You are on page 1of 7

THRY: Evaluation of LPATS Data

Evaluation of LPATS data using VHF interferometric observations of lightning flashes during the Eulinox experiment
Claire Thry
ONERA, DMPH/EAG

Abstract. During the EULINOX experiment, lightning flashes survey was performed at the same time by a LPATS and by the ONERA VHF interferometric mapper. We use our detailed VHF observations of lightning flashes to discriminate between cloud-to-ground and intracloud flashes, and to evaluate the performances of the LPATS during five periods from two stormy days. As it was often mentionned, the LPATS performances were better for negative CG flashes than for positive ones, but the results varied a lot from one period to an other. We validated 75 % of LPATS negative strokes, and only 25 % of LPATS positive strokes. The validation ratio of LPATS negative strokes is much higher for multi-strokes CG flashes than for mono-stroke CG flashes. All high intensity positive strokes were validated, and most of rejected ones had a low intensity. On the other hand, the LPATS recorded most of the negative CGs detected by our interferometric system, but missed more than a half of positive CGs.

1. Introduction
During the Eulinox campaign (June to July 1998), the electrical activity of storms was surveyed by means of two systems : a Lightning Positionning and Tracking System (LPATS) and the ONERA VHF interferometric mapper ("ITF"). The LPATS is supposed to detect and localize the strokes of cloud-to-ground lightning flashes (CG), and the ITF detects and maps in two or three dimensions either intra-cloud (IC) or CG flashes through their VHF radiation. In a large part of Western Europ, CGs detection systems has been installed (LPATS, LPP or more recently the Impact hybrid system), and one may wish to use these data as a lightning indicator for atmospheric studies and mesoscale modelling over Europ. The LPATS uses a time-of-arrival technique and it consists in a network of stations that detect the electromagnetic radiation from return strokes, with a microsecond GPS synchronization. The position of a stroke can be deduced from the times of arrival of this electromagnetic radiation in four or more stations. The LPATS detection efficiency is supposed to be 70 % for CG flashes, and 1 % for IC flashes (Bent and Lyons, 1984). It gives the location and time (in milliseconds) of CGs, and for multi-stroke CG flashes, it gives the characteristics of all the successive return strokes. For the EULINOX experiment, the ONERA ITF system was set up around the DLR operational center in Oberpfaffenhoffen (OP on Figure 1). This system is composed of two interferometric stations which detect both IC and CG flashes up to 120 km away, and record, with a 23 s sampling rate, the amplitude and direction of arrival of the VHF radiation from the lightning flashes (azimuth and elevation). The intersection of the directions of arrival in each station gives the position of VHF sources all along the propagation of flashes in the cloud or to the ground.

EULINOX, 1998-1999 - FINAL REPORT

Figure 1. Locations of the two stations of the ONERA VHF interferometric mapper, and of two LPATS stations close to the DLR operational center (OP). In order to evaluate the performances of the LPATS around Munich, we compared simultaneous sets of data from these two systems (ITF and LPATS). We chose five periods from two stormy days (three periods on the 21st of July, and two on the 25th of June). We didn't estimate the spatial discrepancy between the two sets of data, although we checked that the positions of LPATS strokes were consistent with ITF observations during the same flash, that is within few kilometers. We must note that the two systems are not detecting the same part of flashes : the LPATS gives time and position of strokes at ground level, and the ITF maps the whole flash, including its intra-cloud development (Defer et al, 2000).

