Development of micro-meso scale atmospheric environmental simulator with CIP finite volume method

点击翻译按钮获取中文摘要

High performance numerical models play an essential role in accessing and predicting natural disasters and environmental hazards in earth's atmosphere. We, under this research project, carried out some fundamental studies concerning the key issues in constructing numerical models for atmospheric dynamics with high accuracy, efficiency and robustness. Our major achievements are as follows.1)We have proposed a unified numerical procedure for both compressible and incompressible flows using the CIP/multi-moment finite volume framework. The model eliminates all the singularities for Mach number ranging from zero to infinite and possesses rigorously the numerical conservativeness.2) A new grid based on the multi-moment concept, so called "M-grid", has been devised. Theoretical analysis and numerical experiments show that the M-grid has better numerical dispersion than any Arakawa-type finite difference grid for all grid resolutions.3)We have built a numerical model for direct simulation of the interactions between air and water of micro-scale flows. The air/water interface is explicitly resolved by an interface capturing method we developed recently. The numerical model has been parallelized and tuned on Earth Simulator and other supercomputer platforms.4)The numerical formulations on unstructured meshes have been developed to deal with the geometrical complexity in topography. We obtained 4^<th>-order accuracy on triangular unstructured mesh.5)Toward the implementation in the spherical geometry and the global model, we have constructed several accurate and robust numerical formulations using three new-type global grids, i.e. the Yin-Yang grid, the gnomonic cubic grid and the icosahedral grid. The complete conservativeness of the Yin-Yang grid was obtained by including a flux modification across the overset boundaries. All prototype models on different spherical grids are verified with numerical benchmark tests and found promising.