You are on page 1of 20

Journal of Instrumentation

Lightning observations using broadband VHF interferometer and electric


field measurements
To cite this article: A. Chilingarian et al 2020 JINST 15 P07002

View the article online for updates and enhancements.

This content was downloaded from IP address 130.88.237.223 on 03/07/2020 at 18:37


Published by IOP Publishing for Sissa Medialab
Received: March 23, 2020
Revised: April 27, 2020
Accepted: May 19, 2020
Published: July 2, 2020

Lightning observations using broadband VHF


interferometer and electric field measurements

2020 JINST 15 P07002


A. Chilingarian,a,b,c M. Dolgonosov,c,d,1 A. Kiselyov,c Y. Khanikyantsa and
S. Soghomonyana
a Yerevan Physics Institute, Alikhanyan Brothers 2, Yerevan 0036, Armenia
b National Research Nuclear University MEPhI, Moscow Engineering Physics Institute,
Kashira Hwy, 31, Moscow 115409, Russia
c Space Research Institute of Russian Academy of Sciences,

Profsoyuznaya str., 84/32, Moscow 117997, Russia


d National Research University Higher School of Economics,

Myasnitskaya Ulitsa, 20, Moscow 101000, Russia

E-mail: mdolgonosov@hse.ru

Abstract: A broadband radio interferometer for locating lightning emissions was built and deployed
at the high-altitude research station on Mount Aragats (Armenia). The instrument operates in the
frequency range from 24 to 78 MHz and employs three identical broadband flat-plate antennas
which are located in the horizontal plane and form two orthogonal baselines of 13 m length. The
time differences of arrival of the signals at pairs of antennas determined from the peak of the
cross-correlation function are used to measure the azimuth and elevation angles of the radiation
source as a function of time. This paper describes the operation of the interferometer and reports
observations of lightning flashes complemented by synchronous measurements of fast wideband
electric field waveforms and changes in the near-surface electrostatic field. Combined analysis
of these data makes it possible to identify the type of lightning discharge. The capability of the
system to classify lightning types is demonstrated by example observations of an intracloud and a
cloud-to-ground lightning flashes.

Keywords: Interferometry; Avalanche-induced secondary effects; Detector modelling and simu-


lations II (electric fields, charge transport, multiplication and induction, pulse formation, electron
emission, etc)

1Corresponding author.

c 2020 IOP Publishing Ltd and Sissa Medialab


https://doi.org/10.1088/1748-0221/15/07/P07002
Contents

1 Introduction 1

2 Broadband VHF interferometer 2


2.1 Basic principle 2
2.2 Instrumentation 3
2.3 Data processing algorithm 5

2020 JINST 15 P07002


3 Measurements of electric fields 6
3.1 Wideband electric field 6
3.2 Near-surface electrostatic field 7
3.3 Other data sources and sign convention 8

4 Results and discussion 8


4.1 June 14, 2019, 17:58:17.473 UT (inverted IC) 8
4.2 June 18, 2019, 22:13:53.661 UT (-CG) 10
4.3 Angular uncertainty 14
4.4 Angular resolution 16

5 Summary 16

1 Introduction

Broadband radio interferometer systems for lightning location studies have been successfully de-
veloped for more than two decades. Broadband interferometer for lightning location was first
introduced by Shao et al. [1], and later it has been further developed and improved by other research
groups (e.g., [2]–[12]). With this technique, the very high frequency (VHF) radiation from light-
ning is received by an array of closely spaced antennas, typically 10–50 m apart. Measurements of
phase or time difference of arrival of the signals from pairs of antennas provide two-dimensional
direction of arrival of the lightning radiation as a function of time. Due to short baseline used in
the interferometric systems, the time differences of arrival of the signals are in the range of several
tens of nanoseconds, and therefore, these systems require a digitizing unit with high sampling rate.
Another important parameter of the digitizer is the memory depth, which determines the maximum
record length of the interferometer data. Due to technology limitations, first broadband interfer-
ometers were restricted to continuos recording only a few milliseconds of time series data. During
last decade, the availability of high-speed digitizers with deep onboard acquisition memory has
made it possible to record the interferometer data continuously over the duration of lightning flash.
Significant improvement of broadband lightning interferometer over similar instruments used in
the past, and a comprehensive analysis of its operations has been reported by Stock et al. [8]. In

–1–
addition to the broadband interferometry, a capability to measure simultaneously the polarization
state of the corresponding broadband VHF sources has been recently introduced by Shao et al. [10].
The high time resolution, on the order of 1 µs, and small angular uncertainty, less than 1◦ , has
made the broadband VHF interferometers helpful tools to study the progression of lightning flashes
in great detail.
In this paper we describe a broadband radio interferometric system deployed at the high-altitude
research station on Mount Aragats (Armenia), and report observations of lightning flashes detected
by the interferometer. We present combined analysis of interferometric data, synchronous measure-
ments of fast wideband electric field waveforms, and changes in the near-surface electrostatic field

2020 JINST 15 P07002


produced by lightning flashes. Based on this analysis, we demonstrate a capability of the system to
identify the type of lightning flash.

