COMPARATIVE ANALYSIS OF IONOSPHERIC VARIATIONS …geo.phys.spbu.ru ›...

COMPARATIVE ANALYSIS OF IONOSPHERIC VARIATIONS BEFORE STRONG EARTHQUAKES

T.V. Gaivoronskaya

Pushkov Institute of Terrestrial Magnetism, Ionosphere and Radio Wave Propagation RAS

(IZMIRAN), Troitsk, Moscow Region, 142190, Russia e-mail: [email protected]

Abstract. At two ionospheric stations Petropavlovsk and Magadan the series of deviations of critical frequencies foF2 from their moving average values are considered before strong earthquakes. The irregular variations at both stations are compared to exclude the identical changes connected with geomagnetic disturbances and to reveal seismo-ionospheric effects. It has been found that positive deviations are predominated in the time of preparation of strong earthquakes. They become more intense with increase of magnitude of earthquakes.

Introduction

Anomalies of electronic concentration of the ionosphere at middle latitudes, which are connected with seismic activity, are usually less than ionospheric disturbances observed during global geomagnetic storms, therefore the seismic effects are quite often difficult to identify. The earthquakes occur both in geomagnetic quiet and disturbed periods. It is a note in this connection that geomagnetic storms are global phenomena, whereas premonitory ionospheric effects of earthquakes are essentially local. In general, characteristics of F2-layer are qualitatively identical at stations if they are located from each other on distances no more than 500-700 km. However at such distances one of the stations can be in a zone of seismic activity, and another - outside of it, so that seismo-ionospheric disturbances should be marked at the first station and to be absent or be less intense at other. In order to reveal seismo-ionospheric disturbances, the comparative analysis of data at two stations of radiosounding Petropavlovsk and Magadan has been made. Earlier, data at two stations are compared with each other, and daily correlation coefficients of ionospheric parameters are calculated. It is received that before strong earthquakes the correlations are quite often broken (Gaivoronskaya et al., 2002; Gaivoronskaya, 2005). Now hourly variations of critical frequencies foF2 are examined in more detail.

Methods

There have been considered ten strong earthquakes in 1992-1994 with magnitudes from М=5.4 up to М=7.5 at seismic region - Kamchatka. Zone of active preparation of earthquakes usually covers the area in radius of 200-300 km (Bowman et al., 1998; Dobrovolsky et al., 1979), thus the earthquakes with epicenter on distances not farther than 250 km from Petropavlovsk-na-Kamchatke have been chosen. The ionospheric F2-layer is the most variable of regular layers, and its critical frequency foF2 is one of the most accurately measured parameters of ionosphere, therefore that frequency has been taken for analysis of data series obtained at stations located near earthquake areas. We have examined the series of ΔfoF2 deviations of critical frequencies of F2-layer from their moving average values at two ionospheric stations Petropavlovsk and Magadan. Usually the average values are calculated for a month, they are derived from the data for the previous and following 15 days. In case of forecasting of seismic events it is necessary to calculate average values only for previous period of 15 days when results of radiosounding are known. The hourly irregular variations received at two stations Petropavlovsk and Magadan are compared with each other:

Proceedings of the 7th International Conference "Problems of Geocosmos" (St. Petersburg, Russia, 26-30 May 2008)

437

ΔfoF2 (P-М) = ΔfoF2 (P) - ΔfoF2 (М) = [foF2 (P) - foF2 (М)] - [C (P) - C (М)], C is the moving average value of frequencies on previous 15 days. Thus it is possible to exclude the significant ionospheric variations connected with geomagnetic storms, and also some distinctions between average values foF2 at two stations. Fig. 1 illustrates results of calculations of variations ΔfoF2 (P) at station Petropavlovsk (above) and relative variations ΔfoF2 (P-М) by comparison of data of stations Petropavlovsk and Magadan (below). On horizontal axis the hours of local time of Petropavlovsk are marked. Irregular variations are calculated within 12 days, from 23 February till 5 March 1992, when there are a series of earthquakes, the greatest of which with magnitude M = 6.8 was registered on 2 March 1992. A few days before the geomagnetic storm take place. In the top figure the significant negative irregular variations connected with geomagnetic storm are visible, while at the bottom figure the negative disturbance practically is absent. However at bottom figure the positive disturbance 7 day prior to the main seismic shock is noticed. Fig. 2 shows another example of relative variations ΔfoF2 (P-М) before an earthquake on 8 June 1993. In that case there is a catastrophic earthquake with magnitude М=7.5 and a series of aftershocks. Considerable positive deviations are observed before the main shock and one of the aftershocks.

0 24 48 72 96 120 144 168 192 216 240 264 288-8-6-4-2024

.fo

F2(P

), M

Hz

Petropavlovsk 23 February-05 March 1992

6.8 6.0 6.3

0 24 48 72 96 120 144 168 192 216 240 264 288hours LT

-4

-2

0

2

4

.fo

F2(P

-M),

MH

z

Petropavlovsk-Magadan 23 February-05 March 1992

6.8 6.0 6.3

Fig.1. Variations ΔfoF2 (P) at station Petropavlovsk (above) and relative variations ΔfoF2 (P-М), received at comparison of data of Petropavlovsk and Magadan (below),

before earthquake occurred 2 March 1992.


438

0 24 48 72 96 120 144 168 192 216 240 264 288-3

-1

1

3

.fo

F2(P

-M),

MH

z Petropavlovsk-Magadan 02-13 June 1993

7.5 5.1 5.5 5.8 6.3

hours LT

Fig.2. Relative irregular variations ΔfoF2 (P-М) at Petropavlovsk as compared with Magadan before earthquake on 8 June 1993 with magnitude М=7.5.

Results

So only a small part of energy of earthquakes penetrates into ionospheric altitudes, the disturbances of critical frequencies foF2 caused by seismic activity are difficult to reveal on the background of daily irregular variations, particularly on the background of ionospheric storms. Comparative analysis of data of radiosounding at two stations Petropavlovsk and Magadan allows us to exclude the disturbances connected with geomagnetic storms and to select seismo-ionospheric effects.

It is received that 10-12 days before strong earthquakes in 1992-1994 the irregular positive variations are predominated at station Petropavlovsk as compared with station Magadan. Calculations show that positive deviations become visible in the period preceding strong earthquakes with magnitude M=5.8 and more. Positive variations are more expressed with larger magnitude of earthquake. It confirms that the effect is just connected with seismic activity.

References Bowman, D.D., G. Quillon, C.G. Sammis, A. Sornette, and D. Sornette (1998), An observation test of the critical earthquake concept, Journal of Geophysic Research, 103, 24359-24372. Dobrovolsky, I.R., S.I. Zubkov, and V.I. Myachkin (1979), Estimation of the cize of earthquake preparation zones, Pure Applied Geophysics, 117, 1025-1044. Gaivoronskaya, T.V., and S.A.Pulinets (2002), Analysis of the F2-layer variability in the areas of seismic activity, 20 pp., Preprint IZMIRAN, N2 (1145), Moscow. Gaivoronskaya, T.V. (2005), Ionospheric variations in seismically active regions, Izvestia, Physics of the Solid Earth, 41, N3, 56-60.


439

COMPARISON OF EXPERIMENTAL AND MODEL Q-BURSTS IN TIME DOMAIN

M. Hayakawa1, A.P. Nickolaenko2, T. Ogawa3, M. Komatsu4

1Department of Electronic Engineering, UEC, Chofu-city, Tokyo, Japan, e-mail:

[email protected] 2Usikov Institute for Radio-Physics and Electronics, NAS of the Ukraine, Kharkov, Ukraine, e-mail:

[email protected] 3Science Laboratory International, Kamobe, Kochi, Japan, e-mail: [email protected]

4Polytechnic College Kochi, Noichi, Kochi, Japan, e-mail: [email protected]

Abstract. We compare natural extremely low frequency pulses (Q-bursts) with computations of analytical time domain solution. The wide band receiver was used in experiment with the upper cut-off frequency about 11 kHz. The ball antenna allowed for ELF-VLF records of vertical electric field in the fair weather conditions with a unique resolution of 22 kHz sampling frequency. We use the uniform Earth – ionosphere cavity model, the linear frequency dependence of propagation constant, and we find an excellent agreement between the observation and modeling.

Introduction and experimental setup

Term ‘Q-burst’ is related to discrete natural ELF radio pulses detected worldwide and lasting for 0.3 – 1.5 seconds. These signals originate from powerful lightning strokes whose pulsed amplitude exceeds the continuous Schumann resonance (SR) background by a factor ten or greater. Some Q-bursts were related to the ‘red sprites’. In the present study we compare the precise electric field measurements in the ELF-VLF band with the time domain solution for the uniform Earth–ionosphere cavity [see Nickolaenko, Hayakawa, Ogawa, and Komatsu, 2008 for detail].

Fig. 1. Frequency response of ELF – VLF receiver. The wide band receiver detects the ELF – VLF pulses, and the narrow band filtering corresponds to typical response of a receiver for the global electromagnetic

(Schumann) resonance studies.

A ball antenna was istalled at Tochi station (33.3° N, 133.4 ° E) a few kilometers away from the center of Kochi City, Japan. Electric fields were observed in the fair weather afternoons covering the four seasons since November 2003. The ‘narrow band’ and ‘wide band’ frequency responses of our receiver are


440

shown in Fig.1. We discuss the wide band data acquired with the 22 kHz sampling rate [Ogawa and Komatsu, 2007].

Waveforms observed

Typical waveforms are shown in Fig. 2, which usually corresponded to the source – observer distances D close to 20 Mm. These pulses contain the direct and antipodal waves that merge into a W type Q-burst. Two kinds (the wide – and narrow–band) of waveforms were acquired, and we use below the wide band records for a comparison with model computations.

Fig.2. A Q-burst of W type in the wide and narrow frequency bands.

One may conclude by comparing the upper and lower panels in Fig.2 that wide– and narrow

band waveforms are different. The wide band signal contains the resolved leading and rear parts. The first one contains the direct (initial pulse) and antipodal (second pulse) waves forming the W pattern. The secondary or rear pulse is a combination of merged round–the–world waves, and it is delayed from the first pulse by ~0.16 s. The delay allows for estimating the signal velocity in the Earth – ionosphere cavity being about 250 thousands kilometers per second [Ogawa and Komatsu, 2007]. The fine details of the record are absent in the narrow band channel: individual atmospherics become invisible. We know [Nickolaenko, Hayakawa, 2002] that for large source – observer distances cavity losses concentrate the pulse spectrum below 100 Hz. Therefore, the waveform modifications although being present in the narrow band do not unrecognizably change the waveform. One can identify the W pattern at the beginning. The dispersion pertinent to the narrow band filtering provides a greater impact on the delayed ELF waveform by turning it into a kind of attenuating sinusoid.


441

Model

We compute waveforms directly in the time domain for the uniform Earth – ionosphere cavity model. The ‘linear’ propagation constant is used ( ) ( ) 10062 fiff −−=ν that was found from the Schumann resonance cross-spectra measured at large longitudinal separation of two observatories [Nickolaenko and Hayakawa, 2002]. Computations were performed with the algorithm accelerating the convergence of time series.

Fig.3. Evolution of E – and H – pulses with the source distance. Signal focusing is obvious around the source antipode.

A flat infinite frequency response of the receiver was assumed. Figure 3 depicts a series of

pulsed waveforms computed for a set of source – observer distances ranging from 2 to 20 Mm. The field increase is clearly seen at the vicinity of antipodal distance of 20 Mm. We used the ‘white’ source current moment of a lightning discharge with Ids (f) = 108A*m. The above current moment is related to a stroke with the peak current of 25 kA and the channel length of 4 km. Since we compare experimental and computational waveforms, the particular source amplitude is insignificant. The stroke polarization was positive. Visual comparison of pulse amplitudes indicates that the above current moment underestimates the observed value by a factor of up to five, which is in agreement with regular Q-bursts observations.

Comparison of experimental (black) and model (color) data

Frames below present individual pulses: black lines depict the results observed and the model data are shown by the red dotted lines. The time is shown along the abscissa, and the pulse middle sub-peaks were aligned to facilitate the comparison. The major feature of all data sets was their high reciprocity. Once the


442

sub-peaks coincide, the rest of the waveforms became synchronized also, provided that a correct source – observer distance was found.

Pulse No.1. Industrial 60 Hz interference is clearly seen.

Pulse No.2. A few discrete pulses and a distinct negative slow tail atmospheric are superimposed on the waveform of the Q – burst. Industrial interference is noticeable.


443

Pulse No.3. A positive slow tail atmospheric is combined with the Q – burst together with continuous 60 Hz industrial interference signal.

Pulse No.4. A W type Q – burst arrived from antipodal distance. Direct and antipodal waves almost merge.

Many ‘short distance’ VLF atmospherics were detected indicating on a high activity of the Asian thunderstorm center.


444

The left ordinates of each frame show experimental amplitude of vertical electric field in all the frames. The right ordinate shows the computed amplitude corresponding to the red dotted curves. As one may see, the source amplitude used in our computations was underestimated in comparison with measurements by a factor of 2 – 5.

The wide band record in the above frames resolves small ‘regular’ spikes responsible for the Schumann resonance background signal. Different kinds of atmospherics occur during the records including the ‘slow tails’ denoted in the plots. These signals were not processed during this particular study. Our wide band measurements applied an exceptional sampling frequency of 22 kHz, they covered both the slow tail (VLF) and the Schumann resonance (ELF) bands, and very pure records were obtained. The high quality of experimental records is clear. In particular, the industrial interference at 60 Hz frequency (radiation from local power supply lines) is very small.

Conclusion

Data presented in this report allow for the following conclusions:

Experimental and model data are similar: if one aligns the first (arrival) peak, positions of all other peaks practically coincide. The pulses have rather sophisticated waveforms, which do not resemble an attenuating sinusoid.

Results of precise ELF – VLF observations correspond to the radio propagation model within the uniform Earth – ionosphere cavity having the linear frequency dependence of propagation constant. Similarity of model data to observations confirms the accuracy of such a model.

Insignificant deviations between the plots are easily explained by the finite pulse – to – background ratio (~ 10), by an impact of industrial 60 Hz interference, and by the random (hence, unknown) details of the spectrum of a parent lightning stroke.

References Nickolaenko, A. P., and M. Hayakawa (2002), Resonances in the Earth-ionosphere Cavity, 380 pp., Kluwer

Acad., Dordrecht, Netherlands. Ogawa, T., and M. Komatsu (2007), Analysis of Q-burst waveforms, Radio Sci., 42, RS2S18,

doi:10.1029/2006RS003493. Nickolaenko, A. P., M. Hayakawa, T. Ogawa, and M. Komatsu (2008), Q-bursts: A comparison of

experimental and computed ELF waveforms, Radio Sci., 43, RS4014, doi:10.1029/2008RS003838.


445

FINE STRUCTURE OF ELECTRIC PULSED FIELD ABOVE THE BENT STROKE OF LIGHTNING

I.G. Kudintseva1, A.P. Nickolaenko2, and M. Hayakawa3

1Karasin National University, Kharkov, Ukraine,

2Usikov Institute for Radio-Physics and Electronics, National Academy of Sciences of the Ukraine, Kharkov, Ukraine, e-mail: [email protected]

3The University of Electro-Communications, Chofu-city, Tokyo, Japan, e-mail: [email protected]

Abstract. We present the pulsed electric fields computed in the neutral atmosphere above a powerful positive lightning stroke with the bent channel containing vertical and horizontal sections, each 10 km long. The fine structure of field is demonstrated arising in the space due to a combination of delayed pulses arriving from the stroke sections. The electric field distribution depends on time and on the stroke orientation with respect to an elevated observer. Characteristic size of ‘filaments’ in the transient electric field is about 1 km along the horizontal direction, and it reaches a few tens of kilometers in the height.

Introduction The present paper is related to the red sprite phenomena. The detailed description of the

phenomenon and its model might be found in the review paper by Pasko (2006). A specific feature of atmosphere transient luminous events (TLE) is the fine tendril structure of sprites possibly associated with the horizontal sections of the parent lightning channel.

We do not model development of a sprite in the present paper. Instead, we discuss the property overlooked in literature: the transient electric field has a structured spatial distribution. We treat the simplest possible model of lightning stroke. The ‘engineering model’ of a return stroke is bent (broken) in the middle, so that the current wave moves along the Γ–shaped channel: initially it goes upward and afterwards it turns horizontally. We demonstrate that such a stroke provides three pulses in the mesosphere. An interference of these pulses in space causes a sophisticated structured field distribution. The fine structure of the field may induce the bunching of charged particles, which form the filaments of a transient luminous event. Model description

When computing electric pulses in the mesosphere, we use the model of a stroke with the bent channel. The stroke channel is 20 km long. Initial, vertical half of the channel is 10 km; it coincides with the vertical Z-axis of the Cartesian coordinate system. The Γ–shaped stroke is bent at 10 km altitude and the current proceeds horizontally here along the X-axis. We use the ‘engineering model’ with the following

current at the base of a positive stroke: ( )∑=

−⋅=4

1

exp)(k

kk tItI τ , t ≥ 0. Amplitudes Ik and relevant time

constants are: I1 = 569 kA, I2 = −460 kA, I3 = −100 kA, I4 = −9 kA, τ1 = 16.6 µs, τ2 = 333.3 µs, τ3 = 0.5 ms, and τ4 = 6.8 ms. Current wave propagates along the stroke with the velocity ( )Vo tVtV τ−⋅= exp)( , Vo = 8⋅107 m/s, and τV = 0.25 ms. Total length of the channel is 20 km. Initial electric moment is vertical. The

horizontal current appears when

−

=>koV

oVk zV

Vttτ

τ ln , the latter is necessary for reaching the ‘turning’

altitude zk = 10 km. Reflections from the perfectly conducting ground are also included. Radiation component of electric field (components EX, EY, and EZ) was computed for an arbitrary

point in neutral atmosphere. In the present treatment we find the spatial distribution of electric field above the Γ–shaped discharge and demonstrate its fine structure never mentioned in the literature.


