This simulation of moisture available for rain or snow (precipitable water, shown in millimeters) at 18 UTC on 27 September 2005 was produced by the Nested Regional Climate Model, or NRCM (see related article) at 36-kilometer resolution. The NRCM allows the short-term, regional Weather Research and Forecasting model to depict atmospheric features and interactions over larger areas of the globe and longer time spans. This simulation was driven by previously analyzed data on the north and south boundaries and by sea-surface temperatures. NCAR scientists are now analyzing 50-year NRCM predictions driven by the Community Climate System Model. (Image courtesy James Done.)
by Bob Henson
In this issure are eight snapshots of work in progress related to the Community Climate System Model (CCSM) and the Weather Research and Forecasting model (WRF). Much more information on the models can be found online (see “On the Web"). In future issues, we’ll be covering data assimilation, climate scenario development, and other model-related topics.
Weather prediction models and global climate models have largely followed separate paths over their decades of evolution—and with good reason. Many of the physics and dynamics components needed to produce realistic weather simulations would cause problems if they were included in a global model. But there’s been another key constraint: computer time and cost. It’s normally far too expensive to even try simulating decades or centuries of global climate using the much greater detail provided by a weather model’s higher resolution.
The move toward massively parallel computers is now changing the scene. In the years to come, supercomputers boasting as many as 100,000 processors could arrive at major modeling centers around the nation and the world. These machines pave the way for models that can tackle many small tasks at the same time, which is a key requirement for portraying weather-scale features across the entire globe.
But enhancing detail alone doesn’t fix some of the key physical and dynamical problems that can take years to resolve. Thus, while developers at NCAR and elsewhere explore long-range ideas for next-generation global modeling, they’re also forging ahead with a variety of more immediate improvements to the Community Climate System Model and the Weather Research and Forecasting model.
- CCSM—which brings together ocean, land, sea ice, and atmosphere components—was the largest single source of guidance for the 2007 assessment of the Intergovernmental Panel on Climate Change (IPCC). With important long-term support from NSF, the U.S. Department of Energy (DOE), and other partners, as well as its freely available source code and long history in academia, the CCSM is the global model of choice for university researchers.
- WRF—developed by an international consortium of agencies, universities, and commercial organizations, with strong NCAR leadership—has vaulted past its peers in the last decade. The NCAR-maintained Advanced Research WRF (ARW) now serves more than 10,000 users at weather services, campuses, labs, and private firms worldwide. The most recent upgrade to ARW, version 3.1, was released on 9 April.
Each of these two modeling systems has a growing family tree.
- Specialized versions of WRF have been tailored for chemistry, fire-weather, and hurricane applications. Sophisticated data assimilation systems, including ensemble and variational approaches (3D-Var and 4D-Var), allow model runs to begin with a broad set of current data. Early results from WRF 4D-Var appear in the January issue of Monthly Weather Review. In addition, several universities teamed with NASA to develop planetWRF, a global-scale version released in March that’s designed to simulate planetary atmospheres.
- CCSM includes four recently upgraded components (atmosphere, sea ice, land, and ocean) that can be run independently or together. The model also serves as the numerical backbone for the Whole Atmosphere Community Climate Model, which can simulate processes up through the stratosphere and beyond (see related article). The upcoming CCSM 4.0 will track carbon interactively as it passes between the atmosphere and other parts of the Earth system.
NCAR/ESSL director Guy Brasseur. (Photo by Carlye Calvin.)
Along with these improvements, scientists at NCAR and elsewhere are looking at how to draw on the strengths of both weather and climate models to advance global simulation. Guy Brasseur, head of NCAR’s Earth and Sun Systems Laboratory (ESSL), notes that one of NCAR’s original goals was to improve weather forecasting. Thanks to basic and applied research at NCAR and other institutions since the 1960s, day-to-day forecasts are now markedly better than even a few years ago.
“We still have to predict the weather, but people want information on longer time scales,” says Brasseur. “Now we need to expand that goal into a knowledge system that can help nations and regions assess their vulnerability to climate change and help define the next adaptation policies.”
NCAR’s science divisions have worked more closely than ever in recent years on modeling efforts that cross traditional lines. NCAR-wide groups that focus on chemistry and cloud physics have helped facilitate improvements in both CCSM and WRF. Through such efforts as the first ESSL-wide scientific retreat, held last fall, and a strategic plan for the lab now being developed, Brasseur hopes to enhance the cross-pollination. “It is clear that this new interdisciplinary approach and the participation of several divisions on Earth system modeling projects have been facilitated by the existence of ESSL,” he says.