2 Broadband VHF interferometer

2.1 Basic principle


A short baseline system for lightning location using the VHF signals emitted by lightning discharge
was first suggested by Oetzel and Pierce [13]. The principle of the metod is illustrated in figure 1.

Figure 1. Principle of lightning location with a short baseline. VHF radiation is received by two antennas
which are distance d apart.

In a short baseline system, the baseline length d is small compared with the distance to the
source, thus the incoming signal is treated as a plane wave at the receiver locations. The difference
τ of time of arrival between a pair of horizontally spaced antennas is used to determine the arrival
angle of the plane-wave radiation. For two antennas separated by a distance of d, the arrival angle
α is given by
c · τ 
∆φ · λ

α = arccos = arccos (2.1)
d 2π · d
where λ is the wavelength of the radiation, ∆φ is the phase difference between the two signals, and
c is the speed of light in air.
To determine the direction of arrival of the signal for sources located in two spatial dimensions
second measurement is required which can be obtained from additional baseline. With three

–2–
antennas forming two orthogonal horizontal baselines, two time differences of arrivals, τ1 and τ2
measured for each of two baselines provide the direction of arrival (azimuth and elevation) of VHF
lightning radiation. For two orthogonal baselines 1 and 2 corresponding to east-west and north-
south directions, respectively, the azimuth (Az) and elevation angles (El) are given by the following
equations [8, 14, 15]

τ1
 
Az = arctan (2.2)
τ2
 q 
c
El = arccos τ + τ2 .
2 2 (2.3)
d 1

2020 JINST 15 P07002


Generalization of equations (2.2) and (2.3) for the case of nonorthogonal baselines with
arbitrary orientation has been reported by Stock et al. [8].

2.2 Instrumentation
The interferometer is installed at an altitude of 3200 m above sea level on Mount Aragats. The
instrument uses three identical circular flat-plate resistively coupled antennas of 30 cm diameter.
Three antennas are arranged to form two orthogonal baselines of 13 meter length, as shown in
figure 2.

Figure 2. Antenna configuration of the interferometer.

The block diagram of the interferometer is shown in figure 3. Each of three measuring channels
of the interferometer includes flat-plate antenna, coaxial lightning surge protector, amplifier, and
bandpass filter. The surge protector (Model AL6-NF-14-9, L-com Global Connectivity) features
wideband operation from DC to 5.8 GHz, and has a breakdown voltage of 90 V. The antenna
signal passes through the lightning protector and is fed to a broadband low-noise amplifier of 50
Ω impedance. The amplifier (Model HD25116, HD Commmunications Corp.) has a bandwidth
of 1–1000 MHz, a gain factor of 40 dB, and a noise figure of 3.3 dB. The bandpass filter (Model

–3–
KR Electronics 2804, KR Electronics Inc.) has a bandwidth from 24 to 82 MHz. Due to resistive
coupling of the antennas the output signals are proportional to the time derivative dE/dt of the
incident electric field E. A four-channel Picoscope 6403D digitizing oscilloscope is used to capture
the signals from the antennas. The maximum real-time sampling rate of the oscilloscope is 5 GS/s
(1.25 GS/s for four-channel operation), and the memory depth is 1 GS (250 MS per channel).
The amplitude resolution is 8-bit. The interferometer data presented in this study were digitized at
156.25 MS/s sampling rate (sample interval of 6.4 ns). With this sampling rate, the cut-off frequency
fc =78.125 MHz (Nyquist frequency) of the built-in low pass filter of the oscillosope determines the
higher boundary frequency of the recorded signal. The lower boundary of 24 MHz frequency is set
by the external bandpass filter.

2020 JINST 15 P07002


Figure 3. Block diagram of VHF interferometer.

The oscilloscope is triggered by the signal from a commercial MFJ-1022 active whip antenna
that covers a frequency range of 300 kHz to 200 MHz. The signal from the triggering antenna is
fed to one of four channels of the oscilloscope (Channel D). With each trigger, the record length
is 500 ms including 100 ms pretrigger time and 400 ms posttrigger time. The triggerout pulse of
the oscilloscope is relayed to the National Instruments (NI) myRIO board which produces the GPS
time stamp of the record. Main parameters of the interferometer are shown in table 1.

Table 1. Parameters of the interferometer.


