Long-Range Magnetic Studies Explain How Delayed "Sprites" Get Their Energy

Their studies show that previously undocumented strong cloud-to-ground electrical currents can persist between the first and the follow-up discharges, maintaining enough energy flow for the second burst to actually trigger the sprite.

Sprites are faint, colorful and exceedingly brief flashes that are now known to erupt high in Earth's atmosphere in a region just below the ionosphere, which begins at an altitude of about 50 miles. While previously sensed by keen-eyed observers, such as airline pilots, they only have been scientifically documented in the past decade.

Previous research, some of it by Steven Cummer, a Duke assistant professor of electrical and computer engineering, has linked sprite development to the extra-powerful thunderstorms observed in places like the Midwestern United States.

Earlier work also ties each sprite to an unusually strong cloud-to-ground lightning bolt, followed by a second discharge at heights of 25-50 miles.

Cummer, who uses an antenna in a remote forest location near Duke to study such Midwestern storms, first documented that the energy in the second discharge, rather than the initial bolt, is time-aligned in every detail with the waxing and waning of a sprite.

In a new report, published in the Thursday issue of the journal Geophysical Research Letters, Cummer and Martin Fuellekrug of Frankfurt University's Institute for Meteorology and Geophysics answer a question that had continued to perplex sprite investigators in a minority of cases. About 75 percent of the time, sprites evolve quickly in a "very causal" way, Cummer said in an interview. That explainable chain of events begins when an exceptional lightning burst builds up a high-altitude electric field sufficient to spawn the second spark that turns into a sprite.

"But with that remaining 25 percent there can be a really long delay from any lightning stroke to when the sprite starts, from longer than 10 to as long as 200 milliseconds (thousandths of a second), which is longer than most lightning processes happen," he said.

In such cases, Cummer's antenna studies – which analyze low-frequency electromagnetic returns characteristic of lightning strikes – would show the initial bolt and the second discharge as peaks on a graph separated by an unaccountable gap in activity. "It was a mystery what was connecting the two," he noted.

"It had been known, independent of sprites, that some lightning discharges have what are called ‘long continuing currents' that can last 100 milliseconds or longer," he said. "But most measurements of that current are much smaller than the lightning peak current, just not enough current to make anything like a sprite happen.

"There were suspicions that something like these long continuing currents were acting in some sprite-producing lightning. But those would need to be at least 10 times, and probably more like 50 times, bigger-than-ordinary continuing currents. And nobody had observed continuing currents long enough and strong enough to make these delayed sprites."

Cummer found signs of the missing currents by teaming up with Fuellekrug, who works with magnetic field sensors exceptionally sensitive to ultra-low electromagnetic frequencies. Stationing those sensors in the summer of 1998 at Santa Cruz, Calif., Soccoro, N.M., and Saskatoon, Saskatchewan, Fuellekrug focused on three different cases – in Michigan, Minnesota and Oklahoma – where high-altitude sprites followed lightning strikes below by more than 40 milliseconds. The lightning and sprite events were linked by their timing and locations, the first being logged by the National Lightning Detection Network, while the sprites were video-imaged by University of Alaska researchers.

Applying mathematical modeling analysis to Fuellekrug's much more sensitive measurements, Cummer found continuing cloud-to-ground currents in one event that varied from about 4,000 to 7,000 amperes over a period of about 150 milliseconds. "That number is extremely big," he said. "Most measurements of continuing currents like this in less spectacular lightning are on the order of 100 to 200 amps."

His analysis showed 10,000 amperes of continuing current flowed between lightning and sprite discharge during another event. "In an ordinary lightning discharge, you may have a peak current of 10,000 to 20,000 amps," Cummer noted, but for a much shorter time. "These continuing currents are approaching the peak currents in ordinary lightning, but we're talking about durations that are more than 100 times longer."

The difference between a large and small interval in this study may still seem like an indistinguishable instant. "You can probably have a sense of 100 milliseconds, which is 0.1 seconds, by looking at a stopwatch. But that tenth of a second is still pretty short," Cummer said.

"The question is: does this always happen in the case of long-delayed sprites? We've only sampled three. Beyond the sprite implications, these measurements raise the ceiling of just how big continuing lightning currents can get."

Sprites that erupt so long after an initial megabolt of lightning "aren't necessarily any more energetic than the ones that come soon after," Cummer said.

"There's no question that these kinds of discharges are the ones that can start forest fires. If you pump that much current into a tree for that long, you give it so much time to heat up you can't help but start a fire."