The Cascadia subduction zone in the Pacific Northwest has well-documented geological records of megathrust earthquakes with the most recent Mw 9 rupture occurring in 1700 A.D. The paleoseismic observations suggest that Southern Cascadia is mature for future earthquakes since the last event. Consequently, it is crucial to investigate the potential rupture scenarios. Various interseismic locking models are developed along Cascadia, including offshore uncertainties and different material assumptions. Although they all share similar moment deficits, whether future earthquakes may rupture the entire margin or be segmented, as found in the paleoseismic records, remains unknown. Accordingly, we aim to investigate: (1) possible rupture segmentation patterns, (2) whether south Cascadia can host margin-wide ruptures, and (3) whether the existing locking models suggest similar future rupture scenarios. We estimate the stress distribution constrained by the locking models from static calculation and discover that they lead to different stress distributions, indicating distinct seismic potentials despite their similar moment deficits. Our dynamic rupture scenarios show that the south can generate both segmented ruptures (> Mw 7.3 - 8.4) and margin-wide ruptures (> Mw 8.6) depending on hypocenter locations. The extent of Schmalzle-based segmented scenarios matches the proposed historical segmented events, and the margin-wide scenarios are well consistent with the coastal subsidence records of 1700 A.D. Therefore, we propose that three high-slip trench-breaching patches are sufficient for reproducing historical subsidence records. Our reasonable dynamic simulations can be applied in future studies for assessing seismic and tsunami hazards, and also serve as a comparison for non-trench-breaching scenarios.

The 2021 M7.4 Maduo (China) earthquake ruptured a 170 km-long left-lateral fault within the Bayan Har tectonic block in the northeast Tibetan Plateau. We use Sentinel-1 and ALOS-2 Interferometric Synthetic Aperture Radar, and Global Navigation Satellite System data to investigate the mechanisms of coseismic and postseismic deformation due to the Maduo earthquake. We present a refined coseismic slip model that features variations in both strike and dip angles, constrained by the rupture trace and precisely located aftershocks. The postseismic displacements are discontinuous along the fault trace, indicating shallow afterslip and velocity-strengthening friction in the top 2-3 km of the upper crust. Postseismic displacements that have the same sense as the coseismic ones are also observed at larger (> 50 km) distances away from the fault trace. The observed surface deformation is qualitatively consistent with either deep afterslip or viscoelastic relaxation, but does not exhibit obvious features that could be attributed to poroelastic effects. We developed a fully coupled model that accounts for both stress-driven creep on a deep localized shear zone and viscoelastic relaxation in the bulk of the lower crust. The mid- to near-field data can be reasonably well explained by either deep afterslip or non-Maxwellian visco-elasticity. However, a good fit to both the near and far-field (> 150 km) GNSS data cannot be achieved assuming the bulk viscoelastic relaxation alone, and requires a contribution of deep afterlip and/or a localized shear zone extending through much of the lower crust.

Temporary seismic network deployments are quite common both in land and offshore. The acquired data have significantly helped improve our understanding of earthquake processes and internal structure of the Earth. However, some temporary stations, especially these all-in-one units without external GPS timing system, suffer from incorrect timing record and thus pose a challenge to fully utilize the valuable data. To inspect and fix such time problems, ambient noise cross-correlation function (NCCF) is widely adopted by using daily waveforms. However, it is difficult to identify short-term time drift after stacking the NCCF output for several days to months. To detect such clock errors, travel times of local and distant earthquakes are utilized along with NCCF. We apply such a strategy on an Ocean Bottom Seismograph (OBS) dataset from southern Mariana subduction zone and a dataset from a temporary dense network from Weiyuan shale gas field, Sichuan, China. By inspecting travel times from local and distant events, we identify a very short-term clock drift (~25 sec) on the OBS data that was not detectable using NCCF only. To overcome the problem, short segments (3, 6, 12 hours) of daily wavefrom data is inspected as clock errors become stable within the selected segments. In addition, the data quality is carefully inspected with impact of different interstation distance and period band on NCCF. In particular, we find that the 6-hour segment with a period band of 2-5 sec is able to detect and correct short term changes, including linear drift. For the dense array data, we observe that NCCF symmetry is well-preserved for short interstation distance (within 1 km) but becomes distorted for larger interstation distances. Therefore, we split our dense array (79 stations) into 16 groups with a maximum interstation distance of 500 meters and 1-2 sec period band was selected after testing. Short data segments improve the time-drift detection efficiency in NCCF results, which is consistent for both local and distant events. In a nutshell, the carefull selection of data length and NCCF parameters can be helpful to identify and correct the time drift errors of temporary seismic stations.

