Non-linear Normal Mode Initialization and Its Effect on Data Assimilation and Forecasting

The non-linear normal mode initialization (NNMI) for a spectral model of mediumrange NWP in NMC, SMA is described, and its effect on data assimilation and forecasting is discussed. It is experimentally shown that in terms of two iterations by using first five vertical modes, a balanced initial field appropriate to data analysis is obtained. It also shows that NNMI is highly successful in eliminating the spurious high frequency gravity wave oscillation in forecasting, therefore, it plays an important role in improving data assimilation and model forecasting.