30th International Conference on Lightning Protection - ICLP 2010

(Cagliari, Italy - September 13th -17th, 2010)

SIMPLIFIED ESTIMATION OF CHARGE TRANSFER IN POINT-CHARGE

MODEL THROUGH MEASUREMENT OF E-FIELD
Koji Michishita Shinji Kawamoto
Shizuoka University Chugoku Electric Power Co.
[email protected] [email protected]

In all stations but one at Higashihiroshima, the two-

ABSTRACT
dimensional horizontal components of magnetic field
A charge transfer is one of the most important parameters were measured by using loop antennas and the vertical
when the damage to the equipments on a power electric field was measured by using the electrostatic
transmission line due to the high energy of the lightning antenna. The measuring system for the electric field and
discharge is discussed. The amount of charge transfer
the magnetic field had lower cutoff frequency of 1.6 Hz,
associated with a lightning flash or individual strokes can be
estimated by using the electric field measured with a slow
important for the accurate measurement of the slow E-
antenna. In this paper, we propose a simplified method field change. At Higashihiroshima, only the vertical
where the height and the amount of a charge is estimated electric field was measured and the system had the lower
from the measured electric field change, and the accuracy of cutoff frequency range of 8.9Hz. The sampling interval
the estimated results is roughly estimated. of the recording system was 1 µs and the field waveforms
for 250 ms prior to the trigger and for 750 ms after the
1 INTRODUCTION trigger were recorded. The triggered time of the recording
A charge transfer is one of the most important system was stamped by a clock with a precision of about
parameters when the damage to the equipments on a 200ns.
power transmission line by the huge energy of the
lightning is discussed. One can accurately evaluate the
charge transfer by integrating the current associated with
lightning flash, however, it takes a lot of time to
accumulate the statistically meaningful amount of data. It
is helpful to estimate the energetic parameters based on
the measured electric field waveforms [1] in
accumulating those in the wide area efficiently. Please
refer to [2] for the detailed review on the method of
analysis of the charge transfer from the measured electric
field.
The authors have proposed a simplified method of
estimation of charge transfer associated with a return
stroke [3, 4]. In this paper, we propose a simplified
method where the height and the amount of a charge is
estimated from the measured electric field change, and
the accuracy of the estimated results is roughly estimated..

2 MEASUREMENT
Figure one shows the location of one of the four
measuring stations of the electric field at Kake in the
Chugoku area in Japan. Kake lies in a basin at 200m
above sea level. Other stations are located at Iwakuni,
Hofu, and Higashihiroshima, respectively. Iwakuni
located at 53km in the south-southwest direction from
Kake, Hofu located at 93km in the southwest direction
from Kake, and Higashihiroshima located at 50km in the
southeast direction from Kake. The sensor at Hofu moved
to Yamaguchi in 2007, about 10 km north of Hofu. Fig.１,Location of Measuring station and striking point

