For over a century, numerous undocumented reports have appeared about unusual large-scale luminous phenomena above thunderclouds1,2,3,4,5,6 and, more than 80 years ago, it was suggested that an electrical discharge could bridge the gap between a thundercloud and the upper atmosphere7,8. Since then, two classes of vertically extensive optical flashes above thunderclouds have been identified—sprites9,10,11 and blue jets12,13,14. Sprites initiate near the base of the ionosphere, develop very rapidly downwards at speeds which can exceed 107 m s-1 (ref. 15), and assume many different geometrical forms16,17,18,19. In contrast, blue jets develop upwards from cloud tops at speeds of the order of 105 m s-1 and are characterized by a blue conical shape12,13,14. But no experimental data related to sprites or blue jets have been reported which conclusively indicate that they establish a direct path of electrical contact between a thundercloud and the lower ionosphere. Here we report a video recording of a blue jet propagating upwards from a thundercloud to an altitude of about 70 km, taken at the Arecibo Observatory, Puerto Rico. Above an altitude of 42 km—normally the upper limit for blue jets and the lower terminal altitude for sprites—the flash exhibited some features normally observed in sprites. As we observed this phenomenon above a relatively small thunderstorm cell, we speculate that it may be common and therefore represent an unaccounted for component of the global electric circuit.
The reported video recording was obtained during night-time observations using a Sony DCR TRV 730 charge-coupled device (CCD) video camera equipped with a blue extended ITT Night Vision GEN III NQ 6010 intensifier with a 40° circular field of view. The operation wavelength region of the intensifier was 390–870 nm at 77% sensitivity and 350–890 nm at 44% sensitivity. The intensifier provided a monochrome (predominantly green) image output. The camera was deployed at the Lidar Laboratory of Arecibo Observatory, Puerto Rico (18.347° N, 66.754° W, elevation 305 m above the sea level) (Fig. 1).
In the late evening of 14 September 2001, a cluster of thunderstorm cells developed at approximately 200 km range northwest of the observation site. Figure 1 shows an infrared image acquired from the GOES 8 satellite at 03:15 ut on 15 September 2001. The storm cloud top and the associated, almost continuous, lightning activity were clearly seen by the naked eye from Arecibo Observatory. The lower edge of the camera's field of view was aligned with the distant cloud top (altitude approximately 16 km) to avoid penetration of the direct light from bright lightning channels into the intensifier. The exact pointing direction of the video camera for the event was determined using the recorded star field (azimuth 336°, elevation 19.8°, corresponding to the centre of the camera field of view). GPS time stamps were recorded using the same video camera several hours before and several hours after the reported event, providing timing information with ±33 ms accuracy.
The event was observed starting at 03:25:0.782 ut on 15 September 2001, lasted a total of 24 video frames (∼33 ms each), and concluded with an intense lightning flash in the underlying thundercloud in frame 25. Figure 2 shows a sequence of nine images extracted from frames 6–14. The full video sequence, including odd and even video fields for each frame, is available as Supplementary Information.
Figure 1 shows the camera's pointing direction and field of view. The dashed line represents the direction toward the apparent starting point of the event in frame 1 (see Supplementary Information), which intersects the coldest (and hence highest) cloud tops at ∼200 km range. The 20-km-diameter circle drawn around the intersection point corresponds to the estimated transverse dimension (at ionospheric altitudes) of the observed phenomena. The maximum uncertainty in the estimated range of ∼200 km is ±20 km, corresponding to the width of the thunderstorm region, which translates to a ±2 km altitude uncertainty of features shown in Fig. 2, too small to affect any conclusions reported here.
The apparent speed of upward propagation of the observed phenomena remained remarkably stable during the first five frames, and is estimated to be 0.5 × 105 m s-1 (±0.07 × 105 m s-1), consistent with known speeds of the leader process in conventional lightning20. The speed increased to 1.6 × 105 m s-1 between frames 5 and 6, and to 2.7 × 105 m s-1 between frames 6 and 7. The analysis of the two video fields corresponding to frame 8 indicates that the large altitude change between frames 7 and 8 happened in two steps. During the first field the left branch, clearly visible in frame 7, extended up to altitude ∼70 km, while during the second field the right branch formed with a wider tree-like structure. The altitude change of ∼32 km for the left branch and ∼37 km for the right branch happened faster than the duration of one video field (16.7 ms). The estimated speed in the range (1.9–2.2) × 106 m s-1 is therefore a lower bound on the actual speed, which was probably higher.