Number of antennas 3
Baseline length 13 m
Bandwidth 54 MHz
Amplifier gain 40 dB
Sampling rate 156.25 MS/s
Sample interval 6.4 ns
Capture length 500 ms

–4–
2.3 Data processing algorithm
Our data processing approach is similar to that developed and described by Stock et al. [8]. The
processing algorithm is based on the well known generalized cross correlation technique, where
the time delay between two signals is determined from the position of the dominant peak of the
cross-correlation function of the signals. The cross correlator is implemented in the frequency
domain. First, the antenna signals are transformed by fast Fourier transform into frequency domain,
and then, the cross spectrum is calculated by multiplication of one signal by the complex conjugate
of the other. The inverse Fourier transform of the cross spectrum yields the cross-correlation
function in the time domain, where the position of the dominant peak indicates the time delay. The

2020 JINST 15 P07002


“generalized” cross-correlation adds a windowing function prior to the inverse Fourier transform.
The purpose of the windowing function is to improve the estimation of the time delay by suppressing
the contributions from secondary peaks of cross correlation. In this study, we use the smoothed
coherence transform (SCOT) [16] windowing function.
Each of three records of antenna signals contain 78.125 MS resulting from the digitization of
500-ms record sampled with 6.4 ns interval. During processing procedure, the records of entire
flash are divided into a short time duration windows, and for each window the cross-correlation
function of pairs of signals is computed. In our study, a 1.63 µs (256 samples) sliding window was
used for signal processing. To improve the detection efficiency and reduce the erorrs in location of
radiation sources, successive windows are overlapped by an adjustable amount. Usually we use a
75% overlap, that is sliding window step is 64 samples (192 out of 256 samples overlap). In order
to improve accuracy of the time delay estimation, the waveforms are upsampled by a factor of 16,
thus providing effective time resolution of 0.4 ns in the determination of time delay. The waveforms
are upsampled in the frequency domain, by zero-padding the two spectra used in determining the
cross spectrum.
All three baselines are used for the data processing, including the 18.4 m diagonal baseline of the
antenna array. For each window, time delays for the three baselines (AB, AC, BC) are computed, and
the source direction is determined by the average of the three baseline measurements. The window
is considered to contain a radiation source if the maximum amplitude of the cross-correlation in the
window exceeds the manually set threshold.
The interferometer data are, as a rule, contaminated by noise produced by a number of differents
sources, such as high-voltage overhead transmission lines, local power transients, reflections of
lightning emission, etc. To reduce the noise, the data are filtered by using several selection criteria
proposed by Stock et al. [8]. First one is the closure delay defined as τ123 = τ12 + τ23 − τ13 , where τi j
are the time delays between each of antenna pairs. If the detected emission is produced by a distant
point source, the closure delay should ideally be zero. In practice, it must be smaller than sampling
interval ∆t, that is τ123 /∆t ≤ 1. Second criterion is the standard deviation of source coordinates
obtained for three possible channel pair configurations, which should be very small. Another two
selection criteria used are the amplitude of the cross-correlation function, and the event multiplicity,
defined as the number of windows where the event has been found.

–5–
3 Measurements of electric fields

In addition to the VHF interferometer, we used a measurement system to record the lightning
wideband electric field waveforms, and a network of electric field mills to measure the electrostatic
field changes caused by nearby lightning flashes.

3.1 Wideband electric field


The techniques to measure electromagnetic signatures produced by lightning are well known (e.g.,
book by M. Uman, appendix C [17], and book by V.A. Rakov, chapter 7 [18] and references therein).
A detailed analysis of measurement system for recording of the lightning wideband electric field can

2020 JINST 15 P07002


be found in [19, 20]. We give below a brief overview of the measurement principle by considering
the wideband electric field (E) measurement system used in the present study and also in our earlier
studies [21, 22]. The block diagram of the measurement system is shown schematically in figure 4.
A sensor that is commonly used to measure the lightning vertical electric field is a metallic flat-plate
antenna.

Figure 4. Block diagram of the lightning wideband electric field.

According to the boundary conditions on the surface of a perfect conductor, the electric field
vector on the surface of metallic plate has only a vertical component. The charge Q induced by
electric field E on the antenna plate having a surface area S is given by

Q (t) = ε0 · S · E (t) (3.1)

where ε0 = 8.85 × 10−12 F/m is the permittivity of free space.


Any change in the vertical electric field E causes a change in the induced charge Q. The
resultant short-circuit current I is proportional to dE/dt:
dQ (t) dE (t)
I (t) = = ε0 · S · . (3.2)
dt dt
In order to measure E, it is necessary to use an integrating capacitor CCa , where Ca is the
capacitance of the antenna (typical value of Ca is ≈50–100 pF). The passive integrator (figure 4)
is formed by the integrating capacitor C and resistor R (which is the input resistance of the
oscilloscope). The resistor provides a discharge path for the capacitor C, so that the output voltage

–6–
in response to a step-function excitation decays exponentially with a time constant τ = RC. The value
of τ must be much larger than the variation time of the signal of interest. If frequency content of the
electric field E(t) satisfies the condition ω  1/RC, the voltage U at the output of the integrator is
directly proportional to the vertical component of the electric field and given by

