Abstract:

On March 25, 2023 at 13:53 local time, an earthquake of M_L3.2 struck Weishan, Shandong Province. By April 4, three aftershocks with magnitudes greater than M_L0 had occurred. On April 6, another M_L3.3 event struck the same epicentral area, after which seismicity intensified, forming the Weishan earthquake swarm. The swarm effectively ended on June 30 following an M_L0.7 earthquake.
In this study, we used PhaseNet, a deep learning-based detector, to identify seismic events, and then applied the HypoDD algorithm for precise relocation, yielding a high-precision catalog for the Weishan swarm. In addition, using observation reports of M_L≥0 earthquakes from January 2009 to March 2024 from the Shandong seismic network and neighbouring provinces, we performed double-difference tomography to invert the three-dimensional velocity structure of the source region, providing a detailed image of the subsurface architecture. The three-dimensional Poisson's ratio was then calculated from the inverted P-and S-wave velocity models using \sigma = \frac{V_{\mathrm{P}}^2-2V_{\mathrm{S}}^2}{2(V_{\mathrm{P}}^2-V_{\mathrm{S}}^2)} . Furthermore, employing the P-wave primitive-polarity picker POI(Probability of arrival time and polarity based on Order statistics and Information theory) together with the HASH method, we inverted focal-mechanism solutions for 10 earthquakes with M_L≥2.0. Integrating the precise relocation, the 3-D velocity structure, and these focal mechanisms, we identified the seismogenic faults responsible for the swarm. Finally, we undertook a comprehensive analysis of the seismotectonic setting and seismic environment of the Weishan earthquake swarm.
The results show that the Weishan swarm defines a clear NWW-trending linear zone ~3km long and<1km wide, with focal depths tightly clustered at 4~8km. Nodal parameters from the 10 M_L≥2.0 focal mechanisms are consistent with the fault geometry revealed by precise relocations. Together, these indicate that the seismogenic fault is a previously unrecognized, high-angle, left-lateral strike-slip fault trending northwest-west(NWW) and dipping gently to the southwest(SW). The rupture surface is relatively small(~3km×2km), with a narrow damage zone and a comparatively planar fault plane. During the swarm, rupture initiated at ~8km depth and propagated upward with bilateral growth along strike, but did not reach the surface.
Near the epicentral area, seismic velocities vary markedly. Prominent high-velocity and high-Poisson's-ratio anomalies occur northwest of the Sunshidian Fault, whereas low-velocity and low-Poisson's-ratio anomalies appear below ~15km southeast of the Fushan Fault. Velocity and Poisson's ratio also show clear layering across the region. The swarm itself is situated within a high-velocity anomaly for both P and S waves, and Poisson's ratio near the epicentre is relatively low, indicating that the source rocks are relatively hard.
Based on the inferred fault properties and regional crustal structure from precise locations and focal mechanisms, we conclude that the Weishan swarm reflects brittle failure of hard layers in the source region-i.e., a concentrated release of stress in a localized volume. These findings refine our understanding of the mechanisms and seismogenic environment governing the Weishan earthquake swarm.

Key words: Deep-Learning, earthquake detection, accurate earthquake location, 3-D velocity structure, seismogenic structure, Weishan earthquake swarm

摘要：

文中基于深度学习方法, 对微山震群附近地区进行了地震检测和地震精定位, 并通过双差层析成像方法反演了震源区的三维速度结构。基于地震精定位结果和反演的三维速度结构, 并结合部分M_L≥2.0地震的震源机制解, 对微山震群的孕震环境和发震构造进行了深入分析。研究结果表明, 微山震群的发震断裂为一条高角度左旋走滑断裂, 走向NWW, 略向SW向倾斜; 发震断裂破裂面尺度较小, 破裂由8km深度处开始向地表方向破裂, 同时沿断裂走向两侧破裂, 但未破裂至地表。地下介质波速和泊松比在孙氏店断裂和凫山断裂两侧变化剧烈, 且不同深度具有一定的分层特征。微山震群位于P波、 S波高速异常体内, 同时泊松比相对较低。基于发震断裂的性质与区域地壳结构特征, 分析认为震源区坚硬地层的脆性破裂是微山震群的直接成因, 反映了区域应力的集中释放过程。上述发现深化了对微山震群地震活动机制和孕震环境的理解。

关键词: 深度学习, 地震检测, 地震精定位, 3-D速度结构, 发震构造, 微山震群