A comprehensive GCM-based climate and carbon cycle model: Transient simulations to year 2100 using prescribed carbon emissions

We use an interactive global climate and carbon-cycle model to perform simulations of climate change from 1870 AD to 2100 AD forced by anthropogenic emissions of greenhouse gases. The goal of this effort is to produce a simulation tool capable of improved realism in future climate predictions. In particular, we seek to include and better understand feedbacks between the climate system and the carbon cycle. We use the NCAR Parallel Climate Model 2 (PCM-2) which includes a version of NCAR's CCM3 for the atmospheric GCM and a version of the POP model for the ocean GCM. The horizontal resolutions for atmosphere and ocean are T42 truncation (about 280 km) and 0.7 degrees (about 75 km), respectively. The ocean carbon model is based on OCMIP protocols, but modified to eliminate the phosphate-restoring restriction. The terrestrial biosphere model is IBIS-2 which simulates biophysical and biogeochemical surface fluxes and includes a dynamic model of natural vegetation. The model is integrated to a pre-industrial equilibrium using a prescribed atmospheric CO2 concentration of 290 ppmv. This equilibrated state serves as the starting point for transient simulations. The Control case assumes zero net anthropogenic carbon emissions and is used as a reference for model drift. The Full Interaction case assumes historical carbon emissions up to year 2000, then uses SRES A2 emissions up to year 2100. Simulated atmospheric carbon dioxide concentrations from this case closely track the observed concentrations during the twentieth century. A Carbon Cycle Only case is the same as the Full Interaction case, except that the greenhouse radiative forcing of climate is held fixed at preindustrial values. We will discuss the model's simulation of the twentyfirst century climate and atmospheric CO2 concentrations, the strength of carbon cycle feedbacks, and compare our simulation results to those of Cox et al. (2000) and Friedlingstein et al. (2001).