Scaling Properties Of The Vrancea (Romania) Seismic Source At Intermediate Depths

8/13/2019 Scaling Properties Of The Vrancea (Romania) Seismic Source At Intermediate Depths

1/16

Rev. roum. GOPHYSIQUE / Rom. GEOPHYS. J., tome 54, p. 317, 2010, Bucureti

SCALING PROPERTIES OF THE VRANCEA (ROMANIA)

SEISMIC SOURCE AT INTERMEDIATE DEPTHS

EMILIA POPESCU, FELIX BORLEANU, ANICA O. PLCINT,CRISTIAN NEAGOE, MIRCEA RADULIAN

National Institute of Research and Development for Earth Physics, P.O. Box MG-2,

077125 Bucharest, Romania; e-mail: [email protected]

For the present study we selected 126 earthquakes of moderate magnitudes (3.2 MD 6.2) occurred in the timeinterval 19972005 in the Vrancea seismic region at intermediate depths (62 h 166 km). Two relativedeconvolution techniques (spectral ratios and empirical Greens function) are applied to retrieve the sourceparameters. To this purpose, the data set was divided into pairs of main and empirical Greens function events. Wecould select 28 main events and 98 empirical Greens functions. Six main events are generated in the upper segmentof the subducting lithosphere (60 h < 110 km) and twenty two in the lower one (h 110 km). In all cases, thesignal/noise ratio as recorded by the Romanian seismic stations is acceptable for our purpose. Once the seismicsource parameters (seismic moment, source dimension, stress drop) are estimated, we investigate the scalingproperties for the Vrancea subcrustal source. A slight deviation from a homogeneous rupture process and significantvariations on depth and on magnitude of the stress drop and source size are emphasized. They can be tentativelyexplained by a more efficient seismic energy release in the deeper segment of the subducting lithosphere and by therole played by fluids at intermediate depth.

Key words: subcrustal earthquakes, source scaling, spectral ratios, empirical Greens function, Vrancea.

1. INTRODUCTION

Seismic source study cannot be achieved inmost of the cases other way than using theseismic motion effects recorded at the Earthsurface, therefore corrections for the propagationeffects on the hypocenter seismic station pathand for the site effects induced by the localstructure are fundamental to properly understandphysics of the source process.

The corrections are substantially moredifficult to introduce at high frequencies, whichare strongly influenced by the structural andprocess heterogeneities at small scale (muchmore difficult to understand and control than thelarge-scale heterogeneities). For this reason,most of the source studies are limited to lowfrequencies and the large-scale processes anddetailed source features are simply ignored andany step forward in extending the frequencyrange considered to higher values represents forseismologists a real challenge.

However, to simulate the strong groundmotion, which is of highest interest for seismic

hazard assessment and engineering purposes, weneed to introduce the high-frequency contributionsince this one plays an important role incharacterizing the earthquake motion both aswaveform in time and spectra.

Looking to the seismic signal at highfrequencies is the main objective of this paper.The chance of correctly decoding the source andstructure effects in the Vrancea region from therecorded seismograms is greater now because adense seismic network has recently been installedin Romania: 38 digital stations with high quality

recordings of moderate and strong earthquakes(Fig. 1) are operated at present by the NationalInstitute of Research and Development for EarthPhysics (www.inp.ro). Starting with 1995 thisnetwork has been continuously developed in theframework of Romanian-German program StrongEarthquakes: A Challenge for Geosciences andCivil Engineering.

We apply relative deconvolution methods(spectral ratios method and empirical Greensfunction method) to retrieve the source parameters


2/16

Emilia Popescu, Felix Borleanu, Anica O. Plcint, Cristian Neagoe, Mircea Radulian 24

from the acceleration records. Basically, themethods remove the path and instrument effects by

using pairs of earthquakes with close hypocenters

recorded by common stations. They are verysuitable for seismic sequences, characterized by

occurrence in space and time clusters.

Fig. 1 The K2 stations distribution.

