Is ball lightning “Anderson Localization’’?: Localized and enhanced fields in a corridor with irregular-shaped metal walls Kazuo Tanaka* and Masahiro Tanaka Depariment of Electrons and Computer Engineering of Gift University, Yanagida I-, Japan 501-1] (Received 7 July 1997; accepted for publication 24 October 1997) ‘We propose that “Anderson Localizat a possible explanation for ball lightning on the basis of computer simulation of electromagnetic wave propagation in a corridor rectal walls. It was found that the electromagnetiz wave could be sirongly localized and its strength ‘could be strongly enhanced in strall resions in the coridor under specific conditions. Similar effects may create electic fields strong enough to cause electiical treakdown conditions in the open ‘imosphere, and this phenomenon may be observed a ball lightning. © 1997 American Instine of Physics. [S0003-6951(97}02352-8] Ball lightning is considered a real phenomenon which occurs in the natural atmosphere. Although a number of the- retical works concerning ball lightning have been pub- lished, @ number of mysteries still remain, One interesting theory for ball lightning, proposed first by Kapitsa,! is that ball lightning is a fireball resulting from electric discharges created by the interference of strong electromagnetic waves from thunder clouds. Kapitsa’s theory states that the energy of ball lightning is supplied by electromagnetic waves, end this proposal has beon developed both theoretically?> end experimeneally.”* Onisuki and Onurvton,” in one of their 2x- periments with microwaves, produced a plasma fireball in a metal cavity. They showed that this plasma fireball could travel through the cieleciric slab withoat destruction and could move against the wind, which are just the mysterious characteristics of ball lightning that have beon reported by eyewitnesses, In this clestromagnetic-inierference hypeth- cals, it Is necessary to employ some kind of focusing or localization of the electromagnetic waves in order to make the electrie field strong enough to establish the breakdown condition in a localized region of the open atmosphere. Anderson indicated that waves may be localized into a finite spatial region in disordered or random media because of constructive interference occurring within the randomly scattered waves, and this effect is often called “Anderson Localization””? The localization of electromagnetic waves in randoi media has been extensively investigated."™""" Frich etal. analyzed the reffection of electromagnetic waves by one-dimensional random media,'S They showed theoretically that the wave can be localized and its amplitude can be strongly enhanced due to “stochastic resanance” in random edie, Since there are many scattering objects located at random positions in the complicated topography, the natural environment can be regarded as a disordered or random me- dia for elestromagnetic waves. This disordered condition is necessary for localization and enfrancing the waves. In this letter, we propose that Anderson Localization of electromagnetic waves in natural environments may be @ cause of bal lightning. It is possible to ecasider the random fluctuations of the index of refraction in the air as a random media for electromagnetic waves. However, it was found that Electronic ms tnaka@ikin6 gifea.scjp ‘Appl. Phys. Ltt. 74 (26), 29 Cecember 1997 (0008-695 119771 (26)'379818:8 10.00 the normal fluctuation of the index of refraction in the air due to changes in temperature, atmospheric pressure, and bumid- ity are too small to cause significant localization ox.enkance- ‘ment of electromagnetic waves. In order to investigate this hypothesis. concerning the cause of ball lightning, we considered, the propagation of electromagnetic waves in a corridor which has. irregular- shaped metal walls. Simulation results showed that the elee- tric field inside the comridor with the irregular walls could become far stronger tham that of the incident electromagnetic ‘waves in localized regions. This effect ean produce electric discharges in small spatial regions, ic., ball-lightning. ‘The