The Benxi Formation coal bed in the Ordos Basin represent one of the most important reservoirs for Coalbed Methane (CBM) in China. However, the heterogeneous distribution of gangue significantly affects coal bed continuity, gas content, reservoir quality, and the design of horizontal well trajectories. Accurately characterizing gangue distribution is therefore essential for improving CBM exploration and development efficiency. This study aims to establish an integrated geological-geophysical workflow for high-precision geomorphology reconstruction and gangue prediction within the Benxi Formation, addressing the limitations of conventional seismic data such as insufficient vertical resolution and blurred stratigraphic boundaries. To achieve this objective, we introduce the Steerable Pyramid seismic micro-attribute enhancement method, which effectively improves the clarity of seismic reflection geometries and enhances vertical resolution while preserving the original structural framework. Based on the enhanced seismic volume, the "imprint method" is employed to reconstruct the paleogeomorphology of the 8 # coal bed in eastern Yulin. Log data analysis indicates that natural gamma (GR) is the most sensitive indicator for gangue development; thus, GR is selected as the key constraint in seismic inversion. By integrating high-precision paleogeomorphologic boundaries into the inversion workflow, a refined three-dimensional gangue distribution model is constructed. Result demonstrates that the development of gangue within the Benxi Formation is predominantly controlled by subtle paleogeomorphologic undulations. Areas corresponding to gentle paleodepressions exhibit minimal gangue content and better coal continuity, representing favorable zones for CBM enrichment and production. The gangue prediction outcomes show high consistency with log and data across multiple well profiles, confirming the reliability and geological validity of the proposed workflow. In conclusion, this study establishes a robust and practical method for accurately characterizing gangue spatial distribution in coal bed through integrated paleogeomorphology analysis and seismic inversion. The findings not only clarify the coupling relationship between paleogeomorphology and gangue development but also provide valuable method support for well placement optimization, sweet-spot identification and efficient CBM development in the Ordos Basin.