QIE Xiushu,ZHANG Yijun,YUAN Tie,ZHANG Qilin,ZHANG Tinglong,ZHU Baoyou,

LU Weitao2,MA Ming6,YANG Jing1,ZHOU Yunjun7,and FENG Guili81Key Laboratory of Middle Atmosphere and Global Environment Observation,Institute of Atmospheric Physics,Chinese Academy of Sciences,Beijing 100029

2Laboratory of Lightning Physics and Protection Engineering,Chinese Academy of Meteorological Sciences,Beijing 100081

3College of Atmospheric Science,Lanzhou University,Gansu 730000

4College of Atmospheric Physics,Nanjing University of Information Technology,Nanjing 210044

5Laboratory for Climate Environment and Disasters of Western China,Cold and Arid Regions Environmental and Engineering Research Institute,Chinese Academy of Sciences,Lanzhou 730000

6School of Earth and Space Sciences,University of Science and Technology of China,Hefei 230026

7College of Atmospheric Sciences,Chengdu University of Information Technology,Chengdu 610225

8Shandong Research Institute of Meteorology,Jinan 250031

1.Introduction

The importance of atmospheric electricity research has been increasingly recognized in recent decades.Thunderstorms are a major player in the global atmospheric electrical circuit,the main components of which are the ionosphere,clear air,conducting earth,thunderstorms(especially the electric charge structure inside the storm)and lightning.In the last decade,new detection and location technologies for lightning with high temporal and spatial resolutions have greatly enhanced studies on atmospheric electricity in China,especially with respect to lightningphysics and thunderstorm electricity.

Lightning is a type of disastrous weather characterized by high voltage,high peak current with large amplitude variation,and severe electromagnetic(EM)radiation.The probabilityofgroundfacilities struckdirectlybylightninghasbeen greatly reduced since the innovation of the lightning rod by Benjamin Franklin.However,the economic losses caused by lightning have been increasing because of today’s wide utilization of micro-electronics.Lightningresearch is important not only in terms of scientif i c research objectives,but also from the viewpoint of lightning protection practices.

Areas covered in this review include the physics and effects of lightning;rocket-triggered lightning and the physical processes of discharge;thunderstorm electricity over the Qinghai-Tibetan Plateau and its surrounding areas;lightning activities associated with severe convective storms;the effect and response of lightning to climate change;the numerical simulation of thunderstorm electrif i cation and lightning discharge;and lightning detection and location techniques.Fi-nally,recent f i ndings relating to sprites in the middle atmosphere above thunderstorms are brief l y described.

2.The physics and effects of lightning

Knowledge regarding the physics of lightning and its effects at f i ne temporal resolution is very important not only in terms of scientif i c research objectives,but also from the viewpoint of lightning protection engineering,in particular when considering today’s wide utilization of microelectronics nowadays.In the last decade,many f i eld experiments of natural lightning have been conducted continuously in the areas of Gansu(e.g.,Qie et al.,2000a,2002b;Zhang et al.,2008),Guangdong(e.g.,Qie et al.,2002a;Dong et al.,2002;Zhang et al.,2009e;2014e),Qinghai(e.g.,Qie et al.,2005a;Zhang et al.,2009c;Wang et al.,2013;Fan et al.,2014),the Tibetan Plateau(e.g.,Zhang et al.,2004a;Zhao et al.,2004;Qie et al.,2005b),Shanghai(Zhu et al.,2003;2014)and Shandong(Qie et al.,2007;Kong et al.,2008;Qie et al.,2014a).The main results from these experiments are outlined in the following sections.

2.1.New insights into the progression of the stepped leader and dart leader

The stepped and dart leaders are important processes in cloud-to-ground(CG)lightning fl ashes.Knowledgeof leader progression is important for the interpretation of lightning initiation.Pro fi ting from high-speed video with a temporal resolution of higher than 1000 frames per second in correlations with broadband electric(E) fi eld change signatures,some new insights into the stepped leader in negative CG fl ashes have been documented(Kong et al.,2005;Qie and Kong,2007;Lu¨ et al.,2008a,Kong et al.,2009;Zhang et al.,2009e).

QieandKong(2007)studiedindetailtheprogressionfeatures of the stepped leader with multiple grounded branches.From the time-expanded waveform of the E fi eld change during the return stroke stage,four sequential peaks could be clearly found.The corresponding time differences between two adjacent peaks were approximately 4,9 and 10μs,respectively.The four peaks corresponded to four return strokes induced by four different leader branches initiated from one channel trunk.The average 2D speed of the four branches was about 1.1×105m s?1.Lu¨ et al.(2008a)analyzed the optical pulse characteristics of a downward stepped leader with six ungrounded branches.They found that the pulses originated from the main channel and the branches were almost identical in terms of geometric mean(GM)values of 10%–90%rise time and half-peak width,which were around 0.4μs and 1.1μs,respectively.

Qie et al.(2002a)found that the E fi eld change waveforms produced by the stepped leader in negative CG fl ashes could be divided into three types,in terms of the distant E fi eld characteristics between the fi nal leader pulse and the return stroke pulse,and the amplitude of the last leader pulse correlated well to the following return stroke radiation fi eld peak.Zhang et al.(2005)further analyzed the close E fi eld characteristics of stepped leader/return stroke process in negative CG fl ashes within 20 km.The leader-stroke waveform at distances of less than 3.4 km appeared as V-shaped pulses with negative leader E fi eld change,while it appeared as MP(monotonous positive)-shaped pulses with positive leader E fi eld change at distances greater than 5.1 km.

Jiangetal.(2014a)detectedanupwardnegativeCG lightning fl ash initiated from a high structure by a high-speed camera operated at 10 000 frames per second.They found,for the fi rst time,that the bidirectional propagation of a dart leader developing through the preconditioned channel.The leader initially propagated downward through the upper channelwith decreasingluminosityandspeedandterminated at an altitude of about 2.2 km.Subsequently,it restarted the development with both upward and downward channel extensions.The 2-D partial speed of the leader’s upward propagation with positive polarity ranged between 3.2×106m s?1and 1.1×107m s?1,while the speeds of the downward propagation with negative polarity ranged between 1.0 and 3.2×106m s?1.

According to the current continuity equation and DU(Diendorfer and Uman)return-stroke model,Zhang et al.(2009b)studied the spatiotemporal characteristics of charge distribution along the lightning channel below 5 km.They found that the charge density deposited along the channel during the return-strokeprocess neutralized the leader charge and decreased upward along the channel.The transferred chargedensitydecreasedwith time,the currentin the channel ceased to fl ow,and the transferred charge became zero after a suf fi ciently long time.

2.2.Electromagneticf i eld ofthe return strokeandits nonlinear fractal nature

The return stroke is the optically brightest and most intense lightning process,and produces the most readily identi fi able EM fi eld signature and the most serious lightninginduced damage.On the basis of the electric fi eld changes from widebandslow antenna,Zhang et al.(2003b)foundthat the zero crossing time of the fi rst stroke radiation fi eld was 63μs and 66μs for positive CG and negative CG fl ashes,respectively,and the corresponding overshoot depth was 20%and 31%,respectively.The initial peak of the return stroke radiation fi eld was followed by several subsidiary peaks.The time interval between the successive peaks increased,whereas their amplitudes decreased in sequence.Zhang and Qie(2003)reconstructed the E fi eld waveforms of the return stroke in three different distance ranges(15–50,50–100 and 100–150 km).A relationship between the return stroke radiation fi eld change and return stroke current pulse was also rebuilt based on the traditional transmission line model.Qie et al.(2001)found that 54%of the negative multiple-stroke CG fl ashes had at least one subsequent stroke with a peak E fi eld change larger than that of the fi rst strokes.Furthermore,about20%ofthesubsequentstrokeshadapeakE fi eldamplitude largerthan those of the fi rst strokes.The GM of the peak fi eldratioofthesubsequenttothe fi rstreturnstroke was 0.46.

On the basis of a statistical analysis of the waveform signatures of 59 f i rst return strokes of negative CG f l ashes in Guangdong,Kong et al.(2009)found that 15.3%of the f i rst strokes were characterized by two or more peaks separated in time by 4–486 μs,which meant one CG f l ash could strike at more than one ground point.Based on this research,it is necessary to re-evaluate the present lightning density distribution,which assumes that one f l ash of lightning strikes only one point—an assumption that has been widely employed when determining regional lightning protection levels.

Yang et al.(2008b)studied the characteristics of induced voltage in a horizontal conductor due to lightning.The simulated results showed that the induced voltage on both ends of the conductor would increase with increasing returnstrokevelocityandheightoftheconductor.Thevoltage would also increase with increasing of the matched grounding impedances on both ends of the conductor but the relationship was nonlinear.

The f i nite conductivity of the ground causes distortion of the EM f i eld,whose amplitude decreases and rise time increases,when the lightning EM f i eld propagates along the ground surface(Zhang et al.,2012a,2012b,2012c).For example,based on the 2D fractional Brownian motion(fBm)model describing the nature of the rough ground surface,Zhang et al.(2012a)analyzed the propagation effects of the rough ground surface on the vertical electric f i eld generated by lightning return strokes,and found that the extra f i eld attenuation increment caused by the roughness decreases with the decrease of the ground conductivity.When the ground conductivityis largerthan0.1S m?1(wet earth),thefrequencies higher than about 2 MHz are attenuated signif i cantly by a rough ground surface with a mean square height of 10 m(Zhang et al.,2012d).Li et al.(2014)developed a 3D f i nitedifference time-domain(FDTD)method for simulating the lightning-radiated EM f i eld over the 2D rough ground,and found that the effect of the 2D surface roughness on the horizontal f i eld could not be ignored even at a distance of 100 m from the lightning channel,and the increase of the land roughness resulted in a lower magnitude of the horizontal fi eld waveform.

Zhanget al.(2014a,2014c)furtheranalyzedthein fl uence of the horizontallyand mixedstrati fi ed conductinggroundon the lightning-induced voltages on the overhead line by using the 2D FDTD method and the Agrawal coupling model,and found that the strati fi ed conducting ground has much effect on the lightning-induced voltages on the overhead lines.Zhanget al.(2014b)also studiedthe effectof strike totall objects on the far lightning-radiated electromagnetic fi eld and presented the fi eld-to-current conversion factors(FCCFs)for current peak inferred from observed magnetic fi eld on the ground level.Assuming that a return stroke current contains two components,a breakdowncurrent and a coronacurrent,Zhang et al.(2009b)calculated each of the two components using the analytical expression with Heilder function,and foundthat the simulated currentwaveforms were in good agreementwith the optical measurementif the dischargetime constantswere properlychosen,andthe DU modelwas physically more reasonable and preferable in simulating the current along the lightning channel.Wang et al.(2012a)also found that the Heidler function component could ref l ect the physical characteristics of the subsidiary peaks in the current waveforms of return strokes.Zhao and Zhang(2009)found that,when the tortuosity of the channel was taken into account,the spherical or cylindrical symmetry vanished,and then the whole lightning channel could be regarded as a fractal antenna composed of a series of single line radiators in Cartesian space,suggesting that the tortuosity of lightning channel should be taken into account in the calculation of lightning EM f i eld.

Lightning can be considered as a large-scale cooperative phenomenon,evolving in a self-similar cascaded way(Gou et al.,2007).On the basis of the E f i eld change waveforms recordedby the slow antenna system,Gou et al.(2006)found that the H¨older exponent sharply decreased to its minimum with the occurrence of the return stroke,and the mean value was?0.1 by using a technique of wavelet-based local effective H¨older exponent.The time exponent variation was concave during the active stroke period.The standard deviation of the H¨older exponent reached its maximum just before the return stroke.Gou et al.(2009)also found that the return stroke process,in terms of its E f i eld waveform,had apparent fractality and a strong degree of multifractality.They suggested that wavelet and scaling analysis might be a powerful tool in the interpretation of a lightning-produced E f i eld and therefore in the understanding of lightning physics.

2.3.Lightning characteristics as revealed by VHF radiation location techniques

Lightning radiation in the very high frequency(VHF)band is associated with air breakdown processes.The VHF radiation characteristics from lightning discharge processes(Zhu et al.,2003;Chen et al.,2005;Wang et al.,2007;Cao et al.,2008)have been studied based on observation data.The location techniques of VHF radiation pulses are used to map or track the lightning discharge channel and infer the charge structure inside the thunderstorm(Zhang et al.,2008;Zhang et al.,2009f).

By using the data from a self-developed narrow band VHF interferometer system and the synchronous E fi eld changes produced by lightning,Zhang et al.(2008)studied the processes of a negative CG fl ash containing 19 strokes.It was found that the preliminary breakdown events of the CG fl ash started from the negative charge region and exhibited fi rstly a downward and then an upward propagation.Very intense and continuous radiation was found during the stepped leader process,while less and only discrete radiation during the dart leader processes.M-component events produced hook-shaped fi eld changes accompanied by an active burst of radiation at their beginning.Following these active radiation processes,M events appeared to fi nally contact main conductingdischarge channels.K events and attempted leaders were essentially the same as dart leaders except that they could not reach the ground and initiate return strokes.

Three-dimensional images of lightning progression were obtained successfully for the fi rst time in China based on a 3D lightning mapping system working at a frequency of 270 MHz with a 3 dB bandwidth of 6 MHz(Zhang et al.,2010).Signi fi cant differences between the negative CG and positive CG fl ashes in terms of the initiation and propagation of the radiation sources were found.The preliminary breakdown of thenegativeCG fl ashpropagatedataspeedofabout5.2×104m s?1.The stepped leader propagated downward at a speed of 1.3×105m s?1.The initial process of the positive CG fl ash was also associated with propagation processes of negative streamers.

2.4.Spectra of the lightning discharge channel in the visible band

Among many research methods in lightning physics,optical spectra analysis is the only one that can ref l ect the physical features within dischargechannels,which are in the form of plasma because of the ionization by the large discharge current.The spectrum of a discharge channel is closely related to the plasma properties and temperature of the channel.Since 2001,observationsof lightning spectra in the band of 400–700 nm have been carried out using a slitless spectrograph,and the average temperature and electron density in lightning channels have been deduced from the spectra in the visible band(e.g.,Yuan et al.,2002,2004a,2006;Ouyang et al.,2006;Wang et al.,2009b).

Two spectrum lines of wavelengths 604.6 nm and 619.4 nm for the f i rst return stroke of negative CG f l ashes were recorded for the f i rst time in lightning spectra by Yuan et al.(2004b).Based on the spectra structure,the lines in the lightning spectra could be classif i ed into two categories:essential lines and characteristic lines.The essential lines can be recorded in most lightning return strokes,which can,to some extent,ref l ect the common characteristics of the dischargechannel.Thecharacteristiclines carrythe information ref l ecting the trait of each individual discharge process.The essential lines include NII(399.5 nm),NII(480.3 nm),Hβ(486.1 nm),NII(568.0 nm),NI(648.2 nm),and so on.

Notable differenceshavebeen foundbetweenthe spectral features over Qinghai Plateau and the region of Guangdong(Yuan et al.,2002).The spectral energy is concentrated in the band of relatively short wavelength in Guangdong,while it is in the band of longer wavelength in Qinghai,indicating the discharge energy and channel temperature is higher in Guangdong.In Guangdong area,the transitions between exited states of n=3 in NII ions were the main compositions of lightning spectra,corresponding to upper excited energy of around 23 eV,and lines with higher exited energy(30 eV)from NII and OII ions could be recorded.On the other hand,the spectra over the plateau area were relatively weak,and the transitions from natural NI and OI atoms were strong,with the upper excited energies being around 13–14 eV.

Wang et al.(2009b)showed that the spectra of IC f l ashes demonstrate two different kinds of structural characteristics.One had a similar structure to that of a CG f l ash discharge,characterized by the lines of 500.5 and 568.0 nm being the strongest.The other was completely different,with the lines of 517.9 and 532.8 nm being the strongest;the upper excited energies of around 30 eV and the lines of 500.5 and 568.0 nm were very weak,in contrast.Meanwhile,more lines of OII with high excited energy were found in the spectra of IC lightning discharges in comparison with those of CG f l ashes in the same region.

