Professional Documents
Culture Documents
Atmospheric Research
journal homepage: www.elsevier.com/locate/atmosres
Key Laboratory of Meteorological Disaster, Ministry of Education (KLME), Joint International Research Laboratory of Climate and Environment Change (ILCEC),
Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters (CIC-FEMD), Key Laboratory for Aerosol–Cloud–Precipitation of China
Meteorological Administration, Nanjing University of Information Science & Technology, Nanjing 210044, PR China
Keywords: A stochastic discharge parameterization scheme of upward lightning is presented and embedded into an existing
Upward lightning parameterization scheme fine resolution two-dimensional thundercloud electrification and lightning model. At present, our lightning
Self-initiated upward lightning model can simulate not only intra-cloud (IC) lightning and cloud-to-ground (CG) lightning but also upward
Nearby-lightning-triggered upward lightning lightning. By this improved model, 139 upward flashes originating from the tall structures were obtained. Out of
these upward flashes, 67 cases occur without any nearby preceding lightning discharges (called self-initiated
upward flashes), whereas 72 cases are triggered by the preceding lightning discharges (called nearby-lightning-
triggered upward flashes). The relationship between the initiation of self-initiated upward flashes and thun-
dercloud electrical environment parameters is discussed. The results show that the initiation of self-initiated
upward flashes is related to the weighted average of the lower charge heights (H) and the average of the lower
charge density (D). As D increases, the critical H increases. In addition, the reason for IC lightning or upward
lightning triggering upward lightning is investigated. IC lightning discharges or upward lightning discharges
neutralize the main negative charge region in the cloud and enhance the influence of the lower positive charge
region on the ground electric field.
⁎
Corresponding author at: School of Atmospheric Physics, Nanjing University of Information Science & Technology, Nanjing 210044, PR China.
E-mail address: ybtan@ustc.edu (Y. Tan).
https://doi.org/10.1016/j.atmosres.2018.10.011
Received 14 May 2018; Received in revised form 12 October 2018; Accepted 23 October 2018
Available online 25 October 2018
0169-8095/ © 2018 Elsevier B.V. All rights reserved.
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
Nearby-lightning-triggered upward flashes are common in summer region under different electrical charge structures in self-initiation up-
thunderstorms. At present, there are many researches on this type of ward flashes and discuss the triggering reason of nearby-lightning-
lightning. Saba et al. (2016) analyzed one hundred high-speed video triggered upward flashes.
recordings of negative upward lightning flashes, the results indicated
that all upward flashes are triggered by preceding nearby lightning, 2. Simulation methods
most of them are positive cloud-to ground (CG) flashes. Furthermore,
several hypotheses on how positive CG flashes and IC flashes trigger In this model, a stochastic discharge parameterization scheme of
upward lightning have also been proposed (Rakov and Uman, 2003a, upward lightning is built and embedded into an existing two-dimen-
2003b; Warner et al., 2012a, 2012b; Zhou et al., 2012). One is that sional thundercloud electrification and lightning model (Tan et al.,
negative leaders associated with either IC development or following a 2006, 2014b). The electrification scheme includes modified inductive
positive CG return stroke horizontally propagate toward the triggered (Ziegler et al., 1991; Shi et al., 2016) and non-inductive charging (Shi
towers. One is that a positive CG return stroke propagates through a et al., 2015), while the parameterization scheme of discharge can refer
previously formed leader network that is near the triggered towers. The to (Tan, 2006, Tan et al., 2006; Tao et al., 2009), allowing us to simplify
other is that multiple upward flashes could be triggered by the same IC our description. Note that a two-dimensional model has been adopted
lightning discharge, and the number of triggered upward lightning is in this paper due to its ability of simulating the lightning discharge with
largely depended on the horizontal extent of IC lightning discharges. high resolution, where 250 m resolution (grid spacing) and 2 s time step
Nevertheless, self-initiated upward flashes mainly occur during the non- are used to simulate the microphysical and electrification processes in a
convective season or the thunderstorm dissipating stage, where the low 70 km × 20 km domain, and a fine resolution of 12.5 m is adopted to
electrical charge structure and inactive lightning activities are all fa- calculate the discharge process (Tan, 2006; Tan et al., 2006). The dis-
vorable conditions for the slow enhancement of electric field under the tortion of ambient electric field caused by tall structures is the main
cloud (Takagi et al., 2006; Wang and Takagi, 2012; Zhou et al., 2012; reason for triggering upward lightning. Tan et al. (2014a) presented
Liu et al., 2013). that the spatial resolution of the simulation domain has a positive
It is worth noting that current research results of upward lightning correlation with the distortion of electric field at the top of a building.
mainly come from experimental observations. Taking the limitation of In other words, it is hard to reach the initiation and propagation
observation methods and experiment conditions into account, it is threshold of upward lightning in a coarse resolution for 3-dimensional
difficult to obtain comprehensive thundercloud electrical parameters domain. That is why a two-dimensional fine resolution is used in the
before and after upward lightning occurs. This lead to a result that most lightning model.
