Mesospheric sprite current triangulation

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106, NO. D17, PAGES 20,189-20,194, SEPTEMBER 16, 2001 Mesospheric sprite current triangulation Martin Fiillekrug, 1 Dana R. Moudry, 2 Graham Dawes, 3 and Davis D. Sen[man 2 Abstract. A network of three time-synchronized high-precision induction coil magnetometers is installed in North America to measure sprite-associated lightning flash waveforms in the frequency range Hz during the Energetics of Upper Atmosphere Excitation by Lightning, 1998, sprite campaign in July Simultaneous intensified video observations on board an aircraft are used to investigate 16 sprites with long time delays >33.33 ms relative to the parent lightning discharge reported by the National Lightning Detection Network. Three different long-delayed sprite-associated waveforms can be distinguished: 38% do not exhibit any significant magnetic intensity variation, 25% exhibit slow variations -o100 ms, and 25[70 exhibit short pulses 4 ms. The source locations of the sprite-associated short pulses are triangulated by use of arrival time difference analysis. One source location exhibits a substantial spatial displacement 060 km relative to the parent lightning discharge, in agreement with the azimuths of sprite luminosity edges determined f?om the corresponding background star field of the video observations on board the aircraft. It is concluded f?om the temporal and spatial coincidence of the secondary short pulse and the sprite luminosity that this particular sprite is associated with current in the mesosphere. 1. introduction Transient optical emissions in the mesosphere above thunderstorms, denoted sprites [Sentman et al., 1995; Lyons, 1996], are generally preceded by positive cloudto-ground lightning discharges associated with particularly large ELF electromagnetic field pulses [Boccippio et al., 1995; Reising et al., 1996]. Curernet et al. [1998] reported secondary ELF magnetic field pulses with time delays 5 ms from the parent lightning discharge, coincident with the sprite luminosity as verified by simultaneous high-speed photometer recordings. The analysis of 17 similar cases showed that the calculated sprite brightness is well correlated with the charge moment inferred from the secondary ELF magnetic field pulse [Reising et al., 1999]. These results are attributed to mesospheric electrical current within the sprite, supported by theoretical model calculations [Pasko et al., 1998]. Simultaneous ELF magnetic field recordings and Paper number 2001JD / 01 / 2001JD high-speed video camera observations constrained the mesospheric current location to the sprite head in the altitude range km [Cummer and Stanley, 1998]. However, the cited experimental results are only based on the temporal coincidence of ELF magnetic field radiation and optical sprite luminosities. In fact, remote ELF magnetic field measurements are sensitive to vertically oriented current and cannot distinguish between tropospheric and mesospheric current. An alternative explanation of the observations may be secondary current enhancement in the parent lightning channel, for example, as a result of the redistribution of charges within the thundercloud [t ell et al., 1998]. In this picture, horizontally outward branching lightning channels in mesoscale convective systems drain charge centers at distances of some tens of kilometers and on timescales of several hundreds of milliseconds [Krehbiel et al., 2000]. This charge is subsequently lowered to ground within the parent lightning channel as continuing current after the initial return stroke. This view simultaneously explains the occasional horizontal displacement of sprites [Lyons, 1996; Wescott et al., 1998] and the long time Institut fiir Meteorologie und Geophysik, Universit/it Frankfurt am Main, Frankfurt am Main, Germany. 2Geophysical Institute, University of Alaska, Fairbanks, delays ms between the parent lightning discharge Alaska. and sprite formation [Fiillekrug and Reising, 1998]. To 3Department of Geology and Geophysics, University of distinguish between the two hypotheses of mesospheric Edinburgh, Edinburgh, Scotland, United Kingdom. electrical sprite current and secondary current enhancement in the parent lightning channel, this contribution Copyright 2001 by the American Geophysical Union. investigates the spatial coincidence of horizontally displaced sprite luminosity and ELF magnetic field pulse 20,189 source location.

