Abstract
In the context of operational numerical weather prediction, using actual data as the initial conditions, aliased spectral models are compared with dealiased spectral models (operational European Centre for Medium-Range Weather Forecasts spectral models). For the same number of grid points, forecasts of aliased spectral models are more accurate than those of dealiased spectral models. For the same maximum truncated wavenumber, the computational requirements of aliased spectral models are only 58% of those of dealiased spectral models, while the forecasts of both models are generally very similar to each other. The computational nonlinear instability has not yet been encountered in the integrations of aliased spectral models in operational numerical weather prediction.
Corresponding author address: Xiong-Shan Chen, Institute of Atmospheric Physics, Chinese Academy of Sciences, P.O. Box 2718, Beijing 100080, People’s Republic of China.
