Abstract: Constructing the theory and simulation for multilayer phase screen of underwater turbulence were the main problems of the research for underwater turbulence. Considering the statistical property of underwater turbulence, here 3D covariance matrix was used to model the statistical property and continuity of underwater turbulence. Here overlapping allocation matrix was put forward and the simulation of multilayer phase screen was optimized. Furthermore, the relation between computational efficiency and the characterization of statistical property of underwater turbulence was analyzed. The results show that overlapping allocation matrix can improve the computational efficiency for a layer of phase screen. Also utilizing 3D covariance matrix could obtain the correlation of phase screens, which is in conformity with theory for statistical property in a layer. However, utilizing 3D covariance matrix led to deviation in small scale and large scale. This research expends original independent 2D phase screen into the multilayer phase screen which has spatial correlation. And that is why this model is more correspond to the reality of underwater turbulence.