Uncertainties in the Prediction of Regional Climate Change

Abstract

Uncertainties in regional climate change predictions for the 21st century by five coupled atmosphere-ocean General Circulation Models (AOGCMs) (two of them including ensembles of simulations), for different anthropogenic forcing scenarios and 23 regions in the World, are examined. The variables considered are seasonally and regionally averaged precipitation and surface air temperature for the future period of [2071–2100] as compared to the present day period of [1961–1990]. We find that the dominant source of uncertainty in the simulation of regional climatic changes by AOGCMs is due to inter-model variability with inter-scenario and internal model variability playing secondary roles. For both models including ensemble simulations, the spread of predicted average climatic changes by different realizations of the same ensemble is small. In addition, simulated regional climatic changes exhibit a high level of coherency among different forcing scenarios. Overall, uncertainties in predicted regional changes by the five AOGCMs are of the order of 3 K or greater for surface air temperature and 25% of present day values or greater for precipitation. These uncertainties would be transmitted to any regionalization technique used to enhance the regional information of AOGCMs. Differences between AOGCMs and their effects on the model simulations need to be better understood in order to increase confidence in the prediction of regional climate change due to anthropogenic forcings.