On the Possible Origin of Chaotic Pulse Trains in Lightning Flashes

Transcription

1 atmosphere Article On Possible Origin Chaotic Pulse Trains in Lightning Flashes Mohd Muzafar Ismail 1,2, *, Mahbubur Rahman 1, *, Vernon Cooray 1, *, Mahendra Fernando 3, Pasan Hettiarachchi 1 and Dalina Johari 1,4 1 Ångstrom Laboratory, Division Electricity, Department Engineering Sciences, Uppsala University, Box 534, Uppsala SE-75121, Sweden; Pasan.Hettiarachchi@angstrom.uu.se (P.H.); Dalina.Johari@angstrom.uu.se (D.J.) 2 Faculty Electronics and Computer Engineering, Telecommunication Engineering, Universiti Teknikal Malaysia Melaka, Hang Tuah Jaya, Durian Tunggal, Malacca 76100, Malaysia 3 Department Physics, University Colombo, Colombo 03, Sri Lanka; mahendra.fernando@gmail.com 4 Faculty Electric Engineering, Centre for power Electrical Engineering Studies, Universiti Teknologi Mara, Shah Alam, Selangor 40450, Malaysia * Correspondence: Muzafar.Ismail@angstrom.uu.se (M.M.I.); Mahbubur.Rahman@angstrom.uu.se (M.R.); Vernon.Cooray@angstrom.uu.se (V.C.); Tel.: (V.C.); Fax: (V.C.) Academic Editor: Robert Talbot Received: 30 October 2016; Accepted: 30 January 2017; Published: 5 February 2017 Abstract: In this study, electromagnetic field radiation bursts known as chaotic pulse trains (CPTs) and regular pulse trains (s) generated by lightning flashes were analyzed. Through a numerical analysis it was found that a typical CPT could be generated by superimposing several s onto each or. It is suggested that chaotic pulse trains are created by a superposition several regular pulse trains. Since regular pulse trains are probably created by dart or dart-stepped leaders or K-changes inside cloud, chaotic pulse trains are caused by superposition electric fields caused by more than one se leaders or K-changes propagating simultaneously. The hyposis is supported by fact that one can find regular pulse trains eir in beginning, middle or later stages chaotic pulse trains. Keywords: regular pulse train; chaotic pulse train; numerical superposition; HF radiation 1. Introduction Regular pulse trains (s) consist a train pulses each with a signature similar to electric field pulses generated by dart leaders. According to Krider et al. [1], s are generated by a dart-stepped leader or K-changes type discharges propagating inside cloud. Each pulse train had a typical duration µs with a mean time interval between individual pulses about 5 6 µs. The total duration a single pulse was typically 1 2 µs with a pulse width at full width at half maximum (FWHM) about 0.75 µs. According to Rakov et al. [2], amplitude individual pulses in s is about two order magnitude smaller than return stroke (RS) pulses. Moreover, observation made by Rakov et al. [2] shows that s occur both in cloud flashes and during cloud activity between return strokes. Chaotic pulse trains (CPTs) were first observed by Weidman [3]. Since CPTs that Weidman observed occurred during leader stage subsequent return strokes, he coined name chaotic subsequent return stroke to represent subsequent return strokes preceded by chaotic pulse bursts. Subsequent studies demonstrated that CPTs can also occur in electric fields cloud flashes without any return strokes [4 6]. Gomes et al. [6] observed occurrence CPTs in ground flashes eir just before subsequent return strokes or without any association with subsequent return Atmosphere 2017, 8, 29; doi: /atmos

