A layered three-dimensional noalinear numerical model was constructed to simulate the generation and propagation of interanal tides over the continental slope. The simulation was split into external mode computation (...A layered three-dimensional noalinear numerical model was constructed to simulate the generation and propagation of interanal tides over the continental slope. The simulation was split into external mode computation (EMC) and internal mode computation (IMC) to minimize the computational load.IMC was carried out once afte EMC was implemented N time. As to EMC, a semi-implicit numerical scheme was applied in such a way that the pressure gradient terms and the velocity divergence terms were discretized semi-implicitly, but the other terms were discretized explicitly. Eulerian-Lagrangian explicit discretization are applied to the convective terms simultaneously. As a result, the stability of EMC did not depend on the wave celerity and time step was not limited by the CFL condition. More than that, use of the conjugate gradient accelerated Jacobi method further improved the computational efficiency of the model.展开更多
文摘A layered three-dimensional noalinear numerical model was constructed to simulate the generation and propagation of interanal tides over the continental slope. The simulation was split into external mode computation (EMC) and internal mode computation (IMC) to minimize the computational load.IMC was carried out once afte EMC was implemented N time. As to EMC, a semi-implicit numerical scheme was applied in such a way that the pressure gradient terms and the velocity divergence terms were discretized semi-implicitly, but the other terms were discretized explicitly. Eulerian-Lagrangian explicit discretization are applied to the convective terms simultaneously. As a result, the stability of EMC did not depend on the wave celerity and time step was not limited by the CFL condition. More than that, use of the conjugate gradient accelerated Jacobi method further improved the computational efficiency of the model.
