Summary

Through carefully carrying out construction and operation of the NCKU

ULF/ELF/VLF receiving stations in Taiwan, we have recorded valuable sferic data that are associated with the transient luminous events in the upper atmosphere. Also through diligently performing analyses of the TLE-associated sferics, we have made major contributions toward the understanding of the characteristics of the TLE-associated discharges.

When the ISUAL onboard the FORMOSAT-2 satellite began its main mission of performing global survey of TLEs in 2004, a protogroup magnetic ULF recording station had already been installed at the Lulin observatory. The system was equipped with a signal modulator (Figure 2-5) in the signal chain to record the sferics in the 1 to 100 Hz radio band and to notch-filter the 60 Hz power grid noise. The Lulin ULF system had to be re-built a couple of times after it was damaged by nearby lightning. During the testing and calibrating the repaired antennas, we noticed that the signal modulator introduced a time delay in the recorded ULF signals and produced a substantial phase in the signals;

see Figure 2-6. Besides using representative lab-generated waveforms to calibrate the

121

Lulin ULF system, an elementary signal reconstruction method was used to check the fairness of the re-constructed signals. During the in-lab calibrations, the signal modulator was found to cause a ~9-millisecond time delay in the received waves. After the signal reconstruction, the event time delay is corrected as well as some of the amplitudes for below 100 Hz wave components are restored. From reconstructed sferics, the inferred current moment amplitude of sprites were found to increase by 69±55%; the ringing in the recorded sferics is reduced and that results in a shorter duration in the time constant (-52±15%). The sprite N21P brightness and the inferred CMC are further checked. The threshold CMC to initiate sprites is inferred to be ~900 C-km from reconstructed data and it is 25% lower than that inferred from the notch-filtered sferics. Moreover, the

correlation between the charge moment charge of the causative discharges and the brightness of sprites is found to be much tighter in the reconstructed dataset (0.92 in the reconstructed dataset versus 0.76 in the notch-filtered dataset).

Sferics radiated by the TLE-associated discharges carry the information that reveals the nature of the discharges. Therefore, the detection and the analysis of the

TLE-associated sferics are important components of the TLE researches. The Lulin ULF station measures sferics in the frequency ranges of 0.01-700 Hz; see Figure 2-9. The ULF band sferics can propagate over a very long distance with little attenuation and hence is suitable for performing global monitoring of electric discharges. However, the ELF/VLF system can detect sferics at a higher frequency band and is more sensitive to local lightning activities. In 2008, in order to increase the recording frequency range of our protogroup ULF system, the signal modulator was removed from the system. From June 2009, we run some preliminary measurements using a portable VLF system at a few potential sites in Taiwan. In December 2009, a protogroup ELF/VLF system with a

122

simple vertical electric field antenna (Duke VLF-E) was installed at the Cingcao Elementary School in the suburban of Tainan City. In August 2010, we further updated the system with more sensitive sensors, including one vertical electric and two orthogonal magnetic field antennas (Quasar ELF/VLF-E/B). The sampling rate has been set at 100 kHz for the three data channels and the GPS precise time channel. The NCKU group in Taiwan has operated these two radio recording systems (the Lulin ULF station and the Cingcao ELF/VLF station) concurrently since late 2010. The valuable ULF and

ELF/VLF data have been recorded 24 hours a day continuously and are used to support our research on TLEs and their causative lightning.

Over the past decade, more than ten thousand TLEs were recorded by the ISUAL onboard the FORMOSAT-2 satellite and fifteen hundred TLEs were observed during the ground TLE campaigns carried out at various sites in Taiwan; see Table 6-1. On the TLE sferic research front, the most important findings the detection of sferics associate with GJs and blue jets. As it has discussed in Chapter 4, more than 100 TLEs were observed to occur over Typhoon Lionrock on August 31, 2010. Among them, 14 negative GJs have clear optical images and ULF and ELF/VLF sferics were analyzed in detail. From their morphologies, six and three events respectively are the familiar “tree-like” and the

“carrot-like” GJs which were reported in [Su et al., 2003], however a new hybrid form, which is termed as “tree-carrot-like” GJs are categorized. The key features of the recorded ULF and ELF/VLF band sferics were found to be closely associated optical evolution stages, including initiating lightning, leading jet, fully-developed jet (surge current) to trailing jet (continuing current). Furthermore, more than 80% of GJs have recognizable optical emissions from the initiating lightning and all of them have the clear radio emissions in ULF and ELF/VLF bands which are similar with the lightning

123

discharges. Their peak current moments-versus-charge moment changes diagram are also verified in Figure 4-5 and have different trends for different groups of GJs. This feature suggests that the eventual forms of the negative GJs may have been determined at the initiating stage. On the other hand, other evolution stages of GJs only emit radio sferics in the lower frequency (ULF) band. The inferred parameters of the discharging sources indicate that the peak current moments of the surge current and the continuing current are two important factors that determined the optical morphology of the GJs. The peak surge current moments of the tree-like GJs are larger than 60 kA-km, while that in the other two groups are both lower than 36 kA-km. The dividing factor that separates the carrot-like from the tree-carrot-like GJs appears to be the peak continuing current moment. If the peak continuing current moment is less than 25 kA-km, only dim carrot-like GJs will be generated; whereas the peak continuing current moment in the tree-carrot-like GJs ranges from 39-71 kA-km, which are substantially larger than that for the carrot-like GJs and sometimes even larger than the tree-like GJs.