2 Atmosphere 2017, 8, strokes. According to Gomes et al. [6], width individual pulses CPTs is within range a few microseconds with minimum ranges into sub microsecond scale. The inter-pulse duration lies within range 2 20 µs, and duration CPT is µs. The discharge mechanism by which a CPT is generated in lightning flashes is currently unknown. In this paper, we make an attempt to understand origin CPTs in lightning flashes. First, we collect electric field records containing typical signatures s and CPTs generated from lightning discharges. Then we use measured s in a numerical analysis to show that a typical CPT could be a result sum more than one s. This finding n suggests that possible origin CPTs could be several s taking place simultaneously in cloud during a lightning discharge. 2. Methodology 2.1. Measurement Measurements electric field signatures generated by negative cloud-to-ground (CG) flashes pertinent to Swedish thunderstorms were recorded from May to August 2014 during summer at Uppsala ( N, E). The site is located 70 km inland Baltic Sea and 38 m above sea level. The distances to negative CG flashes from measurement site were estimated by using Swedish lightning location network (LLN). The flashes being analyzed here ranged from 10 to 100 km from measuring station. The timing for each event was provided by a global positioning system (GPS). The wideband electric field was recorded by a calibrated flat plate antenna located 1.5 m above ground. The antenna system consisting a parallel flat plate antenna toger with an electronic buffer circuits for fast electric field measurements is identical to ones used previously in studies reported in references [7 9]. Moreover, a narrowband electric field (3 MHz) was also recorded. The high frequency (HF) radiation at 3 MHz was detected by a tuned antenna system whose electric field sensing flat plate was located on ro a 3-m-high van located in vicinity wideband antenna. The field enhancement caused by placing antenna over ro van was obtained by connecting antenna located on ro van to a wideband system and comparing amplitudes wideband signals measured simultaneously by two antennas (i.e., one located aboveground and or located on ro van). A 60-cm-long coaxial cable (RG58) was used to connect antenna to electronic buffer circuit for fast electric fields. The zero-to-peak rise time output was shorter than 30 ns when step inputs pulse was applied to fast electric field antenna system. The decay time constant for fast electric field antenna, which is determined by RC constant electronics, was 15 ms. The decay time constant fast electric field antenna was found to be sufficient for faithful reproduction microsecond scale field changes. The tuned circuit at 3 MHz is a combination passive elements where inductance (47 µh) is connected in series with antenna (58 pf) and 50 Ω termination forming a simple RCL circuit. The bandwidth tuned circuits at 3 MHz is 264 khz. Signals from both wideband and narrowband systems were fed into a digital transient recorder (Yokogawa SL 1000 equipped with DAQ modules ) through proper termination (50 Ω resistor). The wideband signal was fed to this 12-bit digitizer by 10-m-long coaxial cables (RG-58), whereas 3 MHz signal was fed by a 3-m-long cable. The sampling time was set to 250 ms at a sampling interval 10 ns. A pre-trigger time 30% total time window (250 ms) was used in se measurements. The trigger setting oscilloscope was set such that signals both polarities could trigger system Data In this measurement campaign a total 98 lightning flashes were found where both wideband and 3 MHz signals were successfully measured simultaneously. In 64 se flashes eir a (s), or a CPT(s) or both were found. We did not analyse all se records for this study but only considered those cases (18 lightning flashes) where both s and CPTs were present without having any time