of the research results are the conjectures based on observation data,
and these research results have particularity and uncertainty. In addi- 2.1. Upward lightning parameterization
tion, the initiation of upward lightning is relatively difficult compared
to the initiation of IC lightning or CG lightning. It not only requires tall The parameterization scheme of upward lightning also needs to
structures on the ground, but also requires the electric field at the top of consider the initiation of upward lightning, the propagation of upward
the structures to exceed a certain threshold. These factors make it more lightning, the termination of upward lightning, and the calculation of
difficult to obtain a comprehensive data of upward lightning through induced charges in the channel. These are consistent with the para-
observation. Therefore, establishing a corresponding theoretical model meterization schemes of IC and CG lightning. However, upward light-
for an in-depth study is necessary. At present, numerical models on ning is triggered from the high-grounded objects and generally propa-
lightning mainly include the following two categories. One is a nu- gates in one direction to the charged cloud, unlike IC and CG lightning
merical thundercloud model with extensive parameterizations of elec- that are initiated from the clouds and propagate in two different di-
trification mechanisms, which can simulate various types of lightning, rections. Therefore, the initiation and propagation of upward lightning
including IC lightning, negative CG lightning, and positive CG lightning will be described in detail.
(Mansell et al., 2002; Tan et al., 2006; Guo et al., 2016; Iudin and
Davydenko, 2015, Iudin et al., 2017). However, due to the coarse model 2.1.1. Initiation of upward lightning
resolution, the ground electric field can hardly reach the initial It is commonly assumed that upward lightning is triggered from tall
threshold of upward lightning. Therefore, the above models cannot si- structures (Rakov and Uman, 2003a; Rakov and Uman, 2003b; Warner,
mulate upward lightning. The other are physical models without cloud 2012; Smorgonskiy et al., 2013; Jiang et al., 2014). Upward lightning
microphysical process and parameterization schemes of electrification initiating from the low-rise structures is rarely observed. Because, the
(Becerra and Cooray, 2006a, 2006b, Becerra et al., 2007; Guo et al., electric field at the top of a low-rise building can hardly meet the initial
2017; Yang et al., 2017), which can simulate the initial process of up- threshold of upward lightning (Standler and Winn, 1979; Chauzy and
ward leaders from a variety of ground objects and the connection Raizonville, 1982; Qie et al., 1994), while the distortion of ambient
process of downward leaders and upward leaders. Such models have electric field caused by a tall building can satisfy the initial threshold of
made a great progress on the parameterization schemes of upward upward lightning (Diendorfer et al., 2009; Guimarães et al., 2014). That
leaders, but these models mainly focus on the connection process be- is why a 250-m grounded building is designed in our model. Taking the
tween upward leaders and downward leaders within a small area, difference between the initiation of self-initiated upward lightning and
whose scales are far smaller than that of lightning and these models can the initiation of nearby-lightning-triggered upward lightning into ac-
hardly discuss the relationship between the characteristics of upward count (the inception of self-initiated upward lightning will be affected
lightning and the thundercloud electrical environment parameters. by the corona space charges), their initiation thresholds will be set
Under this background, we present a stochastic discharge para- respectively. Firstly, according to the Peek's law (Peek, 1929), for
meterization scheme of upward lightning by the equivalent substitution hemispherical capped lightning rods, the electric field for the onset of
method and some reasonable assumptions, and embed it into an ex- glow corona Ecor is 3 MV/m under the standard condition (Aleksandrov
isting two-dimensional thundercloud electrification and lightning et al., 2001; Guo and Zhang, 2017; Guo et al., 2017), and considering
model (Tan et al., 2006; Tan et al., 2014b). the difference between corona onset and leader inception, we choose 7
In this paper, we first describe the stochastic discharge para- MV/m as the electric field for the initiation of nearby-lightning-trig-
meterization scheme of upward lightning. Then we make a more de- gered upward lightning (E0). As mentioned above, tall structures have a
tailed classification of upward lightning simulated in our model. great distortion on the ambient electric field. Tan et al. (2014a) pre-
Finally, we investigate the relationships between the average con- sented an electric field intensification factor Kr, which represents the
centration of lower charge region and the critical height of lower charge ratio of the distorted electric field to the original ambient electric field
64
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
in continuous space. Kr is associated with not only the height but also 2.1.2. Propagation of upward lightning
the width of a building. Therefore, a universal relation was presented as Upward lightning has a bifurcation structure, which is similar to IC
Formula (1), based on the estimated Kr of all building patterns in lightning and CG lightning. Therefore, the propagation of upward
continuous space. lightning will also adopt the stochastic discharge parameterization
scheme (Mansell et al., 2002; Tan, 2006; Tan et al., 2006). When the
Kr (H , W ) = 14.5797H 0.7478W 0.0124
(1) potential gradient between the lightning channel and the environment
exceeds the propagation threshold (Ecrit) of upward lightning, the
where H is height and W is width. It can be found that the value of Kr lightning channel will expand. Because the temperature of the leader
(H, W) is mainly contributed by height, the effect of the width on Kr (H, top is so high that the air near the leader top is highly ionized, so the
W) is very small. In this paper, H is 250 m and W is 12.5 m (there are propagation threshold of the upward leader is much smaller than the
many choices for these two parameters, as long as they are favorable for initial threshold. In our model, 150 kV/m (the same for both positive
the initiation of upward lightning). Thus, Kr (H, W) can be calculated by and negative leader) is chosen as the propagation threshold of upward
Formula (1). lightning referring to Tan et al. (2006b; 2014b). This value is effective
and largely determined by our spatial resolution of 12.5 m. If there are
Kr (H , W ) = 14.5797 × 250 0.7478 × 12.5 0.0124 878
multiple points meeting the propagation threshold of the upward
At the initial threshold (E0) of 7 MV/m and the electric field in- leader, lightning channels will stochastically choose one point as the
tensification factor (Kr) of 878, the ambient electrical field E required extension point. The treatment method for the stochastic propagation of
for initiating upward lightning is: the upward lightning channel are in agree with the method of Mansell
et al. (2002).
