Explore the vast range of titles available.

In addition to influencing climatic conditions directly through radiative forcing, increasing carbon dioxide concentration influences the climate system through its effects on plant physiology. Plant stomata generally open less widely under increased carbon dioxide concentration, which reduces transpiration and thus leaves more water at the land surface. This driver of change in the climate system, which we term 'physiological forcing', has been detected in observational records of increasing average continental runoff over the twentieth century. Here we use an ensemble of experiments with a global climate model that includes a vegetation component to assess the contribution of physiological forcing to future changes in continental runoff, in the context of uncertainties in future precipitation. We find that the physiological effect of doubled carbon dioxide concentrations on plant transpiration increases simulated global mean runoff by 6 per cent relative to pre-industrial levels; an increase that is comparable to that simulated in response to radiatively forced climate change (11± 6 per cent). Assessments of the effect of increasing carbon dioxide concentrations on the hydrological cycle that only consider radiative forcing will therefore tend to underestimate future increases in runoff and overestimate decreases. This suggests that freshwater resources may be less limited than previously assumed under scenarios of future global warming, although there is still an increased risk of drought. Moreover, our results highlight that the practice of assessing the climate-forcing potential of all greenhouse gases in terms of their radiative forcing potential relative to carbon dioxide does not accurately reflect the relative effects of different greenhouse gases on freshwater resources.

A methodology is described for probabilistic predictions of future climate. This is based on a set of ensemble simulations of equilibrium and time-dependent changes, carried out by perturbing poorly constrained parameters controlling key physical and biogeochemical processes in the HadCM3 coupled ocean-atmosphere global climate model. These (ongoing) experiments allow quantification of the effects of earth system modelling uncertainties and internal climate variability on feedbacks likely to exert a significant influence on twenty-first century climate at large regional scales. A further ensemble of regional climate simulations at 25 km resolution is being produced for Europe, allowing the specification of probabilistic predictions at spatial scales required for studies of climate impacts. The ensemble simulations are processed using a set of statistical procedures, the centrepiece of which is a Bayesian statistical framework designed for use with complex but imperfect models. This supports the generation of probabilities constrained by a wide range of observational metrics, and also by expert-specified prior distributions defining the model parameter space. The Bayesian framework also accounts for additional uncertainty introduced by structural modelling errors, which are estimated using our ensembles to predict the results of alternative climate models containing different structural assumptions. This facilitates the generation of probabilistic predictions combining information from perturbed physics and multi-model ensemble simulations. The methodology makes extensive use of emulation and scaling techniques trained on climate model results. These are used to sample the equilibrium response to doubled carbon dioxide at any required point in the parameter space of surface and atmospheric processes, to sample time-dependent changes by combining this information with ensembles sampling uncertainties in the transient response of a wider set of earth system processes, and to sample changes at local scales. The methodology is necessarily dependent on a number of expert choices, which are highlighted throughout the paper.

Global and local feedback analysis techniques have been applied to two ensembles of mixed layer equilibrium C[O.sub.2] doubling climate change experiments, from the CFMIP (Cloud Feedback Model Intercomparison Project) and QUMP (Quantifying Uncertainty in Model Predictions) projects. Neither of these new ensembles shows evidence of a statistically significant change in the ensemble mean or variance in global mean climate sensitivity when compared with the results from the mixed layer models quoted in the Third Assessment Report of the IPCC. Global mean feedback analysis of these two ensembles confirms the large contribution made by inter-model differences in cloud feedbacks to those in climate sensitivity in earlier studies; net cloud feedbacks are responsible for 66% of the inter-model variance in the total feedback in the CFMIP ensemble and 85% in the QUMP ensemble. The ensemble mean global feedback components are all statistically indistinguishable between the two ensembles, except for the clear-sky shortwave feedback which is stronger in the CFMIP ensemble. While ensemble variances of the shortwave cloud feedback and both clear-sky feedback terms are larger in CFMIP, there is considerable overlap in the cloud feedback ranges; QUMP spans 80% or more of the CFMIP ranges in longwave and shortwave cloud feedback. We introduce a local cloud feedback classification system which distinguishes different types of cloud feedbacks on the basis of the relative strengths of their longwave and shortwave components, and interpret these in terms of responses of different cloud types diagnosed by the International Satellite Cloud Climatology Project simulator. In the CFMIP ensemble, areas where low-top cloud changes constitute the largest cloud response are responsible for 59% of the contribution from cloud feedback to the variance in the total feedback. A similar figure is found for the QUMP ensemble. Areas of positive low cloud feedback (associated with reductions in low level cloud amount) contribute most to this figure in the CFMIP ensemble, while areas of negative cloud feedback (associated with increases in low level cloud amount and optical thickness) contribute most in QUMP. Classes associated with high-top cloud feedbacks are responsible for 33 and 20% of the cloud feedback contribution in CFMIP and QUMP, respectively, while classes where no particular cloud type stands out are responsible for 8 and 21%.