446

Computed results

Electromagnetic pulse (see Fig.1) originates from the vertical section of lightning and its reflection in the ground. Radiation from the horizontal section (the second pulse in Fig.1) appears after the current wave turns horizontally (t = 390 µs from the stroke initiation). Finally, the wave reflected from the ground provides the third pulse. Particular waveforms were computed at an observer altitude z = 80 km. Shown are the “Along” and “Across” positions. The origin of coordinate system is set at the base of the cloud-to-ground stroke. The horizontal branch is oriented along X – axis. Position “Across” is in the YZ plane (or x = 0) at a horizontal distance 50 km from the stroke base. Position “Along” is in the XZ plane (y = 0) and has the same horizontal distance. The abscissa in Fig.1 depicts the time in µs from the stroke initiation. Transient electric field is plotted on the ordinate in V/m. Component EX is depicted by the red line, EY – by blue, and EZ – by the black line.

‘ACROSS’ ‘ALONG’ z = 80, x = 0, y = 50; z = 80, x = 50, y = 0.

Fig.1. Pulses radiated by vertical and horizontal sections of the Γ–shaped stroke.

Pulses from horizontal section are delayed with respect to those arriving from the vertical part of the stroke; this is caused by the finite velocity of the current wave. The third pulse corresponds to reflections from the ground. It has the opposite sign in comparison with the ‘direct horizontal’ signal. To obtain spatial distributions of electric fields, we computed temporal variations at positions ranging from 50 to 100 km vertically and from –50 to 50 km horizontally. The EX, EY, and EZ data allowed us to construct the spatial dependence of all field components for the fixed time moments. The 2D maps of these 3D distributions reflect the pulse interference in space.

We depict in Fig.2 the ‘motion picture’ of vertical electric field EZ distribution for five particular times t ∈ [320; 440] µs varying with the 30 µs step. Two cross-sections are shown: the XZ (Along) and the YZ (Across) planes. Fine structure of the field is seen with a general upward progress of the pulsed fronts. The filament structure of the field arises from delicate interaction of delayed pulses coming to the elevated


447

observation point. The model indicates that a powerful stroke with only two sections of current channel can form a complicated field structures in mesosphere. Such fields ‘warm up’ fragments of atmosphere thus supporting formation, motion, and branching of future sprite filaments.

Fig.2. Spatial distribution of the EZ field component.

Plots in Fig.2 indirectly show the ‘doughnut’ radiation pattern from the vertical section of the stroke, as might be expected (the red structure in the lower left frame and the crimson structure in the right one). The higher rows of plots correspond to the situation when pulses from the horizontal section of stroke reach the mesosphere. Particular arrival times depend on the distance and orientation of horizontal branch with respect to the observer. Therefore, different ‘hair comb’ distributions arise depending on the field component and the


448

plane of observation. The positive fields are shown by reddish colors, and negative – by the blue palette. The first attract free electrons, while the second repel them. As a result, the atmosphere particles are subjected to the field, which might cause free electron bunching. It is worth noting that the field structure strongly depends on the position (along or across) of the observer.

Obviously, electric field in the atmosphere is structured and variable within an arbitrary plane. We have shown two planes for the reason that relevant plots are highly symmetric, which is easily understood from simple considerations. Discussion and conclusion

We considered spatial structures of transient electric field above a simple bent or Γ–shaped positive stroke. A fine constitution of electric field in space arises from the sequential radiation from vertical and horizontal sections of the lightning channel. The model was applied of a return stroke having single vertical and a horizontal section, each 10 km long. The current waveform and the velocity of current wave in the channel were borrowed from usual models of return strokes (e.g., Nickolaenko and Hayakawa, 2002). Physics of a bent lightning stroke was not addressed, as this is a separate and a complicated problem itself. We note that mere presence of two sections in the stroke channel causes the spatial effects demonstrated since a Γ–shaped stroke provides three delayed pulses in the atmosphere. These pulses cause a fine structure in space.

The number of pulses increases when a stroke acquires additional sections. For example, a discharge of two vertical and two horizontal segments provides seven pulses. Hence, relevant spatial distributions of electric field may become much more complicated even when the horizontal sections remain collinear. It is obvious that particular distribution of the field and its temporal variation depend on the stroke morphology and on the particulars of current wave progress along the channel. Real strokes, such as ‘spiders’, have many sections and branches; these may cause an extremely sophisticated distribution of fields in space.

A standard explanation of sprite development exploits the quasi-electrostatic field (Pasko, 2006), which slowly varies in time and space. In contrast, we turned to the radiation component alone and found complicated spatial structures. It is obvious that all the fields – the static, induction and radiation take part in the sprite formation. Radiation field decreases with distance as r1 , but it has a smaller initial value than the static and induction fields. The induction and static field decrease faster, as 21 r and 31 r correspondingly. All three of them become equal at the distance where 1=kr , this is a classical definition of the ‘far zone’. By assuming r = 50 km, we find that the equality is held for f = 955 Hz. Hence, the radiation field will dominate in the mesosphere, as the spectrum of a typical stroke occupies frequencies much higher than 1 kHz.

Influence of atmospheric conductivity was ignored here, as is usually done at ‘high’ frequencies (Pasko, 2006). If one accounts for the finite conductivity, the electric field will decrease faster with distance; the details will be ‘smoothed’ in time and space, separate filaments might ‘blur’ or ‘merge’ owing to the wave dispersion, but the fine structure will not vanish completely.

We found the downward/upward and sideways driving forces, which are highly structured in space. Future modeling should involve the complete fields including the pulsed ones. Incorporated in the motion equations, the transient forces will cause focusing of free charged particles on the time scales exceeding the pulse duration, as we discuss below.

Consider a one-dimensional electron beam that moves upward with the velocity U0 = 5000 km/s, that corresponds to the particle energy of E = 1.14×10−10 erg or to ~71 eV. Let a short pulse ∆E of the ∆t duration interact with the beam at altitude h0 = 60 km. We assume that the pulse simultaneously reduces the velocity

by tm

EeUe

∆⋅∆⋅

−=∆− of all particles from the interval [h0 − ∆h; h0 + ∆h], and they loose the moment

teEp ∆⋅⋅∆−=∆ . Electrons from altitude interval [h0 − ∆h; h0 + ∆h] will lag behind the median flow, and the latter is maintained by the ‘background’ quasi-static field. Thus, a juvenile clustering of particles may appear in the beam, which we ignore. Further, a new pulse of opposite (positive) polarity arrives at the same altitude interval being delayed by τ = 10 ms. The positive pulse accelerates particles by increasing their

velocity by tm

EeUe

∆⋅∆⋅

=∆ .


449

In our speculations, the decelerated electrons leave the altitude h0 with the velocity UU ∆−0 , while those accelerated have the velocity UU ∆+0 , but they are delayed by the time τ. Obviously, the particles of the second group overtake the first group at some altitude above the h0. The vertical coordinate of each group varies in time as: ( ) ( )tUUhtz ∆−+= 001 and ( ) ( )( )τ−∆++= tUUhtz 002 . They become equal at the

time of bunching τδδτ

21

20 +

=∆∆+

=U

UUtb where 0U

U∆=δ . Focusing (bunching) will occur at the height

δδτ

21

00+

+= UhHb . One may observe that there is an important model parameter describing the

‘extension’ of time: δδ

δ21

21

≈+

=R . This is the ratio of focusing time to the initial retard τ of second

electric pulse. Let us substitute a particular value of δ = 1/200 velocity modulation or ∆U = 25 km/s, which is quite

realistic. Indeed, if the characteristic duration of radio pulse is ∆ t = 0.2 µs, the necessary alteration in the electron mechanical momentum is tEeUme ∆∆=∆ . The causative electric pulse has to be of

tU

emE e

∆∆

=∆ amplitude or, ∆E = 7 V/m. In other words, an electric pulse of 7 V/m amplitude and of

0.2 µs duration modulates the above postulated electron velocity U0 by the factor of 1/200. It is easy to calculate now that both groups of electrons will meet at the Hb = 60 + 5 = 65 km altitude, and the encounter is shifted in time by tb = Rτ = 1 ms from the onset of the paired electric pulse. Our elementary estimates provide quantities perfectly agreeing with the sprite observations.

Structure of a real lightning stroke is more complicated than that of a simple Γ – shaped model. However, the simplest Γ – shaped model readily predicts focusing of free charged particles in the atmosphere above the stroke.

To conclude the paper, we list the main results.

1. Structured spatial distribution was obtained of the transient electric fields in the mesosphere above a Γ – shaped lightning stroke. The fine structure appears even in the simplest model containing vertical and a horizontal section.

2. Characteristic size of filaments in the electric field distribution is about 1 km on the horizontal direction and reaches a few tens of kilometers along the vertical.

3. The spatial distribution originates from the pulse series arriving at an elevated observer. Their superposition provides the structured electric field in space.

4. Transient electric fields are able to cause focusing of free charged particles in the atmosphere. Such a mechanism must be included into advanced models of sprite development.

Reference Nickolaenko, A. P., and M. Hayakawa (2002), Resonances in the Earth-ionosphere Cavity, 392 pp., Kluwer

Acad., Dordrecht, Netherlands. Pasko, V.P., (2006). Theoretical modeling of sprites and jets, M. Füllekrug et al (eds.), “Sprites, Elves and

Intense Lightning Discharges”, 253 – 311, © Springer. Printed in Netherlands.


450

STRESSES IN ROCK SAMPLES AND ELECTROMAGNETIC RELAXATION TIMES

Lementueva R.A., Gvozdev A.A., Irisova E.L.

Foundation of the Russian Academy of Sciences Shmidt Institute of Physics of the Earth RAS,

Moscow, Russia, e-mail: [email protected]

Abstract. The results of laboratory studies made on rock samples and models are given in the work. Samples were exposed to mechanical loading by means of press. The method of free relaxation of absorption current has been applied. Curves of the dependence of the decrement of absorption current on time have been obtained at different values of mechanical stress. The curves are not characterized by one, but a number of relaxation times-short and long. Short times changes are pronounced at the increase of stresses, but long relaxation times increase greatly. It is evident that the increase of relaxation times is connected with the formation of micro-cracks in the sample. The formation of the main crack leads to the streak decrease of long relaxation time. Thus relaxation times of absorption current characterize the change in the system of cracks. The results of the investigations can be applied in the development of new methods of natural studies in seismology.

Introduction

One of actual problems in geophysics is studying a process of destruction of rocks and changes of the

geophysical fields observable during the formation of cracks. In a zone of preparing destruction there is a complex intense condition which leads to the generation and germination of cracks. In such zones significant anomalies of various geophysical fields are expected to develop. Fast and slow relaxation processes in electric fields at deformation of rocks are studied poorly. Practically mechanisms of occurrence of polarizability in complex stress conditions are not studied. In a number of works (A.V. Ponomarev and G.A. Sobolev) it was marked that the change of a seeming resistance is an effective precursor of preparing destruction of a file of rocks. However till now completely there are not clear the nature of anomalies and places of the most probable displays. Especially intensive processes of variations of resistance can develop close to nonhomogenities where there are the strongest concentration of pressure. The purpose of the given work is reception of the new information on polarizability of rocks, revealing of possible mechanisms of occurrence and a relaxation of charges in laboratory experiments.

Methods Rocks can be presented as an imperfect dielectric with losses in parallel connected resistance and

capacity. The general current arising under the action of an electric field, it is possible to present it as the sum defined by the formula:

I = I3 + I0 + Ia ,

where I - general current through the dielectric, I3 - current of charge, I0 - residual current, Ia - current gradually weakening in due course. It is named reversible current of absorption, and the phenomenon of its occurrence - dielectric absorption (Sidorov, V.A.). Falling off of a current in due course can be described by means of a set of exhibitors and to present by the formula:

Ia = 1 20 1 .....

t t tn

nA e e eA A ττ τ− − −+ +

Thus, when studying a current of absorption (Ia) it is necessary to measure the rate of recession of the

curve of polarizability. This method is known as a method of free relaxation of a current of absorption in absence of an external circuit (Sidorov, V.A.).

For studying recession of a current of absorption from natural (pirophillit, sandstone, and others) and modeling (concrete) materials, different samples were made (fig.1). Samples of the rectangular form were


451

made by the size 10х7х5 cm. Models of large-scale samples (fig. 1d) reached the size 200х70х50 cm. Cores had the sizes: height: 6- 7 cm, diameter: 5 cm.

For loading we applied the press of 50 ton (Troitsk). Within 1-2 days before loading the measurements of background curves of recession of a current of absorption were spent at F=0, and average curve Ia = f (t) was under construction. In cycles loading measurements were carried out at stages of constant loadings, and before the measurement the sample or model were exposed to constant pressure within 3-4 minutes. The same process was repeated at other loading. On a surface of samples from pirophillit, limestone, and sandstone, measuring electrodes MN were fixed which were established in a stratified zone of lamination and a predicted zone of concentration of pressure. In the top and bottom parts of the sample there are current gauges A and B made of graphite paste; transitive resistance of contact to the sample was less than 2 kOm for models from concrete.

Fig.1. Models. a - a core, b - a sample of the rectangular form, c - a sample of the rectangular form with concentrators, d - a large-scale sample from concrete.

The model from concrete with the additive of a graphite dust was applied. In this case, the influence of

a mineralization on the behavior of electric parameters was studied at loading. On large-scale model process of a relaxation of a current of absorption in a zone of asperity of blocks was studied at development of a condition of instability (Fig. 1d). On this sample in the work of R.A. Lementueva, V.I. Ponjatovskaya, E.I. Irisova, and V.A. Popov variations of various geophysical fields were investigated. In the given work it is experimentally shown that the process of a relaxation of current Ia is possible to study in a local site of the large-scale sample, in the case of break of asperity with the movement of the fault.

Samples were humidified and contained 3 - 5 % of moisture. That provided a condition of rocks close to the natural one.


452

Studying of electric processes was spent on the equipment described by Komarov, V.A. for method of caused polarization (ВП). The scheme of measuring installation presented in fig. 2, includes: MSVP device (Bobrovnikov, L.Z., L.I. Orlov, and V.A. Popov), batteries, the switchboard of a current, ballast resistance (Rk) and milliampermeter (mА). Gauges M and N (of “Radelkis”) were fixed to lateral surfaces of the sample by means of plugs from plexiglas. The stability of own potentials of electrodes is 0.1 - 0.2 mV.

Fig.2. Scheme of measurements.

The technique of supervision consists in the following. In the beginning natural electric potentials (ΔU0) between each pair of gauges N-M1 and N-M2 have been measured. Then through the sample an impulse of the stabilized current (I) runs of duration of 15 s with an amplitude 20 mА. At the passing of the impulse of current potentials ΔU were registered between the same pairs of gauges.

After switching-off of current I = 20 мА potentials of the caused polarization (ΔUвп) were measured in

0.25, 0.5, 1, 2, 7, 11, and 22 s. Laboratory installation allowed registration of potentials in time оf 1 upto 100 sec. And by that limits of registration (t → 200 sec) extended. It is visible that the process of recession of a

current of absorption Ia = Iaехр (-Δtτ

) is characterized for various relaxations times (1, 2,. .n).

The quantitative estimation of the observable phenomenon in an interval (Δ t) can be received

ln Ia = ln I0 — Δtτ

.

Having defined the time of relaxation, it is possible to speak about a degree of polarizability of environment, because it is known that τ = RC , С~ε , ε — dielectric permeability of environment.

Results Curves of recession of a current of absorption depending on loading for various samples are shown in

fig.3. The analysis of experimental data shows that curves of reduction of a current of absorption in due course, received at various mechanical pressure have an expotential appearance. As the received curve in half-logarithmic scale, is also similar to an exhibitor (instead of a straight line), we observe a number of processes of relaxation with various τ. In the received dependences of a current of the absorption presented in fig. 3а, b, c, d, it is possible to allocate conditionally fast and slow processes of relaxation. The first time of relaxation is defined, when Ia falls down in-е-time. In the field of “slow” (τi ≥ 20s) delay of relaxation was observed at increase of loading (F). At change of F (MPa) τi increased or decreased, depending on variation of loading on the sample.


453

Curves of recession of a current of absorption depending on time at various loadings for various samples. a - rectangular sample from pirofillit, b - rectangular sample from limestone, c - model from the concrete mixed with a graphite dust, d - sample from concrete with concentrators of pressure; 1 – state before loading, 5 – state after loading.