1 t ε0 · S · E (t)

U= I (τ) dτ = . (3.3)
C 0 C
The wideband electric field measurement system installed at the Aragats station uses a 52 cm
diameter circular flat-plate antenna followed by a passive integrator with C=3 nF and R=1 MΩ.
Thus, the decay time constant of the integrator is RC= 3 ms. No voltage amplification was used,

2020 JINST 15 P07002


the output of the integrator was connected via a 60 cm double-shielded coaxial cable to a Picoscope
5244B digitizing oscilloscope. The capture length was 1 s including 200 ms pretrigger time and
800 ms posttrigger time. The sampling rate was 25 MS/s (sample interval of 40 ns), and the
amplitude resolution was 8 bit. The bandwidth of the electric field measuring system was from
≈ 50 Hz to 12.5 MHz. The lower boundary of ≈50 Hz is determined by the RC time constant of
3 ms of the integrator which is equivalent to a high-pass filter with cut-off frequency of 1/(2πRC).
The higher boundary of 12.5 MHz is determined by the upper frequency response of the Picoscope
5244B oscilloscope (Nyquist frequency for the sampling rate of 25 MS/s). The antenna calibration
for electric field amplitude is presently not available, so the amplitude of recorded waveforms is
given in voltage units of the oscilloscope. The two oscilloscopes (of the interferometer system and
wideband electric field system) were triggered by the same signal from MFJ-1022 antenna, thus,
the synchronization between the interferometer and wideband E field records was within 1 µs.

3.2 Near-surface electrostatic field


The measurement of electrostatic field change produced by lightning is described in a number of
books [17, 18, 23] and review papers, e.g., [24]. The slowly-varying electric fields (on timescales
of the order of seconds or more) are conventionally measured by field mills (also called electrostatic
fluxmeters or generating voltmeters). Field mills determine electric field strength by measuring
modulated, capacitively induced charges or currents on metal electrodes. The metallic sensing
plate is periodically exposed to and shielded from the electric field by a grounded, rotating shutter.
The charge induced on the sensing plate and the current between the plate and ground are both
proportional to the electric field E normal to the electrode. Thus, the field strength can be determined
by measuring the induced charge or current. If the signal is rectified by a phase-sensitive detector
(relative to the shutter rotation), the output signal shows both the magnitude and polarity of the
electric field.
In the present study, the near-surface electrostatic field changes caused by nearby lightning
flashes were recorded using a network of commercially available electric field mills (Model EFM-
100, Boltek Corporation), one of which is placed at the Aragats station, one at the Nor Amberd
station at a distance of 12.8 km from Aragats, and one at the Yerevan station, at a distance of 39.1 km
from Aragats. The sensitivity distance of EFM-100 is about 33 km, and the response time of the
instrument is 100 ms. The near-surface electrostatic field changes were recorded at a sampling
interval of 1 s. Electrostatic field data is displayed and graphed on a PC using the included software.
For complete electrical isolation, the field mill is connected to the PC using fiber optic cable.

–7–
3.3 Other data sources and sign convention
For lightning classification purposes, we also looked for time coincidences of lightning events
detected by our recording systems with those found in the database of the World Wide Lightning
Location Network (WWLLN). The network is operated by the WWLLN management team, lead
by Prof. Robert Holzworth of the University of Washington in Seattle. The WWLLN detects most
intense very low frequency (VLF, 3–30 kHz) signals from lightning which propagate efficiently
many thousands of kilometers in the waveguide formed by the Earth surface and the ionosphere.
The network has currently over 60 nodes around the globe. One of the WWLLN nodes is installed
in Yerevan Physics Institute.

2020 JINST 15 P07002


Throughout this paper, we use the atmospheric electricity sign convention, according to which
the downward directed electric field or field change vector is positive.

4 Results and discussion

The data presented in this study were acquired at the Aragats Space Environmental Center (ASEC)
during 2019. In this section, we examine the observational data for two lightning flashes of different
type, namely, inverted intracloud (IC) flash, and negative cloud-to-ground flash (-CG). For each of
two events we analyse results of processed interferometric measurements, synchronous measure-
ments of fast wideband electric field waveforms, and changes of the near-surface electrostatic field
produced by lightning. Based on this analysis we identify the type of lightning flash.

4.1 June 14, 2019, 17:58:17.473 UT (inverted IC)


Figure 5 shows oscilloscope traces of the interferometer signals from measuring and triggering
antennas produced by lightning flash that occurred at 17:58:17.473 UT on June 14, 2019.

Figure 5. Oscilloscope traces of signals from three measuring antennas A, B, and C (blue, red, and
green, respectively) and from the triggering antenna (black) produced by lightning flash that occurred at
17:58:17.473 UT on June 14, 2019.

