Whether you’re involved in the high performance computing or not, almost every one of your days is likely planned around the findings of HPC-enabled advanced modeling and simulations for weather forecasting that tell us whether to carry an umbrella, wear extra sunscreen, or if the roads will be safe to drive on.
But as more severe weather events occur and the threat of climate change grows more serious, weather forecasting is taking on even more prominence. And the trend of blaming the weatherman for botched predictions is only set to go down as more sophisticated systems and models enter into the climate equation.
Most recently, Spain made steps toward this end by commissioning a new bullx supercomputer through the Spanish Meteorological Agency (AEMET) with the hopes of bolstering the country’s weather forecasting and climate research.
The addition of the Bull supercomputer is expected to help bridge the decades-wide gap that exists between short-term weather forecasting and the extended climate predictions. The result should be a more accurate understanding of climate change that is specific to each of the country’s regions, in each season of the year.
The bullx system is on track to provide performance in the range of 168 teraflops, which represents 75 times the capability of the Cray supercomputer it will replace. Expected to weigh in as Spain’s third-fastest supercomputer, the system will feature Bull’s direct liquid cooling system, which uses water at room temperature to generate energy consumption savings of 20-40 percent compared with traditional air-cooling or cold water-based cooling systems.
To complement the hardware upgrade, AEMET also plans to implement new forecasting models that are capable of making predictions down to the 1-2 mile scale, to better cope with severe local weather such as thunderstorms.
The modeling system, called HARMONIE/Arome has already made its mark on the European meteorology by offering an initial range of 36 hours and spatial resolution of 1.5 miles. The models will focus on the combination of the Iberian Peninsula and Balearic Islands, as well as the Canary Islands.
A large-scale atmospheric chemistry model already in place at AEMET, called MOCAGE, is expected to see improved resolution, more extensive coverage and more accurate air quality predictions. MOCAGE will also have an emergency mode for real-time predictions in case of a hazardous materials leak or other industrial accident.
This news follows a recent trend across the US and Europe of bolstering its climate infrastructure. Most recently, the US National Weather Service announced that it has signed a $44.5 million contract with IBM to deliver a ten-fold increase in computing power. Since Big Blue spun off its x86 biz to Chinese-owned Lenovo, Cray has been tapped to supply the new systems. When the work is completed later this year, NWS will have a combined 5-petaflops of supercomputing capability, enabling it to run the kind of sophisticated weather predictions that are the lifeblood of its mission.
Similarly, the Met Office in the United Kingdom has signed a $128 million deal with Cray for multiple Cray XC supercomputers and Cray Sonexion storage systems. The three-phase, multi-year project represents the largest supercomputer contract ever for Cray outside of the United States, and is expected to complement other Cray systems seeking to answer weather and climate problems posed at the European Center for Medium Range Weather Forecasting (ECMWF) and several other weather prediction and climate forecasting centers worldwide.
The Met Office, like representatives from the US and Spain, is emphasizing the necessity of the upgrades in order to ensure forecast accuracy, which ultimately means greater planning and safety during severe weather events.
This need was recently illustrated when ECMWF accurately predicted the touchdown of Hurricane Sandy on the United States East Coast with the help of two 2.5 petaflops systems, while domestic (US) models running on twin 213-teraflops systems showed the hurricane heading away from land.
Back in Spain, AEMET is planning to further leverage its upgrades by using the bullx system off of land as well. The agency is expanding its focus to encompass maritime forecasting services, wave prediction and hydrological monitoring, where it will predict and study the effects that rain and snow may have on watercourses.