You are on page 1of 4

30th International Conference on Lightning Protection - ICLP 2010

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

EVALUATION OF 45 NEGATIVE FLASHES BASED ON E-FIELD


MEASUREMENTS, VIDEO DATA AND
LIGHTNING LOCATION DATA IN AUSTRIA
Schulz W., Pichler H., Diendorfer G.
Austrian Lightning Detection and Information System (ALDIS)
Kahlenberger Str. 2A, 1190 Vienna, Austria,
w.schulz@ove.at; h.pichler@ove.at; g.diendorfer@ove.at

time synchronized to the FM system and therefore also


ABSTRACT
the camera data is GPS time stamped. A red filter type 29
In this paper we present an analysis of 45 negative flashes (dark red) was used in front of the camera lens in order to
based on video and field measurement data. The analysis achieve better image contrast during daytime recordings.
based on video and E-field measurement data is an ideal
method to check lightning location data in terms of cloud- 3 DATA AND METHODOLOGY
to-cloud or cloud-to-ground classification. We will show that
the flash and stroke detection efficiency determined by this In summer 2009 a total of 45 negative flashes were
analysis is quite similar to the result from measurements at recorded during two thunderstorms. One thunderstorm
the Gaisberg Tower. We further present some lightning occurred on the 29th of June and one on the 3rd of
parameters for negative flashes and compare them to values August. Both measurements were done at the same
available in the literature. location in the south of Vienna (48.0798° N/13.0096° E).
This recording site was chosen because of the low
1 INTRODUCTION
electromagnetic noise level and it was already used for
During the last years data from the Austrian Lightning other lightning studies in Austria (see e.g. [3] or [4]). The
Location System (LLS) ALDIS were presented at several closest flashes occurred at a distance of 300 m and the
conferences. For negative flashes statistics on most distant flash at a distance of 28 km.
multiplicity, percentage of single stroke flashes and the The video and E-Field data sets are complementing
ratio of first to subsequent stroke electric field peaks in each other because sometimes we could identify strokes
Austria differ significantly from measurements in other only either on the video or in the E-field data and
regions of the world [1]. sometimes the strokes could be clearly identified in both
To be able to validate the LLS data a field data sets. As an example record #57 is shown in Fig. 1.
measurement system (FM-System) was developed which This flash contained three strokes and each stroke
records E-Field data continuously [3-4]. When evaluating exhibits a separated ground strike point. According to the
field measurement data only, there is still an uncertainty LLS data the distance of the individual ground strike
of about 10% regarding flash categorization (cloud-to- points from the measurement system were in the range
ground or cloud-to-cloud). To reduce this uncertainty the from 5 to 7 km.
measurement system was recently extended by a GPS
time synchronized digital video camera. 4 RESULTS

2 MEASUREMENT SYSTEM 4.1 Lightning parameters


The used FM-System was already described in detail The mean multiplicity for the 45 flashes determined
in several papers [3-5]. The monochrome video camera from the FM system and the video camera data is 3.0. For
system integrated to the previously used FM-System is this dataset the LLS determined a mean multiplicity of
able to record up to 200 frames per second with VGA 2.7. A percentage of single stroke flashes of 20% was
resolution (640x480 pixels). This frame rate is equivalent determined from the FM system and the video camera
to an exposure time of 5 ms per frame and should be data. This value is significantly smaller compared to
sufficient to detect almost all the individual strokes in a previous studies made in Austria (e.g. 33% [3] or 50%
flash. [4]). From the LLS data we determined 25% single
At the maximum frame rate of 200 frames per second strokes.
this monochrome camera (8 bit per pixel) produces a data
stream of 60 Mbytes per second. The camera is strictly

1011-1
1st Stroke 2nd Stroke 3rd Stroke
Figure1. Video images and correlated electric field data (in V/m) for flash #57 (03.08.2009, 13:05:54)