1065-1
 3 METHOD OF ANALYSIS the charge transfer by using (1) when the electric field is
measured at least at one station.
To estimate the charge transfer associated with a return
The net electric field change (E) associated with a
stroke, the charge in the cloud is assumed a point charge
return stroke is evaluated by the difference between the
as shown in Fig. 2.
electric field before the beginning of the leader and that
+Q
Q after the end of a return stroke current including the
continuing current.
In [3], we have presented a method of analysis of the
H charge transfer from the measured electric field obtained
at least two stations. In [3], D is estimated from the
above-mentioned way, and two parameters, namely H
and Q in (1) were evaluated by solving the simultaneous
equations in the case that the electric field is obtained at
E－ E two stations. When the electric field is obtained at three
－
stations, H and Q in (1) were evaluated by the least
EE
total
squares error minimization technique.
-Q
－Q
4 RESULTS
D
4.1 Charge transfer
Fig. 2 Point charge model to evaluate the E-field at the An example of the measured electric and magnetic
surface of the ground. field waveform associated with a negative flash is shown
in Fig. 3. Positive field is defined downward regarding
The lower charge (-Q) in Fig. 2 representing the the vertical electric field and is defined from the east to
charge on surface of the ground is the mirror image of the the west with respect to the horizontal magnetic field.
upper charge (Q). Since the electric field in Fig. 3 (b) goes upward, one can
When the electric field produced by the lower charge is know that the field is associated with a negative flash
taken into account, the charge (Q) can be related to the defined bringing down negative charges to the ground.
net electric field (E) as in (1). The time zero in the figure is the time of the trigger of the
recording system, which is shown in the figure.
2π 0 D 3E{1  ( H D ) 2 }3 2
Q  (1) In this case, three negative strokes were identified in
H the measured waveforms. The location of the first stroke
where ε0 is the permittivity of the free space, E is the estimated by the LLS is given in Fig. 1 and the distance
electric field change, D is the horizontal distance between (D) of the first stroke from the station at Kake is
station and striking point, H is the charge height from the estimated 22km. The crest value of current was estimated
ground . -37kA by the LLS. As electric field waveform includes
In the estimation of the charge transfer, the charge is some 60Hz commercial frequency component, the
postulated just above the lightning striking point. In other influence of this component is reduced by moving
words, the lightning channel is postulated to be vertical to average. Then we estimate the change of electric field
the ground. This postulation makes it possible to estimate (E) by the difference between the maximum value and
the horizontal distance D in (1) as the distance from the the minimum value. In Fig. 3 (c), this change is estimated
station to the lightning striking point provided by LLS 68V/m: the minimum value is found before the beginning
(Lightning Location System). In this paper, the location of the leader and the maximum value is given at about
of the return stroke provided by the LLS operated by the 6ms after the first stroke.
Chugoku Electric Power Co. is used where the IMPACT The height of the negative charge center is estimated as
sensors were used as the DF. follows. At first, in electric field waveform we calculate
Furthermore, H in (1) is also estimated as the product the time interval (t) between the beginning of the
of the duration of the stepped leader evaluated from development of the negative leader assumed to be signed
measured electric and magnetic field waveforms and the by preliminary breakdown pulse and the onset of the
average speed of the stepped leader. In the analysis, the return stroke. Speed of stepped leader varies with strokes
duration of the stepped leader is approximated as the time and it is often within 0.1 ~ 0.4 km/ms. The product of the
difference between the beginning of the preliminary average speed of stepped leader of 0.2 km/ms by the
breakdown pulse and the onset of the return stroke. above time interval (t) is assumed to be the height of
If the above two parameters, namely D and H, can be charge center. In this way the value (H) is estimated to be
given by the above-mentioned methods, one can evaluate 4 km.

1065-2
 Substituting 22 km, 4 km, 68 V/m for D, H, ⊿E in (1) Figure 5 shows cumulative frequency distribution of
respectively, Q is estimated 10 C. the estimated impulse charge for negative first strokes by
In the period from July 10 to 31 in 2005, where the the proposed method. Table 1 shows statistics of the
sensor at Higashihiroshima was not in operation, the estimated impulse charge by the proposed method and
charge transfer associated with 254 return strokes were the impulse charge evaluated through current observation
estimated from the measured electric field at Kake by the in Switzerland [5] for downward negative first strokes.
method proposed in this paper. Figure4 shows location of
those 254 striking points of negative first strokes. In the
analysis, the return strokes striking at the distance of less
than 30km from the station at Kake were subject to
analysis. With conventional method [3, 4], we could
estimate charge transfer associated with only 43 (17% of
254) return strokes.
200

150
Bew-Field [nT]

100 Preliminary
breakdown
pulse
50

0
-30 -20 -10 0 10 20 30
-50 Time [ms]
(a) B-field waveform. Fig. 4 Location of 254 striking point of negative first
strokes estimated by LLS
100

80 99.99
05 08/25 18:06:59.9919250
E-Field[V/m]

60 99.9
40
99
Cumulative frequency distribution [%]

20

0 95
-400 -200 0 200 400 600 800 90
-20
Time [ms]
80
-40 70
(b) E-field waveform. 50
80 30
70 20
60 10
 E = 68V/m 50 5
E-Field[V/m]

t = 20.7ms
40 1
30
20 .1
10 05 08/25 18:06:59.9919250
.01
0 0.1 1 10 100
-40 -30 -20 -10 -10 0 10 20 30 40
Charge Transfer [C]
Time [ms]

(c) E-field waveform in expanded scale Fig.5, Cumulative frequency distribution of charge transfer
Fig. 3 Example of measured electric and magnetic field for negative first strokes estimated by the proposed method.
waveforms.