2 20,190 F YLLEKRUG ET AL- MESOSPHERIC SPRITE CURRENT TRIANGULATION 30t I 1000km I"' '"'" 1 / 20 / ',,, O Figure 1. A lightning flash (solid circle) radiates an electromagnetic wave in concentric circles (dotted lines) along the great circle paths (dashed lines) on the spheroidal Earth to the measurement instruments (open circles) at Santa Cruz (STC), Socorro (SOC), and Saskatoon (SKT). The sprite photons illuminate a small portion of a video field (dot-dashed lines) on board the aircraft (plus). The sprite source location is triangulated with the spherical hyperbolas of equal arrival time differences (solid lines). 2. The Experiment During the sprite campaign 1998 the Institut ffir Geophysik, Frankfurt/Main, Germany, installed a network of three high-precision induction coil magnetometers in North America to record continously horizontal magnetic field variations with a resolution of F /V - in the frequency range Hz. The in- struments are located in Santa Cruz (STC), California (37.1øN, 122.2øW), Socorro (SOC), New Mexico (34.0øN, 107.1øW), and Saskatoon (SKT), Saskatchewan, Canada (52.2øN, 107.2øW). The network is timesynchronized with an operational accuracy of- 20 us between different stations by GPS as verified by simultaneous recordings of the measurement instruments at one location. The Geophysical Institute, University of Alaska Fairbanks, performed intensified video camera observations of sprites above thunderstorms on board an aircraft as part of the Energetics of Upper Atmosphere Excitation by Lightning, 1998 (EXL98), sprite campaign [Sentman et al., 1998]. One video frame has a duration of ms and is divided in two fields of odd and even horizontal scan lines of ms duration each. The azimuths of the sprite luminosity edges on a video field are calculated from an analysis of the background star field with an accuracy ø. The optical video observations, the ELF magnetic field recordings, and the lightning occurrences reported by the National Lightning Detection Network (NLDN) [Cummins et al., 1998] are synchronized by GPS timing with an accuracy <1 ms as verified by inspection of individual video field scan lines of a sequence of four successive (or dancing) sprites, the associated ELF magnetic field pulses, and the preceding positive lightning flashes reported by the NLDN on July 15, 1998, 0503: : UT. The experiment combines the sprite observations, the ELF magnetic field recordings, and the NLDN lightning flash reports during the EXL98 sprite campaign. The experiment is illustrated in Figure 1. For example, on July 15, 1998, 0509: UT, a lightning flash occurred at 45.21øN, 94.48øW and radiated an electromagnetic wave that propagated almost at the speed of 13 ø' [ s¾ :46: STC 07115/98 05:09: O.ffS 0.1 O J! [...,...,,...,...,,... soc,? OA o.l SKI' SKT ".1 ' 0.05 [11.1 ii (tl-s ", T Figure 2. The long-delayed sprites illuminate a small portion of a video field between the azimuths of the sprite luminosity edges (dot-dashed lines). Positive lightning discharge occurrences are reported at 0.0 s by the NLDN associated with short ELF magnetic field pulses. The delayed sprites occur within the dashed lines and are associated with ELF horizontal magnetic intensity Ba waveforms, which are classified in three different groups: (2a) no signature (-), (2b) slow variations (SV) ms, and (2c) pulses (P) - 4 ms (compare to Table 1).

