ATMOSPHERIC AND OCEANIC SCIENCE LETTERS, 2014, VOL. 7, NO. 3, 248−253

Characteristics of a Negative Cloud-to-Ground Lightning Discharge Based on Locations of VHF Radiation Sources SUN Zhu-Ling1, 2, QIE Xiu-Shu1, and LIU Ming-Yuan1 1 Key Laboratory of Middle Atmosphere and Global Environment Observation (LAGEO), Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China 2 University of Chinese Academy of Sciences, Beijing 100049, China

Received 29 December 2013; revised 6 March 2014; accepted 28 March 2014; published 16 May 2014

Abstract The lightning very high frequency (VHF) radiation location system based on the short-baseline time-difference of arrival (TDOA) technique provides an effective approach to describe the temporal and spatial development of lightning discharge in two dimensions with high resolution. A negative single-stroke cloud-to- ground (CG) lightning flash was analyzed in detail using the radiation location results and synchronic fast/slow elec-tric field changes. The long-duration preliminary break-down process appeared to propagate with bi-directional leader channels. The two negative simultaneous discharge channels sloped down with a considerable horizontal com-ponent in the lower positive charge region at speeds of about 105 m s−1. The stepped leader was clearly converted from one channel of the preliminary breakdown process and spread downwards with branches. The speeds of the stepped leaders were about 105 m s−1. The K processes after the return stroke could either directly initiate from the start region with negative polarity lightning discharge, or initiate from a new region in the cloud as negative recoil streamers. All K processes propagated along the preceding electrified channel, while not all K processes initiated from the tips of positive breakdowns. The speeds of the K processes were about 106–107 m s−1.

Keywords: short-baseline, very high frequency radiation, cloud-to-ground lightning, discharge Citation: Sun, Z.-L., X.-S. Qie, and M.-Y. Liu, 2014: Characteristics of a negative cloud-to-ground lightning discharge based on locations of VHF radiation sources, Atmos. Oceanic Sci. Lett., 7, 248–253, doi:10.3878/ j.issn. 1674-2834.13.0110.

1 Introduction The cloud-to-ground (CG) lightning discharge process

is a very important issue in the lightning research field because of the damage it can cause to human lives and the devices and technology we rely upon. Since the late 1980s, knowledge of the development process and char-acteristics of CG lightning discharge has been gradually extended, based on measurements of optical or electro-magnetic field observations (Thomson et al., 1984; Rakov and Uman, 1990; Kong et al., 2008). To achieve a clear

Corresponding author: QIE Xiu-Shu, qiex@mail.iap.ac.cn

and detailed understanding of the physical process of lightning, lightning location techniques with high tempo-ral and spatial resolution are required. Very high fre-quency (VHF) location techniques have been an effective approach to describe smaller-scale breakdown processes in lightning discharges (Shao et al., 1995; Morimoto et al., 2005; Nishihashi et al., 2013).

Using a VHF broadband interferometer, Mazur and Ruhnke (1993) reported that positive and negative lead-ers are produced in the preliminary breakdown process and propagate in opposite directions. Dong et al. (2003) observed that the preliminary breakdown process ex-tends upward with the bi-directional channels from the negative charge region, and both channels can be con-sidered as negative breakdowns. Proctor et al. (1988) interpreted that preliminary breakdown is the start of the leader using the imaging system of lightning. Mean-while, Rhodes and Krehbiel (1989) concluded that the channel characteristic of the preliminary breakdown is clearly different from that of the stepped leader. The K process is a kind of discharges occurred frequently both in CG and intra-cloud lightning inside the cloud, with step-like change superimposed on the electric field change. Mazur (1989) interpreted that the K process transports negative charges into the flash origin from other remote regions. Mazur et al. (1997), Rison et al. (1999), and Akita et al. (2010) described the K process as a negative recoil streamer that occurs when a positive leader encounters the negative charge region. Currently, there are still many unanswered questions relating, for example, to the propagation characteristics of the pre-liminary breakdown, the difference and relationship be-tween the preliminary breakdown and the stepped leader, and the mechanism of the K process.

