A survey of ELF and VLF research on lightning-ionosphere interactions and causative discharges

TL;DR: A review of the development of ELF and VLF measurements, both from a historical point of view and from the view of their relationship to optical and other observations of ionospheric effects of lightning discharges is provided in this paper.

Abstract: [1] Extremely low frequency (ELF) and very low frequency (VLF) observations have formed the cornerstone of measurement and interpretation of effects of lightning discharges on the overlying upper atmospheric regions, as well as near‐Earth space. ELF (0.3–3 kHz) and VLF (3–30 kHz) wave energy released by lightning discharges is often the agent of modification of the lower ionospheric medium that results in the conductivity changes and the excitation of optical emissions that constitute transient luminous events (TLEs). In addition, the resultant ionospheric changes are best (and often uniquely) observable as perturbations of subionospherically propagating VLF signals. In fact, some of the earliest evidence for direct disturbances of the lower ionosphere in association with lightning discharges was obtained in the course of the study of such VLF perturbations. Measurements of the detailed ELF and VLF waveforms of parent lightning discharges that produce TLEs and terrestrial gamma ray flashes (TGFs) have also been very fruitful, often revealing properties of such discharges that maximize ionospheric effects, such as generation of intense electromagnetic pulses (EMPs) or removal of large quantities of charge. In this paper, we provide a review of the development of ELF and VLF measurements, both from a historical point of view and from the point of view of their relationship to optical and other observations of ionospheric effects of lightning discharges.

Summary

1 Introduction

• In 2015, there were reports that those suspected of terrorism and other criminal activity were being targeted and killed by Egyptian police during security raids.
• While these approaches have helped to reduce the measurement problems caused by the weaknesses present in their information sources, those weaknesses remain.
• Over the years, as the authors have discussed existing human rights data with human rights advocates and researchers in human rights non-governmental organisations (HROs) around the world, they have heard time after time about the problems that come with relying on public reports for the purposes of measurement.
• This means that many known human rights violations go unreported.

2 What do existing measures of civil and political rights

• According to Goldstein (1986), anyone that attempts to generate quantitative data on human rights will face challenges associated with definitions, data reliability, and data interpretation.
• With regard to definitions, most projects have decided to hew closely to the definitions of various rights found in international human rights treaties, often aided by the various treaty bodies overseeing those documents, and on this front, HRMI is no exception.

2.1 Existing measures of civil and political rights

• There are several existing measures of respect for civil and political rights, often particularly focusing on the subset of those rights known as “physical integrity rights.”.
• Each of these datasets depends on content analyses of annual reports from the US State Department, Amnesty International, and, in the case of PTS, Human Rights Watch.
• Two more recent projects have produced quantitative scales that focus specifically on torture and are also grounded in international law.
• The political participation scale is created in a similar manner, except the components are aggregated by taking their average instead of using a measurement model.
• The survey the authors administer explicitly defines the rights under analysis with references to relevant international treaties and conventions, including the International Covenant on Civil and Political Rights , the Convention against Torture (CAT), 4 Hathaway relies on the Convention Against Torture (CAT) and several regional treaties for her definition of torture.

2.2 Problems of information

• The problem of obtaining reliable, unbiased, and comprehensive information is perhaps the most serious impediment to the collection of quantitative civil and political rights data.
• This focus on maintaining the effectiveness and credibility of the organisation means that HROs are unlikely to report on events that did not happen; however, it also means that many abuses go unreported (Hill, Moore, and Mukherjee 2013).
• As such, some have suggested that regression analyses utilising these potentially biased data should use some statistical method for accounting for that bias (e.g. Bagozzi, et al., 2015; Conrad, Hill, and Moore, 2014).
• The Varieties of Democracy Project (V-Dem) has attempted to sidestep these problems of information by turning to another source of information: experts on the countries being discussed (Coppedge et al., 2017).

2.3 Problems of interpretation

• Given the many problems of information laid out above, it is unsurprising that the interpretation of the limited information to which previous projects have had access has also faced huge hurdles.
• While these dimensions of abuse have long been recognised, every previous project aimed at collecting cross-nationally comparable civil and political rights data has failed to fully capture at least one of these dimensions.
• Similar to PTS, Fariss (2014) produces a single score for all physical integrity rights, and in a method similar to CIRI, V-Dem provides very little information on range (Coppedge et al., 2017).
• In their efforts to do this, the authors particularly intend to (1) use better sources of information than were previously available, (2) provide transparent indicators of uncertainty, and (3) measure the full dimensionality of civil and political rights abuse.