E0 7000 Our model simulates the step-by-step propagation of the lightning
E= = kV / m 8kV / m
K r (H , W ) 878 from an initiation point. When the potential gradient between the po-
sitive leader tips and the ambient grid points exceeds −150 kV/m (the
Although tall structures have a great distortion on the ambient physical sign convention is adopted for the polarity of the electric field
electric field, Tan et al. (2014a) presented that the spatial resolution of in this paper), one new channel segment may be added to the existing
simulation domain is positively correlated with the electric field in- positive leader channel. Similarly, when the potential gradient between
tensification factor Kr. Therefore, taking the resolution of our model the negative leader tips and the ambient grid points is smaller than
(12.5 m) into consideration, the distorted electric field at the top of the 150 kV/m (where positive and negative signs stand for the direction
250-m building still cannot reach 7 MV/m in our model. Based on these rather than the numerical value), one new channel segment may be
factors, we can only adopt the equivalent substitution method that the added to the existing negative leader channel (Mansell et al., 2002; Tan
ambient electric field of 8 kV/m is chosen to be the criterion for the et al., 2007). It is generally known that upward lightning is originated
initiation of nearby-lightning-triggered upward lightning without si- from the top of a building, and the building connected to the ground is
mulating the distortion process. This means that when the ambient considered as a good conductor without resistance. Therefore, upward
electric field at the top of a 250-m building in the model reaches 8 kV/ lightning is usually a one-way propagation process. Taking the above
m, nearby-lightning-triggered upward lightning would initiate from this factors into consideration, when the absolute value of potential gra-
building. Secondly, due to the negative effect of corona space charges dient exceeds 150 kV/m, the upward leader channel will add a new grid
on the electric field near the top of the building, the initiation threshold point. If the potential gradient is positive, the polarity of the leader is
of self-initiated upward lightning will be larger than the initiation positive. Referring to Mansell et al. (2002), the leader channel is treated
threshold of nearby-lightning-triggered upward lightning. According to as a conductor with resistance, thus the simulated leader channels are
the research results proposed in Aleksandrov et al. (2002, 2005) and assumed to have a nonzero internal electric field of 500 V/m (defined as
Bazelyan et al. (2008), the corona shielding coefficient 1.4 is chosen to Eint). Once the new channel point has been chosen, the potential of the
replace the effect of corona space charges on the electric field near the point is given as follow:
top of the building. Therefore, the ambient electric field of 11.2 kV/m is m
chosen to be the initiation threshold of self-initiated upward lightning. (m ) = ref s Eint · di (2)
i=1
There may be some error between the initiation thresholds and the
where s stands for the polarity of the leader channels, and ϕref is the
corresponding electric field thresholds in the actual environment (such
reference potential of the initiation point. m is the segment number
as the determined electric field value for the inception of upward
away from the initiation point, and di is the length of each segment.
lightning observed by Zhou et al. (2012) and Becerra et al. (2018)).
Considering the influence of the channel on the potential distribution of
As we all know, thunderclouds keep moving throughout the thun-
the environment (both extended leader channels and the building are
derstorm, and the strong electric field under the thunderclouds is
the fixed boundary), the ambient potential distribution should be up-
usually in a very small domain. Therefore, the initiation of upward
dated after each extension (Tan et al., 2006; Tan et al., 2007).
lightning is much more difficult than that of IC lightning or CG light-
ning. In this paper, in order to obtain more upward lightning flashes,
the location of the building in the model is not fixed. Based on the 2.1.3. Termination of the leader and calculation of the induced charge
model setting above, as long as the ambient electric field of a certain When the absolute values of the potential gradient between all
point at the altitude of 250 m reaches the initiation thresholds, the channel points and their ambient grid points are smaller than the
building will be set there. If there are multiple points meeting the propagation threshold or the leader propagate to the border (except the
condition, the building will be located where the ambient electric field ground), the upward leader will be terminated.
is the largest. In our scheme the lightning channel is regarded as a conducting
Some researchers have found that positive and negative leaders are channel. After the lightning is over, the Lightning channel will induce
different in initiation and propagation (D'Alessandro and Gumley, charges. According to Poisson's equation, the induced charge on the
2001; Solomon et al., 2001). However, the differences of initiation and channel is given by
propagation threshold between positive leaders and negative leaders = 2
(3)
are not given in detail. At present, most of the models do not take the
differences between positive and negative leaders into account (e.g., where ε is the electric permittivity, ρ is the net charge density at grid
Heldson et al., 1992; Mansell et al., 2002; Iudin et al., 2017). Therefore, points. Each hydrometeor category receives charges released by light-
these differences will not be considered in our model. ning in proportion to its total surface area, regardless of preexisting
65
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
66
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
Fig. 1. (a, b, and c) Upward lightning channel structure and distribution of space charge and (d, e, and f) electric potential. Fig. 1a and d are UL1, Fig. 1b and e are
UL2, and Fig. 1c and f are UL3. The purple diamonds represent the initiation points of upward lightning flashes, the purple channels represent upward positive
leaders, and the blue channels represent upward negative leaders. In Fig. 1a, b, and c, black solid lines are for positive charge and dashed lines are for negative charge
with contour values starting at ± 0.1 nC/m3 with intervals of 0.5 nC/m3. In Fig. 1d, e, and f, black solid lines represent positive potential and dashed lines represent
negative potential with contour values starting at 0 MV. The distributions of charge and electric potential are all from coarse resolution (250 m) model. (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. (a, b, and c) Upward lightning channel structure and distribution of space charge and (d, e, and f) electric potential. The details are same as those in Fig. 1.