We also used the ISUAL TLEs as the guide to search for the associate sferics of TLEs in Lulin ULF to Cingcao ELF/VLF datasets. The main purpose is to discover the electromagnetic signatures for different groups of TLEs for a given event range. After the event time of ISUAL TLEs was properly validated, around ten thousand TLE-associated sferics were uncovered (see Chapter 5). The standard deviations of the event time differences between the ISUAL TLEs and the associate sferics recorded by our Lulin ULF, Duke VLF-E, and Quasar ELF/VLF-E/B are only 1.36, 1.06, 1.14, and 1.27 milliseconds, thus the event association has a very high confidence level. Further

analyses of the TLE-associated sferics produced very fruitful results. An important result from analyzing the ELF/VLF data was to find 38% of the blue jets that occurred within 6

124

Mm have associate sferics; while no corresponding ULF sferics was uncovered. The relative higher frequency nature of the blue jet-associated sferics suggests that blue jets may be closely related to the NBEs. Another kind of sferics which is not radiated by the lightning discharges is the sprite current. In sharp contrast to the blue jets, the sprite currents only radiates in the low frequency ULF band [Cummer et al. 1998; Cummer et al.

2003] and has negligible VLF emissions. Optical emissions from the sprite currents were also found in the ISUAL SP and AP data. After carefully analyzed the sprite current signals in ISUAL SP, AP, and ULF datasets, it was found that the percentage of sprites featuring a sprite current is higher (19%) than 10% reported in [Cummer et al. 1998].

Since most of the sprites detected by ISUAL are very bright and have relative strong sferic signals; this seems to imply that the bright sprites are likely to have sprite current.

On the other hand, the detailed analyses for the pure sprite, sprite-with-elve events, sprite-with-halo events, sprite-with-halo-and-elve events, halo, halo-with-elve events, and elves are discussed together. The ranges of 90% sferic detectability for the

abovementioned TLE groups respectively are 6, 6, 5, 4, 10, 4, and 5 Mm. In the ULF dataset, most of the TLE groups have a similar 90% sferic detection range (4-6 Mm), expect for halos (10 Mm) and this may imply that the halo-associated sferics may have larger amplitudes and can propagate farther. This conjecture has been confirmed through analyzing the spectra of the average sferic waveforms for the TLE groups. The low frequency sferic components for TLEs occurring with halo tend to have larger amplitudes and thus can propagate farther than other TLE groups. However, interestingly, the

normalized frequency spectra of all the analyzed TLE groups possess similar magnitude and frequency distribution, except pure sprites and sprite-with-sprite-current events appear to more energy distribute below 200 Hz. The average sferic waveforms and the rise and decay time derived from these waveforms confirm this signature again. In short,

125

the sferics associate with “TLEs occurring with halo” have larger magnitudes and thus can be observed farther away; also the sferic waves associate with sprites have larger rise and decay time which mean lower frequency components carry more energy than the higher frequency counterparts. Beyond these, other signatures show no notable difference among the TLE groups. Finally, the polarity distribution in each group of TLEs was analyzed. It is interesting that the fractional weight of the negative events increases monotonically from 0% (sprite) to 96% (elve), while the fraction of the negative events over the land area decreases from 48% to 5%. These polarity variations indicate that each group of TLEs not only has a nearly fixed polarity distribution, but also has a more prevalent occurring area, either it be land, oceanic or coastal regions. This polarity asymmetry for each group of TLEs appears to be quite stable as long as the event size is statistically meaningful. The gradual transition of the negative event ratio in each geographic area is emphasized by using different color shadings in Table 5-3.

Table 6-1. The electromagnetic signatures of basic groups of TLEs.

TLE group and its representative sferics Sferic signatures

Pure Sprite

Value Range

80% sferic detectability 6 Mm

Ratio of positive event¹ 100% All Event occurring over land 48% All Event occurring over ocean 28% All Event occurring over coast 24% All Signal duration, ms² 1.2 – 2.8 0 – 5 Mm

% of sprites with sprite current³ ~20% 0 – 6 Mm Time delay between the lightning

stoke and sprite current

4.54±1.54 0 – 6 Mm

1. The ratio of positive polarity agrees with the previous report (>99%) [Boccippio et al., 1995; Williams, 2006; Williams et al., 2007].

2. The signal duration increases gradually with distance due to the

126

signal dispersion.

3. The fraction of sprites with sprite current is a factor of two higher than the 10% report [Cummer et al., 1998; Cummer et al., 2006b].

Pure Halo Value Range

90% sferic detectability 10 Mm

Ratio of positive event¹ 80% All

Event occurring over land 18% All Event occurring over ocean 28% All Event occurring over coast 54% All Signal duration, ms 0.8 – 1.4 0 – 5 Mm

1. [Barrington-Leigh et al., 2001; Bering et al., 2002, Bering et al., 2004a; 2004b; Bhusal et al., 2004; Newsome and Inan 2010]

reported that the ratio of negative events ranges from 50% to 80%.

Pure Elve

Value Range

90% sferic detectability 10 Mm

Ratio of positive event¹ 96% All

Event occurring over land 5% All

Event occurring over ocean 56% All Event occurring over coast 39% All Signal duration, ms 0.8 – 1.8 0 – 5 Mm

1. [Barrington-Leigh et al., 2001; Bering et al., 2002, Bering et al., 2004a; 2004b; Bhusal et al., 2004; Newsome and Inan 2010]

reported that the ratio of negative events ranges from 50% to 80%.

Blue Jet

Within 6 Mm range, 38% of blue jets were found to have associate sferics; the sferic waveform suggests that they may be closely linked to NBEs.

Gigantic Jet

Sferics contain features that correspond to the initiating lightning (the left-most signal in Figure (a)) which kicks off the GJ event and it may be linked to the bolt-from-the-blue (BFB). Sferics associate with the leading jet, the surge current and the continuing current are

127

Three groups of negative GJs:

Tree-like Carrot-like Tree-Carrot-like Peak CM of

surge current >60 kA-km <36 kA-km Peak CM of

continuing current

<27 kA-km >27 kA-km

128