3 HRMI’s approach to civil and political rights

• Further, the authors seek to create measures for every country in the world in a way that ensures cross-national comparability, while remaining transparent in the means by which those measures are created.
• As described above, the authors particularly wanted to improve on 1. the quality of information, 2. the transparency of uncertainty, and 3.
• The authors have attempted to answer these challenges by 1. directly collecting information from human rights researchers and practitioners that are gathering information and monitoring human rights issues in each country, 2. using statistical methods that allow us to accurately and honestly report their uncertainty with regard to the intensity of abuse, and 3.
• The authors describe their pilot approach to collecting civil and political rights data, beginning with a discussion of the pilot version of the HRMI Civil and Political Rights expert survey, followed with a more detailed description of the model used to obtain the intensity score for each right measured.

4 The pilot HRMI civil and political rights expert survey

• In order to directly collect information on civil and political rights performance in countries around the world.
• Torture and ill-treatment may be committed for any specific purpose, including (but not limited to) attempts to obtain information or confessions, punishment for suspected or committed acts, intimidation, coercion, and discrimination.
• Those engaged in or suspected of non-violent political activity (e.g. protesters, journalists, activists).
• Beyond each of the sections focused on a particular right, the authors also include sections focused on asking their respondents to score the intensity of three hypothetical countries on their respect for the rights under consideration.

4.1 Selection of pilot countries and expert survey respondents

• A significant benefit of their approach to measuring civil and political rights is the ability to avoid some of the biases that exist in the public documentation of abuses of these rights, by collecting information directly from experts on the human rights situation in each country being studied.
• In the pilot study, the authors focused primarily on human rights practitioners who are directly monitoring the civil and political rights situation in each country.
• As such, the authors do not intend to rely on academics as respondents in most cases, as they are rarely involved in the collection of primary information and tend to rely more heavily on secondary sources.
• Thirteen countries were nominated, and the authors selected all 13 for inclusion in the pilot, as together they provided significant diversity in government type, country size, level of development, geographic location, and many other factors.
• By the end of the process, the authors had identified between 17 and 43 potential survey respondents per country, each of whom was sent a single-use survey link, to ensure that the survey link was not shared with unintended respondents.

4.2 Producing intensity scores: model description

• The simplest way to combine expert survey responses on the intensity question into a single score for each country would be to report the average of the survey responses for a given country.
• This allows for the fact that survey respondents may respond differently to the same objective conditions.
• Respondents with βs closer to 0 place countries with different human rights performances relatively close together on the scale, while those with more positive βs place countries with different performances relatively far apart on the scale.
• These “anchoring vignettes” combined with the Bayesian factor model described above, allow us to correct for any potential differences in how experts view the underlying scales in their survey.
• The resulting re-scaled distributions have means that ranged from approximately 2.7 up to around 10.1, with standard deviations ranging from 0.03 to 1.44.

5.1 HRMI pilot civil and political rights indicators

• In Figure 1 and Figure 2, mean scores for each country are presented as dots.
• There is only a 0.03 probability that Angola has better practices on the right to be free 26 from torture than does Australia; similarly, the probability that Fiji, New Zealand, and United Kingdom have worse torture practices than Angola are all practically indistinguishable from 0.
• As mentioned above, their survey did not only collect information from their expert respondents on the intensity of the state’s respect for these rights.
• In particular, their respondents stated that those especially vulnerable to torture and ill-treatment by government agents in Australia included: Detained asylum seekers, refugees, and immigrants, including children, and especially those held in offshore facilities on Manus Island and Nauru.

5.2 Comparison with V-Dem

• As noted above, the V-Dem project contains several indicators of civil and political rights and employs a methodology similar to ours.
• Even in that case, it would appear that VDem uses a slightly more constrained definition than the authors do, as they limit torture to acts committed with the aim to “extract information or intimidate victims, who are in a state of incarceration” (Coppedge et al., 2017, 221), whereas they allow for torture to be for “any specific purpose.”.
• Nevertheless, even in the presence of this differences in definition, the authors find that their measures in comparable thematic areas correlate highly and positively.
• While both definitions would appear to focus purely on extra-legal killings, it is possible that, particularly in the case of Saudi Arabia, respondents to the V-Dem survey considered Saudi Arabia’s use of the legalised death penalty against political opponents as something that should lower Saudi Arabia’s score on political killings.
• Still, while interesting, more study would be needed to determine why the difference exists.

6 Conclusion

• The HRMI civil and political rights pilot has demonstrated the benefits of collecting information on the full scope, intensity, and range of government respect for civil and political rights directly from human rights experts in countries around the world.
• Further, the statistical methods the authors use to convert this information into quantitative metrics allow us to be honest about uncertainty, and permit sensible cross-country comparisons.
• This work represents a significant advance over existing human rights data projects and the authors plan to extend coverage to a wider sample of countries as soon as possible.
• To accomplish these goals, the authors will continue to need help.

Figures

Figure 4. (left) LEP events and (right) early events contrasted. LEP events are displaced poleward from the causative discharge, are delayed up to 1–2 s, and are most commonly negative DA. Adapted from Peter [2007]. Early events, on the other hand, occur directly above the causative discharge and affect a much smaller region; they have <100 ms delay from the causative CG (between the red dashed lines) and are most commonly positive DA.