1065-3
 Table 1 Statistics of Impulse charge for negative first Although the property of lightning between the
strokes Chugoku area in Japan and in Switzerland might not be
the same, it is possible that there is not large difference
95% 50% 5%
between them. Therefore, fairly good agreement between
This study 1.3C 4.4C 23C
Ref [5] 1.1C 5.2C 24C
the statistical values estimated by the proposed method in
this paper and those in [5] might indicate that the
proposed method of estimation of the charge transfer
In [5] it is reported that the 5%, 50% and 95% values
including the measurements of the slow electric field
of the charge transfer associated with downward negative
change is effective.
first strokes were 1.1C, 5.2C and 24C, respectively. Our
For negative first strokes, correlation coefficient of
corresponding statistical values of the charge transfer
peak current and impulse charge by authors’ method is
were 1.3C, 4.4C and 23C. The 50% value of the charge
0.37. Compared with Berger s’ coefficient 0.61, this is
transfer in this study is 15% smaller than that reported in
fairly small. This is partly because the return strokes with
[5]. The 95% values and the 5% values estimated by the
small peak currents are subject to analysis in this study.
proposed method were differed from those reported in [5]
In other words, the duration of the current plays dominant
with the discrepancy of less than 20%. In the authors’
roles in the determination of the charge transfer in this
opinion, the estimated values by the authors’ method of
study.
analysis approximately agree with the values in [5].
The average value of H (height of charge center) is Table 2. Statistics of current peak for negative first
6.9km and this value fairly agrees with the height of the strokes
temperature of –10 °C in summer thunderstorms in
Florida [6]. 95% 50% 5%
Figure 6 shows cumulative frequency distribution of This study -10k A -20k A -50k A
the current peak for negative first strokes estimated by Ref [5] -14k A -30k A -80k A
LLS, and Table 2 shows statistics of the current peak
estimated by LLS and the current peak observed in Table 3 shows statistics of impulse charge estimated
Switzerland [5] for downward negative first strokes. The by the proposed method and that observed by Berger et al
estimated peak by LLS in this study is smaller by 30% for downward positive first strokes. In the table, the
than the value observed in Switzerland [5]. We statistics of the current peak estimated by LLS is shown
hypothesize that this is the cause of the above-mentioned together with the current peaks reported in [5].
small value of the impulse charge in this study. It is reported that the 50 % and 95 % values of the
charge transfer associated with downward positive first
99.99
strokes were 16C, 150C, respectively. The 50% and 95%
99.9 values of the charge transfer estimated by the proposed
method were 5C and 64C, respectively, for positive first
strokes in summer. These statistical values of the charge
Cumulative frequency distribution [%]

99
transfer estimated in this study are less than 15% of the
95 reported values in [5]. Although the 95% value of current
90 peak is almost the same value, the charge estimated in
80 this study is 30% of the 95% value of the charge transfer
70
reported in [5].
50 This might indicate that the point charge model
30 adopted in the estimation of the charge might not
20 appropriate in the case of positive strokes. The
10 assumption that the charge center of the positive stroke is
5 above the striking point might not be appropriate either.

1 Table 3. Statistics of impulse charge and current peak

for positive first strokes
.1
No. of 95% 50% 5%
.01 events value value value
1 10 100
impulse This study 26 0.6 5 64
Current [kA]
charge [C] Ref [5] 25 2.0 16 150
current peak This study 26 5 14 57
Fig.6, Cumulative frequency distribution of current peak for [kA] Ref [5] 20 4.6 35 250
negative first strokes.

1065-4
4.2 Charge height 10
Figure 7 shows the distribution of the height of the 9
charge center evaluated from the duration of the leader
development preceding the return strokes observed on 8

Height [km]
July 27, 2008. Among 377 field waveforms associated 7
with negative first strokes on July 27 2008, the
preliminary breakdown pulse can be identified in 143 6
cases. The height of the charge center was in the range
from 1.9 km to 26 km, with the average and the standard 5
deviation being 6.5 km and 3.3 km, respectively. 4
Although the standard deviation is not small, the number
of data whose height was estimated to be less than half or 3
larger than twice of the average height of 6.5 km, was -25 -20 -15 -10 -5 0
only 13 and 4 , respectively. Therefore, it might be Temperature [°C]
possible that the statistics of the lightning parameters are
not greatly influenced by the variation of the estimated Fig.8. Distribution of high-altitude temperature on seven
height of the charge center. thunderstorm days on July 2008.
Figure 8 shows the distribution of the upper
Table 4 shows the average and the standard deviation
atmospheric temperature reported in [7] on 7
of the height of the charge estimated from the leader
thunderstorm days in July 2008. The temperature
duration. The average for 4 thunderstorm days is from 5.3
measured at two stations close to Kake was employed:
km to 6.5 km, in the range with temperatures from -5 to -
One station was at Yonago, about 130 km north-east of
15 °C in Fig. 8. This might indicate that the negative
Kake, and the other station was in Fukuoka, about 210
charge center of the first stroke can be estimated fairly
km south-west of Kake. At those stations, the
well from the duration of the leader development.
temperature was measured at 9 and 21 o’clock by using
the radio sonde. The charge center of the negative strokes Table. 4 Average value of estimated height of a charge
is inferred in the range whose temperature is -5 to -15 °C. in the point-charge model on 4 thunderstorm days.
This range is found in the height from 5.2 km to 7.9 km
in Fig. 8 with the median of 6.6km which agrees with the 7/19 7/27 7/28 7/30
height of the temperature of -10 °C for Florida Estimated height [km] 5.3 6.5 6.5 5.9
thunderstorm in summer [6]. It is interesting that the Standard deviation [km] 2.3 3.3 2.8 1.4
range of the height for a constant temperature from -5 to -
15 °C is found within 1 km. Therefore, it is reasonable to
assume the height of the charge center to be a constant 4.3 Comparison of estimated charge transfer
value around 6.5 km in the estimation of the charge Figure 9 shows the location of return strokes estimated
transfer. For the conservative-side estimate, the height of by LLS, picked out on condition that the electric field
the charge center should be assumed about 5 km. waveforms are obtained at more than 3 stations and the
40 pulses characteristic to the preliminary breakdown can be
seen in the electric or magnetic field waveforms before
the onset of return strokes. Those recordings are obtained
30 on July 30 2008 and the height of charge center estimated
from the duration of the leader development distributed
Frequency

