Accessible Climate Computing for 'Downstream' Science
2010-12 Seed Grant
The predictive understanding of climate change, and the detection and attribution of climate change to anthropogenic and natural components are among the leading scientific issues of our time. While the planetary-scale response of climate to anthropogenic forcing seems now to be confirmed, the cutting edge of research, policy/decision-making and public interest are now looking for an improved understanding and predicting capability of the regional scale signatures of climate change.
Computer simulation and data analysis are key pillars of climate science research. Scientists at NOAA’s Geophysical Fluid Dynamical Laboratory (GFDL) have developed among the best climate models in the world today and have been important contributors to international climate assessments, such as those issued by the Intergovernmental Panel on Climate Change (IPCC).
In this Siebel Energy Challenge project, Jaswinder Singh, professor of computer science, and Venkatramani Balaji, head of the modeling system group in atmospheric and oceanic sciences are looking to integrate the leading climate research being done at GFDL with the scalable computing and interdisciplinary research and education expertise on the Princeton campus, as well as to develop new collaborations with researchers at the Center for High Performance Computing in Cape Town, South Africa.
Professor Singh is working with colleagues to lead the development of a new course on parallel computing at a variety of scales, to be incorporated into the undergraduate curriculum of the Computer Science Department. The course will be targeted at qualified juniors and seniors, and open to approved graduate students as well. Students will study the fundamentals of parallel computing (from desktop scale to internet scale), learn how to design and implement parallel/distributed algorithms for the tasks at hand, learn now to use popular software tools and technologies for large-scale parallel computing and data analysis, and review how the characteristics of applications impact software/hardware system design and vice versa. The course will be application-driven, with climate modeling being a driving application area.