Lithological identification for mixed carbonate and clay sediments employing principal component analysis of...

This study proposes a novel method for lithological identification to address the challenge of accurate classification in complex sedimentary successions. Leveraging principal component analysis, the methodology was developed using Ordovician data from the Western Ordos Basin, where the lithological assemblage predominantly comprises of marine carbonate rocks, claystones, and their associated transitional rocks. Five well-log curves highly sensitive to lithology were selected as original variables, from which two principal components (PCs) were extracted to capture key compositional discrepancies. The first PC effectively differentiates clay and carbonate minerals, thereby enabling identification of transitional rocks between carbonates and claystones. The second PC reflects variations in calcite and dolomite contents, allowing discrimination of transitional lithologies between limestone and dolostone. Following the geological interpretation of these two principal components, a lithological cross-plot was constructed, which reveals distinct clustering among various rock types and supports a well-defined classification framework for transitional lithologies. The reliability and accuracy of the proposed method were validated against X-ray fluorescence element logging data from an active borehole. By prioritizing geological interpretability through feature transformation rather than treating the classification as a "black box", this study establishes a transparent, observable, and highly practical framework for lithological identification.