20 in the range from 2.5 km to 12 km, with the average of

5.9km.
As mentioned above, the height and the charge transfer
10 of a point charge is estimated by using the least square
error minimization technique in the conventional method
when the electric field is obtained at more than 2 stations.
0 In the calculation, the height is varied from 0.1 km to 25
<3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 >10 km and in 19 cases out of 34, the best fit is given when
Height [km] the height of the charge center is either the upper or the
lower limits of the assumed range. This might be because
the charge center is not above the lightning striking point
Fig. 7 Distribution of estimated height of a charge in the and/or the measurement accuracy of the electric field
point-charge model on July 27 2008.
becomes low as the amplitude of the electric field change

1065-5
decreases with the increase of the distance to the 15
Correlation Coefficient = 0.53
lightning striking points. Further, other four cases are

Height by proposed method [C]

also excluded due to the low resolution of the electric
field amplitude.
10
Figure 10 and 11 show the relation between the charge
transfer and the height of the charge estimated by the
method proposed in this paper and that estimated by
applying least squares error minimization technique [3, 4], 5
respectively.

0
0 5 10 15
Height by conventional method [C]

Fig.11. Correlation between the charge height estimated by

using least square minimization technique and that
estimated by method proposed in this paper.

5 CONCLUSION
We propose the new method of estimation of charge
amount associated with return stroke using measurement
of the electric field. With this method one can estimate
Fig. 9 Location of 34 striking points of negative first strokes the charge transfer from the electric field at one station.
estimated by LLS subject to analysis of charge transfer. Statistics of the charge transfer for negative first
strokes approximately agree with the published values [5].
20 The average height of the negative charge center of the
Correlation Coefficient = 0.93 first strokes for 4 thunderstorm days is from 5.3 km to 6.5
km, in the range with temperatures from -5 to -15 °C.
Charge by proposed method [C]

15 These might indicate that the proposed method of

estimation of the charge including the estimation of the
charge height might be fairly appropriate for negative
10 first strokes. However, it is also indicated that the charge
transfer associated positive first strokes might not be
estimated with good accuracy by the proposed method.
5 6 REFERENCES
[1] P. R. Krehbiel, M. Brook, R. McCrory, “An analysis of the
charge structure of lightning discharges to the ground”, J.
0 Geophys. Res.,84, 2432-2456 (1979)
0 5 10 15 20 [2] W. J. Koshak, E. P. Krider and M. J. Murphy, “A
Charge by conventional method [C] Multipole expansion method for analyzing lightning field
changes”, J. Geophys. Res.,104, 9617-9633 (1999)
Fig.10. Correlation between impulse charge estimated by [3] K. Michishita, S. Kawamoto, K. Maeda, “Estimation of
using least square minimization technique for electric fields Charge Transfer Based on Measurement of Vertical
and estimated by method proposed in this paper. Electric Field”, IEEJ-EIT Joint Symposium on Advanced
Technologies on Power System, HV-07-116, PSE-07-175,
The correlation coefficient of the charge is 0.93 and Bangkok, Thailand(2007, 11, 19)
the ratio of the charge estimated by the proposed method [4] S. Kawamoto, K. Maeta, K. Michishita, “Proposal of
charge estimation method associated with return stroke”,
to that estimated by applying the least square error
2009 CIGRE SC C4 Meeting & Colloquium in Japan, S3-5,
minimization technique is from 0.78 to 1.35. If one Kushiro (2009.6.8)
speculates that the charge transfer estimated by the least [5] K. Berger, R. B. Anderson, and H. Kroninger: "Parameters
squares error minimization technique is accurate, the of lightning flashes", ELECTRA, No.41, pp.23-37 (1975)
maximum error by the proposed method can be estimated [6] M.A.Uman：”The lightning Discharge” ,Academic Press
35%. However, the height estimated by the proposed pp.61 (1987)
method seems practical in Fig. 11. [7] http://www.jma.go.jp/jma/index.html

1065-6