Figure 11. Gamma ray flare event. From Inan et al. [1999]. (a) VLF great circle paths from the NPM, NLK, and NAA transmitters to the VLF receiver at Palmer Station. The shading shows the region in daylight during the event. (b) Gamma ray counts on the RHESSI satellite. (c‐f) VLF amplitude and phase narrowband signals observed at Palmer. Figures 11e and 11f are zoomed out in time to show the full hour‐long recovery signature.

Figure 1. Physical mechanism of LEP. (1) Lightning injects VLF energy into the Earth‐ionosphere waveguide and into the magnetosphere. (2) VLF energy propagates as whistler mode waves in the plasmasphere. (3) Whistler waves interact in gyroresonance with relativistic radiation belt electrons, scattering them in energy and pitch angle. (4) Relativistic electrons scattered into the loss cone precipitate in the upper atmosphere through collisions with neutrals. (5) Secondary ionization is detected by subionospheric VLF probing techniques. Adapted from Lauben et al. [2001].

Figure 5. Examples of sprites and simultaneous early VLF events. From Marshall and Inan [2006].

Figure 9. Strong ELF radiation from sprite‐producing lightning. The ELF (<1 kHz) amplitudes of (left) typical and (right) sprite‐producing lightning vary by more than a factor of 35, while the peak currents vary only by a factor of 7. This reflects the relatively much larger total charge moment change in the sprite‐producing stroke compared to the typical stroke. Adapted from Cummer and Lyons [2005].

Figure 10. Example of a sferic burst. From Marshall et al. [2007]. (a) Map showing the location of the VLF receiver at Yucca Ridge (YR) and the Lightning Mapping Array (LMA), as well as locations of LMA‐detected pulses, colored in time. (b) Broadband VLF data, exhibiting strong burst activity over most of this 10 s sample. (c) LMA pulses versus altitude. (d) NLDN‐detected CG strokes.

Figure 6. Contrasting examples of early/fast and early/ slow events. From Haldoupis et al. [2006].

Figure 7. (top) Long recovery and (bottom) step‐like early VLF events. From Cotts and Inan [2007]. In the LR event, note that a long recovery in amplitude can sometimes occur simultaneously with a standard ∼100 s recovery in phase. In the step‐like event, the amplitude shows no discernible recovery, but the phase recoveries in a long 6 min.

Figure 2. Example LEP events on 28 March 2001. Taken from Peter [2007]. (top left) Map showing the location of the causative lightning discharges and the HAIL great circle paths. (bottom left) Data for an event at 0709:48 UT. (right) Data for another event at 0642:01 UT.

Figure 8. (a) Elve observed by ISUAL and (b) simultaneous early/fast event observed on the HWV signal at Crete. From Mika et al. [2006]. (c) Optical emissions in N2 1P band system and (d) electron density changes predicted for an E100 = 30 V/m (100 kA) CG discharge. From Marshall et al. [2010].

Figure 3. Example of the first early/fast events observed by Armstrong [1983]. (left) The geometry of the problem, where a storm just south of the NSS transmitter affects the path of that transmitter to Palmer Station, Antarctica. (top right) Palmer broadband VLF data in spectrogram form. A lightning‐generated sferic is observed at 0 s, and an associated sferic is detected 1 s later. (bottom right) The narrowband data from Palmer, exhibiting an early/fast event with zero delay from the causative sferic.

Figure 12. An example TGF with VLF sferic association, modified from Inan et al. [2006], including corrected RHESSI timing from Grefenstette et al. [2009]. See Cohen et al. [2010] for a modification of Figure 12. (a) Map showing the RHESSI footprint and propagation directions to Palmer Station. (b) RHESSI gamma ray counts. (c) VLF broadband sferic and direction‐finding results. (d) Histogram of all sferics over a 30 min period.