2. DESCRIPTIONOF RELATIVE METHODS

The relative methods of the spectral ratios(SR) and empirical Greens function (EGF)deconvolution are widely used in analyzingground motion recordings at earthquakes. Themethods are applied to pairs of earthquakeslocated approximately in the same place andrecorded by common seismic stations. Thisclass of methods allows the elimination of the

propagation, site and instrument effects for themain event through the deconvolution of thewaveform of the smaller co-event, considered asEGF. A pair main EGF events should fulfillthe following conditions:

waveforms should be similar (therefore thefocal mechanisms should be similar);

hypocenters should be close relative to eachother;

the difference between the magnitude of mainevent and EGF event should be at least one unit;

the width of the EGF pulse should besufficiently small relative to the width of mainevent pulse (to approximate EGF with a delta-type function).

EGF and SR methods have been developedrelatively recently in seismology and have beenapplied in a series of seismic zones of the World(e.g., Mueller, 1985; Frankel et al., 1986; Houghet al., 1989; Mori and Frankel, 1990). In additionto EGF method, the SR method allows thesimultaneous determination of the source parametersfor both main and EGF event in a selected pair,

if broadband and high signal-to-noise ratiorecordings are available (Lindley, 1994).For a source model with uniform rupture and

spectral fall-off at higher frequencies of -2type, the spectral ratio can be approximated bythe theoretical function:

( )

( )1/22P

cG

0

1/22Gc

P

0

f/f1

f/f1

R(f)

+

+

= (1)


3/16

3 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 5

where 0P, 0

G are low-frequency levels ofdisplacement amplitude spectra of the principal

event and Greens event, respectively, and fcP

,fcGare the corner frequencies of the two events.

They are the spectral free parameters which aredetermined by a procedure of non-linear regressionwhich find the best fit of the observed spectralratios with a function of type (1). The ratio ofthe low-frequency spectral levels is equivalentto the ratio of the seismic moments.

According to Brune (1970), the cornerfrequency is directly related to the size of therupture area:

r= 0.28/fc (2)

r representing the equivalent radius of the sourceand , the velocity of shear waves at sourcedepth. Thus, knowing the corner frequencies forthe main and EGF events, we can apply (2) toestimate the source dimension (average radius).

Once we know the seismic moment andsource radius values, we can evaluate the sourcestress drop (Brune, 1970):

30

B16r

7M= (3)

Alternatively, applying the EGF technique,we can estimate in time domain the width of thesource pulse, which is a measure of the durationof the rupture process. In this case, the radius ofthe source was determined using the formula ofBoatwright (1980):

r = (1/2v)/(1-v/sin) (4)

where 1/2is the rise time (approximately half ofthe source duration), v is the rupture velocity(taken as a fraction of the shear wave velocityat the source depth), is the P-wave velocity in

source, is the angle between the normal to thefault and P-wave emergent direction.

3. DESCRIPTION OF DATA SETAND DETERMINATION

OF SOURCE PARAMETERS

The data set selected to estimate the sourceparameters consists of 126 intermediate-depthearthquakes (62 h 166 km) with magnitudes

range 3.2 MD 6.2. The magnitudes arecalculated from the duration measured on the

seismograms recorded by Vrncioaia (VRI) andMuntele Rou (MLR) stations, using the calibrationtechnique proposed by Trifu and Radulian (1991).This technique assures a homogeneous scale formagnitude. We consider the earthquakes separatedin two segments of the subducted lithosphere,the upper segment (60 h


4/16


5/16


Table 1

Hypocentral parameters of main shock of May 3, 2002 and the associatedempirical Greens function of May 15, 2002 (bold line is the main event)

Data hh:mm Lat(oN)

Lon(oE)

h(km)

MD

M 2002/05/03 18:31 45.57 26.33 162 5.2

EGF 2002/05/15 04:26 45.55 26.36 153 4.3

Table 2

Main event of 2002/05/03 and associated earthquake of 2002/05/15

Station a fcG

(Hz)fc

P

(Hz)rsp

G

(m)rsp

P

(m)1/2

P

(s)regf(m)

