Published

2026-09-05

Hybrid Ground Motion Prediction in Iran Using Regional GMPE and Machine Learning Residuals with SHAP-Based Interpretation and Uncertainty Analysis

Predicción Híbrida del Movimiento del Suelo en Irán Usando GMPE Regional y Residuos de Aprendizaje Automático con Interpretación Basada en SHAP y Análisis de Incertidumbre

DOI:

https://doi.org/10.15446/esrj.v30n(2).122761

Keywords:

Ground Motion Prediction Equation,, Shapley Additive Explanations,, XGBoost, Peak Ground Acceleration,, Multilayer Perceptron,, Quantitative uncertainty. (en)
Ecuación de Predicción del Movimiento del Suelo, Explicaciones Aditivas de Shapley, XGBoost, Aceleración Máxima del Suelo, Perceptrón Multicapa, Incertidumbre cuantitativa (es)

Downloads

Authors

  • Elham Boostan Islamic Azad University, Karaj Branch, Faculty of Basic Science, Department of Geology, Iran 0000-0002-2777-4941
  • Nima Vaziri Department of Geology, Ka. C., Islamic Azad University, Karaj, Iran
  • Ali Shafiee Department of Civil Engineering, California State Polytechnic University, Pomona, United States

Accurate prediction of peak ground acceleration (PGA) is vital for seismic hazard assessment, especially in tectonically complex regions like Iran. Global Ground Motion Prediction Equations (GMPEs) often lack local reliability, while machine learning (ML) methods can improve accuracy but remain opaque. This study introduces a hybrid framework that integrates a regionally calibrated GMPE—developed from Iranian seismic data—with ML-based residual modeling. Residuals were captured using an ensemble of Extreme Gradient Boosting (XGBoost) and a Multilayer Perceptron (MLP), combining structured and nonlinear learning. The hybrid model achieved an R² of 0.419 and RMSE of 0.089 in the log domain, outperforming standalone GMPE and ML models. To improve transparency, SHAP (SHapley Additive Explanations) identified the relative influence of magnitude, distance, and site conditions, while empirical confidence intervals quantified prediction uncertainty. By uniting regional calibration, predictive accuracy, interpretability, and uncertainty analysis, the proposed framework advances ground motion prediction in Iran and offers a transferable approach for other seismically active but data-limited regions.

La predicción precisa de la aceleración máxima del suelo (PGA) es vital para la evaluación del peligro sísmico, especialmente en regiones tectónicamente complejas como Irán. Las ecuaciones globales de predicción del movimiento del suelo (GMPE) suelen carecer de fiabilidad local, mientras que los métodos de aprendizaje automático (ML) pueden mejorar la precisión pero siguen siendo opacos. Este estudio introduce un marco híbrido que integra una GMPE calibrada regionalmente—a partir de datos sísmicos iraníes—con modelado de residuos basado en ML. Los residuos se capturaron utilizando un conjunto de Extreme Gradient Boosting (XGBoost) y un perceptrón multicapa (MLP), combinando aprendizaje estructurado y no lineal. El modelo híbrido alcanzó un R² de 0.419 y un RMSE de 0.089 en el dominio logarítmico, superando a los modelos GMPE y ML individuales. Para mejorar la transparencia, SHAP (SHapley Additive Explanations) identificó la influencia relativa de la magnitud, la distancia y las condiciones del sitio, mientras que los intervalos de confianza empíricos cuantificaron la incertidumbre de la predicción. Al unir calibración regional, precisión predictiva, interpretabilidad y análisis de incertidumbre, el marco propuesto mejora la predicción del movimiento del suelo en Irán y ofrece un enfoque transferible para otras regiones sísmicamente activas pero con datos limitados.

