Science and Weather

U.S. supercomputer now doubles power for hurricane forecasts

July 30, 2013 at 2:20 p.m.

According to the U.S National Weather Service, the computers running their weather forecast models have just received a substantial upgrade — more than doubling the computing power they have access to — and this is just the first step in what promises to be a 'quantum leap' in weather forecasting accuracy.

Up until last week, the supercomputer that the National Weather Service uses — called 'Tide' — was able to run around 90 trillion calculations per second. Using this power, they could track the motion of weather patterns across the nation and around the world, including the movement and rotation of the powerful hurricanes that spin up during the summer and fall. However, the incredible complexity of our atmosphere — how it moves and interacts to create weather — taxes even that level of computing power. As a result, weather models have to break down the atmosphere into 'packets', several kilometres long on a side, treating each packet as one point, in order to run all the computations fast enough that we can get a timely forecast.

The problem with having to break everything down into these packets is that the packets are sometimes larger than the storms they're trying to predict. The model can get some general indication that something's going to happen inside that packet, but it's hard to pull out specifics. One way to solve this problem is to increase the amount of operations your computer can perform. That will let you make the packets smaller, or it will let you combine several runs of the model in the same time it used to take you to just run it once. Either way, it increases the accuracy of your forecast.

As of last Thursday, both Tide, and its backup — known as Gyre — are now running at 213 trillion calculations per second, more than double the computing power they had before.

"These improvements are just the beginning, and build on our previous success. They lay the foundation for further computing enhancements and more accurate forecast models that are within reach," said Louis W. Uccellini, the director of the National Weather Service, in a NOAA statement.

The first model to benefit from this improvement is the Hurricane Weather Research and Forecast System (HWRF), likely due to the 'active to extremely active' hurricane season that was forecast by the NWS back in May. The boost in computing power for this model will improve forecasts of storm tracks and intensity, and they've apparently already seen a 15 per cent improvement over model runs from last year. That may not seem like a whole lot, but when your dealing with a storm that's 1,500 kms wide (like Hurricane Sandy was), 15% can improve estimates of where the storm is going to hit by over 200 kms. Also, for storm intensity that could improve the forecast by as much as a category of hurricane strength, thus warning people much sooner of the potential for a devastating storm.

This is just the beginning, though.

"Next comes the quantum leap," Uccellini said in the statement.

By summer of 2015, the NWS plans on increasing the computing power of Tide and Gyre to nearly 10 times what they can do now, bringing them both up to 1.95 quadrillion calculations per second. According to Uccellini, that will let them run an enhanced version of their Global Forecast System (GFS), which will result in better forecasts for all kinds of weather events — thunderstorms, tornadoes, heat waves, snowstorms, etc.

[ More Geekquinox: Edmonton museum hosts Russian meteorite fragments ]

This is excellent news for both weather forecasters and for the public.

There's a general perception in the public that weather forecasters are always wrong. The reason for that is because noone goes around commenting about it when the weather forecast was right. It's only when the forecast was wrong that it becomes 'noteworthy' and sticks in our memory.

Weather forecasting is tough, though. There are still uncertainties that can foil even the best forecasters, and that is only going to get worse in the future. Climate change due to global warming is going to shift weather patterns and make the atmosphere even more chaotic than it already is, thus making it even harder to predict what it will do.

However, with increases in computing power like this latest development, and the further advances to come, forecast accuracy will improve, regardless. Public perception may not improve with it, but the lives and property that will likely be saved by these improvements, continued 'ribbing' from the public is a small price to pay.

(Images courtesy: NASA/NOAA)

Geek out with the latest in science and weather.
Follow @ygeekquinox on Twitter!