According to the relative intensities of spectral lines and transition parameters,Ouyang et al.(2006)calculated the temperatures for individual lightning stroke at different heights of the discharge channel using a multiple-line method.The temperature in return stroke channels varied from 29 000 to 36 000 K.For a certain return stroke channel,temperature along the discharge channel showed a decreasing tendency with height.Zhang et al.(2007a)calculated the electron density according to the Hα line Stark broadening formula.The electron density varied from 4.68×1017to 5.03×1017cm?3.Simultaneously,with the Saha equation,the electron density was found to range from 9.03×1017to 17.5×1017cm?3.Generally,the more intense(i.e.large peak discharge current)the lightning discharge,the higher the channel temperature,the electron density and the relative concentration of highly ionized particles,but the lower the concentration of neutral atoms.

2.5.New observational evidence on positive CG f l ashes

A positive CG f l ash lowers positive charge to the earth,and usually neutralizes more charge than a negative f l ash.It is generally thought that less than 10%of CG f l ashes are positive on average,and positive CG f l ashes are less understood than negative ones because of the lack of observational data.Kong et al.(2008)reported a positive CG f l ash with a pronounced stepped leader.The fast E f i eld change of the positive leader immediately prior to the return stroke showed clearly pronounced pulses,indicative of a step-like development.The time intervals between the 26 leader pulses ranged from 3 to 31 ms,with a mean value of approximately 17 ms.The 2D propagation speed,estimated from the two adjacent frames,increased from 0.1×105to 3.8×105m s?1as the leader approached the ground.Kong et al.(2008)suggested that positive CG lightning can be produced by branching of the in-cloud discharge channels,probably when these channels occur near or below the cloud base.

Using data from the Lightning Mapping Array(LMA),Zhang et al.(2006a)analyzed the 3D spatial and temporal developmentof positive CG lightningdischarges.The results indicated that a positive CG f l ash could be divided into three stages.The f i rst stage was the discharge process in cloud with a long duration preceding the return stroke.This process propagated at a velocity of 105m s?1,and produced intensive radiation with a magnitude equal to that of the negative leader.During this stage,the lightning channels developed horizontally in the positive charge region with fewer branches as the negative polarity breakdown.During the stage after the return stroke,the lightning channels propagated at a velocity of two times faster than that before the return stroke.This stage involved many positive fast impulses and corresponded to the continuing current process produc-ing less and dispersed radiation points and more intensive radiation power.During the f i nal stage,the lightning channels developed at a velocity equal to that before the return stroke and the radiation points appeared mainly at the end of channel.All of the radiation points of the positive CG f l ash appeared in the positive charge region of the cloud.Little or no radiation was detected during the positive leader just before the return stroke.

Qie et al.(2013)analyzed 185 positive CG f l ashes containing 196 return strokes in Da Hinggan Ling forest region(50.4°N,124.1°E)of northeastern China documented with a multi-station network of fast and slow antennas.It was found that 71.9%of the positive CG f l ashes contained continuing current,but the average duration of continuing current was short with a GM value of 16.7 ms,because of the small size of the storm cell in this relatively high latitude region.According to the electric f i eld waveforms indicative(or not indicative)of IC discharge,positive CG f l ashes can be classif i ed into four types,i.e.,ordinary positive CG f l ash(63.8%),hybrid positive CG-IC f l ash(21.1%),hybrid IC-positive CG f l ash(5.4%),and hybridIC-positive CG-IC f l ash(9.7%).About 15.1%of the recorded positive CG f l ashes were byproduct of IC lightning discharge.

2.6.Narrow bipolar events

Narrow bipolar event(NBE)is a type of lightning discharge event.It is markedly different from regular CG and IC lightning in many respects.An NBE is associated with strong radio frequency emissions and narrow bipolar waveforms(Zhu et al.,2007,2010a).The 3D propagation of NBEs was observedfor the f i rst time in China by Zhang et al.(2010).The NBE channels originatedat an altitude of～10.5 km in the upperpositive-chargeregionandextendedhorizontally all around.The source power of an NBE can approach 16.7 kW,which is much greater than that of normal lightning discharge,which ranges between 100 mW and 500 W.The vertical scale of NBEs found by Liu et al.(2012)was in the range of 0.40–1.9km,with an average speed of 0.44×108to 1.0×108m s?1.

Negative NBEs produce larger electric f i eld changes on average and are more isolated from other dischargeprocesses compared to positive NBEs(Wu et al.,2011).Wu et al.(2012)found that the positive NBEs occur between the main negative charge layer and the upper positive charge layer,whilenegativeNBEs occurbetweentheupperpositivecharge layer and the negativescreeningchargelayer at the cloud top.L¨u et al.(2013)found that the NBE occurrence at 51 degrees N appeared to differ signif i cantly from that in most lower latitude regions.Speci fi cally,no NBEs with negative electric fi eld pulses(positive charge moving up)were observed.

Wang et al.(2012)documented 236 NBEs in Shandong province.Of between,32 occurred in isolation and 204 occurred in association with either IC or CG lightning discharges.Among the latter,130 appeared to initiate lightning discharges,while 72 embedded in lightning discharges and the remaining 2 terminated the lightning discharges.No apparent difference among all types NBEs was found on the parameters of NBEs radiation waveforms.They found the NBEs occurred at a height ranging from 7 km to 16 km with a peak power ranging from 12 kW up to 781 kW in the 267–273 MHz passband.

Zhu et al.(2010b)introduced a direct technique to measure the time takenby the currentfrontto propagatealongthe channel from distant radiation f i eld pulses of the NBE on the basisofthetransmission-linemodel,whichinvolvedintegrating over the initial half-cycle of the narrow bipolar waveform of the NBE.The ratio of the integral result to the initial peak amplitude made a good approximation to the time taken by the current front to travel along the channel.

2.7.Tallstructures-relatedlightningandlightningattachment process

Tall structures are usually used for lightning studies because of the high lightning incidence probability.Recently,high-speed video cameras have been used to observe the attachment process of lightningto tall structures with relatively high spatial and temporal resolution.Using high-speed images oftwo naturaldownward fl ashes struckontwo tall structures in Guangzhou,Lu et al.(2012)analyzed 45 unconnectedupwardleaders(UULs)occurredin19downwardnegative fl ashes.Each observed UUL was initiated by a downwardsteppedleaderbeforea newstrike pointwas struck.The maximum distances for the downward leaders to attract the UULs with inception heights from 100 to 200 m,200 to 300 m,and over 400 m were approximately 350 m,450 m,and 600 m,respectively.

Lu et al.(2013)recorded a downward negative lightning fl ash that terminated at a 440 m high building.The attachment process in this fl ash exhibited an unexpected behavior in that the downward leader tip connected to the lateral surface of the similar to 400 m upward connecting leader(UCL)belowits tip.It appearsthat theeffectofthedownwardleader on the UCL was signi fi cant,while the effect of the UCL on the downward leader was negligible,except for the f i nal 80μs preceding the beginning of the f i rst return stroke.The ratio of speeds of the downward leader and the UCL tends to decrease with time,ranging from 1.8 to 0.12.

Jiang et al.(2014b)studied the lightning f l ashes striking at a 325-m-tall meteorology tower in Beijing.Among eight upward lightning f l ashes documented during two thunderstorms,four were self-initiated events without lightning activity nearby prior to their initiation,two were triggered by the nearby positive CG with the initiation of the upward leaders from the tower lagged 0.4 ms and 5 ms behind,respectively,and the remaining two were triggered by nearby IC lightning activities.The average 2-D speed of the upward positive leader was 1.0×105m s?1within several hundred meters above the tower tip.

3.Rocket-triggered lightning and the physical processes of discharge

A common technique for triggering lightning with this method involves launching a small rocket trailing a thin,grounded copper wire toward the charged cloud overhead,and is called classical triggering.In the altitude triggering technique,the rocket usually spools out 50–100 m of insulating Nylon followed by several hundred meters of copper wire.The electric f i eld at the ground is usually used as a reference to launch a rocket for triggering lightning,although the electric f i eld aloft is more indicative(Qie et al.,1994),but hard to measure.The surface electric f i eld is usually 5–10 kV m?1when lightning is triggered successfully.Over 100 lightning f l ashes have been triggered with classical or altitude triggering techniques in several regions of China since then,including Gansu,Beijing,Jiangxi,Shanghai,Guangdong,the Tibetan Plateau and Shandong from 1989 to 2014.A new model rocket,made of composite material and assembled with a parachute,was newly developed and utilized successfully in recent experiments(Qie et al.,2010).Most of the results in the last decade are outlined below.

3.1.Characterization of lightning currents and the close EM f i eld

The most serious lightning-induced damage is usually caused by close lightning discharges.Rocket-triggered lightning provides a unique opportunity for measuring the discharge current and close EM f i elds,which are essential for both understandingthe physics of lightning and the design of lightning protection systems.

The Shandong Artif i cially Triggering Lightning Experiment(SHATLE)started from 2005 in Binzhou,Shandong Province(Qie et al.,2007;Zhang et al.,2006;2007b).Zhang et al.(2009a)and Zhao et al.(2009a)studied the current waveform characteristics and corresponding close E f i eld change during SHATLE.The whole discharge process of all triggered f l ashes lasted from 518 ms to 1900 ms.The GM value of the current peak was 12.1 kA(Qie et al.,2014a).Other current parameters are given in Table 1.The E f i eld changes produced by the dart leader/return stroke sequences appeared as V-shape pulses at 60 m,and the distance(r)dependence of the dart leader E f i eld change was r?1.18(Qie et al.,2009a;Zhang et al.,2009b).

The GuangdongComprehensiveObservationExperiment on Lightning Discharge(GCOELD)started from 2006 in Conghua,Guangdong.For the triggered lightning from 2006–2011(Zhang et al.,2014e),the peak current of the returnstrokes rangedfrom6.67to 31.93kAwith aGM valueof 15.9 kA.The maximum induced voltage generated by return strokes on the overhead power line(1200 m in length and 2 m above the ground)exceeded 10 kV.The maximum induced voltage on a vertical 10 m signal line was 3.10 kV.The triggered-lightning technique was also used to test the detectionef fi ciencyandlocationprecisionofGuangdonglightning locationsystem(LLS)inGuangdong.Itwas exploredthatthe LLS yielded detection ef fi ciency and location error of 92%and 760 m,respectively,for triggered fl ashes.For RSs of the triggered lightning,the peak currents given by the LLS deviated from those measured at the base of the lightning channel by 16%on average.

Yang et al.(2008a)developed a magnetic fi eld measuring system with two rectangular loops perpendicular to each other,withwhichthetotalhorizontalmagnetic fi eld produced by lightning discharges was detected.The magnetic fi elds at 60 m,based on 32 return strokes,varied from 18 mT to 148 mT with a GM of 52 mT(Yang et al.,2010).By using Ampere’s law of magnetic fi elds,the currents were inferred from the measured magnetic fi elds,which were in good agreement withthe directlymeasuredcurrentatthe baseofthedischarge channel.The system proved to be a useful tool for current retrieval and measurement of the close EM environment of lightning fl ashes(Yang et al.,2008a,2010).

Zhang et al.(2003c)studied the statistical characteristics of the leaders in fi ve altitude triggered negative lightning discharges in Guangdong in 1998.The E fi eld change at close distance was characterizedby negativediscontinuous oscillating pulses superimposed on slow positive change during the stable propagation stage of the bidirectional leader.Meanwhile,the E fi eld change at far distance began with bipolar pulses followed by unipolar pulses.Zhang et al.(2011a)employed two existing models,a “source charge”leader modeland a return-strokemodel of the modi fi ed transmission line model with linear current decay with height(MTLL),both based on the assumption of uniform leader chargedistributionalongthe channel,tosimulate theV-shape structural characteristics of the close dart leader/return stroke fi eld change.They suggested that at the early stage there was often some uncertainty regarding whether the charge deposited by the dart leader was completely neutralized by the following return-stroke process.

3.2.Discharge processes and new insights into the positive leader

The high-speed camera has become an important tool in triggered lightning research.Using a high-speed camera system and two electric f i eld antenna systems,L¨u et al.(2008b)documentedthe initial processes of an altitude-triggerednegative lightning event.The discharge began with the incep-tion and propagation of an upward positive leader and then a bidirectional leader process.The 2D propagation speed of the upward positive leader in its inception phase was about 3.8×104to 5.5×104m s?1from about 393 to 452 m above theground.Thestabledownwardnegativeleaderbeganatthe tip of the unstable one,with a 2D propagationspeed of about 1.9×105m s?1.Wang et al.(2012b)and Jiang et al.(2013a)found that the positive leader in the initial stage of the classical triggered lightning shows a similar stepped manner of propagation to that of the negative leader.The induced step length varies from 0.9 m to 3.7 m with a geometrical mean value of 1.7 m.

Table 1.Current waveform parameters based on SHATLE 2005–11.

Yang et al.(2009)analyzed the initial discharge stages of two triggered f l ashes on the basis of the synchronous data of the current and close EM f i eld.Lu et al.(2009)documented the initial processes of an altitude-triggered negative lightning event.The discharge began with the inception and propagation of an upward positive leader,then an almost simultaneous propagationof both the upward positive leader and downward negative leader followed,known as the bidirectional leader process.The 2D propagation speed of the upward positive leader in its inception phase was about 3.8×104to 5.5×104m s?1.The stable downward negative leader propagated with a 2D speed of approximately 1.9×105m s?1.The average step length was about 3 m,and the time interval between steps varied from 6μs to 31μs with a mean value of 15μs.

Dong et al.(2001)observed the weak VHF radiation of the positive leader at a close distance during a triggered f l ash using a broadband interferometer.The speed of the upward positive leader was on the order of 104to 105m s?1.In classical triggered negative f l ashes,the speed of upward positive leaders ranged from 0.35×105to 7.71×105m s?1.

3.3.M-componentswith peakcurrent in the rangeof kiloamperes

Large M-components with peak current in the range of kilo-amperes were found in a rocket-triggered negative f l ash in SHATLE 2009(Jiang et al.,2011).Among the 31 distinct current pulses,there were f i ve large M-components with unusually large peak current in the range of kilo-amperes.The GM value of peak current for the f i ve large M-components was 5.1 kA,the half peak width was 76.3μs,and the rise time from the 10%to 90%peak was 34.6μs,while the correspondingvalues for the 18 typical M-componentswere 243 A,400μs and 319μs,respectively.The waveform parameters of the return stroke and typical M-component were in good agreement with those found in previous triggered lightning(Yang et al.,2010;Zhao et al.,2011).The M-like events were superimposed on a slowly-varying continuing current,while the directly measured current prior to the stroke was not signif i cant.One stroke/M-component(RM)event that exhibited both stroke and M-component features was also found in the same f l ash(Qie et al.,2011).The simultaneous E f i eld and current waveform of RM implied a superposition of the dart leader and M incident wave in the channel.The proposed possible reason was that two branches with a common lower portion existed simultaneously in the upper part of the discharge channel.Zhang et al.(2011b)used the two-wave model,proposed by Rakov et al.(1995),to reveal that the speed of the M-component essentially controlled the electric fi eld but had little effect on the magnetic fi eld.A larger re fl ection coef fi cient resulted in a larger magnetic fi eld but a smaller electric fi eld.Jiang et al.(2013b)proposed a modi fi ed model based on Rakov’s two-wave theory and confi rmed that the evolution of M-component through the lightningchannelinvolvesa downwardwavetransferringnegative charge from the upper to the lower channel and an upward wave drainingthe charge transportedby the downward wave.The upwardwave serves to depletethe negativechargeby the downward wave at its interface and makes the charge density of the channel beneath the interface layer to be roughly zero.