References

Abrahamson, N. A., & Youngs, R. R. (1992). A stable algorithm for regression analyses using the random effects model. Bulletin of the Seismological Society of America, 82(1), 505–510. https://doi.org/10.1785/BSSA0820010505

Abrahamson, N. A., Silva, W. J., & Kamai, R. (2014). Summary of the ASK14 ground motion relation for active crustal regions. Earthquake Spectra, 30(3), 1025–1055. https://doi.org/10.1193/070913EQS198M

AlHamaydeh, M., Tellab, S., & Tariq, U. (2024). Deep learning based strong ground motion signal prediction in real time. Buildings, 14(5), 1267. https://doi.org/10.3390/buildings14051267

Alidadi, N., & Pezeshk, S. (2024). Ground-motion model for small-to-moderate potentially induced earthquakes using an ensemble machine learning approach for CENA. Bulletin of the Seismological Society of America, 114(4), 2202–2215. https://doi.org/10.1785/0120230242

Aram, Z., & Arian, M. (2016). Active tectonics of the Gharasu River Basin in Zagros, Iran, investigated by calculation of geomorphic indices and group decision using analytic hierarchy process (AHP) software. Episodes, 39(1), 39–44. https://doi.org/10.18814/epiiugs/2016/v39i1/89235

Arian, M. (2015). Seismotectonic-geologic hazards zoning of Iran. Earth Sciences Research Journal, 19(1), 7–13. https://doi.org/10.15446/esrj.v19n1.40670

Boore, D. M., & Atkinson, G. M. (2008). Ground-motion prediction equations for the average horizontal component of PGA, PGV, and 5%-damped PSA at spectral periods between 0.01 s and 10.0 s. Earthquake Spectra, 24(1), 99–138. https://doi.org/10.1193/1.2830434

Boore, D. M., Stewart, J. P., Seyhan, E., & Atkinson, G. M. (2014). NGA-West2 equations for predicting PGA, PGV, and 5% damped PSA for shallow crustal earthquakes. Earthquake Spectra, 30(3), 1057–1085. https://doi.org/10.1193/070113EQS184M

Campbell, K. W., & Bozorgnia, Y. (2014). NGA-West2 ground motion model for the average horizontal components of PGA, PGV, and 5% damped linear acceleration response spectra. Earthquake Spectra, 30(3), 1087–1115. https://doi.org/10.1193/062913EQS175M

Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794. https://doi.org/10.1145/2939672.2939785

Chiou, B., & Youngs, R. R. (2014). Update of the Chiou and Youngs NGA model for the average horizontal component of peak ground motion and response spectra. Earthquake Spectra, 30(3), 1117–1153. https://doi.org/10.1193/072813EQS219M

Douglas, J. (2003). Earthquake ground motion estimation using strong-motion records: A review of equations for the estimation of peak ground acceleration and response spectral ordinates. Earth-Science Reviews, 61(1–2), 43–104. https://doi.org/10.1016/S0012-8252(02)00112-5

Farajpour, Z., Javidan, M., & Mirzaei, N. (2019). Seismic hazard zoning of Iran using updated historical and instrumental data. Natural Hazards, 95(1–2), 119–142.

Feng, D. C., Wang, W. J., Mangalathu, S., & Taciroglu, E. (2021). Interpretable XGBoost-SHAP machine-learning model for shear strength prediction of squat RC walls. Journal of Structural Engineering, 147(11), 04021173. https://doi.org/10.1061/(ASCE)ST.1943-541X.0003115

Gandomi, A. H., & Alavi, A. H. (2012). A new multi-gene genetic programming approach to non-linear system modeling. Part II: Geotechnical and earthquake engineering problems. Neural Computing and Applications, 21, 189–201. https://doi.org/10.1007/s00521-011-0735-y

Ghofrani, H., & Atkinson, G. M. (2014). Comparison of NGA-West2 GMPEs with global ground-motion models. Earthquake Spectra, 30(3), 1107–1126. https://doi.org/10.1193/070913EQS197M

Kotha, S. R., Bindi, D., & Cotton, F. (2020). Partially non-ergodic empirical ground-motion model for Europe and Middle East. Bulletin of Earthquake Engineering, 18, 4091–4116. https://doi.org/10.1007/s10518-020-00820-1

Lundberg, S. M., & Lee, S. I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.

