The major interest of research in the Goddard Institute for Space Studies (GISS) is long term climate change. One of the tools to study this change is through the use of GCMs. Therefore a great deal of effort has been spent at GISS to improve its GCM. In order to evaluate how well the new dynamics and physical parameterization schemes work, we need to compare GCM results with observations. GCM runs using the AMIP SST as forcing are convenient ways to evaluate the performance of the new schemes. A 10-year GCM run not only can help us to evaluate how well the GCM behaves in the mean state, but can also tell us how well the GCM captures intra- and inter-annual variabilities.
This study involves comparing the results of the GISS Model II GCM and the GCM results submitted to AMIP. The GCM result submitted to AMIP is from one of the intermediate versions of the new GISS GCM, which is still under development. This version has a quadratic upstream scheme for the advection of temperature and humidity, 8th order Shapiro filter, new physical parameterization schemes of convection, large scale clouds, planetary boundary layer and ground hydrology. The radiation has dust and anthropogenic aerosol included. Global mean states, intra- and inter-annual variabilities of the model results are studied. The meteorological parameters studied include surface air temperature, precipitation, evaporation, surface winds, 500 hPa height, 850 hPa zonal wind, outgoing long wave radiation and incident solar radiation at the Earth's surface.
The global mean states of both the old model and the new model agree with the observational values. This is particularly gratifying because many of the new schemes are based on physical principles and rely less on tuning. Geographical distributions of precipitation, outgoing long wave radiation, incident solar radiation and surface winds show significant improvement of the new model from the old. The surface air temperature and evaporation also show acceptable results except over mountains and arid areas. Seasonal variability of the zonal mean of different parameters indicates considerable improvement in the seasonal distribution of water vapor and cloud, and the movement of the ITCZ. For interannual variability, Hovmöller diagrams of OLR anomaly, 850 hPa zonal wind anomaly and precipitation anomaly show that the new model captures the 1983 and 1987 El Niño events very well while the old model fails to show strong signals for the two El Niños.
Tentatively, we conclude that the new convection, large-scale cloud and planetary boundary layer parameterization schemes show noteworthy improvement. The effects of the other new schemes require more tests.