TEC variations over Southern Europe before and during the M6.3 Abruzzo earthquake of 6 th April 2009

Transcription

1 TEC variations over Southern Europe before and during the M6.3 Abruzzo earthquake of 6 th April 2009 M.E. Contadakis, D.N. Arabelos, Ch. Pikridas and S. Spatalas Department of Geodesy and Surveying, Aristotle University of Thesaloniki, Greece Abstract In this paper the Total Electron Content (TEC) data of 16 Global Positioning System (GPS) stations of the EUREF network which are being provided by IONOLAB (Turkey) were analysed using wavelet analysis and Discrete Fourier Analysis in order to investigate the TEC variations over Southern Europe during the last month before the catastrophic Abruzzo earthquake of M=6.3 of 6 th April The main conclusion of this analysis is that abrupt TEC variations, accompanied by enhanced oscillations in a broad range of frequencies of TEC, occur randomly over a broad area of several hundred km from the earthquake, even 26 days before the earthquake. High frequency oscillations (f Hz, periods T 1h) seems to point to the location of the earthquake with relatively higher accuracy but the fractal characteristics of the frequencies distribution points to the locus of the earthquake with even higher accuracy. Key words: GPS network, ionospheric total electron content, wavelet analysis 1. Introduction It is generally accepted by the scientific community that tectonic activity resulting to earthquakes induces variations in earth ionosphere by means of the so called Lithosphere- Atmosphere- Ionosphere mechanism (Molchanov et al. 2004). This strong opinion emerged from the results of a great amount of research done by means of ground- based experiments (Molchanov et al. 2004; Molchanov et al. 2005; Liperovsky et al. 2002; Shvets et al. 2004; Roznoi et al. 2004; Roznoi et al. 2009; Biagi et al. 2009), Space-born studies( Parrot 2006; Hayakawa et al. 2000) and combined space- born and ground- based studies (Roznoi et al. 2007; Muto et al. 2008) as well. Finally The development of GPS and GLONASS satellite systems provide a perfect opportunity for a simultaneous inspection of TEC variations over a great number of locations around the earth and furthermore to investigate any interrelation of these variations or isolate variations of TEC which may occur over a particular site with enhanced tectonic activity. A lot of work has also be done in this direction (see for instance Afraimovich et al. 2001; Afraimovich et al. 2002; Liu et al. 2001; Contadakis et al. 2008). These studies indicated that over a broader area over the site where a strong earthquake occur (magnitude>5.5) uneven variations of TEC are observed. So it is of great interest by the occasion of Abruzzo earthquake to look for any significant characteristic which potentially may be used in earthquake forecast. In this paper the Total Electron Content (TEC) data of 16 Global Positioning System (GPS) stations of the EUREF network ( which are being provided by IONOLAB ( were analysed using wavelet analysis and Discrete Fourier Analysis in order to investigate the TEC variations over the Southern Europe during the last month before the catastrophic Abruzzo earthquake of 6 th April

