The multi- polarization magnetotelluric method (MPMT) is a novel electromagnetic technique for detecting geological bodies characterized by contrasts in both resistivity and magnetic susceptibility. Based on the requirements for deep geological surveys in the Yueshanpu ore block of Ningxiang, Hunan, this study established an electromagnetic model of carbonate- hosted manganese deposits by integrating petrophysical statistics with existing borehole data. Forward modeling of MPMT was conducted to evaluate the method' s effectiveness. Using the Nonlinear Conjugate Gradient (NLCG) inversion technique, we first inverted data from known profiles to determine optimal MPMT inversion parameters. We then calculated 2D distributions of resistivity and magnetic susceptibility for three survey lines. By synthesizing the petrophysical characteristics of manganese ores and surrounding rocks, an exploration model was developed to delineate the 3D spatial distribution of ore layers in the Yueshanpu block. Verification drill holes confirmed that the dual- parameter information (resistivity and magnetic susceptibility) obtained from MPMT effectively constrains the spatial geometry of manganese orebodies. This approach reduces inherent ambiguities in geophysical interpretation and enhances prediction accuracy.