CFR 1.48 7.40 3.27 214 484 0.093 604GHR 1.40 4.50 2.02 352 784 0.098 640

VAR 1.48 6.32 2.62 306 720 0.114 744VRI 1.36 5.02 2.34 251 605 0.128 832TUD 1.40 6.32 2.62 352 714 0.091 594SIR 1.42 5.17 2.20 316 672 0.149 973Average 1.420.05 5.491.15 2.450.45 29956 663106 0.1120.023 731150

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5log f (Hz)

-1

0

1

2

3

log(sr)

main event: 2002.05.03, M=5.2, h=162 kmGreen's function:2002.05.15, M=4.3, h=153 km

CFR station

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

-1

0

1

2

3

GHR station

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

-1

0

1

2

3

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

-1

0

1

2

3SIR station TUD station

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

-1

0

1

2

3

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

-1

0

1

2

3VRI stationVAR station

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5log f (Hz)

-1

0

1

2

3

log(sr)

TUD GHR

SIR VAR

CFR

VRI

(a) (b)

Fig. 3 (a) The spectral ratios for the main earthquake of May 3, 2002 and the EGF of May 15, 2002 at different stations;(b) spectral ratio (dashed line) obtained by averaging the individual spectral ratios for the 6 available stations.


6/16


0.0 0.5 1.0 1.5 2.0 2.5 3.0

t (s)

-1.0

-0.5

0.0

0.5

1.0

relativeamplitude

CFR

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

TES

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

SIR

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

TUD

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

-1.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

GHR

VRI

VAR

Main event: May 3, 2002

Green's function: May 15, 2002

0.0 0.5 1.0 1.5 2.0t (s)

-0.5

0.0

0.5

1.0

sp

ectralamplitude

Average source time function of the main event (continous line)on May 3, 2002 deconvolved with May 15, 2002, empirical Green'sfunction; dashed line represent standard deviation

(a) (b)

Fig. 4 (a) Apparent source time functions of the main earthquake of May 3, 2002 resulted from the deconvolution withevent Green function of May 15, 2002 for the common stations available; (b) Average source time function in the case

of the same earthquake (continuous line) and the standard error (dashed line).

Table 3

Source parameters and associated errors for the considered main events; Nsr number of values used to obtain the average

estimation by SR technique; Negf number of values used to obtain the average estimation by EGF technique.No. Data fc

(Hz)rsr(m)

Nsr 1/2(s)

regf(m)

Negf

1 1997/10/11 3.350.82 571136 23 0.1100.027 729177 122 1997/11/18 2.530.27 74282 10 0.1500.02 952148 83 1997/12/30 3.560.59 53699 8 0.0900.020 595129 64 1998/01/19 4.250.81 44084 8 0.0860.005 56434 45 1998/03/13 2.400.40 795158 11 0.1200.030 793201 106 1998/07/27 2.830.69 712270 11 0.1300.020 846117 67 1999/03/22 2.790.51 687130 12 0.1210.015 78795 48 1999/04/28 1.840.49 1051344 57 0.1560.036 1015233 269 1999/04/29 4.300.83 44378 5 0.1000.041 653270 6


7/16


Table 3(continued)

10 1999/06/29 3.080.76 635149 11 0.1010.025 659160 10

11 1999/11/08 2.910.15 63933 11 0.1170.020 766129 1212 1999/11/14 3.030.71 658216 14 0.1140.021 745140 413 2000/03/08 2.550.68 756185 17 0.1150.004 755 36 414 2000/04/06 1.880.46 1044245 58 0.1600.028 1047189 3615 2000/05/10 3.320.69 581121 13 0.0770.015 50499 1616 2001/03/04 1.790.39 1196292 18 0.1600.020 1065154 1317 2001/05/24 1.730.41 1131270 37 0.1500.040 967237 2618 2001/07/20 1.870.47 1063302 46 0.1400.030 918210 3419 2001/10/17 2.900.63 649130 7 0.1240.005 80831 420 2002/05/03 2.450.45 663106 6 0.1120.023 731150 621 2002/09/06 3.761.03 501137 2 0.0760.019 498121 322 2002/11/30 1.920.72 933340 39 0.1660.047 1087308 1423 2003/10/05 3.270.56 586114 14 0.1220.036 795 238 1324 2004/02/07 3.170.40 59073 8 0.1400.060 848342 7