2 2. The data 2.1 TEC valus and Dst index In this paper we are interesting in the variation of TEC over a broad area over the catastrophic Abruzzo earthquake so we use the TEC estimates provided by IONOLAB ( (Arikan et al. 2009) for the GPS stations of EUREF which cover the south of Europe for the time period between 11/03/2009 and 09/04/2009. Unfortunately the provided TEC for the stations of EUREF of central Italy was not available. So the TEC values over ROME was gently provided to us by Dr. Ciraolo (Ciraolo 2009). Table 1 displays the 16 EUREF stations. The TEC values are given in the form of a Time Series with a sampling gap of 2.5 minutes. However in time periods of uneven variations of TEC the provider change the sampling gap to 2.0 or 1.0 or even 0.5 minute in an unpredictable way, a fact which hardens the FFT elaboration of the Time Series. So special attention was given in order to analyze segments of data with the same sampling gap. This was not necessary in the case of wavelet analysis. The variations of the geomagnetic field were followed by the Dst- index quoted from the site of the Space Magnetism Faculty of Science, Kyoto University ( for the time period of our data. Figure 1 displays the Dst-index variations on March of It can be seen that in general the geomagnetic field was rather quiet on March of 2009 with some slight variations on 13 (-30nT), 21(-27nT) and 24(-24nT) of March Seismic activity on Central Italy and the broader area Since we are studying the TEC variations over the Southern Europe we take into consideration the seismic activity in a broader area of the Central Italy. Table 2 displays the seismic activity in the broader area of Central Italy, quoted from the web site Orfeus Wilbert II, Table 1. The EUREF stations of the Southern Europe No GPS Station Distance (km) Longitude(deg) Latitude(deg) Location 1 YEBES Yebes(Spain) 2 LRC LaRochelle(France) 3 TLSE Toulouse(France) 4 MARS Marseille(France) 5 AJAC Ajaccio(France) 6 GEN Genova(Italy) 7 CAGL Ajaccio(France) 8 MEDI Medicina(Italy) 9 ROM Rome(Italy) 10 NOT Noto(Italy) 11 MATE Matera(Italy) 12 OHRI Ohrid(FYROM) 13 SOFI Sofia(Bulgaria) 14 BUCU Bucuresti(Romania) 15 ISTA Istambul(Turkey) 16 ANKR Ankara(Turkey) It is seen that apart of the seismic activity in Central Italy with main event the Abruzzo earthquake of 6 th of April, which is thoroughly studied (Atzori et al. 2009,Cirella et al. 2009) there is a weak seismic activity in Ionian sea and Albania started before and developing in parallel with Central Italy activity and a seismic activity in Northwestern Balkan Peninsula started one month later. 2

3 Figure 1. Dst variations in March, 2009 Table 2. Seismic activity in a broader area of Central Italy (Events within 5 distance of selected Lat: Lon: ) Date Time Mag Lat Lon Depth Description :58: CentralItaly :20: NorthwesternBalkanPeninsula :00: Greece :03: NorthwesternBalkanPeninsula :55: CentralItaly :43: Albania :37: NorthwesternBalkanPeninsula :23: NorthwesternBalkanPeninsula :17: NorthwesternBalkanPeninsula :39: Greece :59: Greece :44: SouthernGreece :14: CentralItaly :22: CentralItaly :38: CentralItaly :53: CentralItaly :34: CentralItaly :47: CentralItaly :26: CentralItaly :15: CentralItaly :56: CentralItaly :37: CentralItaly :32: CentralItaly :10: Greece-Albania border region :16: IonianSea :13: IonianSea :57: Greece :04: Albania 3. Data analysis The data consist of TVEC values sampled every 2.5 minutes for each station and for the time interval between 11/03/2009 and 09/04/2009. We analyze the data time series using both Wavelet Analysis and Fast Fourier Transform Analysis. As it was mentioned earlier in periods of uneven TEC variations the provider change abruptly the sampling gap to 2.0, 1.0 or 0.5 minutes. These alterations can be easily compensate in wavelet analysis by considering 3

4 the local frequency alteration but in the FFT analysis special attention was paid in order to analyze segments of the same sampling gap. For the Wavelet Analysis we use the wavelet Biorthogonal 6.8. This is a compactly supported wavelet for which symmetry and exact reconstruction are possible and can be used for Discrete as well as for Continues wavelet analysis. Figure 2 displays the characteristic of the wavelet Biorthogonal 6.8 (Misiti et al. 1996). We choose this wavelet because it is symmetric and sharp cutting something that intuitively we expect to properly suit for the analysis of TEC variation. Figure 2. Characteristics of wavelet Biorthogonal 6.8. Upper part: Decomposition: scaling function phi (left) and wavelet function psi (right)& bellow them& the low- and high- pass filters respectively. Lower part: the same as for the reconstruction. 4. Results 4.1 Wavelet analysis The results of the Wavelet Analysis are shown in figures 3 to 7, in which a sample of the analyzed variations of the TEC over the GPS stations for the time period which covers the pre- co- and post- earthquake period of the main shock of l Aquila earthquake are presented i.e. from 11/03/2009 to 09/04/209. Only the moment of the main shock is marked on the figures of MATE (Figure 6) and ROME (Figure 7) by an black arrow. The scale of the diagrams are: TECU in Y- axis (1TECU=10 16 electrons m -2 ) and the sampling gap is in minutes. This sampling gap is 2.5 minutes for the diagrams of LRC,MEDI, CAGL, MATE, (occasionally changing to 2, 1 even 0.5minutes) and 1 minutes for the diagram of ROME. 4

