The Application of Multiple Grid Method in A Long-range Numerical Weather Prediction Model
多网格法在长期数值天气预报模式中的应

In this paper, the basic idea of Multiple Grid method (MG) and the improved nonsteady atmosphere-earth surface coupled anomaly model are introduced, then MG is used to solve the geopotential tendency equation (an ellipse equation) in the model. It is shown, through the seasonal prediction experiment of circulations in the Northern Hemisphere from June to August in years from 1982 to 1989, that the iteration precision is increased when using MG. The defects that anomaly field is too smooth and amplitude is too small are overcomed. Large-scale low-frequency signal in the model, which plays a key role in the long-range weather prediction, is solved. Comparing experiments in high and low precisions proves that MG is better than the step-by-step over relax method increasing the computing speed more than one time. MG is more efficient in high precision than in low one.