Monterrubio-Velasco, M., Callaghan, S., Modesto, D., Carrasco Jiménez, J. C., Pallarés Font de Mora, P., Vázquez-Novoa, F., Quintana-Ortí, E. S., Pienkowska, M., & de la Puente, J. (2024). A machine learning estimator trained on synthetic data for real-time earthquake ground-shaking predictions in Southern California. Communications Earth & Environment, 5, Article 258. https://doi.org/10.1038/s43247-024-01436-1

Molnar, C. (2020). Interpretable Machine Learning (2nd ed.). https://christophm.github.io/interpretable-ml-book/

Mori, F., Mendicelli, A., Falcone, G., Acunzo, G., Spacagna, R. L., Naso, G., & Moscatelli, M. (2022). Ground motion prediction maps using seismic microzonation data and machine learning. Natural Hazards and Earth System Sciences, 22(3), 947–966. https://doi.org/10.5194/nhess-22-947-2022

Mousavi, S. M., & Beroza, G. C. (2023). Machine learning in earthquake seismology. Annual Review of Earth and Planetary Sciences, 51, 105–129. https://doi.org/10.1146/annurev-earth-071822-100323

Pozos-Estrada, A., Gómez, R., & Hong, H. P. (2014). Use of neural network to predict the peak ground accelerations and pseudo spectral accelerations for Mexican inslab and interplate earthquakes. Geofísica Internacional, 53(1), 39–57. https://doi.org/10.1016/S0016-7169(14)71489-8

Razaghian, G., Beitollahi, A., Pourkermani, M., & Arian, M. (2018). Determining seismotectonic provinces based on seismicity coefficients in Iran. Journal of Geodynamics, 119, 29–46. https://doi.org/10.1016/j.jog.2018.05.007

Soriano, C., Tarazona, S. F., Almeida, M. C. F., Almeida, M. S. S., & Borges, R. G. (2021). Selection of earthquake ground motion accelerograms for the continental margin of Southeastern Brazil. DYNA, 88(217), 228–236. https://doi.org/10.15446/dyna.v88n217.89687

Wang, X., Wang, Z., Wang, J., Miao, P., Dang, H., & Li, Z. (2023). Machine learning based ground motion site amplification prediction. Frontiers in Earth Science, 11, 1053085. https://doi.org/10.3389/feart.2023.1053085

Xie, Y., Ebad Sichani, M., Padgett, J. E., & DesRoches, R. (2020). The promise of implementing machine learning in earthquake engineering: A state-of-the-art review. Earthquake Spectra, 36(4), 1749–1771 https://doi.org/10.1177/8755293020919419

Zhan, W. W., & Chen, Q. S. (2024). Machine learning ground motion models for a critical site with long-term earthquake records. IOP Conference Series: Earth and Environmental Science, 1334, 012032. https://doi.org/10.1088/1755-1315/1334/1/012032

Dimensions

PlumX

Article abstract page views

90

Downloads

Download data is not yet available.

How to Cite

Boostan, E., Vaziri, N., & Shafiee, A. (2026). Hybrid Ground Motion Prediction in Iran Using Regional GMPE and Machine Learning Residuals with SHAP-Based Interpretation and Uncertainty Analysis. Earth Sciences Research Journal, 30((2), 155-162. https://doi.org/10.15446/esrj.v30n(2).122761

License

Copyright (c) 2026 Earth Sciences Research Journal

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Earth Sciences Research Journal holds a Creative Commons Attribution license. 

You are free to:

Share — copy and redistribute the material in any medium or format
Adapt — remix, transform, and build upon the material for any purpose, even commercially.
The licensor cannot revoke these freedoms as long as you follow the license terms.

The Earth Sciences Research Journal is the copyright holder for these license attributes.