The time-difference of arrival (TDOA) technique is one of the major lightning location techniques used in short-baseline VHF radiation location, which was first used to estimate the elevation and azimuth of radiation sources at a bandwidth from 20 to 80 MHz by Taylor (1978). Zhang et al. (2003) employed this technique with a center frequency of 280 MHz and a bandwidth of 10 MHz to im-age lightning branches in two dimensions, although the system detection efficiency was relatively low for the limit of the experimental field and signal bandwidth. Subse-quently, a new short-baseline lightning location system

NO. 3 SUN ET AL.: CHARACTERISTICS OF A NEGATIVE CLOUD-TO-GROUND LIGHTNING DISCHARGE 249

using the same technique but with the bandwidth from 125 to 200 MHz was developed to map lightning channels (Cao et al., 2012), and the generalized cross-correlation time-delay estimation method with filtering and parabolic interpolation algorithm was used to provide higher location accuracy to a certain extent (Sun et al., 2013). The accuracy and availability of the system has been demonstrated through comparisons between radiation source location results and synchronic high-speed video camera observa-tion for a rocket-triggered lightning flash.

In this paper, by means of the short-baseline TDOA technique, we present 2D lightning observations of a negative single-stroke CG lightning flash of long duration together with the preliminary breakdown and leader de-velopment. We provide a detailed analysis of the temporal and spatial development of the major discharge processes with high temporal and spatial resolution, such as the pre-liminary breakdown processes, leader development, and K processes.

2 Experiment In the summer of 2013, the lightning VHF radiation lo-

cation system based on short-baseline TDOA technology was used to obtain 2D precise radiation maps in the SHan-dong Artificial Triggering Lightning Experiment (SHATLE) (37.82°N, 118.11°E) in Shandong Province, China (Qie et al., 2007, 2009). The system consisted of four identical broadband flat plane antennas placed in an or-thogonal antenna array. The lightning VHF radiation sig-nals from each of these antennas were processed through a band-pass filter and amplifier sequentially, and then ac-quired by a LeCroy oscilloscope at a sampling rate of 1 GS S−1. In this study, the configurations of the system were slightly different from those described in Sun et al. (2013). The orthogonal baseline length was decreased from 10 m to 8 m due to space constraints. The bandwidth of the band-pass filter was expanded to 140–300 MHz and the data acquisition vertical resolution was increased from 8 bits to 12 bits. This enhanced the sensitivity of the system.

In addition, lightning electric field changes caused by the lightning flash were also measured synchronously by the fast and slow antennas, whose bandwidths were 2 MHz and 0.5 MHz, with time constants of 0.1 ms and 220 ms, respectively. By convention, a positive electric field change in this study correlates with the removal of nega-tive charge in the cloud.

3 Analysis and results The negative single-stroke CG lightning occurred at

2133:03 UTC 2 August 2013. Using the time interval be-tween the perception of lightning and the first sound of thunder, the lightning was estimated to be about 2 km away from the TDOA location system. Figure 1 shows the VHF radiation field detected by the short-baseline TDOA location system and simultaneous electric field changes of the lightning recorded by the fast and slow antennas. Time zero represents the incipient moment of the fast electric

field change signal. The flash lasted about 240 ms and the return stoke happened at about 81 ms, followed by a short-duration continuing current to ground lasting 3.5 ms. Finally, the flash finished with a sequence of typical K processes. The VHF radiation mainly concentrated in the phases before the return stroke and near the times of the K processes; while after the return stroke, the VHF radiation lasted just 0.2 ms and became almost quiescent in the continuing current.

Figure 2 presents the radiation source locations for the whole CG lightning in two dimensions. The azimuth in-creases in a clockwise direction from the north. The color changes with time from blue to red. It can be seen that the lightning started in the south of the observation station at 68° elevation, which is marked by an “S” in Fig. 2. The whole lightning channel expanded widely with the azimuth exten-sion from −30° to 250°, and the lightning made contact with the ground in the south of the observation station at about 68° azimuth. Using the location result of the lightning radia-tion sources, the major lightning discharge developments are described in detail in the following sections.

Figure 1 The very high frequency (VHF) radiation field detected by the short-baseline time-difference of arrival (TDOA) location system and simultaneous electric field changes of the lightning recorded by the fast and slow antennas. R indicates the return stroke and K1–K6 indicate the six K processes.

Figure 2 2D VHF radiation source locations of the whole cloud-to-ground (CG) lightning flash. S indicates the start region of the lightning discharge.