454

The change of loading Fmax up to 0, both τ1 and τ2 decreased (“fast” and “slow” relaxations). The stated

laws of behavior of curves are well visible in fig. 3а (the sample from pirophillit) and 3b (the sample from limestone). Similar dependences have been observed for cores from sandstone. Curves are presented of model from the concrete mixed with a graphite dust (fig. 3c) and with concentrators of pressure (fig. 3d).

Fig.3е. Attenuation of a current of absorption depending on time of large-scale model from concrete. Characteristics of the curve Ia for these models constructed in half-logarithmic scale is similar for

pirophillit, however constants of time τ1 and τn of the relaxation are less and after loading removal (F→0) it is almost a straight line.

Development of a status of instability in zone of modeled break as shown in the experiments [Lementueva, R.A., V.I.Ponjatovskaya, E.I.Irisova, and V.A. Popov], is in a direct communication with the process of accumulation of deformations under the influence of a tension and structural changes in the environment. It, in turn, is reflected in the character of dependence Ia = f (t).

Experiments with large-scale model from concrete are presented in fig. 3е. In fig. 3e results of recession Ia for one pair gauges N – M1 are presented. At occurrence of a local break in asperity it is visible that there is an accelerated process of relaxation Ia (Fig. 3e, curve 2). However loading increase (curve 3) results in increase of time of relaxation. Durability of asperity in this time interval has increased because of a congestion destroyed parts of the material in the fault. Failure of asperity with advancement on a break is characterized by a curve 4 with character change exponent dependences at the point t = 11s. Current Ia falls down at first very slowly and curve — 4 shows that the state before of destruction of asperity is characterized by the big times of relaxation. Thus, as well as in the previous experiments, in a local deformable zone at modeling of a status of instability in a fault occurrence of big times of relaxation before the asperity failure is noted. There is a long steady polarization.


455

Conclusions Results are presented of research of electrophysical characteristics of rocks under the influence of

mechanical loadings on samples. The applied method of free relaxation of a current of absorption has allowed us to reveal the basic distinctions in behavior of a current of absorption Ia at micro- and macrodestruction of samples and models of rocks and at development (in model from concrete) statuses of mechanical instability in the environment. Relaxation time (τn) is a key parameter of process of an establishment and disappearance of volume polarization of rocks. On the basis of the resulted schedules acceleration of process of relaxation can testify to the occurrence of considerable infringements in the environment of rocks. Occurrence of big times of relaxation testifies that in a deformation zone there is an essential destruction, or the beginning of the main crack is formed. The constant relaxation τ in the absorption current is the characteristic of environment in the given time interval and can be used for the description of polarizability of rocks in various stages of loading. Definition τi on a curve constructed in half-logarithmic scale, allows us to calculate in each time interval times τn relaxations Ia. The received results, probably, will promote understanding of the processes which are taking place at destruction of rocks and for improvement of techniques of geophysical measurements.

Acknowledgements The work was partly supported by the grant RFFI 06-05-64888-a.

References

Ponomarev, A.V., and G.A.Sobolev (2003), Earthquakes Physics and Precursors, Nauka, Moscow, 270 pp. (in Russian)

Sidorov, V.A. (1987), About electric polarizability of non-uniform rocks, Izvestia Russian Academy of Sciences, Physics of the Solid Earth. No 10, 58–64.

Lementueva, R.A., V.I. Ponjatovskaya, E.I. Irisova, and V.A. Popov (2007), Results of ultrasonic diagnostic and electrometric measurements at modeling of a fault of complex structure, Geophysical Researches, 7, 65–73 (in Russian)

Komarov, V.A. (1980), Electroinvestigation by a method of the caused polarization, Nedra, Leningrad, 90 pp. (in Russian)

Bobrovnikov, L.Z., L.I. Orlov, and V.A. Popov (1986), Multichannel station of a method of the caused polarization, In: Electroprospecting Equipment: Handbook, Nedra, Moscow, 79–85 (in Russian)


456

OPPORTUNITIES OF USING OF ELECTROMAGNETIC SIGNAL OF LIGHTNING DISCHARGES FOR THE REMOTE SENSING OF SEISMIC

ACTIVITY

V.A. Mullayarov1, V.I. Kozlov1, A.V. Ambursky1

1Yu.G. Shafer Institute of Cosmophysical Research and Aeronomy SB of RAS, Yakutsk, 677980, Russia,

e-mail: [email protected]

By using of earthquake events in the Kamchatka region the opportunity of using electromagnetic signals of lightning discharges (atmospherics) for the remote sensing of seismic activity is considered. Variations of the amplitude of atmospherics, propagating in subionospheric waveguide above areas of earthquake epicentres reflect changes in the structure of electron concentration in the lower ionosphere occurring under the influence of litospheric processes. The characteristic attributes of variations preceding strong seismic events with no deep focus, (electromagnetic precursors of earthquakes) are revealed. The application of algorithm to the extended set of observational data for the winter periods of 2004-2006 has shown that the possibility of a short-term prediction of such earthquake events can reach 60 %. The reasons of "false alarms" (probability of 25 %) are not yet revealed.

1. Introduction As is known the ionospheric variations were often associated with the seismic activity (Liu et el., 2000,

Silina et al., 2001, Pulinets et al., 2004). They appeared a few days or hours before the seismic shocks of large intensity. These effects of seismic phenomena on ionospheric parameters were accompanied by changes in radio station signals whose paths are over quake epicenters (Gokhberg et al., 1989, Molchanov and Hayakawa, 1998, Rozhnoi et al., 2004, Soloviev et al., 2004). The nature of these changes depends on the signal frequency, path length, and on the conditions alone the path. At the same time there are natural sources of VLF emissions that can be used for probing the areas of seismic activity. Electromagnetic emissions during lightning discharges (atmospherics) are the most widespread and most suitable natural sources for this purpose. Although electromagnetic signals deriving from thunderstorm sources are substantially non-stationary, a sufficiently large flow of atmospherics should facilitate the acquisition of statistically significant results.

In this work we examine the elementary algorithm of defining of alert days by a "blindly" retrospective analysis of amplitude variations in VLF signals caused by thunderstorms propagating above the earthquake epicenters in the Kamchatka region and recorded in Yakutsk (ϕ = 62º N, λ = 129.7º E).

2. Methods The influence of processes of a preparing earthquake on the level of VLF-signals received in Yakutsk will

take place, if the area of earthquake corresponds to the first Fresnel zones above the path "a storm source - point of signal receiving". The size of Fresnel zones F is determined by the distance d and wave length λ as F = (λ n d1d2/d)1/2, where d = d1+d2, d1 is the distance from a point up to a source, d2 is the distance from a point up to the receiver, n is the ordinal number of zone.

Therefore for each chosen earthquake we defined the azimuth and distance up to Yakutsk by which we determined the first Fresnel zones. We choose those atmospherics whose azimuths correspond to the calculated Fresnel zones. And a minimal distance up to sources (discharges) must exceed the distance up to earthquakes by 25 %. The average atmospherics amplitude received within an hour (average quantity is of the order 1000 and more) is determined. Preliminarily the signal amplitudes are led to the amplitude of one distance (a distance up to a seismic center), using as a first approximation the dependence of factor of attenuation, inversely proportional to a squared distance.

3. Results Fig. 1 shows the variations of the hourly averaged amplitude of electromagnetic signals of lightning

discharges - atmospherics, which has been already published for the Koryak event of strong earthquake (magnitude 7.6) that occurred on 20.04.06 (20th of April) at 23:35:05 UT. The azimuth on the Koryak seismic


457

center is 76.3 degree, and the distance is 1953 km. The feature of this earthquake was that the great number of aftershocks was observed, and the high magnitude was up to 6.5. Variations of average atmospherics amplitude in the interval 06-26.04.2006 determined for night (a) and day time (b) hours are presented. The preliminary maximum of night atmospherics amplitude was observed on 10.04.06, 10 days before the earthquake, and the minimum - was on 16.04.06 (4 days before the earthquake). The effect of earthquake was displayed the next day after the event as a significant peak of atmospherics amplitude, exceeding the minimum level by a factor of 3. In the level of day time atmospherics the effect of earthquake is expressed even in the greater degree - the excess of the previous level was exceeded by a factor of 10. The day amplitude after that peak did not decrease, and night amplitude dropped on 3rd day, then it has again increased a little. Such behaviour of amplitude can be connected with a significant aftershocks' magnitude.

The short-term (1-2-day) strengthening of amplitude of atmospherics, passing "precisely" above the future epicenter (within limits of the first Fresnel zone) was observed 10 days prior to the earthquake. Maximum in the distribution correspond to the earthquake azimuth (76 degree, Fig. 2).

Then, after the minimum within 3-4 days before the earthquake

has reached the maximum on the next day after the earthquake. Thatmospherics amplitude) was observed in an enough wide area corrconsequent Fresnel zones.

The similar picture of behaviour of average atmospherics ampliof events of strong earthquakes. Results of consideration of amplitunature which are passing above areas of earthquakes with magnitudstationary state of their sources, their amplitude drops 3-6 days


458

Fig.2. The azimuth distribution of the nightamplitude of atmospherics at 10 days before the earthquake (20.04.06).

Fig.1. Variations of the mean night amplitude of the atmospherics received from the various azimuths before and during Koryak earthquakes 20-22.04.06.

there was an increase in amplitude which e effect of earthquake (the increase of the esponding to sizes of the fifth (and more)

tude has been observed in a significant part de variations of VLF-signal of the storm e of more than 5, show that despite non- before earthquakes with its subsequent

restoration by the day of events. The effect is similar to signal amplitude variations of low-frequency radio stations.

It allows us to consider such behaviour as typical and try to formulate one of the possible elementary (primitive) algorithms for the detection of earthquake precursor. A steady increase of atmospherics amplitude within 4 days was defined as "alert" periods - or, really, the total increase of amplitude for such period must exceed the mean-square value of amplitude variations for the previous five days multiplied by a correcting factor (it has been calculated, that it equals 3).

Such an algorithm has been checked up on the Kamchatka peninsula region. Without using of the earthquakes catalogue at the first stage by data of atmospherics received in Yakutsk for winter periods of 2004-2006 according to the considered algorithm the possible "alert" days of earthquakes have been defined. The azimuth 120º (from the northern direction), corresponding to the "center" of Kamchatka has been chosen as the most probable azimuth on seismic centers (within the limits of fifth Fresnel zone).

The analysis of atmospherics amplitude variations for the specified period has shown 8 "alert" days which according to the algorithm can be interpreted as a precursor of earthquakes. Then these alert days were compared with the earthquake catalogue with magnitude more than 4.5, really occurred on the Kamchatka peninsula or in the nearest region. In Fig. 3 the earthquake epicenters are presented, where "+" represents the earthquakes with "alert" and "o" - represents the earthquakes with no "alert". One can see, the centers of earthquakes during the analyzed period were mainly located in two areas, one of which is closer to the chosen direction of atmospherics arrival. In this area practically all earthquakes have "alert" (5 of 7 days). At the same time, in the other region far from the selected direction of atmospherics arrival, only two alerts were defined. The recalculation of atmospherics amplitude in case when the azimuth corresponds to the direction of the second area of earthquakes gives an increase in probability of correct alerts.

Fig. 3. Location the epicenters of earthquake with M > 4.3 at Kamchatka region for winter 2004-2006. The line is the direction from st. Yakutsk to “center” of Kamchatka peninsula.

4. Discussion Thus, a control "blindly" ("in the dark") retrospective analysis using the elementary algorithm of defining

alert days by average atmospherics amplitude has shown that the probability of such short-term forecast of earthquakes can be up to 60-70 %.

Unfortunately, except the omitted earthquakes during the considered period the algorithm has defined 3 false alerts. To find the possible reasons of such false alerts the Kamchatka volcano activity in these days and geomagnetic disturbances have been considered. Any specific activity of volcanoes has not been observed. Ар index is used as the parameter describing geomagnetic disturbances. The behaviour of average values of Ар index in the previous days by the moment's false alerts, is presented in Fig. 4 where a zero day corresponds to false alert days. One can see that the false alerts are generated on the 4th day after Ар maximum and they can be connected with such character of magnetic planetary disturbances. However, if we construct the picture of behaviour Ар (using the statistical superposed method) for all considered earthquakes (Fig. 5), then it is seen that against the background of quasi-periodic variations of Ар the event of earthquakes falls on the second day after the maximum of Ар index.


459

Thus, false alert days do not differ from events of earthquakes from the point of view of geomagnetic disturbances. So, we think that geomagnetic disturbances cannot be used to explain the false alert days. At the same time we see that the event of earthquakes were perhaps closely connected to geomagnetic disturbances.

Fig.4. The behaviour of average values of Ар index in the previous days to the moments of 3 false alerts.

Fig.5. The behaviour Ар (by the statistical superposed method) for all considered earthquakes.

5. Conclusions Results of consideration of amplitude variations of VLF-signal of the storm nature which are passing

above areas of earthquakes with magnitude of more than 5, show that despite non-stationary state of their sources, their amplitude drops 3-6 days before earthquakes with its subsequent restoration by the day of events. The similar picture of behaviour of average atmospherics amplitude has been observed for a significant part of events of strong earthquakes. The effect is similar to signal amplitude variations of low-frequency radio stations.

It allows us to consider such behaviour as typical and try to formulate one of the possible primitive algorithms for the detection of earthquake precursor. Such an algorithm has been checked on the Kamchatka peninsula region by data of atmospherics received in Yakutsk for winter periods of 2004-2006. A control "blindly" retrospective analysis using of the elementary algorithm of defining alert days by average atmospherics amplitude has shown that the probability of such short-term forecast of earthquakes is 60-70 %.

References Gokhberg, M. B., Gufeld, I. L., Rozhnoi, A. A., Marenko, V. F., Yampolshy, V. S., and Ponomarev, E. A. (1989), Study of seismic influence on the ionosphere by superlong wave probing of the Earth-ionosphere waveguide. Phys. Earth Planet. Inter., 57, 64–67. Liu J.Y., Chen Y.I., Pulinets S. A., Tsai Y.B., and Chuo Y.J. (2000), Seismo-ionospheric signatures prior to M ≥ 6 Taiwan earthquakes. Geophys. Res. Lett., 27, 3113-3116. Molchanov, O. A., and Hayakawa, M. (1998), Subionospheric VLF signal perturbations possibly related to earthquakes. J. Geophys. Res., 103, 17489–17504. Pulinets S. A., Gaivoronska T. B., Contreras A. Leyva, and Ciraolo L. (2004), Correlation analysis technique revealing ionospheric precursors of earthquakes. Natural Hazards and Earth System Sciences, 4, 697–702. Rozhnoi, A., Solovieva, M. S., Molchanov, O. A., and Hayakawa, M. (2004), Middle latitude LF (40 kHz) phase variations associated with earthquakes for quiet and disturbed geomagnetic conditions, Phys. Chem. Earth, 29, 589–598. Silina A. S., Liperovskaya E. V., Liperovsky V. A., and Meister C.-V. (2001), Ionospheric phenomena before strong earthquakes. Natural Hazards and Earth System Sciences, 1, 113–118. Soloviev, O.V., Hayakawa, M., Ivanov, V.I., and Molchanov, O.A. (2004), Seismo-electromagnetic phenomenon in the atmosphere in terms of 3D subionospheric radio wave propagation problem, Phys. Chem. Earth, 29, 639–647.