Full text

Click
Here
for
Full
A
rticl e
A survey of ELF and VLF research on lightning‐ionosphere
interactions and causative discharges
U. S. Inan,
1,2
S. A. Cummer,
3
and R. A. Marshall
1,4
Received 14 August 2009; revised 20 November 2009; accepted 6 January 2010; published 18 June 2010.
[1] Extremely low frequency (ELF) and very low frequency (VLF) observations have
formed the cornerstone of measurement and interpretation of effects of lightning
discharges on the overlying upper atmospheric regions, as well as near‐Earth space. ELF
(0.3–3 kHz) and VLF (3–30 kHz) wave energy released by lightning discharges is
often the agent of modification of the lower ionospheric medium that results in the
conductivity changes and the excitation of optical emissions that constitute transient
luminous events (TLEs). In addition, the resultant ionospheric changes are best (and often
uniquely) observable as perturbations of subionospherically propagating VLF signals.
In fact, some of the earliest evidence for direct disturbances of the lower ionosphere in
association with lightning discharges was obtained in the course of the study of such
VLF perturbations. Measurements of the detailed ELF and VLF waveforms of parent
lightning discharges that produce TLEs and terrestrial gamma ray flashes (TGFs) have also
been very fruitful, often revealing properties of such discharges that maximize ionospheric
effects, such as generation of intense electromagnetic pulses (EMPs) or removal of
large quantities of charge. In this paper, we provide a review of the development of ELF
and VLF measurements, both from a historical point of view and from the point of view
of their relationship to optical and other observations of ionospheric effects of lightning
discharges.
Citation: Inan, U. S., S. A. Cummer, and R. A. Marshall (2010), A survey of ELF and VLF research on lightning‐ionosphere
interactions and causative discharges, J. Geophys. Res., 115, A00E36, doi:10.1029/2009JA014775.
1. Introduction
[2]TheD region ionosphere spans the altitude range
between 60 and 100 km, a region often dubbed the
“ignorosphere” due to the difficulty of systematic mea-
surements [Sechrist, 1974]; the altitude range is t oo high
for balloons, and too low for in situ satellite measure-
ments. Typical methods for studying the D reg ion include
in situ rocket measurements, which are nece ssarily tran-
sient; VHF radar measurements of the mesosphere–lower
thermosphere (MLT) region; and long‐wave (i.e., VLF
and LF) probin g. The VLF technique for probing the D
region takes advantage of the fact that waves in the VLF
frequency range (3–30 kHz) reflect from the l ower iono -
sphere at an altitude of ∼60 km in the daytime and ∼85 km at
night, and the received signal inherently contains informa-
tion about that region of the ionosphere and its variability.
Furthermore, these frequencies propagate over long dis-
tances (tens of megameters) with low loss (∼2 dB per Mm)
in the Earth‐ionosphere waveguide, thanks to the ∼90 km
thickness of the waveguide at night.
[
3] The sources of VLF energy of interest to us in this
paper are twofold: man‐made, ground‐based VLF transmit-
ters, which typically operate from as low as 12 kHz (Russian
alpha transmitters) to as high as 40.75 kHz (the NAU trans-
mitter in Puerto Rico); and lightning, whose radio energy
spans up to 10 GHz, but peaks in the 5–10 kHz range [Rakov
and Uman, 2003, p. 6]. The peak radiation component of the
lightning electric field at ionospheric altitudes can reach 15 V/m
or higher [Lu, 2006; Marshall et al., 2010], exceeding the
thresholds of ionization and optical emissions from atmo-
spheric constituents, and thus causing direct D region iono-
spheric modification. VLF transmitters are not powerful
enough to cause ionization in the D region, but they can heat
the ionosphere enough to observably modify its conductivity
[Rodriguez et al., 1994]. Furthermore, these VLF transmitter
signals, when observed at long range, can be used to “probe”
the D region ionosphere, providing an important measure-
ment technique of localized D region disturbances.
[
4] In this review, we discuss the uses of ELF and VLF
observations for studying the D region ionosphere and pro-
vide a survey of the literature in this area. In sections 2 and 3
we provide an introduction to the VLF probing technique, and
its use in studying lightning‐induced electron precipitation
(LEP) events and “early” perturbations. Section 4 deals with
1
STAR Laboratory, Stanford University, Stanford, California, USA.
2
Koç University, Istanbul, Turkey.
3
Electrical and Computer Engineering Department, Duke U niversity,
Durham, North Carolina, USA.
4
Center for Space Physics, Boston University, Boston, Massachusetts,
USA.
Copyright 2010 by the American Geophysical Union.
0148‐0227/10/2009JA014775
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, A00E36, doi:10.1029/2009JA014775, 2010
A00E36 1of21

