Three 20-day winter forecasts have been carried out using the Florida State University Global Spectral Model to examine the systematic errors of the model. Most general circulation models and global forecast models exhibit the same kind of error patterns even though the model formulations vary somewhat between them. Some of the dominant errors are a breakdown of the trade winds in the low latitudes, an overprediction of the subtropical jets accompanied by an upward and poleward shift of the jets, an error in the mean sea level pressure with overintensification of the quasi-stationary oceanic lows and continental highs, and a warming of the tropical mid- and upper troposphere. In this study, a number of sensitivity experiments have been performed for which orography and model physics are considered as possible causes of these errors.
A parameterization of the vertical distribution of momentum due to the subgrid scale orography has been implemented in the model to address the model deficiencies associated with orographic forcing. This scheme incorporates the effects of moisture on the wave induced stress. The parameterization of gravity wave drag is shown to substantially reduce the large-scale wind and height errors in regions of direct forcing and well downstream of the mountainous regions.
Also, a parameterization of the heat and moisture transport associated with shallow convection is found to have a positive impact on the errors, particularly in the tropics. This is accomplished by the increase of moisture supply from the subtropics into the deep tropics and a subsequent enhancement of the secondary circulations.