Process-Based Emergent Constraints on Global Physical and Biogeochemical Feedbacks

点击翻译按钮获取中文摘要

We use complex computer models of the climate system to make predictions of how climate will change in the future under scenarios of global warming. Despite the use of large supercomputers, we have to make approximations in building climate models so that we can produce century-long climate simulations. These approximations lead to 'errors' in models i.e. we know that climate models cannot reproduce exactly the observations we have made of past average climate and climate change. Climate model errors lead to uncertainties when we make predictions.We constantly strive to build new and better climate models in order to reduce uncertainty in predictions. We are guided by our past experience and intuition in targeting resources in specific areas of model improvement. But can we be more efficient? If we knew which of the many errors that exist in our ability to simulate present day climate and climate change are most important in determining the uncertainties in future predictions, could we better use our limited resources to improve models more quickly?In this proposal we aim to develop a generic technique to better target resources on improving climate models. The central concept, which we call a process-based emergent constraint, is to link errors in climate models to uncertainties in predictions. We will target three different areas; water vapour and lapse rate (the rate of change of atmospheric temperature with altitude), low-level clouds and processes that affect the rate of exchange of carbon in the land-surface and vegetation with the atmosphere. All these processes are know to have a leading-order influence on uncertainty in global temperature predictions and their regional consequences. We aim to exploit state-of-the-art model simulations and data from new observational platforms or data from existing observations that we will process in novel ways. There will be a focus on data from satellites. A key aim of our project is to identify processes that are responsible for both errors in models and uncertainties in predictions and to develop measures (metrics) that can be used to evaluate those processes in models.The project will provide multiple examples of how to target climate model improvement on processes that really matter for projections. The ultimate aim is to reduce uncertainty in predictions. However, in addition, the project will provide information on how to better target new observations of specific climate processes. We aim to make recommendations for future observational programmes (or the continuation of existing programmes) that will lead to more rapid progress in climate modeling and prediction.