The diffuse attenuation coefficient (Kd) is a crucial parameter in ocean optics, representing an apparent optical property influenced by the inherent optical characteristics of seawater and the surrounding light field. It is closely related to factors such as seawater quality and chlorophyll concentration. As an active remote sensing instrument, marine polarized lidar emits light in the blue-green wavelength band capable of penetrating seawater, offering all-weather detection potential, and possesses a distinct edge in mapping the vertical distribution of Kd within the ocean. By combining Fernald"s backward iteration and slope approaches, this study proposed a layered inversion method for oceanic profile Kd estimation, utilizing dual-polarization channel signals. The vertical polarization channel is specifically used to suppress surface signals and enhance near-shore oceanic backscatter. Conducted in the Yellow and East China Seas, the ocean lidar was mounted on a marine experimental platform, with a 10-meter water depth used to validate the stratification algorithm. Results show a polarization degree of 0.479 at the sea surface for the dual-polarization channel signal. With a vertical resolution of 1 meter, the stratified inversion of the oceanic profile Kd using dual-polarization channels yields a root mean square error of 0.049 compared to actual in-situ measurements, representing a 52.4% improvement in accuracy over non-polarized channel signals. Additionally, the layered inversion algorithm outperforms the traditional Fernald algorithm, demonstrating a 32.4% improvement in precision.