Fig. 2a and d are UL4, Fig. 2b and e are UL5, and Fig. 2c and f are UL6.
initiation of UL1 and UL3, their leaders both enter the negative po- In observation, nearby lightning activities produce a strong instant
tential well (potential extremum region). The potential wells attract the electric field, which will trigger upward flashes in a few milliseconds to
leaders of opposite polarity into the central area and prevent their hundreds of milliseconds (Zhou et al., 2012), but there is still not a clear
outward expansion (MacGorman et al., 2001; Tan, 2006). As shown in definition on the nearby-lightning-triggered upward flashes in our
Table 1, the maximum potentials of lower negative potential wells model (Because there is no time factor for the lightning simulated in our
(PLPW) for these two upward negative flashes are less than −63.2 MV model). In order to make the definition of the upward flashes simulated
(reference potential of the ground is 0 MV). in the model close to the actual observation, we define that upward
67
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
flashes triggered by nearby lightning discharges within two steps are on the thundercloud electric field but also on the height of the building
nearby-lightning-triggered upward flashes. As shown in Fig. 2, UL4 and the neutral aerosol particle concentration of the site. Zhou (2015)
propagating into the lower positive charge region is triggered by the quantitatively investigate the effect of building height on the initiation
preceding upward lightning discharge shown in Fig. 1c. UL5 and UL6 of upward flashes. Because the ambient electric field is chosen to be the
are triggered by IC lightning discharges and propagate into the lower criterion for the initiation of upward flashes, and the magnitude of the
positive charge region. UL4, UL5, and UL6 are all nearby-lightning- ambient electric field mainly depends on the weighted average of
triggered upward flashes, where UL5 and UL6 have some similarities, bottom charge heights (defined as H) and the average of the bottom
for example, when these two upward flashes initiate, the thundercloud charge density (defined as D). Note that the average of the bottom
top heights in Fig. 2b and c are both 12.2 km, and the charge structures charge density is calculated by dividing the total surface charge density
are both tri-pole. However, the msl of the ground in the SEET case is of the bottom charge region by the area of the bottom charge region.
3 km, which is far greater than that in the NJ812 case, therefore, the The greater D is, the easier it is to initiate upward flashes. The higher H
thundercloud shown in Fig. 2c develops more exuberant. In addition, is, the harder it is to initiate upward flashes. Therefore, we will discuss
UL6 has more channel branches, and the maximum potential of the how D and H combine to affect the initiation of self-initiated upward
lower potential well for UL6 is greater than that for UL5. These dif- flashes in the following section.
ferences are related to the case adopted for simulation. Due to the difference of charge structures between SUNL A and
UL2, UL4, UL5, and UL6 are all upward positive flashes, their lea- SUPL B, these two types of upward flashes will be discussed separately.
ders propagate into the lower positive potential wells shown in Fig. 1e, The relationship between D and H in 45 SUNL A is shown in Fig. 3a. The
2d, e, and f. Table 1 shows the maximum potentials of the lower po- coefficient of determination R2 is 0.87. This means that the fitting effect
tential wells for them, the maximum potentials are always greater than is very good, and H and D have a good linear relationship. The fitting
36.8 MV. Combining with the analysis on the upward negative flashes function is marked in Fig. 3a. As D increases, the critical H for initiating
mentioned above, upward leaders in Figs. 1f and 2f propagate into the the SUNL A also increases. In other words, as D increases, upward
lower negative potential wells and form many branches, moreover, flashes can initiate when H is higher than just when H is low.
these upward leaders are confined to the lower strong potential wells. SUPL B has the similar characteristics to SUNL A. The relationship
While the other four upward leaders shown in Figs. 1d, e, 2d, and e pass between H and D in 22 SUPL B is shown in Fig. 3b. Although H center
through the lower positive potential wells and extend to the zero po- and D have a linear relationship, the coefficient of determination R2 is
tential line. This phenomenon is probably related to the maximum only 0.62. This means that the fitting effect for SUPL B is not as good as
potential of the lower potential well, and further research will be car- the result for SUNL A.
