Anthropogenic climate change is usually perceived as a comparatively slow trend towards higher temperatures, while associated changes in interannual variability are often considered small. In a recent paper, we have proposed that land-surface processes may give rise to a pronounced increase in interannual variability during the summer season. Such an increase in variability would have important repercussions for the frequency of extreme summer heat waves, and would also represent a challenge to adaptive response strategies designed to cope with climate change. In the current study, we present detailed process studies and an intercomparison of global and regional climate models.

We simulate future climates in scenarios with increasing atmospheric greenhouse-gas concentrations, using output from high-resolution GCM simulations and spatially refined simulations using a regional climate model over Europe. Analysis is presented for Europe and North America. Simulations suggest that interannual summer temperature variability may increase by up to a factor two within the current century. Analysis of the simulated water cycle suggests that the increase in variability is partly related to land-surface processes, and associated with an increasing frequency of summer droughts. The representation of the associated non-linearity requires a realistic simulation of the continental-scale seasonal cycle of the terrestrial water storage. The realism of control simulations to capture this aspect of the climate is assessed using diagnostic estimates of the continental-scale water balance.

Intercomparison of model simulations suggests that there is considerable agreement between different models regarding the occurrence of the effect, but there are substantial differences regarding its seasonal evolution as well as geographical distribution and amplitude. Analysis suggests that these differences related to a wide range of model differences, among them the representation of the soil hydrology and the simulation of large-scale circulation anomalies.