3 F LLEKRUG ET AL.: MESOSPHERIC SPRITE CURRENT TRIANGULATION 20,191 Table 1. Occurrence Times and Locations of Lightning Flashes Reported by the NLDN With Long- Delayed Sprites NLDN Sprite ELF Number Date Time, LT Latitude Longitude Delay, ms Distance, km a Group b 1 July 15, : ø ø 77 [-25.2, 47.5] SV 2 July 15, : ø ø 154 [ 31.4, 98.5] P 3 July 16, : ø ø 36 [-22.1,-22.8] - 4 July 16, : ø ø 75 [ -3.8, 5.4] - 5 July 17, : ø ø 34 [ 5.4, 23.4] - 6 July 17, : ø ø 48 [-47.6, 22.7] - 7 July 17, : ø ø 109 [-47.3, 25.2] P 8 July 17, : ø ø 69 [-38.0, 0.0] SV 9 July 17, : ø ø 93 [-14.1, 33.2] 10 ms 10 July 17, : ø ø 144 [ 0.0, 20.2] 30 ms 11 July 19, : ø ø 34 [ 49.6, 82.3] - 12 July 19, : ø ø 40 [-31.6, 13.4] P 13 July 27, : ø ø 114 [ 10.3, 13.9] SV 14 July 27, : ø ø 104 [ -6.6, 4.9] - 15 July 29, : ø ø 65 [-24.6, 20.5] P 16 July 31, : ø ø 51 [ 13.4, 29.3] SV a The sprite luminosity edges are given in distances relative to the lightning flash locations. b The sprite-associated ELF magnetic waveforms are classified in three groups (compare to Figure 2): (2a) no signature (-), (2b) slow variations (SV), and (2c) short pulses (P). Two sprites are associated with 10 and 30 ms long waveforms and remain unclassified. light in concentric circles along the great circle paths on the spheroidal Earth to the measurement instruments. The lightning flash excited a mesospheric transient optical emission, and the sprite photons traveled at the speed of light to the aircraft, where they illuminated a small portion of the video field between the edges of the two measured azimuths (see Figure 2). the occurrence time of the lightning flash with a typical duration - 2 ms (see Figure 2). We attribute this signature to the band-limited measurement instrument finite impulse response of the parent lightning discharge return stroke. The subsequently occurring ELF horizontal magnetic intensity waveforms can be classified in three different groups (see Table 1 and Figure 2): 6 waveforms (38 ) do not exhibit any significant horizontal magnetic intensity variation (Figure 2a), 4 wave- 3. Long-Delayed Sprites forms (25 ) exhibit slow variations ms (Figure During July 1998, 216 optical sprite observations 2b), 4 waveforms (25%) exhibit short pulses - 4 ms were collected in the framework of the EXL98 sprite campaign. The sprite occurrence times are used to determine the times and locations of the causative lightning flashes reported by the NLDN within the observed thunderstorm. The sprites are usually preceded by (Figure 2c), and 2 waveforms are associated with 10 ms and 30 ms long variations and remain unclassified. It is evident that the long-delayed sprites of Figure 2a are not associated with any current above the detection threshold determined by the natural background positive cloud-to-ground discharges to within - 33 ms. noise and the instrumental resolution. The waveforms From the 216 sprite observations, 16 exhibit time delays from the parent lightning discharge ms, hereinafter denoted long-delayed sprites (see Table 1). This percentage (7.4%) is in agreement with the likelihood of long-delayed sprite occurrences reported in of Figure 2b and the unclassified waveforms remain ambigious since the onset and the duration of the sprite luminosity do not necessairly coincide with the onset, the maximum, or the duration of the horizontal magnetic intensity enhancement respectively. Finally, the previous work [Fiillekrug and Reising, 1998]. Sprites sprite-associated ELF waveforms of Figure 2c clearly with long time delays ms usually exhibit electromagnetic waveforms with little interference from the parent lightning flash signature. Consequently, they exhibit a secondary short pulse comparable to the short pulse of the parent lightning discharge. The observed percentage of secondary ELF pulse occurrences (25%) are particularly well suited for subsequent analysis. is in agreement with an estimated 20% likelihood of The orthogonal north-south (NS) and east-west (EW) ELF magnetic field recordings (BN$ and BBw) are secondary ELF pulse occurrences determined from 98 sprite observations on August l, 1996 [Reising et al., used to calculate the horizontal magnetic field intensity 1999]. Note that no relationship between the classi- Bl -- V/B v$ + B w for classification of the resulting fied secondary ELF magnetic waveform and the sprite waveforms. All ELF waveforms exhibit a short pulse at appearance (bead, columnar, carrot, etc.), its appar-

