Five simple indices of large-scale surface temperature variations will be used to assess the performance of a number of climate models in simulating climate variability and change during the 20th century. These indices are based on the spatial fingerprint of greenhouse climate change and include the area-mean surface temperature, the land-ocean temperature contrast, the meridional temperature gradient in the Northern Hemisphere, the magnitude of the annual cycle in temperature over land, and the magnitude of the diurnal temperature range over land.
The indices will be calculated for the global domain and for the North American region from observational data for the period 1880-2003 and compared with data from model simulations for this period. The models to be analyzed include simulations from all three U.S. models (Community Climate System Model, at the National Center for Atmospheric Research at Boulder, Colorado; Geophysical Fluid Dynamics Laboratory model (GFDL) at Princeton; and Goddard Institute for Space Studies model (GISS) at New York City) being made available for the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report, together with several other international models. The variability and correlation structure of the indices will be compared between the observations and the models on three different timescales: interannual variability, decadal variability and trends on 50 year and century timescales.
This evaluation will provide greater confidence in using these models for climate change detection and attribution studies and for projection of future climate change.
