Enhancing Geothermal Exploration with HybridSHAP: A Machine Learning-Based Geothermal Gradient Prediction Using Explainable AI

ABSTRACT

The global drive for renewable energy has intensified interest in geothermal energy, a sustainable resource harnessed from the Earth’s internal heat. Predicting geothermal gradients, which show temperature rises with depth, is essential for geothermal exploration, especially in country like Colombia with obvious volcanic activity. Conventional techniques for estimating geothermal gradients depend on expensive and time-consuming in-situ measurements. This paper proposes HybridSHAP, an intuitive hybrid ensemble learning approach that combines different algorithms such as XGBoost, LightGBM, Random Forest, and CatBoost, to accurately predict geothermal gradients. HybridSHAP integrates sophisticated feature engineering methods and explainable AI, particularly SHAP values, to improve model transparency and comprehend the impact of geological features on predictions. The performance of the model is evaluated using cross-validation metrics, obtaining a high R-squared (R²) value of 0.9525 and low RMSE of 0.00729, highlighting optimal predictive accuracy and generalization. The results obtained by the proposed model demonstrate that features such as elevation, fault types, and free-air anomalies evidently impact the prediction outcomes. The study claims that HybridSHAP efficiently predicts geothermal gradients, providing a practical platform for guiding exploration and optimizing drilling strategies with the ability to transform geothermal exploration and support sustainable energy development.

References  

• [1] Alfaro, J. B. Rueda-Gutiérrez, Y. Casallas, G. Rodríguez, and J. Malo, “Approach to the geothermal potential of Colombia,” Geothermics, vol. 96, p. 102169, 2021, doi: 10.1016/j.geothermics.2021.102169.
• [2] Alfaro, C. A. Vargas, and J. Gutiérrez, “Mapa preliminar de gradientes geotérmicos de Colombia,” Servicio Geológico Colombiano, 2009.
• [3] Alqahtani et al., “Machine learning techniques in predicting bottom hole temperature and remote sensing for assessment of geothermal potential in the Kingdom of Saudi Arabia,” Sustainability, vol. 15, no. 17, p. 12718, 2023, doi: 10.3390/su151712718.
• [4] ANH (Agencia Nacional de Hidrocarburos), “Llanos orientales,” https://www.anh.gov.co/es/hidrocarburos/oportunidades-disponibles/procesosdeseleccion/ronda-colombia-2010/tipo-1/llanos-orientales/, 2010.
• [5] ANH (Agencia Nacional de Hidrocarburos), “Valle medio del magdalena,” https://www.anh.gov.co/es/hidrocarburos/oportunidades-disponibles/procesosdeseleccion/ronda-colombia-2010/tipo-1/valle-medio-del-magdalena/, 2010.
• [6] Assouline, N. Mohajeri, A. Gudmundsson, and J. L. Scartezzini, “A machine learning approach for mapping the very shallow theoretical geothermal potential,” Geotherm. Energy, vol. 7, no. 1, pp. 1–50, 2019, doi: 10.1186/s40517-019-0135-6.
• [7] Bachu, J. C. Ramon, M. E. Villegas, and J. R. Underschultz, “Geothermal regime and thermal history of the Llanos Basin, Colombia,” AAPG Bull., vol. 79, no. 1, pp. 116–128, 1995, doi: 10.1306/8D2B14D0-171E-11D7-8645000102C1865D.
• [8] Barrero, A. Pardo, C. A. Vargas, and J. F. Martínez, Colombian Sedimentary Basins: Nomenclature, Boundaries and Petroleum Geology, a New Proposal, vol. 1, Agencia Nacional de Hidrocarburos, 2007. [Online]. Available: https://www.anh.gov.co/documents/12/colombian_sedimentary_basins.pdf
• [9] Békési et al., “An updated geothermal model of the Dutch subsurface based on inversion of temperature data,” Geothermics, vol. 88, p. 101880, 2020, doi: 10.1016/j.geothermics.2020.101880. A.Burton-Johnson, R. Dziadek, and C. Martin, “Geothermal heat flow in Antarctica: Current and future directions,” Cryosphere Discuss., vol. 1, pp. 1–45, 2020, doi: 10.5194/tc-14-3843-2020.
• Chen and C. Guestrin, “Xgboost: A scalable tree boosting system,” in Proc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining, vol. 78, pp. 5–794, 2016, doi: 10.1145/2939672.2939785.
• Clavijo, L. Mantilla, J. Pinto, L. Bernal, and A. Pérez, “De San Lucas, norte del valle medio del Magdalena y noroeste de la Cordillera Oriental,” Boletín Geol., vol. 30, no. 1, pp. 45–62, 2008.
• de Porta, “La formación del istmo de Panamá: su incidencia en Colombia,” Revista de la Academia Colombiana de Ciencias Exactas, Físicasy Naturales, vol. 27, no. 103, p. 2062, 2003, doi: 10.18257/raccefyn.27(103).2003.2062.
• Dyment et al., “The world digital magnetic anomaly map: version 2.1,” in EGU General Assembly Conf. Abstracts, p. 22098, 2020, doi: 10.5194/egusphere-egu2020-22098.
• H. Friedman, “Greedy function approximation: a gradient boosting machine,” Ann. Statist., vol. 118, pp. 9–1232, 2001, doi: 10.1214/aos/1013203451.
• [15] Gómez, N. E. Montes, A. Nivia, and H. Diederix, Mapa Geológico de Colombia, Escala 1:1.000.000, Servicio Geológico Colombiano, 2015. [Online]. Available: https://datos.sgc.gov.co/maps/c05c6dbf27f645eb883bae3a9cd0d08f.
• Gómez-Tapias et al., “Physiographic and geological setting of the Colombian territory,” Geol. Colombia, vol. 1, pp. 1–6, 2020, doi: 10.32685/pub.esp.35.2019.01.
• Goutorbe, J. Poort, F. Lucazeau, and S. Raillard, “Global heat flow trends resolved from multiple geological and geophysical proxies,” Geophys. J. Int., vol. 187, no. 3, pp. 1405–1419, 2011, doi: 10.1111/j.1365-246X.2011.05228.x.
• Hasterok, M. Gard, and J. Webb, “On the radiogenic heat production of metamorphic, igneous and sedimentary rocks,” Geosci. Front., vol. 9, no. 6, pp. 1777–1794, 2018, doi: 10.1016/j.gsf.2017.10.012.
• Kolawole and J. C. Evenick, “Global distribution of geothermal gradients in sedimentary basins,” Geosci. Front., vol. 14, no. 6, p. 101685, 2023, doi: 10.1016/j.gsf.2023.101685.

Download all article in PDF

WSN 198 (2024) 139-162

ADVERTISEMENT