3 Atmosphere 2017, 8, separation or any or events (for example downward dart or dart-stepped leader pulses) between m. We have observed that s and CPTs had appeared differently in different flashes so we classified m into four categories. Category 1 ( CPT): a CPT that started with an (16 trains in eight lightning flashes); Category 2 (CPT ): a CPT that ended with an (10 trains in six lightning flashes); Category 3 ( CPT ): a CPT that both started and ended with an (6 trains in two lightning flashes); and Category 4 (CPT CPT): a CPT that had an in middle it (4 trains in two lightning flashes). Figure 1a d shows examples se four categories respectively. Three megahertz signals for respective wideband records are also shown in se figures. Figure 1a shows a pulse train that started with a positive regular pulse train which lasted for about 170 µs, which was n followed by a chaotic pulse train that lasted for about 75 µs. In example shown in Figure 1b, chaotic train started at about 200 µs before continuing up as a regular pulse train, which was about 190 µs long. Figure 1c shows a pulse burst where chaotic pulses appear in middle regular pulse trains. Observe that polarity pulses in s is opposite to each or. Also in certain instances, regular pulses same polarity appeared in middle pulse train as shown in Figure 1d. The convention to define polarity pulses is an atmospheric electricity sign convention. According to this convention, a negative charge in a cloud produces a negative field at ground level, and a negative return stroke produces a positive field change. It should be noted that all s and CPTs irrespective categories above were observed in between return strokes. There were a total 36 trains. The average time separation from end previous RS to beginning trains (all categories) was about 24 ± 14 ms, whereas time separation from end trains (all categories) to beginning following RS was about 4 ± 2 ms. Figure 2a shows an example a flash where a sequence -CPT (category 1) was found between first and second RS. Sixteen trains found for category 1 were distributed in between RS in way such that eight trains appeared between first and second RS, six trains appeared between second and third RS, and two trains appeared between third and fourth RS. Note that for category 1, 16 trains mean 16 distinct s and 16 distinct CPTs. In category 2, 10 trains were found, where six trains appeared between first and second RS, three trains appeared between second and third RS, and one train appeared between third and fourth RS. Note that for category 2, 10 trains mean 10 distinct s and 10 distinct CPTs. In category 3, six trains were found, where two trains appeared between first and second RS, three trains appeared between second and third RS, and one train appeared between third and fourth RS. Note that for category 3, six trains mean six distinct CPTs but 12 distinct s. Because, for each CPT re were one before and one after. Finally, in category 4, four trains were found, where two trains appeared between first and second RS, one train appeared between second and third RS, and one train appeared between third and fourth RS. Note that for category 4, four trains mean 4 distinct s but 8 distinct CPTs as re were two CPTs for each. Thus, total sum all distinct s and CPTs were 42 and 40 respectively. 3. Results and Discussion 3.1. Characteristics Regular Pulse Train () Pulses in occur at quite regular time intervals. Each pulse in begins with a fast large-amplitude portion, followed by a small and slowly varying overshoot. However, in some pulses, opposite overshoot is not that pronounced. If we consider any pulse with an opposite overshoot to be bipolar, n almost all pulses we observed in present study can be categorized as bipolar. However, in a study conducted by Kolmasova and Santolik [10], each individual pulse is considered to be unipolar if immediately following overshoot opposite polarity does not exceed one-half peak amplitude original s pulse. If we use this criterion, n most individual pulses in should be categorized as unipolar. In Figure 2b an is shown which is an expansion from Figure 2a. Note that polarity pulses is negative here.