the interaction between the lightning EMP and the ionosphere
and the resulting density changes and elves. In section 5 we
discuss the use of ELF and VLF signatures of lightning that
produce sprites, halos, and elves, while section 6 discusses
VLF “ sferic bursts” and their relationship with in‐cloud (IC)
lightning discharges. Finally, section 7 discusses the rela-
tionship between VLF measurements and gamma ray events,
in particular the recently discovered terrestrial gamma ray
flashes (TGFs).
2. Lightning‐Induced Electron Precipitation
[5] The basic mechanism for lightning‐induced electron
precipitation (LEP) events is illustrated in Figure 1. The
electromagnetic pulse from lightning (1) propagates primar-
ily in the Earth‐ionosphere waveguide, but a small fraction of
its energy penetrates through the ionosphere and into the
magnetosphere, where it propagates (2) as a whistler mode
wave [Helliwell, 1965]. This whistler energy propagates at
∼0.1c to 0.01c either obliquely in the plasmasphere (as
shown), or along field‐aligned density enhancements known
as ducts. In the equatorial region, the whistler energy in the 1–
10 kHz range interacts with radiation belt electrons of ∼100–
300 keV in cyclotron resonance and may cause pitch angle
and energy scattering (3). Electrons scattered from their
trapped orbits into the bounce loss cone precipitate (4) in
the upper atmosphere (at 60–120 km) due to collisions
with neutral molecules. This burst of precipitation causes an
ionospheric disturbance, i.e., an electron density enhance-
ment, which can be detected via phase and/or amplitude
changes on subionospheric VLF transmitter signals (5), or by
in situ rocket (200–300 km altitude) or satellite (300–700 km
altitude) measurements.
[
6] Whether the whistler waves propagate obliquely
(nonducted, i.e., waves which do not propagate along
magnetic field lines) or ducted (along the field lines, guided
by field‐aligned density depletions) can be determined
based on the location of the precipitation region: for oblique
(nonducted) whistlers, this precipitation occurs over a wide
region centered poleward of the lightning discharge, as
shown in Figure 1, while the location of ducted precipitation
is more localized and determined by the location of the
“duct,” which may or may not be directly above the light-
ning discharge.
[
7] LEP events have been observed on transmitter signals
at frequencies as low as 2.45 kHz (at Siple Station,
Antarctica [Carpenter et al., 1985]), and at frequencies as
high as 780 kHz (from a broadcast transmitter at Santa
Cruz, Argentina [Carpenter et al., 1984]) . Most reported
observations have amplitude changes ranging from 0.04 dB
[Carpenter et al., 1984] to 6 dB [Helliwell et al., 1973] and
phase changes as high as 12° [Inan and Carpenter, 1987].
[
8] What follows is a brief history of LEP observations
and theory.
2.1. History of LEP Research
[
9] Early detections of electron precipitation were made
from rockets and satellites: Rycroft [1973] observed a single
electron burst event on a rocket that was associated in time
with an observed whistler, marking the first connection
between lightning and radiation belt electrons. Goldberg et al.
[1986] simultaneously observed a lightning flash and electron
precipitation on a rocket, and noted that the lightning‐induced
Figure 1. Physical mechanism of LEP. (1) Lightning injects VLF energy into the Earth‐ionosphere
waveguide and into the magnetosphere. (2) VLF energy propagates as whistler mode waves in the plas-
masphere. (3) Whistler waves interact in gyroresonance with relativistic radiation belt electrons, scattering
them in energy and pitch angle. (4) Relativistic electrons scattered into the loss cone precipitate in the
upper atmosphere through collisions with neutrals. (5) Secondary ionization is detected by subionospheric
VLF probing techniques. Adapted from Lauben et al. [2001].
INAN ET AL.: ELF AND VLF SIGNATURES OF LIGHTNING A00E36A00E36
2of21