The mean number of ground strike points per flash is 4.2 LLS performance
1.93. This number was determined from 40 flashes Detection efficiency (DE) and location accuracy are
because in five cases no video data were available. 53% the most important performance parameters of LLSs and
of the flashes had more than one ground contact point. therefore we tried to determine those performance
Due to the reason that the time resolution for video parameters from the data recorded during this
recordings with 200 fps is only 5 ms continuing current measurement campaign. During the measurement
parameter analysis is limited to the so called long camping on the two days the LLS detected 44 out of 45
continuing current (CC). According to [7] or [8] a long negative flashes (Flash DE 98%). The one flash that was
CC is defined as a CC with a duration of more than 40 not detected was a single stroke flash. This single stroke
ms. In our data set we found 10 strokes in eight flashes flash was actually located by the LLS but was finally
(out of 45) with long CC. The mean CC duration of these rejected by the applied quality criteria of the location
10 CCs was 197 ms. The longest CC we found in our algorithm. The peak current of this stroke was about -3
dataset lasted for 445 ms. kA according to the LLS data.
The stroke peak current is determined from the LLS The LLS detected 114 strokes out of 135 strokes
data because for several close flashes the FM-system was completely correct (Stroke DE 84%). It is important to
saturated and therefore it is impossible to determine the note that all real negative CG strokes were categorized by
peak field. Median peak current of the 114 correctly the LLS correctly as CG lightning. We have never seen
detected strokes is -13 kA. The median peak current of all that a CG stroke was erroneously categorized as cloud-to-
first strokes is -14.5 kA and the median peak current of cloud discharge. On the other hand we found in the LLS
all first strokes in multiple-stroke flashes is -16.5 kA. The data some initial breakdowns or cloud processes
median peak current of all subsequent strokes, erroneously categorized as CG strokes (8 strokes not
independent whether it is from a stroke in the same included in the 135 strokes mentioned above and
channel or a stroke in a new channel to ground, is -12.1 categorized as CG were cloud discharges).
kA. It is interesting to note that the ratio of peak currents In order to be able to asses the location accuracy of the
from first and subsequent strokes is much smaller than 2 LLS we searched in the video data for strokes using the
(for a review of publications related to this ratio see [2]) same channel as the previous stroke and therefore striking
although the data set is too small for a more to the same ground strike point. Similar to the procedure
comprehensive statistical analysis. Percentage of flashes described in [6] the differences between the stroke
with subsequent stroke peak current greater than the first positions within a flash have been computed from the
stroke peak current in this study was 37% (35 flashes position distances in the LLS data by scaling them by (1
analyzed).
/ 2 ). This scaling is necessary because both positions