ried out in the future. Table 1 shows RLP for the six upward lightning As shown in Fig. 3, H under the dipole charge structure is generally
flashes, except for that for UL3, RLP for other five upward lightning higher than that under the tri-pole charge structure. H under the dipole
flashes are all greater than 45.3%. In addition, RLP before the two IC charge structure is mainly in the range of 7 km to 8 km, whereas H
lightning discharges (triggering UL5 and UL6) are also given, they are under the tri-pole charge structure is mainly in the range of 5.5 km to
10.1% and 21.5% respectively. RLP when upward flashes occur is sig- 7 km. In addition, D under the dipole charge structure is also overall
nificantly greater than RLP when upward flashes cannot occur. The greater than that under the tri-pole charge structure. D under the dipole
configuration relationship between lower charge and upper charge may charge structure is mainly in the range of −0.3 nC/m3 to −0.6 nC/m3,
determine whether upward flashes can occur and propagate into the whereas D under the tri-pole charge structure is mainly in the range of
lower charge region. Quantitative discussion will be given in sub- 0.25 nC/m3 to 0.35 nC/m3. The coefficients of the fitting function in
sequent studies. Fig. 3a and b are 2.14 and 8.46 respectively. The main reason for this
Table 2 shows the frequency of different types of lightning simu- difference is that the value of the ambient electric field is not only
lated by the SEET case and the NJ812 case. Out of 139 upward flashes, proportional to the square of H but also to D. As H increases, H has an
67 cases are self-initiated upward flashes, 72 cases are nearby-light- increasing influence on the value of the ambient electric field. As
ning-triggered upward flashes, and the SEET case is more favorable for mentioned above, H and D under the dipole charge structure are gen-
the initiation of self-initiated upward flashes than the NJ812 case. Ac- erally greater than those under the tri-pole charge structure. Therefore,
cording to the previous analysis, we think that the lower cloud base when D increases, the increment of H under the dipole charge structure
height in the SEET case (mountain thunderstorm) may be the main is much smaller than that under the tri-pole charge structure.
reason for this result. Compared with observations (Prentice and
Mackerras, 1977), the ratio of IC to CG lightning flashes is relatively
small, which may be related to excessive upward flashes. Upward fla- 3.3. Nearby-lightning-triggered upward flashes
shes neutralize the bottom charge, which causes the CG lightning to be
hard to occur. Nearby-lightning-triggered upward flashes are assumed to be trig-
gered by preceding lightning discharge activities. What needs to be
emphasized here is that this type of upward flashes can occur not only
3.2. Self-initiated upward flashes during the preceding lightning development, but also after the end of
the preceding lightning (Zhou et al., 2010, 2012; Warner et al., 2012a).
SUNL A and SUPL B are both self-initiated upward flashes. The re- The latter will be discussed in this section. As shown in Fig. 2b and c,
search on the triggering reason of self-initiated upward flashes has been UL5 and UL6 are triggered by preceding IC flashes or upward flashes.
carried out all these years. At present, there are some research results Wang et al. (2008) described the electric field changes (physics sign
on the triggering reason of self-initiated upward flashes. Becerra et al. convention) associated with the nearby-lightning-triggered upward
(2007) suggested that the initiation of upward flashes not only depends flashes from the lightning protection tower. The electric field had first a
Table 2
Frequency of different types of lightning in two thunderstorm cases.
Self-initiated upward flashes Nearby-lightning-triggered upward flashes IC lightning CG lightning
Total 67 72 85 12
SEET 66 34 40 6
NJ812 1 38 45 6
68
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
Fig. 3. Relationship between the weighted average of the bottom charge heights and the average of the bottom charge density under the dipole charge structure (a)
and the tri-pole charge structure (b). The abscissa is the average of the bottom charge density. The ordinate is the weighted average of the bottom charge heights. R2
is the coefficient of determination.
UL Initiation Site
UL Initiation Site
UL Initiation Site
UL Initiation Site
Fig. 4. Distribution of (a and b) electric field and (c and d) electric potential before IC lightning and after IC lightning. Purple diamonds represent the initiation points
of upward lightning (UL), Purple triangles represent the initiation points of IC lightning. Black solid lines are for positive electric field (physics sign convention) and
dashed lines are for negative electric field with contour values starting at ± 0 kV/m. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
positive (negative) change that lasted tens of milliseconds followed by a The electric field at the UL initiation site is about 300 V/m before the
negative (positive) change. Taking UL5 as an example, which conforms occurrence of IC lightning. Then the IC lightning occurs, which neu-
the characteristic described above. Fig. 4 give the distribution of (a and tralize the main negative charge region shown in Fig. 2b. The electric
b) electric field and (c and d) electric potential before and after IC field at the IC lightning initiation site decreases quickly accompanying
lightning. It can be found in Fig. 4a that the contour value of electric by a sudden increase of the electric field at the UL initiation site. After
field around the upward lightning (UL) initiation site is smaller than the occurrence of the UL, the lower positive charge region shown in
8 kV/m at 55.03 min (IC flash occurs at 55.06 min). However, the Fig. 2b is neutralized, and the electric field at the UL initiation site is
contour value rapidly increases and exceeds the initiation threshold reduced to the original level.
after the occurrence of IC lightning shown in Fig. 4b. This electric field Why IC lightning can trigger the upward flash? According to the
changes around the UL initiation site can be seen more clearly in Fig. 5. above analysis, IC lightning discharges neutralize the main negative
69
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
70
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
Fig. 6. Distribution of (a and b) electric field and (c and d) before preceding UL and after preceding UL. Purple diamonds represent the initiation point of UL
triggered, Purple squares represent the initiation point of preceding UL. Preceding UL is the self-initiation upward flash shown in Fig. 1c, UL is the upward flash
triggered by preceding UL. The details are same as those in Fig. 4. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
References
Aleksandrov, N.L., Bazelyan, E.M., Carpenter, R.B., Drabkin, M.M., Raizer, Y.P., 2001.