250 ATMOSPHERIC AND OCEANIC SCIENCE LETTERS VOL. 7

3.1 The preliminary breakdown and the leader

Figure 3a shows the electric field change of the period preceding the return stroke, which was found to be con-sistent with the so called BIL model defined by Clarence and Malan (1957). In this model, the whole phase pre-ceding the return stroke can be divided into three suc-cessive discharge processes: the preliminary breakdown (B), intermediate (I), and stepped leader (L).

The long-duration preliminary breakdown stage lasted about 26 ms. As shown in Fig. 3a, the fast electric field change was relatively small in amplitude and superim-posed several negative pulses during the initial 4 ms. Meanwhile, the slow electric field change negatively in-creased at a steady rate, indicating that the negative charge moved toward the observation site or positive charge moved away from the observation site. As seen in Fig. 3e, the lightning began with repeated breakdowns near point S, then spread in opposite directions and de-veloped simultaneously, as indicated by arrows b1 and b2. Subsequently, another small branch, b3, initiated from the start point S and propagated for 6 ms with increasing ele-vation. The propagation of these discharge channels also can be shown in Figs. 3c and 3d .Thereafter, discharge b1

Figure 3 Radiation sources of the discharge before the return stroke in the CG lightning: (a) electric field changes recorded by the fast and slow antenna; (b) VHF radiation field; (c) elevation versus time; (d) azimuth versus time; and (e) azimuth-elevation display. In (a), B, I, and L indi-cate the preliminary breakdown, intermediate, and stepped leader, re-spectively. In (e), the arrows b, i, and l indicate channels of the prelimi-nary breakdown, intermediate, and stepped leader, respectively.

extended the lightning channels downward to about 80° azimuth and 52° elevation with several fine branches; and discharge b2 appeared to progress horizontally to about 223° in azimuth and 70° in elevation. The extents of dis-charges b1 and b2 on a spherical projection were esti-mated to be 4.2 km and 5.4 km. The average velocities of discharges b1 and b2 in two dimensions were 1.6 × 105 m s−1 and 2.0 × 105 m s−1. Actual 3D extents and velocities will be larger than these, due to the possible channel slope to the observation site. Both discharges b1 and b2 had similar characteristics for VHF radiation, and the radia-tion sources were mainly distributed on the tips of the lightning channels. The preliminary breakdown ended with a continuous burst of positive pulses lasting about 10 ms in the fast electric field change, indicating that the negative breakdowns occurred inside the lower positive charge region. Kasemir (1960) proposed the bi-directional leader concept that both negative and positive leaders propagate simultaneously in opposite directions; while in observations of the nature of lightning, positive break-downs are almost undetectable by the VHF radiation lo-cation system because of their weak VHF radiation. In this case, it can be inferred that the preliminary break-down might have happened between the main negative charge region and the lower positive charge region, and the negative breakdown discharges, b1 and b2, simulta-neously transferred negative charges toward the observa-tion site in a downward sloping direction. Using a broad-band interferometer, Dong et al. (2003) also observed that the preliminary breakdown process originates from the negative charge region and extends upward with the bi-directional channels. However, the development direc-tion in the current study was downward with a large hori-zontal component. This may be because of the difference of the charge distribution in different thunderclouds. Fur-thermore, the preliminary breakdown process had consid-erable long-duration horizontal discharges in the cloud before the appearance of the leader toward the ground. Therefore, it can be demonstrated that the lower positive charge region may have great charge density and large horizontal extent, and play a significant role in the pre-liminary breakdown process.

The intermediate period after the preliminary break-down process lasted 21 ms. Unlike the preliminary breakdown, there was no significant change in the fast and slow electric field change, which was accompanied by discrete and weak VHF radiation during this phase as shown in Fig. 3b. Discharge arrow i1 (Fig. 3) progressed in the direction of increasing elevation upward, and the elevation ascended from about 52° to 68° with an aver-age speed of about 2.3 × 105 m s−1. Meanwhile, dis-charge b2 continued to develop horizontally, and finally extended the channel to about 246° azimuth after about 6 ms, as indicated by the arrow i2. The average speed of the radiation source progression was estimated to be 1.4 × 105 m s−1. It can be inferred that discharge i1 trans-ported negative charge vertically upward, and affected the electric field change with discharge i2 in common.

