Abstract:

Seismic activity across the Sichuan Basin has historically remained low. Since 2014, however, seismicity has increased dramatically in the Weiyuan region of the southern Sichuan Basin, where large-scale shale gas hydraulic fracturing(HF)operations have been conducted. In particular, three successive earthquakes with M≥4.0 struck Rongxian during February 24-25, 2019. Just six months later, a moderate M5.4 earthquake occurred in the vicinity of Weiyuan County, the largest event recorded in the study area to date.

To enhance local seismic monitoring capabilities, a temporary seismic network was deployed across the Weiyuan region by the Institute of Geophysics, China Earthquake Administration(CEA) in 2019. The network consists of 40 broadband stations equipped with CMG-3ESPC/40T sensors with an average interstation spacing of ~8km. Using waveform data recorded by this network, we first constructed a high-precision earthquake catalog using a machine-learning workflow. The b-value for the entire catalog was then estimated with the maximum-likelihood method. Robust and timely estimates of b-values are essential for quantitative constraints on the evolution of earthquake sequences.

In this study, we detected and relocated a total of 49 982 events using the machine-learning-based LOC-FLOW from continuous waveforms recorded by the temporary network between October 2019 and December 2020. Cross-correlation differential travel times of these events were incorporated to constrain the relative location. We further applied a local magnitude calibration to calculate magnitudes for small-to-moderate earthquakes with hypocentral distances of ≤50km. The resulting high-resolution earthquake catalog has a completeness magnitude(M_c) of 0.4, and the largest event corresponds to the M_L5.2 Zizhong mainshock. HypoDD relocation results show that the majority of earthquakes are confined to depths less than 7km, with an average depth of 3.5km, and generally align with the burial depth of the shale gas reservoir in the study area. Specifically, the hypocenter of the Zizhong M_L5.2 earthquake lies at a focal depth of 4.43km. In map view, seismicity in the Weiyuan region delineates linear clusters adjacent to known faults, predominantly striking NW or nearly NS to NNE.

The b-value from the Gutenberg-Richter frequency-magnitude law is a critical parameter for characterizing regional seismicity. Based on the enhanced earthquake catalog, we further estimate b-values with the maximum-likelihood estimation technique to investigate their spatiotemporal variations within the study region. The spatial distribution of b-values shows clear lateral heterogeneity across the region. The estimated b-values are 0.92, 0.93, and 0.81 for southwestern(Region I), southeastern (Region Ⅱ), and northeastern(Region Ⅲ) Weiyuan regions, respectively, with corresponding completeness magnitudes of 0.6, 0.4, and 0.5. Particularly, Region III(northeastern Weiyuan) has a lower b-value than that estimated from the entire catalog(0.93), suggesting a locally elevated crustal stress within this zone.

Moreover, we detect a systematic decrease in b-values with increasing depth, likely reflecting the increase in rock confining pressure. An exception appears at the southern segment of the Molin fault, where the events at shallow depth generally correspond to low b-values, indicating concentrated elastic stress accumulating near the fault termination. Additionally, we identify a near-vertical seismic cluster characterized by high b-values. Some local earthquakes with relatively high b-values are also concentrated around the burial depth of the shale gas reservoir, which are likely triggered by shear failure via high-pressure fluid diffusion.

Time-dependent b-values reveal that decreases in b-values generally coincide with elevated seismicity, implying that small-to-moderate earthquakes mainly release stress accumulation within the shallow crust. Before the 2019 Zizhong M5.2 earthquake, the b-values decreased rapidly over a short period, followed by gradual recovery after the mainshock. Such temporal evolution is likely associated with the deformation process within the source region, including preseismic stress accumulation, coseismic stress relaxation, and post-seismic stress adjustment. The high-completeness catalog and detailed spatiotemporal b-value distributions in our study provide a fundamental dataset for traffic light monitoring frameworks and assessing seismic hazard in the Weiyuan region.

Key words: Weiyuan gas field, LOC-FLOW, HypoDD relocation, b-value, spatio-temporal variation

摘要：

近年来, 川南威远地区地震频度显著增加。文中利用2019年10月—2020年12月期间威远地区宽频带流动台网的连续波形数据, 基于深度学习的LOC-FLOW地震检测和定位流程, 构建了包含49 982个地震事件的高精度目录。在此基础上, 开展了震级校正和b值时空分布特征研究, 获得的地震目录中最小完备震级为M_c0.4。基于波形互相关的双差定位结果显示, 大部分地震呈线性或条带状分布在一些断裂附近。研究区的b值空间分布存在明显的横向变化, 东北部地区b值(约为0.81)明显低于整体平均值(0.93)。同时, 地震丛集的b值分布随深度增加呈减小趋势, 但这种现象在墨林场断裂下方却并不明显。此外, 还观测到一个近垂直的地震簇, 具有较高的b值。在页岩气赋存深度处, 也存在若干高b值的微震震群, 推测可能与流体扩散引起的剪切破裂有关。整体上, 研究区的b值变化与中小地震活动频次具有一定的相关性。

关键词: 威远地区, LOC-FLOW流程, 双差定位, b值, 时空分布特征