248 249
4. Jets with significant red emissions 250
251
Figure 4(a)-(c) exhibits a set of cropped images recorded by the three cameras for a 252
sprite occurring about 400 km away. Images taken without filter and with red filter both 253
show clear sprite emissions, whereas the sprite emissions in the blue filter image are 254
weak but recognizable after enhancement. Figure 4(d)-(f) contains images from a blue jet 255
with a similar event distance as the sprite shown in Figure 4(a)-(c). In the red camera 256
image the emissions from the base of a blue jet are clearly visible, but the brightness is 257
much lower than that in the sprite. However due to the atmospheric scattering, the blue 258
emissions from this blue jet are not discernible. We use the MODTRAN (MODerate 259
resolution atmospheric TRANsmission) atmospheric code [Ontar Corporation, 2002] to 260
estimate the molecular scattering from the jets to observatory. Figure 5 shows the 261
effective response curves of our filter-equipped cameras at the Lulin Observatory to a 15 262
13 km altitude emission source that locates 400 km away. Note that these effective response 263
curves have taken the atmospheric transmittance, the camera passing band, and the CCD 264
gain setting all into account. For convenience, the major emission lines of nitrogen in the 265
wavelength range of 350-1000 nm are also overlapped.
266
In this ground observation, the red camera has a passing-band of 550-1000 nm, 267
which is broader than the red-band component (560-710 nm) of the color TV camera 268
deployed in the Sprite94 campaign [Wescott et al., 1995; 1998]. The quenching height of 269
the 1PN2 band is ~53 km [Vallance-Jones, 1974, p.119]. Hence the red emissions in blue 270
starters and blue jets should be very dim at the low, 15-20 km altitudes. However, due to 271
the broader passing-band in the red camera, more 1PN2 emissions can pass through and 272
result in the discernible red-band images. Also as exhibit in Figure 5, the blue emissions 273
should be intrinsically dim due to a very low and narrow effective passing band of the 274
blue-filter-equipped camera; even the blue-band camera’s gain was set at five times that 275
of the red-band camera. Therefore, it is no surprising that the blue emissions from jets are 276
buried under noise and cannot be discerned.
277
The main process for streamer discharges in the Earth atmosphere is the collision 278
between electrons and the molecular nitrogen. The main resulting emissions from 279
nitrogen are 1PN2 (478-2531 nm), 2PN2 (268-546 nm), LBH N2 (100-240 nm), and 280
1NN2+ (286-587 nm) [Kuo, 2007, Table 3-2]. In Figure 5, the 1PN2, 2PN2, and 1NN2+ 281
band emissions and their relative intensity are calculated using the method reported in 282
Kuo et al. [2008]. It is known that the electric field enhancement around the streamer
283head can reach ~ 5 Ek [Pasko and George, 2002, and references therein; Ek is the 284
conventional breakdown threshold in air and is ~32 kV/cm at ground]. Under the 285
14 assumption of a steady state emission [Kuo et al., 2005, Equation 2, and references
286
therein], the intensity ratio for emissions recorded by different filter-equipped CCD 287
cameras can be computed. By considering the emissions from nitrogen under 5 Ek driving 288
electric field and after taking the quenching effect at ~20 km altitude (the blue jet/starter 289
altitude), the atmospheric transmittance, the camera passing band, and the CCD gain 290
setting into account, the intensity ratio between our red-band and blue-band cameras is 291
estimated to be 2.3 and the intensity ratio between the unfiltered camera and the red-band 292
camera comes out to be 6.1.
293
We analyze some of the recorded blue starters/jet events, and find that the brightness 294
ratio for the jets recorded by the unfiltered finder camera and the red-filter camera is ~5, 295
which is close to the theoretical ratio of ~6. The red emission of blue starters/jets would 296
be 1PN2 band emissions. It is also consistent with the ISUAL recorded images of the blue 297
starters/jets through a 1PN2 band filter [Chou et al., 2010, Figure 3]. The detail analyses 298
will be presented in another paper.
299 300
5. Lightning activity in the jet-producing storm 301
302
Optical images recorded during the observation period indicate that the clouds in the 303
FOV are constantly illuminated by lightning discharges. Hence lightning data from 304
WWLLN (World Wide Lightning Location Network; [Rodger et al., 2006]) was initially 305
used to locate lightning in the observed region. However, it may be that most of the 306
lightning flashes in this area are intra-cloud or CG lightning with relative weak peak 307
currents, only one WWLLN lightning was found within the 10 second windows of these 308
15 jets. Therefore, we revolve to use the cloud illuminations of the lightning in the recorded 309
images as the indicators of the lightning activity.
