【Objective】Aiming at the key challenges in 3D geological modeling of deep concealed manganese deposits, including insufficient multi-source data integration, lack of synergistic constraints, and difficulties in characterizing ultra-thin orebodies, this study takes the super-large fully concealed Daotuo manganese deposit in Guizhou Province as the research object. By integrating multi-source heterogeneous data such as DEM, geological mapping, drilling logs, exploration line sections, remote sensing images, and audio magnetotelluric (AMT) sounding data, an integrated 3D geological model covering the surface, strata, structures, and orebodies is constructed.【Methods】A systematic workflow was established, including standardized data processing, multi-source information synergistic constraints, and incremental simulation of ultra-thin orebodies. A total of 18 modeling units were defined, and surface fitting, topological reasoning, and Kriging interpolation were adopted to achieve accurate construction of geological bodies. Subsequently, cross-section analysis, virtual borehole verification, model dissection, and geostatistical reserve estimation were carried out.【Results】The results show that the model accurately reproduces the three-dimensional spatial structure of the study area from the surface down to an elevation of -2000 m. It clearly reveals the spatial occurrence regularities of the manganese-bearing rock series of the Datangpo Formation, Nanhua System, the F3 fault, and the stratiform orebodies. The model can effectively verify the reliability of raw data and validate the spatial relationships between structures and orebodies.【Conclusion】The proposed method significantly improves the accuracy of deep ore prospecting deployment and resource evaluation.