Estimation of Model Error for the Global GRAPES Model

The model error exists for the reason that a certain difference is present between the weather state described by the model and the true weather state though more and more physical processes and complicated boundary conditions are considered in numerical weather prediction models. Filter divergence was found in the ensemble Kalman filter system when the model errors were neglected in previous studies. In this paper, it is shown that the model errors increase linearly with the decrease of resolution and the increase of the model error is obvious with the increase of model forecasting length through estimating the discriminations among the forecasting fields at different resolutions.