SCD logo
Table of Contents | Community Service | Highlights | Publications | Staff, Visitors & Collaborators


Welcome to SCD's FY1999 Annual Scientific Report. The table of contents allows you to access any part of the report. You may read the report in sequence by following the "Next" links on each page.

Message from SCD Director Al Kellie

Having assumed the directorship of SCD in December 1998, I am pleased with SCD's contribution to the overall scientific mission of NCAR and its university community.

SCD Director, Al Kellie
SCD Director Al Kellie announces the arrival of blackforest, a 288-processor IBM SP cluster. UCAR President Rick Anthes and IBM vice president for SP systems Lou Bifano participated in the blackforest dedication ceremony.

We have much to applaud during the 12 months covered in this review. Most notably is the articulation and actualization of the first steps in our new computing hardware roadmap. The first significant step is the acquisition of the new IBM RS/6000 SP cluster named "blackforest." This machine provides the means by which NCAR scientists can make the transition to Distributed Shared Memory (DSM) architectures that eventually will permit significant achievements in the execution of climate and weather codes.

In the U.S., market forces (particularly in the commercial arena) and price performance are moving high-performance computing away from expensive, specialty-built supercomputers. Supercomputers today are composed of components that yield profits in the commercial and e-commerce sectors of the vendors' portfolios. In fact, the marriage of technical scientific computing with commercial computing has benefits for both. We see an emphasis on reliability that is demanded by the Charles Schwabs of the world because downtime is a direct hit on the corporate bottom line.

NCAR's scientific computing brings a need to push the hardware and software out to their limits; we challenge vendors to rise to the occasion. We take the risks that cannot be pursued by the commercial users. Distributed shared memory cluster architectures are likely to become the most important component of the SCD computational equipment hierarchy for the next several years. Indeed, during the period covered by this report, SCD has developed an aggressive roadmap of equipment acquisition that presents options for placing NCAR in the Teraflops arena during FY2000. Attendant with this increase is our commitment to provide user training and code conversion assistance so that our client community of researchers will be able to take full advantage of these new computing tools.

Our goal is simply this: to provide university and NCAR scientists with the fastest, most reliable and efficient computing environment available to meet the requirements of atmospheric science research.

Because of the ongoing rapid increase in computing cycles, we have prepared plans for advancing our Mass Storage System (MSS), long regarded as the crown jewel of NCAR's many unique assets. We are developing the fourth generation of mass storage systems at NCAR. Modern object-oriented methods have been incorporated into the design. For the first time, the metadata and charging information will be available through web interfaces. This new generation of user interface will keep pace with increased data storage and retrieval demands into the next century, while at the same removing many of the machine dependencies and relying more on open standards.

With MSS-4, the promise of distributed hierarchical storage is closer to a reality. With more than 200 Terabytes of data stored on the system (and a current monthly growth rate of 5 TB per month), NCAR's MSS is a storage system that is without peer anywhere in the world.

SCD's Visualization Laboratory continues to push the envelope of stereo, three-dimensional scientific visualizations. Empowering scientists to visualize large, complex tera-scale datasets in a manner that is both intuitive and instructive, SCD continues to lead the way in providing NCAR scientists with an invaluable tool for analyzing and understanding their data. Further, the laboratory is a frequent venue for visitors and other interested parties to literally see the results of NCAR's simulations of geoscience processes.

SCD's web-based information infrastructure continues to provide timely information on the machines we support and to supplant the traditional means of getting the message out. Moreover, as part of the UCAR ITC plan, SCD has undertaken to host and manage UCAR's web sites. We foresee continued advances in web-based technologies and expect they will be more ingrained into the fabric of tools provided by SCD. SCD will remain an advocate for and implementor of these technologies to simplify the use of our computing infrastructure and disseminate information.

Data Support continues to provide a wide spectrum of data for atmospheric and oceanographic research studies. Numerous extant datasets are routinely updated, and participation in worldwide-recognized efforts make the SCD archives invaluable. Preparing input and receiving output from atmospheric reanalysis projects (e.g. the NCEP/NCAR Global Atmospheric Reanalysis 1948-1999, ECMWF 1979-1993 Reanalysis, and ECMWF 1958-2001 Reanalysis, forthcoming) has brought together a unique combination of many historical data into significant research collections. Furthermore, data from the continental-scale hydrological cycle and energy-exchange studies within the GCIP project, plus comprehensive marine surface datasets from the COADS project are important components for the broad scope of modeling and climate research needs throughout NCAR/UCAR.

The FL4 Uniform Network project was part of the ongoing fulfillment of UCAR's strategic plan for the network infrastructure. It will bring all UCAR sites into cabling standardization by the end of calendar year 2000. Another of the important projects for the networking strategic plan is the completion of the LAN cabling upgrades. A flexible LAN structure at UCAR is vital to UCAR's future.

SCD has embarked on a new relationship with the other NSF supercomputer centers (NCSA and SDSC), known as Partnerships for Advanced Computational Infrastructure (PACI) centers. In signing a new Memorandum of Understanding this summer, we outlined three broad themes under which activities can be initiated. Projects in the three areas of training and outreach, earth system visualization, and virtual computing environments offer enhanced cooperation between the three centers and are strongly supported by the CISE directorate of the National Science Foundation.

I invite you to read our divisional report to learn more about the scope and breadth of activities undertaken by SCD this past fiscal year. Much has been accomplished, but there is much yet to do. SCD intends to continue to pursue major areas of technological innovation in support of atmospheric science at NCAR.

Al Kellie