4.Thunderstorm electricity on the Tibetan Plateau and its surrounding areas

The Tibetan Plateau is the largest and highest area in the world,with an average elevation exceeding 4000 m.Thunderstorms occur frequently in the summer season over the Tibetan Plateau because of its unique dynamic and thermodynamic effects.In the summers of 2003–05,comprehensive observations on thunderstorm electricity were conducted on the Tibetan Plateau and Chinese inland plateau regions(Qie et al.,2009b).

4.1.Thunderstorm electricity in four different plateau regions of China

Thunderstorm electricity has been examined experimentally in a number of studies across four different plateau regions,including Nagqu located in the central Tibetan Plateau(31°29′N,92°03′E;4508 m MSL),Datong on the Qinghai Plateau(37°04′N,101°35′E;2560m MSL),and Zhongchuan(36°36′N,103°39′E;1970 m MSL)and Pingliang(35°57′N,106°69′E;1630 m MSL)in Gansu Province on the Chinese inland plateau.

Qie et al.(2003c,2005b)found that thunderstorms,usually of small scale and short duration,occurred frequently in these plateau regions in the monsoon season.Hailstones with diameter less than 1 cm were usually observed with duration shorter than 10 min duringthunderstorms.Sometimes,more than one thunderstorm process was observed during a single thunderstorm day.However,the lightning activity was weaker compared with that in other prominent lower regions.Using the Lightning Imaging Sensor/Optical Transient Detector(LIS/OTD)data from 1995 to 2002,Qie et al.(2004)and Zhang et al.(2004)found that the mean total fl ash density over the Tibetan Plateau was 3–5 fl ashes yr?1km?2,that fl ash activity exhibited a seasonal variation and mainly occurred from June to August with a maximum lightning activity period from late June to mid-July,and that the diurnal variation peak appeared from 1400 to 1600 Local Standard Time(LST).

According to the polarity of the surface E fi eld,Qie et al.(2009b)divided thunderstorms in the four regions into two categories.(1)Special-type:The surface E f i eld underneath the thunderstorms had the same polarity as the clear sky,i.e.,the surface electric f i eld was controlled by positive charge inside the thunderstorms(def i ned as positive,as mentioned above).Zhang et al.(2004a)and Qie et al.(2005b)suggested that this kind of thunderstorm is characterized by an unusual tripole charge structure with a larger-than-usual lower positive charge center(LPCC)at the base of the thunderstorm,and usually the thunderstorm is characterized by IC f l ashes that occur mostly in the lower dipole.(2)Normal-type:The surface E f i eld was negative when the thunderstorms were overhead,consistent with the normal thunderstormsobserved in the other prominent lower altitude regions during the summer season.This kind of thunderstorm also showed a tripole charge structure,but the LPCC was weaker than the former.

Thecharacteristicsofthe surfaceE f i eld ofthunderstorms in the four plateau regions were similar to each other,but the percentage occurrence of the two types of thunderstorm was different(Qie et al.,2009b;Zhang et al.,2009d).Table 2 shows statistical results for the two types of thunderstorm in the four regions.The percentage of special-type thunderstorms increased with the altitude of the region.The special-type thunderstorms represented around 73%,60%,54%and 46%of the total in Nagqu,Datong,Zhongchuan and Pingliang,respectively.The f l ash rate in the four plateau regions was quite low compared with that in other low altitude regions.Zhang et al.(2010)found that the f l ash rate of special-type thunderstorms was slightly larger than that of normal-type ones in the plateau regions.

In differentstages ofthe thunderstorm,the surfaceE f i eld changesandlightningdischargetypescan bedifferent.Qie et al.(2005a)found that,in the mature stage of a thunderstorm at Datong,most IC f l ashes occurred between the LPCC and the main negative charge center aloft,and CG f l ashes were rare in this stage.In the later stage,a weakenedLPCC played a dominant role in the initiation of negative CG f l ash discharge(Qie et al.,2005b).According to the type of predominant f l ashes associated with the thunderstorm,Zhang et al.(2009d)found that special-type thunderstorms over the Chinese inland plateau can be divided into three types:(1)IC-dominated:no occurrence of CG f l ashes;(2)negative CG-dominated:＞50%of CG f l ashes were negative;and(3)positive CG-dominated:the dominant CG f l ashes were positive.Among 22 cases of special-type thunderstorms in the Nagqu region,four thunderstormsproducedno CG f l ashes.The percentage of CG f l ashes ranged from 1.88%to 76%for the other 18 thunderstorms,only in 6 cases the percentage of CG to total f l ash number was larger than 50%,and 15 cases were negative CG-dominated.Only three of the cases were mainly positive CG f l ashes and one of them occurred on 13 August 2003,which seemed to be the strongest thunderstorm in the central Tibetan Plateau during the 2-year observation period and produced 50 CG f l ashes with 49 being positive(Qie et al.,2009b).

4.2.Charge structure of thunderstorms in the plateau regions

Multi-station measurements on the E f i eld changes caused by lightning discharges is an effective way to estimate the charge centers inside thunderstorms.Point-charge and point-dipole models are usually used to analyze the neutralized charge centers for CG and IC f l ashes,respectively.For thunderstorms with a larger LPCC,it was found that IC f l ashes were usually polarity-inverted and occurred between the main negative charge center and the LPCC(Qie et al.,2000b).Zhang et al.(2004a)and Qie et al.(2005b)inferred the electric charge structure of thunderstorms and the characteristics of lightning discharges at the initial stage of thunderstorms using VHF location techniques and E f i eld changes in the Nagqu region.They found that most of the IC f l ashes were polarity-inverted and occurred between the negative charge region in the middle and the positive charge regionatthebottomofthethunderstorm,suggestingthethunderstorms might have had a tripole charge structure.Recently,based on 3D localization of wideband electric f i eld change pulses,Li et al.(2013)analyzed the charge structure of a thunderstorm in Qinghai Province,China.They found an inverted dipole charge structure at the development and mature stage of the thunderstorm,with four charge layers(positive–negative–positive–negative)at the dissipating stage,at heights of 5.0,4.0,3.0,and 1.8 km,respectively.

The existence of a middle negative charge and large LPCC over the Chinese inland plateau and Tibetan Plateau is widely recognizedand accepted.However,evidence of upper positivechargein storms was not founduntil 2008.Using data from E f i eld changes from a seven-site network of slow antennas synchronized by a Global Position System(GPS)with a 1μs time resolution in the region of Zhongchuan,Cui et al.(2009)found that the upper dipole was also a source of IC f l ashes.Among 10 IC f l ashes,f i ve occurred betweenthe upper dipole and the other fi ve between the lower.The heights of IC discharge moments were located between 3.3 and5.6km MSLforthelower fi veIC fl ashes andbetween6.8 and 7.7 km MSL for the upper fi ve,respectively.Analyzing 16negativeCG and2positiveCG fl ashes inDatong,Zhanget al.(2009c)respectively found that the negativecharge region was located at a height of 5.5–8.0 km MSL(mostly around 6.5 km MSL)and the positive charge height was around 8.5 km MSL,indicating that the charge structure of special-type storms could be basically represented by a tripole structure but with a larger-than-usual LPCC.The height range of the main negative charge region is in good agreement with the result given by Qie et al.(2000a).

Table 2.Statistical results of two types of thunderstorm in four regions.

In situ E fi eld measurement is a direct and effective method to determine more accurately the charge structure inside thunderstorms.A balloon-borne E fi eld sounding system,based on the principle of point discharge,was designed by Zhao et al.(2009b).The fi rst E fi eld pro fi le insideaspecial-typethunderstormwasobtainedintheregionof Pingliang.There were four charge regions with three layers inside the storm and one at the lower boundary of the storm.The LPCC region was between 4.5 and 5.3 km MSL(corresponding to a temperature region of 3°C to ?2°C).The main negative charge layer was between 5.4 and 6.6 km(?3°C to ?10°C).The upper positive charge layer was located between 6.7 and 7.2 km(?11°C to ?14°C),and a negative screening charge layer was also detected at the lower boundary of the thunderstorm.These observational results confi rmed that thunderstorms in the plateau regions are usually characterized by a tripole charge structure with a larger-thanusual LPCC.

A 3D thunderstorm model coupled with dynamical and electrical processes has been developed for theoretical studies on the spatial and temporal evolution of charge structure in the plateau regions(Guo et al.,2003,2007).It was found that the lower maximum disturbing central potential temperature,the reversal temperature and relative humidity in the middle layer were key parameters for the formation of the charge structure.The simulation results by using real sounding data indicated that both types of thunderstorms appeared to begin with the lower dipole of a normal tripole structure,rather than with the upper dipole followed by the development of a weaker lower positive charge region.

4.3.Initial discharge processes of lightning f l ashes

The characteristics of the preliminary breakdown process involved in CG f l ashes are dissimilar in different geographical areas,which may be associated with the charge structure of thunderstorms.Due to the special charge structure inside the plateau’s thunderstorms with a larger-than-usual LPCC,negative CG f l ashes usually proceed with a long preliminary breakdown process lasting several hundreds of milliseconds,similar to IC discharges(Qie et al.,2000a;Kong et al.,2006).Qie et al.(2000b)found that the K-type breakdown process could occur during the preliminary breakdown process,which they named as IC discharges.Using the timeof-arrival(TOA)method,they investigatedthe K-type breakdown processes duringthe long IC dischargeprocess through fi ve-station measurements of a wideband slow antenna system in the Zhongchuan region,and found that the K processes occurred in the lower part of the storm.It was found that both positive CG and negative CG fl ashes usually followed long lasting IC discharge processes with a duration of 170–300 ms,and K-type breakdown processes during initial IC discharge started from the negative charge region and propagated downward to the LPCC with an average speed of 1.5×107m s?1.Wang et al.(2009a)also found that the initial discharge of IC fl ashes developed from the middle negativechargeregiontotheLPCC basedonthelocationofpulses from a seven-station network of fast antennas.

Long IC discharges just before negative CG fl ashes were also found in Datong,with an average altitude of about 2650 m MSL.On the basis of slow antenna and high-speed digital camera observation data,Qie et al.(2005a)found that long-duration IC processes occurred just before the stepped leader/returnstrokesequence.One suchIC dischargeprocess lasted approximately 160 ms and occurred in the lower part of the cloud with the lowest point at around 1.7 km above the ground.Zhang et al.(2003a)found that the preliminary discharge showed a bi-layer structure,by using a shortbaseline lightning VHF pulse location system with the TOA technique.Using LMA data,Zhang et al.(2009f)also found that the preliminary breakdown process with longer duration time in negative CG discharges was an IC discharge process.

A large LPCC may play an important role in longer preliminary breakdown processes.However,from one case of typical thunderstorms on the central Tibetan Plateau(4508 m MSL),Qie et al.(2005b)foundthe existenceofthe LPCC did not cause positive CG fl ashes,and only negative CG fl ashes were observed in the late stage of the thunderstorm.The quite large LPCC prevented negative CG fl ashes from occurring because abundant lower positive charges could make IC discharges between the lower dipole possible.In the late stage of the storm,when the LPCC decreased greatly with the fall down of the most positive charge carriers(rain particles and graupel or hail)to the ground,negative CG fl ashes couldbetriggeredfrequentlybytheLPCC.Thissuggeststhat a weak LPCC is conducive to the occurrence of negative CG fl ashes,while a large LPCC is conducive to polarity-inverted IC fl ashes ornegativeCG fl ashes with longerpreliminarydischarge.

5.Lightning activities associated with severe convective weather

Severe convectiveweather,such as hailstorms,mesoscale convective systems(MCSs)and so on,generally produces not only heavy precipitation,damaging wind and hailstone,but also lightning discharges which sometimes cause serious damage.The lightning activity and its relationship with dynamic processes and precipitation structure in severe convectiveweathersystemshasbeenstudiedinthelast decadebased on data from CG lightning location networks,SAFIR3000 lightning data,Doppler radar,meteorological satellites,and Tropical Rainfall Measuring Mission(TRMM)-based sensors.

5.1.Lightning characteristics in different thunderstorms

Lightning is an indicator of vigorous convection.The lightning activities in different kinds of thunderstorms such as hailstorms,MCSs,and squall lines have been studied in China.Feng et al.(2006a,2007,2008)and Liu et al.(2009)found that hailstorms usually presented higher ratio of positive CG fl ashes during periods of hail fall in Shandong Province.The positive CG fl ashes represented more than about 45%of total CG fl ashes,which was much higher than the climatic mean value(12.5%)in the region.The falling of hail was often reported in the region of dense positive CG fl ashes.Sometimes,hailstones appeared slightly on the right fl ank of the dense CG fl ash region.There was a distinct CG fl ash rate increase in hailstorms during the period of fast development,while a rapid reduction in the CG fl ash rate occurred in the dissipating stage.The hail fall corresponded to an active positive fl ash period,and the increase of the positive CG fl ash rate was generally accompanied by a decrease of the negative CG fl ash rate.The peak negative CG fl ash rate usually occurred 0–20 min earlier than hailstone fall,but the peak positive CG fl ash rate usually appeared at the time of or after the advent of hail fall.When the polarities of CG fl ashes changed,it often indicated the advent of severe weathersuch as hailfall,damagingwindandheavyprecipitation.Both the ratioof IC to CG fl ashes andIC fl ash densityin hailstorms were much larger than those in common thunderstorms.Most positive CG fl ashes usually occurred in or near the strong echo regions in hailstorms,but the CG fl ash density or CG fl ash rate were usually lower than those in common thunderstorms due to higher cloud top and frequent IC fl ashes.

Zheng et al.(2009)analyzed the characteristics of the lightning activity and electrical structure of a hailstorm in Beijing by using total lightning information from the SAFIR3000 3D lightning location system.The results indicated that the peak of the lightning rate came about fi ve minutespriortohailfall.Only6.2%ofthetotallightningwasCG fl ashes,among which 20%were positive.In the stage of hail fall,theelectricalstructureofthehailstormwasinverted,with the main negative charge region located around the ?40°C level and the main positive charge region around the ?15°C level.In addition,a weak negative charge region brie fl y existed below the positive charge region.After the hail fall,the electrical structure underwent fast and persistent adjustments and became a normal tripole structure.The lightning activity andelectrical structurewere closely related to the dynamic and microphysical processes of the hailstorm.It was believed that severe storms with stronger updrafts were more conduciveto an inverted tripolar electrical structure than normal thunderstorms,and the inverted distribution could then facilitate more+CC lightning in the severe storms.

Data from the Beijing SAFIR 3000 lightning detection system and Doppler radar have also provided some insights into the 3D lightning structure and evolution of a leadingline and trailing-stratiform(LLTS)MCS over Beijing(Liu et al.,2011;2013b).Most of the lightning in the LLTSMCS was IC lightning.Using CG location data,Feng et al.(2006b)and Liu et al.(2008)also found that almost all the CG f l ashes were negative in the f i rst developing stage,and the CG f l ash rate was high(more than 10 min?1)and negative CG f l ashes were predominant during the mature stage of the MCS.The CG f l ash rate declined rapidly with the increase of the positive CG f l ash ratio in the dissipating stage.The majority of CG lightning occurred in the convective region of the radar echo,particularly at the leading edge of the front.Little IC and positive CG lightning occurred in the stratiform region.During the storm’s development,most of the IC lightning occurred at an altitude of around 9.5 km above the ground and the IC lightning rate reached its maximum at 10.5 km above the ground,in the mature stage of the storm.When the thunderstorm began to dissipate,the altitude of the IC lightning decreased gradually.The spatial distribution of lightning was well correlated to the rainfall on the ground,although the peak value of rainfall appeared 75 min after the peak lightning rate(Liu et al.,2011).Convective region of the LLTS could be characterized by a tripole charge structure with a negative charge region in middle or a multi-layer charge structure with three layers of positive charge and a two-layer negative charge region in between(Liu et al.,2013a,b).