NO. 3 SUN ET AL.: CHARACTERISTICS OF A NEGATIVE CLOUD-TO-GROUND LIGHTNING DISCHARGE 251

The leader began at about 47 ms and lasted 33 ms until the return stroke. During the leader phase, there were clusters of positive pulses in the fast electric field change, corresponding to the stepped leaders. The slow electric field change had a relatively large negative slope in am-plitude, and the VHF radiation again became intense and continuous with time. As shown in Fig. 3e, the leader started from ahead of discharge i1, and was highly branched while developing to the ground. In the high-elevation region, numerous distinct branches were superimposed together on the 2D map. When the main channel descended to about 12° azimuth and 46° elevation, the leader split into three branches extending simultane-ously towards the ground, as shown by arrows l1, l2, and l3. Finally, discharge l2 reached the ground and induced the return stroke. As the leader approached the ground, the radiation sources appeared more widespread, and the time interval between leader pulses became shorter and irregu-lar, due to the increasing branches dispersed from the leader channels. The estimated velocities of the three main leader branches were 1.3 × 105 m s−1, 1.2 × 105 m s−1, and 1.5 × 105 m s−1, which are similar to the speed of 1.5 × 105 m s−1 reported by Yoshida et al. (2012).

Generally, the preliminary breakdown process is consid-ered to be necessary for the initiation of the stepped leader. Zhang et al. (2009) found that the stepped leader results from the K processes along the previously formed channel through the start point of the lighting. A similar observa-tion, that the stepped leader is initiated by recoil streamers during the final stage of preliminary breakdown process, was reported by Cao et al. (2012). In the present study, the stepped leader developed directly from the preliminary breakdown in the flash, and the speeds of the three succes-sive discharge processes prior to the return stroke were in the same order of magnitude, indicating that the stepped leader might be an extension of the preliminary breakdown process propagating outside the cloud. However, more ob-servations and studies are needed to understand the initia-tion mechanism of the stepped leader.

3.2 K processes About 72 ms after the start of the return stroke, a se-

quence of typical K processes occurred with step-like changes in the slow electric field change, as shown in Fig. 1. The average time interval of the six prominent K proc-esses was about 15 ms.

The step-like change of the K process, K1, lasted for about 1 ms, accompanied by 1.7 ms of relatively strong VHF radiation. Figure 4a shows the radiation source loca-tion results of the K1 process. It initiated at the start of region S, and progressed quickly along the extent of dis-charge b2 of the preliminary breakdown toward the ob-servation site, as shown by the arrow K1a. About 0.2 ms after the initiation of the K1a process, the other break-down, K1b, started from the discharge channel K1a near the region S, and synchronously developed along the pre-vious path of discharge b3. Discharge K1a terminated at the branch point of discharge b3. The average velocities of discharges K1a and K1b were about 3.2 × 106 m s−1

Figure 4 2D VHF radiation source locations of the K processes in the CG lightning: (a) processes of K1; (b) processes of K2. The arrows K1a, K1b, K2a, and K2b indicate channels of the K processes K1 and K2.

and 1.6 × 106 m s−1, and are typical of those reported by other investigators (Akita et al., 2010; Winn et al., 2011). The K1 process was considered to be a negative break-down discharge and transferred negative charge down-ward from the start region of the lightning with an obvi-ously horizontal component, like discharge b2, as evi-denced by the negative field change. Meanwhile, the pos-itive charges deposited by the return stroke were also neutralized. The results show that process K1 had the same progress direction of the positive breakdown caused by the return stroke, and was different from the negative recoil streamers mentioned by Akita et al. (2010).

After a delay of 16 ms, process K2 occurred for about 1.2 ms and also radiated intense VHF radiation. Figure 4b shows the radiation source location results of the K proc-ess, K2. The discharge began from the region of about 162° in azimuth and 75° in elevation, and spread down-ward to the start region, S, as shown by arrow K2a. The VHF radiation lasted 30 μs in this period, and the velocity was estimated to be 5.4 × 107 m s−1. The high propagation speed and negative slow field change suggest that the fast negative streamer progressed along the existing positive breakdown channel, which might have been generated by the return stroke and could not be detected. Process K2 was considered to be a negative recoil streamer and transferred the negative discharge from another region in the cloud to the lightning start region.