–8–
2020 JINST 15 P07002
Figure 6. Image of lightning discharge in rectangular (left) and in polar (right) coordinates. Lightning was
registered at 17: 58: 17.473 UT, June 14, 2019. Color coding for both panels shows time on the same scale.

Figure 6 shows the image of the lightning flash in rectangular and polar coordinates. In
rectangular coordinates, the azimuth angle is plotted on the horizontal axis, and the elevation angle
is on the vertical axis. In polar coordinates, the cosine of the elevation angle is plotted along the
radius, zenith is in the center of the diagram, and horizon on the periphery. The azimuthal angle
is counted counterclockwise from the north (0◦ ), the angle 90◦ corresponds to the east. The color
coding in all diagrams indicates time. The time span of diagrams displayed by the colorbar in the
middle of the figure is determined by the time interval where solutions (radiation sources) were
found by the processing software. Obviously, this time span is less than or equal to the full record
length of 500 ms (from −100 ms to 400 ms), and depends on the criteria used for the filtering of the
observational data. For this particular case shown in figure 6, the solutions (radiation sources) were
found in time interval from −66 ms before the trigger to 229 ms after trigger.
More distinct images can be obtained for separate stages of lightning discharge, by selecting
corresponding time window in the visualization of processed data. An example is demonstrated in
figure 7, which shows a fragment of the same lightning flash (shown in figure 6) for a time interval
from 90 to 190 ms relative to the trigger. Development of this flash can be seen from animations
of the data which are available in the supporting information (supplementary data as video 1 and
video 2). As seen from figures 6 and 7, the minimum value of elevation angle is about 23–25◦
(ignoring few scattered points in the range of elevation angles from 10◦ to 23◦ ), that is, the lightning
does not strike the ground. Therefore, we can classify this lightning as a cloud discharge. This
classification is supported by the analysis of electrostatic filed changes measured by two field mills
at Aragats and at Nor Amberd stations, and by the examination of fast wideband E field record. In
this analysis presented below we use the methodology of lightning type classification developed in
our ealier studies [21, 22].
Electrostatic field changes produced by lightning flash that occurred at 17:58:17.473 UT on
June 14, 2019 are shown in figure 8. As seen from panels (a) and (b) in figure 8, the electrostatic

–9–
field changes detected by two field mills at Aragats and at Nor Amberd have opposite polarities. The
larger field change at Aragats is negative but at the Nor Amberd station it is positive; that is, polarity
reversal with distance is detected. Therefore, this lightning can be identified as an intracloud flash,
because the polarity reversal with distance is expected only when an elevated vertical dipole is
neutralized. The larger field change detected at Aragats corresponds to a closer distance and its
polarity is negative, which is indicative of inverted intracloud flash neutralizing a dipole whose
positive charge was below negative.
Identification of this event as a cloud flash is further supported by the fast wideband E field
record shown in figure 9. Examination of the record shows that it contains only short bipolar pulses
of microsecond and sub-microsecond duration and no signatures characteristic of return strokes are

2020 JINST 15 P07002


seen. Thus, the lightning is identified as an inverted intracloud (IC) flash.

Figure 7. Fragment of the images of lightning discharge shown in figure 5 for a time interval from 90 to
190 ms relative to the trigger. Color coding for both panels shows time on the same scale.

4.2 June 18, 2019, 22:13:53.661 UT (-CG)


Second event analysed in this study is a cloud-to-ground lightning flash detected by the interfer-
ometer. Figure 10 shows the images of the lightning flash in rectangular and polar coordinates for
time interval from 5 ms to 45 ms relative to trigger. As seen from figure 10, the minimum value
of the elevation angle is about 6–7◦ , that is the discharge propagates practically up do the horizon
and strikes the ground. Therefore, this lightning can be classified as a cloud-to-ground (CG) flash.
To validate this classification and determine the polarity of CG flash, we examine electrostatic filed
changes produced by this lightning, and synchronized records of wideband field and interferometer
antenna signal.
Electrostatic field changes produced by lightning flash that occurred at 22:13:53.661 UT on
June 18, 2019 are shown in figure 11. As seen from panels (a) and (b) in figure 11, the electrostatic
field changes detected by two field mills at Aragats and at Nor Amberd have same positive polarity.

– 10 –
2020 JINST 15 P07002
Figure 8. Electrostatic field changes recorded by the field mills at Aragats (a), and at Nor Amberd (b)
(separated by 12.8 km) for lightning flash that occurred at 17:58:17.473 UT on June 14, 2019.