4 Atmosphere 2017, 8, As described in Section 2.2, re were a total 42 distinct s and total number individual pulses in se regular pulse trains was 800. On average, each had about 22 pulses. The individual pulse duration (PD) has an arithmetic mean and standard deviation about 2 ± 1.0 µs and one typical individual pulse is shown in Figure 2c, where PD also is marked. A sequence pulses normally persists for µs with an arithmetic mean and standard deviation inter-pulse duration (IPD) about 5 ± 1.3 µs. The IPD is time duration between maximum amplitude two consecutive pulses in train as shown in Figure 2d. As mentioned in Section 2.2, all s took place in between return strokes and distribution se trains was following: 20 distinct s were located between first and second RS, 16 distinct s were located between second and third RS, and 6 distinct s were located between third and fourth RS. The amplitude pulses in s were calculated as a fraction following return stroke amplitude. The average electric field amplitude pulses in was about one-fifth electric field peak return stroke, and this ratio varied from 0.1 to 0.4. All values presented here regarding parameters PD and IPD agreed with values found by Krider et al. [1] but values for total duration pulse trains in our case were much lower than that Krider et al. This can be explained by fact that values inter-stroke interval for temperate regions (Uppsala) are shorter than that for subtropical regions such as Florida and Arizona [7] Characteristics Chaotic Pulse Train (CPT) There were a total 1120 individual pulses in those 40 distinct CPTs found in this study. Each distinct CPT had about 28 pulses on average. Figure 3a shows an example CPT which is an expansion CPT in Figure 2a. Because ir chaotic behavior, especially with regard to ir amplitudes, pinpointing exact beginning and end each pulse in CPT with good precision is difficult. The total duration (TD) CPT is estimated using criterion adopted by Gomes et al. [6]. According to this criterion, TD is time duration between regions pulse activity at start and end pulse train where pulse amplitude is equal to 10% maximum amplitude in CPT. In several cases where noise level was high, we had to increase this limiting value to 20%. In our study, observed duration, TD, CPTs varied from 20 to 120 µs. We estimated TD, PD and IPD manually with an estimated uncertainty about ±0.5 µs. Pulse duration (PD) is width an individual pulse when amplitude at start and end individual pulse is equal to 10% maximum amplitude in CPT. The width each pulse or its duration (PD) varied within range 0.5 to 13 µs with an arithmetic mean and standard deviation about 4 and 3 µs, respectively. The inter-pulse duration (IPD) is time duration between maximum amplitude two consecutive pulses in train. The IPD varied within range from 2 to 13 µs with an arithmetic mean and standard deviation about 8 and 5 µs, respectively. In Figure 3b which is an expansion Figure 3a, PD and IPD are shown. As mentioned in Section 2.2, all CPTs took place in between return strokes and distribution se trains was following: 20 distinct CPTs were located between first and second RS, 14 distinct CPTs were located between second and third RS, and 6 distinct CPTs were located between third and fourth RS. As for amplitudes, sometimes pulses with largest amplitudes occur immediately after initiation chaotic pulse train, and in some cases, occurrence largest peak happens eir in middle and/or at end chaotic pulse train. The arithmetic mean and standard deviation ratio maximum amplitude pulses to following return stroke peak were 0.4 ± 0.3. A comparison pulse characteristics found in our study to Gomes et al. [6] shows a good agreement but total train duration is not comparable due to fact that Gomes et al. analyzed and presented data from three different geographical regions toger, namely from Sweden, Denmark and Sri Lanka.

5 Atmosphere 2017, 8, HF Radiation (3 MHz) We recorded HF radiation at 3 MHz simultaneously with CPT and. The HF radiation associated with CPTs was previously reported by Gomes et al. [6] and Mäkela et al. [11], but as far as we know re were no reports on measured HF radiation associated with s. Figure 1a d, Figures 2e and 3c show several examples HF radiation associated with CPT and. Our general observation was that 3 MHz signal corresponding to an had also a regular pattern. Moreover, we found that as a regular pulse had an initial peak followed by an opposite overshoot, 3 MHz signal density behaved consistently with less oscillation compared to a chaotic pulse train. As a chaotic pulse had an erratic shape, 3 MHz signal density showed also an erratic nature. On or hand, 3 MHz signal corresponding to a CPT showed more irregular pattern where amplitudes could eir be increased or decreased significantly compared to amplitudes a 3 MHz signal in response to an. This information was important because it was easier to find transition from one type pulse train to anor type pulse train in 3 MHz signal and this was used to confirm transition between trains in wideband signal. This can be seen in Figure 1a d where transition is marked with dotted line Numerical Analysis: Construction a CPT through Numerical Superposition s The similarity in width individual pulses in both regular pulse trains and chaotic pulse trains and occurrence regular pulse trains at start, in middle, or at end chaotic pulse trains suggested to us hyposis that se chaotic pulses are probably created by superposition several regular pulse bursts at random times. This is a reasonable physical scenario because during lightning flashes, re could be instances where several dart or dart-stepped leaders, like discharge processes, are simultaneously active in cloud. Each dart-stepped like leader produces an, and sum signals from all dart-stepped-like leaders will appear as a CPT. In this section, we will demonstrate this by superimposing s numerically. We will also show that CPTs constructed in this manner also have Fourier spectrums similar to ones estimated from measured CPTs. To realize this, out 42 distinct s we chose one randomly to start with. After selecting a 15- to 60-µs-long section, we took anor section from same with same length but this time start new section was shifted in time to eir direction with 2 5 µs compared to previous section. In case category 3 ( CPT ) where two distinct s were present, sometimes we chose sections s from both s and sometimes from only one distinct. Then we made a superposition se two sections. The simulated signal was n compared to measured CPT belonging to corresponding train. The comparison was based on ir erratic behaviour and temporal characteristics. This procedure was repeated as a trial and error method. Number sections that were superimposed, duration train section and position in time chosen sections were changed until best combination was found that corresponded measured CPT satisfactorily. In Figure 4a,b results our numerical simulation are shown, indicating best combinations with three (Figure 4a) and four s (Figure 4b) and with 60-µs-long section. In numerical superposition, we tried superimposing eir same polarity or s with mixed polarities (i.e., adding both positive and negative s to create CPTs). Both procedures gave rise to chaotic-type pulse bursts. The reason we tried constructing a CPT with mixed polarity was because our records showed that we occasionally observed polarity pulses in s changing ir polarities. In or words, may start with pulses one polarity and n change pulses toward end polarity. This showed us that in some situations it is possible for an one polarity to interact with an opposite polarity in creation CPTs. Interestingly, resulting pulse trains have all characteristics similar to those observed in chaotic pulse trains. The similarity simulated CPT to ones measured strongly suggests that se pulse trains are created by a series dart-stepped leader like discharges propagating simultaneously in cloud.