25 2004/07/10 2.920.58 661138 19 0.1310.038 878241 1426 2004/09/27 2.160.36 882153 39 0.1200.020 782142/ 3827 2004/10/27 1.600.46 1222360 74 0.1700.040 1140242 4728 2005/05/14 1.720.57 1190358 59 0.1610.045 1052292 41

-0.5

0.0

0.5

1.0

relativ

eamplitude

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0t (s)

0.0 0.5 1.0 1.5 2.0

CFR GHR TES

0.0 0.5 1.0 1.5 2.0

-0.5

0.0

0.5

1.0

0.0 0.5 1.0 1.5 2.0

-0.5

0.0

0.5

1.0

TUD VAR

0.0 0.5 1.0 1.5 2.0

-0.5

0.0

0.5

1.0

VRI

Fig. 5 Approximation of source time function for the main earthquake of May 3, 2002with a theoretical function of the type sin x/x.


8/16


Table 4

The source parameters and associated errors for the EGF events estimated through spectral ratios method

N Data fc

(Hz)r

(m)1 1997/03/19 6.822.16 290722 1997/07/14 4.19 4433 1997/11/11 8.241.34 224384 1997/12/18 5.30 3505 1998/01/14 9.012.69 234526 1998/01/31 7.611.93 258667 1998/02/19 4.710.40 397428 1998/03/06 6.020.03 3081.419 1998/06/06 6.90 26910 1998/08/24 11.60 16011 1998/09/21 7.472.05 26468

12 1998/11/14 7.531.29 2524113 1998/12/12 5.651.93 36111814 1998/12/17 10.472.53 1895715 1998/12/28 9.651.69 1973816 1999/01/06 11.520.94 1621317 1999/01/09 9.162.29 2155618 1999/01/23 7.401.83 2596419 1999/03/09 7.140.40 2611520 1999/03/17 6.491.29 2865421 1999/03/23 3.880.71 4918922 1999/04/04 8.131.25 2333923 1999/04/15 9.110.96 2002024 1999/04/30 5.571.67 277116

25 1999/05/05 6.720.01 2770.7126 1999/06/06 11.23 16527 1999/06/22 6.572.22 31011328 1999/07/15 5.26 35329 1999/10/12 7.342.79 37422830 1999/11/24 10.943.84 1866731 1999/12/17 8.122.27 2427832 2000/05/13 7.702.34 2599733 2000/05/28 10.620.95 1661734 2000/07/01 6.881.96 2839035 2000/07/27 7.703.15 30219936 2000/08/06 6.332.00 2998037 2000/10/12 7.062.56 21750

38 2000/12/19 3.910.83 4899939 2000/12/28 3.940.40 4663340 2001/01/17 4.741.02 4078341 2001/02/03 10.222.96 2047942 2001/02/27 8.143.85 23711643 2001/03/18 4.642.12 45222444 2001/03/28 5.911.51 3157645 2001/05/20 5.101.21 39012646 2001/07/06 5.912.52 3209347 2001/07/23 4.741.41 42212848 2001/07/29 8.342.85 24061


9/16


Table 4 (continued)

49 2001/09/25 3.561.26 562134

50 2001/09/28 10.401.84 1783451 2001/10/17 5.904.5 40027352 2001/12/14 11.271.58 1622253 2002/01/25 4.811.25 4069754 2002/03/16 3.51.55 61524255 2002/05/15 5.831.52 2994756 2002/05/26 10.310.46 176857 2002/06/14 7.102.97 30111258 2002/07/14 8.212.10 2354859 2002/08/04 4.610.56 4064960 2002/08/05 6.421.11 2624461 2002/08/16 8.610.51 2161362 2002/08/27 9.043.43 22579

