Tony Fountain and Lixin (pronounced ‘Lee-shin’) Lu have recently joined the ranks -- remotely -- of the LTER Network Office. Tony is a staff scientist at the San Diego Super Computer Center (SDSC) and a liaison between SDSC and the Long Term Ecological Research Network. Lixin is helping LTER to implement climate models, using the SDSC. Both are on a one-year assignment.
The mission of SDSC is to provide world leadership in advancing knowledge through the development and application of advanced computing technologies. In operation since 1985, SDSC is a campus research unit of UCSD and the foundation for activities of the National Partnership for Advanced Computational Infrastructure (NPACI).
The Partnership, funded by a five-year, $170 million grant from the National Science Foundation, teams 37 leading institutions in high-performance computing to create and deploy a leading-edge, ubiquitous, pervasive, and continuous computational infrastructure to support disciplinary and interdisciplinary research by the national research community.
Tony Fountain’s research interests are in the development and dissemination of methods and tools for exploiting high-performance computing resources (computers, software, storage devices, and networks) for earth scientists. "High-performance computing enables scientists to address questions that are difficult or impossible to study by other means," says Tony.
"The complexity and scale of earth system processes make them ideal candidates for focusing computer science research and development, Tony says."
Concentrating on parallel computing and super computing, Tony is particularly interested in large-scale earth system simulations, and data management and analysis.
"Especially data mining, machine learning, and computer-intensive statistical methods," Tony says. "In general, I am interested in the application of methods from artificial intelligence to problems in earth systems science." Other interests include data visualization, graphical representations and animations of scientific data, and the use of visual depictions to understand and explain scientific phenomena and scientific models.
Tony’s immediate goals include the creation of an experimental testbed to facilitate the experiments outlined at the 1998 Biological Scale Process Modeling Workshop. "This testbed will feature an on-line archive of models and data, which will be connected to the NPACI supercomputing platforms and the mass storage systems at SDSC, and accessible to LTER researchers." (For more about NPACI, visit the website: http://www.npaci.edu/)
A second project is the development of analysis and visualization tools and techniques for processing LTER data and model output. These tools will employ commercial packages such as IBM’s Data Explorer and SGI’s MineSet. Particular applications include fly-through animations of selected LTER sites incorporating digital elevation maps, aerial photography, and various land-cover data layers.
As with other visualization techniques, fly-through animations provide perspective and insight to data. "For large data sets, a single image is either too large or too coarse for convenient analysis," Fountain says. "Fly-through animation provides visualization at scales that are appropriate for the analysis tasks-far away for general characteristics, close in for details." Fly-through animations are also useful for education and outreach. "They convey a lot of information about a site in a manner that is interesting and enjoyable."
A second set of tools and techniques will involve the application of analytic data mining techniques to LTER data. Data mining is the process of extracting useful information from large databases. The goal is to discover useful patterns that can aid understanding and assist problem solving. Data mining includes techniques from statistics (e.g., regression, clustering) and machine learning (e.g., rule induction, neural networks).
A third project involves enhancing collaborative research between SDSC and LTER by developing high-speed network connections between SDSC and selected connections between SDSC and selected LTER sites. Initial explorations will focus on connecting SDSC to Colorado State University and to Michigan State University.
A fourth project is to learn more about the mission and activities of the LTER Network, to meet more of the LTER members, and to identify significant computing issues within the LTER community. "We plan to create an archive of models, data, and tools for use by the LTER researchers. This will lay the groundwork for future project proposals to strengthen the collaborative relationship between the SDSC and the LTER Network," Fountain says.
Lixin’s expertise in climate and ecosystem modeling will help facilitate the interactions between numerical modeling and field observation in the the LTER Network, enhancing understanding of the interactions between the earth and the atmosphere.
"To move those ecosystem models (maize model, etc.) and hydrological models (splash, etc.) to the Supercomputer, and to make them accessible to various research communities can be a highly rewarding project for the LTER Network Office and a crucial step for model integration efforts to the scientific community as a whole," Lixin says. "I am excited about the possibility of being part of it. At the same time, I am confident that to run the coupled model at high resolution over a long term on a super computer will help further investigate the two-way feedbacks between the atmosphere and biosphere, and also it can be a very good pilot project to demonstrate the importance of cooperations between different research groups."