6 Atmosphere 2017, 8, Since, according to our suggestion, chaotic pulse burst is a superposition several regular pulse bursts, HF associated with chaotic pulse burst can be treated as sum HF radiation associated with regular pulse bursts. When different HF radiation pulses overlap, depending on ir phases, sometimes resulting amplitude is smaller than that, and sometimes it is larger. This means that peak amplitudes HF radiation pulses from chaotic pulse trains will be smaller Atmosphere than2016, that 7, 29 an in some cases, and it could be larger in or cases Figures 2e and 3c show several examples HF radiation associated with and CPT. sometimes it is larger. This means that peak amplitudes HF radiation pulses from chaotic Figurepulse 1b shows trains will be HF smaller radiation than that associated an with in some a CPT cases, that and ended it could as be larger an. in or Observe cases. that amplitude Figures HF 2e and radiation 3c show associated several examples with HF regular radiation pulse associated train iswith less than and amplitude CPT. HF radiation Figure 1b associated shows with HF radiation chaotic associated pulse train. with a However, CPT that ended Figureas 1c,d an. shows Observe that that amplitude HF amplitude radiation associated HF radiation with associated an with is larger regular than pulse amplitude train is less than HFamplitude radiation associated with ahf CPT. radiation In se associated cases, one with can see chaotic pulse occurrence train. However, where Figure CPT 1c,d isshows in that middle amplitude an and HF radiation associated with an is larger than amplitude HF radiation associated is in middle a CPT. The different trends amplitude HF for a CPT and an in with a CPT. In se cases, one can see occurrence where CPT is in middle an and some cases are due to destructive and constructive interference. The sum amplitudes several s is in middle a CPT. The different trends amplitude HF for a CPT and an can bein less some than cases are amplitudes due to destructive all original and constructive s, or interference. less than The sum amplitudes amplitudes only several a few original s s can but be higher less than than amplitudes amplitudes all original remaining s, or less original than s, amplitudes or sum only a amplitudes few several s original cans evenbut be higher zero because than amplitudes destructive interference. remaining original Meanwhile, s, or when sum sum amplitudes amplitudes several several s s lines can up, even re be zero is constructive because destructive interference. interference. This one Meanwhile, can immediately when be seen from sum recorded amplitudes data shown several in s Figure lines 1a d. up, re is constructive interference. This one can immediately be seen from recorded data shown in Figure 1a d. We also analysed Fourier spectrum measured CPT and compared results with We also analysed Fourier spectrum measured CPT and compared results with ones ones obtained obtained from from simulated simulated CPT. CPT. The The results results comparison comparison are shown are in shown Figure 5a,b. in Figure Note 5a,b. Note that that Fourier spectrums constructed and and measured measured CPTs CPTs are almost are almost identical, identical, and y and have y a have a peak peak around around khz. khz. This This similarity Fourier spectrums also also suggests suggests common common origin origin s and s CPTs. and CPTs. Figure 1. Cont.