252 ATMOSPHERIC AND OCEANIC SCIENCE LETTERS VOL. 7

After a short interruption of the VHF radiation of about 0.1 ms, the discharge resumed near region S and propa-gated, retracing the path of discharge b1, without branch-ing, as shown by arrow K2b. Finally, discharge K2b stopped at 48° elevation, where the main stepped leader branched. The average velocity of discharge K2b was about 7.6 × 106 m s−1. As the progression speeds of the K processes were one or two orders of magnitude faster than those of the preceding preliminary breakdown along the same paths, it can be recognized that the channels inside the cloud might still have retained good conductivity after the return stroke; while those outside the cloud had al-ready cooled down and cut off, or there were not enough negative charges to support the channel to move forward, and the K process ultimately did not reach the ground, which was similar to the attempted leader.

Other K processes later on all developed in a similar manner, initiating from the start region of the lightning or a new region in the cloud, and propagating along the pre-ceding electrified channel. The speeds of all the K proc-esses were 106–107 m s−1.

4 Summary In this study, the main discharge processes, including the

preliminary breakdown process, stepped leader and K proc-esses, of a negative single-stroke CG lightning flash were investigated using the lightning VHF radiation location sys-tem based on the short-baseline TDOA technique. The major observation results can be summarized as follows.

The preliminary breakdown process occurred between the middle negative charge region and the lower positive charge region with bi-directional leaders, and transferred negative charges downward to the lower positive charge region. Two discharge channels developed simultaneously with a large horizontal component in different directions, both considered to be negative breakdown. The results indicate that the long-duration preliminary breakdown process likely corresponds to the lightning channels propagating horizontally with branches, and further con-firm that the existence of the lower positive region has a considerable effect on the characteristics of the prelimi-nary breakdown process.

The stepped leader was directly transformed from one of the branches of the preliminary breakdown process, inferred to be the further development of the latter. Abundant small branches extended away from the main leader channel in the large elevation, and three branches of stepped leader progressed simultaneously downward to the ground, with average velocities of about 105 m s−1.

All K processes presented in this paper propagated along the channel established by preceding breakdowns, and were also found to have negative polarity. It should be noted that the K processes were not always initiated from the tips of positive breakdowns. They might have been initiated in the start region of the lightning, or a new negative region in the cloud, and then transferred negative charges to the lightning channel to neutralize the positive

charge deposited there. The propagation velocities of K changes were about 106–107 m s−1, greater than the pre-liminary breakdown and stepped leaders.

The lightning VHF radiation location system based on the short-baseline TDOA technique shows good perform-ance in locating simultaneous radiation sources, and pro-vides an effective approach to infer the distribution char-acteristics of electric charge inside the thunderstorm. Compared with the long-baseline TDOA technique de-picting the lightning discharge in three dimensions (Rison et al., 1999; Zhang et al., 2010), the short-baseline TDOA technique described here only provides location results of lightning propagation in two dimensions, but with higher temporal and spatial resolution. Acknowledgements. The research was supported by the Instrument Developing Project of the Chinese Academy of Sciences (Grant No. YZ201206), the National Natural Science Foundation of China (Grant Nos. 40930949 and 41175002), and the National Science and Technology Support Project (Grant No. 2008BAC36B03).

References Akita, M., Y. Nakamura, S. Yoshida, et al., 2010: What occurs in K

process of cloud flashes? J. Geophys. Res., 115, D07106, doi: 10.1029/2009JD012016.

Cao, D., X. Qie, S. Duan, et al., 2012: Lightning discharge process based on short-baseline lightning VHF radiation source locating system, Acta Phys. Sinica, 61(6), 069202, doi:10.7498/aps. 61.069202.

Clarence, N. D., and D. J. Malan, 1957: Preliminary discharge pro-cesses in lightning flashes to ground, Quart. J. Roy. Meteor. Soc., 83, 161–172.

Dong, W., X. Liu, C. X. Chen, et al., 2003: Broadband interferome-ter observations of the bi-directional breakdown process in natu-ral lightning, Chinese J. Geophys. (in Chinese), 46(3), 317–321.

Kasemir, H. W., 1960: A contribution to the electrostatic theory of lightning discharges, J. Geophys. Res., 65(7), 1873–1878.

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

Mazur, V., 1989: Triggered lightning strikes to aircraft and natural intracloud discharges, J. Geophys. Res., 94(D3), 3311–3325.