1011-2
used are subject to random errors [6]. As stated in [6] the two storms.
there is a possibility that the channel geometry and/or the Evaluation of the LLS performance revealed a high
actual ground contact varied slightly from stroke to stroke flash and stroke DE of the Austrian LLS ALDIS. Flash
and were not resolved by the video camera. Therefore the DE of 98% (for peak currents greater 2 kA) [13] at the
differences determined by this method should be regarded Gaisberg tower for flashes containing at least one
as upper bounds of the actual position differences. subsequent stroke (or β pulse) is probably an
For 20 subsequent strokes we determined a median overestimation because the upward initiated flashes from
value of 440 m and a standard deviation of 1000 m. The the tower exhibit an overall higher multiplicity compared
median value is slightly larger as determined from strokes to natural downward flashes. Nevertheless the Flash DE
to the Gaisberg tower (median 368m, STD 768m [13]). determined in this paper is the same and interestingly the
This indicates that the argument mentioned above about only missed flash in this analysis is a single stroke flash.
the limitation of this type of analyses seems to be correct. Stroke DE of 84% determined from video and FM data
The significantly greater standard deviation in these (smallest peak current in this study determined from LLS
measurements is probably related to the small sample data is -3.5 kA) and a DE for subsequent strokes at the
size. Further the large standard deviations prevent to see Gaisberg Tower (I>2kA) of 85% [13] are in the same
any significant differences of the distributions from the range.
Gaisberg Tower and the video measurements (according The location accuracy determined in this study is
to the Mann-Whitney-U-test). somewhat greater as determined for strokes to the
Gaisberg tower. As stated in [6] we also think that the
5 DISCUSSION/SUMMARY location differences determined by the method used in
Some lightning parameters determined with the this paper are somehow upper bounds because it is not
Austrian LLS and presented in the past are significantly clear that we see always the true ground contact of the
different from what is published in the literature for other channel.
regions (e.g. Florida). Therefore a lot of efforts have been
6 REFERENCES
made in the past to clarify whether these differences are
real or a result of the used LLS technology and its [1] Schulz, W., K. Cummins, G. Diendorfer, and M. Dorninger
performance. (2005), Cloud-to-ground lightning in Austria: A 10-year
In this paper we have shown that for flashes in two study using data from a lightning location system, J.
thunderstorms we can determine parameters which are Geophys. Res., 110, D09101, doi:10.1029/2004JD005332.
[2] Nag A., V. A. Rakov, W. Schulz, M. M. F. Saba, R.
more similar to what is known from the literature (e.g. the Thottappillil, C. J. Biagi, A. O. Filho, A. Kafri, N.
percentage of single stroke flashes or the percentage of Theethayi, T. Gotschl (2008), First versus subsequent
flashes with subsequent stroke peak current greater than return-stroke current and field peaks in negative cloud-to-
the first stroke peak current) but nevertheless the ratio of ground lightning discharges. Journal of Geophysical
first stroke to subsequent stroke peak currents seems to be Research, Vol. 113, D19112, doi:10.1029/2007JD009729.
still far from what is reported in the literature (although [3] Schulz W., Lackenbauer B., Diendorfer G., Pichler H.
the sample size is too small to give a final answer to this (2005), LLS data and correlated continuous field
topic). It is interesting to note that for the data from the measurements. SIPDA, Sao Paulo, Brazil.
video and FM system used in this paper and which are [4] Schulz W. and Diendorfer G. (2006), Flash Multiplicity
and Interstroke Intervals in Austria. ICLP, Kanazawa,
more similar to the values in the literature also the ALDIS Japan.
LLS reported values in the same range for the particular [5] Schulz W., Sindelar S., Kafri A., Götschl T., Theethayi N.,
storms. A possible reason that the results still differ is that Thottappillil R. (2008), The ratio between first and
the data used in this study are from two thunderstorms subsequent lightning return stroke electric field peaks in
only (two convective storms) and therefore the results are Sweden, ICLP, Uppsala, Sweden.
probably not representative for the overall lightning [6] Biagi, C. J., K. L. Cummins, K. E. Kehoe, and E. P. Krider
activity in Austria (2007), National Lightning Detection Network (NLDN)
The mean number of ground strike points determined performance in southern Arizona, Texas, and Oklahoma in
in this study is 1.93 and high compared to the values for 2003–2004, J. Geophys. Res., 112, D05208,
doi:10.1029/2006JD007341.
the number of ground strike points available in the [7] Brook, M., N. Kitagawa, and E. J. Workman (1962),
literature [9-12] ranging from 1.45 to 1.7. Quantitative study of strokes and continuing currents in
Statistics about long CC from other studies are given lightning discharges to ground, J. Geophys. Res., 67, 649–
by M. Saba (arithmetic mean CC duration is 151 ms from 659.
40 events in 27 storms [personnel communication]) or [8] Kitagawa, N., M. Brook, and E. J. Workman (1962),
Kitagawa et al. [8] (arithmetic mean 181 ms – 94 flashes Continuing current in cloud-to-ground lightning
in 3 storms). In this study we determined an arithmetic discharges, J. Geophys. Res., 67, 637– 647.
mean of 197 ms for 10 long CC events captured during [9] Hermant, A. (2000), Traqueur d’Orages, Nathan/HER,

1011-3
[10] Rakov, V.A., Uman, M.A., and Thottappillil, R. (1994),
Review of lightning properties from electric field and TV
observations. J. Geophys. Res. 99: 10,745-50.
[11] Saba M. M. F., M. G. Ballarotti, and O. Pinto Jr. (2006),
Negative cloud-to-ground lightning properties from high-
speed video observations, J. Geophys. Res., 111, D03101,
doi:10.1029/2005JD006415.
[12] Valine W. and Krider E.P. (2002), Statistics and
characteristics of cloud-to-ground lightning with multiple
ground contacts. J. Geophys. Res., 107(D20), 4441,
doi:10.1029/2001JD001360.
[13] CIGRE WG C4.404A (2009), Cloud-to-Ground Lightning
Parameters Derived from Lightning Location Systems -
The Effects of System Performance. Brochure No. 376,
CIGRE.

1011-4

You might also like