7 Atmosphere 2017, 8, Atmosphere 2016, 7, (c) Figure 1. Cont.

8 Atmosphere 2016, 7, Atmosphere 2017, 8, Atmosphere 2016, 7, (d) (d) Figure 1. Several examples wideband and HF radiation associated with mixed chaotic pulse trains Figure 1. Several examples wideband and HF radiation associated with mixed chaotic pulse trains Figure (CPT) 1. Several and regular examples pulse trains wideband (). and CPT HF that radiation starts as associated a positive with ; mixed CPT chaotic that ends pulse as trains a (CPT) and regular pulse trains (). CPT that starts as a positive ; CPT that ends as (CPT) negative and regular ; (c) pulse CPT that trains occurred (). in CPT middle that starts a positive as a positive (preceding) ; and CPT a negative that ends as a a negative negative (following); ; ; (c) (c) (d) CPT CPT that that in occurred occurred middle in incpt. middle middle a a positive positive (preceding) (preceding) and and a negative a negative (following); (following); (d) (d) in in middle middle CPT. CPT. Figure 2. Cont.

9 Atmosphere 2017, 8, Atmosphere 2016, 7, (c) (d) (e) Figure Figure 2. A 2. A sample sample.. preceded preceded CPT CPT located located between between first first and and second second RS. RS. The distance first RS was about 20 km from measuring system; An after expanded from The distance first RS was about 20 km from measuring system; An after expanded from Figure 1a; (c) A single pulse from in Figure 1b; (d) section several regular pulses with Figure 1a; (c) A single pulse from in Figure 1b; (d) A section several regular pulses with inter-pulse duration about 9 µs; (e) HF radiation associated with in Figure 1b. inter-pulse duration about 9 µs; (e) HF radiation associated with in Figure 1b.

10 Atmosphere 2017, 8, Atmosphere 2016, 7, (c) Figure Figure A sample sample CPT. CPT. A CPT CPT expanded from Figure 1a; 1a; Asection section CPT CPTexpanded expandedfrom from Figure Figure 2a 2a with with pulse pulse duration duration (PD) (PD) and inter-pulse and inter-pulse duration duration (IPD) indicated; (IPD) indicated; (c) HF radiation (c) HF radiation associated with associated CPT in with Figure CPT 3a. in Figure 3a.

11 Atmosphere 2017, 8, Atmosphere 2016, 7, Figure 4. Chaotic pulse burst produced by numerical superposition s. Superposition three Figure 4. Chaotic pulse burst produced by numerical superposition s. Superposition three s; Superposition four s. s; Superposition four s.

12 Atmosphere 2017, 8, Atmosphere 2016, 7, Figure Figure Comparison Comparison Fourier Fourier spectrums simulatedand andmeasured measuredcpt. CPT. Superposition Superposition three three s; s; Superposition Superposition four four s. s. 4. Conclusions 4. Conclusions In this paper, electromagnetic field radiation bursts known as CPTs and s generated by lightning In this flashes paper, were electromagnetic analysed. In addition field radiation to providing bursts known statistical as data CPTs for and features s generated chaotic by lightning pulse trains flashes (CPTs) were analysed. and regular In pulse addition trains to providing (s), we hyposized, statistical data based for on features features chaotic pulse measured trains (CPTs) s and andcpts, regular that pulse CPTs trains are (s), possibly weformed hyposized, by a superposition based on features s. The validity measured s and CPTs, that CPTs are possibly formed by a superposition s. The validity

