Random forest-based peak ground acceleration prediction for shallow crustal earthquakes in Northern Sumatra: A comparison of hyperparameter optimization methods

Raka Syahfaqih Silalahi; Lailatul Husna Lubis; Russell Ong; Angga Wijaya

doi:10.58524/cesl.v1i2.167

Authors

Raka Syahfaqih Silalahi Department of Physics, Faculty of Science and Technology, Universitas Islam Negeri Sumatera Utara Author
Lailatul Husna Lubis Department of Physics, Faculty of Science and Technology, Universitas Islam Negeri Sumatera Utara Author
Russell Ong Department of Physics, Faculty of Science and Technology, Universitas Islam Negeri Sumatera Utara Author
Angga Wijaya Agency for Meteorology, Climatology, and Geophysics (BMKG) Author

DOI:

https://doi.org/10.58524/cesl.v1i2.167

Keywords:

Peak ground acceleration, Random Forest, Hyperparameter optimization, Bayesian Optimization, Shallow crustal earthquakes, Northern Sumatra

Abstract

Reliable prediction of peak ground acceleration (PGA) is essential for seismic hazard assessment in tectonically active regions such as northern Sumatra, where shallow crustal earthquakes are associated with complex source, propagation-path, and site effects. This study evaluates three hyperparameter optimization strategies—Grid Search, Randomized Search, and Bayesian Optimization—for improving Random Forest-based PGA prediction. The dataset consists of 2,509 PGA records from 453 shallow crustal earthquakes recorded in northern Sumatra during 2017–2023. Moment magnitude, epicentral distance, hypocentral depth, dominant soil period, and altitude were used as predictor variables, while PGA was transformed to its natural logarithm before model development. Model configurations were optimized using five-fold cross-validation and evaluated using mean squared error (MSE), root mean squared error (RMSE), and the coefficient of determination (R²). Bayesian Optimization yielded the lowest cross-validation MSE of 0.9847, compared with 0.9855 for Randomized Search and 0.9857 for Grid Search, indicating only marginal differences in predictive accuracy among the three approaches. Randomized Search was computationally most efficient, requiring 754.2 s, compared with 1,353.6 s for Bayesian Optimization and 1,677.7 s for Grid Search. The optimized Random Forest model achieved a test MSE of 0.9851, RMSE of 0.9925, and (R²) of 0.6078. These results indicate that Bayesian Optimization provides a slight advantage in predictive accuracy, whereas Randomized Search offers a more favorable trade-off between accuracy and computational efficiency. The findings highlight the importance of hyperparameter optimization strategy in developing data-driven ground-motion models for shallow crustal earthquakes in northern Sumatra.

References

1. Abrahamson, N. A., Gregor, N., and Addo, K. (2016). BC Hydro Ground Motion Prediction Equations for Subduction Earthquakes, Earthquake Spectra, Vol. 32, No. 1, 23–44. doi:10.1193/051712EQS188MR.

2. Atkinson, G. M., and Boore, D. M. (2003). Empirical Ground-Motion Relations for Subduction-Zone Earthquakes and Their Application to Cascadia and Other Regions, Bulletin of the Seismological Society of America, Vol. 93, No. 4, 1703–1729. doi:10.1785/0120020156.

3. Baltay, A. S., Abrahams, L. S., and Hanks, T. C. (2020). When Source and Path Components Trade Off in Ground-Motion Prediction Equations, Seismological Research Letters, Vol. 91, No. 4, 2259–2267. doi:10.1785/0220190379.

4. Baltay, A. S., Hanks, T. C., and Abrahamson, N. A. (2017). Uncertainty, Variability, and Earthquake Physics in Ground-Motion Prediction Equations, Bulletin of the Seismological Society of America, Vol. 107, No. 4, 1759–1771. doi:10.1785/0120160164.

5. Boore, D. M., and 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, Vol. 24, No. 1, 99–138. doi:10.1193/1.2830434.

6. Boore, D. M., Stewart, J. P., Seyhan, E., and Atkinson, G. M. (2014). NGA-West2 Equations for Predicting PGA, PGV, and 5% Damped PSA for Shallow Crustal Earthquakes, Earthquake Spectra, Vol. 30, No. 3, 1057–1085. doi:10.1193/070113EQS184M.

7. Bozorgnia, Y., Hachem, M. M., and Campbell, K. W. (2010). Ground Motion Prediction Equation (“Attenuation Relationship”) for Inelastic Response Spectra, Earthquake Spectra, Vol. 26, No. 1, 1–23. doi:10.1193/1.3281182.

8. Campbell, K. W., and Bozorgnia, Y. (2008). NGA Ground Motion Model for the Geometric Mean Horizontal Component of PGA, PGV, PGD and 5% Damped Linear Elastic Response Spectra, Earthquake Spectra, Vol. 24, No. 1, 139–171.

9. Chiou, B. S.-J., and Youngs, R. R. (2008). An NGA Model for the Average Horizontal Component of Peak Ground Motion and Response Spectra, Earthquake Spectra, Vol. 24, No. 1, 173–215. doi:10.1193/1.2894836.

10. Cornell, C. A. (1968). Engineering Seismic Risk Analysis, Bulletin of the Seismological Society of America, Vol. 58, No. 5, 1583–1606.