geometry of the problem considered in this letter is shown in Fig, 1. Two-dimensional straight waveguides, whose widths are given by koa, were aligned to a corridor ith iregular-shaped metal walls whose length is given by ‘ob as shown in Fig. 1, where ko is a wave number resulting from 2e/wavelength. It wes assumed that both stiaight ‘waveguides satisfied the single-mode condition. The space in the waveguides and the comidor were filled with air. A ‘TE dominart guided mode, which has only the ; component of the electric field E(-,y), was assumed to be incident from the left straight waveguide. It was possible to solve the prob- Jem of reflection and transmission of the guided mode by @ ‘corridor numerically, as shown in Fig. |, with high accuracy by use of the boundary-clement method based on guided- miode extracted integral equations (GMBIES)."© 11 was con- FIG. 1. Geomety of te problem, © 1997 American instinte of Physics 9798w w “lly IG. 2, Transmission characteris of the corridor as shown in Fi. firmed that all the numerical results in this letter satisfy the energy conservation law within an aocuracy of 1%. ‘The shapes of the irregular walls in the corridor were determined by the summation of Gaussian shot pulses that obey 2 stochastic Poisson process.” The iregular-shaped ‘walls were placed such thet their average wall positions co- incided with the position of the straight waveguides as shown in Fig. 1. The average values of the upper and lower ‘all position of this corridor are zero measuring each wall position from ech straight waveguide. The auto-cortelation function of the wall positions can be writen, using Camp- bell’s theory," as Glk7)= 07 -expl—(kor)*Mkol)"s a where o? = (0) is a variance of the wall position and ko! is the transverse correlation length. Tt should be noted that all lengths in this letter were normalized by the wave number io. Parameters used in the numerical calealations are given by kya 1.99% (0.995Xwavelength), kb=230 (3631 Xwavelength), kyl =4 (0.04xwavelengit). Given the concition that the construction of the ieregular shaped walls was fixed to a given sample stochastic Poisson process. the tansmission characteristics were caleu- Inted, The dependence of the power transmission coefficient ‘on the normalized standard deviation of the wall position defined by p= o/(kya) is shown in Fig. 2. It was found that, when values of p were small (ic., p<0.015), the electromag retie waves could easily trensmit through the corridos. When values of p were large (ie., p>0.05), the electromagnetic waves could not easily transmit through the corridor die to interference by the waves scattered from the irregular walls ‘These results are reasonable from a physical viewpoint. Tt was also found that there are several resonance points as showa in Fig. 2, which give small or large transmissions. ‘The distributions of the electric fielés in the corridor were tlso calculated, The distributions of the electric fletds inside the corridor are shown in Figs. 3 and 4 for the reso- rant points (a) and (b) marked by arrows in Fig. 2. The z axis in Figs. 3 and 4 represents the absolute velue of the clectic field normalized by the amplitude of the incident TE mode. ‘The practical shape of the condor for exch numerical ex- ample is also shown in the upper insot of Figs. 3 and 4. 5794 Appl. Phys. Let, Val. 71, No. 28, 29 December 1997 FIG. 3, Atsolue valve of felddistuation (x,)| tn tbe conor a point (a) in Fie. 2 (p= 00119, Twas found that the electromagnetic wave was localized and its field was eninanced in small regions in the corridor at ‘each resonance poiat, as showa in Fig. 3 for p=0.0119, and in Fig. 4 for p=QO712, The electric field for p=O0119, [point (a) in Fig. 2] was up to 100 times larger than that of the incident wave es shown in Fig. 3. The electric field for (p=0.0712 [point (b) in Fig. 2] was up to 30 times larger, as shown in Fig. 4. (Note that the z-axis units in Fig. 3 are different from those in Fg. 4.) The phenomenon at point (a) in Fig. 2 resembles the stochastic resonance in. one- dimensional random media proposed by