Fengetal.(2009)studiedatypicalsqualllinesystemwith damaging wind and hailstones causing great economic loss.It was shown that positive CG f l ashes accountedfor 54.7%of total CG f l ashes.During the initial developing stage,the CG fl ash rate was less than 0.5 min?1and most of the CG fl ashes were positive.It increased signi fi cantly,up to 4.5 min?1,along with the rapid development of the squall line,and the percentage of positive CG to total CG was more than 75%during this period.The CG fl ash rate began to decrease but the percentage of negative CG fl ashes to the total increased graduallyandexceededthat of positive CG duringthe mature and dissipating stages.positive CG fl ashes tended to occur on the right fl ank and negative ones on the left fl ank.Strong wind at the surface occurredin or near the regionswith dense positive CG fl ashes.Almost all positive CG fl ashes occurred near the strongradar echo regionsand in the frontparts of the squall line.However,the negative CG fl ashes almost exclusively occurred in the regions with weak and uniform radar echoes.The total fl ash rate in the storm was very high,up to 136 min?1,and the ratio of IC to CG fl ashes was 35:1.The CG distributionfeatures in the squall line were obviouslydifferent from those of ordinary MCSs.

Zhang et al.(2006b)found that the charge structures in the main part(convective region)of two supercell thunderstorms were the inverted tripole type.The positive CG fl ash dischargesoccurredinthe mainpartofthe thunderstormsand originatedfrom the positive chargeregionlocated in the middle part of the thunderstorms,while the negative CG fl ash discharges occurred in the anvil of the thunderstorm.The charge structure was the inverted dipole type in the anvil re-gionduetothe slant ofthe chargestructurein the mainregion towardtheanvilregion.Thenegativechargeregionlocatedin the upper part of the anvil produced many negative CG f l ash discharges.

Pan et al.(2009)examined the spatial and temporal distribution of lightning in seven typhoons over the Northwest Pacif i c using data from World Wide Lightning Location Network(WWLLN).They found three distinct f l ash activity regions in mature typhoons,a weak maximum in the eyewall regions(20–80 km from the center),a minimum between 80 and 200 km from the center,and a strong maximum in the outer rainbands(＞200 km radius).The lightning in the outer rainbands was greater in frequency than that in the inner rainbands,and less than 1%of f l ashes occurred within 100 km of the center.Few lightning f l ashes occurred near the center after landfall.Each typhoon produced eyewall lightningoutbreaksduringits intensif i cationperiodandbeforethe maximum intensity,indicating that lightning activity might be used as an indicator of typhoon intensity change.Zhang et al.(2012e),using the CG location data,also conf i rmed this kind of lightning distribution in Typhoon.Pan et al.(2014)studied lightning in 69 tropical cyclones over Northwest Pacif i c,and found that in more than half of the weak(Category 1–3)and strong(Category 4–5)typhoons,the peak value of lightning usually occurred before the maximum wind speed was attained.

5.2.Relationship of lightning with dynamical processes

Liu et al.(2009)found that most fl ashes of hailstorms occurred in the region with temperature lower than ?40°C,while dense positive CG fl ashes occurred in the region between ?40°C and ?50°C.Negative CG fl ashes occurred mostly in the relatively weak radar echo region,and positive CG fl ashes were distributed in the strong echo region,especially with a large gradient of echo intensity.The CG fl ashes tended to occur in the cloud region with re fl ectivity between 25 and 35 dBZ.

For the case of an MCS,Feng et al.(2006b)found that negative CG fl ashes mainly occurred in the region with temperature lower than ?50°C and a high temperature gradient in the front of the MCS,especially in cloud with temperature lower than ?60°C,but there were few CG fl ashes in the regionwith temperaturehigherthan?30°C.Therelationship between positive CG and negative CG fl ash number and cloudtopbrightnesstemperaturecouldbe fi tted preferablyby a three-power polynomial distribution.According to the appearance of peak values,the hourly fl ash rate lagged behind minimum brightness temperature and the area of cold cloud shield with temperature＜ ? 45°C lagged behind the hourly fl ash rate.The cloud top continued to extend horizontally shortly after the CG fl ash rate reached its maximum.Downburst and damaging winds were possibly producedwhen that the bow echo was associated with the jump in the CG fl ash rate.

Feng et al.(2009)found that dense positive CG fl ashes usually corresponded to updraft regions of the squall line system,and did not occur in the core of the updraft,but instead just behind and close to the main updraft.Negative CG fl ashes usually clustered in the intense echo(＞40 dBZ)region and their duration coincided with the strong convection,which suggestedthat negativeCG fl ashes couldbe usedto indicate the strongconvectiveregion.YuanandQie(2010)also foundthat,for a squall line system,most lightning fl ashes occurred in the regionof low brightness temperature,especially the region of lower than 200 K,and a few fl ashes could also be observed in the region of 240–260 K,which usually corresponded to the stratiform region of the squall line.

5.3.Relationship of lightning with the precipitationstructure of the thunderstorm

Feng et al.(2007)found that the probability of lightning occurrence was 20 times higher in the convective region than in the stratiform regionon the basis of TRMM data.The convective rain contributed much more rainfall to the total than stratiformrain,and the convectiverain representedmore than 85%of the total in two hailstorms.The results suggested that the vertical distribution of cloud water content,cloud ice water content and precipitation-sized ice content are helpful to judge the developing stage and to nowcast the weather system’s evolution.A linear relationship between f l ash rate and ice water content was obtained,and its correlationcoeff i cient was about 0.69.Most lightning f l ashes corresponded to regions with updraftat 5 km MSL,and the intensity of updrafts at 5 km MSL could be used as an indicator of lightning activity.

Yuan and Qie(2010)investigated lightning activity and its relationship with precipitation structure for a strong squall line over South China using TRMM satellite data.The results showed that most lightning f l ashes occurred near the strong convective region,and a few f l ashes occurred in the stratiform region of the squall line.There was a strong relationship between f l ash rate and ice precipitation content at 7–11 km MSL at the convective cell scale,and the correlation coeff i cient was 0.92.Yuan and Qie(2008)studied the lightning activity and precipitation characteristics during the South China Sea summer monsoon season,and found that when maximum radar ref l ectivity at 7 km MSL reached 36 dBZ,the probability of lightning occurrence was 50%in the pre-monsoon season,and increased to 38 dBZ in the monsoon season.The f l ash rate of precipitation systems could be expressed as a function of maximum storm top height,maximumsnow depth andminimumpolarizationcorrected temperatures(PCTs).Among those,the most stable was the relationship between f l ash rate and maximum snow depth.

5.4.Assimilation of lightning data and application in MCS

With the accumulation of high quality lightning location data,lightning data assimilation has become an important research topic.Recently,Qie et al.(2014b)established empirical relations between total lightningf l ash rate and the ice particle(graupel,ice,and snow)mixing ratio.The constructed nudging functions were applied in a MCS simulation with the Weather Research and Forecasting(WRF)model.They found that the representation of convection was signif i cantly improved one hour after the total lightning data assimilation,even during the assimilation period.The precipitation center,amount,and coverage were all much closer to the observation in the sensitivity run with lightning data assimilation than in the control run without lightning data assimilation.This simple and computationally inexpensive assimilation technique showed promising results and could be useful when the event is characterized by moderate to intense lightning activity,especially in the region where radar data is unavailable,for example in mountainous regions and over the oceans.

6.Effect on,and response of,lightning to climate change

The importance of lightning in climate studies has been increasingly recognized.The following three aspects in this fi eld are reviewed:(1)the lightning climatology in China and its surrounding areas;(2)lightning-induced NOx,which is very important not only for studies of atmospheric chemistry and climate change in both the free troposphere and planetary boundary layer,but also for understanding of the global nitrogen cycle;(3)the response of lightning to climate change.

6.1.Lightning climatology in China and its surrounding areas

Climatic characteristicsofglobalorregionallightningactivities have received much attention,and represent an active research topic.In the last decade,the variation of lightning activity in China has been studied using satellite-based LIS/OTD lightning data(Qie et al.,2003a,2003b;Yuan and Qie,2004;Ma et al.,2005a).There were four belts of lightning activity that run parallel to the seashore,near the sea region,central region,western region,and western boundary region.The lightning density distribution over mainland China showed a distinctive large-scale variation trend with distance from the coast and latitude,which was consistent with the annual mean precipitation variation trend.The Tibetan Plateau,China’s three-step staircase topography and latitude are three important factors affecting lightning distribution.The irregular distribution of lightning density was closely related to the irregular distribution of ground thermal and dynamical forcing.

The lightning activity on the Tibetan Plateau exhibited a seasonal variation in which it mainly occurred from June to August with a maximum f l ash activity period from late June to mid-July(Qie et al.,2003c).The diurnal variation of f l ash rate peaked during 1400–1600 LST,with the exceptions of the prominent high mountainous regions,which peaked earlier,and the prominent low basins,which peaked later.The lightning f l ashes over the Plateau responded strongly to the topography and surface thermodynamic features.Toumi and Qie(2004)found that the thermodynamic parameters and rainfall obtained from meteorological reanalysis data were broadly consistent with the observed seasonal cycle of lightning on the Tibetan Plateau.However,there was more lightning in spring than one might expect from a simple relationship with rainfall,temperature or cloud buoyancy.The cloud buoyancy and rainfall showed a better seasonal relationship when they were multiplied by the ratio of the sensible to latent heat fl ux(the Bowen ratio).This suggested that sensible heat fl ux plays an important role,at least on the Tibetan Plateau,in modifying the ef fi ciency of generating lightning from cloud buoyancy.

6.2.Lightning-induced NOx

The chemical processes in the troposphere caused by lightning activity are very important,and also complicated.Lightning discharges produce nitrogen oxides(LNOx).It is expected that with an increase of temperature,the total amount of LNOxwill also increase.However,due to the dif fi culty in measuring the exact amount of NOxgenerated by a single fl ash,it is dif fi cult to assign a certain concentration of NOxto a speci fi c lightning discharge in a speci fi c storm.

Zhou and Qie(2002),utilizing NOxanalyzer,observed the NOxevolution under conditions of a thunderstorm.The results showed that the peak values of the average volume mixing ratio of NOxin the air corresponded to the lightning fl ashes during the process of thunderstorm,but there were time lags between the peak values of volume mixing ratio of NOxand lightning f l ashes.The order of the transportation time of NOxgenerated by lightning could be f i tted with a quadratic relationship,and the coeff i cient was rather high,but no strict linear relationship between the transportation time and distance was found(Zhou et al.,2005).Zhang et al.(2014d)used a 3D-space cell-gridded approach to extract the lighting channel from VHF lightning locations,and a relationship between the NO productionper unit arc length and atmospheric pressure is applied to the NOxproduction.The averageNOxproductionsper CG and IC f l ash were estimated to be 1.89×1025and 0.42×1025molecules,respectively,in northeastern Qinghai-Tibet Plateau.The average annual total productionof LNOxin East Asia was about 2.30 Tg(Zhou et al.,2004).

Guo et al.(2006)discussed the transportation of LNOxby advection and turbulence in thunderclouds using a 3D dynamic–electrif i cationcoupledmodel.Theresults indicated that strong discharges were mostly located in the region of the upper edge of the middle negative charge region corresponding to the ambient region of updraft and center of the horizontal speed f i eld.LNOxwas transported by advection and turbulence after f l owing out from the discharge channel,and formed a density center in the weak wind f i eld.Due to the variety of lightning and thunderstorms,the parameters of lightningdischarge(energy,length,peakcurrent,channeltortuosity,initial altitude of leader,number of return strokes,etc.)also have a wide range of distribution.Therefore,the measurements in individual f i eld experiments may not be reliable for extrapolating globally.The global amount of NOxproduced by lightning and thunderstorms is still highly uncertain.

6.3.Response of lightning to climate change

Lightning activity,as a kind of extreme climate event,has drawn more attention in terms of its response to climate change.The global or regional relationships between lightning activity and some meteorological parameters,such as sea surface temperature,terrestrial surface temperature and relative humidity,have been examined based on lightning data observed by satellite-borne LIS/OTD and National Centers for Environmental Prediction(NCEP)meteorological data.

The responses of lightning f l ash rates to El Ni?no events and the interannual variation of surface wet bulb temperature and air temperature were studied by Ma et al.(2005b,2005c).During the 1997/98 El Ni?no event,a relatively signif i cant positive anomaly of lightning activity occurred in Asia/Australia and the Indian Ocean,and the maximum anomaly percentage reached 30%and 50%,respectively.One of the most sensitive positive anomaly areas was from southeastern China to the IndochinaPeninsula,where the position of the anomaly center for each season during the El Ni?no,as compared with normal years,had a westward shift;and,especially in winter or spring,there was a simultaneous northward shift.In addition,analysis of the interannual variation in the lightning density anomaly percentage,convective precipitation and high convective available potential energy(CAPE)days showed that each one among the three anomaly percentages was correlative with the other for the positive anomaly zone,and that the response of lightning activity to the El Ni?no event was the most sensitive.In 1997the anomaly percentage of the positive anomaly areas in winter reached 498%.

Research on whether global and regional lightning activity functions as a sensitive indicator of climate change has shown that,on the interannual time scale,the global total fl ash rate has hada positiveresponseto the variationin global surface air temperature,with a sensitivity of 17%K?1±7%K?1(Ma et al.,2005b).Also,the seasonal mean fl ash rate of continents all over the world,and that of continents in the NorthernHemisphere,hada sensitivepositiveresponseto the increase in global surface air temperature and wet bulb temperature,with a sensitivity of about 13%K?1±5%K?1.Although the increase in global lightning activity might serve as an indicator of the increase in global air temperature,this may not be the case on the regional scale.

Pan et al.(2013)found that the diurnal variation of lightning above the sea show two peak values,occurringin the afternoon and morning respectively.Xiong et al.(2005,2006)found that higher relative humidity resulted in more lighting activity in dry regions and less lighting activity in wet regions.The watershed of relative humidity for lightning production was about 72%–74%.Yuan and Qie(2008)found that the lightning activity over the South China Sea began to enhance in April,peaked in May,and decreased after June.Compared to the pre-monsoon season,the mean cell-level fl ash rate decreased by 13%and the mean fl ash optical radiance increased by 15%during the monsoon season,respectively.The mean fl ash rate was higher during the premonsoon season.The vertical development of precipitation systems in the pre-monsoon season was also stronger than that in the monsoon season,meaning frequent lightning activity was consequently observed.

The relationships between lightning activity and a series of convective indices have been investigated using 10-yr LIS lightning data over nine monsoon-proneareas in which highimpact weather events are frequently observed(Dai et al.,2009).Correlation analysis for each study area showed that a higher lightning fl ash rate and lightning probability were associated with more unstable air and smaller vertical wind shear in a nearly saturated lower layer in most of the studied regions.However,the correlation varied from region to region.The best correlation between lightning activity and the convective indices was found in eastern and southern China,whereas correlation was worst in some inland or basin regions in which topographic effects were more signif i cant.

Although many studies have revealed some objective facts that there exist certain correlations between climate change and lightning activity,the mechanisms and physical processes involved in these correlations remains unclear at present.The diurnal solar heating,the latitudinal temperature gradient,the general circulation of the atmosphere,the location of regions of convergenceand divergence,static and baroclinic instabilities etc.,all inf l uence the global distribution of thunderstorms.From short time scales(hourly,daily,monthly and annual)there seems to be obvious positive correlation between tropical lightning activity and surface temperature,upper tropospheric water vapor,cloud cover,and anvil ice content.Whether these relationships exist on longer time scales is still uncertain,althoughclimate models do support positive correlation between lightning and global temperature.Thus,with regard to research on the relationship between lightning and climate,not only the functions of temperature but also other factors should be taken into consideration.