11. 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, Vol. 61, No. 1–2, 43–104. doi:10.1016/S0012-8252(02)00112-5.

12. Douglas, J. (2011). Ground-Motion Prediction Equations 1964–2010, PEER Report 2011/102, Pacific Earthquake Engineering Research Center, University of California, Berkeley.

13. Douglas, J. (2024). Ground Motion Equation Predictions 1964–2023 (Incomplete), University of Strathclyde.

14. Douglas, J., and Edwards, B. (2016). Recent and Future Developments in Earthquake Ground Motion Estimation, Earth-Science Reviews, Vol. 160, 203–219. doi:10.1016/j.earscirev.2016.07.005.

15. Erlangga, W. (2020). Karakteristik dan Parameter Subduksi Sumber Gempa Pulau Jawa, Teknisia, Vol. 25, No. 2, 88–98. doi:10.20885/teknisia.vol25.iss2.art4.

16. Zhao, J. X., Zhang, J., Asano, A., Ohno, Y., Oouchi, T., Takahashi, T., Ogawa, H., Irikura, K., Thio, H. K., Somerville, P. G., Fukushima, Y., and Fukushima, Y. (2006). Attenuation Relations of Strong Ground Motion in Japan Using Site Classification Based on Predominant Period, Bulletin of the Seismological Society of America, Vol. 96, No. 3, 898–913. doi:10.1785/0120050122.

17. Kanno, T., Narita, A., Morikawa, N., Fujiwara, H., and Fukushima, Y. (2006). A New Attenuation Relation for Strong Ground Motion in Japan Based on Recorded Data, Bulletin of the Seismological Society of America, Vol. 96, No. 3, 879–897.

18. Kotha, S. R., Cotton, F., and Bindi, D. (2018). A New Approach to Site Classification: Mixed-Effects Ground Motion Prediction Equation with Spectral Clustering of Site Amplification Functions, Soil Dynamics and Earthquake Engineering, Vol. 110, 318–329. doi:10.1016/j.soildyn.2018.01.051.

19. Kotha, S. R., Weatherill, G., Bindi, D., et al. (2020). A Regionally-Adaptable Ground-Motion Model for Shallow Crustal Earthquakes in Europe, Bulletin of Earthquake Engineering, Vol. 18, 4091–4125. doi:10.1007/s10518-020-00869-1.

20. Kramer, S. L. (1996). Geotechnical Earthquake Engineering, Prentice Hall, Upper Saddle River, New Jersey.

21. McCaffrey, R. (2009). The Tectonic Framework of the Sumatran Subduction Zone, Annual Review of Earth and Planetary Sciences, Vol. 37, 345–366. doi:10.1146/annurev.earth.031208.100212.

22. McGuire, R. K. (2004). Seismic Hazard and Risk Analysis, Earthquake Engineering Research Institute, Oakland, California.

23. PuSGeN. (2017). Peta Sumber dan Bahaya Gempa Indonesia Tahun 2017, Pusat Studi Gempa Nasional, Kementerian Pekerjaan Umum dan Perumahan Rakyat, Indonesia.

24. Sahakian, V. J., Baltay, A. S., Hanks, T. C., Buehler, J. S., Vernon, F. L., and Kilb, D. (2018). Decomposing Leftovers: Event, Path, and Site Residuals for a Small Magnitude ANZA Region GMPE, Bulletin of the Seismological Society of America, Vol. 108, No. 5A, 2800–2813. doi:10.1785/0120170376.

25. Scherbaum, F., Cotton, F., and Smit, P. (2004). On the Use of Response Spectral-Reference Data for the Selection and Ranking of Ground-Motion Models for Seismic-Hazard Analysis in Regions of Moderate Seismicity: The Case of Rock Motion, Bulletin of the Seismological Society of America, Vol. 94, No. 6, 2164–2185. doi:10.1785/0120030147.

26. Stewart, J. P., Douglas, J., Javanbarg, M., Bozorgnia, Y., Abrahamson, N. A., Boore, D. M., Campbell, K. W., Delavaud, E., Erdik, M., and Stafford, P. J. (2015). Selection of Ground Motion Prediction Equations for the Global Earthquake Model, Earthquake Spectra, Vol. 31, No. 1, 19–45. doi:10.1193/013013EQS017M.

27. Youngs, R. R., Chiou, S.-J., Silva, W. J., and Humphrey, J. R. (1997). Strong Ground Motion Attenuation Relationships for Subduction Zone Earthquakes, Seismological Research Letters, Vol. 68, No. 1, 58–73. doi:10.1785/gssrl.68.1.58.

28. Zhang, M., and Pan, H. (2021). Application of Generalized Pareto Distribution for Modeling Aleatory Variability of Ground Motion, Natural Hazards, Vol. 108, No. 3, 2971–2989. doi:10.1007/s11069-021-04809-3.

29. Irsyam, M., et al. (2020). Peta Sumber dan Bahaya Gempa Indonesia Tahun 2017, Edisi Revisi 2020, Pusat Studi Gempa Nasional (PuSGeN), Kementerian Pekerjaan Umum dan Perumahan Rakyat, Indonesia.