Accurate weather forecasts can save both lives and a lot of money. But overstated predictions can often be costly too, as businesses shutter ahead of warnings of a monster storm that turns out to be nothing more than a little bit of sleet. No great fun yet not quite the snowpocalypse.
Why do forecasters often get it so wrong?
New York City was wondering that on Tuesday (March 14). Fear of a late-winter storm was stoked by the thought of as much as 18 inches (45 cm) of snow. Airlines cancelled more than 5,000 flights, schools were closed, subway service was curtailed and residents stocked up on food like it was their last supper. (Apparently, we’re all having mac and cheese when the end comes).
But the biggest snowfall hit north of the city. And much less than predicted fell in New York, leaving a slushy mess all over the Big Apple.
Snowfall totals may be the most difficult aspect of weather to predict
Meteorologists don’t lack data. They generate thousands of computer models based on information gathered from balloons the US sends up each day, and from satellite imagery.
But a large storm like the one that slammed into the northeast is so complex, there’s a greater chance of error than with other types of weather, says Gino Izzi, a meteorologist for the US government’s National Weather Service. A difference of less than one degree in temperature could determine whether precipitation falls as snow or a mix of sleet and rain, he says.
Please note: The atmosphere is three dimensional
That means not every square inch of the sky can be analyzed. “We don’t have precise values of everything that’s going on in the atmosphere,” Izzi says. For example, a rise in temperature at a certain altitude affects what precisely falls below. If that temperature value at that point wasn’t factored in or even recorded, forecasters relying computerized models may miss a critical change.
The US weather forecasts have fallen short compared with their European counterparts in some notable cases, perhaps most famously ahead of Hurricane Sandy in October 2012. The European Center for Medium-Range Weather Forecasting, based in Reading, England, had correctly predicted Sandy’s sharp left turn landfall six days before it slammed into the densely populated northeast US, causing more than $68 billion in damage. That forecast was two days earlier than US forecasters.
More data would help, but it must be used with caution
After the late Sandy forecast embarrassment, the US invested more than $44 million to improve its forecasts. Last year, The National Oceanic and Atmospheric Administration said it had tripled the computing capacity of its”supercomputers”—one in Virginia and one in Orlando—to increase the data analyzed, and to make more accurate predictions earlier.
That’s no guarantee of improvement in every forecast. Meteorologists can become overly reliant on computer models. If the original model is wrong, Izzi notes, subsequent calculations will be, too.
“It becomes a little bit more of an art than a science,” he says. “Unfortunately many of the equations are unsolvable.”
Lucky for the kids who get to stay home. For businesses, not so much.