The effect of coronae on leader initiation and development under thunderstorm
conditions and in long air gaps. J. Phys. D. Appl. Phys. 34 (22), 3256–3266.
Aleksandrov, N.L., Bazelyan, E.M., Drabkin, M.M., Carpenter, R.B., Raizer, Y.P., 2002.
Corona discharge at the tip of a tall object in the electric field of a thundercloud.
Plasma Phys. Rep. 28 (11), 953–964.
Aleksandrov, N.L., Bazelyan, E.M., Raizer, Y.P., 2005. The effect of a corona discharge on
a lightning attachment. Plasma Phys. Rep. 31 (1), 75–91.
Bazelyan, E.M., Raizer, Y.P., Aleksandrov, N.L., 2008. Corona initiated from grounded
objects under thunderstorm conditions and its influence on lightning attachment.
Plasma Sources Sci. Technol. 17 (2), 431–438.
Becerra, M., Cooray, V., 2006a. A self-consistent upward leader propagation model. J.
Phys. D. Appl. Phys. 39 (16), 3708–3715.
Becerra, M., Cooray, V., 2006b. A simplified physical model to deter- mine the lightning
upward connecting leader inception. IEEE Trans. Power Del. 21 (2), 897–908.
Fig. 7. Changes of electric field at the UL initiation site and at the main ne- Becerra, M., Cooray, V., Soula, S., Chauzy, S., 2007. Effect of the space charge layer
gative charge region. Red line represents the changes of electric field at the created by corona at ground level on the inception of upward lightning leaders from
initiation site of the triggered UL, blue line represents the changes of electric tall towers. J. Geophys. Res. 112, D12205. https://doi.org/10.1029/2006JD008308.
field at the density center of the main negative charge region. (For inter- Becerra, M., Long, M., Schulz, W., Thottappillil, R., 2018. On the estimation of the
lightning incidence to offshore wind farms. Electr. Power Syst. Res. 157, 211–226.
pretation of the references to colour in this figure legend, the reader is referred
Berger, k, 1967. Novel observations on lightning discharges, results and research on
to the web version of this article.) Mount San Salvatore. J. Frankl. Inst. 283 (6), 478–525.
Berger, K., 1977. The Earth flash. In: Golde, R.H. (Ed.), Lightning: Physics of Lightning.
Vol. 1. Elsevier, New York, pp. 119–190.
Acknowledgments Berger, k, Vogelsanger, E., 1969. New results of lightning observations. Planet.
Electrodyn. 1, 489–510.
The work was supported by National Key Research and Berger, K., Anderson, R.B., Knoninger, H., 1975. Parameters of lightning flashes. Electra
41, 23–37.
Development Program of China (2017YFC1501504), National Natural Chauzy, S., Raizonville, P., 1982. Space charge layers created by coronae at ground level
Science Foundation of China (Grant No. 41875003, 41805002), Startup below thunderclouds: measurements and modeling. J. Geophys. Res. 87 (C4),
Foundation for Introducing Talent of NUIST (2016r042), Natural 3143–3148. https://doi.org/10.1029/JC087iC04p03143.
Coleman, L.M., Marshall, T.C., Stolzenburg, M., Hamlin, T., Krehbiel, P.R., Rison, W.,
Science Foundation of Jiangsu Province (No.BK20180808) and Natural Thomas, R.J., 2003. Effects of charge and electrostatic potential on lightning pro-
Science Fundamental Research Project of Jiangsu Colleges and pagation. J. Geophys. Res. 108 (D9), 4298. https://doi.org/10.1029/2002JD002718.
Universities (No. 18KJB170010). D'Alessandro, F., Gumley, J.R., 2001. A “Collection Volume Method” for the placement of
air terminals for the protection of structures against lightning. J. Electrost. 50 (4),
279–302.
71
Y. Tan et al. Atmospheric Research 217 (2019) 63–72
Diendorfer, G., Pichler, H., Mair, M., 2009. Some parameters of negative upward-initiated Shi, Z., Tan, Y.B., Tang, H.Q., Sun, J., Yang, Y., Peng, L., Guo, X.F., 2015. Aerosol effect on
lightning to the Gaisberg tower (2000–2007). IEEE Trans. Electromagn. Compat. 51 the land-ocean contrast in thunderstorm electrification and lightning frequency.
(3), 443–452. Atmos. Res. 164, 131–141.
Flache, D., Rakov, V.A., Heidler, F., Zischank, W., Thottappillil, R., 2008. Initial-stage Shi, Z., Tang, H.Q., Tan, Y.B., 2016. Effects of the inductive charging on the electrification
pulses in upward lightning: leader/return stroke versus M-component mode of charge and lightning discharges in thunderstorms. Terr. Atmos. Ocean. Sci. 27, 241–251.
transfer to ground. Geophys. Res. Lett. 35, L13812. https://doi.org/10.1029/ https://doi.org/10.3319/TAO.2015.12.10.01(A.
2008GL034148. Smorgonskiy, A., Rachidi, F., Rubinstein, M., Diendorfer, G., Schulz, W., 2013. On the
Fuchs, F., Landers, E.U., Schmid, R., Wiesinger, J., 1998. Lightning current and magnetic proportion of upward flashes to lightning research towers. Atmos. Res. 129, 110–116.
field parameters caused by lightning strikes to tall structures relating to interference Solomon, R., Schroeder, V., Baker, M.B., 2001. Lightning initiation-conventional and
of electronic systems. IEEE Trans. Electromagn. Compat. 40 (4), 444–451. runaway breakdown hypotheses. Q. J. R. Meteorol. Soc. 127 (578), 2683–2704.
