Researchers at institutions including the Department of Energy’s Oak Ridge National Laboratory have developed a new method for statistical analysis of climate models that more accurately predicts future conditions. This method provides a method of adjustment for models with high temperature sensitivity, which is a known problem in the community. By assigning different weights to the models and combining them, the researchers estimate that global temperatures will increase by 2-5°C by the end of the century.
This projection, published in the journal Nature Communications Earth & Environment, is consistent with previous estimates; however, this new framework is more inclusive and avoids the abandonment of models that was common practice in previous methods.
“We don’t evaluate the models individually,” said Elias Massoud, a computational ecohydrologist at ORNL. “Instead, we see how they can be combined, using aggregated information to make predictions about the future.”
A key parameter for these models, known as equilibrium climate sensitivity, or ECS, describes the relationship between changes in carbon dioxide and corresponding warming. Although the Earth system has a real ECS, it is not a measurable quantity. Different evidence can provide a reasonable picture of the Earth’s true ECS, which can reduce uncertainty in simulation models.
However, many models assume a high ECS and predict higher temperatures in response to the amount of carbon dioxide in the atmosphere being greater than that occurring in the real Earth system. Because these models provide scientists and policymakers with predictions of future conditions, it is important to ensure that they reflect Earth conditions as accurately as possible.
Previous methods alleviated this problem by excluding models with high ECS values. “It was a difficult approach,” Masoud said. “The discarded models may have the good information we need to understand particularly extreme targets.”
“Instead, we applied a tool called Bayesian Model Averaging, which is a way to combine models with different effects to estimate their distributions,” Massoud said. “We used this to constrain the ECS in these models, allowing us to predict future conditions without the ‘hot model problem’.
The fact that many models come from similar code or have the same parameters raises concerns about model independence. “If two models are dependent, they give the same information,” Masoud said. “In our study, we use the results of the weight coefficients to estimate the degree of independence of each and then take into account the impact of the same information so that it is not counted twice.”
This new method lays the foundation for a better understanding of a range of climate models. Model weights included in this study informed the Fifth National Climate Assessment, a report released Nov. 14 that assesses the impacts of climate change in the United States. This project also supports the Earth System Grid Federation, an international collaboration led by the U.S. Department of Energy that manages and provides access to climate models and observations.
“Our study combines model data with observational data to obtain the best estimate of the state of the Earth system,” Masoud said. said. “This allows scientists to make more accurate predictions about how the Earth and climate are changing.”
Source: Port Altele