460

ANOMALOUS EXCITATION OF SCHUMANN RESONANCES

ASSOCIATED WITH HUGE EARTHQUAKES, CHI-CHI (CHINA, 1999)

NIIGATA-CHUETSU (JAPAN, 2004), NOTO-HANTOU (JAPAN, 2007),

OBSERVED AT NAKATSUGAWA IN JAPAN

K. Ohta1, J. Izutsu2, K. Furukawa1, M. Hayakawa3 1Department of Electronic & Information Engineering, Chubu University, Kasugai Aichi 487-8501

Japan, e-mail: [email protected] 2Research Center of Earth Watch Safety Net, Chubu University, Kasugai Aichi 487-8501 Japan,

e-mail: [email protected] 3Department of Electronic Engineering, The University of Electro-Communications, Chofu, Tokyo

182-8585 Japan, e-mail: [email protected] Abstract. There are two kinds of methods to estimate a precursory signature of earthquakes by means of electromagnetic waves. The direct method is measuring emissions from a hypocentral region. An example of this method is ULF/ELF electromagnetic emission because its propagation loss in the ground is very low. Another method can be called “indirect method”, by measuring the change in amplitude and/or phase of VLF - VHF subionospheric propagation due to the change in the ionosphere disturbed by the energy from the epicenter. VLF/VHF signals are used in order to study the anomaly in their propagation. Observing and analyzing these two methods (direct and indirect) are very important to prove the generation mechanism of electromagnetic anomalies associated with earthquakes. We have carried out the ULF/ELF observation at Nakatsugawa in Japan (35°25’ N, 137°32’ E in Gifu Prefecture near the foot of Mt. Ena). In this station, we have observed ULF/ELF electromagnetic emissions as the “direct method”, and we have observed the Schumann resonance as the “indirect method”. We have observed and analyzed anomalous excitation of Schumann resonances possibly associated with the 1999 Chi-Chi earthquake (M7.6) in Taiwan. After this earthquake, we observed anomalous excitations of Schumann resonances before large earthquakes, Mid-Niigata Prefecture earthquake (2004, M6.8), Noto-Hantou earthquake (2007, M6.9) in Japan. We present further experimental results on these excitations in this paper. 1. Introduction

The research of Schumann resonance as a precursor of the earthquake has been reported in 1980’s (Schumann, 1952, Maki and Ogawa, 1983). To observe and analyze seismogenic effects by two methods (direct and indirect) is very important to elucidate the generation mechanism of electromagnetic anomalies associated with earthquakes. We have carried out the ULF/ELF observation at Nakatsugawa in Japan (Geographic lat. 35°25’ N, long. 137°32’ E in Gifu Prefecture near the foot of Mt. Ena). In this station, we have observed ULF/ELF electromagnetic emissions as the “direct method (Hayakawa 1999, Molchanov et al., 2001)”, and we have observed the intensity changes of Schumann resonance as the “indirect method (Hayakawa et al., 1996, Ohta et al., 2000)”. We observed the anomalous electromagnetic emissions before large earthquakes at the Nakatasugawa station (Ohta et al., 2006). We have observed anomalous excitation of Schumann resonances before the 1999 Chi-Chi Earthquake (M7.6, depth 10km), main shock occurred at L.T. 02:47 (L.T.=U.T.+9) on September 21, 1999 (Hayakawa et al., 2005), and have performed a statistical study on the anomalous effects in Schumann resonances in Nakatsugawa for large earthquakes in Taiwan (Ohta et al., 2006). In this paper we will present further observational results on abnormal effects in Schumann resonances and additional resonances at Nakatsugawa for a few large earthquakes in Japan


461

2. Observation system and excitation of Schumann resonances on the 1999 Chi-Chi earthquakes We have three orthogonal induction coils (1.2 m permalloy) as magnetic sensors and we observe

simultaneously three magnetic field components (Bx: north-south direction, By: east-west direction and Bz: vertical direction). The signals detected by the sensors are amplified by means of pre-amplifiers with the low pass filter (with cutoff of 30 Hz) and they are fed to main-amplifiers. Then the signals are digitized by means of an A/D converter with sampling frequency of 100 Hz, and they are stored on a hard disc every six hours. The details of our ULF/ELF observation system have already been described in Ohta et al. (2001). Signal analysis is based on the FFT with the data length of 1024, so that the temporal resolution is about 10 seconds and the corresponding frequency resolution is about 0.1 Hz. We can measure the amplitude ratio and phase difference among the three components in the frequency range from 0Hz to 50 Hz. In this observation system, we can estimate from the phase difference between Bx and By whether the observed signal has been propagated over the long distance or short distance. For example, if the phase differences of electromagnetic signals are concentrated around ±180° or 0°, we can understand that these signals have been propagated over longer distances as compared with their wavelength. We can also estimate the direction of electromagnetic signals from the ratio of amplitudes of Bx and By (Hayakawa and Ohta, 2006). However, our induction coils are short (1.2 m), so that there should be direction errors (up to 10°) due to some inaccuracy in installations of the coils. Schumann resonance is the global electromagnetic resonance phenomenon excited by lightning discharges in the cavity formed by the Earth surface and the ionosphere (Schumann, 1952; Nickolaenko and Hayakawa, 2002). The frequency of the fundamental mode is approximately 8 Hz (fundamental mode: n=1) and the frequencies of higher modes are approximately 14 Hz (second mode: n=2), 20 Hz (third mode: n=3), 26 Hz (fourth mode: n=4), and so on. The frequencies of these modes are known to be so stable that the fluctuation is only about 0.2 Hz because of the propagation characteristics and conditions (Nickolaenko and Hayakawa, 2002).

Figure 1. A sonogram of By on September 1-30 in 1999.

Figure2. The change of intensity of n=4 （26.36 - 26.56Hz） before and after earthquakes.


462

But the intensity of each resonant mode depends on the intensity of lightning activities and the distance between the lightning source and the observatory (Hayakawa et al., 2005). Figure 1 illustrates an example of the dynamic spectra (sonograms) of By components during September 1 through October 1, 1999. The vertical axis is the frequency from 0 Hz to 50 Hz, and the intensity is plotted in color. Figure 2 shows the change of intensity of n=4 (26.36-26.56 Hz) before and after Chi-Chi earthquakes in September 1999. The rise-up of the intensity (n=4) seems precursor phenomena for huge earthquakes. The earthquakes occurred near Taiwan area (lat 20.5°N-26.0°N, long. 119.0°E-122.5°E) with magnitude greater than M5 from 1999 to 2004 were 29 events in number. The earthquakes occurred in land were 7 events, and these 7 earthquakes accompanied precursor phenomena, excitation of Schumann resonances. On the other hand the earthquakes in sea were 22, and only 2 events accompanied excitation of Schumann resonances. Among these 2, one is the biggest earthquake, and another one is the nearest from the land. 3. Anomalous resonance before the 2004 Mid-Niigata Prefecture earthquake

The 2004 Mid-Niigata Prefecture earthquake (M6.8, depth 20km) occurred at 17:56 L.T. on October 23, 2004. The anomalous resonance was observed during the period from the afternoon on October 20 to the morning on October 22. Figures 3 shows the largest anomalous resonance on the By and the Bx components, respectively, observed at 18:00 – 23:55 L.T. on October 20, 2004, three days before the earthquake. There are two strong resonances in the n=3 frequency range in both By and Bx components. In both components, the intensity of these two resonances changes simultaneously.

Firstly, we have analyzed the strong resonance (upper one of the two strong resonances in Fig. 3, Marked with “n=3”) of the By component. We have tallied the strongest intensity (dB) and the frequency (our frequency resolution is 0.097 Hz) at that time at every period (our temporal resolution is 10.24 sec) during the six hours in Fig. 3. The maximum intensities are located at 20.60 Hz, 20.70 Hz, and 20.80 Hz. The intensity is gradually attenuated by the increase and decrease of the frequency from these median frequencies.

Fig. 3. Excitation of Schumann resonances at 18:00-23:55 on Oct. 20, 2004 (By component).

The 2004 Mid Niigata Prefecture earthquake (M6.8, depth 20km) occurred at 17:56 L.T. on Oct. 23, 2004.

050

100150200250

-1

80

-1

50

-1

20

-9

0

-6

0

-3

0 0

30

60

90

12

0

15

0

18

0

Phase di ffe rence (deg. )

Oc

cu

rre

nc

e n

um

be

r

Figure 4. Occurrence number of phase difference between By and Bx at 20.60Hz-20.80Hz.


463

We have analyzed further the characteristics of emissions at these median frequencies (20.60 Hz, 20.70 Hz, and 20.80 Hz). Figure 4 is the histogram of phase differences between By and Bx at those median frequencies. The phase differences are centered around ±180°, therefore we understand that this resonance is a linearly polarized wave and propagated over longer distances (Hayakawa et al., 2007). So, we can conclude this resonance is identified as Schumann resonance (n=3).

020406080

100

-1

80

-1

50

-1

20

-9

0

-6

0

-3

0 0

30

60

90

12

0

15

0

18

0

Phase di ffe rence (deg. )

Oc

cu

rre

nc

e n

um

be

r

Fig. 5. Occurrence number of phase difference between By and Bx at 16.11Hz，16.21Hz, and 16.30Hz.

shown “A” in Fig. 3 Next, we have analyzed another strong resonance (lower one, marked with “A”) in the same way. Figure 5 illustrates the distribution of the frequency with the maximum intensity of the lower strong resonance shown in Fig. 4. The median frequencies of maximum intensity of this resonance are 16.11 Hz, 16.21 Hz, and 16.30 Hz, and these frequencies are about 2 Hz higher than the typical frequency of n=2 (14 Hz). Figure 5 is the histogram of phase differences between By and Bx at these median frequencies (16.11 Hz, 16.21 Hz, and 16.30 Hz), which exhibit a nearly flat distribution over the phase difference and which is completely different from Fig. 4. Since this resonance has a different median frequency and a different polarization from the conventional Schumann resonance, it seems that this resonance might be due to a generation mechanism different from the Schumann resonance.

The phase difference has almost a homogeneous distribution over the phase difference. We can conclude that this resonance might be generated by a mechanism different from the usual Schumann resonance. Therefore it is considered that they are likely to be generally locally or relatively close to the observation station and might be due to a different generation mechanism from Schumann resonance. However, it is very interesting that the temporal changes of intensity of resonances are almost the same between the anomalous Schumann resonance (n=3) and another anomalous resonance. 4. Anomalous resonance before the 2007 Noto-Hantou earthquake

Figure 6. Excitation of Schumann resonances at 06:00-11:55 on 25 Mar., 2007.

The 2007 Noto Hantou earthquake (M6.9, depth 11km) occurred at 09:41 L.T. on March 25, 2007.


464

The anomalous resonance was observed during the period from the afternoon on March 24 to the morning on March 31.

Figures 6 shows the largest anomalous resonance of the Bx component observed at 06:00 – 11:55 L.T. on March 25, 2007, just before and after the earthquake. There is a strong anomalous resonance in Bx component marked with “C”, but there is no obvious anomaly in By component differently from the case of the 2004 Mid-Niigata Prefecture earthquake. Therefore we make a histogram of phase differences between By and Bx at the all frequency ranges from 19.53 Hz to 21.48 Hz. It is just the same as Figure 4. We can conclude this resonance is usual Schumann resonance (n=3). Next we analyzed the distribution of the frequency with the maximum intensity of anomalous strong resonance of Bx component, marked with “C”. The median frequencies of maximum intensity are 18.06 Hz, 18.16 Hz, and 18.26 Hz, which are about 2 Hz lower than the typical frequency of Schumann resonance (n=3), being the same as the case of the 2004 Mid-Niigata Prefecture earthquake. The phase difference between By and Bx at the median frequencies (18.06 Hz, 18.16 Hz, and 18.26 Hz) is the same as Fig.5. As mentioned above, the temporal changes of the intensity of the anomalous resonances are almost the same between the anomalous Schumann resonance (n=3: 20 Hz) and another anomalous resonance (16.21 Hz and 18.35 Hz). Let us suppose that the intensity change of these anomalous resonances (20 Hz, 16.21 Hz, and 18.35 Hz) was precursor of the earthquake and caused by the propagation anomaly in the ionosphere and/or atmosphere disturbed by any energy source from the epicentral region.

In the case of the 2007 Noto Hantou earthquake, the third mode of the Schumann resonance is not so strong and the bandwidth is broad. This Schumann resonance arrived from the broad direction centered at 20° - 25° by goniometer method. On the other hand, the anomalous resonance at the frequency centered at 18.16 Hz has a strong intensity and a narrow bandwidth, and its resonant frequency is 2 Hz lower than the conventional frequency of n=3. This anomalous resonance is found to be very similar to the anomalous resonance observed before the 2004 Mid-Niigata Prefecture earthquake (lower strong resonance shown in Figure 3). However, the generation mechanism of this anomalous resonance is quite unknown and there is no convincing theory at the moment to explain both cases of the 2004 Mid-Niigata earthquake and the 2007 Noto Hantou earthquake. However, there exists one possible generation mechanism of these anomalous ULF-ELF resonances (Sorokin et al., 2003). 5. Conclusion We have observed the anomalous excitation of the Schumann resonances possibly associated with earthquakes since 1999 at Nakatsugawa station in Gifu prefecture in Japan (Hayakawa, et al., 2005; Ohta et al., 2006). In this paper we analyzed the anomalous Schumann resonance and an additional anomalous resonance observed before the 1999 Chi-Chi earthquake, the 2004 Mid-Niigata Prefecture earthquake and the 2007 Noto Hantou earthquake. The characteristics of anomalous resonances are: 1. The intensity of a particular mode of the Schumann resonance increased before the large earthquake near

the observation station, and decreased after the occurrence of earthquake. 2. An excitation of another anomalous resonance was also observed at the frequency shifted by about 2 Hz

from the typical frequency of the Schumann resonance. This anomalous resonance had a high Q factor and a strong intensity.

3. Since the temporal changes of the intensity of the anomalous Schumann resonance and another anomalous resonance were almost the same, there is a possibility that another anomalous resonance was closely related with the Schumann resonance. However, we need to consider any convincible generation mechanism of anomalous resonances in future.

4. There is a possibility that another anomalous resonance was received at Nakatsugawa as the induced magnetic field from the epicentral region (Sorokin et al. 2003). However, these anomalous resonances


465

were always detected only at one component (Bx component or By component). To solve these characteristics of anomalous resonances, the multipoint simultaneous observation is absolutely necessary. We have established new observation stations at Shinojima in Aichi prefecture and at Minami-Izu in Shizuoka prefecture and have already started the preliminary observation. We expect to acquire new information by our new observation network in near future in order to have better understanding of the anomalous seismogenic effect. Acknowledgements The observation of the electromagnetic anomalies associated with earthquakes in Chubu University is supported by “Academic Frontier” Project for Private Universities: matching fund subsidy from Ministry of Education, Culture, Sports, Science and Technology (2006-2010). We would like to express acknowledgements to concerned members and organizations. The observation at Nakatsugawa is partially supported by a scholarship of Sanwa Technos Corporation to which we like to express great thanks. One of the authors (M. H.) is grateful to NiCT for its support (R&D promotion scheme funding international joint research). References Hayakawa, M., K. Hattori, and K. Ohta (2007), Monitoring of ULF (ultra-low-frequency) geomagnetic

variations associated with earthquakes, Sensors, 7, 1108-1122. Hayakawa M., and K. Ohta (2006), The importance of direction finding technique for the study of VLF/ELF

sferics and whistlers, IEEJ Trans. FM, 126, 2, 65-70. Hayakawa M., K. Ohta, A.P. Nickolaenko, and Y. Ando (2005), Anomalous effect in Schumann resonance

phenomena observed in Japan, possibly associated with the Chi-chi earthquake in Taiwan, Ann. Geophysicae, 23, 1335-1446.

Hayakawa, M., and O. A. Molchanov (Editors) (2002), Seismo Electromagnetics: Lithosphere -Atmosphere-Ionosphere Coupling, TERRAPUB Tokyo, 477p.

Hayakawa, M. (Editor) (1999), Atmospheric and Ionospheric Electromagnetic Phenomena Associated with Earthquakes, TERRAPUB Tokyo, 996p.

Hayakawa M., O. A. Molchanov, T. Ondoh, and E. Kawai (1996), The precursory signature effect of the Kobe earthquake on VLF subionospheric signals, J.Comm.Res.Lab., 43, 413-418.

Maki K., and T. Ogawa, (1983), ELF emission associated with earthquakes, Res. Lett. Atmos. Electr., 3, pp. 41-44.

Molchanov, O.，A. Kulchisky, and M. Hayakawa (2001), Inductive seismo-electromagnetic effect in relation to seismogenic ULF emission, Natural Hazards and Earth System Sciences, 1, 61-67.

Nickolaenko, A. P., and M. Hayakawa (2002), Resonances in the Earth-ionosphere Cavity, Kluwer Acad. Pres, Dordrecht, 380.

Ohta K., K. Makita, and M. Hayakawa (2000), On the association of anomalies in subionospheric VLF propagation at Kasugai with earthquakes in the Tokai area, Japan, J. Atmos. Electr., 20, 85-90.

Ohta K., K. Umeda, N. Watanabe, and M. Hayakawa (2001), ULF/ELF emissions observed in Japan, possibly associated with the Chi-Chi earthquake in Taiwan, Natural Hazards and Earth System Sciences, 1, 37-42.

Ohta K., N.Watanabe, and M.Hayakawa (2006), Survey of anomalous Schumann resonance phenomena observed in Japan, in possible association with earthquakes in Taiwan, Physics and Chemistry of Earth, 31, 397-402.

Schumann W. O. (1952), On the free oscillations of a conducting sphere which is surrounded by an air layer and an ionosphere shell, Z. Naturforschaftung, 7a, 149-154.

Sorokin V. M., V. M. Chmyrev, and A. K. Yaschenko (2003), Ionospheric generation mechanism of geomagnetic pulsations observed on the Earth’s surface before earthquake, J. Atmos. Solar-terr. Phys., 64, 21-29


466

VARIATIONS OF VLF SIGNALS RECEIVED ON DEMETER SATELLITE

IN ASSOCIATION WITH SEISMICITY

A. Rozhnoi1, M. Solovieva

1, Molchanov O.

1

1Institute of the Earth Physics, RAS, Bolshaya Gruzinskaya 10, Moscow, Russia,

e-mail:[email protected]

Abstract. We present two methods of the global ionosphere diagnostics using VLF signals received

on board the DEMETER satellite in association with two cases of strong seismic activation. The

method of reception zone changes reveals an evident effect before and during the great Sumatra

earthquake with long-time duration of the order of one month. The result leads to the conclusion on

the size of perturbation area of the order of several thousands kilometers. Disadvantage of this

method is in its difficulty to separate preseismic and postseismic effects. In contrast, the difference

method allows us to overcome this difficulty and it shows the appearance of preseismic effect for

several days for seismic activation near Japan. However, we need such an analysis for reliability in

regular satellite data to be checked by ground reception of subionospheric VLF signals. It is not so

obvious that both of these satellite and ground effects are excited by the same generation

mechanism on the ground. So, these look as complementary.