As we know,increases in greenhouse gases can lead to climate change,which may increase the intensity of strong thunderstorms and lightning activity.Meanwhile,thunderstorms will increase the amount of water vapor and ice crystals in the upper troposphere.Due to the production of NOx,lightning activity increases the amount of O3and thus further increases the amount of greenhouse gases in the atmosphere.Therefore,an increase in lightning activity may make the climate even warmer throughthis positive feedback mechanism.Aerosolcorrelateswith thunderstormsand lightning activity,as well as with climate change.While aerosol inf l uences climate change,it also inf l uences the electrif i cation process to some extent by changing cloud microphysical characteristics.With the development of global lightning detection techniques,long-term observations of global lightning activity have been realized,from which important information concerning strong convection can be obtained.Such information,through its relationships with certain cli-mate parameters,couldprovidea useful referenceand further promote the application of lightning data in climate change research.

7.Numerical simulations of thunderstorm electrif i cation and lightning discharge

VHF observations of lightning discharges and multiparameter radar provide important knowledge for understanding the interaction between lightning discharge behavior and dynamic and microphysical f i elds in thunderclouds.However,the problem is that no technology in the foreseeable future is capable of simultaneously observing all of the dynamic,microphysical processes and E f i elds in evolving thunderstorms with high enough temporal and spatial resolution to delineate all their signif i cant behaviors and interactions.Numerical modeling is able to provide insight into thunderstormelectrif i cationand dischargeprocesses andhelp discriminate the interactions between dynamic,microphysical processes and lightning discharges.

7.1.Lightning discharge simulation with a 2D f i neresolution lightning model

A 2D f i ne-resolution(12.5 m)lightning model,which was modif i ed according to a stochastic lightning parameterization model(Mansell et al.,2002),was developed by Tan et al.(2006a,2006b,2007).The lightning discharge and electrif i cation scheme,including non-inductiveand inductive charge electrif i cation mechanisms,has been integrated into a 2D and 3D cumulus model.The hydrometeors considered in the model included cloud droplets,rain,ice crystals,snow,graupel and hail.The microphysical process and electrif i cation within a 250 m resolution were simulated in the thunderstorm domain.The cloud charge distributions at the f i ne resolution(12.5 m)were derived through an interpolation technique before the initiation of the lighting discharge.Figure 3 in the paper by Tan et al.(2006a)shows a simulated IC f l ash and the corresponding charge distribution background with a tripole charge structure before the f l ash initiation.The bilevel branched channel structures,horizontal extension and maximum changes of vertical E f i eld simulated by the f i neresolution lightning model were in good agreement with previous observational results than those from a coarser model.After IC f l ash initiation(black diamond)at the boundary between positive and negative potential zones,potential wells attracted the leaders of opposite polarities into the central area and prevented their outward expansion.It was possible for leaders to extend all throughout the opposite charge region,but they avoided the isolated charge area of the same polarity.Tao et al.(2009)added a CG f l ash scheme into the above mentioned 2D f i ne-resolution model,and produced the f i ne branched channel structure of a CG f l ash with different types of cloud charge distributions,such as dipole,tripole,bidipole and multi-layer charge structures to describe the relationships between CG f l ash channel propagation features and cloud charge distribution.The model results showed that the inducedchargesof oppositepolaritywere depositedin local volumes where the bidirectional leaders passed during a CG f l ash discharge.These charges were f i nally embedded in the pre-existing positive and negative dispersed cloud charge zones.This sub-process caused a more complicated charge distribution in thunderclouds,like a “multi-layer cake”.Although the embedding only affected the charge structure in a pair of positive and negative charge regions immediately next to the ground,the electrostatic energy of a thundercloud was signif i cantly consumed when the discharge terminated and the E f i eld strength weakened acutely.It was suggested that the observed bipolar CG f l ash is possibly due to the intense changes in electrical potential and polarity reversal of induced charges caused by the contact of the downward leader channel to the ground.

The subsequent neutralization of the residual charges in the channel volumes with surrounding dispersed cloud charges during the IC and CG f l ash was also discussed by Tan et al.(2007)and Tao et al.(2009).It was found that some residual charges were deposited in the local volumes of cut-off and non-conductingleader channels after the lightning discharge terminated and these charges were gradually neutralized with surrounding dispersed cloud charges.This process should relate to the turbulence exchange,advection transport,and gravitational sedimentation etc.in thunderclouds.The simulation also indicated that potential at initiationpointis akeytodecidewhetherdownwardleaderreaches ground(Tan et al.,2014).The absolute values of initiation potential of CG f l ashes are greater than 30 MV,while the absolute values of initiation potential of IC lightning are basically less than 30 MV.Since potential f i eld is determined by space charge distributions,polarities and types of lightning discharges are also dependent on relative locations and magnitudes of positive and negative charge zones near initiation points.

7.2.Simulation of charge structure with a 3D thunderstorm model

A3D dynamicsandelectrif i cationcoupledmodelwas developedtoinvestigatethecharacteristicsofmicrophysics,dynamics and electrif i cation inside thunderstorms(Sun et al.,2002).The model included a full treatment of small ions with attachment to six classes of hydrometeors(cloud drops,rain,ice,snow,graupel and hail),f i ve electrif i cation processes,which included ionic diffusion,electric attraction,inductive charging,non-inductive charging and secondary ice crystal charging.The larger precipitation particles were also forced by the vertical component of electrical force in electrically intensive thunderstorms.For the horizontal and vertical advection terms,fourth-order and second-order f i nite differences were used,and the super-relaxation method was used to determine electric potential.The model had the capacity to reproduce many of the observed characteristics of thunderstorms in dynamical,microphysical,and electrical aspects.

A regional thunderstorm model,coupled with two pri-mary non-inductive electrif i cation mechanisms,the Takahashi(1978)and Saunders et al.(1991)schemes,was developed and used to simulate a thunderstorm that occurred in Beijing,based on Regional Atmospheric Modeling System(RAMS)by Li et al.(2012)and Liu et al.(2014).The results were in agreement with some observations and results of other models.However,the evolution processes and shapes of cloud charge distribution in the two schemes were different.The result of Takahashi(1978)scheme produced a tripolar charge distribution in the cloud before the f i rst lightning discharge.The Saunders et al.(1991)scheme produced a transformation from an inverted dipole distribution to a tripolar charge distribution.The results from both schemes showed that the positive charge carrier at a low level of a thunderstormis rain droplets,that aggregatesand graupel are located at high levels,and that the charge center distribution of graupel is similar to the distribution of total charge in the thunderstorm cloud.Zhou and Guo(2009)also developed a 3D numerical model for simulating the electrif i cation and discharge processes in a hail storm and performedsimulation tests.

The effects of electrif i cation on microphysical and dynamical processes were examined by performingtwo numerical experiments,with and without electrif i cation processes(Sun et al.,2002).The model results showed that,when electrif i cation processes were included,the mass transfer among hydrometeors in microphysical processes,especially collection and coalescence processes,changed considerably as a result of signif i cant modif i cation of the terminal velocities of large precipitation particles.The change of mass transfer in microphysical processes affected cloud buoyancy by changing the amount and distribution of hydrometeors,and latent heat release in the middle region of the thunderstorm increased,i.e.convection strengthened by including electrifi cation processes.The amount and diameter of solid precipitation particles at the surface increased because a stronger updraft sustained large precipitation particles and prevented them from falling out of the cloud earlier.

The spatial and temporal evolution of charge structure and the interactions between electri fi cation,convection and rainfall have been studied numerically by Guo et al.(2007).The results indicated that the inductive and non-inductive charging mechanism played crucial roles in the evolution of electrical structure within thunderstorms,and the electrical development depended highly on ice-phase microphysical processes.The appearance time of the maximal E fi eld was the same as that of the maximal solid rainfall density and that of the maximal ascending velocity starting to decrease,but later than the maximal liquid rainfall density.

The effects of the temperature and relative humidity profi le on the charge structure in thunderclouds have also been analyzed by Guo et al.(2003)and Zheng et al.(2007).In southern China,the value of convective available potential energy(CAPE)was large,and the main positive and negative charge centers were raised to a high level,before a dipole charge structure formed.The humidity could also affect the charge structure.Enhanced middle relative humidity would increase the instability of the whole thunderstorm.A dipole charge structure corresponded to a maximum midlayer humidity,and a quasi polarity-invertedcharge structure to a minimum one.With reference to the effects of initial disturbance,it was noted that in the same temperature profi les,the lower maximumdisturbingcentral potentialtemperature(?θc)corresponded to a weak thunderstorm and quasi polarity-inverted charge structure,while higher?θccorresponded to a severe thunderstorm and dipole charge structure.For intermediate ?θcvalues,the storm had a tripole charge structure.

The effects of electrical activity on hail fall at the surface and hail growth during thunderstorms were simulated by Zhang et al.(2004b).The results indicated that,compared to duringthe non-electri fi cationprocess,hail with chargeand a strong E fi eld made precipitationincrease by 50%.Furthermore,the average diameter of the hail particles was 0.7 mm larger,and the time of hail fall laggedby three minutes.Electrical activity mainly in fl uenced the collection and melting process of hail.Since electri fi cation and discharge processes are very complex,comparisons of simulated results with observed data of the E fi eld distribution and lightning features during thunderstorms are necessary.

7.3.The f l ash rate simulated by numerical models at the thunderstorm scale

The f l ash rate at the scale of the thunderstorm has been studiedvianumericalsimulation.Assumingthatthe collision between ice crystals and rimed graupel particles was a dominant mechanism for charge separation in thunderstorms,Xie et al.(2005)studied the effect of two ice glaciation mechanisms of crystal(Fletcherand Hallett-Mossopglaciation)and liquid water content on f l ash rate.The results showed that there was a large disparity in ice crystal number concentration distribution with increased pressure and temperature in the two glaciation mechanisms,which directly resulted in a difference of electrical activity in the thundercloud.With an increase in liquid water content,the time of the f i rst lightning f l ash would be delayed,the location of the breakdown process would be lower,and the lightning f l ash rate would decrease.

8.Lightning detection and location techniques

Knowledge on lightning relies on the progress of lightning detectiontechnologies.EM radio-frequencydetectionis the main technology for detecting and locating lighting discharge sources,because lightning emits signif i cant EM radiation covering a very broad range of frequencies,from below 1 Hz to almost 300 MHz.In the microwave band(300 MHz to 300 GHz),and even in the visible light band(about 1014to 1015Hz),lightning is also detectable.Although the durations of various lightning processes are very short,the processes produce rich observable EM radiation.Lightning detection and location techniques in the frequency band at very low frequency(VLF)(3–30 kHz),low frequency(LF)(30–300 kHz),high frequency(HF)(3–30 MHz),VHF(30–300 MHz)and ultra-high frequency(UHF)(300–3000MHz)ranges have been developed worldwide.Each physical process in a lightning f l ash is associated with its characteristic electric and magnetic f i eld,so different techniques are used to detect different discharge processes.There are three EM radio-frequencylocation techniques that are most commonly used:magnetic direction f i nding(MDF),TOA,and interferometry.2D/3D lightning location systems based on longbaseline and short-baseline TOA VHF radiation pulse location techniques,and broadband and narrowband VHF interferometric,have been developedsuccessively for the purpose of lightning research and warning systems in the last decade in China.

8.1.TOA-based VHF radiation pulse location techniques

TOAtechnologylocatestheradiationpulsesemittedfrom lightning discharges,by measuring the time of arrival of the individual VHF pulses from lightning to different receivers.TOA-based location systems can be divided into two types:long-baseline and short-baseline.

A 3D lightning mapping system based on TOA and GPS technology was developed in China by Zhang et al.(2010).The principle of system was similar to the LMA(Rison et al.,1999),but worked at 270 MHz with a bandwidth of 6 MHz and was composed of seven aff i liated stations.The time and peak values were recorded every 25μs.The digitization rate was 20 MHz.The 40 MHz high precision clock was synchronized and calibrated by 1 pulse-per-second(PPS)output of a GPS receiver.3D images of lightning progression were obtained successfully for the f i rst time in China by using this system.The location error was estimated to be less than 50 m.

Fast and slow antennas,which respectively detect the fast and slow E f i eld changes produced by lightning discharge processes with microsecondor sub-microsecondtime resolution,are very useful in research on the physics of lightning.Multi-station observations of fast or slow antennas can also be used to locate lightning radiation sources using TOA technology.Through multi-station measurements of broadband slow antenna systems synchronized by GPS,some special featuresof lightningdischargesin the eastern TibetanPlateau were investigatedby Qie et al.(2000a).The bandwidthof the slow antenna was 0.2 to 2 MHz and the decay time constant was about 5 s.The E f i eld change signals were digitized at a sampling rate of 1 MHz with an amplitude resolution of 12 bits.Five observation stations covered an area of 10×10 km.The K-type breakdown processes during one IC discharge were found to develop from the main negative charge regionto theLPCC byusing theTOA techniques.Wang et al.(2009a)developed a similar TOA algorithm-based lightning locationsystem using a fast antennasystem with a bandwidth of 0.1 to 5 MHz and a time constant of 2 ms.The system contained seven stations with baselines of several kilometers.Thehightime-resolutionGPSwithatimingaccuracyof50ns was usedto synchronizethe signals fromeachstation.Radiation pulses in the initial stages of f i ve IC lightning discharges werewell locatedin3D,andtheradiationsourceswerefound to be nicely associated with the radar echo of the storm,indicating the technique could effectively locate the lightning radiation sources.

Zhang et al.(2003a)developed a short-baseline TOA lighting radiation detection system with antenna separation of 10 m,central frequency of 280 MHz and bandwidth of 6 MHz.The signals were recorded with a sampling rate of 2 GHz and a record length of 500 MB.The segmented triggering mode was used to resolve the conf l ict of a very high sampling rate and relatively small capacity.Cao et al.(2012)modif i ed the hardware with wide-band receivers(125–200 MHz)and data processing software of the system.To reduce noise and improve estimation accuracy of time delay,a general correlation time delay estimation algorithm based on direct correlation method and wavelet transform was adopted by Sun et al.(2013).Moreover,parabolic interpolation algorithm was used in the fractional delay estimation to improve the time resolution of the positioning system.The location results for a rocket-triggered lightning and an IC lightning indicated that the modif i ed short-baseline time-difference of arrival(TDOA)technology could effectively map the lightning radiation sources in 2D.Short-baseline TOA VHF radiation pulse location systems can only locate the azimuth and elevation of the lightning discharge sources occurring nearby.Time synchronization is not a problem in this kind of system,which is,in contrast,very critical in multi-station long-baseline systems.Therefore,the cost of this system is relatively cheap and suitable for locating local lightning dischargesin some keyareas.However,it neverthelesshassome shortcomings,such as only providing 2D location information,its relativelylargeelevationlocationerror,andrelatively short detection range.Essentially,the short-baseline timedifference of arrival technology is interferometer.

8.2.VHF/UHF interferometer lightning location systems

Lightning usually produces some noise-like bursts of EM radiation lasting for tens to hundreds of microseconds.It is very dif fi cult to identify individual pulses from these bursts.The interferometer measures the phase difference between the signals from different sensors,and provides an ef fi cient way for locating the noise-like pulses.The VHF/UHF interferometer technique locates the azimuth and elevation of the radiation source by using three to four antennas with two orthogonal baselines.To locate the sources in 3D,two or more synchronized interferometers are needed.

Dong et al.(2001,2002)developed a broadband interferometer system by employing three identical broadband antennas,which were separated horizontally with distances of 10 m.These antennas were located at three apexes of a square and all connected to a Digital Storage Oscilloscope(DSO)via a 50-m-long coaxial cable,terminated with characteristic impedance of 50 ? and through 25 MHz high pass fi lters.The sampling rate was 1 GHz,and the memory of each channel was divided into 2000 segments.Each segment recorded up to 1000 points.By using this system,the positive leader and negative breakdown processes in arti fi-cially triggered lightning and bidirectional breakdown processes in natural lightning were observed(Dong et al.,2002,2003).Qiu et al.(2009)proposed a phase f i ltering algorithm which combined circular correlation with translationinvariant de-noising for the VHF broadband interferometry.The application of this algorithm can produce clearer 2D images of lightning discharges than the conventional algorithm did.