63 2002/09/10 4.271.72 47419064 2002/11/03 7.771.41 2444165 2002/11/27 7.331.13 2584166 2002/12/15 10.331.62 1782667 2002/12/23 14.720.23 123168 2002/12/30 6.312.25 29213669 2003/01/03 7.081.24 2694870 2003/01/05 9.301.76 2003771 2003/04/06 9.503.01 2075672 2003/05/02 7.272.26 28810573 2003/05/19 4.621.14 42310474 2003/05/26 9.760.61 1911275 2003/08/02 4.150.62 4556676 2003/08/27 4.941.44 41014577 2004/01/21 7.21087 2543078 2004/02/13 8.242.37 2468279 2004/03/17 3.801.38 50924980 2004/04/02 7.771.79 2394881 2004/04/04 4.390.87 4358182 2004/04/06 8.932.63 2115583 2004/04/15 10.662.34 1795784 2004/04/22 8.731.06 2152785 2004/06/03 14.50 12586 2004/07/02 9.483.82 2198787 2004/09/12 6.451.31 409194

88 2004/10/24 2.85 65189 2004/11/17 7.041.20 2704990 2005/01/10 10.412.91 1896791 2005/01/10 8.930.61 2021792 2005/01/29 14.351.59 1281593 2005/02/17 8.301.94 2295694 2005/03/06 3.940.75 49310895 2005/03/07 12.91 14496 2005/04/15 5.830.75 3379697 2005/05/09 6.931.85 2867798 2005/05/14 4.861.15 405108


10/16


4. SCALING OF SOURCE

PARAMETERS

On the basis of the source parameters obtainedas presented in the previous section (cornerfrequency or source radius and seismic moment)we estimate all the other parameters of interest(seismic moment, source radius, rupture duration,rise time, stress drop) using the conversion formula(2) (4) and the independent estimations of themagnitude and seismic moments for the mainevents.

The analysis of the source parameters scalingis carried out over the entire depth domain and

separately on two segments on depth, characteristicfor the Vrancea subcrustal seismogenic zone(Trifu, Radulian, 1994; Enescu, Enescu, 1998;Popescu et al., 2000): zone A, 60 h 110 km

and zone B, 110 < h 180 km. Notably, theseismic rate is approximately five times higher

in the lower segment (B) than in the uppersegment (A). Specifically for our data set, 28earthquakes (6 main events) occurred in zone A,while 98 earthquakes (22 main events) occurred inzone B. According to the previous investigations,the two segments on depth are assumed to becharacterized by different seismicity patternsand triggering mechanisms.

First we considered the scaling between theseismic moment and duration magnitude (MD),plotted in Fig. 6 (a). The duration magnitudewas estimated using the total duration of the

seismogram and the difference between S-wavearrival time and P-wave arrival time. Theequations of the regression lines are as follows:

Zone A: lg M0= (1.39 0.14)MD+ (8.730.61) (5)R = 0.88, = 0.47

Zone B: lg M0= (1.57 0.07)MD+ (7.56 0.32) (6)R = 0.92, = 0.38

Zone A+B: lg M0= (1.43

0.07)MD+ (8.26

0.30) (7)R = 0.89, = 0.43

The slope has values between 1.40 and 1.58,which is close to the theoretical value (1.5),obtained for shallow earthquakes. The tendencyof the slope of the regression line to increasewith increasing depth is probably due to the

increase of the stress drop and the more efficientrelease of the seismic energy at depth.

Similarly, we determined the relationshipbetween the seismic moment and the momentmagnitude (Mw), plotted in Fig. 6 (b). Theequations of the regression lines are given by:

Zone A: lg Mo= (1.510.15)Mw+ (8.83 0.58) (8)R = 0.89, = 0.46

Zone B: lg Mo= (2.05 0.07)Mw+ (6.45 0.40) (9)R = 0.90, = 0.40

Zone A+B: lg Mo= (1.80 0.09)Mw+ (7.53 0.34) (10)R = 0.88, = 0.44

Again the slope of the regression line tendsto increase with increasing depth (from 1.5 which is standard value, see Hanks and Kanamori,1979 to 2.0).