4 20,192 FOLLEKRUG ET AL.: MESOSPHERIC SPRITE CURRENT TRIANGULATION SKT 07/15/98 A 05:09: UT SOC STC [t]=ms Figure 3. A positive lightning flash occurs at 0.0 ms. The time delays of the horizontal magnetic intensities Bn at the measurement instruments result from the propagation of the electromagnetic wave from the lightning flash location to the instruments along the great circle paths within the Earth-ionosphere waveguide (compare to Figure 1). ent size, its intensity, or the time delay between the parent lightning discharge and sprite formation can be quantified since the video recordings are not absolutely calibrated or normalized with respect to the distance between the sprite and aircraft location. The azimuths of the sprite luminosity edges are used to calculate the minimum distance between the lightning flash location and the observed sprite (see Table 1). Alternating signs indicate that the sprite may have occured above the lightning flash, while equal signs indicate a horizontal displacement of the sprite. For example, the delayed sprite 2 in Table I exhibits a particularly large hori- zontal displacement from the parent lightning channel, and the sprite is associated with an ELF magnetic field pulse (see Figure 2c) which is well suited for further analysis. 4. Current Triangulation Figure 3 shows in more detail the ELF waveforms of the lightning flash and the long-delayed sprite 2 in Table I and Figure 2c recorded at the network of measurement instruments. The positive lightning flash occurs at 0.0 ms. The obvious time delays result from the mean wave propagation velocity of the electromagnetic wave at ""85% of the speed of light from the lightning flash location to the measurement instruments along the great circle paths (compare to Figure 1). The attenuation of the peak horizontal magnetic intensity with increasing distance results from the geometric spreading of the electromagnetic energy and from the absorption within the ionosphere and the Earth, i.e., along the conductive boundaries of the Earth-ionosphere waveguide. The network of three GPS time-synchronized measurement instruments makes possible source location trian- gulation by use of arrival time difference analysis [Lee, 1986]. Since the ELF wave propagation velocity depends on the variable ionosphere [Hughes and Gallenberger, 1974], the analysis is divided in two parts. In the first step we calculate the arrival time differences of the first ELF pulse from the lightning flash (see Figure 3) between each pair of two horizontal magnetic intensity recordings from the interpolated cross-covariance function in the time domain and determine the ionospheric transfer function from the known lightning flash location of the parent lightning discharge. In the second step of the analysis we use the determined ionospheric transfer function to triangulate the source location of the secondary ELF pulse under the assumption that the ionospheric variability is small during the short time interval of the investigated delays ms. This source location triangulation is based on the spatial coincidence of the three lines of equal arrival time differences (or equivalent distance differences) between each pair of two stations, i.e., spherical hyperbolas (see Figures 4 and 1). Each intersection of two spherical hyperpolas indicates a source location. In practice, these intersections do not coincide as a result of instrumen- tal errors. This instrumental error can be estimated since the arrival time differences are not independent of each other. Theoretically, the total sum of all cyclic permuted arrival time differences needs to be zero. In practice, we find this sum to be ""12 s, which relates to the operational time synchronization accuracy of the network measurement instruments (compare to the experiment section). Assuming a maximum instrumental synchronization error of ""20/ s, and a mean wave propagation velocity of ""85% of the speed of light, the c 45.2 II /15/98 05:09: UT SKT/STC / ;TC/SOC SOC/SKT /.. Iii IIi iiii iii iii //// I /// iii /// /// / i II / [ Ing. ] = deg.. Figure 4. Triangulation of the secondary ELF pulse source location by use of arrival time difference analysis (solid lines) results in a substantial displacement frown the parent lightning discharge (solid circle) in agreement with the sprite luminosity recorded on a. small portion of a video field (dashed lines) on board the aircraft (compare to Figure 1). The maximum instrumental errors (dot-dashed lines) are small compared to the observerd horizontal displacement.