Guimarães, M., Araujo, L., Pereira, C., Mesquita, C., Visacro, S., 2014. Assessing currents Standler, R.B., Winn, W.P., 1979. Effects of coronae on electric fields beneath thunder-
of upward lightning measured in tropical regions. Atmos. Res. 149, 324–332. storms. Q. J. R. Meteorol. Soc. 105 (443), 285–302.
Guo, F., Lu, G., Wu, X., Wang, H., Liu, Z., Bao, M., Li, Y., 2016. Occurrence conditions of Suzuki, T., 1992. Long term observation of winter lightning on. Japan Sea coast. Res. Lett.
positive cloud-to-ground flashes in severe thunderstorms. Sci. China Earth Sci. 59 (7), Atmos. Electr. 12, 53–56.
1401–1413. Takagi, N., Wang, D., Watanabe, T., 2006. A study of upward positive leaders based on
Guo, X.F., Zhang, Q.L., 2017. Effects of geometrical parameters of two height-unequal simultaneous observation of E-fields and high-speed images. IEEJ Trans. Fundam.
adjacent objects on corona discharges from their tips during a thunderstorm. Atmos. Mater. 126 (4), 256–259.
Res. 190, 113–120. Tan, Y.B., 2006. Numerical Simulations of the Relationship of the Lightning Discharge
Guo, X.F., Zhang, Q.L., Zhang, J.B., 2017. Improvement of corona discharge model and its with the Space Charge and Potential Distribution in Thundercloud. in Chinese. PhD
application on simulating corona discharge in the presence of wind. Math. Probl. Eng. dissertation. University of Science and Technology of China.
2017, 9853439. https://doi.org/10.1155/2017/9853439. Tan, Y.B., Tao, S.C., Zhu, B.Y., 2006. Fine resolution simulation of the channel structures
Heidler, F., Manhardt, M., Stimper, K., 2014. Self-initiated and other-triggered positive and propagation features of intra-cloud lightning. Geophys. Res. Lett. 33, L09809.
upward lightning measured at the Peissenberg Tower, Germany. In: Paper presented https://doi.org/10.1029/2005GL025523.
in 2014 International Conference on Lightning Protection (ICLP). IEEE, pp. 157–166. Tan, Y.B., Tao, S.C., Zhu, B.Y., Ma, M., Lu, W.T., 2007. A simulation of the effects of intra-
Heldson, J.H., Wu, G., Farley, R.D., 1992. An intra-cloud lightning parameterization cloud lightning discharges on the charges and electrostatic potential distributions in a
scheme for a storm electrification model. J. Geophys. Res. 97, 5865–5884. https:// thundercloud. Chin. J. Geophys. 50 (4), 916–930.
doi.org/10.1029/92JD00077. Tan, Y.B., Guo, X.F., Zhu, J.R., Shi, Z., Zhang, D.D., 2014a. Influence on simulation ac-
Iudin, D.I., Davydenko, S.S., 2015. Fractal model of a compact intracloud discharge. I. curacy of atmospheric electric field around a building by space resolution. Atmos.
Features of the Structure and Evolution. Radiophys. Quantum Electron. 58 (7), Res. 138, 301–307.
477–496. Tan, Y.B., Tao, S.C., Liang, Z.W., Zhu, B.Y., 2014b. Numerical study on relationship be-
Iudin, D.I., Rakov, V.A., Mareev, E.A., Iudin, F.D., Syssoev, A.A., Davydenko, S.S., 2017. tween lightning types and distribution of space charge and electric potential. J.
Advanced numerical model of lightning development: application to studying the Geophys. Res. 119, 1003–1014. https://doi.org/10.1002/2013JD019983.
role of LPCR in determining lightning type. J. Geophys. Res. 122, 6416–6430. Tao, S.C., Tan, Y.B., Zhu, B.Y., Ma, M., Lu, W.T., 2009. Fine resolution simulation of
https://doi.org/10.1002/2016JD026261. cloud-to-ground lightning and thundercloud charge transfer. Atmos. Res. 91,
Jiang, R.B., Qie, X.S. Wu, J, Z., Wang, D.F., Liu, M.Y., Lu, G.P., Liu, D.X., 2014. 360–370.
Characteristics of upward lightning from a 325-m-tall meteorology tower. Atmos. Wang, D., Takagi, N., 2012. Characteristics of winter lightning that occurred on a
Res. 149, 111–119. windmill and its lightning protection tower in Japan. IEEJ Trans. Power Energy 132
Liu, D.X., Qie, X.S., Pan, L.X., Peng, L., 2013. Some characteristics of lightning activity (6), 568–572. https://doi.org/10.1541/ieejpes.132.568.
and radiation source distribution in a squall line over North China. Atmos. Res. 132, Wang, D., Takagi, N., Watanabe, T., Sakurano, H., Hashimoto, M., 2008. Observed
423–433. characteristics of upward leaders that are initiated from a windmill and its lightning
Lu, W.T., Wang, D.H., Zhang, Y., Takagi, N., 2009. Two associated upward lightning protection tower. Geophys. Res. Lett. 35, L02803. https://doi.org/10.1029/
flashes that produced opposite polarity electric field changes. Geophys. Res. Lett. 36, 2007GL032136.
