al., 2002; Su et al., 2003; Hsu et al., 2005; Su et al., 2005;
van der Velde et al., 2007], ISUAL provides high temporal resolution (0.05 ms for the AP), detailed spectroscopic information, and the ability to observe blue emissions due to a higher atmospheric transmittance for satellite observa-tions. Previous work has shown that the E field magnitudes and the average electron energies derived from the ratio SP2/SP3 ratio are consistent with the predictions of the sprite streamer model [Kuo et al., 2005; Liu et al., 2006].
Below we assume that most of the GJ luminous emissions
are from streamers, and analyze them using the methods discussed by Kuo et al. [2005].
[15] The relative response functions, R(l), for all of the ISUAL instruments have been calibrated in the preflight tests [Mende et al., 2005]. Here we consider the major band emissions (1PN2, 2PN2, N2LBH and 1NN2+) of molecular nitrogen [Kuo et al., 2007, and references therein]. The percentage of the total band emission into an ISUAL sensor unit is defined as the band percentage Bk(h) from the kth band. Bk(h) also is a function of the altitude h and can be expressed as
Bkð Þ ¼h X
l
Ikð ÞT l; hl ð ÞR lð Þqkð ÞDlh X
l
Ikð ÞDll ð1Þ
Figure 3. (a) The associated AP photometric data for the GJ in Figure 1. Data from the red and the blue modules are indicated by the red and the blue lines, respectively. (b) The first frame of the ISUAL Imager data for this GJ. (c) Expanded view of the AP signal traces between 1.5 and 4 ms (the dashed rectangular in Figure 3a). The background-subtracted AP data with (d) blue filter and (e) red filter for the time ranging between1.1 and 0.7 ms.
A04314 KUO ET AL.: ISUAL-RECORDED GIGANTIC JETS
5 of 10
A04314
where Ik(l) is the intensity of emission lines in the kth band emission as a function of wavelength l, T(l, h) is the atmospheric transmittance, R(l) is the instrument response, and qk(h) is the quenching ratio. The major attenuation mechanisms include O2 absorption, O3 absorption and molecular Rayleigh scattering [Kuo et al., 2007, and references therein]. The quenching factor qk(h) is defined by [Vallance-Jones, 1974]
1
1þ k q;N2NN2ð Þ þ kh q;O2NO2ð Þh
=Ak
where kq,N2and kq,O2are the collisional quenching rates for molecular nitrogen and oxygen; NN
2(h) and NO
2(h) are the number densities of molecular nitrogen and molecular oxygen as a function of altitude h; Ak is the Einstein coefficient for the kth band emission. The collisional quenching rates and Einstein coefficients for the various band emissions are given in Kuo et al. [2007, and references therein]. The number densities of molecular nitrogen and oxygen are calculated using the MSIS model [Hedin, 1991].
[16] We use the known band percentages of ISUAL SP2 and SP3 to infer the total band emissions in 2PN2and 1NN2+
[Kuo et al., 2005, 2008]. The ratio 1NN2+/2PN2reflects the relative ratio of the ionization rate for 1NN2+
to the excita-tion rate for 2PN2. Compared with the projected rate ratio of 1NN2+
to 2PN2calculated using the ELENDIF code [Kuo et al., 2007], the same ratio derived from the ISUAL observed events can be used to deduce the reduced E field and the average electron energy [Kuo et al., 2005]. The validity of calculating the emission intensity ratio 1NN2+
/2PN2is jus-tified under the steady state conditions.
[17] The altitude range of the major emissions in the streamer region of the GJ precursor is about the height of one or two AP channels (12 – 24 km) between altitudes of
40 to 80 km. The altitude uncertainty for the upward propagating luminous emissions in the fully developed stage of the GJ is about half of an AP channel, 6 km.
Taking the possible error due to the altitude of the cloud top into account, we estimate the error on altitudes to be
±10 km. Using the time- and altitude-resolving ISUAL AP data shown in Figure 3, the time in the SP readings (Figure 2) can be converted into the occurrence height.
The altitude-varying ratio of 1NN2+/2PN2is computed and shown in Figure 4a. Figures 4b and 4c present the inferred reduced E field and the average electron energy in the representative GJ, computed using the methods reported by Kuo et al. [2005]. The dashed lines in Figure 4 mark the maximum and minimum values of the derived quantities, after taking the possible error of ±10 km on altitudes into account. In the streamer region of this GJ event, the ratio (1NN2+/2PN2) is0.07. The reduced E field is thus 394 Td (1 Townsend = 1021 V-m2) and the average electron energy is8.5 eV.