1. Introduction

During the last 10-20 years there have been attracted noticeable attention in the effects of ionospheric plasma

related to seismicity, with keeping in mind both possibilities to use them for earthquake forecast and to study

the fundamental problem of lithosphere-ionosphere coupling. There are two directions of this research. The

first is in-situ observation of those effects, i.e. satellite observation. Numerous papers have already been

published on such observations (see the reviews by Parrot et al., 1993; Molchanov et al., 2002). The second

direction is far-distant remote sounding of the ionospheric perturbations in connection with seismic events by

means of electromagnetic signals. Results of VLF sounding in different frequency bands have been

published in many papers (Gufeld, et al., 1992; Hayakawa, et al., 1996; Molchanov and Hayakawa, 1998;

Rozhnoi et al., 2004; Horie et al., 2007).

We are going to discuss here the reception of VLF transmitter signals on board the DEMETER satellite.

Such a reception was undertaken on many satellites for the investigation of VLF wave propagation and VLF

wave interaction with ionospheric plasma (e.g. Inan and Helliwell, 1982; Molchanov, 1985). However, in the

application of VLF signals to long-time seismic effects we need a special data processing both on board the

satellite and on the ground in addition to the reception itself. Therefore it can be considered as a new method

of ionospheric sounding in association with seismicity.

2. Satellite VLF Data

Here we will discuss only the data of electric field from a powerful Australian NWC (19.8 kHz) transmitter.

As the main characteristic of a VLF signal, we compute the signal to noise ratio (SNR) as follows:

SNR=<A0>/Amin, where <A0> is an average amplitude spectrum density in the frequency band including the


467

transmitter frequency F0 and Amin is the minimum value outside of the signal band..Due to large longitudinal

distances between adjacent orbits (about 2500 km at the middle latitudes) we need the averaging period of, at

least, 3-4 weeks in order to obtain statistically significant results and longitudinal spatial resolution of

100-200 km. It dictates a selection of rather long periods of seismic activity related to very strong

earthquakes or to a series of large earthquakes. We need to average the signal over fast variations and try to

seek for slow changes in the reception zones during seismically active periods. We discuss two methods for

revelation of the seismic influence: analysis of changes in reception zones for finding the large-scale space

variation and analysis of SNR differences for finding of temporal variation during seismic activation.

3. Change in reception zone

As an example we analyse the reception zone of the NWC transmitter in relation to a strong earthquake

activity near Sumatra which started with a great seismic shock on December 26, 2004 (magnitude M = 9).

There were several strong shocks with following weaker ones until January 3, 2005 and the aftershock series

continued until the summer of 2005 with new explosions of activity on March 28, 2005 (M=8.7, so-called

Sumatra-2 earthquake) and on July 24, 2005 (M= 7.5). Fig. 1 shows the area of Sumatra activity and two

reference zones with rather weak seismic activity during this period.

Fig. 1. Seismic activity during the period from October 1,

2004 to December 31, 2005. A triangle indicates the place

of NWC transmitter, positions of the strong earthquakes

with M> 6.6 are shown by solid circles, and the area of

Sumatra activity is outlined by an open circle. Reference

zones with rather weak seismic activity are outlined by

circles with the numbers 2 and 3.

We use DEMETER data during the period of more than

one year from October 24, 2004 to December 31, 2006. In

order to find a possible EQ influence we divide the

observation period into the intervals with duration of 30

working days, taking into account the missing days and

keeping about the same volume of recording points. Then

we compare the reception in the different time intervals

both in the Sumatra area and in the reference zones. In

such a way some seasonal variations can be expected. So we computed values <SNR> averaged over the

central part of the NWC transmitter reception zone for each month from October 2004 to August 2006 and

found a small seasonal variation of the order of 25% with an increase in the winter time. Therefore we

analyse the normalized SNR values: Sn = SNR/<SNR> and calculate the ratio p = N(Sn 1)/N0, where N0 is

the total number of recording points and N(Sn 1) is the number of points with increased value Sn 1 inside

the Sumatra area or inside the reference zones. The result is presented in Fig. 2. We conclude an essential

depression of VLF signal intensity during the period of November-December 2004 only above the Sumatra


468

area, which is probably connected with the preparation of seismic activity in this area.

4. Difference method

We use the method for correlated analysis of VLF subionospheric signals from transmitters located in Japan

(JJY and JJI) and in Australia (NWC) that were measured at the ground stations and on board the

DEMETER satellite. So we develop the satellite data processing which is similar to the processing of data at

the ground station (Rozhnoi et al., 2004).

Fig.2. Ratio of reception points with strong signals

(Sn 1 ) to the total number of recording points

inside the Sumatra area (second panel from

above) and inside the reference zones (shown in

Fig.1, first and third panels) together with

magnitude difference (M-7) of the great

earthquakes inside the Sumatra area (forth panel)

and seismic energy released (fifth panel).

For ground observation we use a residual signal of

phase dP or amplitude dA defined as the difference

between the observed signal and the average of

quiet days of the current month. For satellite

observation we calculate a “reference surface”

(model) over the region of interest as a function of

longitude and latitude. In this study, a simplified

approach to compute this surface was used. The

method consists in averaging all the data available

in the considered region, regardless of the global

disturbances, in particular, of the magnetic activity.

The length of this period was selected equal to 2

months in order to be not affected by the seasonal variations. The method of the local polynomial

interpolation was used to build the reference surface with a longitude and latitude resolution of 0.32°. Using

the reference surface, at any time and for any longitude and latitude, it is possible to define the residial VLF

signal as the difference between the measured amplitude S (t, longitude, latitude) and the reference value Sm

(longitude, latitude).

We present here a comparison of satellite and ground reception during July-September 2005. There were

several large EQs (M 6.2) at the area of Japan in this period, which occur inside or close to the sensitivity

zone of our ground reception and inside of the satellite reception zone for NWC transmitter (19.8 kHz).

Positions of these EQs are shown in Figure 3a for the ground reception and in Figure 3b for satellite

reception (right rectangle). Additionally for the control of satellite reception we select another rectangle that

is also inside of the reception zone but it is free from seismic activity (left rectangle in Fig. 3b).


469

A comparison of results of the satellite and ground observations is presented in Fig.4. Here we show the

differences averaged over night time for the ground reception and differences averaged along the orbit

crossing the seismic or reference area depicted in Fig. 3. By paying attention to the hatched regions, we

notice an evident decrease in VLF signal both on the ground and on the satellite in association with

seismicity.

Fig. 3 (a) Map of earthquakes with M≥6.0 during the period of July-September, 2005. Sensitivity zones for

the ground reception of Japan transmitters are shown by red line, NWC and NPM transmitters - by dash

lines; (b) Reception zone of the NWC transmitter during the period of July-September, 2005 (model)

registered on the satellite DEMETER together with the positions of large EQ epicenters. Red circles – EQ in

sensitivity zone of wavepath JJY-PTK. The selected area for the analysis of VLF signal reception above

Japan is outlined by A) rectangle on the right, and the control rectangle B) that is free from earthquake

activity is shown on the left.

5. Discussion

We have presented two methods of the global ionosphere diagnostics using VLF signal received on board a

satellite in association with two cases of strong seismic activation. The method of changes in reception zone

reveals an evident effect before and during great Sumatra earthquakes with long-time duration of the order of

one month. The result leads to the conclusion on the size of perturbation area of the order of several

thousands kilometers, which is found to be in good agreement with the suggestion by the ground-based

subionospheric VLF propagation (NWC-Japan path) (Horie et al., 2006). The disadvantage of this method is

in its difficulty to separate preseismic and postseismic effects. In contrast, the difference method allows us to

overcome this difficulty and shows the appearance of significant preseismic effect for several days for

seismic activation near Japan. However, we need such an analysis for reliability in regular satellite data and

in the check by ground reception of subionospheric VLF signal. It is not obvious that both the effects are

excited by the same generation mechanism at the ground, so that those methods are complementary to each

other.

In general the methods of diagnostics discussed here have perspective to be global due to the world-wide

(a) (b)


470

positioning of powerful VLF transmitters and satellite reception. However, they have specific disadvantage

because they require a rather long time period of analysis due to large longitudinal distances between

satellite orbits. Above a fixed area in the same local time the satellite appears only once a day. As the result,

we need, at least, one month period of registration for the longitudinal spacing of about 1000km.

Fig.4. VLF signal differences in the ground observation for the wave paths: JJI-Kamchatka, JJY-Kamchatka,

NWC-Kamchatka and NPM-Kamchatka and averaged satellite VLF signal differences observed on board the

DEMETER from the reception of NWC transmitter: the solid line for the data above Japan, and dash-dot line

for the data aside of Japan area (see Fig. 3b). Two panels below are Dst variations and EQ magnitude

values.

As a mechanism of the observed effects, we suggest the following:

- Of course such a long-time and large-scale perturbation in the ionosphere cannot be produced by a seismic

shock itself (duration of minutes) and we need to suppose some long-lasting agent, which influences the

ionosphere around the date of an earthquake.

- We believe that this initial agent is an upward energy flux of atmospheric gravity waves (AGW) which are

induced by gas-water release from earthquake preparatory zone (e.g. Liperovsky et al., 2000; Molchanov,

2004).

- Penetration of AGW waves into the ionosphere leads to the modification of natural (background)

ionospheric turbulence, especially for space scales ~ 1-3 km and wave numbers kT~ 10-4

-10-3

m-1

. This

weak but reliable effect is revealed from direct satellite observations (Molchanov et al., 2004; Hobara et al.,

2005).

- Resonant scattering of the VLF signals is possible on the following condition of the frequency- wave


471

number synchronism : 0= s + T , k0 = ks+ kT, where 0, k0 are for the incident VLF wave, T,

kT are for the turbulence and s, ks are for the scattering waves. It can be found that the amplitude of

incident wave A0 decreases exponentially during the course of propagation through the perturbed medium:

A0~ exp(- nATH), where n is the coefficient of nonlinear interaction, and H is the length of interaction

region. In our case of VLF signals: T<< 0 ~ s, and the interaction is especially efficient because k0~ ks~

kT (Molchanov, 1985). Therefore, even though the amplitude of the turbulence AT is small, the scattering

could be significant if the length H is large.

References

Gufeld, I.L., A.A. Rozhnoi, S.N. Tyumensev, S.V. Sherstuk, and V.S. Yampolsky (1992), Radiowave

disturbances in period to Rudber and Rachinsk earthquakes,. Phys. Solid Earth, 28 (3), 267– 270.

Hayakawa, M., O. A. Molchanov, T. Ondoh, and E. Kawai (1996), The precursory signature effect of the

Kobe earthquake on subionospheric VLF signals, J. Comm. Res. Lab., Tokyo, 43, 169–180.

Hobara, Y., F. Lefeuvre, M. Parrot, and O. A. Molchanov (2005), Low latitude ionospheric turbulence and

possible association with seismicity from satellite Aureol 3 data, Ann. Geophys., 23, 1259-1270.

Horie, T., S. Maekawa, T. Yamauchi, and M. Hayakawa (2007), A possible effect of ionospheric

perturbations for the Sumatra earthquake, as revealed from subionophseric very-low-frequency (VLF)

propagation (NWC-Japan), Int. J. Remote Sensing, 13, 3133-3139.

Inan, U.S., and R.A. Helliwell (1982), DE-1 observations of VLF transmitter signals and wave-particle

interaction in the magnetosphere, Geophys. Res. Lett., 9, 917-23.

Liperovsky, V.A., O.A. Pokhotelov, E.V. Liperovskaya, M. Parrot, C.V. Meister and T. Alimov (2000),

Modification of sporadic E-layers caused by seismic activity, Surveys in Geophysics, 21, 449-486.

Molchanov O.A. (1985), VLF Waves and Iinduced Emissions in the Near-Earth Plasma, Moscow, “Nauka”,

223 p.

Molchanov, O. A, M. Hayakawa, V. V. Afonin, O. A. Akentieva, and E. A. Mareev (2002), Possible influence

of seismicity by gravity waves on ionospheric equatorial anomaly from data of IK-24 satellite 1. Search

for idea of seismo-ionosphere coupling, in Seismo Electromagnetics:

Lithosphere-Atmosphere-Ionosphere Coupling, Edited by Hayakawa, M. and Molchanov, O., Terrapub,

275–285.

Molchanov, O. A., O. S. Akentieva, V. V. Afonin, E. A. Mareev, and E. N. Fedorov (2004), Plasma

density-electric field turbulence in the low-latitude ionosphere from the observation on satellites; possible

connection with seismicity, Phys. Chem. Earth, 29, 569– 577.

Molchanov, O. A. (2004), On the origin of low- and middle-latitude ionospheric turbulence, Phys. Chem.

Earth, 29, 559–567.

Parrot, M., J. Achache, J.J. Berthelier, E. Blanc, A. Deschamps, F. Lefeuvre, M. Menvielle,J.L. Planet, P.

Tarits, and J.P. Villain (1993), High-frequency seismo- electromagnetic effects, Phys. Earth Planet. Inter.,

77, 65-83.

Rozhnoi, A., M. S. Solovieva, O. A. Molchanov, and M. Hayakawa (2004), Middle latitude LF (40 kHz)

phase variations associated with earthquakes for quiet and disturbed geomagnetic conditions, Phys. Chem.

Earth, 29, 589–598.


472

SEISMIC TRAVELLING IONOSPHERE DISTURBANCES AT F2-REGION IN TIME OF HELIOGEOPHYSICAL DISTURBANCES

N.P. Sergeenko, M.V. Rogova, A.V. Sazanov

Pushkov Institute of Terrestrial magnetism, Ionosphere and Radio waves propagation,

Troitsk, Moscow region, 142190, Russia, e-mail: [email protected]

Abstact. The properties of travelling ionospheric disturbances (TID) (their horizontal sizes are 1~ 4 thousand kms, excesses from the background are 15-30 %), formed in F2 layer as a result of preparation of the strong earthquake in time of heliogeophysical perturbations are explored. The inhomogeneities arise 10-15 h before earthquakes and move horizontally with a transonic speed on distances of a few thousand kms up to round – the - world trajectories focused approximately along an arc of a major circle, transmitted above an epicenter region. Select of macro-scale irregularities was caused by spatial - time differences of dynamics of F2-layer of the ionosphere in time of the heliogeophysical disturbances from dynamics of quasi-causative seismic macroscale ionospheric inhomogeneities. The complex analysis of the arrays of critical frequencies of F2 - layer of the ionosphere (foF2) of a world network of automatic ground ionospheric stations of vertical sondage was carried out. As a result of these examinations, the geophysical patterns of ionospheric effects incipient at superimposition of macroscale inhomogeneities and ionospheric disturbances of solar and magnitospheric origin were synthesized.

INTRODUCTION

This work is devoted to the study of the possibilities of detection of traveling disturbances (TIDs) of

seismic origin in the main peak of ionosphere at the time of heliogeophysical disturbances. The sizes of irregularities reach 1~3 thousand kms, lifetimes exceed 104 sec, and moving distances are 10~15 thousand kms, the speeds are approximate to sonic velocity. It was also established that the formation of TID precedes 10~15 hrs to the moment of catastrophic earthquakes with magnitude М ≥ 6 and localization ~1 thousand kms from the epicenter of the earthquake (Kalinin and Sergeenko, 2002; Kalinin et al, 2003; Sergeenko and Kharitonov, 2005). Perturbations of electron concentration, commensurable in sizes and contrast range with seismic TIDs, but completely different character of temporal and spatial changes, can appear on the conditions of ionospheric storms and substorms. Below we review features of occurrence of TID in quiet and disturbed heliogeophysical conditions.

DATA ANALYSIS

Files of values of relative variation of critical frequencies of F2 layer of the ionosphere by the global network of ionospheric stations of vertical sounding δfoF2 were used in the analysis as a main source of information about seismic TID: δfoF2 = (foF2- foF2m)/foF2m , foF2m – moving median [Gaivoronskaya et al., 1971). δfoF2 are used for detection and numerical characterization of ionosphere disturbances (Zevakina et al., 1990). It does not depend on multiplicative errors of representation foF2 due to the use of various scales at stations of vertical sounding, gives "protection" against inadequate influence of major "emissions" and excludes regular seasonal and daily variability.

The seismic TIDs are selected as a local irregularity with extreme value (δfoF2)max ≥15 % at the middle latitudes and (δfoF2)max ≥10 % at low latitudes. It is also supposed that during of 2~3 hours diversion should be more than 10 %. Kp, Dst, and AE indices were used for the description of geomagnetic situation. The quite conditions. In fig. 1a we can see an example of variations δfoF2(t) for different ionospheric stations for the earthquake, which occurred at Alyaska on 09.03.1985 (ϕ=66,6N; λ=150,5W; t0=14h08m UT; M=6,2), a vertical line marks


473

mailto:[email protected]

the moment of the earthquake, an oblique line represents variant of propagation of signals most likely related to a developing earthquake. The appearing disturbances before the earthquake travel to Asia, Indian ocean and up to La Renion ionosphere station. The first signals are registered ∆t = -13 hours ahead of time. Amplitudes of signals are δfoF2(t) > 20 %. According to the data in fig. 1c, positive impulses appeared on quiet geomagnetic background: Kp ≤ 2, AE<200 nT, Dst did not fall below -20 nT.

