Physicists in Japan have made the best study yet of the gamma rays that are produced in the minutes leading up to a lightning flash. In addition, the team also observed for the first time emissions that ended abruptly less than a second before the exact moment the flash occurs. The finding provides important information about the relationship between the mysterious atmospheric accelerators that produce the gamma rays and the lightning that we see in the sky.

Physicists have known for some time that gamma rays are sometimes produced when lightning strikes. Indeed, gamma-ray pulses from thunderclouds that vary in length from sub-millisecond to several minutes have been detected for the last 30 years. Most researchers agree that there are two types of bursts: very short, higher-energy bursts that coincide with lightning; and longer, lower-energy pulses that are sometimes not associated with a specific lightning event. While all of these bursts are thought to be created when charged particles are accelerated by the huge electric fields that build up in a thundercloud, the exact mechanism – or mechanisms – that produce them remains a mystery.

In this latest study, Harufumi Tsuchiya of the RIKEN High-energy Astrophysics Laboratory and colleagues at several other Japanese institutes looked at data collected in 2010 by the Gamma-Ray Observation of Winter THunderclouds (GROWTH) experiment at the Kashiwazaki-Kariwa nuclear power plant. The experiment includes several different gamma-ray detectors that are used in tandem with plastic detectors – the latter ensuring that charged particles such as muons are not mistaken for gamma rays. The system detected gamma rays at energies between 40 keV and 30 MeV.

One stormy night

At about 9:30 p.m. on the evening of 30 December 2010, the team noticed an increase in the rate at which gamma rays were being detected by the experiment. Over the next three minutes the rate increased in a way that is consistent with previous observations of prolonged gamma-ray emissions from thunderclouds. But then in the space of 800 ms, just as a lightning event was recorded by the experiment's optical detector, the detection rate dropped back to the background level.

Unless it was an incredible coincidence that the burst ended a second before a flash of lightning occurred, it appears that the two events are linked. Indeed, a check of meteorological records shows that there were no other lightning strikes within 5 km of the experiment at around the same time. The team also saw an increase in the average energy of the gamma rays as the pulse evolved with time. For example, an excess of photons with energies greater than 10 MeV only appeared about 2 min into the pulse.

Small acceleration region

As the researchers used several detectors, they were also able to work out where the gamma rays were being produced, finding that photons with energies greater than 10 MeV were created in a region stretching across about 180 m in the thundercloud. This suggests that the gamma rays are produced in a relatively small section of the much larger cloud. Furthermore, the 800 ms delay between the end of the gamma-ray pulse and the lightning flash suggests that the lightning is initiated some distance away from the acceleration – although the process that connects the two is still unknown.

The measurements are described in Physical Review Letters.