Deep seismic sounding by microtremor (SSMT) broadband signals for Vrancea seismic zone

Paper structure:Introduction Methodology ResultsDiscussion and Conclusions

Introduction

Vrancea earthquake source is one of the most specific and curious among different seismic generation regions. Its most important peculiarity is the very narrow intermediate depth zone, able to generate high magnitude earthquakes at relatively high rate. The strong seismic potential, the location of the zone near big cities (Bucharest for example), its relatively deep seismogenic volume, concentrated in a very small volume, make this zone dangerous and put it at high seismic risk. The high occurrence rate and the large macroseismic field, observed in many cases, need very careful and deep investigations to unreveal the specific seismogenesis of the local earthquakes.The most important purpose in this case could be defined as to establish the seismic generation properties of the Vrancea seismc zone. To do this, it is necessary to study in details the deep interior around the seismogenic source and to try to establish all peculiarities, possibly in high resolution.

Hypotheses review

The most explored hypotheses about the seismogenic properties of Vrancea seismic zone are related to the plate tectonic models, developed for the zone.In general, the zone can be considered the result of a subducting part of the Thetis ocean slab. There are several indicators about the existence of paleosubduction like findings of ophiolites, flish, glaukofan shale, but up to now all these facts can not be combined in a logically stable ensemble. There are lot of contradictions about the strike and the slip directions of subduction. Some authors (Balla.Z; Csontos et. al) [1,2] consider northward and/or eastward direction (Sandulescu M.) [3], others (Gribacea R. et al) [4] suggest a south – southward direction from the Moesian platform. The fact of Neogen magmatism is another argument, supporting the subduction hypothesis. A lot of authors [1,2, 3,4,5] consider that the oceanic lithosphere was joined to the East-European platform and subducted to west and southwestward direction under the Carpatian arc during the Miocene. The ophiolite complex location trough the Transilvanides and Auseni Mountains and the subduction geometry to the north-west, also support a possible subduction of the Moessian platform.The generalization of all geophysical data gives three alternative models of the geodynamics of Vrancea seismic zone (Fig.1.)

Fig.1. Alternative models of the geodynamics of Vrancea seismic zone: A) the slap is detached and freely down drowns in the mantle; B), the slab is still consolidating with upwarding under the Carpathians lithosphere of the Moessian platform; C) Considers the detached continental lithosphere (Knapp J.H. et al.) [6].

For the first model (1A) the slap is detached and freely down drowns in the mantle. Second model - the slab is still consolidating with upwarding under the Carpathians lithosphere of the Moessian platform (Fig 1B). The third model (Fig.1C) considered the detached continental lithosphere. Each of the models supposes that a relatively cold and dense lithosphere body is located in the upper mantle under the Carpathian arc at the moment, and that it generates seismicity in a certain depth interval. The first two models supposed that subduction of the oceanic lithosphere took place, but they differ in which part of the orogen the process took place and in what direction the subduction developed. All models try to explain the different peculiarities of the crust-mantle coupling. They aim at explaining the relationships between upper and lower parts of the structures, observed and analyzed by geophysics. Particularly, it is interesting to explore whether the ocean basin has been incorporated during the formation of the East Carpathians, for which there are many signs at the boundaries of the two lithosphere plates. It is expected that they should should be reflected to the geological formations at the surface or in depth. Each of the models has its drawbacks and many internal contradictions, which means that the geodynamics of the Vrancea zone and its seismogenic properties are still under discussion. In 2005 he patented a new method of passive seismic study based on the analysis of the spatial variation of the spectrum of local micro-seismic area [7]. The method is based on experimentally verified assumption that the vertical component of displacement in micro seismic noise is represented mainly by the vertical component of moving fundamental fashion wave of the Rayleigh. Such an assumption is tenable first because micro seismic field comprises more than surface waves and second surface wave field contains predominantly Rayleigh waves and Love. The wave of Love appears horizontally polarized SH-wave, so no contribution to the vertical component of hesitation. Third, as shown by the numerous field measurements, waves Rayleigh are represented mainly by fundamental (zero) fashion, contributions by first fashion constitutes no more than a few percent in amplitude contribution of higher modes is insignificant in relation to the first.The possibility of using spatial variations of the energy spectrum of micro seismic field study of deep geological structure of the medium is formulated later as a method of low frequency seismic micro drilling, and was identified by Russian researcher in conducting

