Citation :

Deepak Kumar Sinha, Sujata Kulkarni, "An Explainable Machine Learning Framework for Processing Heterogeneous Seismic Sensor Data," International Journal of Electronics and Communication Engineering, vol. 13, no. 7, pp. 77-95, 2026. Crossref, https://doi.org/10.14445/23488549/IJECE-V13I7P106

Abstract

Predicting the severity of seismic events relies on a large amount of data collected from remote sensor networks. However, processing long-term telemetry records is computationally challenging. For example, the data are typically subject to an extremely unbalanced class distribution, and sensor type variations are extremely high over time, thus reducing the effectiveness of the standard classification methods. This study presents an explainable machine learning framework for classifying event severity using long-term data from mixed seismic probes. Event records of the Indian tectonic region from 1947 to 2022 were extracted from the USGS ComCat database. To process the different information of the sensors, a processing pipeline based on a class-weighted random forest was assembled, which was assisted by data harmonization and feature engineering procedures. This pipeline was combined with SHapley Additive exPlanations (SHAP) to trace where and when the model considers spatial constraints and measurement uncertainty, that is, the focal depth and azimuthal gaps. Even with the complete elimination of explicit magnitude scalar objects to prevent target leakage, the system attained an overall accuracy of 91 %, along with a macro F1-score of 0.43. SHAP diagnostics were applied to demonstrate the existence of a hard physical boundary. Although secondary sensor proxies can be successful in classifying normal baseline tremors, there is a mathematical failure during extreme events. Catastrophic hazard detection is performed using direct waveform measurements. This work demonstrates that XAI can be used as a powerful diagnostic tool to identify the actual predictive limitations of legacy telemetry networks.

Keywords

Explainable Artificial Intelligence, Machine Learning, Random Forest, SHAP, Seismic event classification, Seismic Hazard Assessment.