For broadband interferometer systems,a high speed digitizer and a large capacity recording system are needed to record broadband signals.Generally,the segmented triggering mode for each event is used to reduce the capacity requirement,so it is hard to record the lightning f l ash continuously.Furthermore,it is diff i cult to have two or more completely identical coaxial cables for a broadband interferometer.Therefore,the system error of broadbandinterferometers could be a little larger than that of narrowband interferometers.

Zhang et al.(2008)developeda narrowbandinterferometer system using a f i ve-antenna array consisting of short and long baselines along two orthogonal directions.The interferometer was operated at a central frequency of 280 MHz with a 3 dB bandwidth of 6 MHz.The signal received by the central antenna of the array was separately interfered with the signals from all the remainingantennas.These output signals were digitized with a sampling rate of 1 MHz and a resolution of 16 bits.The system error that arose from frequency conversion was reduced through phase detection by directly using high frequency amplif i ers.An interactive graphic analysis procedurewas usedto removethe fringeambiguitiesthat exist inherently in interferometry and to determine the direction of lightning radiation sources in 2D as a function of time with a resolution in the order of microseconds.

9.Transient luminous events above thunderstorms

Transient luminous events(TLEs)are very short-lived discharges that occur above thunderstorms,e.g.sprites,elves and jets.These f l eeting optical emissions in the mesosphere can initiate from the tops of thunderclouds up to the ionosphere,providing direct evidence of coupling from the lower atmosphere to the upper atmosphere.The f i rst ground-based video recordings of sprites were obtained in northern America in 1989(Franz et al.,1990).Since then,several groundand aircraft-borne observations have been used to explore these kinds of discharges.In addition to sprites,another two types of discharges,termed elves and jets(blue jets,gigantic jets),were recorded.The term “Elves”refers to “Emission of Light and VLF”perturbations from an EM pulse(EMP)source,which appear as a ring(at 90 km altitude)and can spread laterally at the speed of light over 300 km.Blue jets emanate from the top of thundercloudsup to an altitude of 40 km.Gigantic jets propagate upwards from thunderclouds to altitudes of about 90 km(Su et al.,2003).Observations of TLEs have been conducted recently in China.

9.1.Observationalstudies on sprites associatedwith lightning

The Chinese Sprites Observation Campaign(CSOC),aimed at understanding the characteristics of sprites over mainland China and the relationships between sprites and lightning fl ashes,has been conducted since the summer of 2007.The fi rst observationsite is locatedin ZhanhuaCounty,ShandongProvince(37°49′42′N,118°05′06′E).The camera system used in the campaign is Watec902H camera,and the fi eld of view(FOV)of the observation system is 31.1°(horizontal)by 21.2°(vertical).

A total of 17 sprites were fi rst observed over two thunderstorms in 2007(Yang et al.,2008c).All of the observed sprites occurred in a cluster,and their appearances were different,including “columniform sprites”,“columniform sprites”with angel-like wings,“carrot sprites”,“dancingsprites”,etc.Theestimatedbottomelevationofoneofthe columniform sprites was about 47±12 km and the top was 86±15 km.The vertical length of one of the carrot sprites was about 42 km,with the bottom at 39±10 km and the top at 81±14 km.The duration of the sprites varied from 40 ms to 160 ms,with a mean value of 61 ms.

Yang et al.(2013a,2013b)studied the characteristics of sprite-producing and non-sprite-producing summer thunderstorms.They found that the average positive CG lightning peak current in sprite-producing storms was larger than that in the non-sprite-producing one.The convection was also stronger in sprite-generated thunderstorms,but there was no obvious difference in the microphysical characteristics.The parental CGs of sprites were positive and located in regions with a cloud top temperature of?40°C to ?60°C and radar re fl ectivity of 15–35 dBZ.Most of the sprites appearedin the period characterized by a sharp decrease in negativeCGs and an increase in positive CGs.

9.2.Observational studies on Gigantic jets

Gigantic jet is a type of large-scaled transient discharge which occurs above thunderstorms.It connects the thunderstorms and ionosphere directly.Compared with sprites,gigantic jet is very dif fi cult to be observed from the ground.Yang and Feng(2012)reported a gigantic jet event in eastern China,near the Huanghai Sea.The top altitude of this gigantic jet was estimated to be about 89 km.The gigantic jet-producingstormwas a multi-cellthunderstormand the gigantic jet event occurred in the storm developing stage,with the maximum radar echo top around 17 km.Different from results from other countries that positive CGs dominatedduring the time period of gigantic jet.It is by far the furthest from the equator gigantic jet recorded over summer thunderstorm.

Although some studies on TLEs have been conducted in the past few years in China,more cases are needed to provide statistically reliable characteristics of TLEs and associated lightning and thunderstorms.There are also signi ficant questions raised that deserve study in the years to come.One question is the in fl uence of TLEs,especially in terms of quantitative estimations of the chemical effect on the mesosphere.Another question is the E fi eld established by a lightning discharge;no existing model considers the complete E fi elds produced by CG and IC discharges.Furthermore,the in fl uence of TLEs on Earth’s environment and space weather should also be studied.

1 0.Concluding remarks

New understanding on the physics of lightning and thunderstorm electricity has been achieved during the last decade through continuous observations in some representative weather system regions in China.The f i rst lightning experiment on the Tibetan Plateau clearly revealed the special charge structure and discharge phenomenon.Some exciting results have been documented in aspects of in situ soundings of thunderstorms and lightning mapping technologies.Progresses has also been achieved in the area of rocket-triggered lightning and its application.However,our understanding of lightning activity and atmospheric electricity is still limited and not satisfactory for lightning protection in the context of modern society.Some key questions remain unanswered,such as the predominant electrif i cation mechanism in different thunderstorms with very different manifestations of precipitation,the connection processes between CG discharges and ground objects,the physical mechanism of sprites and narrow bipolar events,and the long-term response of lightning and the global circuit to temperature change.High quality detection of lightning,in association with thunderstorm microphysics and dynamics,and long-term data accumulation will serve as crucial measures.

Acknowledgements.This research was supported by the National Key Basic Research and Development(973)Program of China(2014CB441400)and the National Natural Science Foundation of China(Grant No.41475002).

Cao,D.J.,G.S.Zhang,T.Zhang,and Y.H.Wang,2008:Analyseson VHFradiationof Cloud-to-Ground f l ashesinPingliang loess plateau.Plateau Meteorology,27(2),365–372.(in Chinese)

Cao,D.J.,X.S.Qie,S.Duan,Y.J.Xuan,and D.F.Wang,2012:Lightning discharge process based on short-baseline lightning VHF radiation source locating system.Acta Physica Sinica,61(6),069202.(in Chinese)

Chen,C.P.,X.S.Qie,Q.Zhang,G.S.Zhang,and T.Zhang,2005:Analysisof VHFradiations of discharging of Cloud-to-Ground f l ashes.Proceedings of the CSEE,25(19),122–126.(in Chinese)

Cui,H.,X.Qie,Q.Zhang,T.Zhang,G.Zhang,and J.Yang,2009:Intracloud discharge and the correlated basic charge structure of a thunderstorm in Zhongchuan,a Chinese Inland Plateau region.Atmospheric Research,91(2–4),425–429.

Dai,J.,and Coauthors,2009:A comparison of lightning activity and convective indices over some monsoon-prone areas of China.Atmospheric Research,91,438–452.

Dong,W.S.,X.S.Liu,Y.Yu,and Y.J.Zhang,2001:Broadband interferometer observations of a triggered lightning.Chinese Sci.Bull.,46(18),1561–1565.

Dong,W.S.,X.S.Liu,Y.J.Zhang,and G.S.Zhang,2002:Observations on the leader-return stroke of cloud-to-ground lightning with the broadband interferometer.Sci.China:Earth Sci.,45(3),259–269.

Dong,W.S.,X.S.Liu,C.X.Chen,Y.J.Zhang,and H.B.Wang,2003:Broadband interferometer observations of the bi-directional breakdown process in natural lightning.Chinese J.Geophys.,46(3),449–456.

Fan,X.P.,G.S.Zhang,Y.H.Wang,Y.J.Li,T.Zhang and B.Wu,2014:Analyzing the transmission structures of long continuing current processes from negative ground f l ashes on the Qinghai-Tibetan Plateau.J.Geophys.Res.,119(5),2050–2063.doi:10.1002/2013JD020402

Feng,G.L.,X.S.Qie,T.Yuan,and Y.J.Zhou,2006a:A case study of cloud-to-ground lightning activities in hailstorms under cold eddy synoptic situation.Acta Meteorologica Sinica,20(4),489–499.

Feng,G.L.,X.S.Qie,and Y.J.Zhou,2006b:A case study on characteristics of lightning activity in a mesoscale convective system.Plateau Meteorology,25(2),220–228.(in Chinese)

Feng,G.L.,X.S.Qie,T.Yuan,and S.Z.Niu,2007:Analysis on lightning activity and precipitation structure of hailstorms.Science China-Earth Science,50(4),629–639.

Feng,G.L.,X.S.Qie,and S.J.Wu,2008:Cloud-to-ground lightning characteristics of hail clouds in Shandong province.Chinese.J.Atmos.Sci.,32(2),289–299.(in Chinese)

Feng,G.L.,X.S.Qie,J.Wang,and D.L.Gong,2009:Lightning and Doppler radar observations of a squall line system.Atmospheric Research,91(2–4),466–478.

Franz,R.C.,R.J.Nemzek,and J.R.Winckler,1990:Television image of a large upward electrical discharge above a thunderstorm system.Science,249(4964),48–51.

Gou,X.Q.,Y.J.Zhang,X.S.Qie,and W.S.Dong,2006:Characterizationofradiationf i eldofreturnstrokeof cloud-to-ground lightning with local effective Holder exponent.Proceedings of the CSEE,26(3),75–78.(in Chinese)

Gou,X.Q.,Y.J.Zhang,W.S.Dong,and X.S.Qie,2007:Waveletbased multifractal analysis of the radiation f i eld of f i rst return stroke in cloud-to-ground discharge.Chinese J.Geophys.,50(1),101–105.(in Chinese)

Gou,X.Q.,M.L.Chen,Y.J.Zhang,W.S.Dong,and X.S.Qie,2009:Wavelet multiresolution based multifractal analysis of electric f i elds by lightning return strokes.Atmospheric Research,91,410–415.

Guo,F.X.,Y.J.Zhang,X.S.Qie,and M.H.Yan,2003:Numerical simulation of different charge structures in thunderstorm.Plateau Meteorology,22(3),268–274.(in Chinese)

Guo,F.X.,M.H.Yan,and Y.J.Zhang,2006:A model study of lightning-generated NOx(LNOx)and it’s transportation.Plateau Meteorology,25(2),229–235.(in Chinese)

Guo,F.X.,Y.J.Zhang,and M.H.Yan,2007:A numerical study of the charge structure in thunderstorm in Nagqu area of the Qinghai–Xizang Plateau.Chinese J.Atmos.Sci.,31(1),28–36.(in Chinese)

Jiang,R.,X.Qie,C.Wang,J.Yang,Q.Zhang,J.Wang,and D.Liu,2011:Lightning M-components with peak current in the range of kilo amperes and their mechanism.Acta Physica Sinica,60(7),079201.(in Chinese)

Jiang,R.,X.Qie,C.Wang,and J.Yang,2013a:Propagating fea-tures of upward positive leaders in the initial stage of rockettriggered lightning.Atmospheric Research,129–130,90–96.

Jiang,R.,X.Qie,J.Yang,C.Wang,and Y.Zhao,2013b:Characteristics of M-component in rocket-triggered lightning and a discussion on its mechanism.Radio Sci.,48,597–606.doi:10.1002/rds.20065.

Jiang,R.,Z.Wu,X.Qie,D.Wang,and M.Liu,2014a:High-speed video evidence of a dart leader with bidirectional development.Geophys.Res.Lett.,41,doi:10.1002/2014GL060585.

Jiang,R.B.,X.S.Qie,Z.J.Wu,D.F.Wang,M.Y.Liu,G.P.Lu,and D.X.Liu,2014b:Characteristics of upward lightning from a 325-m-tall meteorology tower.Atmospheric Research,149,111–119,doi:10.1016/j.atmosres.2014.06.007.

Kong,X.Z.,X.S.Qie,G.S.Zhang,and T.Zhang,2005:Research on stepped-leader and return-stroke of the cloud-toground lightning with multiple-grounding-point in the same return stroke.Proceedings of the CSEE,25(22),142–147.(in Chinese)

Kong,X.Z.,X.S.Qie,Y.Zhao,T.Zhang,G.S.Zhang,and W.S.Dong,2006:An analysis of discharge processes of one cloudto-ground lightning fl ashon theQinghai-Xizang Plateau.Chinese J.Geophys.,49(4),993–1000.(in Chinese)

Kong,X.Z.,X.S.Qie,and Y.Zhao,2008:Characteristics of downward leader in a positive cloud-to-ground lightning fl ash observed by high-speed video camera and electric fi eld changes.Geophys.Res.Lett.,35,L05816,doi:10.1029/2007 GL032764.

Kong,X.Z.,X.S.Qie,Y.Zhao,and T.Zhang,2009:Characteristics of negative lightning f l ashes presenting multiple-ground terminations on a millisecond-scale.Atmospheric Research,91(2–4),381–386.

Li,D.S.,Q.L.Zhang,Z.H.Wang,and T.Liu,2014:Computation of lightning horizontal f i eld over the two-dimensional rough ground by using the three-dimensional FDTD.IEEETransactions on Electromagnetic Compatibility,56(1),143–148,doi:10.1109/TEMC.2013.2266479.

Li,W.L.,D.X.Liu,X.S.Qie,S.M.Fu,S.Duan,and Y.C.Chen,2012:Evaluation of noninductive charging mechanisms and simulation of charge characteristic structure in the early thunderstorm based on RAMS V6.0.Acta Physica Sinica,61(5),059202,doi:10.7498/aps.61.059202.(in Chinese)

Li,Y.J.,G.S.Zhang,J.Wen,D.H.Wang,Y.H.Wang,T.Zhang,X.P.Fan,and B.Wu,2013:Electrical structure of a Qinghai-Tibetan Plateau thunderstorm based on threedimensional lightning mapping.Atmospheric Research,127,90–115.

Liu,D.X.,X.S.Qie,G.L.Feng,and S.J.Wu,2008:Analyses on lightning temporal and spatial characteristics in the severeconvective weather innorth China.PlateauMeteorology,27(2),358–364.(in Chinese)

Liu,D.X.,G.L.Feng,and S.J.Wu,2009:Temporal and spatial characteristics of cloud-to-ground lightning of hailstorms over north China.Atmospheric Research,91(2–4),459–465.

Liu,D.X.,X.S.Qie,Y.J.Xiong,and G.L.Feng,2011:Evolution of the total lightning activity in a leading-line and trailing stratiform mesoscale convective system over Beijing.Adv.Atmos.Sci.,28(4),866–878,doi:10.1007/s00376-010-0001-8.

Liu,D.X.,X.S.Qie,L.X.Pan,and L.Peng,2013a:Some characteristics of lightning activity and radiation source distribution in asquall lineover north China.Atmospheric Research,132–133,423–433.

Liu,D.X.,X.S.Qie,Z.C.Wang,X.K.Wu,and L.X.Pan,2013b:Characteristics of lightning radiation source distribution and charge structure of squall line.Acta Physica Sinica.,62(21),219 201–219 201.

Liu,D.X.,X.S.Qie,L.Peng,and W.L.Li,2014:Charge structure of a summer thunderstorm in North China:Simulation using a regional atmospheric model system.Adv.Atmos.Sci.,31,1022–1034,doi:10.1007/s00376-014-3078-7.