Fault weakening process controls earthquake rupture propagation and is of great significance to impact the final earthquake size and seismic hazard. Critical slip-weakening distance (Dc) is one of the key parameters, which however is of difficult endeavours to be determined on natural faults, mainly due to its strong trade-off with the fault strength drop. An estimation method of Dc value proposed by Fukuyama et al (2003, 2007) provides a simple and direct reference of Dc on real faults from the near-fault ground displacement at the peak of ground velocity (Dc”). However, multiple factors may affect the observed near-fault ground velocity and thus need to be considered when estimating Dc. In this work we conduct 3D finite element numerical simulations to examine the effects of finite seismogenic width and near-fault low velocity zones (LVZ) on the results of Dc”. In uniform models with constant prescribed Dc, the derived Dc” values increase with seismogenic width. With a near-fault LVZ, Dc” values show significant magnification. The width of the LVZ plays a more important role in enlarging Dc estimation compared to the depth of LVZ. Complex wavefields and multiple wiggles introduced by LVZ could lead to delay pick and then cause large deviation. Overestimation should be considered when using Dc” from limited station to infer Dc on fault. Furthermore, the scaling between Dc” and final slip in models with a constant Dc indicates that the scale-dependent feature of Dc” might not be related to variations in friction properties.

We modeled the bending deformation of the subducting plate at the southern Mariana using a 3-D elastic plate flexural model. Intraplate stresses were investigated, with the along-strike variable boundary loadings. In order to match the observed plate geometry, boundary vertical loading near the Challenger Deep had to be set twice of that in the region further east within our study area. By comparing results between 2-D and 3-D models, we found that the difference on estimating plate stress can exceed 20% when there is an along-strike variation in plate bending. Finally, we found that the sharp lateral variation in the σ and the σ corresponded to an outer-rise earthquake cluster at the southern Mariana subduction zone, indicating that along-strike variation in σ and σ may be a significant mechanism to cause non-uniform distribution of outer-rise earthquakes.

Lithospheric flexure at subduction zones is a major contributor to induce outer rise earthquakes. Here, we modeled the bending deformation of the subducting plate at the southern Mariana using a 3-D plate flexural model. Intraplate stresses were investigated, with the along-strike variable boundary loadings. In order to match the observations of plate geometry, boundary vertical loading near the Challenger Deep has to be set twice of that in other areas. We also compared results between 2-D and 3-D models and found that the difference on estimating plate stress can exceed 20% when there is an along-strike variation in plate bending. Finally, we found that the sharp lateral variation in the σ and the σ corresponded to an outer rise earthquake cluster at the southern Mariana subduction zone, indicating that along-strike variation in σ and σ may be a significant mechanism to cause non-uniform distribution of outer rise earthquakes.

Earthquake migration patterns are important to reveal various triggering mechanisms, including the tectonic process and those caused by anthropogenic activities. Mapping out the spatial-temporal seismicity pattern is traditionally conducted using reference marks either in spatial or time. However, such mapping is particularly challenging for induced earthquakes because most industrial records that provide reference marks are unavailable to the public. Moreover, advances in earthquake detection techniques proliferate earthquake catalogs and thus require labor-intensive investigation. Therefore, a new methodology is demanded to automatically investigate spatial-temporal patterns of seismicity without reference marks. Here, we present a deep learning-based method to automatically identify the timings and locations of anomalous seismicity, defined by the sudden change of earthquakes in a region. We first rasterize multi-dimensional earthquake catalogs into 2-D distribution maps. Then, we identify the maps with anomalous seismicities and extract their timings and locations to generate condensed catalogs to reduce the manual effort in further investigation. We choose Changning and Weiyuan in Sichuan Basin as our study areas due to their high seismicity rates in recent years. We use the Changning catalog to train the method and the Weiyuan catalog to test the method's spatial transferability. Our approach successfully condenses both the Changning and Weiyuan catalogs with the accuracy of 0.87 based on the F1 score. The anomalous seismicities identified by our network include both earthquakes associated with hydraulic fracturing and aftershocks following strong quakes. As such, our method could be applied to broader areas with more complex migration patterns, including natural earthquake sequences.