zone (LVZ). The LVZ properties have profound impacts on earthquake nucleation and rupture propagation. Moreover, the shallow structure may significantly impact near-fault ground shaking intensities during earthquakes.

To derive high‐resolution FZ structure of the Chenghai fault in Yunnan, southwestern China, we deployed a linear dense array crossing the fault from January to February 2018. The array consisted of 158 short‐period (5 s) three‐component instruments and spanned an aperture of ~8 km with average station spacing of 40–50 m. We derived high‐resolution subsurface structure beneath the array.

Figure 1 in Yang et al. (2020). Map showing the location for the study region (black dashed box) and faults there (black lines).

Our ambient noise tomographic results showed a distinct LVZ with a width of 3.4 km across the CHF. The LVZ extended to 1.5 km in depth and likely represents nonuniform distribution of sediments during the formation and evolution of Binchuan basin, to which the CHF was one of the controlling factors. Considering the large population in the Binchuan basin, the newly discovered wide LVZ associated with the CHF could be an important factor for seismic hazard assessment, as it may amplify ground motion in future earthquakes.

Figure 7 in Yang et al. (2020). Top: station locations and altitudes. Bottom: shear wave velocity models derived from ambient noise tomography.

Induced earthquakes

Anthropogenic activities such as underground liquid injection are known to likely cause earthquakes. In recent decades, there raise social concerns on the damage caused by those induced events. The Hutubi underground gas storage facility in Xinjiang, China, with a maximum gas storage capacity of 10.7 billion m3, provides a good opportunity to study seismicity potentially induced by the annually cyclic injection and extraction of natural gas.

Figure 1 in Zhou et al (2019). The Geographic map of the study region. The lower left inset marks the location of the Hutubi County in China.

To statistically distinguish induced seismicity from the tectonic background, we investigate the background seismicity probability of each event using the space‐time epidemic‐type aftershock sequence model (ETAS) and a stochastic declustering method. Moreover, we relocate earthquakes by incorporating a dedicated mobile seismic network after refining the regional 1D velocity model by utilizing an artificial source. Both our ETAS modeled results and our high‐resolution relocations suggest gas‐injection‐induced seismicity adjacent to the Hutubi UGS during the first and second injection periods. Focal mechanism solutions of the two largest earthquakes (Mw 2.8 and 3.0) in August 2013 show a possibly unmapped reverse fault gently dipping to the south. Based on our high‐resolution earthquake locations, we propose that these on‐fault earthquakes are not hydrologically connected with the reservoir formation but are likely induced by poroelastic stress perturbations due to gas injection.

Figure 8 in Zhou et al (2019). Relocated earthquakes near the Hutubi UGS. (a) Locations of studied cases. (c&b) waveforms on station STZ of relocated events. (d) Relocated earthquakes projected onto the geological formations.