Abstract: Prediction of total organic carbon (TOC) content represents a critical step in the geological evaluation of source rocks. Accurate TOC prediction based on well logging data serves as the foundation for comprehensive source rock evaluation. Current logging⁃based TOC evaluation methods can be categorized into four types: linear statistical method, curve overlay method, elemental logging method, and artifi⁃ cial intelligence⁃based method. Focusing on the marine shale reservoirs of the Permian Dalong formation in the eastern Sichuan Basin, an in ⁃depth analysis was conducted on the contributors to high TOC in the lower section of the formation. Then, a new logging⁃based TOC evalua⁃ tion method was proposed by incorporating classified clay mineral types. This method effectively addresses the limitations of conventional logging⁃based TOC evaluation methods which suffer from low accuracy due to the huge abundance of bentonite and pyrite and the incom⁃ plete illitization of smectite in the Dalong formation. The results indicate that, compared with conventional logging⁃based TOC evaluation methods, the new method achieves higher computational accuracy, with a mean relative error of 2.16%, and suggests a robust adaptability to the complex geological setting and strong reservoir heterogeneity in the study area. Furthermore, when applied to validation wells in other marine blocks, the proposed logging⁃based TOC evaluation method incorporating classified clay mineral types demonstrates a better agree⁃ ment with the trend derived from core measurements.

Key words: Sichuan Basin, Permian, Dalong formation, shale gas, TOC, source rock, ΔLogR