Physical and observational constraints on 21st-century global warming

Physical understanding and simulations by computer climate models indicate that global-average temperature will rise in coming decades, due to human greenhouse-gas emissions. Regional climate changes are what people experience, for example in temperature and rainfall; impacts are caused especially by extremes such as heatwaves. Most changes in regional climate and extremes are expected to get larger as global-average warming increases. Given future greenhouse-gas emissions (which depend on societal choices), climate models give a range of projections of global-average temperature rise. For example, for the low-emissions scenario in the most recent assessment of the Intergovernmental Panel on Climate Change (IPCC), the range of projected global warming is 1.3–2.4degC by 2100, encompassing the Paris targets of 1.5 and 2.0degC. This large uncertainty in projections has hardly diminished over the last 30 years, despite huge improvements to the models' simulations of the present-day average climate. The uncertainty remains a serious obstacle for climate policy, both for planning adaptations to the climate consequences of a particular future scenario, and for working out how rapidly emissions must be reduced to keep below a particular temperature. The largest contribution to the uncertainty is due to clouds, which affect the climate both by reflecting sunlight and by inhibiting heat loss from the Earth to space, thus opposing or amplifying global warming. The net result involves diverse changes in clouds in different regions. It depends on cloud physical processes, which differ among models and from place to place, and on the geographical pattern of surface warming. The pattern of surface warming in turn depends on where the atmosphere adds heat to the sea surface, and on the ocean circulation, which moves heat around and transfers the added heat to the deep ocean. Through their influence on the pattern of surface warming, both ocean and atmosphere processes affect changes in cloud. Constraints on model projections have been developed that use observations of climate change over the last 150 years. But such constraints are an unreliable guide to the future, because they focus on global-average temperature change, disregarding its geographical pattern, which has been quite different in recent decades from what is projected in response to greenhouse gases. While the pattern has a strong effect on relationships between global-average quantities, the physical principles governing local processes are invariant, regardless of global change and its pattern. Therefore we will attack the problem of uncertainty in a different way. Using newly available observational datasets and simulations, and new methodologies that we have developed, we will analyse the differences of models from reality and among themselves, regarding the physical processes in atmosphere and ocean that have determined local changes in cloud and the geographical pattern of change in surface temperature in the recent past. We aim to reduce the uncertainty of projections for the 21st century, by giving greater weight to existing results from models whose simulated processes are more similar to the observed processes, and by evaluating the effect of making the model processes agree better with the observed processes. On the basis of our results, we will recommend areas for improvement of models, having the potential to make them more realistic and precise. The new knowledge arising from this novel approach could have a transformational effect on the usefulness to society of climate change projections.