Climate modelers improve novel methods to corral uncertainty

August 1, 2016, Pacific Northwest National Laboratory
Climate modelers improve novel methods to corral uncertainty
An international workshop tackled the issue of uncertainty quantification in climate modeling. The workshop is another step to help scientists around the world understand climate system behavior and improve high-powered, next-generation Earth-system models. Credit: Pacific Northwest National Laboratory

Riffing on the theory that two heads are better than one when tackling a tough problem, it was all hands on deck for nearly 70 scientists and students from some 30 countries around the world attending a five-day workshop in Trieste, Italy. Gathered at the Abdus Salam International Centre for Theoretical Physics (ICTP), they got right down to sharing insights and strategies to quantify uncertainty in climate model projections. The outcome was ambitious but attainable: assess the reliability and quantify the uncertainty of climate change information for decision-making and map out the path toward next-generation climate modeling. The workshop was designed and organized by researcher Dr. Yun Qian at Pacific Northwest National Laboratory.

"Uncertainty quantification is a focus for the U.S. Department of Energy," said workshop lead and PNNL atmospheric scientist Dr. Yun Qian. "DOE has gathered eight national laboratories and six partner institutions to collaborate in developing a next generation and Earth-system model called the Accelerated Climate Modeling for Energy, or ACME. This workshop will help us address critical gaps in scientific computing and develop the resources needed to fill them."

Overall uncertainty in climate projections has not been significantly reduced since the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) issued in 2001. Because of the rapidly increasing complexity of Earth system models, the problem becomes more challenging every year. But the exponential growth in climate model complexity and the rapid build-up of wide-ranging data increase the imperative to understand the uncertainty boundaries. Uncertainties will always exist that may be beyond reach. The key is to develop strategies that can evaluate risks and use information in a way that informs vulnerability, impact, and adaptation issues.

The workshop aimed to provide participants information on strategies to quantify the in climate model projections and assess the reliability of climate change information for decision-making. The program included lectures on fundamental concepts in applications and Bayesian inference and sampling, and hands-on computer exercises on importance sampling, and global sensitivity analyses.

The lectures also covered a range of scientific issues underlying the evaluation of uncertainties in , such as the effects of uncertain initial and boundary conditions, uncertain physics, and limitations of observational records. Progress in quantifying uncertainties in hydrologic, land surface, and atmospheric models at both regional and global scales was reviewed.

Significant challenges still remain in applying the various UQ approaches to and their projections. One way forward could be to make use of information at shorter weather timescales and process levels.

Explore further: Gauging the impact of climate change on US agriculture

More information: Yun Qian et al. Uncertainty Quantification in Climate Modeling and Projection, Bulletin of the American Meteorological Society (2016). DOI: 10.1175/BAMS-D-15-00297.1

Related Stories

Gauging the impact of climate change on US agriculture

July 8, 2016

To assess the likely impact of climate change on U.S. agriculture, researchers typically run a combination of climate and crop models that project how yields of maize, wheat, and other key crops will change over time. But ...

Better soil data key for future food security

June 21, 2016

To project how much food can be produced in the future, researchers use agricultural models that estimate crop yield, or how much of a crop can be produced in a certain amount of space. These models take into account factors ...

Recommended for you

Machine learning-detected signal predicts time to earthquake

December 18, 2018

Machine-learning research published in two related papers today in Nature Geoscience reports the detection of seismic signals accurately predicting the Cascadia fault's slow slippage, a type of failure observed to precede ...

0 comments

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.