The electromagnetic signatures of lightning discharges (‘sferics’) associated with the observed luminous event were recorded at two locations: Dominguito (18.421° N, 66.740° W), Puerto Rico, and at Palmer Station (64.774° S, 64.054° W), Antarctica. Dramatic rebrightening of the event was observed in frames 18 and 25 (see Supplementary Information) in association with large sferics. The sferic polarity indicated an upward transport of negative charge during the rebrightening. Also, an upward motion of subsequent breakdown was clearly evident between frames 18 and 19 (see Supplementary Information). As it is commonly known that subsequent breakdown of a given polarity in lightning generally occurs in the same direction as the initial breakdown of the same polarity, it is likely that the jet itself was created by upward negative breakdown, as was first suggested in ref. 28. A detailed report on the electromagnetic signatures will be presented elsewhere.
The apparent diameter of the breakdown filaments in the recorded images is estimated to be 1.26 km (±0.26 km), similar to the estimated diameters of stars recorded in the same images. Thus our video recordings cannot be used to reliably estimate the minimum scale of the structure, although they provide conclusive evidence for the presence of filamentary branching structure as an essential component of the observed phenomena. This finding agrees with a recently reported colour photograph showing details of streamers in blue jets14, and provides the most detailed evidence presented to date of the theoretically predicted internal streamer structure of blue jets21,22,23.
The transition apparent in frame 10 between the upper region dominated by hotspots and the lower region dominated by relatively smooth filamentary structure is estimated to be at an altitude of 42 km. This transition height is similar to the normal upper terminal altitude of blue jets and the lower terminal altitude of sprites.
The upward branches of the reported phenomena exhibit a diffuse termination at an altitude of approximately 70 km in frame 8. The 70 km termination closely coincides with the steep ledge of the lower ionospheric conductivity profile typically recorded during night-time rocket measurements in equatorial regions (for example, in Kenya24 and Peru25).
The initial stage of the observed phenomena closely resembles the general geometrical shapes and propagation speeds of previously documented blue jets12,13,14, and the event was seen visually to be blue in colour. We therefore speculate that it can be classified as a blue jet, which propagated upward beyond the previously documented altitude. Our results represent, to our knowledge, the first video observation of blue-jet phenomena from the ground, although we emphasize that the possibility of such observations was demonstrated by previous ground-based video recordings of blue starters26, which are probably related to the initial phases of blue jets23,27.
The subsequent dynamics of the upper part of the phenomenon closely resembles some of the features often observed in sprites15,16,17 (that is, the shape of branching discharge trees, the diffuse termination of the breakdown branches on the lower ionospheric boundary, the evolution of the discharge trees into hotspots, and the high propagation speed). We note also that there are some similarities in visual appearance of the observed phenomenon and the so-called palm-tree events, which follow the occurrence of large groups of sprites and exhibit the same red colour as sprites18,19. However, some other features of the observed phenomenon, such as its long duration, the altitude extent, and no apparent association with a positive cloud to ground lightning discharge, do not match typical properties of sprites.
The phenomenon we report here, which effectively demonstrates that an electrical contact between the thundercloud and the lower ionosphere was established in this event, may have been facilitated by the relatively low night-time middle atmospheric conductivity typically observed in the tropics24,25. The lower conductivity values correspond to longer dielectric relaxation timescales (that is, εo/σ, where εo is the permittivity of free space and σ is the conductivity), and therefore allow easier penetration of quasi-static thundercloud electric fields—which presumably drive the observed phenomena23—to higher altitudes. Given the fact that the phenomenon was produced by a relatively small (∼2,500 km2) thunderstorm cell, such cloud-to-ionosphere discharges may be very common in the tropics and may constitute an important, but as yet unaccounted for, component of the global electric circuit29,30.
The GEN III intensifier was provided by ITT Night Vision Industries; we thank M. Robinson for support of our work. We thank W. Lyons for discussions, and S. Gonzalez, Q. Zhou, M. Sulzer, C. Tepley, J. Friedman, E. Robles, A. Venkataraman and E. Castro for support of our observations at Arecibo Observatory. The Arecibo Observatory is a component of the National Astronomy and Ionosphere Center, which is operated by Cornell University under a cooperative agreement with the National Science Foundation. This work was supported by a Small Grant for Exploratory Research from the National Science Foundation to Pennsylvania State University. Stanford participation was also supported by the Office of Polar Programs of the National Science Foundation.
About this article
Journal of Meteorological Research (2014)