Editor's note: This article originally was published March 4, 2009. We are re-posting it because thundersnow was observed in New York City during the storm that hit the East Coast in the past few days.

It's been more than 30 years—during the Blizzard of 1978 to be exact—since Neil Stuart saw "thundersnow," a weather phenomenon featuring the unusual combination of thunder, lightning and snow. The National Weather Service (NWS) meteorologist was 10 years old, living near Boston. The storm—which he says "is famous in meteorological circles" and influenced his career path—dumped 27 inches (67 centimeters) of snow on the ground over two days. The heaviest snow, however, came during a six-hour thundersnow storm that delivered one foot of snow over a six hour period.

Seeing thundersnow come down is "like watching a time-lapse movie of the snow building up, because it falls so quickly," Stuart says.

Thunder and lightning during a snowstorm is different from a run-of-the-mill snowstorm; it is extremely rare—fewer than 1 percent of observed snowstorms unleash thundersnow, according to a 1971 NSW study. But recorded observations of the phenomenon date back to 250 B.C., say ancient Chinese records translated in 1980 by atmospheric scientist Pao-Kuan Wang, now of the University of Wisconsin–Madison.

Today, researchers are interested in thundersnow for its predictive value. According to Patrick Market, an associate professor of atmospheric science at the University of Missouri, a 30-year study of snowfall found that when lightning is observed during a snowstorm, there is an 86 percent chance that at least 15 centimeters of snow will fall within 113 kilometers of the flash. Researchers are trying to determine the combo of atmospheric conditions required to create thundersnow to help them better predict heavy snowfall—which they define as at least 20 centimeters falling at a rate of 7.5 to 10 centimeters per hour—and issue warnings about hazardous weather before it hits, giving people time to prepare, take cover and get off the road. (The National Highway Traffic Safety Administration estimates that more than 800 deaths were caused by snow-related traffic accidents in 2007.)

By the time the lightning flashes during a thundersnow-storm, it is often already too late to prepare local residents for the whiteout on the way. "If we're talking about the observation of thundersnow," Market says, "the predictive value is on the order of minutes to hours."

Thundersnow-storms share some characteristics with summer thunderstorms. In both, a region of relatively warm air causes moisture to condense into clouds. A temperature gradient then forms with colder air farther up and warmer air closer to Earth's surface. If the relatively warm air begins to rise, the turbulence causes some water molecules to lose electrons and others to gain them, forming charges within the atmosphere that lead to electrification (discharged as lightning) and a sudden heating and expansion of the air. Thundersnow is unique, scientists believe, because due to the subzero temperatures, interactions between supercooled liquid water, ice crystals and larger ice particles can also generate lightning. In both types of storms, thunder results from the sound waves created by the rapid cooling and contraction of the air superheated by the lightning.

In the U.S. thundersnow is most likely to form in mountainous regions like the Rockies (thanks to warm air pockets caused by sudden changes in elevation) as well as in the vicinity of comparatively warm and large bodies of water such as the Great Lakes. Snow requires a cold environment, adequate moisture to form clouds, and rising air; thundersnow makes an appearance when a fourth ingredient is added: thermal instability, which is created by the addition of relatively warm air. (Market estimates that temperatures need to get cooler by at least four degrees Celsius per 1.6 kilometers of altitude as warm air travels upward to create the needed turbulence. Scott Steiger, an assistant meteorology professor at the State University of New York at Oswego, recently discovered that there are about six thundersnow storms a year in the lower Great Lakes (Erie and Ontario) region, most of them in November and December.

"These storms don't move, so they can dump up to seven feet [two meters] of snow in one day," he says. "They are very intense snowstorms, but they are very local."

When thundersnow occurs away from mountains and lakes, its heat sources aren't found near the ground but rather at altitudes upward of 10,000 feet (3,000 meters). It is these less-frequent occurrences—in more populous areas in the Great Plains and the Northeastern U.S., as during the 1978 blizzard—where this type of storm has the greatest potential to cause damage. During a March 1 thundersnow storm that covered parts of Georgia, South Carolina and North Carolina, five to to 7.6 centimeters fell per hour, an extremely rare occurrence in that part of the country, Stuart says. Hartsfield-Jackson Atlanta International Airport reported visibility of only about 402 meters for more than an hour around noontime that day, and there were power outages in some areas of northern Georgia due to the heavy, wet snow, he adds.

Market last month joined a team of storm-chasing University of Illinois at Urbana–Champaign researchers using various radars to examine what takes place inside storm clouds to cause snowfall. The team is surveying atmospheric conditions in several locations in Indiana, Illinois and Wisconsin. A field mill, a device that measures electric fields near the ground, will be used to determine whether there is an accumulation of charged ice particles in the clouds above. The team next year plans to fly into snowstorms in NWS planes and drop parcels containing thermometers, barometers and other devices that, like weather balloons, will measure temperature on their way down. If the team encounters thundersnow during its study, it may be able to confirm the conditions needed to produce it, making such icy tempests easier to forecast.

"With some lead time, [be it] hours or even a day or two," Stuart says, "we can see a big storm and predict which areas will see extreme snowfall."