A geosciences consortium for high performance computing
A scientific opportunity: The geosciences are poised for an era of rapid scientific progress. We are on the verge of more accurate advance warning of extreme weather and earthquakes, skillful prediction of space weather events, detailed simulation of precipitation and stream flow in river basins, high-resolution climate modeling, and eddy-resolving ocean models. The single most important factor in achieving these advances is a major expansion of supercomputing capabilities dedicated to geosciences research, so this is the highest priority in NCAR's new strategic plan. This expansion would capitalize on NCAR's 30+-year position as the only major supercomputing center within the geosciences community, and one that provides an integrated set of research, services, and outreach.
There are three elements to the needed expansion; one is creating a new supercomputing facility that can support the technological developments foreseen over the next several decades (which cannot be accommodated at NCAR's Mesa Laboratory), the second is expanded and sustained investment in new supercomputing systems as they become available, and the third is the engagement of the broad geosciences community to utilize and support this initiative.
NCAR plans to develop an integrated science, facility, and funding prospectus bringing the three elements together. NCAR is beginning to work with partners to create the facility based on an advanced conceptual design that is flexible enough to cope with the uncertainty of multiple future funding and computing technology scenarios.
NCAR's criteria for site selection seeks to minimize objective risk, provide space for complete build-out over 20 years, and provide access to utilities, water, and wide-area network connectivity. NCAR has assumed and plans to maintain a leadership role in building the appropriate scientific and facility partnerships that will propel this critical project forward to completion.
A consortium approach: Building on the community service model that has proven successful for NCAR over the last four decades, we believe that a consortium approach is required to take full advantage of the scientific opportunity. We aim to create both a set of physical facilities and a more integrated "national geosciences supercomputing community" that includes facility providers, developers and providers of services and applications, computer and computational scientists, scientific users from geosciences disciplines, and educators. The facilities that are needed include a highly capable, extensible, and networked central supercomputing facility and a group of associated sites that are largely dedicated to grand challenge computational problems in the geosciences.
Community governance: NCAR management, sponsor agencies, and the user community jointly govern NCAR's current supercomputing assets, setting overall direction and priorities, and allocating machine time among current user disciplines. Specific allocation of machine cycles is done through a peer-review process. We believe that community governance for a larger, more disparate group of users will require a multi-disciplinary process to allocate time among defined scientific areas coupled with discipline-specific processes to allocate within areas.
|
 |
FY2005 Annual Report |
|
|
|