The positive change of electrostatic field can be associated with negative cloud-to-ground flash
(-CG) that serves to reduce the main negative charge of the thundercloud.
Alternatively, it coud be a normal-polarity intracloud (IC) flash, which is shown not to be
the case below. We note, that polarity reversal with distance expected for the cloud flash was not
detected for this event.
Synchronized records of wideband field and interferometer antenna signal produced by the
lightning are shown in figure 12. The signal of the interferometer antenna shown in panel (a) in
figure 12 is proportional to the time derivative of electric filed dE/dt, whereas the signal of fast
wideband electric field in panel (b) is proportional to the electric field E. The records are shown for
a time interval which contains typical signatures of cloud-to-ground lightning. The record in panel
(b) contains four relatively wide pulses of positive polarity at 0.44 ms, 10.2 ms, 12.8 ms and 17.5 ms
after the trigger. The risetime of these pulses is about 10–15 µs, and the peak-to-zero fall time is 40–
60 µs. The positive polarity of these pulses is the same as the polarity of electrostatic field changes
shown in figure 11. We attribute these four pulses to return strokes of negative CG lightning. It
is worth noting that this lightning event has been also detected by the WWLLN network. It is
generally believed that the detection efficiency of WWLLN for CG flashes is significantly higher
than that for cloud flashes [25]. Therefore, if the lightning in question is detected by the WWLLN,
we assume that it is more likely to be a CG flash. Certainly, the WWLLN detection is only a minor
argument, and the basic criterion for the identification of lightning as a CG flash is the detection of
return stroke pulse(s) in the wideband electric field record. Thus, based on the analysis of availabale
data, we identify this lightning event as -CG.

– 11 –
2020 JINST 15 P07002

Figure 9. Fast wideband electric field record of lightning flash that occurred at 17:58:17.473 UT on June
14, 2019, shown on different time scales. Entire waveform is shown in panel (a), expanded waveforms are
shown in panels (b)–(f). The noise in (a) and (b) is caused by the 50 Hz electromagnetic interference.

– 12 –
2020 JINST 15 P07002
Figure 10. Fragments of the images of lightning that occurred at 22:13:53.661 UT on June 18, 2019 for a
time interval from 5 to 45 ms relative to the trigger. Color coding for both panels shows time on the same
scale.

Figure 11. Electrostatic field changes recorded by the field mills at Aragats (a), and at Nor Amberd (b)
for lightning flash that occurred at 22:13:53.661 UT on June 13 18, 2019. Both field changes have positive
polarity.

– 13 –
4.3 Angular uncertainty
In order to distinguish between CG and IC flashes in the examples analized above, we compared
the minimum values of elevation angles which were measured to be 6–7◦ and 23–25◦ for CG and
IC, respectively. The angular uncertainty of the interferometer is determined by the uncertainty of
the time delay estimation. To estimate the accuracy of the elevation angle measurement we use the
methodology proposed by [8]. The angular uncertainties of the azimuthal and the elevation angles
can be found from the error analysis of equations (2.2) and (2.3) introduced in section 2.1 and are
given by
c 1
σAz = στ

2020 JINST 15 P07002


(4.1)
d cos (El)
c 1
σEl = στ (4.2)
d sin (El)
where στ is the timing uncertainty (standard deviation of the time delay determination).
As seen from equations (4.1) and (4.2), the azimuthal uncertainty σAz is smallest on the
horizon (El=0), and increases towards the zenith, whereas the uncertainty of the elevation angle σEl
is smallest at the zenith (El=90◦ ) and gets worse towards the horizon.
In theory, the accuracy of any delay estimator is limited by the Cramér-Rao bound [26],
according to which the minimum standard deviation of the time delay estimation is given by
s
3 (1 + 2SNR) 1
στ ≥ 2 2
(4.3)
T f2 − f13
3

8π SNR

where T is the integration time, SNR is the linear signal-to-noise ratio, and f1 and f2 are lower and
higher boundary frequencies of the signal, respectively. Thus, the timing uncertainty and hence,
the angular uncertainty of the interferometer, depends on signal-to-noise ratio. For the parameters
of this study: f1 =24 MHz, f2 =78 MHz, T=1.63 µs (T is the width of sliding window used in the data
processing algorithm), the best possible timing uncertainty of the interferometer for signals with
an SNR of 10 dB is στ ≈ 0.1 ns. Table 2 shows the timing and elevation angle uncertainties of the
interferometer for a medium elevation angle of 45◦ and a small elevation of 5◦ near the horizon,
calculated for signals with various values of SNR.

Table 2. Timing and elevation uncertainties of the interferometer calculated for three different values of
signal-to-noise ratio (SNR) and for two elevation angles.
Elevation uncertainty
Timing uncertainty
SNR [dB] σEl [deg]
στ [ns]
El=45◦ El=5◦
3 0.25 0.47 3.8
5 0.19 0.36 2.9
10 0.1 0.19 1.5

As seen from table 2, the lower bound of theoretical angular uncertainty near the horizon (at
the source elevation of 5◦ ) for signals with SNR of 10 dB is 1.5◦ .