We determined subsequently the scaling ofthe seismic moment with the corner frequency(respectively, source radius), represented in Fig.7.The regression lines approximating the observationdata are given by:


11/16


Zone A: lg Mo= -(3.510.39) lg fc+ (17.59 0.36) (11)R = 0.87, = 0.49

Zone B: lg Mo= -(3.500.28) lg fc+ (17.24 0.21) (12)R = 0.79, = 0.58

Zone A+B: lg Mo = -(3.320.23) lg fc + (17.18 0.18) (13)R = 0.80, = 0.57

3.0 4.0 5.0 6.0 7.0

M

13

14

15

16

17

18

logM

(Nm)

o

3.0 4.0 5.0 6.0 7.0

12

13

14

15

16

17

18

12

13

14

15

16

17

18

3.0 4.0 5.0 6.0 7.0

D

A zone

B zone

A+B zone

2.0 3.0 4.0 5.0 6.0

M

13

14

15

16

17

18

logM

o

3.0 4.0 5.0 6.0

M

12

13

14

15

16

17

18

W

12

13

14

15

16

17

18

2.0 3.0 4.0 5.0 6.0 7.0

W

A zone

B zone

A+B zone

(a) (b)

Fig. 6 Seismic moment magnitude dependence: (a) magnitude from duration; (b) magnitude from seismic moment.

In all cases the slope is larger than 3 (absolutevalue), which is the characteristic slope for auniform rupture process in the source. Thisdeviation suggests a slightly complex process of

breaking for the earthquakes in the Vranceasubcrustal domain.

The scaling of the seismic moment and stressdrop is represented in Fig. 8. The regressionlines are given by:


12/16


Zone A: lg = (0.33 0.07) lg Mo- (3.83 1.07) (14)R = 0.66, = 0.49

Zone B: lg = (0.41 0.04) lg Mo- (5.49 0.59) (15)R = 0.73, = 0.37

Zone A+B: lg = (0.38 0.04) lg Mo- (4.84 0.57) (16)R = 0.66, = 0.41

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

log f (Hz)

13

14

15

16

17

18

lo

gM

(Nm)

c

o

12

13

14

15

16

17

18

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

12

13

14

15

16

17

18

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2

zone A

zone

zone A+

12 13 14 15 16 17 18

log M (Nm)

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

log

(MPa)

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

12 13 14 15 16 17 18

o

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

12 13 14 15 16 17 18

zone A

zone B

zone A+B

Fig. 7 Seismic moment-corner frequencydependence for the two segments of

subducted lithosphere and for the whole area.Fig. 8 Stress drop seismic moment dependence.


13/16


The standard scaling implies a constantstress-drop, independently of source size (slope

zero). For our data set and for the magnituderange involved, 3.2 MD 6.2 (2.9 MW 6.0), the scaling indicates an increasing stressdrop with increasing earthquake size. A similarresult was obtained by Popescu et al. (2003a)for a data set of 28 Vrancea events occurred inthe time interval 19972000.

The increase of stress drop with increasing

source size shows a more efficient and rapidrelease of strain for the larger earthquakes than

the smaller ones. Perhaps, this result indicatesthe role of fluids in triggering the larger events, likein a percolation process (Trifu, Radulian, 1991).

Finally, we analyzed the scaling of sourceduration (more precisely, half of the sourceduration which approximates the rise time) withsource size (Fig. 9). The linear dependencies aregiven by:

Zone A: 1/2= (0.030 0.08) lg Mo- (3.38 0.12) (17)R = 0.91, = 0.02

Zone B: 1/2= (0.029 0.06) lg Mo- (0.33 0.10) (18)R = 0.71, = 0.018

Zone A+B: 1/2= (0.030 0.05) lg Mo- (0.36 0.08) (19)R = 0.77, = 0.017

15 16 17 18

log M (Nm)

0.05

0.10

0.15

0.20

(s)

1/2

o

0.05

0.10

0.15

0.20

15 16 17 18

0.05

0.10

0.15

0.20

15 16 17 18

zone A

zone B

zone A+B

Fig. 9 Scaling of rise time vs. seismic moment.


14/16


The increase of 1/2 with source size is practicallythe same all along the Vrancea subducting

lithosphere, as shown by relations (17) (19).In support of this hypothesis comes and forma complex function of time the source obtainedin the case of empirical Green functiondeconvolution.

