August 12, 2013
The deadly EF5 tornado that hit Moore, Oklahoma on May 20 was unique in several ways. Not only was it one of the strongest twisters ever recorded, but forecasters were able to issue a tornado warning 36 minutes in advance, saving lives. Playing a part in that forecast was a Cray supercomputer at the National Institute for Computational Sciences (NICS).
NICS' new Cray XC30 supercomputer, known as Darter, was used by researchers at the Center for Analysis and Prediction of Storms (CAPS) at the University of Oklahoma from April 22 through June 7, right in the heart of the spring storm season. The weather prediction work was done as part of the spring Hazardous Weather Testbed (HWT) experiment sponsored by the National Oceanic and Atmospheric Administration (NOAA).
Darter, which has nearly 12,000 Intel Sandy Bridge cores and 250 teraflops of peak capacity, was used to calculate the detailed Storm-scale Ensemble Forecasts (SSEF) that regional weather forecasters--such as the National Weather Service office in Norman, Oklahoma that issued the 36-minute, life-saving warning on May 20--rely on to predict tornados and other severe weather events.
According to a story on the NICS website, Darter generated 30 SSEFs every day from May 6 through June 7. Each of those forecasts contain severe weather predictions for the entire continental United States across a horizontal grid with 4-kilometer spacing over a period of 48 hours.
The CAPS forecasters ran several numerical weather models on Darter to generate the SSEFs, including the Advanced Research version of the Weather Research and Forecast model (WRF-ARW), the Advanced Regional Prediction System (ARPS) and the Navy COAMPS model system. As part of these models, real-time data was collected from more than 140 Doppler weather radar installations and conventional observations around the country, and assimilated using the ARPS 3DVAR and Complex Cloud Analysis package, according to the NICS story.
The NICS supercomputer at the University of Tennessee is credited with helping the CAPS forecasters outperform other severe storm forecasts during the spring 2013 storm season. According to an analysis by CAPS published on the NICS website, the SSEFs it generated on Darner were more accurate in predicting precipitation than all other ensemble forecast members, including the National Center for Environmental Prediction (NCEP) operational North American Mesoscale (NAM) model forecast.
"The dedicated use of the powerful NICS-managed Darter supercomputer made a big contribution to a very successful 2013 Spring Experiment season," Fanyou Kong of CAPS tells NICS. "It allows CAPS to not only produce real-time baseline storm-scale ensemble forecasts for HWT but also to run several combinations of ensemble configurations to find optimization for more accurate prediction of severe-weather events."
The HWT is a yearly experiment that investigates the use of models in the prediction of hazardous convective weather, such as tornadoes and severe thunderstorms. In addition to CAPS, several other organizations participated in NOAA's spring HWT this year, including NOAA Storm Prediction Center (SPC) and the National Severe Storm Laboratory (NSSL).
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?