– 14 –
2020 JINST 15 P07002

Figure 12. Fragments of oscilloscope records of signals produced by lightning flash that occurred at
22:13:53.661 UT on June 18, 2019. (a) — signal from the interferometer antenna, (b) and (c) -fast wideband
electric field. Red arrows indicate positions of four return stroke (RS) pulses at 0.44 ms, 10.2 ms, 12.8 ms
and 17.5 ms after the trigger. The strongest return stroke pulse (RS1) is shown on an expanded time scale in
panel (c).

– 15 –
In practice, the actual timing uncertainty may be several times greater than the theoretical lower
bound.
With the 6.4 ns sampling period of the digitizer and upsampling of the waveforms by a factor
of 16 used in our data processing procedure (mentioned in section 2.3), the effective subsample
time resolution of 0.4 ns is obtained. With this time resolution considered as the actual timing
incertainty, the elevation uncertainty near the horizon for signals with typical SNR of 10 dB is
6◦ . Thus, the actual elevation uncertainty is a factor of 4 above the lower bound, but it is still
noticeably smaller than the difference of 17◦ between minimal elevation angles measured for CG
and IC flashes. Therefore, the interferometer observations allow us to reliably distinguish between
CGs and ICs.

2020 JINST 15 P07002


4.4 Angular resolution
In addition to the angular uncertainty, the interferometer also can be characterized by an angular
resolution, i.e., the minimum angular distance between two adjacent radiation sources that can be
discerned by the instrument. From the diffraction theory, the angular resolution can be estimated
as ∆ϕ ≈ λ/D, where for a broadband interferometer system λ is the shortest wavelength and
D is the length of longest baseline [8]. For our interferometer, with the shortest wavelength of
3.84 m corresponding to the higher boundary frequency of 78.125 MHz, and the longest baseline
of 18.4 m (the length of the hypotenuse BC in figure 2), the angular resolution is estimated to be
∆ϕ ≈ 0.21 rad ≈ 12◦ . Emissions from two sources the angular separation of which is smaller
than the angular resolution and which arrive at the interferometer simultaneously (within the
integration time of signal processing) will be indistinguishable. In practice, due to very short
integration time which is equal to the width of sliding window (1.63 µs in present case), the angular
resolution limitations are essential only for situations when emissions from two close sources arrive
simultaneously, within this short time interval.

5 Summary

A broadband radio interferometer for lightning studies was built and deployed at the Aragats research
station. The interferometer was successfully used for lightning detection during the spring, summer,
and fall campaigns of 2019. The obtained results show that interferometer observations of lightning
flashes complemented by synchronous measurements of fast wideband electric field waveforms and
changes of the near-surface electrostatic field produced by lightning make it possible to identify the
type of lightning flash. Certainly, not every lightning flash could be reliably classified using the
instrumentation and methodology described above. The statistics and accuracy of lightning type
classification will be presented in a follow-up study.
A number of improvements are possible for the interferometer, which include the use of longer
baseline, the choice of a better place for the antenna installation, and the use of higher amplitude
resolution and higher sampling rate of the digitizer. With longer baseline, the angular uncertainty
and angular resolution of the interferometer, which are inversely proportional to the baseline length,
can be improved. The antenna installation at a larger distance from buildings can reduce the noise in
the interferometer data produced by reflections. Higher amplitude resolution and higher sampling
rate of the digitizer can improve accuracy in the determination of angular coordinates of VHF

– 16 –
radiation sources. Further experimentation with data processing is required to reduce the noise in
the interferometer data.

Data availability. The data that support the results of this study are available via the multivariate
visualization software ADEI on the Web page of the Cosmic Ray Division (CRD) of the Yerevan
Physics Institute http://adei.crd.yerphi.am/adei/. Interferometer data files (.psdata) are available
at http://crd.yerphi.am/vhf_interferometr/VHF%20INTERFEROMETER%20DATA/2019/ and can
be opened by free software available from https://www.picotech.com/downloads.

2020 JINST 15 P07002


Acknowledgments

The authors would like to thank the staff of the Aragats Space Environmental Center (ASEC)
for the uninterruptible operation of Aragats research station facilities. The expedition to Aragats
high-altitude station was supported by the Armenian National Science and Education Fund grant
18T-1C042. This work was also supported in part by the by Russian Science Foundation grant 17-
12-01439P.

