Climate scientists put multiple models to the test
Researchers with Oak Ridge National Laboratory’s (ORNL’s) Climate Change Science Institute (CCSI) are part of a multipartner team that is evaluating how the structures of terrestrial biosphere models (TBMs) influence the results of model simulations. TBMs are just one component of global climate models and are also used independently to study carbon exchange between the land and atmosphere in the form of carbon dioxide, methane, and other carbon compounds.
Predicting the amount of carbon that will be released into the atmosphere and under what circumstances is complicated by the variability between TBMs, which have been developed over time by multiple research groups across the globe with different project goals and resources at their disposal. By conducting ensemble runs that compare a host of models, the team hopes to inform researchers how they can account for differences and improve the predictive accuracy of TBMs, and by extension, global climate models.
Over the last 3 years, the Multi-scale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP) team, led by Debbie Huntzinger of Northern Arizona University and Anna Michalak of Carnegie Institution for Science, has compared 20 models by simulating terrestrial ecosystem processes that influence carbon flux for the years 1900–2010. Identical environmental and weather drivers, including prescribed inputs such as rainfall, temperature, soils, land cover, and vegetation, are run through each TBM’s unique suite of modeling approaches that then calculates processes such as net ecosystem carbon exchange and nutrient uptake based on these drivers.
“Each model represents ecosystem processes differently, so research teams are viewing the carbon cycle through slightly different interpretations,” said CCSI’s Robert Cook, MsTMIP coinvestigator.
The team determines which modeling approaches lead to the same conclusion and which diverge despite identical initial conditions. By further comparing results to observed conditions in the twentieth century, researchers can have confidence in model output.
“This was the first step in the project,” Cook said. “With an additional 3 years of NASA funding, we’ll develop another set of driver data out to the year 2100 and allow each model to simulate that period.”
An early paper examining results from MsTMIP data determined models generally agree that hot and dry periods lead to larger carbon releases, while cold and wet periods lead to small carbon uptakes in soil and vegetation. Jakob Zscheischler of the Max Planck Institute for Biogeochemistry is lead author on the paper, which appeared in Global Biogeochemical Cycles in June. CCSI’s Cook, Dan Ricciuto, Yaxing Wei, Xiaoying Shi, Jiafu Mao, and Anthony King are coauthors.
ORNL is playing a big role in MsTMIP data management by archiving the driver and output data in ORNL’s Distributed Active Archive Center.
“We want other researchers to be able to use the data like Jakob did by accessing the results from all 20 models without having to replicate simulations or convert the output into a common format for analysis,” Cook said. “From there they can go in and look at what they’re interested in—maybe snow and ice cover on land, for example—and compare the influence of snow and ice cover on carbon fluxes between models for a more complete analysis of model behavior.”
Cook said several other papers defining the progress of the MsTMIP project are under way.
The Department of Energy’s Office of Science Biological and Environmental Research Program is funding ORNL’s contribution to model simulation, while the MsTMIP project and data management are funded by the National Aeronautics and Space Administration.
D.N. Huntzinger et al. “The North American Carbon Program Multi-Scale Synthesis and Terrestrial Model Intercomparison Project – Part 1: Overview and experimental design,” Geoscientific Model Development 6 (2013): 2121–2133. doi: 10.5194/gmd-6-2121-2013.
Y. Wei et al. “The North American Carbon Program Multi-Scale Synthesis and Terrestrial Model Intercomparison Project – Part 2: Environmental driver data,” Geoscientific Model Development 6 (2013): 5375–5422. doi: 10.5194/gmdd-6-5375-2013.