Abstract
The sensitivity of a low resolution, spectral general circulation model (GCM) to specification of physical processes is examined using a new version of the model with refined parameterizations. Specific refinements in parameterization include: 1) smoothing the original orography to greatly diminish undesirable topographic “ripples” occurring near high mountain ranges; 2) adding snowcover on the Tibetan plateau and representing winter snowcover in middle latitudes more realistically; 3) decreasing the land ground wetness and adjusting the drag coefficient and parameters governing condensation-moist convective adjustment.
Results of comparative 5-year integrations show that better parameterization in the low resolution model produces significant improvement in simulation without resorting to the use of higher horizontal or vertical resolution. The combined changes in ground wetness, drag coefficient and condensation-moist convective parameters produce more realistic zonal banding of precipitation belts and a better representation of continental precipitation relative to the ocean. In addition, mass is more nearly conserved and mean sea level pressure and temperature patterns are in better agreement with observations than in the previous model. Major deficiencies in simulation that are not improved include zonal jet and stratospheric temperature structures. Overall, the improvements in simulation suggest a wider applicability of the low resolution model for use in climate sensitivity studies.
Analysis of sensitivity experiments assessing specific effects of parameterization indicate that decreased ground