5 Figure 3. TEC variations over LRC in TECU and the variations of the components with periods 5min, 10min, 20min,40min, 80min, 160min,320min, 640min corresponding to the marks of the right scale d 1, d 2, d 3, d 4, d 5, d 6, d 7, d 8, for the time period of 11/03-09/04/2009. In these figures the first diagram (in red) shows the variations of TEC (i.e. signal) while the following 8 (or 9, in the case of Rome) diagrams shows the variations of the components with periods from 2 minutes up to about minutes. From this sample it is obvious that the Signal and component variations for the nearest to the l Aquila EQ epicenter are larger. Figure 8 present an overall view of the TEC variation over the Network showing the range of TEC variations for the time period of 11/03/2009 to 09/04/2009 while Table 3 displays the date and the GPS station with substantial TEC variations. It is obvious that the larger variations take place over stations which are situated, either in the outer parts of an ellipse which has its major axis roughly parallel to the earthquake fault and a semi-major axis 650km& or on an arc which penetrate the Balkan peninsula in the SW-NE direction. 5

6 Figure 4. TEC variations over MEDI in TECU and the variations of the components with periods 5min& 10min& 20min&40min& 80min& 160min&320min& 640min corresponding to the marks of the right scale d 1& d 2& d 3& d 4& d 5& d 6& d 7& d 8 & for the time period of 11/03-09/04/2009 The same distribution is indicated by the enhanced value of the high frequency components (frequencies ranging between 0.003Hz (period 5min) and Hz (period 100min))& which according to Molchanov et. al ( 2004& 2006) correspond to the frequencies of the turbulent induced by the LAIC coupling process to the ionosphere& during an earthquake preparation period. This is shown (as an example) in Figures 11 and 12. In these figures the range and the standard deviations of the constituents with period of 10 (red) and 20 (blue) minutes over the network are shown. 6

7 Figure 5. TEC variations over CAGL in TECU and the variations of the components with periods 5min& 10min& 20min&40min& 80min& 160min&320min& 640min corresponding to the marks of the right scale d 1& d 2& d 3& d 4& d 5& d 6& d 7& d 8 & for the time period of 11/03-09/04/2009 Finally the same distribution is shown by the enhanced value of the low frequency components too (tidal frequencies). These observational facts were also observed in the past by the occasion of other earthquakes (Contadakis et al. 2008& Contadakis et al. 2007) and have an observational counterpart in atmospheric tides exaltings (Arabelos et. al 2004& Arabelos et al. 2008&) as well as in tidal frequency exalting of the VLF/LF radio transmission (Biagi et al. 2003). We realize also that the large variations in TEC over the specified stations is sporadic& not synchronous and are being spread over a period of a month or possibly more before& during and after the main shock& as it is happened with their counterparts. 7

8 Figure 6. TEC variations over MATE in TECU and the variations of the components with periods 5min& 10min& 20min&40min& 80min& 160min& 320min& 640min corresponding to the marks of the right scale d 1& d 2& d 3& d 4& d 5& d 6& d 7& d 8 & for the time period of 11/03-09/04/2009 Since the geomagnetic field in this periods was rather calm we reasonably may consider the possible correlation of the anomalous TEC variations with the tectonic activity of the area which is covered by our observational data. The sporadic nature also indicates that they are connected with different phenomena which are developed in the broader earthquake preparation area of one of the three earthquake activities in the area: the Central Italy& the Ionian Sea-Albania and the Northwestern Balkan area. 8

9 Figure 7. TEC variations over ROM in TECU and the variations of the components with periods 2min& 4min& 8min&16min& 32min& 64min&128min& 252min& 504min corresponding to the marks of the right scale d 1& d 2& d 3& d 4& d 5& d 6& d 7& d 8 & d 9 for the time period of 27/03-7/04/2009 Table 3. Δ(ΤECU) over the GPS stations of the network between 11/03/2009 and 09/04/2009 Table 3. Δ(ΤECU) over the GPS stations of the network between 11/03/2009 and 09/04/2009 Station MARS MEDI MATE OHRI SOFI BUCU Date 11/03/ /03/ /03/ /03/ /03/ /03/ /03/

