Article Detail

İrem Hatice  DOĞAN

Keywords

Raw Seismic Data Deep Learning 1D-CNN East Anatolian Fault Line Earthquake Classification

Doi : 10.71350/jner.2026167

Abstract

This study aims to rapidly and automatically classify earthquakes on the Hatay-Malatya fault line, which experienced intensified seismic activity following the February 6, 2023 earthquakes, using raw seismic data. Over 1,200 seismic events from 2019-2023, obtained from the Kandilli Observatory (KOERI) network, were analyzed directly using a 1-Dimensional Convolutional Neural Network (1D-CNN) architecture, without resorting to traditional feature extraction methods. As a unique contribution of this study, the Logarithmic Transform technique, which preserves amplitude information in seismic signals, was compared with the standard Linear Normalization method. Experimental results showed that the proposed logarithmic preprocessing strategy increased the model's discriminability, raising the classification success rate from 68% to 79%, and the detection rate of large earthquakes to 87%; demonstrating that the data representation method plays a critical role in deep learning models based on raw data, as much as the model architecture.

References

1. Ahi, K., Chen, Y., & Wang, L. (2023). ReQuakenition: An earthquake early warning system using LSTM. IEEE Access, 11, 2345–2356.
2. Asim, K. M., Khan, S., & Ali, R. (2021). Earthquake magnitude prediction using hybrid neural networks. IEEE Transactions on Geoscience and Remote Sensing, 59(5), 4500–4510.
3. Apriani, D., Nugroho, S., & Santoso, H. (2022). Magnitude estimation from single-station P-wave using deep neural networks. Seismological Research Letters, 93(2), 890–901.
4. Batı Sumatra Çalışma Grubu. (2023). Classification of seismic signals using deep learning in Sumatra subduction zone. Journal of Asian Earth Sciences, 245, 105560.
5. Doğan, F. (2023). Machine learning approaches for earthquake prediction in Northwest Turkey. Natural Hazards, 116, 220–235.
6. Gürlen, E., Yılmaz, H., & Kaya, M. (2023). Marmara Bölgesi depremlerinin Markov zincirleri ve yapay sinir ağları ile analizi. Gazi Üniversitesi Fen Bilimleri Dergisi, 36(2), 112–125.
7. Karcı, A. V., & Karcı, A. (2022). Earthquake magnitude and time prediction with the UKSB neural network. Earth Science Informatics, 15, 1–15.
8. Li, Z., Chen, X., & Liu, Y. (2021). Generative adversarial networks for earthquake P-wave detection. Geophysical Journal International, 227(1), 345–356.
9. Salmanoğ, A., Köse, A., Yıldız, B. (2025). Misconceptions About Correct Behaviors During Earthquakes: A Field Survey on Earthquake Awareness in Hatay. Journal of Natural and Engineering Research, 1(1), 1-8, doi : 10.5281/zenodo.18663198
10. Salam, A., Lee, J., & Park, S. (2020). Earthquake prediction in Southern California using FPA-LS-SVM. Applied Soft Computing, 95, 106532.
11. Uyar, M., & Özdemir, S. (2025). Comparison of LSTM and ANN models for earthquake magnitude prediction. Journal of Seismology and Engineering, 12(1), 45–58.
12. Zhang, J., Wu, Q., & Zhao, H. (2022). Real-time earthquake monitoring using fully convolutional networks. Nature Communications, 13, 1–12.
13. Zhu, W., Lin, Y., & Huang, J. (2023). DHLnet: A data-knowledge driven hybrid learning network for earthquake early warning. Geophysical Research Letters, 50, e2023GL103456.