青藏高原复杂地形优化网格与多源信息同化研究

The Tibetan plateau is famous for its large areas, high altitude and complex terrain. In order to overcome the computation error and computation instability problems aroused by complex terrain in numerical weather prediction model, the optimization grid meshes and the multivariate information assimilation means are used to solve two problems including the computation error near steep terrain and the obtaining of initialization data on fine grid mesh respectively. Firstly, the posteriori error estimator is used to guide the subdivision of terrain grid mesh and to form nesting optimization grid mesh. Secondly, the techniques that were used in nesting grid mesh are applied to treat the terrain, computing noise and high false reflective waves on the interface of coarse grid mesh and fine grid mesh in optimization grid mesh for the computational stability. Last, the multivariate information assimilation method is used to obtain the initialization data on the fine grid mesh so that we can improve the data quality of the initialization field on fine grid mesh. The derivation of posteriori error estimator that guide the subdivision of grid mesh is the key this project. The treatment of interface conditions on the region with fine grid mesh and its multivariate information assimilation are the difficulty of this project. It is the innovation point of the project that optimization grid mesh method is used to solve the computational problems caused by complex terrain. If this project is.accomplished smoothly, the ability of numerical predict the Tibetan plateau is improved greatly.

青藏高原以面积大、海拔高、地形复杂而著称。为克服由复杂地形引起的数值天气预报模式的计算误差和计算不稳定问题，本项目应用优化网格与多源信息同化技术分别解决陡峭地形附近计算误差大及细网格初始条件确定难两大问题。首先，用后验误差估计式指导复杂地形网格的加密，形成嵌套优化网格。然后，借用嵌套网格技巧处理优化网格中粗细网格交界处的地形值，计算噪音及寄生伪反射波等影响计算稳定性的问题。最后，应用多源信息同化方法获得细网格上的初始条件，以提高细网格初始场的数据质量。推导指导优化网格加密的后验估计式是本项目的关键。处理细网格边界条件及其多源信息的同化是该项目的难点。用优化网格方法解决因复杂地形引起的计算问题是本项目的创新点。本项目的顺利完成能较大程度地提高青藏高原的数值预报能力。