precipitation of electrons >40 keV overwhelmed that pro-
duced by the nearby 21.4 kHz NSS transmitter.
[
10] Voss et al. [1984] were the first to document a direct
connection between satellite measurements of transient (1 s)
enhancements of loss cone electrons directly correlated with
whistlers observed at Palmer Station, Antarctica. They found
flux increases of two orders of magnitude over the back-
ground in energies of 45–200 keV. These events were ana-
lyzed in more detail by Voss et al. [1998], who found that
these LEP events were caused by ducted whistlers, and cal-
culated that a single LEP burst of 10
−3
erg s
−1
cm
−2
depleted
∼0.001% of the particles in the affected flux tube. Using the
SAMPEX satellite, Blake et al. [2001] found enhancements
of drift loss cone electron fluxes in the range 150 keV to
1 MeV directly associ ated with thunderstorms. Most
recently, Inan et al. [2007 ] made observations of LEP on
the D EMETER satellite, constraining the electron energies
to 100–300 keV, and directly correlated the events with
lightning data from a ground‐based network. Fur thermo re,
these results demonstrated enhanced precipitating fluxes
when the satellite passed over thunderstorm regions.
2.1.1. Subionospheric VLF Probing
[
11] Perturbations to VLF transmitter signals have been
observed since the use of such transmitters began, and have
been correlated with geomagnetic storms for many decades
[e.g., Bracewell et al., 1951; Potemra and Rosenberg, 1973].
Numerous authors [Lauter and Knuth, 1967; Belrose and
Thomas, 1968] have suggested that such perturbations were
caused by enhancements of the D region nighttime electron
density due to precipitating electrons with energies >40 keV.
The first direct connection between sudden changes in the
VLF transmitter amplitude and phase and lightning‐induced
whistlers was made by Helliwell et al. [1973], who postulated
cyclotron resonance between the electrons and the whistler
wave as the mechanism for electron precipitation. Lohrey and
Kaiser [1979] reported phase perturbations that suggested
a counterstreaming interaction exclusively (no costreaming
interaction) between the whistler waves and the radiation belt
electrons.
[
12] Inan et al. [1985] and Inan and Carpenter [1986,
1987] used phase and amplitude perturbations to VLF
transmitter signals to interpret event sizes in terms of pre-
dicted precipitation fluxes, to associate events with one‐hop
whistlers, and together with a single‐mode propagation the-
ory, inferred precipitated fluxes of 10
−4
–10
−2
erg cm
−2
s
−1
.
Inan and Carpenter [1987] found that amplitude perturba-
tions require a factor of ∼3 larger ionospheric perturbation
compared to phase perturbations, making the latter far more
regularly observed. Inan et al. [1988a] used the subiono-
spheric method to show that LEP events can involve pre-
cipitation of >1 MeV electrons, due to the observation of
events on paths at L ≤ 1.8 and an unusually fast recovery,
suggestive of low‐altitude energy deposition. Inan et al.
[1988b] provided the first direct and extensive association
between many tens of LEP events and individual cloud‐to‐
ground lightning flashes, and Inan et al. [1988c] discussed
the importance of longitude dependence of the bounce loss
cone edges in both hemispheres. Inan et al. [1989] compared
results of a test particle model with observations of an LEP
event on the SEEP satellite to calculate the pitch angle and
energy distribution of precipitated electrons. Simultaneous
observations of ionospheric disturbances in the Northern and
Southern hemispheres were made by Burgess and Inan [1990,
1993], showing that both hemispheres can be affected by the
same lightning event.
[
13] Dowden and Adams [1988] observed that most LEP
events display positive phase changes and negative ampli-
tude changes. These are in agreement with the single‐mode
theory of Inan and Carpenter [1987]: precipitation lowers the
ionospheric reflection height, typically resulting in increased
absorption, and thus negative DA, and negative Dh, the
ionospheric reflection height. They further show that phase
changes D ∼−1/Dh, so a negative Dh makes a positive
phase change. They go on to attribute negative phase to
interference between multiple modes. Dowden and Adams
[1988] further noted the shift from North to South (in the
Southern Hemisphere, i.e., poleward) of the precipitation
region in time. Dowden and Adams [1989] also noted that
mode interference in the Earth‐ionosphere waveguide may
change the sign of the phase or amplitude change. Using two
receivers and monitoring two VLF transmitter frequencies
only 100 Hz apart, Dowden and Adams [1990] and Adams
and Dowden [1990] were able to make estimates of the dis-
tance along the path to LEP events, placing them over active
lightning regions.
2.1.2. Implications for Optical Detection of LEP
[
14] While most of the work cited above concentrated
on precipitation of 100–300 keV electrons, as those reach to
D region altitudes to be probed by subionospheric techni-
ques, Jasna et al. [1992] calculated whistler‐induced pre-
cipitation of ∼100 eV electrons, and showed that they may
constitute 30 times larger energy fluxes with the same input
wave power density. This prediction may be significant
for optical detection of LEP events, a method which can
observe higher altitudes and lower‐energy electrons; some
early optical observations of precipitation induced by mag-
netospheric chorus emissions have been made by Helliwell
et al. [1980] at Siple Station of the 4278 Å emission from
Nitrogen, and by Doolittle and Carpenter [1983] at the
conjugate point in Roberval, Québec. The first detection of
optical emissions from LEP is yet to be made; however, the
intensity is expected to be very low on account of low
particle fluxes, in comparison with natural aurora, and
the time scales are immensely sh orter (∼1 s for a single
LEP event, versus minutes or hours for auroral displays).
Peter and Inan [2007] find that LEP precipitation peaks
at 10
−2
ergs s
−1
cm
−2
for a 100 kA lightning discharge,
while the precipitated flux associated with visible aurora
is generally in the range 0.1–10 ergs s
−1
cm
−2
[e.g., Meng,
1976; Rees, 199 2].
2.1.3. Ducted or Nonducted Whistlers?
[
15] The question of whether ducted or nonducted whis-
tlers are involved in LEP events, or both, has been a topic of
great interest. Observations of both ducted and nonducted
whistlers was first made by the OGO 1 satellite [Smith and
Angerami, 1968]. These provided evidence of the presence
of both types of whistlers in the magnetosphere, while only
ducted whistlers are thought to be able to reach the ground
[Helliwell, 1965]. This experiment also discovered the mag-
netospherically reflecting (MR) whistler [Edgar, 1976].
Burgess and Inan [1993] studied 74 perturbations to sub-
ionospheric VLF, LF, and MF signal paths, and found that
INAN ET AL.: ELF AND VLF SIGNATURES OF LIGHTNING A00E36A00E36
3of21