Frich er al. because the incident wave % almost totally rofloctod! It is possible that an electtic discharge will occur in a region where the ‘enhanced electric fleld satisfies the breakdown condition, and it is possible that this discharge will be observed as ball lightning, It should be noted that the cross-section size of the discharge region, where the enhanced field strength exceeds a given value, can become smaller than that of the wave- ength of electromagnetic waves, as shown in Figs. 3 and 4 (Note thet the width ofthe straight waveguide was about one. wavelength.) ‘The derivatives of the electric Geld normal to the bound- ary on the walls (£(x,y)/dn) are propertional to the current, tributions on the walls, ané their absolu‘e values (|3E(:,y)/dn)) on the upper wall are shown for p=0.0015,, p=0.0119, and p=0.0712 in Fig. 5. The results for p. straight waveguide, It was found that large current dietribu- tions existed on the wall in the vicinity of the localized and 01 eee eos SS FIG. 4, Absolute valve of fel dstbution (s,)] fn the corr at pone () in Fig.2 (p 0.0719, K. Tanaka and M. Tanaka£ a8 HIG, 5. Absolute value of rormal devivative of the elecric told eats. )anl on the upper wal (90.0015, »= 0.0119, ad p=0.0712) enhanced fields of Figs. 3 and 4. If this theory is valid as an explanation of ball lightning, it would mean that large cxr- reuts exist on metal eonducting objects in the vicinity of the ball lightning. These large currents would influence conduct- Ing objects in the vicinity of the ball lightning, and svch phenomena have been reported by eyewitnesses,"* The envi- ronmental situation considered inthis leter is similar to that in a valley, in a street, in a submarine, or in the fuselage of sn aisplane. “To surnmarizo, we have progosed Anderson Localization 4s a possible explanation for tall lightning on the basis of compiler simulation of electromagnetic wave propagation in 4 coridor with irregular-shaped metal was." It was found that the electromagnetic waves could be strongly localized and their strength could be strongly enhanced in small re= gions of the coridor under specific conditions. These effects ‘Appl. Phys. Lett, Vol. 71, No. 26, 29 December 1997 could create electric flelds strong enough to cause elecisical breakdown conditions in the cpen atmosphere, i. ball light- ning. In this leter, only the simplest proslem was considered duc to the limited ability of the computer systems. However, the results presented here can be used as a starting point for further invest gations toward more accurate theory of ball lightning ‘The authors gratefully acknowledge te suppor of JAPAN SKEFTICS. "PL Kapts, Dokl. Akad. Nauk SSSR 101, 245 (1955) CW. Hamilton, Nature (London) 188, 1098 (1960) SY, G. Endean, Nature (London) 263, 753 (1976. 44,1. Lowe, J. Phys. D 29,1237 (96). SPH. Handel, 1 Geophys. Res, 99, 10689 (1998). 5D. M, Handel, Science of Ball Lighting, edited by Y. Obisuki (World Scientie, Lonon, 1589), p. 245, TY, Ohtwulé and H. Ofuraton, Nature (London) 380,159 (1991). Sic Yasui, Phys. Let A 27% 451 (1993) ®p.W, Anderson, Phys. Rev. 109, 1492 (i958). OK Arya, Wave Phenomena, edited by L. Lam and H.C. Moris (Springer, [New York, 1988). p. 259. "KM. Yoo and R.R.Alfero, Phys Rev. B 39, $806 (1989) (Ogurs, M. Kitano, N. Talahash, and 7. Kewanishi, Proceedings of the 1992 URSI Inematinal Symposiim on Hleceomagietic Theory, 1992 (onpubieed), >. 81 "M4 Salle, Appl. Oe. 32,3354 (1992) NS, Takowshi, H Hiowe, §. Nit, 7. Nob, and S. Nonomurs, Sol. Energy Mater. Sol. Cells 34,439 (199. 'SU.Frich, C. Freichle, J-P. Schelecker and P, L, Sem, Phys.Rev. A 8, 16 (1973). "K Tanaka and M. Nakahara, IEEE Trans. Microvave Theory Tech. ‘MTT -it, 1647 (1992. "A, Papontis, Probably, Rondon Variables and Sochastec Processes (ieGrawFil, Tokyo, 1965), Chap. 10. "9G, Bgciy, A Titokewos Ginbulldim (The Mysterious Ball Lighning) (teak Kinyvkingn, Budiper, 1986). Transated isto Japanese by (Osaki and K Sekai (Marzen, Tokyo, 1990). °K ‘Tanaka and M, ‘Tanaka, Procedings of the Sth Intemational Sympo- sim oo Ball Lighting, 1997, p. 137 K. Tanaka and M.Tanaka 8795