The Mercator projection of world map is represented in fig. 1b, the location of epicenter is shown by cross. Dark circles show the locations of ionospheric stations, where positive disturbances were observed. The dotted curve shows the trajectory of moving of this disturbance. The trajectory is focused approximately along the part of an arc of a major circle, transmitted above the epicenter region. The light circles show location of the other ionosphere stations, where there were either negative disturbances or no disturbances or no observations in this time for some reasons.

Fig. 1 δfoF2(t) variations at different ionospheric stations for the Alyaska earthquake on 09.03.1985 The disturbance conditions 1. Ionosphere substorm. Fig. 2 represents the data for the Caribbean earthquake that took place on 16.03.1985 (ϕ=17,5N; λ=62,5W; t0= 14h53mUT, M=6,8). Fig.2c shows there was a substorm (AEmax ~ 400 nT) at 14 h UT, 1 hour before the earthquake. This substorm took place on a quiet geomagnetic background, Dst did not fall below -20 nT. The δfoF2(t) data (fig.2 a,b) allows us to draw a common line of delays for all stations altogether. This line gives ∆t = -13 hrs as the estimate time of the appearance of irregularity above the epicenter. The interval of distances is significant here: - 6~17·103 km. Positive impulses appeared on a quiet geomagnetic background. The substorm began approximately 4 hrs later. The trajectory of a motion of seismic disturbances corresponds to a site of an arc of a major circle, directional in opposite sides from the epicenter. 2. Ionosphere storm. Fig. 3 represents data for the Indonesian earthquake on 01.03.1985 (ϕ=1,4S; λ=119,6E; t0=17h11m UT; М=6,4). δfoF2(t) variations are showed for the various stations located in Asia, Europe, Australia,


474

Northen America at the left part of this figure. The variations of geomagnetic indices are presented in fig. 3c which testify that 01.03.1985 the moderate geomagnetic disturbance (Dstmax ~ -50 nT, Kpmax=6) was observed.

Fig.2 δfoF2(t) variations at different ionospheric stations for the Caribbean earthquake on 16.03.1985

The substorm has also taken place at ~5 h UT on disturbance background. The δfoF2(t) data in fig. 3b testify a biphase ionosphere disturbance in the F2 layer. All diagrams contain clear positive perturbations with amplitude 15~30 % and duration 5~7 hrs and in day local time beginning (5 h UT) on the first day of the geomagnetic disturbance. Approximately at 18 h LT the perturbation has transformed in a negative phase at all stations. It is obvious that when observed simultaneously on ionospheric stations located at various latitudes and longitudes the positive phase of ionosphere storm cannot be identified as seismic precursor, though they have happened in "the necessary" time.

However, fig.3а illustrates the δfoF2(t) data on a chain of stations from Manila and Vanimo, posed in a region of forthcoming earthquake, through domestic stations in the Asia and Europe (Irkutsk, Tomsk, Sverdlovsk, Gorky, Moscow), posed approximately 5 thousands kms from the earthquake epicenter, up to north Africa station Ouagadougou. The positive impulses - harbingers of earthquake have appeared in a region of epicenter ~13 hrs before the beginning earthquake. The time delays of their occurrence at the relevant stations specify an apparent velocity of their travel ~1000 kms/hrs. In fig. 3c dotted line indicated a possible direction of a motion of these TIDs.

In fig. 3а the chain of δfoF2(t) impulses with duration 3~4 hrs was observed both in day time, and in evening local time, while the positive phase of a biphase ionosphere disturbance usually occurs only in day time (06~18 hrs LT) and last not less than 7~8 hrs (Zevakina et al., 1990).


475

Fig.3 δfoF2(t) variations at different ionospheric stations for the Indonesian earthquake on 1.03.1985

DISCUSSION

Situations are presented when apparent TIDs in F2 layer of the ionosphere are presumably linked with developing earthquake. We have given examples when they arose in quiet conditions, in conditions of isolated substorm on the backgrounds of quiet ionosphere and during ionosphere storm with positive initial phase. Undoubtedly, the above given examples do not cover entire spectrum of ionosphere variations.

The changes of electron concentration of F2 layer from storm to storm are significant in the sense that two storms never existed with the same behavior of ionosphere parameters. The basic approach to select macro-scale irregularities consists of the use of differences in dynamics of ionosphere variations, characteristic for disturbed heliogeophysical conditions and in dynamics of TIDs, inconvertible under the shape and velocity of a motion, but with the casual moments of occurrence. Identifications of the dynamics of TIDs require processing data for the territories of hundreds of thousand square kms.

Basic differences between TIDs and effects of magnetosphere perturbations in F2 layer are: ♦During an ionosphere substorm electron concentration is incremented as a result of activity of zonal and meridian electric fields as well as under IGW in the day time after preliminary decrease (Brunelly and Namgaladze, 1988). The trajectories of movement of TIDs coincide with the arcs of the major circle, whereas the perturbations, which arise as a result of IGW generation during a substorm, are usually spread from high latitudes to low in a meridian direction. IGW generation is delayed by 1 hr with respect to the AE-index; ♦Ionosphere storm represents a continuous process lasting from several hours to several days. Parameter variations have global character. In the absence of AE outbursts, positive impulses recorded in δfoF2(t) on the background of negative storm phase can be related to seismic TIDs. Positive perturbations of electron


476

concentration can be observed during ionosphere storms at the transitional and low latitudes; as well as at the midlatitudes in the early stages ionosphere storm, if this storm begins at a day time.

CONCLUSION

The above reviewed situations demonstrate that diagnostics of seismic TIDs is quite possible in many

cases, as there are essential differences in the character of time and spatial changes allowing us to distinguish these disturbances from perturbations of solar and magnetosphere origin. However data configurations that are submitted in figs. 1~3, are not possible for all earthquakes. One of the reasons is in actual properties of a network of ionosphere stations - nonuniform spatial arrangement and, naturally, only on land. Other difficulties are the detection of TIDs on the background of other disturbances in the ionosphere. It is represented that this short-term harbinger could be the essential block in blanket algorithm of the seismic-ionosphere forecast. REFERENCES Kalinin, U.K., Sergeenko, N.P. (2002), Mobile solitary macroinhomogeneities appearing in the ionosphere

several hours before catastrophic earthquakes. Doklady of Russian Akademy of Sciences / Earth Science Section, 387(1), 105-107.

Kalinin, U.K., Romanchuk, A.A., Sergeenko, N.P. Shubin, V.N. (2003) The large-scale isolated disturbances dynamics in the main peak of electronic concentration of ionosphere. Journal of Atmospheric and Solar Terrestrial Physics, v.65, issue 11-13, 1175-1177.

Sergeenko, N.P., A.L. Kharitonov (2005) Short-time magnetosphere – ionosphere predictors of catastrophe earthquakes. Investigations of Earth from Space. 6, 61-68. (Russian)

Gaivoronskaya T.V., Sergeenko N.P., Udovich L.A. (1971) Deviations of critical frequencies of F layer from median values, in Ionospheric disturbances and their influence on a radio communication, Moscow, Nauka, 55-73 (Russian)

Zevakina, R.A., N.P., Sergeenko, E.M.Zgulina, G.N. Nosova (1990) Text-book of short-time forecast of ionosphere, 71 pp, Materials of the World Data Center B. (Russian)

Brunelly B.E., Namgaladze A.A. (1988) The Physics of Ionosphere, 527 pp. (Russian)


477

TESTING OF THE METHOD FOR THE CONVERSION OF THE MT APPARENT RESISTIVITY CHANGES INTO THE RELATIVE CHANGES

IN THE ROCK ELECTRICAL RESISTIVITY USING THE 2D MODEL OF THE GEOELECTRIC STRUCTURE

Marina E. SholpoSPbF IZMIRAN, Russian Academy of Sciences, Muchnoy 2, St.Petersburg, 191023, Russia,


Abstract. A method for direct conversion of observed variations in magnetotelluric apparent resistivity ρa into relative variations in the resistivity of elements of a well-studied geoelectric structure is tested on a 2-D model structure of Petropavlovsk geodynamical polygon. It is shown that (1) the study of the frequency and spatial dependences of the sensitivity of ρa to changes in the resistivities of the structure elements allows us to choose the optimal observation regime and (2) the application of the proposed method makes it possible to define relative changes in the rock resistivity with sufficient accuracy.

The MT apparent resistivity is extensively investigated at present as a possible prognostic parameter in many geodynamic research areas. But it is clear that these researches can hardly be effective, if the source of the variations in the apparent resistivities ρa is unknown. If we have in mind that the cause of variations in ρa is the changes in the rock resistivity (that is the seismic- electrical effect of first kind) then the direct relative variations in the resestivity of elements of a geoelectric structure are a more effective prognostic indicator compared to observed variations in the magnetotelluric apparent resistivity. In this connection the method for the separation of the contributions of individual elements of the geoelectric structure to the variations in the apparent resistivity is proposed. It is the method for the estimation of the relative variations in the resistivities ρi of some elements of the geoelectric structure responsible for variations in ρa.. It reduces to the construction and solution of the following system of equations:

∏ xi aij = сj, j = [1, m]. i Here i is the number of the structure element ( it is comfortable to give to this number the value of the structure element resistivity in Ω m), xi = ρi2 /ρi1 is the relative change in the resistivity of the ith element of the structure, aij = εiav(Tj) is the [ρi1, ρi2]-averaged sensitivity of ρa to changes in the resistivity of the ith element of the structure at the electromagnetic field period Тj, cj=ρa2j /ρa1j , were ρa1j and ρa2j are the apparent resistivities at two different moments at the jth period, and m is the number of the periods used. (We remind the reader that the sensitivity of ρa to changes in the resistivity of the ith element of the structure is defined here as a value that determines the relation between relative variations in the resistivity of the ith element of the structure and the corresponding relative variations in the apparent resistivity: εi (ρi,Tj) = dlogρa/dlogρi = (dρa/ρa)/ (dρi/ρi) [Sholpo, 2003].) Taking the logarithm of these equations, we obtain the system of linear equations. The solution of the latter involves no fundamental difficulties if the values ρa1, 2 are determined with a sufficient accuracy in the entire range of required periods. The main problem in the realization of this method is the correct determination of the aij = εiav (Tj) using numerical modeling of a well-studied geoelectric structure. The results of testing of this method on a 1D model structure were published in [Sholpo, 2006]. This article is dedicated to its testing on a 2D structure. Testing of the proposed method was performed using 2D model of Petropavlovsk geodynamical polygon (Kamchatka). The geoelectric model of this polygon includes three deep and a few surface high conductive elements (Fig.1. The numbers within the model are electrical resistivities in Ω m). Figure 2 presents spatial dependences of maximal values of the function εi(√ T) calculated for i = 8, 10, 20, 30 at the number of points of profile AA. Taking into account the investigation purpose and using these dependences it is possible to determine the best place for MT monitoring.


478

mailto:[email protected]

A x, km A

depth, km

Fig. 1. Geoelectric structure of the earth crust at Petropavlovsk geodynamical polygon (Kamchatka)

0 40 80

0.0

0.4

0.8

1.2

x km

ε i max

8

1020

30

Fig. 2. Spatial dependences of maximal values of the function εi(√T) on the profile AAi = 8, 10, 20, 30 – the numbers of the structure elements and its resistivities in Ω m

ε8 --------------------------------- √T = 3.5 - - - - - - - - - - √T = 5ε10 --------------------------------- √T = 14 - - - - - - - - - - √T =17ε20 --------------------------------- √T = 12 - - - - - - - - - - √T =17

ε30 --------------------------------- √T = 1

At the same time it is necessary to study the frequency responses εi in the different points of the profile, because the possibility of the separation and the accuracy of the estimation of contributions of the individual elements of the geoelectric structure are dependent on the correlation between the periods of its maximums. Thanks to that it is possible to determine the optimal frequency interval. Figure 3 presents the frequency responses of ρa sensitivity to variations in the resistivity of four structure elements for the profile points with x = 37, 45, 60km.


479

1E-4 1E-3 1E-2 1E-1 1E+0

0.0

0.5

1.0

1E-4 1E-3 1E-2 1E-1 1E+0

0.0

0.5

1.01E-4 1E-3 1E-2 1E-1 1E+0

0.0

0.5

1.0

x

x

Hz

Hz

Hz

108

20

10

20 8

10

20

8

ε i

x = 37 km

f

= 47 km

f

= 60 km

f

30

30

30

Fig. 3. The frequency responses of the ρa sensitivities to variations in the resistivities of the structure elements for three points of the profile AA: x = 37, 45, 60 km

i = 8, 10, 20, 30 – the numbers of the structure elements and its resistivities in Ω m

The results of the method testing at these points for two variants of changes in the resistivities of three deeps elements are given in Table 1. The matrices of the coefficients aij were calculated for these points. It should be noted that the matrix A is the characteristic of the given point placed on the surface of the given structure. It can be calculated once for all if geoelectric structure is studied well enough to perform numerical modeling. Owing to that the conversion of apparent resistivity changes into relative changes in rock resistivity can be easily produced. In Table 1 ρi1 and ρi2 are the resistivities of the ith structure element at two different moments; (ρi2/ρi1)r

– the real relative changes in the resistivities of the ith element; (ρi2/ρi1)c – the values calculated using the tested method; δ% – the relative error of (ρi2/ρi1)c; ”noise” – the errors introduced into the apparent resistivities ρa1 and ρa2; εi max – the maximal value of the function εi(Tj) at the present points. To construct the columns of free terms of system of linear equations bj = lg cj = lg (ρa2/ρa1), the values ρa1 and ρa2 are calculated for the sets of given values ρi1 and ρi2 for the optimal interval of periods Tj using the program by Vardaniants. To make the model problem even more realistic, the errors typical of the accuracy of the present-day MTS studies (3-5 %) are introduced into the values ρa1 and ρa2. As can be seen from Table 1 the variations in the resistivity of 20th element are determined rather roughly at the all points. This could be expected because ε20max is small and at the point x=37 km the frequency response maximums of ε10(Tj) and ε20(Tj) are located at close frequencies (Fig.3). Because of this the relative changes in resistivity of 10th element are determined at this point with the greater errors as at others. The point x = 47 km is most favorable for the investigations of 10th and 8th elements but at this point the errors δ20 are also too great. To observe the variations in the resistivity of the 20th element it is reasonable to choose the observation point at the beginning of the profile, for example at x=14.5. Here it is possible to neglect the influence of the 8th element and to solve the system of linear equations for two unknowns (Table 2). Thus, in present case it is


480

impossible to observe the changes in the resistivity of all three deep structure elements with the sufficient accuracy using only one observation point.

Table 1. Results of the testing of the method at three points of the profile AA( x = 37, 45, 60 km )

for two variants of the changes in the resistivity of three structure elements

Table 2. Defining of the changes in the resistivity of 20th structure elementsby the observations at the point x = 14.5 km

i 10 8 20

(ρi2/ρi1)и .905 1.286 .818

εi max .45 .0038 .40(ρi 2/ρi 1)э

δ%0.906

0.1

1.685

31

0.799

2.3δ%

(noise 3%)6.6 281 12.4

( ρi 2 /ρi 1)э

δ%0.882

2.5

0.859

4.9δ%

(noise 3%)4.9 10.1

1 Variant 1 Variant 22 i 10 8 20 10 8 20

3 ρi1 Ω m 9 7 15 10.5 7 22

4 ρi2 Ω m 11 9 25 9.5 9 18

5 (ρi2/ρi1)и 1.222 1.286 1.667 .905 1.286 .818

6 x = 37 km, ε10 max = 0.75, ε8 max = 0.20, ε20 max = 0.25

7 (ρi 2/ρi 1)э

δ%1.2875.3

1.1897.6

1.48411.0

.8733.5

1.3243.0

.8989.8

8 δ%(noise 3%) 20 13.4 38 13.3 7.7 759 x = 47 km, ε10,max = 0.77, ε8,max = 0.91, ε20,max = 0.1710 (ρi 2/ρi1)э

δ%1.2340.9

1.2602.1

1.6730.3

.8990.5

1.2980.9

.8250.9

11 δ%(noise 3%) 2.8 1.8 17 1.4 1.2 7.1

12 δ%(noise 5%) 7.6 1.6 37.8 7.1 1.7 23

13 x = 60 km, ε10,max = 0.74, ε8,max = 0.51, ε20,max = 0.14,14 (ρi2/ρi1)э

δ%1.2190.2

1.2294.4

1.7535.2

.9040.1

1.3112.0

.8120.7

15 δ%(noise 3%) 2.9 4.4 10.5 3.1 3.0 2016 δ%(noise 5%) 6.0 5.0 23.5 6.0 2.8 32


481

To investigate the influence of the surface layer on the results of the estimations of the values of the relative variations in the resistivities ρi there was studied a model where to three deep elements the fourth one with the number i = 30 was added (see Fig.1). The results of the calculations of the values ρi2/ρi1 at the point x = 47 km for the changes in the resistivity of four structure elements are given in Table 3. As is seen the introduction into the model of the surface element capable of time- change of its resistivity does not increase the errors of the determining of the relative variations in the resistivity of the deep structure elements, which is evidently the consequence of the isolated position of the frequency response maximum of ε30 (Fig.3).