2. Work Performed: Discrimination between IN and CG flashes


The channel of a flash, CG or IN, is created by the propagation of a slow bi-polar leader discharge, then high current and high speed discharges can propagate inside this conductive channel (intra-cloud recoil streamers, or return strokes). In the case of a cloud-to-ground flash, one of the leaders goes down to the ground and it is immediatly followed by a return stroke. The polarity of a CG is given by the sign of the charge transferred from the cloud to the ground, that is also the polarity of the leader that reaches the ground before the return stroke. The negative part of the leader discharge has a typical VHF radiation, which is nearly continuous at our time resolution (23 s) and which lasts for far more than a millisecond. This radiation can be clearly observed with our system when the lightning flashes are not too far from one of our station, that is less than 35-40 km. The VHF radiation of the positive part of the leader discharge has a very low level, and can hardly be detected by our ITF system. The high current discharges, that emit light, exhibit a very short VHF radiation (less than a few hundreds of microseconds) with a much higher amplitude than the negative leader radiation. A negative CG flash starts with the bipolar propagation of the leader discharge : the positive part of the leader grows inside the cloud, while the negative one goes downward to the ground and triggers a return stroke when it reaches the ground. After a quiet period (a few tens of milliseconds) we can observe an intra-cloud activity and in some cases others downward successive leaders followed by further return strokes (subsequent strokes). In our
2

THRY: Evaluation of LPATS Data

VHF records, a negative CG can be easily recognized by its VHF signature consisting in a long lasting low amplitude radiation of the downward negative leader that ends with a short and intense peak corresponding the first return stroke, followed by a noticeable quiet period. The reliability of this VHF signature of negative CGs has been successfully tested for other campaigns (Defer, 1999; Defer et al, 2000; Solomon et al, 1999). During a positive CG, the negative part of the leader discharge spreads inside the cloud, and the return stroke occurs when the "silent" positive leader reaches the ground. This return stroke suddenly brings a certain amount of negative charges up in the cloud and changes the electrical potential of the flash channel. Consequently the intra-cloud negative leader is sharply re-activated, and exhibits a new continuous high amplitude VHF radiation just after the VHF peak due to the return stroke. This continuous radiation slowly decreases and last for a few tens of milliseconds or more. This VHF signature (a short event of high VHF amplitude immediatly followed by a continuous radiation) seems to be typical of positive CGs. We tested it succesfully on other sets of data, and we then supposed it gives a reliable criteria of positive CGs discrimination, although it has been less studied than the negative CGs signature. We applied these criteria on the VHF radiation from lightning flashes recorded by our ITF system in order to discriminate between intracloud flashes, positive and negative cloud-toground flashes. We selected flashes close enough from one of our interferometric stations (less than 35 km), we checked each of them and we thus determined their type. We then consider this set of data as a reference for the comparison with LPATS data available during the EULINOX campaign. We sort out from LPATS lists the CG strokes (excluding the IN LPATS flashes) during the same periods and inside the same limited areas around the ITF stations, and we associated strokes in CG flashes when strokes are very close in time and position (within 5 km and 1 s). We then compare the two sets of data in order to evaluate the performances of the LPATS in CG detection.

3. Results
3.1 General Remarks

We give in Table 1 some electrical characteristics observed during the five periods, and the overall mean values. The total flash rates during the periods we chose are quite high (about 20 flashes per minute), but the CG flash rates are often smaller than one CG per minute. The ratio of CG flashes among all flashes (IN and CG) varies a lot from one period to an other, especially concerning negative CG. The overall ratio of CG flashes (positive or negative) is about 10 %. As it is often observed, most of the negative CGs have one or more subsequent return strokes after the first one (multi-stroke CGs), and most of the positive CGs have only one return stroke (monostroke CGs). The mean number of strokes are 2.8 per negative CGs, and 1.2 per positive CGs. As we compared the LPATS and ITF data, we notice that LPATS strokes always correspond to high amplitude peaks in the VHF radiation, with a very small time discrepancy (less than 3 ms). Our analyse showed that these LPATS "CG" strokes correspond either to return strokes or to intra-cloud recoil streamers. This misinterpretation by the LPATS, between intra-cloud and cloud-to-ground flashes is more frequent for positive LPATS "CGs". This problem was signalled before by one of the authors of the LPATS (McGorman et Rust, 1998). We also notice a few cases of mistakes in the polarity of strokes (positive CGs detected as negative by the LPATS), or some missing strokes in LPATS records of multi-strokes negative CGs.
3

EULINOX, 1998-1999 - FINAL REPORT

Period Period duration (minutes) Total number of flashes % of CG- (CG-/all flashes) % of CG+ (CG+/all flashes) Total flash rate (per minute) CG- flash rate (per minute) CG+ flash rate (per minute)

