Supercomputers Aid Weather Forecasts

INTERVIEW NOAA, the government organization tasked with monitoring climate and weather, has flipped the switch on a new line of supercomputers from IBM. The computers, named "Red," "White" and "Blue," will each be tasked with various functions, with one serving as a backup in case of failure.

According to the NOAA, the computational power available to forecasters has increased from 450 billion calculations per second to 1.3 trillion calculations per second.

But, what can we expect from this increase of power in terms of the forecasts we see on television, and the accuracy and timeliness of warnings? To get the answers, BetaNews sat down with the National Center for Environmental Prediction's Director of Central Operations, Kevin Cooley.

BetaNews: Thanks Kevin for taking a moment to answer our questions. To begin, why did you come to the NOAA?

Kevin Cooley: I came to the NOAA NWS after spending the first part of my career in the private sector. After serving in the Marines, I worked for EDS and IBM. After this, I worked for a few years as a consultant specializing in program management and enterprise architecture.

From my days serving in the Marines, I came to appreciate the value of public service and had always looked to return to public service during my career. When the opportunity emerged to join the NOAA NWS, I jumped at the chance. It is a great organization with an important mission. I am very happy with the work!

BetaNews: You were hired in March 2002, but the announcement for the IBM partnership was not made until May of that year. Were things already in motion for these computer upgrades, or were you tasked with the responsibility of modernizing the system?

Kevin Cooley: When I arrived at the NOAA NWS, the acquisition for the CCS was underway. Given the computational requirements of the NOAA NWS numerical weather and climate prediction models, an upgrade of the NOAA NWS high performance computational systems is necessary about every 24 months. The acquisitions activity that was already underway when I arrived was chartered by NOAA NWS and NOAA IT executive management in recognition of this need.

It was my conclusion that the leadership team of this acquisition was performing well. I saw no need to change the functioning of this team. They made a sound acquisition recommendation. My organization (NCEP Central Operations) has the responsibility for managing the CCS contract and as such I have direct responsibility for all aspects of the implementation and daily operations of the CCS.

BN: Will the public notice any immediate benefit from these computer upgrades, or could you speculate on how long it would take to see some kind of benefit from this increased computing power?

KC: We will see the benefit of the new computational capacity delivered in the upgraded systems immediately. Because of the new capacity, we have instituted additional runs of our new Climate Forecast System. Through out the late spring and early summer of this year, we will increase the resolution of our hurricane model from 18km to 9km, and our Rapid Update Cycle (RUC) model used to support aviation weather forecasting from 20km to 13km. These resolution increases allow us to view the atmosphere in smaller chunks and thereby improve the quality of the forecasts.

Additionally, we will upgrade the internal physics calculations of our North American Mesoscale model to improve the accuracy of its predictions. These and other improvements to the NOAA NWS production model suite will directly improve the quality of the weather forecasts provided to the nation.

BN: Will our weather models need to be "upgraded" in order to take advantage of this?

KC: As I indicated previously, we will enhance our models to take advantage of the additional computational capability afforded by the new system. However, we have not had to engage in significant porting of the model code to move from the old to the new machines. The level of compatibility of the computers provided by IBM was excellent and we were able to commence executing our model suite on the new computer with a minimum of effort.

BN: Some in the scientific community have criticized the American weather models as being inferior in capability to their Asian and European counterparts. Does the NOAA think this will help to level the playing field?

KC: Ed, we have a great and highly dedicated group of public servants engaged in the scientific development of new and improved models to support weather and climate prediction. While we always appreciate the resources provided to us by the taxpayers, we continue to strive for excellence and to fully and efficiently use these resources; regardless of whether they are as large as those used by other organizations in other countries. Our measure of success is not how big our computers are but rather how much value we generate for the public using the resources that we have.

These things being said, this new increment of computational technology will certainly improve our model and forecast performance and, in addition to providing value for the taxpayer, this new technology will make us more competitive with other forecast operations.

BN: Last question: In your announcement, it says NOAA has struck a nine-year contract with IBM. When does the partnership end and what can we expect from this partnership in the future?

KC: The contract expires in 2011. The taxpayers can expect that the partnership with IBM for weather and climate high performance computing will lead to the fielding of the computing systems necessary to support the generation of the weather and climate forecast products required to preserve life and property and support the Nation's economy through the remainder of the decade. We expect to field systems by this time that are at least 48 times more capable than those in use when we started the partnership in 2002.

BN: Thank you for your time, Kevin.