The RS method and the EGF deconvolutionprovide two independent ways to estimate thesource radius formulae (2) and (4). Fig. 10 showsdependence between the two radius estimatesfor the 28 main events considered. Note that theradius obtained from the rise time tends to behigher than that obtained from spectral ratios at

smaller magnitudes and becomes smaller athigher magnitudes.

lg regf= (0.63 0.08) lg rsr+ (1.11 0.22), (20)R = 0.84, = 0.05

2.5 2.6 2.7 2.8 2.9 3.0 3.1

log r

2.5

2.6

2.7

2.8

2.9

3.0

3.1

log

r

RS

RT

Fig. 10 The relationship between the source radius of themain earthquakes as obtained from deconvolution with

empirical Greens function and from spectral ratios.

The errors of determination were alsoestimated and in the case of corner frequencyand source radius of 98 earthquakes used asempirical Greens functions using spectral ratiosmethod. In this case the errors are larger and arepresented in Table 4. The errors in the case ofcorner frequencies and the source radius ofempirical Green functions are contained in therange of [0.2, 76%].

5. CONCLUSIONS

The main objective of the present work is toanalyze the Vrancea seismic source characteristicsfrom high-frequency recordings of moderatesize earthquakes using the improvements of thedigital stations operated by the National Instituteof Research and Development for Earth Physics(Bucharest) recently installed on the Romaniaterritory (Fig. 1). The seismic source parametersare retrieved applying to alternate deconvolutionmethods: spectral ratios and empirical Greensfunction. These are relative techniques which veryefficient in removing the path and instrumenteffects if adequate pairs of earthquakes withclose hypocenters recorded by common stationsare available.

The analysis is made on a set of 126intermediate-depth earthquakes recorded between1997 and 2005 (62 h166 km; 3.2 MD6.2).The source corner frequency, half duration (risetime) and seismic moment are first estimated forthe 28 earthquakes selected as main events. Thenthe source radius and stress drop are computedusing standard relationships (Brune, 1970;Boatwright, 1980). Finally, different scalingrelationships are determined on the basis of the

source parameter values.The variation of scaling properties on depth

(in this study we simply considered two activesegments on depth, associated to the triggeringof Vrancea major events) indicate a tendency ofincreasing stress drop with depth and respectivelyof a more efficient seismic energy release in thedeeper part of the Vrancea subducting slab. Thisbehavior correlates also with the decreasenoticed in the slope of the frequency-magnitudedistribution (Popa, Radulian, 2001). The scalingof the seismic source with corner frequency (or

source duration) suggests the presence of acertain complexity in the rupture process. Theincrease of stress drop with increasing sourcesize shows a more efficient and rapid release ofstrain for the larger earthquakes than the smallerones. Perhaps, this result indicates the role offluids in triggering the larger events, like in apercolation process (Trifu, Radulian, 1991).

Acknowledgements. This study benefited of the datarecorded by the K2 network installed in the framework ofCollaborative Research Center 461 Strong Earthquakes


15/16


of Karlsruhe University in collaboration with the NationalInstitute for Earth Physics, BucharestMgurele (Bonjer etal., 2000). The data recorded for each year are available on

CDs for the partners (Bonjer, Rizescu, 2000; Bonjer et al.,2002).

REFERENCES

BOATWRIGHT, J. (1980), A spectral theory for circularseismic source: simple estimates of source dimensions,dynamic stress drop and radiated energy,Bull. Seism.Soc. Am., 70, 128.

BONJER, K.-P., RIZESCU, M. (2000), Data Release 19961999 of the Vrancea K2 Seismic Network, Six CDswith evt-files and KMI v1-, v2-, v3-files. KarlsruheBucharest, July 15, 2000.

BONJER, K.-P., RIZESCU, M., GRECU, B. (2002),DataRelease 20002001 of the Vrancea K2 SeismicNetwork, Two CDs (evt-files). BucharestKarlsruhe,October 24, 2002.