Table 3. Results of the testing of the method at the point x = 47 km in case of the changes

in the resistivities of four structure elements.

I 10 8 20 30ρi1 Ω m 10.5 7 22 20ρi2 Ω m 9.5 9 18 40(ρi2/ρi1)и .905 1.286 .818 2

εi max .77 .91 .17 .51(ρi 2/ρi 1)э

δ%.903.2

1.3031.3

.812.7

2.052.5

δ%(noise 3%)

4.4 1.8 21.9 2.7

Conclusions(1) To choose the regime of observations of variations in the rocks conductivity suitable for the investigation aim it is needed to study the frequency and spatial dependences of the sensitivities εi(T, x) using numerical modeling of a well-studied geoelectric structure.(2) The application of the proposed method in optimal points of observation allows us to estimate values ρi2 /ρi1 with accuracy sufficient for monitoring of relative changes in the electrical conductivity of rocks.

ReferencesSholpo, M.E. (2003), Magnetotelluric monitoring of variations in the electrical resistivity of a conducting

layer: numerical modeling. Physics of the Solid Earth, Vol. 39, No. 2, 112-117.Sholpo, M.E. (2006), Monitoring of relative changes in the electrical conductivity of rock from

observations of the magnetotelluric apparent resistivity (numerical modeling). Physics of the Solid Earth, Vol. 42, No. 4, 323-329.


482

FRACTAL CHARACTERISTICS OF ULF EMISSIONS REGISTERED IN THE HIGH LATITUDE

SEISMIC-QUIET REGION OF SPITSBERGEN ISLAND

N.A. Smirnova, A.A. Isavnin

Institute of Physics, St.Petersburg University, St.Petersburg, 198504, Russia, e-mail: [email protected]

Abstract. ULF frequency range (f = 0.001-5Hz) is now considered as a promising one for manifestation of electromagnetic earthquake precursors. To support the registration of ULF emissions in seismic-active areas, and to exclude magnetospheric effects from the lithospheric ones, the reference stations in seismic quiet but magnetic active regions are required. Here we are starting to analyze the magnetic field data obtained with high resolution (10 Hz sampling rate) in the high latitude seismic-quiet region of Spitsbergen Island (Barentsburg station, Φm=76° N; Λm=115°). Fractal analysis of the ULF data has been fulfilled using Higuchi method and Burlaga-Klein approach. The first preliminary results of the fractal analysis are presented. Variations of the fractal dimensions of the ULF emissions time series are considered along the local time. The possible physical interpretation of the results obtained is suggested. A possibility of using the Barentsburg observatory as a reference station in seismo-electromagnetic research is discussed.

Introduction

One the most important tasks in geophysics is the study of precursors of earthquakes. Nowadays it is known that earthquakes are related to weak seismogenic ULF emissions (see Hayakawa and Fujinawa (eds), 1994). But the problem of analyzing these signals is that they are strongly screened by ULF emissions of the magnetospheric origin (Smirnova, 1999). So the analysis may be strict only during the calm “space weather” time. In this work we are starting to analyze the magnetic field data in the high latitude seismic-quiet region of Spitsbergen Island. This region is situated at the cusp latitude – the most sensitive area to magnetic field disturbances. Because of its extremely high sensitivity to the solar wind conditions it can be used as a reference station for the research of seismogenic electromagnetic emissions. There is also an important task to monitor the position and size of the cusp area.

Experimental data

In Fig.1 you can see a principle scheme of the magnetosphere of the Earth. There are two cusp areas in the magnetosphere corresponding to two poles of the dipole field of the Earth. Each of the cusps is situated between closed and opened magnetic field lines. It is clear that dependent on the solar wind conditions the geographical position of the station can fall into the zone of opened or closed magnetic field lines. So it is important to learn out if the station is situated in the cusp during the analyzed period of time (the magnitude gap of cusp area is very narrow). For this aim we will use the Tsyganenko model of magnetosphere to project the cusp area on the Earth's surface according to parameters of the solar wind.Comparing the magnetic field behavior during the periods when the station falls into the projection of the cusp area on the Earth's surface with parameters of solar wind we will exclude the fluctuations of the magnetic field caused by inhomogenities in the solar wind. After such filtering we will obtain the magnetic field data with most of fluctuations in ULF frequency range caused by lithospheric effects, not magnetospheric. Analyzing such “clean” data and confronting it in time with registered lithospheric events and finding correlation parameters between them could give us some sort of

Fig.1. The scheme of the Earth's magnetosphere. One can see the polar cusps between opened and closed magnetic field lines (the so-called points of bifurcation).


483

prediction mechanism.In this work we are giving preliminary results of data analysis showing the possibility of filtering magnetic field data from the magnetospheric effects. Until recent times few measurements were held at the cusp latitudes and most of them were recorded with low resolution such as 1/60Hz. Using such data we couldn't resolve ULF-range effects in the magnetosphere of the Earth. In this work we are starting to analyze for the first time the magnetic field data with frequency of 10Hz obtained from the Barentsburg observatory – high latitude station with coordinates 78°N, 14°W (local time UTC+1 hour) and magnetic coordinates Φm=76°N, Λm=115°. Currently we have available data for the January 2008. An example of the magnetic record at Barentsburg (H-, D- and Z- components) for the chosen date – 5 th

January 2008 is presented in Fig.2.

Fig.2. The magnetic field data for the 5th January 2008 obtained at the Barentsburg observatory. The x-axis denotes local time (LT). One can see two rather strong distrubances near 21:00 LT and 00:00 LT.

Methods of data analysis

We use two fractal methods of spectrum analysis in our research. These are Higuchi method (Higuchi, 1988, 1990) and Burlaga & Klein method (Burlaga and Klein, 1986). Here we present a short discription of each of them.1. Higuchi method.Consider a set data X(n) with fixed sampling rate. K new time series can be obtained from these data using formula:

Then we can get the length of the curves representing the obtained time series as:

Finally we find the total length of the curve depending on k as

If then the time series X(n) is considered to be a fractal one with dimension D.

k),=(mkk

mN+mX,),+X(mk),+X(mX(m),=X k

m ...2,1,...2k

⋅

−

| |k

k

mN

kk

mNN

k))(i+X(mik)+X(m=(k)L=i

m

111

1

⋅

−

⋅

−

−⋅⋅−−∑

k

(k)L=L(k)

k

=mm∑

1

( ) DkkL −∝


484

2. Burlaga & Klein method.The main difference between the Burlaga & Klein method and the Higuchi method lies in their operation of averaging the time series before calculating the curve length. If assume k = 3 then the curve lengths are:

As the k interval becomes larger the curve length defined by Burlaga & Klein method tends to be shorter, although the fractal dimension D tends to be larger than the real value.So it is clear that the Higuchi method is better suited for analyzing time series with low resolution. This is especially actual in the case of ULF emissions with frequences lying in the range In our work we use the magnetic field data with resolution 10 Hz obtained in Barentsburg, Spitsbergen. Such a high resolution allows us to investigate appropriately the fractal properties of ULF emissions in the frequency range from f=0.001 Hz up to f=1-2 Hz. We have applied the Higuchi method of data processing to obtain more stable values of scaling exponents and fractal dimensions of the ULF emission time series.

Results

Here we present the results of our analysis of the magnetic field data for the 5th January 2008. To find out how fractal dimension varies during the 24-hour period we calculated the dynamics of fractal dimension (see Fig.3).

We held our calculation using the following algorithm. For every 10th-minute interval we compose the corresponding time series of the H-, D- and Z- component variations, each of which is contains 6000 points (taking into account the 10Hz resolution). Then we apply the Higuchi method to the obtained time series beginning from the current timestamp. So given that, every horizontal stroke on our plots denotes 10 minutes gap between calculations.

From Fig.3 one can see the pronounced daily dynamics of the ULF emissions fractal dimensions in all of the three magnetic components (H, D, and Z). There are several extremums in those dynamics, the most of which correspond to the critical times (00:00, 03:00, 06:00, 09:00, etc.). But also there are extremums, which are not concerned with daily rotation of the Earth. These peculiarities might show the movement of the station between zones with opened and closed magnetic field lines and also they can reflect the interaction of the magnetosphere with solar wind (or both effects mixed).

Fig.3. The dynamics of fractal dimension calculated for the H- (top), D- (middle), and Z- (bottom) components of ULF emissions registered at Barentsburg station on 5th January 2008. The Higuchi method was used for this calculation.

Hz1010 3 ÷−

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )3/

3

654

3

987

3

321

3

6543

…

−

− +

X+X+XX+X+X+

X+X+XX+X+X=LBK

( ) ( ) ( )| | ( ) ( )| | ( ) ( )| | ( ) ( )| | ( ) ( )| | ( ) ( )| |3/

3

695847

3

3625143

…−−−−−−

+XX+XX+XX

+XX+XX+XX

=LH


485

Discussion and conclusions

From the results obtained we see the convenience of fractal methods for study of the behavior of the ULF emission source system. We see that fractal dimension of the ULF emissions time series in its dynamics “feels” the changes in magnetosphere such as terminators and other fluctuations of magnetic field and reflects them as extremums. The next step of our research is to consider the correlation between the dynamics of the ULF emissions fractal dimensions and solar wind parameters and exclude variations caused by magnetospheric effects.

To get the more definite conclusions concerning the origin of the extremums revealed, it is necessary to monitor the position and size of the cusp. For such a forthcoming research we will use the Tsyganenko model (http://geo.phys.spbu.ru/~tsyganenko/Geopack-2008.html), which allows us to calculate the projection of the cusp on the Earth's surface. That will give us a possibility to relate appropriate peculiarities in the ULF emissions fractal dynamics with moment of ingress of the Barentsburg station into the entire cusp region.

Acknowledgements

We thank Polar Geophysical Institute for provided data of Barentsburg station and especially Yury Katkalov – the developer of the DMS (the database system for collecting and processing geophysical data).The work was supported by RF President Grant “Leading Scientific School” 1243.2008.5 and Program RNP.2.2.2.2.2190 of Russian Ministry of Education “Intergeophysica”.

References

Burlaga L.F. and L.W. Klein (1986), Fractal structure of the interplanetary magnetic field, J. Geophys. Res., 91 (A1), 347-350.

Hayakawa M. and Y. Fujinawa (eds, 1994), Electromagnetic Phenomena Related to Earthquake prediction, 677 pp.,Terra Scientific Publishing Companym Tokyo, Japan

Higuchi, T. (1988), Approach to an Irregular Time Series on the Basis of Fractal Theory, Physica D, 31, 277-283.

Higuchi, T (1990): Relationship between the Fractal Dimension and the Power-low Index for a Time Series: a Numerical Investigation, Physica D, 46, 254-264.

Smirnova, N. (1999), The peculiarities of ground-observed geomagnetic pulsations as the background for detection of ULF emissions of seismic origin, in: Atmospheric and Ionospheric Electromagnetic Phenomena Associated with Earthquakes, edited by M. Hayakawa, Terra Sci. Pub. Co., Tokyo, 215-232.

Tsyganenko, N.A. (2008), http://geo.phys.spbu.ru/~tsyganenko/Geopack-2008.html


486

PECULIARITIES OF THE ULF EMISSION FRACTAL CHARACTERISTICS OBTAINED AT THE STATIONS OF 210 GM

A.A. Varlamov, N.A. Smirnova Institute of Physics, St.Petersburg University, St.Petersburg, 198504, Russia, e-mail:

[email protected]

Abstract. Magnetic records (1Hz sampling rate) of the 5 stations (Guam, Moshiri, Paratunka, Magadan and Chokurdakh) located from equatorial region to auroral zone approximately along the same geomagnetic meridian (210 MM) have been analyzed using Higuchi method of fractal analysis. The period of 22 months (October 1992 – July 1994) that embodies the date of the strong Guam earthquake of 8 August 1993 has been considered. Comparison of the ULF emission scaling parameters (spectral indexes β and fractal dimensions D) obtained at different latitudes has been fulfilled. Dependence of β and D on Kp index of geomagnetic activity has been separately analyzed for each of 24 local time intervals. The results obtained are considered on the basis of the SOC (Self-organized criticality) concept. A possibility of using the data of the 210 GM stations as reference materials for the Guam seismic active area is discussed.

Introduction In a series of papers by Smirnova and Hayakawa (see References), the specific dynamics of fractal

characteristics of ULF emissions registered at the Guam observatory in relation to the strong Guam earthquake of 8 August 1993 has been reported. Namely the spectral exponents β were decreasing and the corresponded fractal dimension D was increasing when approaching the date of the Guam earthquake. It is also revealed that scaling parameters of the ULF time series are influenced by geomagnetic activity. So the local seismo-electromagnetic phenomena are screened by magnetospheric effects, which are of more global character. To distinguish between these effects the data from reference stations are necessary in addition to the Guam data. Here we consider coordinated magnetic records (1Hz sampling rate) of the 5 stations located approximately at the Guam geomagnetic meridian (210 MM stations). The purpose is to compare the fractal properties of ULF emissions along meridian profile and try to answer the question which station could be used as a Guam reference point in seismio-electromagnetic research. Experimental Data

Magnetic records (ΔН, ΔD, ΔZ – components) used for our analysis were obtained at the 210 MM stations by means of ring-core-type fluxgate magnetometers with sampling rate 1 second. The chain of stations includes Chokurdakh (CHD), Magadan (MGD), Paratunka (PTK), Moshiri (MSR), Guam (GAM) and covers a wide range of latitudes from auroral zone to the equator (see the Table 1)

Table 1. The list of the stations used for analysis

Abbreviation geographic geomagnetic

GAM 13.58° N 144.87° E 5.61° N 215.55° E MSR 44.37° N 142.87° E 37.28° N 213.55° E PTK 52.94° N 158.25° E 46.17° N 226.02° E MGD 59.97° N 150.86° E 53.49° N 218.75° E CHD 70.62° N 147.89° E 64.66° N 212.14° E

The observation period covers 20 months from October 1992 to July 1994 This period embodies the date of

a strong M8 Guam earthquake of 8 August 1993: depth = 60 km, Φ=12.98° N, Λ= 144.80° E. An example of the typical record obtained at Paratunka is given in Fig. 1. In this study we have analyzed the data in each of the 24 daily 1-hour intervals. In the insertion to Fig. 1, one can see the enlarged 1-hour record (ULF emission) for 16-17 UT. So each analyzed time series, which represents ULF emission in 1-hour interval, contains 3600 points. We apply fractal methods to analyze scaling (fractal) characteristics of ULF emissions and study their dynamics in each location in relation to geomagnetic activity as well as in relation to the strong Guam earthquake of 8 August 1993.


487

Fig. 1 An example of the typical magnetic record, (Paratunka, 15/10/92)

Fractal approach

Now it is recognized that the extended dissipative dynamical systems evolve naturally to the state of self-organized criticality (SOC). The SOC state is characterized by high sensitivity of the system to any external perturbations and a fairly broad energy spectrum of dissipation events. The Earth’s magnetosphere and lithosphere were shown to exhibit this type of behavior. The fingerprints of the SOC state are fractal organization (power-law distributions) of the output parameters in both space and time domains (scale-invariant structures and flicker-noise or 1/f fluctuations). Hence we can use fractal methods for analysis of spatiotemporal scaling characteristics of the magnetospheric and lithospheric emissions.

Three methods of fractal analysis have been considered: 1) PSD method (Feder, 1989; Turcotte, 1997); 2) Burlaga-Klein method (Burlaga and Klein, 1986); 3) Higuchi method (Higuchi, 1988, 1990). Time series can be considered as fractal ones in the chosen frequency range, if its PSD (power spectra density) S(f) follows the power law at those frequencies: S(f) ~ 1/fβ Spectral exponent β, and fractal dimension D, characterize the rate of irregularity of the time series. β and D are connected via the Berry equation (Berry, M.V. ,1979 ): D = (5 - β ) / 2. The more advanced fractal approach is Higuchi method. It gives more stable values of fractal dimensions. The method is based on the estimation of length of fractal curve X(N). We consider the ULF data with sampling rate of 1s as a time sequence X(1), X(2), …, X(N). From given time series k new time series are constructed, defined as follows: Then we estimate the length of the curves, which represents the new time series for each k:

kkk

mNNkimXikmXkL

kmN

im

11))1(()()(1

⋅

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

⋅⎥⎦⎤

⎢⎣⎡ −

−⋅⋅−+−+= ∑

⎥⎦⎤

⎢⎣⎡ −

=

where is a normalizing factor. Then the total length of curve is defined as follows: And finally, we plot <L(k)> versus k (ex. k=1, 2, …, 10). If and scales linearly in log-log presentation then we can consider this time series as fractal ones and estimate fractal dimension from the slope of curve (see Fig. 2).

Fig. 2. Examples of the calculation of the ULF emission fractal dimension using Higuchi method.

