Impact of increasing CO2 on monthly-to-annual precipitation extremes: analysis of the CMIP2 experiments

Abstract

The impact of increasing atmospheric CO2 on high and low extremes of monthly-to-annual precipitation is studied using 20 model experiments participating in the second phase of the coupled model intercomparison project (CMIP2). In marked contrast with previous research on daily precipitation extremes, the simulated changes in extremes on these longer time scales are well correlated with the changes in the long-term mean precipitation: wet extremes become more severe especially where the mean precipitation increases, and dry extremes where the mean precipitation decreases. Changes in relative variability play a smaller but discernible role. In an ensemble-mean sense, the variability increases slightly in most areas, so that the contrast between the high and low precipitation extremes grows larger with increasing CO2. The changes in the frequency of extremes (fraction of cases with precipitation above a high or below a low predefined threshold) are much larger than the changes in their magnitude. Most of the ensemble-averaged changes in the frequency of extremes can be reconstructed by using the changes in time mean precipitation alone, provided that the variation in time mean precipitation change between different models is taken into account. The nonlinear relationship between the mean precipitation and the frequency of extremes complicates the interpretation of the frequency changes, especially when averaging frequencies over different models.