A.V.Gorbatikov micro seismic surveys on the island. Lantserot Canary Archipelago and later on the island. El Hierro. In further studies of various geological confirms the relevance of the proposed approach.Deep seismic sounding investigations of Vrancea seismic zone have been performed, based on data from 19 broadband seismic stations, located on the territory of Romania (Fig.1). According to the methodology of the seismic sounding, one of the stations was selected to be a basic station (PLOR). Stations data, location and the equipment used are presented in Table 1. The study was done in an area of approximately 63840 km2 (240x266 km).

1. MethodologyThe method is based on the inversion of the amplitude-frequency domain of the background seismic noise to depths. The main assumption is that the vertical component of the low frequency part of the noise amplitudes is determined by the fundamental modes of the Raleigh waves representing the main part of the seismic noise. The method is based an the assumption that the inhomogeneities in the earth crust and upper mantle change the spectrum of the low frequency seismic noise as follows: The spectral amplitudes of a fixed frequency f decrease over high seismic waves velocity zones and vice versa – they increased over low seismic velocity zones. The technology of measuring and processing provide:

1. Consistent measurement of statistical sustainable microseismic spectra at all points of the network or profile. Reaching statistical sustainable micro seismic signal accumulated over the experimentally determined period of stationarity signal equal to 2 hours.

2. Building on the map or profile of distribution of the amplitude of each frequency in the spectrum.

3. Bind received cards or accounts to the appropriate depth, proceeding from the relationship:

, where H(f) - the depth of the layer of which is being built image, - wavelength of the fundamental fashion relay, f - frequency in the spectrum of micro-seismic signal for which calculations are made, - phase velocity of the fundamental wave mode of relay frequency f, - coefficient of deep attachment estimated in the range 0.4-0.5.The processing and construction of the image consists in that, for each frequency f in the spectrum is plotted spatial curve (or map) the distribution of the variation of the intensity of the micro-seismic signal. This curve (map) is attached to depth. Based on the totality of the profile curves plotted dimensional image where the horizontal axis are the coordinates and the vertical axis – the depth corresponding to the profile (map).The horizontal resolution of the method is about 4% of the wave length and about 8% determine the depth/vertical resolution of the investigated anomalous inhomogeneity body. The vertical resolution thus can reach 15-16%.

Fig.1. Locations of the seismic stations for this research (Google earth).

The frequency f is related to the depth of the inhomogeneity body H and the velocity of the fundamental mode of the Raleigh wave Vr(f) by the relationship [7].

H= 0.5 Vr(f)/f

As a result of the application of this method the velocity contrasts of the seismic waves is the main parameter obtained in 2D and 3D views of the deep Earth’s interior structures. The performance of this methodology allow us on the base of the registrations of the microseismic background noise on the above mentioned broadband stations to construct the deep interior around the Vrancea seismic source. (Fig.2.) The DEM model over the Vrancea deep seismic zone is constructed with a grid 6’ х 6‘.White dots represent the hypocenters of the earthquakes with magnitude greater than 4.0 for the time interval 2000-2013 generated by Vrancea seismic zone. ( Seismic catalogue of National Institute for Earth Physics Romania).According to the methodology created for this research for all stations 1 hour seismic noise records are used. The time interval is 00:00:00-01:00:00 on 01.02.2015. For each station the power spectra have been calculated. The variations of the seismic wave’s velocities are determined according the base station (PLOR). The inversion of the spectral frequencies considered the dispersion of the Raleigh surface waves as the increased velocity is due to the wave’s lengths increase. For this purpose the velocity model suggested by F.Hauser (2001, VRANCEA99) [5] is used.

