BER launches Environmental System Science Program. Visit our new website under construction!

U.S. Department of Energy Office of Biological and Environmental Research

BER Research Highlights

Exploring an Ensemble-Based Approach to Atmospheric Climate Modeling and Testing at Scale
Published: June 09, 2017
Posted: January 26, 2018

Evaluate a methodology to test Earth system models quickly on large scale computing systems.

The Science
In this paper, the utility of ensembles-based testing framework is demonstrated; it was shown that running many short simulations is more efficient than running a single long simulation. It was also demonstrated that aggressive optimizations can lead to simulations with a statistically distinct model state. Also, averaged behavior of many single year runs of the atmosphere are statistically different than one long run, demonstrating atmosphere low-frequency variability.

The Impact
Ensembles-based testing provides faster verification to model developers when developing at scale. With this testing, developers can now more effectively use multicore computing systems when performing development model simulations.

A strict throughput requirement has placed a cap on the degree to which we can depend on the execution of single, long, fine spatial grid simulations to explore global atmospheric climate behavior. Whereas, running an ensemble of short simulations is computationally more efficient. We test the null hypothesis that the climate statistics of a full-complexity atmospheric model derived from an ensemble of independent short simulation is equivalent to that from an equilibrated long simulation. The climate of short simulation ensembles is statistically distinguishable from that of a long simulation in terms of the distribution of global annual means, largely due to the presence of low-frequency atmospheric intrinsic variability in the long simulation. We also find that model climate statistics of the simulation ensemble are sensitive to the choice of compiler optimizations. While some answer-changing optimization choices do not affect the climate state in terms of mean, variability and extremes, aggressive optimizations can result in significantly different climate states.

Contacts (BER PM)
Dorothy Koch
Earth System Modeling Program

(PI Contact)
Salil Mahajan
Oak Ridge National Laboratory

The U.S. Department of Energy Office of Science, Biological and Environmental Research supported this research as part of the Accelerated Climate Modeling for Energy (ACME) project of the Earth System Modeling (ESM) program.

Mahajan S., A. L. Gaddis, K. J. Evans, and M. R. Norman. “Exploring an ensemble-based approach to atmospheric climate modeling and testing at scale.” Procedia Computer Science, 108C, 735-744, (2017). [DOI:10.1016/j.procs.2017.05.259]
(Reference link)

Related Links
BER Highlight: Exploring an Ensemble-Based Approach to Atmospheric Climate Modeling and Testing at Scale

Topic Areas:

  • Research Area: Earth and Environmental Systems Modeling

Division: SC-33.1 Earth and Environmental Sciences Division, BER


BER supports basic research and scientific user facilities to advance DOE missions in energy and environment. More about BER

Recent Highlights

Mar 23, 2021
Molecular Connections from Plants to Fungi to Ants
Lipids transfer energy and serve as an inter-kingdom communication tool in leaf-cutter ants&rsqu [more...]

Mar 19, 2021
Microbes Use Ancient Metabolism to Cycle Phosphorus
Microbial cycling of phosphorus through reduction-oxidation reactions is older and more widespre [more...]

Feb 22, 2021
Warming Soil Means Stronger Microbe Networks
Soil warming leads to more complex, larger, and more connected networks of microbes in those soi [more...]

Jan 27, 2021
Labeling the Thale Cress Metabolites
New data pipeline identifies metabolites following heavy isotope labeling.

Analysis [more...]

Aug 31, 2020
Novel Bacterial Clade Reveals Origin of Form I Rubisco

  • All plant biomass is sourced from the carbon-fixing enzyme Rub [more...]

List all highlights (possible long download time)