References

[1] X.M. Shao, D.N. Holden and C.T. Rhodes, Broad band radio interferometry for lightning
observations, Geophys. Res. Lett. 23 (1996) 1917.
[2] T. Ushio, Z.-I. Kawasaki, Y. Ohta and K. Matsuura, Broad band interferometric measurement of
rocket triggered lightning in Japan, Geophys. Res. Lett. 24 (1997) 2769.
[3] Z.-I. Kawasaki, R. Mardiana and T. Ushio, Broadband and narrowband RF interferometers for
lightning observations, Geophys. Res. Lett. 27 (2000) 3189.
[4] R. Mardiana and Z.-I. Kawasaki, Broadband radio interferometer utilizing a sequential triggering
technique for locating fast-moving electromagnetic sources emitted from lightning, IEEE Trans.
Instrum. Meas. 49 (2000) 376.
[5] S. Qiu, B.-H. Zhou, L.-H. Shi, W.-S. Dong, Y.-J. Zhang and T.-C. Gao, An improved method for
broadband interferometric lightning location using wavelet transforms, J. Geophys. Res. Atmos. 114
(2009) D18211.
[6] D. Cao, X. Qie, S. Duan, J. Yang and Y. Xuan, Observations of VHF source radiated by lightning
using short baseline technology, in proceedings of the 2010 Asia-Pacific International Symposium on
Electromagnetic Compatibility, Beijing, China, 12–16 April 2010, pp. 1162–1165.
[7] Z. Sun, X. Qie, M. Liu, D. Cao and D. Wang, Lightning VHF radiation location system based on
short-baseline TDOA technique — validation in rocket-triggered lightning, Atmos. Res. 129–130
(2013) 58.
[8] M.G. Stock et al., Continuous broadband digital interferometry of lightning using a generalized
cross-correlation algorithm, J. Geophys. Res. Atmos. 119 (2014) 3134.
[9] M. Akita, M.G. Stock, Z.-I. Kawasaki, P. Krehbiel, W. Rison and M. Stanley, Data processing
procedure using distribution of slopes of phase differences for broadband VHF interferometer, J.
Geophys. Res. Atmos. 119 (2014) 6085.

– 17 –
[10] X.-M. Shao et al., Broadband RF interferometric mapping and polarization (BIMAP) observations of
lightning discharges: Revealing new physics insights into breakdown processes, J. Geophys. Res.
Atmos. 123 (2018) 10,326.
[11] F. Lyu, S.A. Cummer, Z. Qin and M. Chen, Lightning initiation processes imaged with very high
frequency broadband interferometry, J. Geophys. Res. Atmos. 124 (2019) 2994.
[12] P. Puričer, P. Kovář and J. Mikeš, New accuracy testing of the lightning VHF interferometer by an
artificial intercloud pulse generator, IEEE Trans. Electromagn. Compat. 6 November 2019.
[13] G.N. Oetzel and E.T. Pierce, VHF technique for locating lightning, Radio Sci. 4 (1969) 199.
[14] C.O. Hayenga and J.W. Warwick, Two-dimensional interferometric positions of VHF lightning

2020 JINST 15 P07002


sources, J. Geophys. Res. Oceans 86 (1981) 7451.
[15] C.T. Rhodes, X.M. Shao, P.R. Krehbiel, R.J. Thomas and C.O. Hayenga, Observations of lightning
phenomena using radio interferometry, J. Geophys. Res. Atmos. 99 (1994) 13059.
[16] G.C. Carter, A.W. Nuttall and P.G. Cable, The smoothed coherence transform, Proc. IEEE 61 (1973)
1497.
[17] M.A. Uman, The Lightning Discharge, Dover Publications, Inc., Mineola New York U.S.A. (1987).
[18] V.A. Rakov, Fundamentals of Lightning, Cambridge University Press, Cambridge U.K. (2016).
[19] J. Jerauld, Properties of natural cloud-to-ground lightning inferred from multiple-station
measurements of close electric and magnetic fields and field derivatives, Ph.D. Thesis, University of
Florida, Gainesville Florida U.S.A. (2007) and online pdf version at
http://purl.fcla.edu/fcla/etd/UFE0021279.
[20] A. Nag, Characterization and modeling of lightning processes with emphasis on compact intracloud
discharges, Ph.D. Thesis, University of Florida, Gainesville Florida U.S.A. (2010).
[21] A. Chilingarian, Y. Khanikyants, E. Mareev, D. Pokhsraryan, V.A. Rakov and S. Soghomonyan, Types
of lightning discharges that abruptly terminate enhanced fluxes of energetic radiation and particles
observed at ground level, J. Geophys. Res. Atmos. 122 (2017) 7582.
[22] A. Chilingarian, Y. Khanikyants, V.A. Rakov and S. Soghomonyan, Termination of
thunderstorm-related bursts of energetic radiation and particles by inverted intracloud and hybrid
lightning discharges, Atmos. Res. 233 (2020) 104713.
[23] V. Cooray, Electromagnetic Fields of Lightning Flashes, in An Introduction to Lightning, chapter 9,
Springer (2015), pp. 135–165.
[24] P.E. Secker and J.N. Chubb, Instrumentation for electrostatic measurements, J. Electrostat. 16 (1984)
1.
[25] C.J. Rodger, J.B. Brundell and R.L. Dowden, Location accuracy of VLF World-Wide Lightning
Location (WWLL) network: Post-algorithm upgrade, Ann. Geophys. 23 (2005) 277.
[26] G. Carter, Coherence and time delay estimation, Proc. IEEE 75 (1987) 236.

– 18 –

You might also like