Table 1. Station

codeLocation Longitude Latitude elevation,

mSensors type

PLOR Plostina 26.649 45.851 680 CMG-40TPLOR 6 Plostina 26.641 45.842 720 CMG-40TPLOR 2 Plostina 26.643 45.842 701 CMG-40TPLOR 1 Plostina 26.646 45.852 706 CMG-40TPLOR 3 Plostina 26.645 45.854 722 CMG-40TPLOR 7 Plostina 26.64 45.86 831 CMG-40TPETR Petestri 27.2311 45.723 85 K2 DigitizerDOPR Dopca 25.388 45.967 544 Q330 DigitizerGHRR Gohor 27.408 46.06 209 Q330 DigitizerMLR Muntele Rosu 25.945 45.49 1392 Q330 DigitizerOZUR Ozunca 25.786 46.095 676 Q330 DigitizerTESR Tescani 26.648 46.511 375 Q330 DigitizerAMMR Amera 27.335 44.61 86 Q330 DigitizerICOR Ion Corvin 27.8009 44.1168 121 Q330 Digitizer CVD 1 Cernavoda 28.0624 44.3207 500 CMG-40TCFR Carcalin 28.1362 45.178 57 CMG-40TISR Istrita 26.5431 45.1188 750 Q330 DigitizerSULR Surlari 26.2526 44.6777 129 Q330 DigitizerMFTR Murfatlar 28.422 44.177 980 Q330 Digitizer

Fig.2. A deep model of the investigated territory up to depths of 400 km.The isosurfaces show the contract of the noise seismic waves velocities in Db.

On the 3D image the contrast of the seismic wave’s velocities is presented. According to the results obtained a low velocity seismic boundary is visible in NE-SW direction (22,97Db). This low velocity structure stretches plan - clearly limited to the south, and continuing behind the check you area north occupying the west and east almost the entire plane horizontally at depth of 200 km. Starting from a depth of 170-180 km, low-speed area is traced to depth 270-280km area in hypotheses earthquakes .There are also two relatively high velocity structures – located to the west and to the east from the hypocenter’s zone (blue area – -0.36 - 4.0 dB). The west structure reaches depths to about 150 km and might be interpreted as part of the Intra Carpathian subplate. The east structure reaching 280 km, might be interpreted as part of the Moessian platform.

The important fact is the localization of the hypocenters zone. It is located in the area of middle values velocity contrast - 20 dB. The zone has higher velocity relatively to the low velocity structure under the focal zone. On the other hand this is a relatively low velocities zone relatively to the high velocity structures to the east and to the west of the focal zone. These results are objective and do not try to explain the generating mechanism of the seismogenesis. This topic is rather complicated, but we believe that discovering the structures around and under the focal zone can help the solution of this complicated task.

2. Comparative analysis of our results with the results of other researchers.

On fig.3. the results of the (SSMT) (b) and P-waves velocity cross-section [Старосенко В.И ] on the same latitude 460N - from longitude 150E to longitude 350E. The cross-section by (SSMT) is from 25038’ to 28042’

The higher velocity zone (SSMT) located on the right of the focal zone down to the 100-120 km can be followed down to 200 km established by the P-waves method.The low velocity structure established by both methods

In all cases the located structures right of the focal zone is characterized by relatively high velocities and to the left of the focal zone – with relatively low velocities.

Other experimental results are related to the gravity Bouger anomalies on the territory of Romania and the created model by SSMT. There is a good correlation with the negative Bouger anomaly and the SSMT model as low velocity structures of the Raleigh waves located to NE and SW of the VRancea source. The same positive correlation is visible to the high velocities to NW and SE and the Bouger anomalies.

а)

б)

Fig.3 a) Latitudinal cross section of the 3D velocity model (P-waves) of the upper mantle ([Старосенко В.И.) [8] [ b) Cross-section on the same profile by (SSMT)

Fig. 4. Bouger anomalies map for Romania territory (upper scheme) according (D.Ioane, 2005)[9]; lower scheme – the SSMT model.

3. Verification of the SSMT accuracy by experiments in situ

To verify the method’s accuracy 4 experiments have been performed. The time interval for the summing of the signal is 1 hour. The experiments are as follows: 1. at 01.02.2015г. in the time interval 00-01AM (GMT); 2. at 01.02.2015 in the time interval 03-04AM; 3. at 10.02.2015 in the time interval 10-11AM; 4. at 10.02.2015 in the time interval 12-13PM . The results obtained are presented on fig. 5.