10 20/03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /03/ /04/ /04/ /03/ /04/ /04/ /04/ /04/ The great resemblance of TEC variations over the network during the co-seismic period (04 to 06/04/2009& Figure 9) and the period between 11/03 and 13/03/2009 (Figure 10) may indicate that both are caused by the seismic activity of Central Italy. We may reasonably conclude that TEC variations in the area within the ellipse& which we mentioned earlier& are due to the seismic activity of Central Italy although the variations over MATE and OHRI may have some connection with Ionian Sea Albania seismic activity& while TEC variations over SOFI and BUCU are connected with the seismic activity of the Northwestern Balkan seismic activity. Figure 8. The range of the signal over the network between 11/03/2009 and 09/04/2009. ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) 10

11 Figure 9. The range of the signal over the network between 04/04-06/04/2009 (co-seismic period). ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) Figure 10. The range of the signal over the network between11/03-13/03/2009 ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) 11

12 Figure 11. The range of the 10 minute (red) and 20 minute (blue) components over the network between11/03-13/03/2009 ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) Figure 12. The standard deviation of the 10 minute (red) and 20 minute (blue) components over the network between11/03-13/03/2009 ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) 4.2 Fast Fourier Transform Analysis The Power Spectrum of TEC variations will provide information on the frequency content of them. Apart of the well known and well expressed tidal variations& for which the reliability of their identification can be easily inferred by statistical tests& small amplitude spacetemporal transient variations cannot have any reliable identification by means of a statistical test. Nevertheless looking at the logarithmic power spectrum we can recognize from the slop 12

13 of the diagram whether the contributed variations to the spectrum are random or periodical. If they are random the slop will be 0& which correspond to the white noise& or -2 which correspond to the Brownian walk otherwise the slop will be different (Turcotte 1997). This means that we can trace the presence of periodical variations in the logarithmic power spectrum of TEC. This method was successfully applied in a previous work (Contadakis et al. 2008). As an example Figure 3 displays the logarithmic power spectrum of TEC over Rome in the time interval between 11/03 and 19/03/2009. It is realized that the spectrum of TEC variations over Rome contain random variations in the high frequency part (f> Hz & period<54 minutes) and periodical variations in the low frequency part (f< Hz & period>54 minutes). Figure 13. Logarithmic Power Spectrum of TEC over Rome in the time interval between 11/03 and 07/04/2009 This is a typical logarithmic power spectrum of TEC& and we have seen that we can trace the presence of periodical variations. The breaking point on the diagram indicate the limited frequency below of which (or correspondently the limited period above of which) periodical variations of TEC exist. 13

14 Figure 14. Logarithmic Power Spectrum of TEC over LRC on 06/04/2009. Limited frequency Figure 14. Logarithmic Power Spectrum of TEC over Rome in 06/04/

15 Figure 15. Limited period for 11/03/2009. ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) Figure 16. Limited period for 06/04/2009 ( =l Aquila EQ& =EQs before l Aquila EQ& =EQs after l Aquila) 15

16 Table 4. Limited frequency (period) of TEC variation over the network between 11/03/2009 and 13/03/2009 Table 4. Limited frequency (period) of TEC variation over the network between 11/03/2009 and 13/03/2009 No GPS Station Frequency(Hz) Period(min) Date 1 YEBES /03 2 LRC /03 3 TLSE /03 4 MARS /03 5 AJAC /03 6 GEN /03 7 CAGL ( h) 8 MEDI /03 9 ROM inf /03 10 NOT /03 11 MATE ( h) MATE (16.56h)-12(07h) MATE ( h) MATE OHRI /03 13 SOFI /03 14 BUCU /03 15 ISTA /03 16 ANKR /03 16

