8 June 2010 • It’s been a busy spring for community climate modeling at NCAR, with one major release accomplished—pushing a veteran model forward—and another on the way that brings a new paradigm into the mix.
Version 4 of the Community Climate System Model (CCSM4.0), which is supported primarily by NSF and the U.S. Department of Energy, became available to university and laboratory researchers on 1 April. Each of the major components of CCSM—modules for ocean, land, sea ice, and atmosphere, as well as the modeling infrastructure—has been significantly improved with many added capabilities, says NCAR’s James Hurrell, chair of the CCSM Scientific Steering Committee.
• an improved portrayal of deep convection (showers and thunderstorms);
• the ability to depict transient changes in land cover;
• a new scheme for tracking melt ponds atop sea ice; and
• a new scheme for parameterizing ocean overflows, which are density-based currents that hug the ocean bottom and flow over the continental slope.
CCSM4 also now includes a single foundation of code to support tasks that may use only a single processor (such as developing a new parameterization) or tens of thousands of processors (such as running simulations at very high resolutions). “We’ve taken a completely new approach with respect to the high-level design of the system,” says Mariana Vertenstein, head of the CCSM’s software engineering group at NCAR.
Moving to CESM
Next to be released is the successor to the CCSM: the first version of the Community Earth System Model (CESM1.0). In addition to the new physical climate system component models included in the CCSM4.0 release, CESM1.0 enables the carbon cycle in the land, ocean, and atmosphere to be fully predicted, rather than being partially specified as in previous versions of the model.
In addition to an interactive carbon cycle, CESM1.0 also includes
• an updated atmospheric chemistry component;
• an updated version of the Whole Atmosphere Community Climate Model (WACCM), which extends the atmospheric projections to altitudes as high as 500 kilometers (300 miles); and
• a new land ice component to study the role of the Greenland and Antarctic ice sheets in climate.
Both CCSM4.0 and CESM1.0 are already being used for an ambitious set of climate experiments to be featured in the fifth assessment (AR5) of the Intergovernmental Panel on Climate Change, whose reports are scheduled for release in 2013–14. Most of the simulations in support of the AR5 are scheduled to be completed and publicly released by the end of 2010, so that the broader research community can easily access them and complete analyses in time for inclusion in the assessment.
The two NCAR-based models will join others by the Climate Model Intercomparison Project (CMIP), part of the World Climate Research Programme. CMIP is an ambitious attempt to coordinate model runs for each IPCC assessment by stipulating a common suite of experiments conducted by various models, each with their own strengths and weaknesses.
The latest phase of CMIP adds several layers of complexity, including a new focus on predictions that range from 10 to 30 years out. These will benefit from the newly added ability of many models (including CESM1.0) to assimilate ocean temperatures, whose influence can extend years into the future. The first such experiment, conducted by the UK Met Office in 2007, correctly predicted little increase in global temperature through 2009 but calls for warming to resume over the next several years.
“Potential forecasting skill has been going untapped because we haven’t been assimilating the present state of the oceans,” says Hurrell. Still, he cautions, “Decadal prediction is a major research challenge, and there are many unanswered questions associated with it. This is just the beginning of research into a very important forecast problem.”
This graphic shows the improvement in the CCSM's depiction of the El Niño/Southern Oscillation within multicentury control simulations. When compared to the last 60 years of sea-surface temperatures for the Niño 3.4 region (top), CCSM4.0 (middle) captures the general behavior of ENSO, including its frequency and amplitude, more closely than does CCSM3.0 (bottom). (Graphic courtesy Adam Phillips, NCAR/CCSM.)