Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches
Abstract
This study delivers a robust earthquake risk analysis for the highly seismically active Pazarcık region in Kahramanmaraş, Turkey. By examining a substantial dataset of 2,148 earthquake occurrences spanning from 1114 to 2023, including three significant historical records, we meticulously calculated earthquake probabilities using a diverse array of seven cutting-edge statistical and machine learning methodologies. Furthermore, we effectively estimated the location and depth of the impactful earthquake that struck on February 6, 2023. The analytical methods employed—ranging from the Poisson process, Gutenberg-Richter law, Weibull distribution, Gamma distribution, and the ETAS model to Extreme Value Theory and the sophisticated XGBoost algorithm—enabled us to rigorously evaluate the probabilities of earthquakes with magnitudes M≥3, M≥4, M≥5, M≥6, and M≥7 over timeframes extending from 1 to 250 years. Our findings convey a striking probability estimate: the likelihood of an M≥7 earthquake occurring in the region within the next 50 years is between 45.6% and 51.2% (averaging at 45.9%), escalating to 70.5% within 100 years and a staggering 94.9% within 250 years. In our pursuit of precise location and depth estimations, the Kernel Density Estimation (KDE) method emerged as the most reliable tool. It achieved an impressively low margin of error of (−0.0004°, −0.0092°) for the coordinates of the February 6, 2023 earthquake, alongside approximately a 0.5 km margin of error in depth estimation. Overall, the depth estimation accuracy reached an astounding 98.5%. This comprehensive study powerfully demonstrates that employing a multi-method approach not only enhances the reliability of results but is essential for accurate earthquake hazard risk analyses.
Keywords
References
- Aksoy, E., İnceöz, M., & Koçyiğit, A. (2007). Lake Hazar basin: A negative flower structure on the East Anatolian Fault System (EAFS), SE Turkey. Turkish Journal of Earth Sciences, 16(3), 319-338.
- Ambraseys, N. (2009). Earthquakes in the Mediterranean and Middle East: A multidisciplinary study of seismicity up to 1900. Cambridge University Press. https://doi.org/10.1017/S0016756810000452
- Arpat, E., & Şaroğlu, F. (1972). Doğu Anadolu Fayı ile ilgili bazı gözlemler ve düşünceler. Maden Tetkik ve Arama Dergisi, 78(78), 44-50.
- Ateş, R., & Bayülke, N. (1977). 26 Mart 1977 Palu Elazığ Depremi. Deprem Araştırma Daire Başkanlığı (Yayımlanmamış Rapor), Ankara.
- B.Ü. KRDAE Bölgesel Deprem-Tsunami İzleme ve Değerlendirme Merkezi. (n.d.). Bölgesel Deprem Veritabanı. http://www.koeri.boun.edu.tr/sismo/zeqdb/
- 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/10.48550/arXiv.1603.02754
- Duman, T., & Emre, Ö. (2013). The East Anatolian Fault: Geometry, segmentation and jog characteristics. Geological Society, London, Special Publications, 372, 495-529. https://doi.org/10.1144/SP372.14
- Erisoğlu, M., Calış, N., Servi, T., Erişoğlu, Ü., & Topaksu, M. (2011). The mixture distribution models for interoccurrence times of earthquakes. Russian Geology and Geophysics, 52(7), 737-744. https://doi.org/10.1016/j.rgg.2011.06.007
Details
Primary Language
English
Subjects
Geology (Other)
Journal Section
Research Article
Authors
Publication Date
December 31, 2025
Submission Date
October 24, 2025
Acceptance Date
November 14, 2025
Published in Issue
Year 2025 Volume: 12 Number: 4
APA
Sevimli, U. İ. (2025). Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches. Gazi University Journal of Science Part A: Engineering and Innovation, 12(4), 1149-1168. https://doi.org/10.54287/gujsa.1809724
AMA
1.Sevimli Uİ. Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches. GU J Sci, Part A. 2025;12(4):1149-1168. doi:10.54287/gujsa.1809724
Chicago
Sevimli, Ulaş İnan. 2025. “Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches”. Gazi University Journal of Science Part A: Engineering and Innovation 12 (4): 1149-68. https://doi.org/10.54287/gujsa.1809724.
EndNote
Sevimli Uİ (December 1, 2025) Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches. Gazi University Journal of Science Part A: Engineering and Innovation 12 4 1149–1168.
IEEE
[1]U. İ. Sevimli, “Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches”, GU J Sci, Part A, vol. 12, no. 4, pp. 1149–1168, Dec. 2025, doi: 10.54287/gujsa.1809724.
ISNAD
Sevimli, Ulaş İnan. “Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches”. Gazi University Journal of Science Part A: Engineering and Innovation 12/4 (December 1, 2025): 1149-1168. https://doi.org/10.54287/gujsa.1809724.
JAMA
1.Sevimli Uİ. Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches. GU J Sci, Part A. 2025;12:1149–1168.
MLA
Sevimli, Ulaş İnan. “Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches”. Gazi University Journal of Science Part A: Engineering and Innovation, vol. 12, no. 4, Dec. 2025, pp. 1149-68, doi:10.54287/gujsa.1809724.
Vancouver
1.Ulaş İnan Sevimli. Multi-Method Earthquake Risk Analysis and Location-Depth Estimation for Pazarcık (Kahramanmaraş): Statistical and Machine Learning Approaches. GU J Sci, Part A. 2025 Dec. 1;12(4):1149-68. doi:10.54287/gujsa.1809724