every perturbation was time‐associat ed with a ducted
whistler. They went on to estimate that ducted whistlers
contribute as muc h as plasmaspheric hiss to radiat ion belt
losses, a ssuming every ducted whistler causes precipitation.
[
16] Poulsen et al. [1993b, 1993a] present the first 3‐D,
multimode model of subionospheric VLF propagation in the
Earth‐ionosphere waveguide, utilizing the Long Wave
Propagation Capability (LWPC) code [Pappert and Snyder ,
1972; Ferguson and Snyder, 1987]. This model enables
simulation of ionospheric disturbances off the GCP from
transmitter to receiver; Poulsen et al. [1993a] show that
disturbances ∼50 km in radius must be within ∼250 km of
the 6000 km path, and that this transverse displacement
from the path depends on both the disturbance size, the path
length, and the t ransmitter frequency. Lev ‐ Tov et al. [1995]
used the model of Poulsen et al. [1993b] to compare simu-
lations with observed LEP events, and from comparisons
extracted information about the altitude profiles of both the
ambient electron density and the disturbed ionosphere.
2.2. Holographic Imaging of Precipitation Regions
[
17] The concept of VLF “imaging” was proposed by
Inan [1990]. Therein, the authors used observations at
three receiver locations from six VLF transmitters in North
America, and the resulting grid of criss‐crossing great circle
paths, to deduce the precipitation region of an LEP event.
Similarly, Dowden and Adams [1993] used an array of five
receivers in New Zealand, monitoring the NWC transmitter,
to find north‐south dimensions of 100–250 km for LEP
disturbances.
[
18] An algorithm for “ holographic imaging ” using an
array of receivers oriented along a meridian was developed
by Chen et al. [1996]. This concept was taken to fruition by
Johnson et al. [1999b, 1999a], through the development of
the Holographic Array for Ionospheric Lightning (HAIL).
Johnson et al. [1999a] went on to use the HAIL array to
make accurate measurements of size (∼1000 km, in agree-
ment with model calculations from Lauben et al. [1999])
and location of precipitation regions, showing both pole-
ward displacement from the lightning strike and a tendency
toward increasing delay from the lightning strike with
increasing latitude (as shown in Figure 2, bottom left).
These observations provided definitive evidence of oblique
whistler propagation. This oblique propagation and pole-
ward displacement was modeled by Lauben et al. [1999,
2001], using the ray‐tracing and wave‐induced particle
precipitation (WIPP) codes of Inan and Bell [1977]. Other
experimental work in determining the size of precipitation
patches has been conducted using an array of VLF receivers
on the Antarctic peninsula [Clilverd et al., 2002], who esti-
mate regions of 600 km in latitude by 1500 km in longitude.
[
19] Extensive studies of LEP events were conducted at
Stanford in recent years using data from the HAIL array.
Peter and Inan [2004] looked at statistics of LEP events
during two 4 h storms and found onset delays, poleward
displacement, and event durations as expected from non-
ducted whistler propagation. Peter and Inan [2005] studied
LEP events associated with Atlantic hurricanes (Isabel 2003
and others), but found no indication that the hurricane‐
associated lightning is more likely to induce electron pre-
cipitation events than lightning associated with other storm
systems. Finally, Peter and Inan [2007] compared HAIL
data of LEP events with modeling results to determine the
ionospheric electron density perturbation corresponding to
Figure 2. Example LEP events on 28 March 2001. Taken from Peter [2007]. (top left) Map showing the
location of the causative lightning discharges and the HAIL great circle paths. (bottom left) Data for an
event at 0709:48 UT. (right) Data for another event at 0642:01 UT.
INAN ET AL.: ELF AND VLF SIGNATURES OF LIGHTNING A00E36A00E36
4of21