[18] During the first 3-year survey (2004 – 2007), 13 GJs were identified from the ISUAL recorded events [Chen et al., 2008]. We selected 5 GJs with clear AP and SP signals, as shown in Table 3, for detailed analyses. The observed and the deduced luminous characteristics of the selected ISUAL GJs are listed in Table 3. The brightness of the FDJ varies from 0.35 – 2 MR at the altitude of the cloud tops (20 km) to the bottom of the ionosphere (90 km). The measured brightness of the TJ is weaker (0.2 – 0.8 MR) in four of the five GJs. The FDJ velocity is measured from the AP data with an altitude resolution of12 km and a time resolution of 0.05 ms. The GJ velocities in Table 3 are all on the order of 107m s1. The ratio of 1NN2+
to 2PN2emission is 0.07 – 0.27, corresponding to a reduced E field of 400 – 655 Td. This in turn yields the average electron energy in the FDJ streamer region to be 8.5 – 12.3 eV. The reduced E field and the average electron energy are significantly higher than those in the ISUAL sprites as reported by Kuo et al. [2005]. However, the results for the GJs are similar to those obtained from a streamer model [Liu and Pasko, 2004; Liu et al., 2006]. It is a good indication that the high E field exists in the streamer tip and the magnitude of E field can exceed 3Ek, where Ek is the conventional breakdown threshold field [Raizer, 1991, p. 135].
[19] The velocity of the observed FDJs is on the order of 107m s1, two orders of magnitude faster than that of the blue jets (105m s1) [Wescott et al., 1995] but comparable to the typical downward and upward sprite streamer veloc-ities (107m s1) measured by 10,000 fps imagery [McHarg et al., 2007] and a multianode photometer array [McHarg et al., 2002]. The highest upward sprite streamer velocity was 1.4 108 m s1, nearly half of the light speed. Similar Figure 4. Altitudinal variation of (a) the 1NN2+to 2PN2emission ratio, (b) the inferred reduced E field
(E/N), where E is the E field strength and N is the neutral density, and (c) the deduced average electron energy for the GJ on 1 October 2005 1122:23.898 UT. The dashed lines represent the lower and the upper bounds of these physical quantities.
A04314 KUO ET AL.: ISUAL-RECORDED GIGANTIC JETS
6 of 10
A04314
downward streamer velocities measured by 10,000 fps imagery and photometer array were also reported in Stenbaek-Nielsen and McHarg [2008]. The acceleration of the FDJ is on the order of 1010 m s2 at altitudes of 40 – 70 km, which is also consistent with observed sprite streamer accelerations; 1.8 1010 m s2 for the upward streamers and 0.5 1010 m s2 for the downward streamers. [McHarg et al., 2007]. Numerical streamer simulations show the propagating velocities to be around 106 m s1, the same as that for laboratory streamers [Briels et al., 2008]. However, in the fast expansion and acceleration phase of overvoltage streamers initiated in 1.1Ek field, the speed of the streamers can reach 2.2 107m s1as reported in numerical simulations [Liu and Pasko, 2004; Liu et al., 2006].
[20] In Table 3, three of the five GJs were found to have associated radio emissions with a positive polarity (i.e., positive charge moving downward inside the cloud) as seen in the data recorded by the Duke University extremely low/
ultralow frequency (ULF/ELF) radio system. The detected polarity ULF/ELF sferics is positive at the time of peak P1.
No strong lightning emission in SP5 (lightning 777.4 nm) infer no strong stroke, which contributes no significant signal of sferics inside the cloud. It implies that currents in the FDJ stage of the GJs generate the ULF/ELF emis-sions. This would be the negative cloud-to-ionosphere (CI) as proposed by Su et al. [2003]. In the CI scenario, the negative streamers carry negative charges and propagate upwardly from the cloud top to the lower ionosphere in the FDJ stage of GJs.
[21] It should be noted that the apparent saturation of the images, Figures 2a – 2f and Figure 3b, from the emissions associated with the lightning activity at the cloud deck level is a processing artifact. To bring out the detailed structures of the GJs, we have to set a very narrow intensity range for these images. Since even the dimmest lightning illumination substantially outshines the brightest GJs, a proper setting for the GJs will make the lightning emissions in the same image frame appear to saturate. In reality, the lightning emissions at the cloud deck level for the GJ presented in Figure 2 are weak, as it can be discerned from the associated 777.4 nm emissions in ISUAL SP channel 5.
[22] To summarize, the inferred reduced E field and the average electron energy in the fully developed jets are 560 ± 110 Td and 10.9 ± 1.7 eV, respectively. These values are
substantially higher than those in sprites, but are similar to those predicted by the streamer models [Liu et al., 2006].
Thus, these results imply the existence of streamer tips in the FDJ stage of the GJs. The speed of upward moving fully developed jets is the same order as that of downward sprite streamers. The upward propagating luminous emissions in the FDJs (P1 in Figure 3c) are thus from streamer tips that extend about1 – 2 AP channels (12 – 24 km), as shown in Figures 3d and 3e. Sferics associated with three of the five GJs listed in Table 3 exhibited positive polarity, which supports the existence of upward moving negative streamers in the FDJ stage of GJs.