5 ,, F TLLEKRUG ET AL.: MESOSPHERIC SPRITE CURRENT TRIANGULATION 20, STC' , I-- " 1.5[ 1 SKT, 5= o.5 "f I 0.05 o !, o.os 0 '%.;5 o.s 0.5 [t]=s 37 %,.,., %%,,, %-' 07129/98 ' ' % 04:25: % % 36.6[- "".. %,% - ' v... /,,'.., /STC' ' ,_. -.,',,',,,, '--'36.4]-.C/S :T' ----,.'. _" '%--.-"... STG./ OC / s?oc! '"' - 1 I/ I, C, [ \\ %...,. %%% 2[ ';;, %% \\\ "" '".. "' %' % / ',, [Ing. ] = deg. -91 Figure 5. (left) The positive lightning discharge 15 in Table I occurs at 0.0 s, and the luminosity of the long-delayed sprite is depicted in the corresponding video field. (right) Triangulation of the secondary ELF pulse (solid lines) results in a source location close to the lightning flash (solid circle) within the azimuths of the sprite luminosity edges (dashed lines) of the video field. No horizontal displacement within the instrumental accuracy (dot-dashed lines) can be distinguished (compare to Figure 4). spherical hyperbolas may be shifted by +5 kin. Consequently, the spatial triangulation error depends on the geometric configuration of the spherical hyperbola in- tersections (see Figure 4). Nevertheless, the triangulated ELF pulse source location of the long-delayed sprite 2 in Table 1 exhibits a particularly large spatial displacement - 60 kin, in agreement with the azimuths of the sprite luminosity edges determined from the star fields of the corresponding video observations (see Figures 4 and 1). Since the sprite luminosity edges are determined with an accuracy ø and the distance between the sprite and the aircraft is -615 kin, the estimated error for the optical sprite observations is +2.7 km (see Figure 4). To test the methodology, it is finally applied to the long-delayed sprite 15 in Table 1, which does not exhibit any substantial horizontal displacement from the lightning discharge, in agreement with the triangulated ELF pulse source location (see Figure 5). 5. Discussion and Summary The secondary ELF pulse source location triangulation of the long-delayed sprite 2 in Table 1 indicates a substantial horizontal displacement from the parent lightning discharge. Consequently, the secondary ELF pulse cannot result from continuing current enhancement within the parent lightning channel since the triangulated source location would need to coincide with the parent, lightning flash location reported by the NLDN. It is thus concluded from the temporal and spatial coincidence of the secondary ELF pulse and the sprite luminosity that this particular sprite is associated with current in the mesosphere. It may be argued that the secondary ELF pulse is associated with a cloud-to-ground lightning discharge that was missed by the NLDN. On the other hand, simultaneous VLF recordings at Stanford University do not exhibit any substantial VLF energy indicative of a subsequent return stroke (C. P. Barrington-Leigh, personal communication, 1999). Long-duration discharges from elevated structures (such as towers) and vertical intracloud discharges may occur and result in ELF pulses without preceding VLF energy (E. R. Williams, personal communication, 1999). However, we are not aware of any reports in the scientific literature that these exceptional discharges have effectively excited sprites, and we consider similar suggestions as unlikely, though not impossible. Wide-angle video camera observations exhibit two extremely weak sprites at 0509: and 0509: UT (see arrows in Figure 2c), which are not associated with any detectable current (compare to the waveforms in Figure 2a and Table 1). Successive sprites (or dancing sprites) strongly indicate that the parent return stroke initiated ongoing reorganization of the charge distribution within the thundercloud. Since secondary ELF pulses rather seem to be the exception than the rule (20% likelihood), it may be speculated that the electromagnetic pulse of the parent return stroke and the subsequent charge redistribution within the thundercloud enhance the ionization in the meso- sphere, which may provide a more favorable condition for mesospheric current to occur. However, we summarize that the temporal and spatial coincidence of the secondary ELF pulse and the sprite luminosity provides one more piece of evidence for sprite-associated electrical current in the mesosphere, in agreement with previous experimental findings [Cummer et al., 1998] and

