Abstract: Elastic wave Kirchhoff prestack depth migration (PSDM) using ocean bottom node (OBN) data is an effective approach for high-resolution imaging of complex submarine geological structures, and its imaging quality is primarily determined by the accuracy of depth-domain P- and S-wave interval velocity models. To address key challenges in OBN multi-wave PSDM, including strong non-uniqueness in velocity model building, insufficient accuracy of initial S-wave velocity models, and poor depth consistency among different wave-mode images, a practical and application-oriented method for multi-wave depth-domain velocity modeling and migration imaging is proposed. The proposed method integrates horizon-matching velocity modeling with residual curvature-based grid tomography. By matching corresponding P- and S-wave seismic horizons, a high-precision initial S-wave interval velocity model is constructed, effectively mitigating error propagation associated with conventional Dix conversion and the assumption of a constant v_p/v_s ratio. Without relying on elastic full waveform inversion, the method further combines horizon-constrained and global residual curvature-based grid tomography to achieve efficient and stable iterative updates of both P- and S-wave velocity model. Numerical experiments and applications to real multi-component OBN seismic datasets demonstrate that the proposed method effectively flattens multi-wave common image gathers and significantly improves depth consistency among different wave-mode images, thereby leading to a substantial improvement in imaging quality. These results indicate that the proposed method is well suited for engineering applications and holds strong potential for practical production and wider industrial adoption.