[M-IS04] Interdisciplinary studies on pre-earthquake processes

[MIS04-05] A Study of the Geomagnetic Network Anomalies Before the 2022 Luding Earthquake

*ZINING YU¹, XILONG JING¹, XIANWEI WANG¹, HAIYONG ZHENG¹ (1.Ocean University of China)

Keywords:Pre-earthquake Anomaly, Geomagnetic data, Topological Network of Multi-stations

Anomalous phenomena in geomagnetic signals are commonly observed before major earthquakes. On September 5, 2022, a Mw6.6 Luding earthquake occurred at 29.59°N and 102.08°E. In order to investigate the variations in geomagnetic signals prior to the earthquake, this study analyzes the geomagnetic data from nine stations around the epicenter. First, the Multi-channel Singular Spectrum Analysis method is applied to reconstruct the low-frequency periodic signals of the geomagnetic data at multiple stations. Then, our prior analysis showed that in the absence of earthquakes, the low-frequency signals between stations exhibited strong similarities, thus we surmise that the signal correlation will decrease when the geomagnetic data is influenced by external disturbances. Subsequently, we employ the K-means clustering to rule out the possibility of reduced correlation caused by a single station. Next, we compute the Pearson correlation coefficients of the low-frequency components of the remaining stations and construct a topological network using the correlation coefficient matrix. The network centrality, is defined as a measure of the network's centralization, where a higher correlation among multiple stations indicates a larger centrality. Finally, we examine the variation of the network centrality during a period of 45 days prior to the Luding earthquake and 15 days after. As shown in Fig. 1, the results indicate that the network centrality initially remains relatively stable and exhibits slight fluctuations around 1. However, about one week before the earthquake, significant anomalies in network centrality are observed, followed by a substantial decrease to about 0.6 on the day of the earthquake.
To validate the significance of the aforementioned geomagnetic anomalies, this study separately discusses the anomalies in terms of time and space. Firstly, we randomly select two groups of 60-day geomagnetic data and extract the centrality anomalies of these two groups of data. It is found that the centrality variations in both groups fluctuate between 0.8 and 1 without any significant anomalies as shown in Fig.2.This indicates that the network anomalies prior to the Luding earthquake are different from normal situations and exhibit temporal significance. Secondly, we select an additional 17 geomagnetic stations to construct a larger-scale topological network and extract the centrality anomalies. The results show that on the day of the Luding earthquake, there is a decrease in centrality, but the centrality anomalies before that day are not significantly pronounced compared to other periods. This suggests that geomagnetic stations closer to the epicenter exhibited more clear anomalies, which indicates the spatial significance of the extracted anomalies.
This study extracts pre-earthquake anomalies among multiple geomagnetic stations of the Luding earthquake, and validates the significance of the anomalies in terms of time and space, which demonstrates the potential of geomagnetic data in the field of earthquake prediction.