AUTHOR=Wang Chao , Zhang Xiufeng , Gao Weiming , Wang Fangfang , Lu Jianqi , Yan Zhaoyang TITLE=Accurate earthquake and mining tremor identification via a CEEMDAN-LSTM framework JOURNAL=Frontiers in Physics VOLUME=Volume 13 - 2025 YEAR=2025 URL=https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2025.1510629 DOI=10.3389/fphy.2025.1510629 ISSN=2296-424X ABSTRACT=The accurate identification of mining tremors and earthquakes is important for establishing a comprehensive mining tremor catalog that can aid in providing regulatory oversight for mining activities. Moreover, using a catalog purely consisting of earthquakes allows for more advanced seismological studies, such as active fault delineation, strong earthquake prediction, and stress field calculations, to be conducted. We focus on the spectral characteristics of mining tremors and earthquakes. By identifying short-period surface waves in the given data and utilizing an improved complete ensemble empirical mode decomposition method with adaptive noise (CEEMDAN) in combination with long short-term memory (LSTM) networks, we conduct a discriminative analysis of seismic events in Liaoning, China, and Japan. After completing basic preprocessing steps for both mining tremors and earthquakes, CEEMDAN is used to decompose the data into different intrinsic mode functions (IMFs). The variance contribution rates of the IMFs are extracted as features, which distinctly identify the short-period surface wave components of mining tremors. These features are subsequently input into an LSTM model for classification training, resulting in an accurate classification model. The results demonstrate that CEEMDAN-LSTM effectively addresses the noise and short-period surface wave aliasing issues encountered within the modes, yielding significantly enhanced classification accuracy. The classification success rate has been significantly improved to 96.5%. Additionally, this study explores the advantages and limitations of various classification features and models, providing effective technical support and new perspectives for the automatic identification of seismic events in the future. This research provides not only an understanding of the characteristics of mining tremors and earthquakes but also a scientific basis for earthquake early warning and disaster prevention. This study suggests that future research can further optimize the model in terms of speed and apply the model to classify more nonnatural seismic events.