Abstract:

Gravity variations provide an important means for studying the processes of earthquake preparation and occurrence, as well as earthquake prediction. In recent years, gravity variations have achieved good applicationin determining the locations of a series of strong earthquakes, including the 2013 Lushan 7.0 earthquake, the 2017 Jiuzhaigou 7.0 earthquake, and the 2022 Menyuan 6.9 and Luding 6.8 earthquakes. However, the current analysis of gravity variation anomalies and the determination of strong earthquake locations is mainly based on manual interpretation, and face issues of low efficiency and reliance on experience. In this paper, automated identification methods of gravity anomaly featuresuch as quadrants, gradient zones, and zero lines are firstly studied. Then the method of focusing gravity variation anomaly features on active block boundary zones is studied to further identify potential strong earthquake locations. Finally, the method is validated using the example of the December 18, 2023 Jishishan M_S6.2 earthquake.
(1)By comprehensively utilizing methods such as machine learning image recognition, total horizontal gradient calculation, and grid data search, gravity variation anomaly features such as quadrants, gradient belts, and zero linesin gravity variation images can be quickly and comprehensively identified, which facilitates the comprehensive statistical analysis of gravity variation anomaly features. Based on the close correlation between seismic activity and the boundaries of active tectonic blocksin China mainland, a method of focusing anomaly featureson the block boundaries has been proposed in order to further delineate potential strong earthquake locations.
(2)Before the Jishishan M_S6.2 earthquake, gravity variation in the northeastern margin of the Qinghai-Xizang Plateau shows a spatial distribution pattern of negative to positive from southwest to northeast. Along the southern segment of the Lajishan Fault through the epicenter of the Jishishan M_S6.2 earthquake, the gravity variation showed distinct zero lines and high-gradient zone features. The maximum difference between the positive and negative centers was approximately 110μGal, with an anomaly range of 200km and a duration of 1-3 years. The gravity variations in terms of pattern, magnitudeand duration are consistent with the existing anomaly indicators for magnitude 6 earthquakes. Combined with information on surface deformation, tectonic background, and geophysics, it is inferred that the gravity observation reflects the change in the distribution of crust-mantle material caused by the NE-SW compression of the northeastern margin of the Qinghai-Xizang Plateau and the reverse thrust movement of the southern segment of the Lajishan fault zone.
(3)By focusing gravity anomaly characteristics on active block boundaries, it is helpful to give more intuitive and precise results for strong earthquake location prediction. The focused results of the three types of gravity anomalies(quadrant, gradient belt, and zero line) in the block boundary zone all show high-value characteristic in the southeastern area of the Qilian Block. The comprehensive statistical results formed by the superposition of the three types of gravity variation anomalies further highlight that the southeastern area of the Qilian Block is the most obvious high-value anomaly zone, and the Jishishan M_S6.2 earthquake is located on the edge of this high-value zone. Compared with traditional manual interpretation methods, the approach presented in this study is not only more intuitive and efficient, but also significantly reduces the size of the earthquake prediction area.
(4)This research is a preliminary exploration of automated extraction methods of gravity variation anomalyfeatures and their application in strong earthquake location prediction combined with active tectonics. In subsequent studies, it is still necessary to refine the focused seismic fault zones, taking into account non-block boundary regions such as strong earthquake rupture gaps, locked fault segments identified by geodetic measurements, sparse small-to-medium earthquake activity zones, and segments with significant Coulomb stress enhancement from numerical simulations, etc. Additionally, a comprehensive statistical analysis of the correspondence between different anomaly features and their seismogenic probabilities should be conducted based on a large number of earthquake cases to assign more reasonable weight coefficients for different gravity variation anomaly features.

Key words: gravity variation anomaly, four quadrants, gradient belts, zero value lines, Determination of strong earthquake location

摘要： 针对重力变化分析和强震地点判定目前仍采用人工判读方式的问题, 文中研究了重力变化图像中“四象限”、 “梯度带”和“零值线”等重力变化异常特征的自动化识别方法, 以及将异常特征区聚焦至块体边界带圈定潜在强震地点的方法, 并以2023年12月18日甘肃积石山 M_S6.2 地震为例进行了测试。结果表明: 1)文中方法可快速识别出“四象限”、 “梯度带”和“零值线”等各类重力变化异常特征区, 以便于对各类异常特征进行全面的统计分析。2)在积石山 M_S6.2 地震前, “四象限”、 “梯度带”和“零值线”3类重力变化异常均显示在祁连块体东南角地区存在高值现象, 3类异常的叠加结果进一步凸显了该地区高值异常特征, 而本次地震即位于该高值异常区的边缘。以上分析表明, 文中方法有助于给出更为精确的强震地点判定意见。

关键词: 重力变化异常, 四象限, 梯度带, 零值线, 强震地点判定