17 Table 5. Limited frequency (period) of TEC variation over the network between 04/04/2009 and 06/04/2009 Table 5. Limited frequency (period) of TEC variation over the network between 04/04/2009 and 06/04/2009 No GPS Station Frequency(Hz) Period(min) Date 1 YEBES /04 2 LRC /04 3 TLSE /04 4 MARS inf /04 5 AJAC ( h) AJAC ( h) AJAC ( h) AJAC inf /04 6 GEN /04 7 CAGL /04( h) CAGL (14.45h)-05(21.50h) CAGL Inf (21.51h)-06(08.35h) CAGL /04( h) CAGL /04( h) 8 MEDI /04 9 ROM inf /04 10 NOT /04 11 MATE (00h)-05(24h) MATE inf ( h) MATE ( h) 12 OHRI /04 OHRI ( h) OHRI ( h) OHRI /04 13 SOFI ( h) SOFI ( h) SOFI /04 14 BUCU /4 15 ISTA /04 16 ANKR /04 In Tables 4 and 5 the distributions of the limited frequencies of TEC variations over the area of Southern Europe for the period of a strong preseismic TEC perturbation 11/03/ /03/2009 and for the co-seismic perturbation (04/04/ /04/2009) are being given. Figure 1 and and Figure 2 show the corresponding distributions. It is realized that the limited period of TEC variations become smaller and smaller as long as we aproach the earthquake epicenter and become zero over the nearest stations. This means that the frequency content of TEC variation is extented to the shorter wavelenghs as we aproach the epicenter and near the epicenter no periodical components are disciminated. This is shown in the Figures 13 and 14 which display for a comparison the logarithmic power spectrum of TEC variations over La Rochel and Rome for the time of l Aquila earthquake. 5. Concluding Remarks 17

18 The serveyance of TEC variations over Southern Europe during the last month before the l Aquila earthquake and the subsequent analysis indicate that abrupt TEC variations& accompanied by enhanced oscillations in a broad range of frequencies of TEC& occur randomly over a broad area of several hundred km from the earthquake& 26 days before the earthquake (and maybe even more). High frequency oscillations (f Hz& periods T 60m) seems to point to the location of the earthquake with relatively higher accuracy but the fractal characteristics of the frequencies distribution points to the locus of the earthquake with an even higher accuracy. References Afraimovich E.L.& Perevalova N.P.& Plotnikov A.V.& Uralov A.M.: The shock acoustic waves generated by the earthquakes& ann.geophys.&.19& & Afraimovich.& Perevalova N.P.& Voyeikov S.V.: Traveling wave packets of total electron content disturbances from global GPS network data& LANL e-print archive&2002& Arabelos& D. N.&Asteriadis& G.&Bloutsos& A.&Contadakis& M. E.&Spatalas& S. D.: Atmospheric tide disturbances as Earthquake precursory phenomena &. Hazards Earh Syst. Sci.&4(1)& pp.1-7&2004. Arabelos& D. N.& Asteriadis& G.& Bloutsos& A.& Contadakis& M. E.& Spatalas& S. D.: Correlation between seismicity and barometric tidal exalting& Nat. Hazards Earh Syst. Sci.& 8& 1129-&2008 Arikan& F.&Yilmaz& A.& Arikan& O.& Sa Yin& I.Gurun& M.&Yildirim& S.A.: Space Weather Activities of IONOLAB Group: TEC Mapi& Geoph. Res. Abstr.&Vol 11&2009 Atzori& S.& Hustad& I.& Chini& M.& Salvi& S.& Tolomei& C.& Bignami& C.& Stramondo& S.& Trasatti& E.& Antonioli& A.& and Boshi& E.: Finite fault inversion of DInSAR coseismic displacement of the 2009 Aquila earthquake(central Italy)& Geophys. Res. Lett.& 36& L15305& doi: /2009gl039293&2009. Biagi P.F.& Piccolo R.& Capozzi V.& Ermini A.& Martellucci S. and Ballecci C.: Exalting in atmospheric tides as earthquake precursors& Natural Hazard and Earth System Sciences& 3(3/4)& & Biagi& P.F.& Castellana& L.&Maggipinto& T.& Loiacono& D.& Sciavulli& L.& Ligonzo& T.& Fiore& M.& Suciu& E.& and Ermini& A.: A pre seismic radio anomaly revealed in the area where the Abruzzo earthquake (M=6.3) occurred on 6 April 2009& Nat. Hazards Earh Syst. Sci.& 9& & Contadakis& M.E.& Arabelos& D.N.& Asteriadis& G.& Spatalas& S.D.& Pikridas& C.: TEC variations over the Mediteranean during the seismic activity period of 20th October& in the area of eastern Aegean.& Geoph. Res. Abstr.&Vol 9&2007 Contadakis& M.E.& Arabelos& D.N.& Asteriadis& G.& Spatalas& S.D.& Pikridas& C.: TEC variations over the Mediteranean during the seismic activity period of the last quarter of 2005 in the area of Greece& Nat. Hazards and Earth Syst. Sci.& 8& &2008. Ciraolo& Privet communication& Cirella& A.& Piatanesi& A.& Cocco& M.& Tinti& E.& Scognamiglio& L.&Michelini& A.& Lomax& A.& and Boschi& E.: Rapture history of the 2009 L Aquila (Italy) earthquake from non-linear joint inversion of strong motion and GPS data& Geophys. Res. Lett.& 36& L19304& doi: /2009gl039795&2009. Hayakawa& O.& Molchanov& O.A.& Kodama& T.& Afonin& V.V.& Akentieva& O.A.: Plasma density variations observed on a satellite possibly related to seismicity& Adv. Space Res. Lab.& 26 (8)& & Korepanov V.& Hayakawa M.&Yampolski Y.& Lizunov G.: AGW as a seismo-ionospheric responsible agent& Physics and Chemistry of the Earth (in press)& Liperovsky V.A.& Meister C.-V.& Liperovskaya E.V.& Vasil eva N.E.& Alimov O.: On Esspread effects in the ionosphere before earthquakes Natural Hazard and Earth System Sciences& 5& No. 1& 59-62&