6 20,194 F LLEKRUG ET AL.: MESOSPHERIC SPRITE CURRENT TRIANGULATION theoretical analysis [Pasko et al., 1998]. This observation also explains in a natural way the remarkable sprite detection efficiency, --80%, by use of Earth-ionosphere cavity resonances [Fiillekrug and Reising, 1998] and ELF sferic energy [Reising et al., 1999]. Acknowledgments. This research was sponsored by the Deutsche Forschungs Gemeinschaft under grants Fu 285/3-1 and Fu 285/4-1 and hosted by the Space, Telecommunications, and Radioscience Laboratory at Stanford University, California. The measurements were logistically supported by the Natural Environment Research Council equipment pool at the University of Edinburgh, Scotland; Lockheed Martin Missiles and Space Company in Sunnyvale, California; Langmuir Laboratory for Atmospheric Research at New Mexico Tech University, Socorro, New Mexico; and the Institute of Space and Atmospheric Studies at the University of Saskatchewan, Saskatoon, Canada. The authors thank V. Valiant, S. Payne, D. J. Murphy, F. Rest, P. R. Krehbiel, D. McEwen, Y. Zhang, and E. R. Williams for support of this project. References Bell, T.F., S.C. Reising, and U.S. Inan, Intense continuing currents following positive cloud-to-ground lightning associated with red sprites, Geophys. Res. Left., 25, 1285, Boccippio, D.J., E.R. Williams, S.J. Heckman, W.A. Lyons, I.T. Baker, and R. Boldi, Sprites, ELF transients, and positive ground strokes, Science, 269, 1088, Curemet, S.A., and M. Stanley, Submillisecond resolution lightning currents and sprite development: Observations and implications, Geophys. Res. Left., 26, 3205, Curemet, S.A., U.S. Inan, T.F. Bell, and C.P. Barrington- Leigh, ELF radiation produced by electrical currents in sprites, Geophys. Res. Left., 25, 1281, Cummins, K.L., M.J. Murphy, E.A. Bardo, W.L. Hiscox, R.B. Pyle, and A.E. Pifer, A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network, J. Geophys. Res., 103, 9035, Fiillekrug, M., and S.C. Reising, Excitation of Earthionosphere cavity resonances by sprite-associated lightning flashes, Geophys. Res. Left., 25, 4145, Hughes, H.G., and R.J. Gallenberger, Propagation of extremely low-frequency (ELF) atmospherics over a mixed day-night path, J. Atmos. Terr. Phys., $6, 1643, Krehbiel, P.R., R.J. Thomas, W. Rison, T. Hamlin, J. Harlin, and M. Davis, GPS-based mapping system reveals lightning inside storms, Eos Trans. AGU, 81, 21, Lee, A.C., An experimental study of the remote location of lightning flashes using a VLF arrival time difference technique,. J. R. Meteorol. $oc., 112, 203, Lyons, W.A., Sprite observations above the U.S. High Plains in relation to their parent thunderstrom systems, J. Geophys. Res., 101, 29,641, Pasko, V.P., U.S. Inan, T.F. Bell, and S.C. Reising, Mechanism of ELF radiation from sprites, Geophys. Res. Left., 25, 3493, Reising, S.C., U.S. Inan, T.F. Bell, and W.A. Lyons, Evi- dence for continuing currents in sprite-producing lightning flashes, Geophys. Res. Left., 23, 3639, Reising, S.C., U.S. Inan, and T.F. Bell, ELF sferic energy as a proxy indicator for sprite occurrence, Geophys. Res. Left., 26, 987, Sentman, D.D., E.M. Wescott, D.L. Osborne, D.L. Hampton, and M.J. Heavner, Preliminary results from the Sprites94 aircraft campaign, 1, Red sprites, Geophys. Res. Left., 22, 1205, Sentman, D.D., et al., The EXL98 sprites campaign, Eos Trans. AGU, 79, Fall Meet. Suppl., F164, A41C-05, Wescott, E.M., D.D. Sentman, M.J. Heavner, D.L. Hamp- ton, W.A. Lyons, and T. Nelson, Observations of 'columniform' sprites, J. Atmos. Solar Terr. Phys., 60, 733, M. Fiillekrug, Institut ffir Meteorologie und Geophysik, Universit it Frankfurt am Main, Feldbergstrafie 47, D Frankfurt am Main, Germany. G. Dawes, Department of Geology and Geophysics, Edinburgh University, Grant Intstitute, West Mains Road, Edinburgh EH9 3JW, Scotland, United Kingdom. D. R. Moudry and D. D. Sentman, Geophysical Institute, University of Alaska Fairbanks, 903 Koyukuk Drive, Fairbanks, AK , USA. (Received April 29, 2000; revised November 2, 2000; accepted November 29, 2000.)