Krystal Temple

Weather’s Next Top Model

Lately, “how’s the weather where you are?” has been a more interesting conversation starter than usual. From superstorms to blizzards, forecasters are predicting dire weather events on an almost weekly basis. Whether these predictions actually come true, however, is another matter and has led to some embarrassing moments for local and national forecasters.

The most recent example of this was the ominously nicknamed, “Snowquester.” With nearly a foot of snow forecast Washington, D.C., the capital prepared for the worse – the government shut down, schools were closed, flights were cancelled, supermarket shelves emptied and shovels and snow blowers were at the ready. So what happened? A slushy mess that didn’t even amount to one inch. Aside from the more serious inconveniences and economic ramifications this presented, the Washington, D.C. Snowball Fight Association also had to cancel a planned battle in DuPont Circle.

Meteorology has never been an exact science although advanced technology has resulted in more accurate forecasts over the years. And forecasters place a lot of faith in certain weather models in order to offer a more confident weather prediction.

These models represent simulations created by powerful supercomputers that analyze complicated numerical data related to temperature, barometric pressure, wind direction and humidity. A lot of attention has been paid to the European model, delivered by the European Centre for Medium Range Weather Forecasting (ECMWF), which has proven to be more accurate than the U.S. models of the National Weather Service. For example, the European model predicted that Hurricane Sandy would make a direct strike on the East Coast, days before the U.S. model identified that same pattern.  In the case of “snowquester,” U.S. forecasters predicted 10 inches of snow, while forecasts based on the European model predicted that it would be too mild for a significant snowfall.  

So why the discrepancy? Perhaps the biggest reason is that ECMWF has more computing power. The ECMWF model can process higher resolution images which allows forecasters to better pinpoint lines of snowfall, rain and other patterns. They’re also able to provide more frequent models – up to 100 per day – than the National Weather Service (84 per day).

A Look at the NWSC Supercomputer's 100 Racks. (©UCAR. Photo by Carlye Calvin.)

However, the U.S. isn’t standing pat and investments are already being made to level the computational – and forecasting – playing field.  For example, Intel is working with organizations like the National Center for Atmospheric Research (NCAR) to deliver the computing power needed to study the climate and atmosphere in the new NCAR-Wyoming Supercomputing Center (NWSC). According to NCAR, the new supercomputers can run one and a half quadrillion operations per second, in other words a one followed by 15 zeros!

In just one node in these supercomputers there are two Intel chips, which includes eight processors and billions of transistors. Check out this video to learn more about why such powerful computers are needed to analyze the vast amounts of weather related data.

Accurate forecasts mean we can better plan out home improvement, travel and recreational activities. Business can better guard against lost work time, state governments can better manage their emergency resources, and when it comes to public safety, a portion of weather-related fatalities could be reduced with more accurate forecasting and better warning systems.

The lesson here is that better computing means better forecasting. Of course, from a personal standpoint, better accuracy also removes the stress level when it comes to those all-important “what to wear” decisions.

Comments are closed.