Fig.5. Results from the four experiments as follows – from left to right and from up to down:1-01.02. 00-01 AM; 2-01.02. 03-04 AM; 3-10.02. 10-11 AM; 4-10.02. 12-13 AM.

The sensitivity of the model is assessed by two parameters – the dispersion and the standard deviation.

Фиг.6 Space distribution of the dispersion for the four models.

Фиг.8 Space distribution of the standard deviation for the four models.

On fig. 6 the space distribution of the dispersion down to 200-250 km varied in the

interval 0.0007-0.36 . For depths 250 – 400 km the dispersion is variable. The

variation is changed from 0.36 to the west to 0.5 to the east.The standard deviation (fig.7.) varied in the interval 0.02-0.3 dB/10 for depth down to 150 km and 0.3-0.8 dB/10 for depths from 150 to 400 km.In general all results tend to present the same picture of the observed velocity properties of the deep structures around Vrancea seismic source and one and the same picture to the contrast of the Raleigh waves velocities. In the processing of Geophysical information is widely used in practice the correlation function and the correlation coefficient. Normally correlation function allows to determine the degree of similarity between the two signals and their relationship relative to one another.The functions of the cross-correlation between the spectra of experiments the separated for any of the stations are shown in figura.5.

Fig.9.

In Table 2 presents the correlation coefficients between categories to experiments and the average correlation coefficients. The total average correlation coefficient is 0.67 , which indicates a relatively good matching results.

code of the

station

cross correlation coefficient between 1

and 2 experiment

cross correlation coefficient between 1

and 3 experiment

cross correlation coefficient between 1

and 4 experiment

cross correlation coefficient between 2

and 3 experiment

cross correlation coefficient between 2

and 4 experiment

cross correlation coefficient between 3

and 4 experiment

the average value of the correlation coefficient

AMMR 0.809555981 0.959533312 0.919169667 0.809406442 0.859274097 0.891832043 0.874795257SULR 0.882216731 0.774857546 0.903986004 0.762700537 0.884497293 0.860953788 0.844868651MFTR 0.637559155 0.821447034 0.833664517 0.743381616 0.792763985 0.910706355 0.789920444ISR 0.438098428 0.858554468 0.744127324 0.644406387 0.797237941 0.841258651 0.720613866ICOR 0.696816104 0.926975376 0.843456078 0.728398839 0.622224548 0.873043841 0.781819131CVD 0.778908703 0.987199402 0.972566861 0.784129071 0.782532737 0.965822604 0.878526563CFR 0.856073891 0.536646603 0.927724504 0.547904851 0.843590039 0.404484992 0.686070813

TESR 0.384664891 0.393677988-

0.131268199 0.249484786 0.159401448 0.674293105 0.288375671PETR 0.518074223 0.914929932 0.878449298 0.579727808 0.551705687 0.867559338 0.718407714OUZR 0.195649349 0.458069841 0.332671823 0.513886413 0.802851805 0.775902724 0.513171993MLR 0.773867159 0.975756612 0.820644341 0.766080728 0.918408174 0.807054894 0.843635318GHRR 0.790853376 0.879746746 0.788771583 0.777661757 0.741646005 0.706448148 0.780854603

DOPR 0.140917379 0.857144661 0.707339178 0.287698492-

0.100804122 0.562946685 0.409207046PLOR7 0.795007419 0.955534451 0.623750945 0.778394835 0.776917728 0.622773521 0.758729816PLOR3 0.470117716 0.834066217 0.888334698 0.381391245 0.472636791 0.777329444 0.637312685

PLOR1 0.108753199 0.670642241 0.580301149-

0.045062349 0.054885501 0.387385276 0.292817503PLOR2 0.676912483 0.875280297 0.783143131 0.827052008 0.767012078 0.856391018 0.797631836PLOR6 0.091863021 0.882100242 0.687177651 0.237912303 0.422883975 0.686301071 0.501373043

Table 2 . Mutual correlations between experiments: first experiment, 01.02.2015g. (00-01 hours); Second experiment, 10.02.2015g. (10-11 hours); Third experiment, 01.02.2015g. (03-04 hours); Fourth experiment, 10.02.2015g. (12-13 hours).

