To ensure focused research and to measure progress, we need observable climate metrics that define the feedbacks sufficiently both to understand the key processes and to test and improve the simulation of these processes in climate models. A climate feedback is a set of numbers that can be derived from both observations and model output, and that characterizes the nature of a climate feedback process. It is important that this characterization be useful for better understanding the feedback process and for assessing the accuracy of its simulation in climate models. Metrics can use observed past climate trends, but should also use the variability of climate on other time scales that are better observed and where forcing is larger, such as seasonal and diurnal time scales. Good metrics must be focused on objectives that will increase confidence in our ability to usefully model climate feedback processes, and must be defined in terms of variables that are well observed. They should evolve as our understanding and observations improve.
Both global and regional metrics that focus on feedback processes responsible for climate sensitivity should be used to more rigorously test understanding of feedback processes and their simulation in climate models.
A good set of diagnostic tests for climate feedback processes should capture the covariation or coupling between the system's components. If effectively employed, these metrics can be an essential tool to help organize and stratify diagnostic analyses, as well as to relate model simulations to the fundamental aspects of observed phenomena. Successful reproduction of these observed metrics by climate models will not guarantee that climate models will give reliable projections of future climates, but testing climate models against a large set of carefully considered metrics is an effective way forward. They can also be a useful tool for observing the evolution of the climate system and thus make important contributions to the field of climate change detection and attribution. The set of metrics will evolve with time as understanding and simulation of the climate system evolve and improve.
A few examples of possible climate feedback metrics can be given. At the global or regional scale, the covariability of sea-surface temperature, clouds, upper-tropospheric water vapor, the vertical profile of atmospheric temperature, and other observations can be studied over a variety of time scales, including well-observed natural scales of variability, such as the diurnal, annual, and El Niño Southern Oscillation (ENSO) signals. These covariance metrics should then be applied to model simulations to pinpoint those aspects of the models that appear to represent nature accurately and those that require further work. A metric that might enable improvement of feedback processes over land would be observed diurnal and seasonal variations of temperature, clouds, precipitation, and soil moisture. Many other possible regional metrics for testing the simulation of climate system feedbacks can be envisioned, and some are discussed further in Chapters 2 through 8.
A step toward developing widely accepted metrics to evaluate feedback processes could be for the relevant agencies to organize a workshop or series of workshops to define a set of observational and diagnostic metrics that can be used to test understanding and modeling of climate feedback processes. These workshops could include scientists engaged in observation, diagnosis, and modeling of climate and climate processes.
Was this article helpful?