In September of 1938, a massive hurricane traversed the Atlantic Ocean and made landfall in New England. Due to inadequate and incorrect forecasting, the storm struck farther north and with greater intensity than had been predicted, leaving residents and authorities with virtually no warning or time to properly prepare. The storm killed 682 people and damaged or destroyed more than 57,000 homes.
Such fallout would be highly unusual today, thanks in part to high-performance computing (HPC). Advanced computing technologies are now helping meteorologists predict weather events earlier and more accurately than ever before – an improvement that has undoubtedly saved countless lives and billions of dollars in damaged property.
Accurate weather forecasts might be helpful when you’re trying to get dressed for the day or wondering if you need an umbrella, but they also have far-reaching effects on many facets of society, including public safety, transportation, manufacturing, and economic growth. For example, a U.S. Census report found that the coastal population is growing rapidly and will continue to expand through 2020, making improved weather predictions and early warnings essential to protect this high-risk population.
The massive amounts of data required for weather prediction make forecasting a highly data- and compute-intensive exercise. For each forecast, data must be collected and assimilated from multiple sources, including satellites, dropsondes, weather stations and buoys, current and historical observations, and simulations of the Earth’s physical patterns and processes. All of this information is then input into complex meteorological models that simulate weather patterns that are likely to occur.
These increasingly complex computer models combined with rapidly expanding datasets comprised of current and historical weather observations from around the world make HPC an indispensable tool for this field. Transporting, processing, and storing weather-related big data requires new IT approaches and a robust compute infrastructure in order to deliver the great computational power that is needed for weather simulations. Additionally, weather centers must assure 100% availability at all times, meaning their computing infrastructures must be highly reliable and built to support mission critical workloads. The HPC toolbox offers the only computing technologies that are capable of handling these requirements.
HPC-driven weather simulations can help improve forecasting capabilities in a number of ways:
- Scientists can more accurately predict weather at local, regional, and global levels
- Weather centers can streamline the management of their IT infrastructure while increasing flexibility and reliability
- Weather-related systems can support operational workloads as well as storage and archival requirements
- Larger and more complex environmental simulations can be run in less time
The use of HPC tools for weather prediction has drastically improved the accuracy of forecasts in a relatively short period of time. The NOAA’s National Hurricane Center (NHC) recently examined the accuracy of storm location predictions each year since 1970, and found that not only are storm forecasts improving, but the ability to predict significant weather events even earlier is on the rise as well. For example, forecasts of the location of severe storms three days in advance are now as accurate as the two-day-ahead forecasts were just 10 years ago, and as good as the one-day-ahead forecasts were 20 years ago.
The weather can’t be controlled, but with HPC, predictive capabilities can be improved and severe weather events can be forecasted more quickly and accurately than ever before. Please follow me on Twitter @Bill_Mannel to stay up-to-date on the latest HPC innovations for the weather and climate industry.