The analysis of Very Low Frequency (VLF, 3-30 kHz) radio back-scattering can be used to measure the impact of lightning on the D region of the ionosphere (60-90 km). Early/fast events are prompt and rapid changes to the D-region ionosphere associated with certain lightning flashes, causing heating, ionization, and attachment. Previous work has observed the behavior of early/fast events and their connection to specific types of lightning flashes through VLF remote sensing and lightning geolocation, but the unique nature of each event makes it difficult to broadly infer the interactions between lightning and the ionosphere using a small number of case studies.
We assembled a massive database of VLF amplitude samples for cases when high intensity lightning occurs near a transmitter-receiver path. We constructed an artificial neural network to detect and label early/fast events. With a large volume of events compiled, we charted detailed statistics of event occurrences and behavior.
We find a correlation between lightning current magnitude and event likelihood, as well as inverse correlation between event likelihood and distance to transmitter-receiver path. We further confirm the asymmetry of the peak current trends, with positive-current strokes being significantly more likely to produce an event. We find that increased distance of the lightning to the transmitter, and to a lesser extent to the receiver, decreases the probability of an ionospheric disturbance. We find that recovery time is largely not a function of the peak current. We do not find evidence that long-recovery events are a distinct class of Early/Fast events.