1 2 3 4 5 15 15 15 45 30 397 327 218 994 182 9.1 % 15.0 5.0 % 0.5 % 13.7 %% 7.8 % 1.9 % 6.0 % % 4.3 % 3.7 26.5 21.8 14.5 22.1 6.1 2.4 3.3 0.7 0.1 0.8 1.1 0.8 1.1 0.4 0.4

Total 120 2,118 5.9 % 3.6 % 17.7 1.1 0.6

Table 1. General characteristics of the electrical activity during the five periods, and overall values.

3.2

Negative CG

Over the five periods we chose, the LPATS detected 441 negative strokes, which can be grouped in 198 negative CG flashes (mono or multi-strokes). Four more negative CGs (18 strokes) were detected by the ITF system only, so the LPATS seems to have a high detection efficiency of negative CGs : it recorded 97 % of negative CGs and 95 % of the negative strokes. The LPATS negative CGs set was compared to the ITF data, and 38 % of CGs (25 % of strokes) were rejected, because they appear on VHF records as intra-cloud flashes or positive CGs ("polarity error"). All these values are summarized on Figure 2.
Total LPATS CG Rejected LPATS CG (rejected for polarity error) Validated LPATS CG ITF alone CG Total ITF CG 4 12 14 122 331 76 198 CG flashes 441 strokes

110

18 126 349

Figure 2. Number of LPATS and ITF negative CGs (black bars) and strokes (grey bars). During our study, we noted that the validation ratio of strokes depends on the number of subsequent strokes in the same flash, so we analyzed closely the number of strokes of validated or rejected LPATS CGs. The Figure 3 presents the distribution of strokes according to the multiplicity of strokes in the CG they belong to, respectively for validated and rejected CGs. Most of multi-strokes CGs were validated (85 %), and more than a half of monostrokes CGs was rejected (53 %). We can also notice a decreasing trend in the number of strokes with the multiplicity of CGs.

THRY: Evaluation of LPATS Data

60 50 # strokes 40 30 20 10 0 0 1 2 3 4 5 6 7 # strokes per flash 8 9 10 11

Figure 3. Distribution of strokes according to the multiplicity of strokes in their CG. Plain line : validated CGs; dashed line : rejected CGs. More details on the relative amounts of validated strokes in each period are given in the Table 2, along with the overall values. The validation ratios of CGs and strokes varied a lot from one period to another, ranging from 20 % to 77 % of CGs (38 % to 85 % of strokes). The amount of validated strokes from multi-strokes negative CGs is more stable, at about 85 %. Period Number of LPATS neg. CGs Number of ITF alone neg. CG Validated LPATS neg. CG Number of LPATS neg. strokes Validated LPATS neg. strokes Validated mono-stroke Validated multiple strokes Table 2. 1 59 0 61 % 106 70 % 55 % 78 % 2 62 2 76 % 196 85 % 57 % 89 % 3 21 1 48 % 35 66 % 41 % 89 % 4 25 0 20 % 34 38 % 5% 86 % 5 31 1 77 % 70 79 % 69 % 81 % Total 198 4 62 % 441 75 % 47 % 85 %

Validation ratios of negative LPATS CGs and strokes.

3.3

Positive CG

The LPATS reported 132 positive strokes (grouped into 107 CG flashes) during the five selected periods, and 75 % of them (72 % of flashes) were not validated by the VHF observations, most of rejected positive LPATS CGs were in fact intra-cloud flashes. A bar diagram presents these values on Figure 4. The ITF found 34 more positive CGs (42 strokes), and 12 extra positive CGs were given by the LPATS as negative CGs. Consequently only 43 % of actual positive strokes can be found in the LPATS list of positive CGs, and 62 % were given in the LPATS list of CGs of both polarity.

