Abstract: To address the challenge of long-range deep-sea localization of surface targets using small-aperture two-dimensional (2D) planar arrays, this study proposes a method that exploits the diversity of information embedded in deep-sea multipath acoustic propagation. By applying spatial beamforming to the 2D planar array, the horizontal azimuth and vertical elevation angles—and the corresponding multipath information—of the target’s radiated noise are directly extracted. Based on geometric relationships, a localization algorithm model is established; its applicable conditions are derived, and the associated errors are analyzed and quantified. Numerical simulation results demonstrate that, with only prior knowledge of ocean depth, the proposed method achieves a localization accuracy better than 6% (i.e., position error less than 6% of the slant range) by utilizing the incident angle information of first-order reflected acoustic paths within a 10-km range from the target. This significantly enhances the generalization capability of 2D horizontal planar arrays for deep-sea localization of surface targets.