Predicting the Geothermal Gradient in Colombia: a Machine Learning Approach

2024-04-08

Geothermics

AI Summary

Sparse in-situ geothermal gradient measurements and low accuracy of conventional empirical models hinder reliable geothermal resource assessment in Colombia. Method: This study develops, for the first time, a high-resolution (1 km) machine learning spatial prediction framework integrating heterogeneous geospatial features—including geological, geophysical, and topographic variables—optimized via advanced feature engineering to enhance generalization. We employ Random Forest, XGBoost, and Geographically Weighted Regression, rigorously validated through k-fold cross-validation and interpreted using SHAP (Shapley Additive Explanations). Contribution/Results: The best-performing model achieves an R² of 0.89—substantially outperforming classical empirical formulas. We generate a national-scale geothermal gradient map and identify five underexplored high-potential target zones. These results provide a robust, data-driven foundation for quantitative resource evaluation and strategic exploration site selection in Colombia.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Machine learning and data science for the Web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Problem

• Predict geothermal gradient using machine learning in Colombia
• Address lack of direct geothermal measurements in large regions
• Estimate geothermal potential with global geophysical datasets

Innovation

• Predicts geothermal gradient using machine learning
• Employs Gradient Boosted Regression Tree algorithm
• Leverages global geophysical datasets for predictions