Today's forecast via Climate Reanalyzer. The blue zone is -20 to 20oF.
It's cold as hell. The polar vortex spinning in the Arctic pushed a brutal chill down into the US, plummeting temperatures to their lowest in 20 years. With most of the country freezing their behinds off right now, weather's been on the brain more than usual. But humans have been thinking about, talking about, and desperately trying to predict the weather for millennia.
There are all kinds of great reasons to want to know what the elements are going to subject us to, running the gamut from knowing whether you should take an umbrella to work to saving lives—the polar vortex has already been blamed for 21 deaths. Unfortunately, weather soothsayers are notorious for getting it wrong. But 21st century tech trends like big data, crowdsourcing, and supercomputing are changing that. The future of weather forecasting actually looks sunny.
Predicting how our chaotic, ever-changing atmosphere is going to behave is extremely complex. Part one is observation. Data on current atmospheric conditions is constantly collected from satellite imagery, radars, weather stations and buoys, sensors on ships and airplanes, and sent back to weather organizations. Scientists feed the deluge into massive computer simulations of the Earth’s climate—part two.
Supercomputers running complex mathematical equations perform trillions of calculations to determine the future properties of the atmosphere. But even with today's super-fast machines, computer models can only forecast about two weeks out, and still spit out hundreds of possible outcomes. It’s up to humans’ best guess from there.
A supercomputer simulation of near real-time wind patterns, via Cameron Beccario
To get more accurate predictions, scientists need to better understand atmospheric patterns, which means collecting larger amounts of more precise information about the elements for observation. To that end, the National Oceanic and Atmospheric Administration has started installing Water Vapor Sensing Systems on airplanes. The sensors measure the changing moisture levels in the air throughout flights and send back the data in near real-time to the National Weather Service. Water vapor is a valuable data point; it lets scientists determine the timing of fog and cloud formation, which helps anticipate thunderstorms.
Video via Youtube/NutsAboutSouthwest
Meanwhile, NASA and Japan’s space agency are launching a global precipitation satellite into the sky this February to study the intricate details of raindrops, snowflakes, and ice particles in the air. The "next-gen observation" will help forecasters quantify when, where, and how much it rains and snows around the world, crucial for understanding weather patterns.
Other bourgeoning climate observation techniques are more scrappy, like the smartphone app pressureNET that's crowdsourcing the data-collection process. The Android app automatically collects atmospheric pressure measurements using barometers in the smartphones. It sends the data from thousands of phones to scientists, who incorporate it into climate models.
As sensors proliferate and citizen scientists feed meteorologists with more and more data, larger and more complex computer models can be developed. And so improving weather forecasts is inextricably linked to bigger, faster computers. This year the NOAA’s supercomputers got a big upgrade; they can now process more than 200 trillion calculations per second. By 2015, the organization expects its machines will be 20 times faster than they are now.
Naturally, the further into the future scientists try to foresee, the less precise their predictions get. More powerful machines extend that time period. If a 20-fold speed boost could be the difference between predicting a storm's activity for six days or seven, imagine what a quantum computer could do. We could theoretically plan a vacation six months in advance and know if they’re forecasting sun that week. Had I that kind of lead time before this frigid-ass weather, I'd have made sure to skip town.