The US National Weather Service will be getting a 10X boost in supercomputing power, thanks to $44.5 million contract with IBM for two new Cray systems. The upgrade will increase the throughput of the agency’s combined supercomputing resources from about .5 petaflops to five petaflops when the project is completed later this year, enabling it to provide “more timely, accurate, reliable, and detailed forecasts.”
Plans to renew the infrastructure have been in place since July 2013 with $25 million set aside from the Hurricane Sandy supplemental bill. IBM was already under contract to supply the machines, but the company’s sale of its x86 server business to Chinese company Lenovo delayed the purchase. With the US opting to forego the security risk of using Chinese-owned parts, the parties worked out an agreement for IBM to use Cray as a subcontractor.
As detailed in this press release from the National Oceanic and Atmospheric Administration (NOAA), there are actually two upgrades planned for the forecasting systems. This month, capacity for each of two identical IBM machines is being tripled from .213 petaflops to 0.776 petaflops for a total combined capacity of 1.552 petaflops. According to NOAA, this is sufficient power for the agency to begin running an upgraded version of the Global Forecast System (GFS) that uses higher resolution and a longer time span (resolution will go from 27km to 13km out to 10 days and 55km to 33km for 11 to 16 days). The Global Ensemble Forecast System (GEFS) will undergo a similar upgrade with the number of vertical levels expanding from 42 to 64 and the horizontal resolution shifting from 55km to 27km out to eight days and 70km to 33km from days nine to 16.
The Hurricane Weather Research and Forecasting model has also been upgraded, and the High-Resolution Rapid Refresh model, which delivers 15-hour numerical forecasts every hour of the day, is now operational, enabling meteorologists with NOAA’s National Weather Service to better predict the path, timing and intensity of adverse weather events.
Over the last few years, US forecasting prominence has slid as rival forecasting groups in Canada and Europe have invested in higher-powered computers. Critics point out that the European model, ECMWF, correctly predicted that Hurricane Sandy would hit the East Coast while the American hurricane model still showed the storm heading off-shore away from land.
Currently, the European Center for Medium Range Weather Forecasting has two computers that run at roughly 2.5 petaflops each, whereas the US has two systems with about 213 teraflops each. Despite having this ten-fold disadvantage in compute power, the NWS is responsible for running more compute intensive local predictions that are not part of the European Center’s mission.
Cliff Mass, a professor of atmospheric sciences at the University of Washington and an outspoken proponent of improved US forecasting, is optimistic about the latest development. “Everything can change now, IF the National Weather Service uses these powerful new computers wisely,” he writes in a recent blog entry.
“Properly used, this new computer power can revolutionize and greatly improve the skill of U.S. numerical weather prediction, with huge positive impacts for the country,” continues Mass. “Problems in U.S. numerical weather prediction have not been limited to lack of computer power and these problems need to be addressed, such as an inability to entrain the huge knowledge based in the huge U.S. academic research community or inefficiencies/duplication of effort in U.S. government research and development. These are issues that must be taken on now. But the excuse of lack of computer power is gone and a renaissance in U.S. NWP is possible.”
Kathryn Sullivan a former astronaut who took over as NOAA administrator in March 2014, has been tasked with balancing the NOAA portfolio, no easy feat in these times of political divisiveness and sequestration-impacted budgets.
“NOAA is America’s environmental intelligence agency,” said Sullivan in a statement. “We provide the information, data, and services communities need to become resilient to significant and severe weather, water, and climate events. These supercomputing upgrades will significantly improve our ability to translate data into actionable information, which in turn will lead to more timely, accurate, and reliable forecasts.”
“We continue to make significant, critical investments in our supercomputers and observational platforms,” added Louis Uccellini, director, NOAA’s National Weather Service. “By increasing our overall capacity, we’ll be able to process quadrillions of calculations per second that all feed into our forecasts and predictions. This boost in processing power is essential as we work to improve our numerical prediction models for more accurate and consistent forecasts required to build a Weather Ready Nation.”