5. Discussion
[23] Starting as a lightning leader but escaping from the cloud top [Krehbiel et al., 2008], the fully developed jet behaves similarly to the streamer-leader phase of a long spark [Bazelyan and Raizer, 2000, pp. 27 – 89; Bondiou and Gallimberti, 1994; Mazur et al., 2000] but with very different length and time scales. The reduced E field derived from the detected photometric signal in the FDJ confirms the existence of the streamer region in the GJs. This suggests that the escaped lightning leader above the cloud top produces streamers that extend from30 km to 90 km;
a length which is substantially longer than the several meters for the laboratory streamers [Bazelyan and Raizer, 2000, p. 86]. Also the luminous duration of the fully developed jet, represented by the P1 photometric peak in Figures 2 and 3c, can last for several milliseconds whereas it is only tens of microsecond for the laboratory leaders [Bazelyan and Raizer, 2000, p. 86]. The associated positive polarity sferics at the time of the peak P1 also suggests that streamers carry negative charges, and thus its leader should also be negative.
[24] In contrast to an arc setting in after a laboratory spark reaches the opposite electrode in the final jump phase, the FDJ streamers of the negative leader exhibit a different behavior when they reach the ionosphere. For the GJ events in the present work, a return-stroke-like process seems to occur after the FDJ has bridged the cloud top and the lower ionosphere. From the temporal and the spatial evolution of the P2 photometric peak, Figure 2 and Figure 3c, the luminosity appeared to originate at 50 km altitude and propagated toward the cloud top. As the FDJ faded away, a Table 3. Observed and Derived Physical Quantities for the Five Selected GJs
Trigger Timea
Geolocationa ULF/ELFb FDJc TJd FDJe
Lon Lat Polarity H (km) Brightness H (km) Brightness H (km) V (x1E7 m s1) 1NN2+
aThe geolocation (longitude and latitude) means the ground projection of the GJ events, which were calculated using the observational geometry shown in Figure 1.
bSource polarity of the sferics was inferred from the data recorded by the Duke University extremely low/ultralow frequency (ULF/ELF) radio system.
The plus symbol (+) indicates that the seferics has positive polarity.
cFDJ denotes the fully developed stage of the GJ, and its estimated altitudes and brightness are in units of km and mega-Rayleigh, respectively.
dSame as footnote c, and TJ stands for the trailing jet of the GJs.
eThe reduced E field and the average electron energy were calculated using the streamer optical model [Kuo et al., 2005, 2007].
A04314 KUO ET AL.: ISUAL-RECORDED GIGANTIC JETS
7 of 10
A04314
trailing jet slowly surged upward. These characteristics all seem to point to a scenario depicted in Figure 5. When the FDJ connected to the lower ionosphere, the ionized gas in the electric path serves as the extension of the local ionosphere and resulted in a lowered local ionosphere boundary. Then positive charges flew from this lowered boundary toward the cloud top or a potential wave propa-gated backward along the original leader channel and produced the ‘‘return-stroke-like’’ phenomenon. This addi-tional current or the strong electric field associated with the potential wave produces additional ionization along the original channel, and further heats the leader channel. Since the lower-altitude region has faster charge attachment rates, the boundary of the ionosphere gradually moves upward and eventually returns to the normal height. In this scenario, the trailing jet actually can be treated as a continuous current that connects the lower edge of the local ionosphere and the cloud top. As the attachment process moves the effective lower edge of the ionosphere upward and the continuous current persists, the contact point also moves up and the continuous current appears as an upsurging trailing jet. Since the image integration time is 29 ms and during this period the tip of the trailing jet clearly moved up, so the charge relaxation time at the 50 – 70 km altitude is longer than 1 ms but certainly shorter than 1 s [Pasko et al., 1998; Sentman et al., 2008].
[25] The trailing jet in Figures 2b – 2f likely is composed of hot leader channels. The photon flux from the continuous emissions in the leader channel was below the detection limit of the AP and cannot be discerned in Figure 3a. The mushroom-like cap of the TJ may be the fan-out streamer region [Raizer et al., 2006]. The leader channel is heated by the joule heating of the injected streamer currents. Because of the dominant associative ionization and the detachment processes above the critical temperature, the emission from the hot leader channels persists for at least five ISUAL Imager frames (>150 ms). The TJ feature of GJs shares some interesting features with the secondary TLEs follow-ing the primary sprites [Marshall and Inan, 2007]. They are
both blue-dominated at altitudes <60 km. The luminous period of the trailing jets can be up to 0.5 s for the event reported by Su et al. [2003] and was >0.3 s for the ISUAL GJs studied here. For the secondary TLEs, the entire luminous duration lasts less than 0.5 s [Marshall and Inan, 2007]. The proposed generating mechanism of the second-ary TLE also needs the preceding sprite to ionize the air in the altitudinal region of 50 – 90 km and lowers the iono-spheric boundary to40 – 50 km.