13 Atmosphere 2017, 8, hyposis was demonstrated by constructing CPTs by superimposing several s. The remarkable similarity Fourier spectrums for constructed and measured CPTs furr demonstrates validity hyposis. Based on this, we suggest that CPTs are nothing but a random superposition a series s created by several dart-stepped leader or K-changes type discharges propagating simultaneously inside clouds. Acknowledgments: The participation Mohd Muzafar Ismail (Ismail M.M.) is funded by funds from Ministry Higher Education Malaysia and Universiti Teknikal Malaysia Melaka. Participation Vernon Cooray and Mahbubur Rahman are funded by fund from B. John F. and Svea Andersson donation at Uppsala University. Participation Dalina Johari is funded by fund from Ministry Higher Education Malaysia and Universiti Teknologi Mara Malaysia. The authors would like to acknowledge Division for Electricity and Lightning Research, Ångström Laboratory, Uppsala University, for excellent facility provided to carry out this research. Author Contributions: The study was completed with cooperation between all authors. Mohd Muzafar Ismail as first author prepared and carried out experiment, collected data, analyzed data, and wrote manuscript. Vernon Cooray contributed to writing manuscript, gave original idea and checked validation measurement and checked simulation analysis. Mahbubur Rahman rewrote, checked simulation analysis, and contributed with knowledgeable discussions and suggestions. Mahendra Fernando contributed with idea and knowledgeable suggestion. This whole idea came from Mahendra Fernando and Vernon Cooray analysis chaotic pulses observed in Sri Lanka. Pasan Hettiarachchi and Dalina Johari supported measurement technique and analysis. All authors agreed with submission manuscript. Conflicts Interest: The authors declare no conflict interest. References 1. Krider, E.P.; Radda, G.J.; Noggle, R.C. Regular radiation field pulses produced by intra-cloud discharges. J. Geophys. Res. 1975, 80, [CrossRef] 2. Rakov, V.A.; Uman, M.A.; Hfman, G.R.; Maters, M.W.; Brook, M. Burst pulses in lightning electromagnetic radiation: Observations and implications for lightning test standards. IEEE Trans. EMC 1996, 38, [CrossRef] 3. Weidman, C.D. The Sub- Microsecond Structure Lightning Radiation Fields. Ph.D. Dissertation, University Arizona, Tucson, AZ, USA, Bailey, J.; Willett, J.C.; Krider, E.P.; Leteinturier, C. Submicrosecond structures radiation fields from multiple events in lightning flashes. In Proceedings International Conference on Atmospheric Electricity, Uppsala, Sweden, April 1988; pp Rakov, V.A.; Uman, M.A. Waveforms first and subsequent leaders in negative lightning flashes. J. Geophys. Res. 1990, D10, [CrossRef] 6. Gomes, C.; Cooray, V.; Fernando, M.; Montano, R.; Sonnadara, U. Characteristics chaotic pulse trains generated by lightning flashes. J. Atmos. Sol. Terr. Phys. 2004, 66, [CrossRef] 7. Ismail, M.M.; Rahman, M.; Cooray, V.; Sharma, S.; Hettiarachchi, P.; Johari, D. Electric field signature in wideband, 3MHz and 30 MHz negative ground flashes pertinent to Swedish thunderstorms. Atmosphere 2015, 6, [CrossRef] 8. Cooray, V.; Lundquist, S. On characteristics some radiation fields from lightning and ir possible origin in positive ground flashes. J. Geophys. Res. 1982, 87, [CrossRef] 9. Galvan, A.; Fernando, M. Operative Characteristics a Parallel-Plate Antenna to Measure Vertical Electric Fields from Lightning Fields from Lightning Flashes; Report, UURIE ; Uppsala University: Uppsala, Sweden, Kolmasová, I.; Santolík, O. Properties unipolar magnetic field pulse trains generated by lightning discharges. Geophys. Res. Lett. 2013, 40. [CrossRef] 11. Mäkelä, J.S.; Edirisinghe, M.; Fernando, M.; Montaño, R.; Cooray, V. HF radiation emitted by chaotic leader processes. J. Atmos. Sol. Terr. Phys. 2007, 6, [CrossRef] 2017 by authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under terms and conditions Creative Commons Attribution (CC BY) license (