CISL hosts Computing in Atmospheric Sciences workshop

Computer experts, scientists, and industry leaders gathered in Annecy, France September 9-13, 2007 to discuss the status and future of high-performance computing for weather prediction and climate modeling. The four-day meeting drew 80 participants and covered topics including hurricane and tsunami prediction, weather prediction, climate change, and challenges and advances in computing technologies. Practical applications extend as far as accurate weather forecasting for the 2008 Olympic Games in Beijing, China. The conference was the eighth biennial session of Computing in Atmospheric Sciences (CAS), a National Center for Atmospheric Research (NCAR) workshop that provides the opportunity for supercomputing industry leaders to hear about their customers' needs and present their product roadmaps, as well as for scientists to exchange ideas and share experiences about computing resources and applications. "We are in grave danger of generating more data than we can ever hope to look at," says Richard Loft of NCAR's Computational and Information Systems Laboratory (CISL). "Processing capability, power requirements, and mass storage all need to keep pace with the ever-increasing volumes of data." The Imperial Palace Hotel in Annecy, France was the venue for CAS2K7. (Photo from Imperial Palace Hotel.)
As computing power advances, so does energy consumption. According to Stefan Heinzel of the Deutsches Klimarechenzentrum (DKRZ) in Hamburg, Germany, electricity to power one sustained Petaflops (1,000 trillion operations per second) could cost up to 15 million Euros per year. John Dennis, a scientist at NCAR working on the Community Climate System Model (CCSM), described collaborative efforts with the U.S. Department of Energy to address the issue of petascale application readiness. By enhancing the model's scalability, he and his colleagues demonstrated the feasibility of using IBM's power-efficient Blue Gene architecture to perform large-scale climate simulations. The positive contributions of high-performance computing also featured prominently in the sessions. A key theme at this year's conference was how high-performance computing can benefit society. Cherri Pancake, a researcher at Oregon State University and one of four keynote speakers, discussed the importance of IT in tsunami modeling and prediction and pointed out that continued improvements in access to historical data, to tsunami warning procedures, and to tsunami models can help save lives. Greg Holland, a scientist at NCAR who works on nested regional climate models, gave a presentation on hurricanes and climate change, noting that "there is an anthropogenic effect on hurricanes. To understand the consequences, we need to keep refining the models so that we can move from hurricane projection to true hurricane prediction." In describing the Australian Community Climate and Earth System Simulator (ACCESS), Kamal Puri from the Bureau of Meteorology Research Centre in Melbourne, Australia discussed the challenges and opportunities of building a next-generation Earth system modeling environment. Many countries face similar demands to provide government leaders as well as general populations with quality environmental services to effectively plan for, and minimize the impact of, future weather events or changes in climate. One challenge is to minimize duplication of efforts across international boundaries and build collaborations among government research laboratories and university research communities. Tom Bettge, from CISL and the Program Chairman of this year's CAS workshop (CAS2K7), commented that "CAS2K7 provides an exciting opportunity for climate scientists, computer scientists, and the supercomputing industry to come from around the world to discuss challenges and envision solutions. In addition to the traditional discussions of high performance computing capacity and capability, the conference allowed exposure to innovative methods of enhancing data distribution mechanisms, constructing data preservation archives, and improving the efficiency and effectiveness of data storage and access. Most importantly, we keep the big picture in mind: this is about advancing science." Industry representatives at the conference included AMD, Cray, IBM, Intel, NEC, SGI, and Sun Microsystems, all of whom provided corporate sponsorship to partially support the meeting. Details about the content of the conference appear in the CAS2K7 Agenda.