19 Misiti M.& MisitiY& Oppenheim G.& Poggi J. M.: Wavelet Toolbox& the Math Works Inc& Molchanov O.& Biagi P.F& Hayakawa M.& Lutikov A.& Yunga S.& Iudin D.& Andreevsky S.& Rozhnoi A.& Surkov V.& Chebrov V.& Gordeev E.& Schekotov A.& Fedorov E.: Lithosphere-atmosphere- ionosphere coupling as governing mechanism for preseismic short-term events in atmosphere and ionosphere & Natural Hazard and Earth System Sciences& 4& 5/6& & 2004 Molchanov O.& Schekotov A.& Solovieva M.& Fedorov E.& Gladyshev V.&Gordeev E.& Chebrov V.& Saltykov D.& Sinitsin V.I.& Hattori K.& Hayakawa M.: Near seismic effects in ULF fields and seismo-acoustic emission: statistics and explanation& Natural Hazard and Earth System Sciences& 5& 1-10& Muto& M.& Yoshida& T.& Horie& M.& Hayakawa& M.& Parrot& M.& Molchanov& O.A.: Detection of ionospheric perturbations associated with Japanese earthquakes on the basis of reception of LF transmitter signals on the satellite DEMETER& Nat. Hazards Earth Syst. Sci.& 8& & Roznoi& A.& Solovieva M.S.& Molchanov& O.A. and Hayakawa& M.: Middle latitude LF (40kH) phase variations associated with earthquakes for quiet and disturbed geomagnetic conditions& Phys. Chem. Earth& 29& & 2004 Roznoi& A.& Molchanov& O.&Solovieva& M.& Gladyshev& V.& Akantieva& O.& Berthelier& J.J.& Parrot& M.& Lefeuvre& F.& Hayakawa& M.& Castellana& L. and Biagi& P.F.: Possible seismo-ionosphere perturbations revealed by VLF signals collected on ground and satellite& Nat. Hazards Earth Syst. Sci.&7& & 2007 Roznoi& A.& Solovieva M.& Molchanov& O.&Schwingenschuh& K.& Boudjada& M.Y.& Biagi& P.F.& Maggipinto& T.& Castellana& L.& Hayakawa& M.: Anomalies in VLF radio signals prior the Abruzzo earthquake(m=6.3) on 6 April 2009& Nat. Hazards Earth Syst. Sci. 9& & Parrot& M.& Berthelier& J.J.& Leberton& J.P.& Sauvaud& J.A.& Santolik& O. and Blecki& J.: Examples of unusual ionospheric observations made by the DEMETER satellite over seismic regions&phys. Chem. Earth& 31& & Svets A.V.& Hayakawa M.& Molchanov O.A& Ando: A study of ionospheric response to regional seismic activity by VLF radio sounding& Phys. & Chem. of the Earth& 29& & Turcotte D.L.: Fractal and Chaos in Geology and Geophysics (2 nd Edition)& Cambridge University Press& Cambridge U. K.&