Accurate and efficient long-range lightning geo-location using a VLF radio atmospheric waveform bank

Abstract

On average there are ∼50 lightning flashes worldwide every second, with activity varying by region and season. Many systems currently exist that detect and locate lightning flashes for a broad range of commercial and scientific applications, including air traffic control, insurance claims, climate modeling, and the investigation of secondary atmospheric and magnetospheric electrical phenomena. These lightning detection systems have varying degrees of coverage area and location accuracy. Commercial ground-based systems that excel at locating return strokes in cloud-to-ground lightning use radio detection in the LF (30-300 kHz) band to provide very accurate location data, with a typical accuracy of ∼0.5 km, but they require a dense network of receivers separated by ∼400 km and are therefore primarily limited to monitoring the land areas within the network.

In addition to radiating in the LF band, each lightning strike generates a broadband electromagnetic pulse containing frequencies from a few Hz through to the optical band with a peak component at VLF (3-30 kHz). Radio waves at VLF propagate through the waveguide formed by the Earth and the ionosphere with relatively low attenuation (∼3 dB per 1000 km), enabling the detection of these pulses, called radio atmospherics, at great distances from the lightning strike. Several existing networks utilize this efficient guiding to geo-locate lightning strikes often at distances greater than 5000 km from a given receiver. However, the Earth-ionosphere waveguide also presents a complex and time-varying channel that heavily disperses the pulse as it propagates away from the strike location. These networks fail to adequately address the path-dependence of the received impulse and suffer a lower location accuracy as a result (∼20 km).

A new technique of long-range global lightning location is presented that both takes advantage of the efficient propagation at VLF and addresses the path-dependence of the propagation channel. This new technique catalogs the dominant variation in expected received waveforms to form a set of waveform banks, which are then used to estimate the propagation distance and identify features on each waveform that allow for a more accurate determination of the arrival time. Using three stations in a trial network, this new technique is used to demonstrate an accuracy of 1-4 km, depending on network geometry and the time of day. Furthermore, this technique provides an estimate of the peak current and polarity in the lightning channel, parameters that existing long-range networks do not measure using VLF radio atmospherics.

The propagation distance estimated at each receiver, together with an arrival azimuth measurement, enables accurate geo-location using as few as three sensors. The redundancy offered by this range and azimuth information mitigates the complexity involved with correlating radio atmospherics from multiple sensors and enables a high detection efficiency. An overall stroke detection efficiency between ∼40-60% is estimated by correlating individual lightning stroke events to data from a commercial LF reference network. There are a significant number of additional events reported by the trial network that do not time-correlate with data from the commercial network, but the tight spatial clustering of lightning strokes between the two networks suggest that many of the unmatched events are not spurious but rather may correspond to weak cloud-to-ground strokes or cloud flashes.