Supercomputers: Machine Meteorologists That Help Issue Your Forecast

If you've seen this recent Intel commercial, you may be asking, what's a supercomputer and how does science use it? 

Supercomputers are extremely powerful, school-bus-sized computers. Their large size comes from the fact that they're comprised of hundreds of thousands (and sometimes millions) of processor cores. (In comparison, your laptop or desktop computer runs one.) As a result of this collective computing capacity, supercomputers are immensely powerful. It's not unheard of for a supercomputer to have a storage space capacity in the neighborhood of 40 petabytes or 500 tebibytes of RAM memory. Think your 11 teraflop (trillions of operations per second) Macbook is fast? A supercomputer can reach speeds of tens of petraflops—that's quadrillions of operations per second! 

Think of everything your personal computer helps you do. Supercomputers do the same tasks, only their kicked-up power allows for volumes of data and processes to be researched and manipulated. 

In fact, your weather forecasts are possible because of supercomputers.

Why Meteorologists Use Supercomputers

Read More

Weather Satellites: Forecasting Earth's Weather From Space

By Tiffany Means

Every hour of every day, billions of weather observations are recorded by weather satellites, weather balloons, ocean buoys, and surface weather stations around the world. Supercomputers provide a home for this tidal wave of weather data to be collected and stored. 

Supercomputers not only house volumes of data, they process and analyze that data to create weather forecast models. A weather model is the closest thing to a crystal ball for meteorologists; it's a computer program that "models" or simulates what the atmosphere's conditions could be at some time in the future. The models do this by solving a group of equations that govern how the atmosphere acts in real life. In this way, the model is able to approximate what the atmosphere is likely to do before it actually does it. (As much as meteorologists enjoy doing advanced math, like calculus and differential equations...the equations used in models are so complex, it would takes weeks or months for them to solve by hand! On the other hand, supercomputers can approximate solutions in as little as an hour.) This process of using model equations to numerically approximate, or forecast, future weather conditions is known as numerical weather prediction.

Meteorologists use forecast model output as guidance when building their own forecasts. The output data gives them an idea of what's currently happening at all levels of the atmosphere and also what's possible in the coming days. Forecasters take this information into consideration along with their knowledge of weather processes, personal experience, and familiarity with regional weather patterns (something a computer can't do) to issue your forecast.

Some of the world's most popular weather forecast and climate monitoring models include the: 

• Global Forecast System (GFS) 
• North American Model (NAM)
• European Center for Medium-Range Weather Forecasting Model (European or ECMWF)

Meet Luna and Surge

Now, the United State's environmental intelligence capabilities are better than ever, thanks to an upgrade of the National Oceanic and Atmospheric Administration's (NOAA) supercomputers.

Named Luna and Surge, NOAA's computers are the 18th fastest in the U.S. and among the top 100 most powerful supercomputers in the world. The supercomputer twins each have nearly 50,000 core processors, a peak performance speed of 2.89 petaflops, and process up to 3 quadrillion calculations per second. (Source: "NOAA Completes Weather and Climate Supercomputer Upgrades" NOAA, January 2016.) 

The upgrade comes at a pricetag of $45 million—a steep figure, yet a small price to pay for the more timely, more accurate, more reliable, and more detailed weather forecasts the new machines offer the American public.

Could our U.S. weather resources finally be catching up to the renowned European model —the UK's bullseye-accurate model whose 240,000 cores led it to accurately predict the path and strength of Hurricane Sandy nearly a week before it hit the New Jersey coastline in 2012?

Only the next storm will tell.