[26] For the processes that produce the P2 pulse, an alternative possibility is the backward streamer-leader from the negative leader FDJ [Mazur et al., 2000 and references therein]. If this was to happen, then the P2 pulse should be from the positive branch of the space leader [Bazelyan and Raizer, 2000, p. 85]. Therefore, the occurrence of the P2 pulse is expected to overlap in time with the P1 pulse associated with the FDJ. Since the P2 pulse trailed the P1 pulse by 1 ms, this implies that if the space leader did appear it must have occurred right below the original ionospheric boundary. Only below 50 km altitude, the luminous emissions from the backward propagating streamer-leader were intense enough and became visible to the ISUAL sensors. The backward streamer-leader fuses with the original negative leader, and together they form the new hot leader channel [Pasko, 2008, and references therein]. In laboratory studies, the negative long spark leader contains three to five steps between a 6 m rod-plane gap [Rakov and Uman, 2003, and references therein]. However, for the ISUAL recorded GJs studied in the article, at most there is only a single stepping process that radiated the P2 peak, which may imply it is a failed step formation [Pasko, 2008].
[27] The accompanying cloud emissions (C3 in Figure 2) were from the lightning activity inside or below the cloud.
From Figures 2a – 2f, the cloud luminous emissions for this event had complicated shapes and distinct layers. This is entirely unlike the ISUAL-observed cloud optical emissions for sprite-producing CGs [Kuo et al., 2005], which have compact, symmetrical shapes since the primary emission source, the cloud to ground channel, is deeply embedded Figure 5. During the fully developed jet stage of the GJ, the local ionosphere boundary could be located
at a much lower altitude because of the presence of the ionized channel in the FDJ. A return-stroke-like process could start at the local ionosphere boundary and extend toward the cloud top. The charge attachment process depletes the low-altitude free electrons faster, causes the ionosphere boundary to move upward, and produces the upward movement of the trailing jet.
A04314 KUO ET AL.: ISUAL-RECORDED GIGANTIC JETS
8 of 10
A04314
below the cloud and the emissions are uniformly scattered by the cloud. This suggests that the GJ-associated lightning emissions originate inside the cloud and near the cloud top.
Moreover, most of the ISUAL detected lightning events possess strong emission peaks in SP5 (777.4 nm), which are highly correlated with the currents in the lightning discharge channels. Since there are essentially no significant 777.4 nm emissions in these GJ events, this implies that there is no strong CG stroke associated with them and that the in-cloud lightning currents are modest at best. We believe that the cloud luminance that lasted for more than 150 ms was closely associated with the redistribution of charges which fed the continuous current in the vertical channel of the trailing jets.
[28] The gigantic jets reported in this article appear to behave like the counterpart of conventional cloud-to-ground lightning, in that they all have a leader (FDJ), a return-stroke-like process, and an ensuing continuous current.
However, not all the ISUAL recorded GJs have a clearly discernible return-stroke-like signal following the FDJs. It is possible that, for the other GJs, the fully developed jet did not connect to the ionosphere and thus there is no return-stroke-like process, the luminous emissions that associated with the return-stroke-like are too dim to be detected by ISUAL instruments, or they simply belong to another category of gigantic jets. For the negative-streamer GJs, Krehbiel et al. [2008] pointed out that they should start as the upward intracloud discharges from the midlevel nega-tive storm charge region. The ISUAL SP data did contain a clean peak about 17 ms before the FDJ peak (P1) and possibly some very transient flashes that followed; see the SP5 trace of the Figure 2. The luminance of this pre-FDJ peak lasted for 1 ms, and the other possible emission peaks were much shorter. Even though this pre-FDJ peak occurred out of the ISUAL AP data range (8 to 232 ms), channel 9 of the blue AP module did register some continuous luminance at the cloud-deck level before the occurrence of the FDJ; see Figures 3a and 3c. Since the initial propagation velocity of the leader could be very slow, it is possible that the luminance associated with it only started to cross into the next channel near the event trigger time and thus produced a near constant reading in the channel 9 of the blue AP. Hence, the data for the GJ presented in Figures 2 and 3 appears to be consistent with the GJ-generating hypothesis proposed by Krehbiel et al.
[2008]. However, the other four GJs studied in this article showed no AP reading at the cloud-deck level and no SP pre-FDJ luminous peak, and thus the validation of the hypothesis is inconclusive.
[2008]. However, the other four GJs studied in this article showed no AP reading at the cloud-deck level and no SP pre-FDJ luminous peak, and thus the validation of the hypothesis is inconclusive.