Mazur, V., and L. H. Ruhnke, 1993: Common physical processes in natural and artificially triggered lightning, J. Geophys. Res., 98(D7), 12913–12930.

Mazur, V., E. Williams, R. Boldi, et al., 1997: Initial comparison of lightning mapping with operational time-of-arrival and inter-ferometric systems, J. Geophys. Res., 102(D10), 11071–11085.

Morimoto, T., Z. I. Kawasaki, and T. Ushio, 2005: Lightning obser-vations and consideration of positive charge distribution inside thunderclouds using VHF broadband digital interferometry, At-mos. Res., 76(1), 445–454.

Nishihashi, M., K. Shimose, K. Kusunoki, et al., 2013: Three-dimensional VHF lightning mapping system for winter thunderstorms, J. Atmos. Oceanic Technol., 30, 325–335.

Proctor, D. E., R. Uytenbogaardt, and B. M. Meredith, 1988: VHF radio pictures of lightning flashes to ground, J. Geophys. Res., 93(D10), 12683–12727.

Qie, X., Q. Zhang, Y. Zhou, et al., 2007. Artificially triggered light-ning and its characteristic discharge parameters in two severe thunderstorms, Sci. China Ser. D-Earth Sci., 50(8), 1241–1250.

Qie, X., Y. Zhao, Q. Zhang, et al., 2009. Characteristics of triggered lightning during Shandong artificial triggering lightning experi-

NO. 3 SUN ET AL.: CHARACTERISTICS OF A NEGATIVE CLOUD-TO-GROUND LIGHTNING DISCHARGE 253

ment (SHATLE), Atmos. Res., 91(2), 310–315. Rakov, V. A., and M. A. Uman, 1990: Some properties of negative

cloud-to-ground lightning flashes versus stroke order, J. Geo-phys. Res., 95(D5), 5447–5453.

Rhodes, C., and P. R. Krehbiel, 1989: Interferometric observations of a single stroke cloud-to-ground flash, Geophys. Res. Lett., 16(10), 1169–1172.

Rison, W., R. J. Thomas, P. R. Krehbiel, et al., 1999: A GPS-based three-dimensional lightning mapping system: Initial observations in central New Mexico, Geophys. Res. Lett., 26(23), 3573–3576.

Shao, X. M., P. R. Krehbiel, R. J. Thomas, et al., 1995: Radio inter-ferometric observations of cloud-to-ground lightning phenomena in Florida, J. Geophys. Res., 100(D2), 2749–2783.

Sun, Z. L., X. Qie, M. Y. Liu, et al., 2013: Lightning VHF radiation location system based on short-baseline TDOA tech-nique—Validation in rocket-triggered lightning, Atmos. Res., 129, 58–66.

Taylor, W. L., 1978: A VHF technique for space-time mapping of lightning discharge process, J. Geophys. Res., 83(C7), 3575–3583.

Thomson, E. M., M. A. Galib, M. A. Uman, et al., 1984: Some fea-

tures of stroke occurrence in Florida lightning flashes, J. Geo-phys. Res., 89(D3), 4910–4916.

Winn, W. P., G. D. Aulich, S. J. Hunyady, et al., 2011: Lightning leader stepping, K changes, and other observations near an intra-cloud flash, J. Geophys. Res., 116, D23115, doi:10.1029/2011 JD015998. .

Yoshida, S., M. Akita, T. Morimoto, et al., 2012: Propagation char-acteristics of lightning stepped leaders developing in charge re-gions and descending out of charge regions, Atmos. Res., 106, 86–92.

Zhang, G., Y. Wang, X. Qie, et al., 2010: Using lightning locating system based on Time-of-Arrival technique to study three-dimensional lightning discharge processes, Sci. China Ser. D-Earth Sci., 53(4), 591–602.

Zhang, Q., X. Qie, and G. Zhang, 2003: Short-baseline time-of- arrival lightning radiation detection system and preliminary loca-tion result, Plateau Meteor. (in Chinese), 22(3), 226–234.

Zhang, Y. J., Q. Meng, W. T. Lu, et al., 2009: Positive charge region in lower Part of thunderstorm and preliminary breakdown proc-ess of negative cloud-to-ground lightning, Acta Meteor. Sinica, 23(1), 95–104.