Differences among the model simulations of the last millennium are related to several factors. The volcanic and solar forcing reconstructions used by the models differ as do their geographic and seasonal implementation. Most climate model simulations published to date used one of the earlier reconstructions of solar irradiance that included an increase in solar irradiance from the Maunder Minimum to present of around 0.2–0.4 percent. None of the long transient simulations have included wave-length-dependent changes in solar irradiance, although this effect has been investigated with shorter sensitivity simulations and has been shown to impact regional surface climate (Shindell et al. 2001). Volcanic reconstructions used in the various climate model simulations show similar timing of major volcanic eruptions and temporal clusters of eruptions. The models differ in the conversion of volcanic sulfate loading into optical depth and in the seasonally dependent horizontal and vertical dispersion of the aerosol cloud. The models also vary with respect to the specifics of the radiative transfer calculations included for volcanic aerosols. For example, in some energy balance models and intermediate complexity models, volcanoes are represented as a negative deviation of the solar irradiance.
Differences between the various model simulations can also be related to differences in the sensitivity of the models (Goosse et al. 2005). Two of the models, CSM and GKSS, are full three-dimensional climate models and have equilibrium climate sensitivities of 2°C and 3.2°C, respectively, for doubling of atmospheric CO2. Some simpler models have an adjustable climate sensitivity (i.e., the Gerber 1.5 × CO2 and 2.5 × CO2 simulations). Different models also exhibit different sensitivities to different external forcings. The full three-dimensional climate models, which include parameterizations for the size distribution and transport of volcanic aerosols, show