BONJER, K.-P., ONCESCU, M.C., RIZESCU, M., ENESCU,D., DRIAD, L., RADULIAN, M., IONESCU, C.,MOLDOVEANU, T. (2000), Source- and site-parametersof the April 28, 1999 intermediate depth Vranceaearthquake: First results from the new K2 network inRomania, XXVII General Assembly of the EuropeanSeismological Commission, Lisbon, Portugal, Book ofAbstracts and Papers, SSA-2-13-O, p. 53.

BRUNE, J.N. (1970), Tectonic stress and the spectra ofseismic shear waves from earthquakes, J. Geophys.Res., 75, 49975009.

ENESCU, D., ENESCU, B.D. (1998), Seismotectonicmodel regarding the genesis and the occurrence of theVrancea (Romania) earthquakes, Rom. Rep. Physics,50, 97122.

FRANKEL, A., FLECHTER, J., VERNON, F., HAAR, L.,BERGER, J., HANKS, T., BRUNE, J. (1986), Rupturecharacteristics and tomography source imaging of ML= 3earthquakes near Anza, Southern California, J. Geophys.Res., 91, 1263312650.

HANKS, T. C., KANAMORI, H. (1979),A moment-magnitudescale, J. Geophys. Res., 84, 23482350.

HOUGH, S.E., JACOB, K., BUSBY, R. (1989), Groundmotions from a M = 3.5 earthquake near Massena, NewYork: Evidence for the poor resolution of corner frequencyfrom small events, Seismol. Res. Lett., 60, 9599.

LINDLEY, G.T. (1994), Source parameters of the 23 April1992 Joshua Tree, California earthquake, its largestforeshock and aftershocks, Bull. Seism. Soc. Am., 84,

10511057.MORI, J., FRANKEL, A. (1990), Source parameters for

small events associated with the 1986 North PalmSprings, California earthquake determined usingempirical Green functions, Bull. Seism. Soc. Am., 80,278285.

MUELLER, C.S. (1985), Source pulse enhancement bydeconvolutions with empirical Greens function,Geophys. Res. Lett., 12, 3336.

POPA, M., RADULIAN, M. (2001), Depth-dependentscaling relations of the source parameters for the

Vrancea subcrustal earthquakes, Rom. J. Phys., 46,499513.

POPESCU, E., BAZACLIU, O., RADULIAN, M.,

ARDELEANU, L. (2000), Clustering properties of theVrancea (Romania) intermediate depth seismicity, St.Cerc. Geofiz., 38, 4151, Publishing House of theRomanian Academy.

POPESCU, E., POPA, M., RADULIAN, M. (2003a),Efficiency of the spectral ratio method to constrain the

source scaling properties of the Vrancea (Romania)subcrustal earthquakes, Report in Physics, 55, Nos. 12, 163181.

POPESCU, E., GRECU, B., POPA, M., RIZESCU, M.,RADULIAN, M. (2003b), Seismic source properties:Indications of lithosphere irregular structure on depthbeneath Vrancea region, Romanian Reports inPhysics, 55, 458509.

RADULIAN, M., BONJER, K.-P., POPESCU, E., POPA,M., IONESCU, C., GRECU, B. (2007), The October27th, 2004 Vrancea (Romania) earthquake, ORFEUSNewsletter, vol. 1, nr. 7, January 2007.

TRIFU, C.-I., RADULIAN, M. (1991),Frequency magnitudedistribution of earthquakes in Vrancea: relevance for a

discrete model, J. Geophys. Res., 96, 43014311.TRIFU, C-I., RADULIAN, M. (1994),Dynamics of a seismic

regime: Vrancea a case history, in Nonlinear Dynamicsand Predictability of Geophysical Phenomena,Geophysical Monograph, 83, IUUG, vol. 18, eds. A.M.Gabrielov and W.I. Newman, AGU, 4353.

Received: October 12, 2010Accepted for publication: March 14, 2011


16/16

Scaling Properties Of The Vrancea (Romania) Seismic Source At Intermediate Depths

Documents

Transcript of Scaling Properties Of The Vrancea (Romania) Seismic Source At Intermediate Depths