【摘要】：In order to quantitatively evaluate the spurious dianeutral mixing in a global ocean model MPAS-Ocean(Model for Prediction Across Scales) using a spherical centroidal voronoi tessellations developed jointly by the National Center for Atmospheric Research and the Los Alamos National Laboratory in the United States, we choose z* vertical coordinate system in MPAS-Ocean, in which all physical mixing processes, such as convection adjustment and explicit diffusion parameter schemes, are omitted, using a linear equation of state. By calculating the Reference Potential Energy(RPE), front revolution position, time rate of RPE change, probability density function distribution and dimensionless parameter χ, from the perspectives of resolution, viscosity, Horizontal Grid Reynolds Number(HGRN), Re?, and momentum transmission scheme, using two ideal cases, overflow and baroclinic eddy channel, we qualitatively analyze the simulation results by comparison with the three non-isopycnal models in Ilicak et al.(2012), i.e., MITGCM, MOM, and ROMS. The results show that the spurious dianeutral mixing in the MPAS-Ocean increases over time. The spurious dianeutral transport is proportional to the HGRN directly and is reduced by increasing the lateral viscosity or using a finer resolution to control HGRN. When the HGRN is less than 10, spurious transport is reduced significantly. When using the proper viscosity closure, MPAS-Ocean performs better than MITGCM and MOM, closely to ROMS, in the 2D case without rotation, and much better than the above-mentioned three ocean models under the condition of 3D space with rotation due to the cell area difference between the hexagon cell and the quadrilateral cell with the same resolution. Both the Zalesak(1979) flux corrected transport scheme and Leith closure in MPAS-Ocean play an excellent role in reducing spurious dianeutral mixing. The performance of Leith scheme is preferable to the condition of three-dimensional baroclinic eddy.