EULINOX, 1998-1999 - FINAL REPORT

Total LPA TS CG + 77 99 30 33 34 42 12 14 76 89

107 f las hes 132 s trokes

Rejec ted LPA TS CG +

V alidated LPA TS CG +

ITF alone CG + LPA TS CG - (polarity error) Total ITF CG +

Figure 4. Number of LPATS and ITF positive CGs (black bars) and strokes (grey bars). The validation rate of positive LPATS strokes varies a lot from one period to the other, from 3 % to 100 %. We studied the distribution of validated and rejected LPATS strokes according to their current intensity (Figure 5). All strokes above 20 kA were validated, and that most of low intensity positive strokes (less than 10 kA) were rejected. No noticeable difference of intensity distribution appears between validated and rejected negative strokes.

# of strokes -50

LPATS intensities

50

Figure 5. Distribution of LPATS strokes according to their current intensity. Plain line : validated strokes; dashed line : rejected strokes.

4. Conclusions
More than two thousand flashes have been studied, among which we found 126 negative CGs and 76 positive CGs, during this two hours cumulated time interval. We deduced some general comments on the efficiency of the LPATS from this study, and we tried here to summarized them, in order to be able to applied them to other LPATS observations of storms. A comparative study of two LPATS networks (South Germany and Switzerland; Hohl and Schiesser, 2000) was performed, and one must be aware that and different LPATS can have very different efficiencies, depending on the number of stations, the release level, etc. From our analyse on South Germany data, we deduced that : The location of LPATS strokes seems to be correct, even if we notice in some case

THRY: Evaluation of LPATS Data

a systematic shift between ITF and LPATS mapping of the same storm. A flash per flash study showed that there is no systematic error between the sensors. This apparent shift is due to the general advection of the clouds, with more IN flashes in the front part of the storm. The ratio CG/(IN+CG) varied a lot from one period to the other (2.5 % up to 20 %), and the overall mean ratio is 9.5 %. Most of negative CGs are multi-stroke, and most of positive CGs are mono-stroke. The mean numbers of strokes per CG are 2.8 strokes (negative CGs) and 1.2 strokes (positive CGs). Nearly all negative strokes was recorded by the LPATS, and 75 % of LPATS negative strokes were valid. Half of mono-strokes LPATS negative CGs were in fact IN flashes, but 85 % of strokes from multi-strokes negative CGs were validated. The LPATS recorded 62 % of positive strokes (43 % with the right polarity), and 75 % of LPATS positive strokes were rejected in our analysis. All high intensity LPATS positive strokes were validated, and most of low intensity LPATS positive strokes were in fact IN flashes. When we cumulated the CG flashes from the two polarities, we found that 66 % of the LPATS strokes actually belong to a CG. Dividing the total number of LPATS strokes (validated and rejected) by the actual number of CG strokes, we estimated the mean LPATS overestimation : (LPATS strokes)/(actual CG strokes) = 1.3.

References
Bent, R.B., W.A. Lyons, 1984 : Theoretical evaluations and initial operational experiences of LPATS (lightining position and tracking system) to monitor lightning ground strikes using a time-of-arrival (TOA) technique. Proc.Int. Conf. Atmos. Electricty, Albany. Defer, E., 1999 : Caractrisation et modlisation de l'activit lectrique de nuages d'orage. Thse de l'Universit Paris 7-Denis Diderot. Defer, E., P. Blanchet, C. Thry, P. Laroche, J. Dye, M. Venticinque, 2000 : Simultaneous cloud-toground lightning detection from ONERA interferometric mapper and NLDN network during the STERAO-A experiment - Case of July 10th 1996 storm. Submitted to JGR. Hohl, R. and H.-H. Schiesser, 2000 : Lightning patterns and their relation to the radar-derived hail kinetic energy. Proc. European Tornadoes and Severe Storms Conf. , Toulouse, Feb. 2000. McGorman, D.R. and W.D. Rust, 1998 : The electrical nature of storms. Oxford Univ. Press, 422 pp. Solomon, R., C. Thry, E. Defer, A. Bondiou-Clergerie, 1999 : Thunderstorm and lightning development : modeling and observations. Proc. 11th Int. Conf. Atmos. Electricty, Guntersville, Alabama.

You might also like