Liu,H.Y.,W.S.Dong,T.Wu,D.Zheng,and Y.J.Zhang,2012:Observation of compact intracloud discharges using VHF broadband interferometers.J.Geophys.Res.,117,D01203,doi:10.1029/2011JD016185.

L¨u,F.,B.Zhu,H.Zhou,V.A.Rakov,W.Xu,and Z.Qin,2013:Observations of compact intracloud lightning discharges in the northernmost region(51°N)of China.J.Geophys.Res.Atmos.,118,4458–4465,doi:10.1002/jgrd.50295.

L¨u,W.,D.Wang,N.Takagi,V.Rakov,M.Uman,and M.Miki.,2008a:Characteristics of the optical pulses associated with a downward branched stepped leader.J.Geophys.Res.,113,D21206,doi:10.1029/2008JD010231.

L¨u,W.T.,and Coauthors,2008b:Analysis of channel luminosity characteristics in rocket-triggered lightning.Acta Meteorologica Sinica,22(3),362–374.

Lu,W.T.,and Coauthors,2009:Simultaneous optical and electrical observations on the initial processes of altitude-triggered negative lightning.Atmospheric Research,9,353–359.

Lu,W.,L.Chen,Y.Zhang,Y.Ma,Y.Gao,Q.Yin,S.Chen,Z.Huang,and Y.Zhang,2012:Characteristics of unconnected upward leaders initiated from tall structures observed in Guangzhou.J.Geophys.Res.,117,D19211,doi:10.1029/2012JD018035.

Lu,W.,L.Chen,Y.Ma,V.A.Rakov,Y.Gao,Y.Zhang,Q.Yin,and Y.Zhang,2013:Lightning attachment process involving connection of the downward negative leader to the lateral surface of the upward connecting leader.Geophys.Res.Lett.,40,5531–5535,doi:10.1002/2013GL058060.

Ma,M.,S.C.Tao,B.Y.Zhu,and W.T.Lu,2005a:Climatological distribution of lightning density observed by satellites in China and its circumjacent regions.Science China-Earth Science,48(2),219–229.

Ma,M.,S.C.Tao,B.Y.Zhu,W.T.Lu,and Y.B.Tan,2005b:Response of global lightning activity to air temperature variation.Chinese Science Bulletin,50(22),2640–2644.

Ma,M.,S.C.Tao,B.Y.Zhu,and W.T.Lu,2005c:The anomalous variation of the lightning activity insouthern China during the 1997/98 El Ni?no event.Science China-Earth Science,48(9),1537–1547.

Mansell,E.R.,D.R.MacGorman,C.L.Ziegler,and J.M.Straka,2002:Simulated three-dimensional branched lightning in a numerical thunderstorm model.J.Geophys.Res.,107(D9),doi:10.1029/2000JD000244.

Ouyang,Y.H.,P.Yuan,X.S.Qie,H.B.Wang,X.D.Jia,and H.M.Zhang,2006:Temperature study on lightning return stroke in the coastal area of Guangdong.Spectroscopy and Spectral Analysis,26(11),1988–1992.

Pan,L.X.,X.S.Qie,D.X.Liu,D.F.Wang,and J.Yang,2009:The lighting activities in super typhoons over the Northwest Pacif i c.Science China-Earth Science,53(8),1241–1248.

Pan,L.X.,D.X.Liu,X.S.Qie,D.F.Wang,and R.P.Zhu,2013:Land-sea contrast in the lightning diurnal variation as observed by the WWLLN and LIS/OTD data.Acta Meteorologica Sinica,27,591–600.

Pan,L.X.,X.S.Qie,and D.F.Wang,2014:Lightning activity and its relation to the intensity of typhoons over the Northwest Pacif i c Ocean.Adv.Atmos.Sci.,31,581–592,doi:10.1007/s00376-013-3115-y.

Qie,X.S.,and X.Z.Kong,2007:Progression features of astepped leader process with four grounded leader branches.Geophys.Res.Lett.,34,L06809,doi:10.1029/2006GL028771.

Qie,X.,S.Soula,andS.Chauzy,1994:Inf l uenceof ionattachment on thevertical distribution of the electricf i eldand charge density below a thunderstorm.Annales Geophysicae,12,1218–1228.

Qie,X.,Y.Yu,X.Liu,C.Guo,D.Wang,T.Watanabe,and T.Ushio,2000a:Charge analysis on lightning discharges to the ground in Chinese inland plateau(verge of Tibet).Ann.Geophys.,18(10),1340–1348.

Qie,X.S.,Y.Yu,X.S.Liu,G.S.Zhang,D.H.Wang,T.Watanabe,N.Takagi,and T.O.Ushio,2000b:K-Type breakdown process of intracloud discharge in Chinese inland plateau.Progress in Natural Science,10(8),607–611.

Qie,X.S.,Y.Yu,H.B.Wang,and C.H.Zhang,2001:Analysis on some features of ground f l ashes in Chinese inland plateau.Plateau Meteorology,20(4),395–401.(in Chinese)

Qie,X.,Y.Yu,C.Guo,P.Laroche,G.Zhang,and Q.Zhang,2002a:Some features of stepped and dart-stepped leaders near the ground in natural negative cloud-to-ground lightning discharges.Ann.Geophys.,20(6),863–870.

Qie,X.,Y.Yu,D.Wang,H.Wang,and R.Chu,2002b:Characteristics of cloud-to-ground lightning in Chinese Inland Plateau.J.Meteor.Soc.Japan,80(4),745–754.

Qie,X.S.,R.Toumi,and T.Yuan,2003a:Lightning activities on the Tibetan Plateau as observed by the lightning imaging sensor.J.Geophys.Res.,108(D17),4551,do:10.1029/2002JD003304.

Qie,X.S.,Y.J.Zhou,and T.Yuan,2003b:Global lightning activities and their regional differences observed from the satellite.Chinese J.Geophys.,46(6),743–750.(in Chinese)

Qie,X.S.,R.Toumi,and Y.J.Zhou,2003c:Lightning activity on the central Tibetan Plateau and its response to convective available potential energy.Chinese Sci.Bulletin,48(4),296–299.

Qie,X.S.,T.Yuan,Y.R.Xie,and Y.M.Ma,2004:Spatial and temporal distribution of lightning activities over the Tibetan Plateau.Chinese J.Geophys.,47(6),997–1002.(in Chinese)

Qie,X,and Coauthors,2005a:The possible charge structure of thunderstorm and lightning discharges in northeastern verge of Qinghai-Tibetan Plateau.Atmospheric Research,76(1–4),231–246.

Qie,X.S.,T.L.Zhang,C.P.Chen,G.S.Zhang,T.Zhang,and W.Z.Wei,2005b:The lower positive charge center and its effect on lightning discharges on the Tibetan Plateau.Geophys.Res.Lett.,32,L05814,doi:10.1029/2004GL022162.

Qie,X.S.,Q.L.Zhang,and Y.J.Zhou,2007:Artif i ciallytriggered lightning and its characteristic discharge parameters in two severe thunderstorms.Science China-Earth Science,50(8),1241–1250.

Qie,X.S.,and Coauthors,2009a:Characteristics of artif i cially triggeredlightning during Shandong artif i cialtriggering lightning experiment(SHATLE).Atmospheric Research,91(2–4),310–315.

Qie,X.S.,T.L.Zhang,G.S.Zhang,T.Zhang,and X.Z.Kong,2009b:Electrical characteristics of thunderstorms in different plateau regions of China.Atmospheric Research,91(2–4),244–249.

Qie,X.S.,J.Yang,R.B.Jiang,J.F.Wang,D.X.Liu,C.X.Wang,and Y.J.Xuan,2010:A new-model rocket for artif i ciallytriggering lightning and its f i rst triggering lightning experiment.Chinese J.Atmos.Sci.,34(5),937–946.(in Chinese)

Qie,X.S.,R.B.Jiang,C.X.Wang,J.Yang,J.F.Wang,and D.X.Liu,2011:Simultaneously measured current,luminosity,and electric f i eld pulses in a rocket-triggered lightning f l ash.J.Geophys.Res.,116,D10102,doi:10.1029/2010JD015331.

Qie,X.,Z.Wang,D.Wang,M.Liu,and Y.Xuan,2013:Characteristics of positive cloud-to-ground lightning in DaHinggan Ling forest region at relatively high latitude,northeastern China.J.Geophys.Res.Atmos.,118,13 393–13 404,doi:10.1002/2013JD020093.

Qie,X.,R.Jiang,and J.Yang,2014a:Characteristics of current pulses in rocket-triggered lightning.Atmospheric Research,135–136,322–329,doi:http://dx.doi.org/10.1016/j.atmosres.2012.11.012.

Qie,X.S.,R.P.Zhu,T.Yuan,X.K.Wu,W.L.Li,and D.X.Liu,2014b:Application of total-lightning data assimilation in a mesoscale convective system based on the WRF model.Atmospheric Research,145–146,255–266,doi:10.1016/j.atmosres.2014.04.012

Qiu,S.,B.-H.Zhou,L.-H.Shi,W.-S.Dong,Y.-J.Zhang,and T.-C.Gao,2009:An improved method for broadband interferometric lightning location using wavelet transforms.J.Geophys.Res.,114,D18211,doi:10.1029/2008JD011655.

Rakov,V.A.,R.Thottappillil,M.A.Uman,and P.P.Barker,1995:Mechanism of the lighting M component.J.Geophys.Res.,100,25 701–27 710.

Rison,W.R.,R.J.Thomas,P.R.Krehbiel,T.Hamlin,and J.Harlin,1999:A GPS-based three-dimensional lightning mapping system:Initial observations in central New Mexico.Geophys.Res.Lett.,26,3573–3576.

Saunders,C.P.R.,W.D.Keith,and R.P.Mitzeva.,1991:The effect of liquid water on thunderstorm charging.J.Geophys.Res.,96(D6),11 007–11 017.

Su,H.T.,and Coauthors,2003:Gigantic jets between a thundercloud and the ionosphere.Nature,423,974–976

Sun,A.P.,H.-Y.Chun,J.-J.Baik,and M.H.Yan,2002:Inf l uence of electrif i cation on microphysical and dynamical processes in a numerically simulated thunderstorm.Quart.J.Appl.Meteor.,41,1112–1127.

Sun,Z.L.,X.S.Qie,M.Y.Liu,D.J.Cao,and D.F.Wang,2013:Lightning VHF radiation location system based on short-baseline TDOA technique—Validation in rockettriggered lightning.Atmospheric Research,129–130,58–66,doi:10.1016/j.atmosres.2012.11.010.

Takahashi,T.,1978:Riming electrif i cation as a charge generation mechanism in thunderstorms.J.Atmos.Sci.,35,1536–1548.

Tan,Y.B.,S.C.Tao,and B.Y.Zhu,2006a:Fine-resolution simulation of the channel structures and propagation features of intra-cloud lightning.Geophys.Res.Lett.,33,L09809,doi:10.1029/2005GL025523.

Tan,Y.B.,S.C.Tao,B.Y.Zhou,M.Ma,and W.T.Lu,2006b:Numerical simulation of the bi-level and branched structure of intra-cloud lightning f l ashes.Science China-Earth Science,49(6),661–672.

Tan,Y.B.,S.C.Tao,B.Y.Zhou,M.Ma,and W.T.Lu,2007:A simulationoftheeffectsof intra-cloudlightningdischargeson the charges and electrostatic potential distributions in a thundercloud.Chinese J.Geophys.,50(4),916–930.(in Chinese)

Tan,Y.,S.Tao,Z.Liang,and B.Zhu,2014:Numerical study on relationship between lightning types and distribution of space charge and electric potential.J.Geophys.Res.Atmos.,119,1003–1014,doi:10.1002/2013JD019983.

Tao,S.C.,Y.B.Tan,B.Y.Zhu,M.Ma,and W.T.Lu,2009:Fineresolution simulation of cloud-to-ground lightning and thundercloud charge transfer.Atmospheric Research,91,360–370.

Toumi,R.,and X.S.Qie,2004:Seasonal variation of lightning on the Tibetan Plateau:A spring anomaly.Geophys.Res.Lett.,31,L04115,doi:10.1029/2003GL018930.

Wang,C.X.,R.B.Jiang,J.Yang,and M.Y.Liu,2012a:Current subsidiary peak in triggered lightning strokes.Radio Sci.,47,RS4002,doi:10.1029/2011RS004933.

Wang,C.X.,X.S.Qie,R.B.Jiang,and J.Yang,2012b:Propagating properties of a upward positive leader in a negative triggered lightning.Acta Physica Sinica,61(3),039203.(in Chinese)

Wang,D.F.,X.S.Qie,T.Yuan,G.S.Zhang,T.Zhang,T.L.Zhang,and Q.L.Zhang,2009a:An analysis of the initial stage of intracloud lightning using the pulse location technique based on the fast electric f i eld change.Acta Meteorologica Sinica,23(6),772–781.

Wang,J.,P.Yuan,F.X.Guo,X.S.Qie,Y.H.Ouyang,and Y.J.Zhang,2009b:The spectra and temperature of cloud lightning discharge channel.Science China-Earth Science,52(7),907–912.

Wang,Y.H.,G.S.Zhang,T.Zhang,and D.J.Cao,2007:Submicrosecond VHF radiation character on broadband wrap in lightning.Proceedings of CSEE,27(9),41–45.(in Chinese)

Wang,Y.,G.Zhang,X.Qie,D.Wang,T.Zhang,Y.Zhao,Y.Li,and T.Zhang,2012:Characteristics of compact intracloud discharges observed in a severe thunderstorm in northern part of China.J.Atmos.Sol.-Terr.Phys.,84–85,7–14.

Wang,Y.H.,G.S.Zhang,T.Zhang,Y.J.Li,B.Wu,and T.L.Zhang,2013:Interaction between adjacent lightning discharges in clouds.Adv.Atmos.Sci.,30(4),1106–1116,doi:10.1007/s00376-012-2008-9.

Wu,T.,W.S.Dong,Y.J.Zhang,and T.Wang,2011:Comparison of positive and negative compact intracloud discharges.J.Geophys.Res.,116,D03111,doi:10.1029/2010JD015233.

Wu,T.,W.S.Dong,Y.J.Zhang,T.Funaki,S.Yoshida,T.Morimoto,T.Ushio,and Z.Kawasaki,2012:Discharge height of lightning narrow bipolar events.J.Geophys.Res.,117,D05119,doi:10.1029/2011JD017054.

Xie,Y.R.,X.S.Qie,F.X.Guo,and M.Baker,2005:Numerical simulation of the effect of liquid water content and ice crystal concentration on lightning f l ash frequency.Plateau Meteorology,24(4),598–603.(in Chinese)

Xiong,Y.J.,X.S.Qie,and F.X.Guo,2005:Temporal and spatial characteristics of South American lightning activities and their relationship with SST in east Pacif i c.Plateau Meteorology,24(3),396–403.(in Chinese)

Xiong,Y.J.,X.S.Qie,Y.J.Zhou,T.Yuan,and T.L.Zhang,2006:Regional responses of lightning activities to relative humidity of the surface.Chinese J.Geophys.,49(2),367–374.(in Chinese)

Yang,J.,X.S.Qie,G.S.Zhang,and H.B.Wang,2008a:Magnetic fi eldmeasuring systemand current retrieval inarti fi ciallytriggering lightning experiment.Radio Sci.,43,RS2011,doi:10.1029/2007RS003753.

Yang,J.,X.S.Qie,J.G.Wang,Y.Zhao,Q.L.Zhang,T.Yuan,Y.J.Zhou,and G.L.Feng,2008b:Observation of the lightninginduced voltage in the horizontal conductor and its simulation.Acta Physica Sinica,57(3),1968–1975.(in Chinese)

Yang,J.,X.S.Qie,G.S.Zhang,Y.Zhao,and T.Zhang,2008c:Red sprites over thunderstorms in the coast of Shandong province,China.Chinese Science Bulletin,53(7),1079–1086.

