Localization and quantification of the acoustical power of lightning flashes

Lightning is a ubiquitous source of infrasound, and an essential climate variable. Acoustic measurements have been carried out by the CEA over the last ten years to characterize thunder within the framework of the HyMeX project. First, during the fall of 2012 in the south of France in the Cévennes region during the intensive measurement campaign (SOP1) and more recently, in the fall of 2018 in Corsica (France), as part of the EXAEDRE campaign. During both the SOP1 and EXAEDRE campaigns, mini-arrays (“AA” for “Acoustic Array”) of four microphones (respectively disposed on a 50m and a 30m-wide triangle) were used. Lightning information were available thanks to three kinds of electromagnetic detection systems. Firstly, classical Lightning Location Systems (LLS) measured the low frequency range (1-350 kHz), giving the flash emission time and location, as well as its peak current. Secondly, a network of 12 antennas, Lightning Mapping Array (LMA), detecting in the very high frequency range (60-66 MHz) was used. It measured the radiation from leaders and intracloud discharges, which occur mostly inside the thundercloud, providing the 3D location of these discharges. Thirdly, the Charge Moment Change (CMC) was provided by broadband Extremely Low Frequency (< 1.1 kHz) measurements.

Time delays between AA sensors inform on the direction of sound arrival, while the difference between emission time and sound arrival provides the source distance. Combining the two allows a geometrical reconstruction of individual lightning flashes, each viewed as a set of point sound sources. Co-localization of acoustic sources with in-cloud detections provided by the LMA and with ground impacts provided by the LLS shows the efficiency and precision of the method. The measured sound amplitude can also be back-propagated, compensating for absorption and density stratification. This allows to evaluate the acoustical power of each detected source, and then the total power of an individual flash.

In both campaigns, very heterogeneous geometrical distributions of source sound powers within a single flash are frequently observed. Most of the power is frequently located in only one portion of the lightning, most of the time in the return stroke, but also sometimes in the intracloud part. A few homogeneous cases are observed, especially in SOP1. The total acoustical power of the flashes turns out to be also extremely variable, extending over at least 4 orders of magnitude with a median value of 3 MW. It correlates quite good with the peak current or the CMC, and the nature of the correlation differs strongly with the category of lightning considered, either typical return strokes or very energetic positive flashes generating sprites. However, a high dispersion of the data is observed, so that it is not possible to correctly predict any electrical parameter using only the total acoustic power of an event, although a trend is statistically observed. This could be overcome by finding other variables to fully explain the relationship between acoustical and electrical parameters, and improving our propagation model to better account for acoustic variability.