L05801. https://doi.org/10.1029/2008GL036598. Warner, T.A., 2010. Observations of simultaneous multiple upward leaders from tall
MacGorman, D.R., Straka, J.M., Ziegler, C.L., 2001. A lightning parameterization for structures. In: Paper Presented at 30th International Conference on Lightning
numerical cloud models. J. Appl. Meteorol. 40 (3), 459–478. Protection. University of Bologna, Cagliari, Italy.
Mansell, E.R., MacGorman, D.R., Straka, J.M., 2002. Simulated three-dimensional bran- Warner, T.A., 2012. Observations of simultaneous upward lightning leaders from multiple
ched lightning in a numerical thunderstorm model. J. Geophys. Res. 107 (D9), 4075. tall structures. Atmos. Res. 117, 45–54.
https://doi.org/10.1029/2000JD000244. Warner, T.A., Cummins, K.L., Orville, R.E., 2012a. Upward lightning observations from
Mazur, V., Ruhnke, L.H., Warner, T.A., Orville, R.E., 2011. Discovering the nature of towers in Rapid City, South Dakota and comparison with national lightning detection
recoil leaders. In: Paper Presented at 14th International Conference on Atmospheric network data, 2004-2010. J. Geophys. Res. 117, D19109. https://doi.org/10.1029/
Electricity, International Conference on Atmospheric Electricity. Brazil, Rio de 2012JD018346.
Janeiro. Warner, T.A., Saba, M.M.F., Ridge, S., Bunkers, M., Lyons, W.A., Orville, R.E., 2012b.
McEachron, K.B., 1939. Lightning to the empire state building. J. Frankl. Inst. 227 (2), Lightning-triggered upward lightning from towers in Rapid City, South Dakota. In:
149–217. Paper Presented at 4th International Lightning Meteorology Conference. Vaisala,
Miki, M., Rakov, V.A., Shindo, T., Diendorfer, G., Mair, M., Heidler, F., et al., 2005. Initial Boulder, Colo.
stage in lightning initiated from tall objects and in rocket-triggered lightning. J. Yang, N., Zhang, Q., Hou, W., Wen, Y., 2017. Analysis of the lightning attractive radius
Geophys. Res. 110, D02109. https://doi.org/10.1029/2003JD004474. for wind turbines considering the developing process of positive attachment leader. J.
Peek, F.W., 1929. Dielectric Phenomena in High-Voltage Engineering. McGraw-Hill, New Geophys. Res. 122, 3481–3491. https://doi.org/10.1002/2016JD026073.
York. Zhou, B.W., 2015. Numerical Simulation of Upward Lightning Trigger and Spread. in
Pichler, H., Diendorfer, G., Mair, M., 2010. Some parameters of correlated current and Chinese. Nanjing University of Information Science and Technology, Nanjing,
radiated field pulses from lightning to the Gaisberg Tower. IEEJ Trans. Electr. Jiangsu, China.
Electron. Eng. 5, 8–13. Zhou, H., Diendorfer, G., Thottappillil, R., Pichler, H., Mair, M., 2010. Simultaneous
Prentice, S.A., Mackerras, D., 1977. The ratio of cloud to cloud-ground lightning flashes in current and electric field observations of upward negative leaders initiated from the
thunderstorms. J. Appl. Meteorol. 16 (5), 545–550. Gaisberg Tower. In: Paper Presented at 2010 Asia-Pacific Symposium on
Qie, X.S., Liu, X.S., Soula, S., Chauzy, S., 1994. Influence of ion attachment on vertical Electromagnetic Compatibility. Tsinghua Univ., Beijing.
distribution of electric field and charge density under thunderstorm. Ann. Geophys. Zhou, H., Diendorfer, G., Thottappillil, R., Pichler, H., Mair, M., 2012. Measured current
12, 1218–1228. and close electric field changes associated with the initiation of upward lightning
Rakov, V.A., Uman, M.A., 2003a. Upward lightning initiated by ground-based objects. In: from a tall tower. J. Geophys. Res. 117, D08102. https://doi.org/10.1029/
Lightning: Physics and Effects. Cambridge University Press, Cambridge, U. K., pp. 2011JD017269.
176–182 Chap. 6. Ziegler, C.L., MacGorman, D.R., 1994. Observed lightning morphology relative to mod-
Rakov, V.A., Uman, M.A., 2003b. Winter lightning in Japan. In: Lightning: Physics and eled space charge and electric field distributions in a tornadic storm. J. Atmos. Sci.
Effects. Cambridge University Press, Cambridge, U. K., pp. 308–320 Chap. 8. 51, 833–851.
Saba, M.M., Schumann, C., Warner, T.A., Ferro, M.A.S., Paiva, A.R., Helsdon, J., Orville, Ziegler, C.L., MacGorman, D.R., Dye, J.E., Ray, P.S., 1991. A model evaluation of non-
R.E., 2016. Upward lightning flashes characteristics from high-speed videos. J. inductive graupel-ice charging in the early electrification of a mountain thunder-
Geophys. Res. 121 (14), 8493–8505. https://doi.org/10.1002/2016JD025137. storm. J. Geophys. Res. 96, 12833–12855. https://doi.org/10.1029/91JD01246.
72