Waveform features and automatic discrimination of deep and shallow microearthquakes in the Changning shale gas field, Southern Sichuan Basin, China

• Published in: Journal of applied geophysics Vol. 241; p. 105850
• Main Authors: Liu, Jianfeng, Xue, Fujun, Dai, Jingjing, Yang, Jianxiong, Wang, Lei, Shi, Xiangchao, Dai, Shigui, Hu, Jun, Liu, Changwu
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2025
• Subjects: Empirical mode decomposition; Feature selection; Machine learning; Principal component analysis; Source depth of microearthquakes; Southern Sichuan Basin of China; Source depth of microearthquakes; Empirical mode decomposition; Feature selection; Southern Sichuan Basin of China; Machine learning; Principal component analysis
• ISSN: 0926-9851
• DOI: 10.1016/j.jappgeo.2025.105850

Identification of microearthquakes at source depth holds significant importance in the field of microearthquake monitoring. Taking 256 microearthquake events (1.5 < ML < 4) in Changning Shale gas exploration area in the south of Sichuan Basin as the engineering background, this paper introduced a method of extracting six feature sets and 6 × 24 feature parameters, which are derived from microearthquake waveform in time and frequency domains based on Empirical Mode Decomposition and Hilbert Transform. The feature importance ranking and 22 key feature parameters closely related to source depth information were obtained using Random Forest (RF) and Extreme Gradient Boosting (XGBoost) algorithms. In addition, principal component analysis (PCA) was used to reduce dimensionality and reconstruct the feature space. The classification performance of multiple algorithms, including XGBoost, Support vector machine (SVM), Logistic Regression (LR), K-Nearest (KN), RF, and Decision Tree (DT) models, was compared. The results show that both the 22-dimensional feature parameters and the feature space reconstructed by PCA can effectively distinguish shallow events with source depths less than 1 km from deep events with source depths greater than 6 km. Using the evaluation indicators of receiver operating characteristic, sensitivity, and specificity, it is believed that XGBoost, SVM, and RF classifiers outperform LR, KN, and DT in identifying source depth. Among them, XGBoost classifiers are the least affected by random sampling and changes in sample proportion. The machine learning technology used in this study can effectively perform automatic source depth classification on seismic signals. •A method of extracting 6 × 24 waveform features in time and frequency domains based on Empirical Mode Decomposition and Hilbert Transform is introduced.•The feature selection algorithms of XGBoost and Random Forest are employed to obtain the feature importance ranking and optimal features, combined with principal component analysis to reduce complexity.•To better detect the depth of microearthquake events, various machine learning methods (i.e., XGBoost, Support vector machine, Logistic Regression, K-Nearest, Random Forest, and Decision Tree) are applied and compared.