)...,,2,1(...,),2(),(),( kmkk

mNmXkmXkmXmXX km =

⎭⎬⎫

⎩⎨⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⎥⎦⎤

⎢⎣⎡ −

+++=

k

kLkL

k

mm∑

==>< 1)(

)(

( ) DkkL −∝

kk

mNN

⋅⎥⎦⎤

⎢⎣⎡ −

− 1


488

Results 1. Dynamics of the ULF emissions fractal dimensions at the stations along 210 geomagnetic meridian.

We have obtained 24 sets of the results represent-ing dynamics of the ULF emissions fractal dimen-sions along the 210 MM profile in each 1-hour daily interval. One example of such dynamics for 11-12 UT interval is shown in Fig. 3. One can see the following peculiarities: 1) Daily fluctuations of D exhibit chaotic dynamics at each of the 5 stations of 210MM. It is difficult to understand whether such fluctuations are random ones or represent deterministic chaos. Those fluctuations may contain the influence of the magnetospheric, ionospheric, lithospheric, processes as well as effects of the man-made origin. 2) Averaging over ± 5 days (running average values of D) exhibits some features of deterministic behavior. Namely the distinct modulation of D with periods near 27 days is outlined, which corresponds to rotation of the Sun around its axes. So it is definitely a magnetospheric effect that is confirmed by the corresponding variations of Kp-index (see the bottom panel). 3) Going to the longer time variations of D, one can reach the effect revealed earlier in the reference papers. Namely that is gradual increase of the ULF emissions fractal dimension before the Guam earthquake of 8 August 1993 (marks by an arrow). That may be an earthquake precursory effect. According to our analysis this effect is more pronounced in the H-component of Guam, and it just vanishes in the Moshiri, Paratunka and Magadan data. What about Chokurdakh, lots of gaps in the data do not allow us to trace such a tendency. 4) The modulation by the 27-day period is the most pronounced in the auroral zone (Chokurdakh), where it is manifested in the H, D and Z components. At the other stations, which are situated inside the plasmasphere, the modulation is observed in the H and D components, and never in Z-component. As to the Guam station, which is situated in the region of equatorial electrojet, such a modulation is seen also in Z component 5) The range of variations of D (see H-component) is wider at the stations of auroral zone (Chokurdakh) and equatorial electrojet region (Guam) in comparison with inside-the-plasmasphere stations (Magadan, Paratunka, Moshiri). Fig. 3. Dynamics of D along 210 MM. Thin curves - daily values, the thick ones - ± 5 day running average value.


489

2. Dependence of the ULF emissions fractal dimensions from Kp index of geomagnetic activity

Some results to study the correlations between fractal dimensions of ULF emission time series and the Kp –index of geomagnetic activity are presented in Fig. 4. From the left part of Fig. 4 one can see that fractal dimensions of the ULF emissions time series decrease with increasing of magnetic activity at all stations, but the rates of the decrease are different at different stations and in different time intervals. The dependence of D on Kp is the most pronounced in the auroral zone (Chokurdakh), and it is rather smooth at the middle latitudes (Magadan and Paratunka). Near the equator (Guam) it becomes again sharper. From the right part of Fig. 4 one can see daily variation of the correlation coefficients between D and Kp. This variation is more regular in the middle latitudes (Magadan, Paratunka, Moshiry) with a pronounced maximum near 15 UT and minimum around 4 UT. As to the auroral zone and equatorial region (Chokurdakh and Guam) the maximum is not so pronounced there, and it is spread over the time interval 10-20 UT. But the minimum is also around 4 UT, which corresponds to local time near the noon (14 LT). So we may conclude that the evening, night and early morning hours are preferable for studying magnetospheric effects whereas the noon hours are the most suitable for analysis of lithospheric effects. Fig. 4. At the left - examples of the plots representing fractal dimensions of the ULF time series (H-component, 11-12 UT) versus Kp-index for all stations. At the right - daily variations of the correlation coefficient between D and Kp along meridian profile (the H-component).

Paratunka

Chokurdakh

Magadan

Moshiri

Guam


490

Discussion and conclusions

We have obtained rather pronounced dynamics of the ULF emission fractal dimension D along 210 MM profile, which reflects the global magnetospheric effects as well as some local processes at each of the 5 stations. One can see magnetospheric effects from the thick lines (± 5 days running average values of D) in Fig. 3 as a modulation of D with the same period near 27 days. That is a characteristic period of the Sun rotation, and it is clearly seen from the corresponding modulation of the Kp index of geomagnetic activity at the bottom panel in Fig. 3. In the horizontal components H and D, this effect is pronounced at all the stations, but in the vertical component Z, it is more visible near the projection of characteristic electrojet locations: auroral electrojet - Chokurdakh station, and equatorial electrojet – Guam station. At the stations situated inside the plasmasphere, which are Magadan, Paratunka and Moshiri, the modulation of the 27-days period is practically disappeared in Z-component. So in order to exclude magnetospheric effects from the horizontal components of the Guam station, all the other stations of 210 MM are appropriate. But to exclude magnetospheric effects from the vertical component of the Guam station, the station Chokurdakh belonging to auroral zone is more recommendable, if there is no other reference station situated near auroral electrojet projection. As to the preferable local hours for analyzing lithospheric effects, the noon hours (near 4 UT or 14 LT) are the most suitable, since influence of geomagnrtic activity on the fractal properties of ULF emissions is depressed near the noon hours (see Fig. 4).

The following conclusions have to be done from the analysis fulfilled. 1. Our analysis shows that the results of the fractal analysis of the data from 210 MM chain of stations may give a support to the results of the Guam station data analysis when we attempt to distinguish between magnetospheric and lithospheric effects. 2. Since magnetospheric effects are of global character, they give the correlated part in the results obtained at all the stations. Lithospheric effects could be of individual and local character at each of the station. So we have to manage how to remove the correlated part from the raw 210 MM results. Now we are working in this direction. 3. We understand that our results are of preliminary character, and they need to be checked and confirmed on the other independent materials. Nevertheless we may conclude that the peculiarities revealed can be used as control factors in the forthcoming investigations of the earthquake precursory signatures based on the analysis of scaling (fractal) characteristics of ULF emissions. Acknowledgements

The work was supported by RF President Grant “Leading Scientific School” 1243.2008.5 and Program RNP.2.2.2.2.2190 of Russian Ministry of Education “Intergeophysica”. We thank Prof. M. Hayakawa and Prof. K. Yumoto for the experimental data of 210 MM. References

Hayakawa, M., T. Ito, and N. Smirnova (1999), Fractal analysis of ULF geomagnetic data associated with the Guam earthquake on August 8, 1993. Geophys. Res. Lett., 26, 2797-2800

Higuchi T. (1988), Approach to an irregular time series on the basis of the fractal theory, Physica D, 31, 277-283 Higuchi, T (1990), Relationship between the Fractal Dimension and the Power-low Index for a Time Series: a

Numerical Investigation, Physica D, 46, 254-264. Smirnova, N. (1999), The peculiarities of ground-observed geomagnetic pulsations as the background for

detection of ULF emissions of seismic origin, in Atmospheric and Ionospheric Electromagnetic Phenomena Associated with Earthquakes, edited by M. Hayakawa, Terra Sci. Pub. Co., Tokyo, 215-232.

Smirnova, N., M. Hayakawa, and K. Gotoh (2004), Precursory behavior of fractal characteristics of the ULF electromagnetic fields in seismic active zones before strong earthquakes, Phys.Chem. Earth, 29, 445-451.

Turcotte, D.L. (1997), Fractals and Chaos in Geology and Geophysics, 398 pp., Cambridge University Press, Cambridge, New York, Melbourne, 2nd edition.


491

SIMULATIONS OF THE EQUATORIAL IONOSPHERE RESPONSE TO THE SEISMIC ELECTRIC FIELD SOURCES

Zolotov O.V.1, Namgaladze A.A.1, Zakharenkova I.E.2, Shagimuratov I.I.2, Martynenko O.V.1

1Murmansk State Technical University, Murmansk, 183010, Russia,

e-mail: [email protected], [email protected]; 2West Department of IZMIRAN, Kaliningrad, Russia,


Abstract. The results of computer simulations of the equatorial ionosphere response to the additional electric fields of presumably seismic origin have been presented and discussed. The effects are the equatorial anomaly reduction (or increase in some cases) with trough deepening and symmetric shifting of the double-peak TEC (and NmF2) maxima from the earthquake epicenter. The obtained results have been compared with GPS (Global Positioning System) IGS network data for the Peru earthquake of 26 September, 2005. The presented numerical results well reproduce observed features of the equatorial ionosphere behaviour before strong earthquakes.

Introduction It has been proposed that earthquakes are preceded by electromagnetic signals penetrating into the ionosphere and detectable from ground- and space-based measurements. Many papers are reported on anomalous ionospheric TEC (total electron content) disturbances as earthquake precursors generated by additional electric charges of a single sign, namely, positive (Pulinets, 1998; Pulinets et al., 2003; Pulinets and Boyarchuk, 2004; Liu et al., 2004). This paper presents the results of the computer Upper Atmosphere Model (UAM) simulations of the equatorial ionosphere response to the hypothetic seismic electric field sources of different spatial configurations. We previously made similar numerical simulations described in the paper (Zolotov et al., 2008) but for the mid-latitudinal cases. This work is a further development of that investigation but for the low-latitudinal sectors where F-region ionospheric plasma is strongly driven by the electric field. The obtained results are compared with the TEC differential maps for the Peru earthquake of September 26, 2005 derived from the GPS TEC data of the IGS (International GNSS Service) network (Dow et al., 2005). Observations Fig. 1 presents differential TEC (%) maps for September 20-28, 2005 calculated relative to the background quiet conditions on the basis of GPS TEC IGS network data. For details on features of the TEC disturbances see (Zakharenkova et al., 2008). Fig. 2 shows the geomagnetic activity for September, 2005. It is clear that such activity cannot generate investigated local large-scale disturbances in the TEC of the ionosphere. Simulations and model results Numerical experiments were carried out by means of the first-principle global 3D Upper Atmosphere Model (UAM) (Namgaladze et al., 1988, 1991, 1998). The UAM model describes the thermosphere, ionosphere and plasmasphere of the Earth as a unite system by means of numerical integration of the corresponding time-dependent three-dimensional continuity, momentum and heat balance equations for neutral, ion and electron gases as well as the equation for the electric field potential. It calculates electric fields both of thermospheric dynamo and magnetospheric origin and seismogenic electric fields as well. The upper atmosphere states, presumably foregone by strong earthquakes, were modeled by means of switching-on additional electric field sources (Fig. 3) in the UAM electric potential equation solved numerically jointly with all other UAM equations (continuity, momentum and heat balance) for neutral and ionized gases. These sources of different kinds and spatial configurations were switched on at the near-epicenter area and maintained as constant during 24 model hours. Two types and eight spatial configurations are taken into considerations:


492

(1) the dipole that consists of a pair of positive and negative potentials at the western and eastern boundaries of the near-epicentral area (Fig. 3, dip1); (2) the dipole elongated in the meridianal direction for ~1500 km (Fig. 3, dip2); (3) the point monopole charge that consists of one positive charge at the epicenter (Fig. 3, mono1); (4) the fulfilled rectangular regions of positive charges (Fig. 3, mono2-mono3); (5) positive charged (equipotent) straight lines of different directions (Fig. 3, mono4-mono6).

-90 -80 -70 -60

02 UT

20.09

02 UT

02 UT 02 UT 02 UT

02 UT 02 UT

02 UT

02 UT

21.09 22.09

23.09 24.09 25.09

26.09 27.09 28.09EQ day

-40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 %

30 20 10 0-10-20-30-40-50

-90 -80 -70 -60-90 -80 -70 -60-90 -80 -70 -60

-90 -80 -70 -60 -90 -80 -70 -60

-90 -80 -70 -60-90 -80 -70 -60-90 -80 -70 -60

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

30 20 10 0-10-20-30-40-50

ϕ

λ

ϕ

ϕϕ

ϕϕϕ

ϕ

ϕ

λ λ

λλλ

λ λ λ

Fig. 1. TEC diff. (%) maps at 02 UT for September 20-28, 2005 relative to the none-disturbed conditions. Black

spot – the earthquake epicenter position.

-200

-150

-100

-50

0

50

EQ01 08 15 22 29

Dst, nT

0 5 10 15 20 25 30 35 40

EQ01 08 15 22 29

Ap

0 0.5 1

1.5 2

2.5 3

3.5 4

4.5 5

EQ01 08 15 22 29

Kp

Fig. 2. Dst-, Ap- and Kp- indexes for the September, 2005. EQ – the earthquake day.


493

Fig. 3. Numerical grid and additional electric field sources: plus for positive sources, minus – for negative ones,

grey points – not modified nodes.

Fig. 4. Latitudinal variations of the electric potential (top left panel), the eastern component of the electric field

(top right panel) and corresponding TEC variations (%) (bottom panel). Red line (top panels) stand for the none-disturbed level.


494

Calculated latitudinal and longitudinal electric field variations and corresponding TEC variations relative to the background quiet conditions are presented graphically in Figs. 4-5.

Fig. 5. Longitudinal variations of the eastern electric field (bottom panel) and corresponding TEC (%)

disturbances (top panel) for earthquake modeled (left) and magnetically conjugated (right) points.

Figs. 4-5 show that magnitudes of the eastward electric field are less than 5-6 mV/m for simulated data both for the epicenter and magnetically conjugate areas. The symmetry of the potential and electric field concerning the geomagnetic equator is caused by the ideal conductivity of the ionospheric plasma along the geomagnetic field lines and, accordingly, by their electric equipotentiality. The corresponding TEC (%) disturbances reach 80% and more. Variations of the TEC disturbances for the low-latitude region (Figs. 4-5) have the form and the structure of the equatorial anomaly (“trough”). The increase of the eastward electric field leads to the deepening of the equatorial anomaly minimum (“trough” over the magnetic equator in the latitudinal distribution of electron density) due to the intensification of the fountain-effect. The dipole-like and single sign (positive) additional electric field


495

sources generate similar disturbances in the TEC of the ionosphere (in contrary to the mid-latitudinal cases). The noticeable difference is the magnitude of these variations: the dipole-like sources generate deeper equatorial trough. The amplitude of the peaks (the “camel case” maxima) remains nearly the same or slightly greater. That behaviour agrees well (at least qualitatively) with the presented Peru earthquake of 26 September, 2005 TEC variations (Fig. 1) detected at the earthquake near-epicenter area for a few days before and after the main shock. The work was supported by grant No. 08-05-98830 of the Russian Foundation for Basic Research. References Dow J.M., Neilan R.E., Gendt G. (2005), The International GPS Service (IGS): Celebrating the 10th

Anniversary and Looking to the Next Decade, Adv. Space Res., Vol. 36, No. 3, pp. 320-326. doi:10.1016/j.asr.2005.05.125.

Liu J.Y., Chuo Y.J., Shan S.J., Tsai Y.B., Pulinets S.A. and S.B. Yu (2004), Pre-earthquake ionospheric anomalies monitored by GPS TEC, Annales Geophysicae, 22, pp. 1585 -1593.

Namgaladze A.A., Korenkov Yu.N, Klimenko V.V, Karpov I.V, Bessarab F.S, Surotkin V.A, Glushchenko T.A, Naumova N.M. (1988), Global model of the thermosphere-ionosphere-protonosphere system, Pure and App. Geophys., Vol. 127, No. 2/3, pp. 219-254.

Namgaladze A.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V., Surotkin V.A., Naumova N.M. (1991), Numerical modelling of the thermosphere-ionosphere-protonosphere system, Journal of Atmospheric and Terrestrial Physics, Vol. 53, N. 11/12, pp. 1113-1124.

Namgaladze A.A., Martynenko O.V., Namgaladze A.N. (1998), Global model of the upper atmosphere with variable latitudinal integration step, International Journal of Geomagnetism and Aeronomy, 1 (1), pp. 53-58.

Pulinets S.A. (1998), Seismic activity as a source of the ionospheric variability, Adv. Space Res., Vol. 22, No. 6, pp. 903-906.

Pulinets S.A., Legen’ka A.D., Gaivoronskaya T.V. and Depuev V.Kh. (2003), Main phenomenological features of ionospheric precursors of strong earthquakes, Journal of Atmospheric and Solar-Terrestrial Physics, Vol. 65, pp. 1337-1347.

Pulinets S.A. and Boyarchuk K. (2004), Ionospheric Precursors of Earthquakes, Springer, Berlin, Germany. Zakharenkova I.E., Shagimuratov I.I., Tepenitzina N.Yu., Krankowski A. (2008), Anomalous Modification of

the Ionospheric Total Electron Content Prior to the 26 September 2005 Peru Earthquake, Journal of Atmospheric and Solar-Terrestrial Physics, doi:10.1016/j.jastp.2008.06.003.

Zolotov O.V., Namgaladze A.A., Zakharenkova I.E., Shagimuratov I.I., Martynenko O.V. (2008), Modeling of the ionospheric earthquake precursors generated by various electric field sources, Proceedings of the XXIX General Assembly of URSI, Chicago, USA, HP-HGE.21.


496

Date post:	23-Jun-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

COMPARATIVE ANALYSIS OF IONOSPHERIC VARIATIONS …geo.phys.spbu.ru ›...

Documents