Scientists on two projects based at NCAR are among many worldwide working with both global and regional models, trying to get much-needed detail using the computers and code that are available now.
The multiagency North American Regional Climate Change Assessment Program (NARCCAP) is drawing on an international smorgasbord of four global and six regional models, including CCSM and WRF. The project’s goal is to study regional-scale uncertainty in climate projections for an area spanning most of North America (see graphic) and to generate climate change scenarios that can be applied to climate impacts research. The project PIs are drawn from universities and national laboratories across the United States, Canada, and the United Kingdom, with support from NSF, DOE, NOAA, and EPA.
In NARCCAP’s second phase, now under way, six regional climate models are driven by four global models for the periods 1971–2000 and 2040–2070 under the IPCC’s A2 emissions scenario. The regional models are run at higher resolution (50 km), thus filling in important details on precipitation and other variables. “This is the largest and most complex regional modeling activity ever created in the United States in support of climate assessment,” says project director Linda Mearns (NCAR). “The value of NARCCAP lies in using multiple global and regional models to more fully explore the uncertainties of regional climate change. The multimodel approach is crucial.”
Those interested in using NARCCAP data can sign up on the project’s website. The data should prove useful for climatologists and impacts researchers, as well as for regional modelers who want to downscale beyond the 50-km NARCCAP output. The U.S. Environmental Protection Program is using NARCCAP scenarios for its water quality program, and other U.S. agencies are expected to follow suit, says Mearns.
Climate projections from global models can shift significantly when the output is used to drive regional models, as shown in analysis from the North American Regional Climate Change Assessment Program. Shown in each plot below is the difference in degrees Celsius in summer average temperatures (June–August) projected for 2050–2070 minus those projected for 1980–1999. At left is the result from Environment Canada’s Coupled General Circulation Model, version 3. At right is the corresponding result—showing even hotter temperatures across the U.S. Midwest and northern plains—when the CGCM data are downscaled using the Canadian Regional Climate Model. (Images courtesy Seth McGinnis/NARCCAP.)
What about running a global and a regional model at the same time, in order to understand and improve the treatment of fine-scale processes? That’s the strategy behind the Nested Regional Climate Model, led by NCAR’s James Hurrell and Greg Holland. The NRCM attempts to seamlessly integrate WRF and the high-resolution Regional Ocean Modeling System, which originated at Rutgers University, within CCSM. A special issue of Climate Dynamics, due out in late 2009, will include 12 papers that elaborate on the strengths and weaknesses of initial simulations with the NRCM.
A special allocation from NCAR’s Computational and Information Systems Laboratory late last year allowed the NRCM team to simulate the period 1995–2005 (see sample graphic, page 1), as well as time slices extending out to 2055 (now in process). The western United States and the hurricane-generating Gulf of Mexico and northern Caribbean are being simulated at 2.5-km resolution.
Much of the NRCM’s current support is drawn from the insurance industry through the Willis Research Network, and from DOE and the offshore oil industry through the Research Partnership to Secure Energy for America. Both user groups are interested in how hurricane tracks and intensities might evolve in a changing climate.
“We think the NRCM offers a practical approach to high-resolution climate modeling. There is an urgent need to provide useful forecasts of changes in high-impact weather, and the NRCM can deliver on this requirement,” says Hurrell.