310
The elevation angle of the cloud top where the jets emerge is 0.10°±0.07°, and the 311
elevation angle of the base of the most intense lightning is -1.34°±0.07°. Since the event 312
distance is 390 km and the Lulin site has a near side-view of the jet-producing 313
thunderstorm, this angular spacing corresponds to a vertical separation of ~10km, which 314
is roughly the vertical extension of the cloud. Moreover, viewing from a single site, the 315
line-of-sight distance of the lightning event cannot be resolved. Only its lateral distance 316
on the image can be determined with sufficient accuracy. From Figure 2(b), the jet-317
producing convective cells are at the edge of the thunderstorm facing the observatory, 318
and its size along the line-of-sight is less than 20 km. Therefore, we only consider the 319
azimuthal distance between the jets and the lightning as the major factor contributing to 320
the interaction of lightning and jets.
321
To avoid bright lightning from saturating the low-light level CCD and hindering the 322
determination of the lightning azimuth, the images from the gain-reduced red camera are 323
chosen for this analysis. To locate lightning on the image frames, we use a tree-search 324
method to find the pixels whose brightness exceeds ten standard deviations of the mean 325
image brightness in a frame. If more than 10 standout pixels are connected to each other, 326
it is defined as a “bright cluster”. Since the vertical span of a bright cluster provides no 327
useful distance information, we sum the image intensity count for all the pixels in a 328
vertical column of a bright cluster. The column possessing the highest intensity counts is 329
taken to be the azimuthal center of the bright cluster. The horizontal extend of the bright 330
cluster is defined as the azimuthal size. The horizontal FWHM (full width at half 331
16 maximum) width of a bright cluster is used to denote the FWHM azimuthal size of a 332
bright cluster on the image. When the azimuth center of a bright cluster is within the 333
FWHM azimuth size of the previous cluster, they belong to the same lightning activity 334
sequence.
335
A bright cluster, which brightness and/or area are greater than those on the previous 336
and the following image frames in the same lightning activity sequence, counts as an 337
“optical stroke” (for an example, see the field 6 in Figure 3). For a lightning stroke, the 338
intensity level of the visible flash goes through the intensifying, peaking, and diming 339
cycle. Since the average interval between strokes of CG lightning is 60 ms or greater 340
[Rakov and Uman, 2003, p.7, Table 1.1; p.222] and while the exposure time of an image 341
field for the NTSC cameras is 16.68 ms (the uncertainty should be less than 1 ms), optical 342
strokes can provide good facsimile representation of lightning strokes. Hence by defining 343
an optical stroke this way, it is a faithful surrogate of a lightning stroke. Therefore by 344
analyzing the activity of the optical strokes before and after a jet, the interplay between 345
the lightning and the jets can be studied.
346
We select the lightning whose lateral displacement in relative to the jets is less than 347
50 km, and count the number of optical strokes occurring within 10 s window of the jets, 348
following the convention adapted in Wescott et al. [1996, 1998]. The cumulative 349
distributions of optical strokes within 50 km lateral distance and within 10 second 350
window of all the observed jets are analyzed. The representation patterns are shown in 351
Figures 6(a1-d1). The vertical red lines denote the relative occurring time of the jets.
352
Figures 6(a2-d2) are the accompanying graphs for the sequence of lightning-jet 353
events depicted in Figure 6(a1-d1), and their aim is to provide additional physical 354
17 information includes the temporal evolution (abscissa), the spatial variation (ordinate), 355
the lateral span (grey line), FWHM of the lateral span (red line), and the brightness (color 356
of the dot; the color palette locates under panel d2) of the optical strokes occurring within 357
5-second window of a targeted jet event.
358 359
6. Correlation patterns between jets and lightning 360
361
From the optical stroke (lightning)-jet correlation patterns, important information on 362
the interplay of the lightning and jets can be inferred. From analyzing the observed events, 363
four representative patterns for the optical strokes occurring around the jets are found.
364
The salient features of the optical stroke patterns are as the followings:
365 366
A. Lightning activity increases before the jets and pauses after for a short period (13 367
events) 368
As indicated in Figure 6(a1), two blue starters occurred almost consecutively 369
near t = 0 ms. The cumulative optical stroke pattern indicates that the occurrence of 370
optical strokes intensifies within 0.5s prior the blue starters, and then falls silent for 371
0.5s after. This lightning-jet correlation pattern is similar to the lightning flash pattern 372
reported in Wescott et al. [1996; 1998]. In Figure 6(a2), the vertical blue dashed lines 373
denote the relative occurrence time of two blue starters. The crosses on the vertical 374