Experiments of the 30–Day Long–range Numerical Weather Prediction in a Seven–level Spectral Model

The present model with the considerations of more physical processes is largely is largely based on the northern hemispheric seven-level spectral model. An efficient method in the model has been given to overcome the problems involving small differences between the two large magnitudes occurring in the divergence equation, as well as in the calculation of the pressure gradient force terms in a σ-coordinate in the vicinity of mountains. The method of calculating nonlinear terms is excellent. The 30-day long-range numerical weather predictions utilizing the present model with initial conditions based on real data on lst May and 7th July 1982 have been made, and the results are satisfactory. The RMS errors of the forecasts are consistently less than those of persistence method even in lower latitudes during the entire 30–day forecasting period.