Yang,J.,X.Qie,Q.Zhang,Y.Zhao,G.Feng,T.Zhang,and G.Zhang,2009:Comparative analysis of the initial stage in two arti fi cially-triggered lightning fl ashes.Atmospheric Research,91,393–398.

Yang,J.,X.S.Qie,G.S.Zhang,Q.L.Zhang,G.L.Feng,Y.Zhao,and R.B.Jiang,2010:Characteristics of channel base currents and close magnetic fi elds in triggered fl ashes in SHATLE.J.Geophys.Res.,115,D23102,doi:10.1029/2010JD014420.

Yang,J.,and G.L.Feng,2012:A gigantic jet event observed over a thunderstorm in mainland China.Chinese Science Bulletin,57(36),4791–4800.

Yang,J.,M.R.Yang,C.Liu,and G.L.Feng,2013a:Case studies of sprite-producing and non-sprite-producing summer thunderstorms.Adv.Atmos.Sci,30(6),1786–1808,doi:10.1007/s00376-013-2120-5.

Yang,J.,X.S.Qie,and G.L.Feng,2013b:Characteristics of one sprite-producing summer thunderstorm.Atmospheric Research,127,90–115.

Yuan,P.,X.S.Liu,Y.J.Zhang,L.Y.Xie,and Z.C.Dong,2002:Theoretical calculationfor thelifetimesof NIIrelatedtolightning process.Acta Physica Sinica,51(11),2495–2502.(in Chinese)

Yuan,P.,X.S.Liu,Y.J.Zhang,X.S.Qie,G.S.Zhang,and H.B.Wang,2004a:Spectral study on lightning return stroke in plateau area.Chinese J.Geophys.,47(1),42–46.(in Chinese)Yuan,P.,X.S.Liu,and Y.J.Zhang,2004b:NII Ion spectra related to lightning discharges.Spectroscopy and Spectral Analysis,24(3),288–291.(in Chinese)

Yuan,P.,X.S.Qie,S.H.L¨u,G.Y.Chen,and G.S.Zhang,2006:The spectral properties of an intense lightning return stroke.Spectroscopy and Spectral Analysis,26(4),733–737.(in Chinese)

Yuan,T.,and X.S.Qie,2004:Spatial and temporal distributions of lightning activities in China from satellite observation.Plateau Meteorology,23(4),488–494.(in Chinese)

Yuan,T.,and X.S.Qie,2008:Study on lightning activity and precipitation characteristics before and after the onset of the South China Sea summer monsoon.J.Geophys Res.,13,D14101,doi:10.1029/2007JD009382.

Yuan,T.,and X.S.Qie,2010:TRMM-based study on lightning activity and its relationship with precipitation structure of a squall line in south China.Chinese J.Atmos.Sci.,34(1),58–70.(in Chinese)

Zhang,G.S.,Y.X.Zhao,X.S.Qie,T.Zhang,Y.H.Wang,and C.P.Chen,2008:Observation and study on the whole process of cloud-to-ground lightning using narrowband radio interferometer.Science China-Earth Science,51(5),694–708.doi:10.1007/s11430-008-0049-9.

Zhang,G.S.,Y.H.Wang,X.S.Qie,T.Zhang,Y.X.Zhao,Y.J.Li,and D.J.Cao,2010:Using lightning locating system based on time-of-arrival technique to study three-dimensional lightning discharge processes.Science China-Earth Science,53,591–602.doi:10.1007/s11430-009-0116-x.

Zhang,H.M.,P.Yuan,S.H.L¨u,and Y.H.Ouyang,2007a:Study on electron density of lightning return stroke.Plateau Meteorology,26(2),264–269.(in Chinese)

Zhang,Q.L.,and X.S.Qie,2003:Reconstruction of return stroke radiation fi eld waveforms and estimation of cloud-to-ground discharges parameters.Plateau Meteorology,22(3),252–258.(in Chinese)

Zhang,Q.,X.S.Qie,and G.S.Zhang,2003a:Short-baseline timeof-arrival lightning radiation detection system and preliminary location result.Plateau Meteorology,22(3),226–234.(in Chinese)

Zhang,Q.L.,X.S.Qie,and H.B.Wang,2003b:Characteristics of the radiation fi elds from return strokes in plateau thunderstorms.Proceedings of thesand CSEE,23(9),94–98.(in Chinese)

Zhang,Q.L.,X.S.Qie,H.B.Wang,C.P.Chen,G.S.Zhang,and T.Zhang,2005:Characteristics and numerical simulation of electric fi eld waveforms produced by close negative cloud-toground fl ashes.Proceedings of the CSEE,25(18),126–130.(in Chinese)

Zhang,Q.L.,and Coauthors,2006:Electric fi eld characteristics of leader and return in triggered lightning.High Power Laser and Particle Beams,18(12),2004–1010.(in Chinese)

Zhang,Q.L.,X.S.Qie,X.Z.Kong,Y.J.Zhou,J.Yang,T.L.Zhang,G.L.Feng,and Q.F.Xiao,2007b:Comparative analysis on return stroke current of triggered and natural lightning fl ashes.Proceedings of the CSEE,27(6),67–71.

Zhang,Q.,X.Qie,Z.Wang,T.Zhang,Y.Zhao,J.Yang,and X.Kong,2009a:Characteristics and simulation of lightning current waveforms during one arti fi cially triggered lightning.Atmospheric Research,91(2–4),387–392.

Zhang,Q.L.,X.S.Qie,Z.H.Wang,T.L.Zhang,and J.Yang,2009b:Simultaneous observation on electric f i eld changes at 60 m and 550 m from altitude-triggered lightning f l ashes.Radio Sci.,44,RS1011,doi:10.1029/2008RS003866.

Zhang,Q.L.,X.D.Liu,J.Yang,R.B.Jiang,Z.H.Wang,and J.C.Bian,2011a:The characteristics and simulation of close leader/return stroke f i eld change waveforms.Radio Sci.,46,RS1017,doi:10.1029/2010RS004469.

Zhang,Q.,J.Yang,M.Liu,and Z.Wang,2011b:Measurements and simulation of the M-component current and simultaneous electromagnetic f i elds at 60 m and 550 m.Atmospheric Research,99,537–545.

Zhang,Q.L.,J.Yang,X.Q.Jing,D.S.Li,and Z.H.Wang,2012a:Propagation effect of a fractal rough ground boundary on the lightning-radiated vertical electric f i eld.Atmospheric Research,104,202–208,doi:10.1016/j.atmosres.2011.10.009.

Zhang,Q.L.,D.S.Li,and Y.F.Fan.,2012b:Examination of the Cooray-Rubinstein(C-R)formula for a mixed propagation path by using FDTD.J.Geophys.Res.,117(D15309),doi:10.1029/2011 JD017331.

Zhang,Q.L.,X.Q.Jing,J.Yang,D.S.Li,and X.Tang,2012c:Numerical simulation of the lightning electromagnetic f i elds alongarough andocean-land mixedpropagation path.J.Geophys.Res.,117,D20304,doi:10.1029/2012JD017851.

Zhang,Q.L.,J.Yang,D.S.Li,and Z.H.Wang,2012d:Propagation effects of a fractal rough ocean surface on the vertical electric f i eld generated by lightning return strokes.Journal of Electrostatics,70(1),54–59,doi:10.1016/j.elstat.2011.10.003.

Zhang,Q.L.,L.Zhang,X.Tang,and J.G.Gao,2014a:An approximate formula for estimating the peak value of lightninginduced overvoltage considering the stratif i ed conducting ground.IEEE Transactions on Power Delivery,29(2),884–889,doi:10.1109/TPWRD.2013.2281982.

Zhang,Q.L.,L.X.He,T.T.Ji,and W.H.Hou,2014b:On the fi eld-to-current conversion factors for lightning strike to tall objects considering the fi nitely conducting ground.J.Geophys.Res.,119(13),doi:10.1002/2014JD021496.

Zhang,Q.L.,X.Tang,J.G.Gao,L.Zhang,and D.S.Li,2014c:The in fl uence of the horizontally strati fi ed conducting ground on the lightning-induced voltages.IEEE Transactions on Electromagnetic Compatibility,56(2),435–443.doi:10.1109/TEMC.2013.2284929.

Zhang,R.,G.S.Zhang,Y.J.Li,Y.H.Wang,B.Wu,H.Yu,and Y.X.Liu,2014d:Estimate of NOxproduction in the lightning channel based on three-dimensional lightning locating system.Science China-Earth Science,57(7),1613–1625.

Zhang,T.L.,X.S.Qie,T.Yuan,Y.J.Xiong,and T.Zhang,2004:The temporal and spatial distribution of lightning activities along theQinghai-Tibet railroad.PlateauMeteorology,23(5),673–677.(in Chinese)

Zhang,T.L.,X.Qie,T.Yuan,G.Zhang,T.Zhang,and Y.Zhao,2009c:Charge source of cloud-to-ground lightning and charge structure of a typical thunderstorm in the Chinese Inland Plateau.Atmospheric Research,92(4),475–480.

Zhang,T.L.,X.S.Qie,M.H.Yan,Y.Zhao,G.S.Zhang,T.Zhang,and Y.H.Wang,2009d:A preliminary analysis on formation of electrical characteristics of thunderstorm in different altitude regions in Chinese inland plateau.Plateau Meteorology,28(5),1006–1017.(in Chinese)

Zhang,T.L.,T.Zhang,Y.Zhao,X.Z.Kong,and Y.H.Wang,2010:Characteristics of thunderstorms and lightning f l ashes in the Chinese inland plateau.Sciences in Cold and Arid Regions,2(3),271–277.

Zhang W.J.,Y.J.Zhang,D.Zheng,and X.J.Zhou,2012e:Lightning distribution and eyewall outbreaks in tropical cyclones during landfall.Mon.Wea.Rev.,140,3573–3586.

Zhang,Y.J.,W.S.Dong,G.S.Zhang,and X.S.Qie,2003c:Characteristics of the leading processes to the artif i cial induced lightning in the air.Chinese J.Geophys.,46(4),446–449.(in Chinese)

Zhang,Y.J.,W.S.Dong,G.S.Zhang,H.F.Zhang,C.P.Chen,and T.Zhang,2004a:Study of charge structure and radiation characteristic of intracloud discharge in thunderstorms of Qinghai-Tibet Plateau.Science China-Earth Science,47(S1),108–114.

Zhang,Y.J.,A.P.Sun,M.H.Yan,F.X.Guo,X.S.Qie,and M.Y.Huang,2004b:Numerical modeling for effects of electric activity during thunderstorms upon the growth of hail particles.Chinese J.Geophys.,47(1),25–32.(in Chinese)

Zhang,Y.J.,Q.Meng,P.R.Krehbiel,X.S.Liu,M.H.Yan,and X.J.Zhou,2006a:Spatiotemporal characteristics of positive cloud-to-ground lightning discharges and bidirectional leader of the lightning.Science China-Earth Science,49,212–224.

Zhang,Y.J.,Q.Meng,W.T.Lu,P.R.Krehbiel,X.S.Liu,and X.J.Zhou,2006b:Charge structures and cloud-to-ground lightning discharges characteristics in two supercell thunderstorms.Chinese Science Bulletin,51(2),198–212.

Zhang,Y.J.,W.T.L¨u,J.Li,W.S.Dong,D.Zheng,and S.D.Chen,2009e:Luminosity characteristics of leaders in natural cloud-to-ground lightning f l ashes.Atmospheric Research,91(2–4),326–332.

Zhang,Y.J.,Q.Meng,W.T.Lu,M.Ma,D.Zheng,and P.R.Krehbiel,2009f:Positive charge region in lower part of thunderstorm and preliminary breakdown process of negative cloudto-ground lightning.Acta Meteorologica Sinica,23(1),95–104.

Zhang,Y.J.,and Coauthors,2014e:Experiments of artif i cially triggered lightning and its application in Conghua,Guangdong,China.Atmospheric Research,135,330–343.doi:10.1016/j.atmosres.2013.02.010.

Zhao,Y.,Y.J.Zhang,W.S.Dong,H.F.Zhang,C.P.Chen,and T.Zhang,2004:Preliminary analysis of characteristics of lightning in the Nagqu area of the Qinghai-Xizang plateau.Chinese J.Geophys.,47(3),405–410.(in Chinese)

Zhao,Y.,and Coauthors,2009a:Analysis on the parameters of the current waveforms of triggered lightning.Acta Physica Sinica,58(9),6616–6625.(in Chinese)

Zhao,Y.,X.S.Qie,M.L.Chen,X.Z.Kong,G.S.Zhang,T.L.Zhanng,and G.L.Feng,2011: Characteristic of M-component in artif i cially-initiated triggered lightning.Plateau Meteorology,30(2),508–517.(in Chinese)

Zhao,Z.K.,and Q.L.Zhang,2009:Inf l uence of channel tortuosity on the lightning return stroke electromagnetic f i eld in the time domain.Atmospheric Research,91(1–4),404–409.

Zhao,Z.K.,and Coauthors,2009b:Electric f i eld soundings and the charge structure within an isolated thunderstorm.Chinese Science Bulletin,55,872–876.

Zheng,D.,Y.J.Zhang,M.Ma,Q.Meng,and W.T.L¨u,2007:Simulation study on the inf l uence of atmospheric stratif i cation on lightning activity.Acta Meteorologica Sinica,65(4),622–632.(in Chinese)

Zheng,D.,Y.J.Zhang,Q.Meng,W.T.Lu,and X.Y.Yi,2009:Total lightning characteristics and electric structure evolution in a hailstorm.Acta Meteorologica Sinica,23(2),233–249.(in Chinese)

Zhou,Y.J.,and X.S.Qie,2002:Mechanism and estimation of lightning generated NOx in Chinese inland area.Plateau Meteorology,21(5),501–508.(in Chinese)

Zhou,Y.J.,X.S.Qie,and T.Yuan,2004:Spatial and temporal distributions of NOx produced by lightning in East Asian region.Plateau Meteorology,23(5),667–672.(in Chinese)

Zhou,Y.J.,S.Soula,V.Pont,and X.S.Qie,2005:NOx ground concentration at a station at high altitude in relation to cloudt-ground lightning f l ashes.Atmospheric Research,75,47–69.

Zhou,Z.M.,and X.L.Guo,2009:A three dimensional modeling study of multi-layer distribution and formation processes of electric charges in a severe thunderstorm.Chinese J.Atmos.Sci.,33(3),600–620.(in Chinese)

Zhu,B.Y.,M.Ma,and S.C.Tao,2003:Measurement and comparison of VHF/VLF radiations of preliminary breakdown of Cloud-to-Ground and intracloud f l ashes.Plateau Meteorology,22(3),239–245.(in Chinese)

Zhu,B.Y.,S.C.Tao,and Y.B.Tan,2007:Initial observations of the lightning narrow bipolar pulses with very powerful VHF radiation.Acta Meteorologica Sinica,65(1),124–130.(in Chinese)

Zhu,B.Y.,H.L.Zhou,M.Ma,and S.C.Tao,2010a:Observations of narrow bipolar events in East China.J.Atmos.Sol.-Terr.Phys.,72(2–3),271–278.

Zhu,B.Y.,H.L.Zhou,M.Ma,F.L¨u,and S.C.Tao,2010b:Estimation of channel characteristics of narrow bipolar events based on the transmission-line model.J.Geophys.Res.,115,D19105,doi:10.1029/2009JD012021.

Zhu,B.Y.,H.L.Zhou,R.Thottappillil,and V.Rakov,2014:Simultaneous observations of electric f i eld changes,wideband magnetic f i eld pulses,and VHF emissions associated with K processes in lightning discharges.J.Geophys.Res.,119(6),2699–2710,doi:10.1002/2013JD021006.