We can conclude that as a whole and all four experiments show relatively same picture contrast of the intensity of the field of Rayleigh waves.

4. Discussion and Conclusions

The velocity contrast established by the SSMT method presented on fig.2 shows several peculiarities:The low velocities of the micro seismic tremors are located at low depths (up to about 50-70 km), where the Moho boundary could be expected. Then a middle contrast area follows down to 200 km In this depth interval the mid contrast area enveloped the well expressed upper part containing the seismic source body from about 80-90 km, down to 140-150 km. The lower 50 km are not seismic active, which means that the seismogenic properties are almost eliminated. The high velocity strata follows deeper parts down to

250-300 km, but still inhomogeneous. After these depths the mantle looks more homogeneous. In general the deep environment of the seismic source Vtancea is characterized by high level of inhomogenity, which is a good reason to expect effective results applying the SSMT method. The comparison of the results obtained with the results of other investigators, using different methods (geothermic, gravity and P-waves velocities), show relatively good coincidence of the concluded peculiarities of the depth behavior of the Vrancea seismic zone.To assess the accuracy of the method, several experimental models have been performed. They confirm the announced accuracy by others [10] in the range of 15-20%.The results obtained show that a relatively simple procedure using the spectrum of the microtremor background noise, the visualization of the intensity of the measured signal on 3D schemes and “connection” the obtained depth by each of the spectral frequency (

), the velocity changes can be determined relatively reliable. If so, such a methodology is a new and effective tool for the deep structures outlining in complicated geological conditions – as Vrancea seismic source for example. It is still difficult to discover the entire earthquakes generation mechanism of the Vrancea source. The very narrow space location is not typical for the convenient subduction zones. The lack of recent active volcanism, the hypocenters of strong events in the depth range from surface to 70-80 km, and the high frequency of the seismic energy emission, does not coincide to the “standard” model of a subduction zone.

REFERENCES

[1] Balla Z. Tertiary palaeomagnrtic data for the Carpstho-Pannonian region in the light of Miocene rotation n kinematics // Tectonophysics. 1987. Vol.139 (1-2). P. 67-98.[2] Csontos L., Nagymarosy A., Horvath F., Kovac M. Tertiary evolution of the intra-Carpathian area; a model // Tectonophysics. 1992. Vol. 208 (1-3). P.221-241.[3] Wenzel F., Achauer U., Enescu D. et al. Detailed look at final stage break-off is target of study in Romania // Eos. Transaction. AGU. 1998. Vol. 79 (48). P.589.[4] Fan G., Wallace T.C., Zhao D. Tomographic imaging of deep velocity structure beneath the Eastern and Southern Carpathians, Romania; implications for continental collision // Journal of Geophysical Research. 1998. Vol. B103 (2). P. 2705-2723.[5] F.Hauser, V.Raileanu, W.Fielitz, ABala, C.Prodehl, G.Polonic, A.Schulze VRANCEA99-the crustal structure beneath the southeastern Carpathians and the Moesian Platform from a seismic refraction profile in Romania. Tectonophysics 340(2001) 233-256[6] Knapp J.H., Knapp C.C., Raileanu V. et al. Crustal constraints on the origin of mantle seismicity in the Vrancea zone, Romania: the case of active continental lithospheric delamination. Tectonophysics. 2005. Vol.410. P.311-323.[7] Горбатиков А.В. Патент на изобретение № RU2271554. „Способ сейсморазведки“. Дата приоритета 25.03.2005

[8] Старостенко В.И., Кендзера А.В., Цветкова Т.А., Бугаенко И.В. Промеждуточные землетясения зоны Вранча и скоростное строение мантии Восточной Европы. Геофизический журнал №3, Т.35, 2013[9] D.Ioane, D.Ion A 3D crustal gravity modeling of the Romanian territory. Journal of Balkan Geophysical Society. Vol..8. №4, November 2005. P.189-198[10] Gorbatikov A.V., N.V.Larin, E.I.Moiseev, A.V.Belyashov. The Microseismic Sounding Method: Application for the study of the buried diatreme structure. / / Doklady Earth Sciences, 2009, Vol. 428, No. 7, pp. 1222-1226