A 10-year integration of the Navy Operational Global Atmospheric Prediction System (NOGAPS) has been performed as part of the Atmospheric Model Intercomparison Project. The intraseasonal variability in the model's tropical heating and 200-mb winds are assessed through a comparison with daily averaged ECMWF analyses and MSU derived oceanic precipitation estimates. The relationships between the primary modes of 20-70 day variability in the Northern Hemisphere winds and tropical heating are established through empirical orthogonal function (EOF) and composite analyses. The results indicate that deficiencies in the simulated variability of tropical heating may have a significant impact on the simulation of extratropical wind field variability.
The intraseasonal variability of precipitation in NOGAPS is much weaker than the observed variability over most of the globe, with the exception of isolated regions around Central America and Madagascar. This suggests that the model simulates the Madden-Jullian Oscillation (MJO) poorly. Hovmöller diagrams of 200 mb velocity potential along the equator show a westward propagating wave number one signal in the simulation, however, it is much weaker and less coherent than in the observations. The simulated MJO also has a maximum amplitude in the Western Hemisphere, while the observed MJO has a maximum amplitude in the Eastern Hemisphere.
The variability of the model's 200 mb winds is too weak in the tropics and subtropics of the Indian ocean Pacific regions, while it is adequate or too strong at higher latitudes and in the Atlantic region. EOF and composite analyses indicate that the model's weak representation of the MJO may be responsible for the deficient subtropical variability in the model's upper-tropospheric wind fields. Observed modes of intraseasonal variability that have a clear connection to tropical precipitation are poorly simulated by NOGAPS, and in regions where these observed modes dominate, such as the North Pacific, the simulated variability is severely underestimated. Observed modes of intraseasonal variability that do not show any significant relationship to tropical convection are well simulated, especially in regions where these extratropically forced modes dominate, such as the North Atlantic.