Mar 18, 2010
Assessing Latin America’s future climate, one terabyte a time
IDB-financed program enables Latin American scientists to use supercomputers to evaluate vulnerabilities to climate change
Inside a nondescript building on the campus of the University of Nebraska Lincoln (UNL) in the mid-western United States, some of the world’s fastest super-computers are grinding through data that could help to predict the future of corn harvests in Mexico and changes in rainfall patterns in Colombia.
The computers – actually a cluster of 1,151 powerful data processing “nodes” working in concert – are helping to answer questions submitted by Latin American climate scientists as part of an IDB-funded training project run by the U.S. National Center for Atmospheric Research (NCAR) and the Regional Climate Modeling Group at UNL.
The project is part of a broader initiative, financed with a $1 million IDB grant, that seeks to help train Latin American scientists in techniques for the analysis and visualization of climatic data generated by supercomputers, and to develop specific case studies that link these techniques to the planning and decision making process in Latin American governments.
The training project tackles two of the principal problems facing climate scientists and policymakers in Latin America: the lack of specific, country- and province-level information about the possible impact of climate change, and the scarcity of computing power necessary to generate such information.
“Our goal is to provide the region’s climate scientists with the training and technology they need to produce more-detailed projections of how climate change may affect specific regions in each country and by doing so identifying cost-effective adaptation options,” said Alfred Grünwaldt, an IDB climate change specialist who designed the program.
Last July a number of seven scientists and climatologists from Colombia, Honduras, Peru, Panama, Jamaica and Mexico participated in the program at NCAR facilities in Boulder, Colorado. Since then, scientists from NCAR and UNL have hosted additional training workshops in Mexico, Colombia and Peru for more than 30 Latin American climate change scientists and practitioners.
Participants received intensive training in the use of climate information and techniques for generating climate change scenarios with the Weather Research and Forecasting (WRF) Model, a sophisticated computerized climate program used by climatologists and weather services around the world.
The first group of scientists that received training back in July 09 was also offered an opportunity to submit a specific climate scenario for processing by UNL’s computers. Such scenarios can consume months of computer time, because they must calculate and project the effects of dozens of physical variables over decades. Though scientists from many disciplines compete for time to UNL’s computer, the IDB-financed program made it possible for the workshops to gain preferential access.
“It was clear from the workshops that the Latin American climate scientists are well aware of the potential environmental and societal problems they could encounter as a result of climate change,” said Bob Oglesby, a professor of climate modeling at UNL who led the training sessions. “What they need are estimates that are as precise and quantitative as possible of what the specific climate change in their region is likely to be, so that they can plan among various adaptation and/or mitigation strategies.”
In Jamaica, for example, policymakers are clamoring for information on how climate change may affect the frequency and intensity of hurricanes. “Access to tools like the WRF and to the computers at UNL is of tremendous importance to our research,” said Trevor Hall, a research fellow in the Department of Physics at the University of the West Indies in Jamaica who participated in one of the training workshops. “We plan to use the WRF as one of three models in a Regional Model Inter-comparison Project on the effects of climate change on hurricanes.”
Franklyn Ruiz, of the Colombian Government’s Institute of Hydrology, Meteorology and Environmental Studies (known as IDEAM), said he and his staff have had a copy of WRF software for two years, but they lack computing power necessary to use it effectively. “We have three data processors that take around four hours to produce a local forecast for the next 36 hours,” he said. Long-term climate modeling is virtually impossible on such equipment, according to Ruiz, so the chance to use the UNL computing cluster is invaluable to Colombia’s climate research efforts.
The first results of climate scenarios submitted by scientists from Colombia, Mexico, Jamaica and Peru are expected to be available by the end of 2010. The information generated by these climate scenarios is expected to be used by these countries in vulnerability and impact assessments. The IDB hopes to continue and expand the training program in the year head.