EUVIMEDEUVIMED entdecken
Open Access · CC BY-SA 4.0

Intelligent Lithology Interpretation for Geological Survey: A Machine Learning Approach Integrating Multispectral Remote Sensing and Terrain Features — A Case Study in a Karst Terrain of Southwestern China

Yuan Jiang, Yaopei Li, Qiaoling Cheng, Guihong Wu

Journal of Computer Science and Frontier Technologies · 30.10.2026 · 4 (3) · 84

DOI: 10.63313/jcsft.9110 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Computer Science and Frontier Technologies, online erschienen am 30.10.2026. Verfasst von Yuan Jiang und 3 weiteren Autor:innen. Die Arbeit nennt 14 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Lithology identification is a fundamental task in geological surveying and mapping, yet the traditional visual interpretation of remote sensing imagery is labor-intensive and strongly dependent on expert experience. This study proposes a machine-learning-based approach for intelligent lithology interpretation that integrates multispectral remote sensing imagery with terrain features, and evaluates its applicability to supporting geological mapping in karst terrains. A typical karst terrain in southwestern China was selected as the study area because of its excellent bedrock exposure, sparse vegetation in rocky-desertified zones, and diverse Permian–Triassic lithologies. A multi-source feature set was constructed, comprising spectral features derived from simulated Landsat-like multi-spectral imagery (original reflective bands, principal components and band ratios), GLCM texture features, and terrain attributes (elevation, slope, aspect and curvature) extracted from a synthetic DEM. Random Forest (RF), Support Vector Machine (SVM) and XGBoost classifiers were trained using training samples derived from a synthetic geological reference map, and were evaluated by confusion matrices, overall accuracy (OA), the Kappa coefficient and per-class accuracies. Three experiments were designed: (1) feature ablation to quantify the contribution of texture and terrain features; (2) a cross-comparison of the three machine-learning models under a unified framework; and (3) a comparison with the conventional maximum-likelihood classifier. The results show that feature fusion improved the OA by 6.5 percentage points relative to the spectral-only scenario, with the most notable gains for the spectrally similar limestone–dolomite pair; XGBoost achieved the best overall performance (OA = 95.5%, Kappa = 0.944), followed by Random Forest and SVM; and the machine-learning classifiers outperformed the maximum-likelihood classifier by 9.3 percentage points in OA. The classified lithology maps were broadly consistent with the reference geological map in well-exposed carbonate areas but revealed discrepancies in vegetated and boundary zones. The findings demonstrate the potential of integrating terrain and texture features with machine learning for karst lithology interpretation and provide practical guidance for AI-assisted geological mapping.

Zitieren

APA

Jiang, Y., Li, Y., Cheng, Q., & Wu, G. (2026). Intelligent Lithology Interpretation for Geological Survey: A Machine Learning Approach Integrating Multispectral Remote Sensing and Terrain Features — A Case Study in a Karst Terrain of Southwestern China. Journal of Computer Science and Frontier Technologies, 4(3), 84. https://doi.org/10.63313/jcsft.9110

Vancouver

Jiang Y, Li Y, Cheng Q, Wu G. Intelligent Lithology Interpretation for Geological Survey: A Machine Learning Approach Integrating Multispectral Remote Sensing and Terrain Features — A Case Study in a Karst Terrain of Southwestern China. Journal of Computer Science and Frontier Technologies. 2026;4(3):84. doi:10.63313/jcsft.9110

BibTeX
@article{Jiang2026,
  author = {Jiang, Yuan and Li, Yaopei and Cheng, Qiaoling and Wu, Guihong},
  title = {Intelligent Lithology Interpretation for Geological Survey: A Machine Learning Approach Integrating Multispectral Remote Sensing and Terrain Features — A Case Study in a Karst Terrain of Southwestern China},
  journal = {Journal of Computer Science and Frontier Technologies},
  year = {2026},
  volume = {4},
  number = {3},
  pages = {84},
  doi = {10.63313/jcsft.9110}
}
Kostenlos starten →