References

1. R.B.S. Yadav et al., “A Probabilistic Assessment of Earthquake Hazard Parameters in NW Himalaya and the Adjoining Regions,” Pure and Applied Geophysics, vol. 169, no. 9, pp. 1619-1639, 2012.
   [CrossRef] [Google Scholar] [Publisher Link]
2. Rakesh Chandra et al., “Seismic Hazard and Probability Assessment of Kashmir Valley, Northwest Himalaya, India,” Natural Hazards, vol. 93, pp. 1451-1477, 2018.
   [CrossRef] [Google Scholar] [Publisher Link]
3. S. Mostafa Mousavi, and Gregory C. Beroza, “Machine Learning in Earthquake Seismology,” Annual Review of Earth and Planetary Sciences, vol. 51, pp. 105-129, 2023.
   [CrossRef] [Google Scholar] [Publisher Link]
4. Hisahiko Kubo, Makoto Naoi, and Masayuki Kano, “Recent Advances in Earthquake Seismology using Machine Learning,” Earth Planets Space, vol. 76, no. 1, pp. 1-32, 2024.
   [CrossRef] [Google Scholar] [Publisher Link]
5. Wenhuan Kuang, Congcong Yuan, and Jie Zhang, “Network-Based Earthquake Magnitude Determination via Deep Learning,” Seismological Research Letters, vol. 92, no. 4, pp. 2245-2254, 2021.
   [CrossRef] [Google Scholar] [Publisher Link]
6. Zebin Yang, and Aijun Zhang, “Enhancing Explainability of Neural Networks Through Architecture Constraints,” IEEE Transactions on Neural Networks and Learning Systems, vol. 32, no. 6, pp. 2610-2621, 2021.
   [CrossRef] [Google Scholar] [Publisher Link]
7. Rui Zhang et al., “Co-Seismic Landslide Susceptibility Mapping for the Luding Earthquake Area Based on Heterogeneous Ensemble Machine Learning Models,” International Journal of Digital Earth, vol. 17, no. 1, pp. 1-33, 2024.
   [CrossRef] [Google Scholar] [Publisher Link]
8. Kaito Fujiwara et al., “Measuring the Interpretability and Explainability of Model Decisions of Five Large Language Models,” Open Science Framework, pp. 1-10, 2024.
   [CrossRef] [Google Scholar] [Publisher Link]
9. Chenzuo Ye et al., “Physically Based and Data-Driven Models for Landslide Susceptibility Assessment: Principles, Applications, and Challenges,” Remote Sensing, vol. 17, no. 13, pp. 1-38, 2025.
   [CrossRef] [Google Scholar] [Publisher Link]
10. Mir Riyanul Islam et al., “A Systematic Review of Explainable Artificial Intelligence in Terms of Different Application Domains and Tasks,” Applied Sciences, vol. 12, no. 3, pp. 1-38, 2022.
    [CrossRef] [Google Scholar] [Publisher Link]
11. Deepak Mane et al., “Unlocking Machine Learning Model Decisions: A Comparative Analysis of LIME and SHAP for Enhanced Interpretability,” Journal of Electrical Systems, vol. 20, no. 2s, pp. 598-613, 2024.
    [CrossRef] [Google Scholar] [Publisher Link]
12. Ahmed M. Salih et al., “A Perspective on Explainable Artificial Intelligence Methods: SHAP and LIME,” Advanced Intelligent Systems, vol. 7, no. 1, pp. 1-8, 2025.
    [CrossRef] [Google Scholar] [Publisher Link]
13. Varad Vishwarupe et al., “Explainable AI and Interpretable Machine Learning: A Case Study in Perspective,” Procedia Computer Science, vol. 204, pp. 869-876, 2022.
    [CrossRef] [Google Scholar] [Publisher Link]
14. Ratiranjan Jena et al., “Explainable Artificial Intelligence (XAI) Model for Earthquake Spatial Probability Assessment in Arabian Peninsula,” Remote Sensing, vol. 15, no. 9, pp. 1-26, 2023.
    [CrossRef] [Google Scholar] [Publisher Link]
15. Prashant Parasar, and Akhouri Pramod Krishna, “Explainable AI-Driven Assessment of Hydro Climatic Interactions Shaping River Discharge Dynamics in a Monsoonal Basin,” Scientific Reports, vol. 15, no. 1, pp. 1-19, 2025.
    [CrossRef] [Google Scholar] [Publisher Link]
16. Mohamed S. Abdalzaher et al., “Comparative Performance Assessments of Machine-Learning Methods for Artificial Seismic Sources Discrimination,” IEEE Access, vol. 9, pp. 65524-65535, 2021.
    [CrossRef] [Google Scholar] [Publisher Link]
17. Sarfraz Khan et al., “Updated Earthquake Catalogue for Seismic Hazard Analysis in Pakistan,” Journal of Seismology, vol. 22, no. 4, pp. 841-861, 2018.
    [CrossRef] [Google Scholar] [Publisher Link]
18. Onur Tan, “A Homogeneous Earthquake Catalogue for Turkey,” Natural Hazards and Earth System Sciences, vol. 21, no. 7, pp. 2059-2073, 2021.
    [CrossRef] [Google Scholar] [Publisher Link]
19. Imtiyaz A. Parvez, Franco Vaccari, and Giuliano F. Panza, “A Deterministic Seismic Hazard Map of India and Adjacent Areas,” Geophysical Journal International, vol. 155, no. 2, pp. 489-508, 2003.
    [CrossRef] [Google Scholar] [Publisher Link]
20. S. Mostafa Mousavi et al., “STanford EArthquake Dataset (STEAD): A Global Data Set of Seismic Signals for AI,” IEEE Access, vol. 7, pp. 179464-179476, 2019.
    [CrossRef] [Google Scholar] [Publisher Link]
21. Michael A. H. Hedlin, Paul S. Earle, and Harold Bolton, “Old Seismic Data Yield New Insights,” EOS, Transactions American Geophysical Union, vol. 81, no. 41, pp. 469-473, 2000.
    [CrossRef] [Google Scholar] [Publisher Link]
22. Carlos Alberto Vargas et al., “Advanced Technology and Data Analysis of Monitoring Observations in Seismology,” Applied Sciences, vol. 13, no. 19, pp. 1-4, 2023.
    [CrossRef] [Google Scholar] [Publisher Link]
23. A. D. Tsampas et al., “Globally Valid Relations Converting Magnitudes of Intermediate and Deep-Focus Earthquakes to MW,” Bulletin of the Geological Society of Greece, vol. 47, no. 3, pp. 1316-1325, 2013.
    [CrossRef] [Google Scholar] [Publisher Link]
24. Domenico Di Giacomo, E. Robert Engdahl, and Dmitry A. Storchak, “The ISC-GEM Earthquake Catalogue (1904–2014): Status after the Extension Project,” Earth System Science Data, vol. 10, no. 4, pp. 1877-1899, 2018.
    [CrossRef] [Google Scholar] [Publisher Link]
25. Filiz Tuba Kadirioğlu, and Recai Feyiz Kartal, “The New Empirical Magnitude Conversion Relations using an Improved Earthquake Catalogue for Turkey and its Near Vicinity (1900–2012),” Turkish Journal of Earth Sciences, vol. 25, no. 4, pp. 300-310, 2016.
    [CrossRef] [Google Scholar] [Publisher Link]
26. Jochen Braunmiller et al., “Homogeneous Moment-Magnitude Calibration in Switzerland,” Bulletin of the Seismological Society of America, vol. 95, no. 1, pp. 58-74, 2005.
    [CrossRef] [Google Scholar] [Publisher Link]
27. Daniel T. Trugman, and Yehuda Ben‐Zion, “Potency–Magnitude Scaling Relations and a Unified Earthquake Catalog for the Western United States,” The Seismic Record, vol. 4, no. 3, pp. 223-230, 2024.
    [CrossRef] [Google Scholar] [Publisher Link]
28. Jiuxun Yin et al., “Earthquake Magnitude with DAS: A Transferable Data‐Based Scaling Relation,” Geophysical Research Letters, vol. 50, no. 10, pp. 1-11, 2023.
    [CrossRef] [Google Scholar] [Publisher Link]
29. A. Laurendeau, C. Clément, and O. Scotti, “A Strategy to Build a Unified Data Set of Moment Magnitude Estimates for Low-to-Moderate Seismicity Regions based on European–Mediterranean Data: Application to Metropolitan France,” Geophysical Journal International, vol. 230, no. 3, pp. 1980-2002, 2022.
    [CrossRef] [Google Scholar] [Publisher Link]
30. Ahmed Ibrahim Ali et al., “Advancements in Artificial Intelligence for Seismic Event Monitoring: A Comprehensive Review,” ERU Research Journal, vol. 5, no. 1, pp. 3814-3837, 2026.
    [CrossRef] [Google Scholar] [Publisher Link]
31. U. V. Anbazhagu et al., AI and Machine Learning in Earthquake Prediction: Enhancing Precision and Early Warning Systems, IGI Global Scientific Publishing, pp. 1-32, 2025.
    [CrossRef] [Google Scholar] [Publisher Link]