LEP events, and found enhancements of ∼15% at 85 km
altitude, consistent with Lev‐Tov et al. [1995] and Clilverd
et al. [2002]. They also calculated that 0.006% of the
electrons in a flux tube with energies 100–300 keV were
precipitated, which can be compared to Voss et al. [1998],
who calculate 0.0015% of electrons with energies >45 keV.
[
20] An example with two LEP events is shown in
Figure 2, taken from Peter [2007]. The data below the map,
for an LEP event at 0709:48 UT on 28 March 2001, dem-
onstrate the differential delay from the lowest latitudes to the
highest. The event at right, an LEP event on the same day at
0642:01 UT, shows how the HAIL array is used to measure
the latitudinal extent of the precipitation region, as the
NAU‐LV and NAA‐WA paths are unaffected.
2.3. Global Rates of Lightning‐Induced Electron
Precipitation
[
21] One of the most important outstanding questions
regarding LEP pertains to its global contribution as a sink
for relativistic particles in the radiation belts. With that goal
in mind, some recent work has focused on quantifying the
global effects of LEP. Clilverd et al. [2004] were able to
relate the amplitude of subionospheric VLF perturbations,
and the precipitated flux, to the causative lightning peak
current. They showed that only 70 kA and larger strokes
created detectable LEP events. Rodger et al. [2005] used as
estimate of the precipitated flux (1–2×10
−3
ergs cm
−2
s
−1
,
from Rodger [2003]), and the global rate of lightning, to
estimate the rate of precipitation over the globe. They con-
cluded a mean rate of 3×10
−4
ergs cm
−2
min
−1
, with peaks
as high as 6×10
−3
ergs cm
−2
min
−1
over active regions.
[
22] The potential impact of LEP on radiation belt electron
densities, especially those in the ∼100 keV to 1 Mev range,
makes them an important area of study. Using the Inan et al.
[1997] WIPP code, Bortnik et al. [2003] has suggested that
magnetospherically reflecting (MR) whistlers may play an
important role in the formation and maintenance of the slot
region, and Bortnik et al. [2006a, 2006b] made estimates of
the differential number flux of particles precipitated from
the radiation belts by MR whistlers. Future work is needed
to compare these model results with observations of whis-
tlers, precipitating electrons, and LEP events on the ground.
Such studies are beginning in recent years with satellites
such as DEMETER [e.g., Inan et al., 2007], but further
long‐term studies are needed to accurately assess the effects
of lightning on radiation belt populations.
3. Early VLF Events
[23] The delay of ∼1 s from the lightning return stroke to
the onset of the VLF amplitude or phase change is a key
feature of LEP events; it is the time required for VLF
whistler mode waves to propagate to the equatorial region of
the radiation belts, plus the time for energetic electrons to
travel from the equatorial region to the lower ionosphere.
Armstrong [1983] was the first to report events that exhibited
a delay of <100 ms (i.e., within the time resolution of the
recording system), precluding any involvement with the
magnetosphere or plasmasphere. An example of such an
event is shown in Figure 3. These events were labeled
“early” or “early/fast” VLF events by Inan et al. [1988b],
who noted that (1) the “early” event s were most often
positive amplitude changes while LEP events (or classic
VLF perturbation events) were negative; (2) they were
confined to the early part of the night anal yzed, while LEP
events came later in the night; (3) t he events had rise times
of any where from <50 ms up to 2 s; (4) a number of events
had “step‐like” amplitude changes with no observed recov-
ery; and (5) a direct coupling mechanism must be responsible
for the observed short delay. However, the authors were
unable to suggest a likely mechanism for these events. The
“fast” moniker for these events denotes their usually rapid
rise time to reach their full perturbation; however, slower rise
times have been observed since Inan et al. [1988b], though
Figure 3. Example of the first early/fast events observed by Armstrong [1983]. (left) The geometry of
the problem, where a storm just south of the NSS transmitter affects the path of that transmitter to Palmer
Station, Antarctica. (top right) Palmer broadband VLF data in spectrogram form. A lightning‐generated
sferic is observed at 0 s, and an associated sferic is detected 1 s later. (bottom right) The narrowband data
from Palmer, exhibiting an early/fast event with zero delay from the causative sferic.
INAN ET AL.: ELF AND VLF SIGNATURES OF LIGHTNING A00E36A00E36
5of21

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "A survey of elf and vlf research on lightning‐ionosphere interactions and causative discharges" ?

In fact, some of the earliest evidence for direct disturbances of the lower ionosphere in association with lightning discharges was obtained in the course of the study of such VLF perturbations. In this paper, the authors provide a review of the development of ELF and VLF measurements, both from a historical point of view and from the point of view of their relationship to optical and other observations of ionospheric effects of lightning discharges. 

Q2. How did Blake et al. find enhancements of drift loss cone electron fluxes?

Using the SAMPEX satellite, Blake et al. [2001] found enhancements of drift loss cone electron fluxes in the range 150 keV to 1 MeV directly associated with thunderstorms. 

Q3. What did Burgess and Inan estimate that whistlers contribute to radiation belt losses?

They went on to estimate that ducted whistlers contribute as much as plasmaspheric hiss to radiation belt losses, assuming every ducted whistler causes precipitation. [16] 

Q4. What is the history of the use of subionospheric VLF probing?

In this review paper, the authors have presented a brief history of the use of subionospheric VLF probing as a technique for studying the lower ionosphere, and in particular the effects of lightning on the ionosphere and, indirectly, in the magnetosphere. 

Q5. How is the absolute amplitude of the received signal measured?

In other words, the absolute amplitude of the received signal is linearly proportional, albeit in a complicated way due to propagation effects, to the absolute amplitude of the source. 

Q6. What was the first direct connection between sudden changes in the amplitude and phase of the V?

The first direct connection between sudden changes in the VLF transmitter amplitude and phase and lightning‐induced whistlers was made byHelliwell et al. [1973], who postulated cyclotron resonance between the electrons and the whistler wave as the mechanism for electron precipitation. 

Q7. What did they do to account for the evolution of the electron energy distribution due to the EMP?

That paper also predicted that the EMP‐ionosphere interaction would lead to “airglow.” Taranenko et al. [1992, 1993a, 1993b] used a 1‐D fully kinetic model to account for the evolving electron energy distribution due to EMP heating of electrons, and furthermore calculated optical emissions due to the EMP. 

Q8. How did Huang and Wright measure the charge moment changes in spriteproducing lightning?

however, these early measurements of charge moment changes in sprite‐producing lightning showed conclusively that the charge moment changes in sprite‐associated lightning strokes are generally large enough to generate sprites according to quantitative model predictions.