You are on page 1of 16

SCALING PROPERTIES OF THE VRANCEA (ROMANIA) SEISMIC SOURCE AT INTERMEDIATE DEPTHS

EMILIA POPESCU, FELIX BORLEANU, ANICA O. PLĂCINTĂ, CRISTIAN NEAGOE, MIRCEA RADULIAN National Institute of Research and Development for Earth Physics, P.O. Box MG-2, 077125 Bucharest, Romania; e-mail: epopescu@infp.ro

For the present study we selected 126 earthquakes of moderate magnitudes (3.2 ≤ MD ≤ 6.2) occurred in the time interval 1997–2005 in the Vrancea seismic region at intermediate depths (62 ≤ h ≤ 166 km). Two relative deconvolution techniques (spectral ratios and empirical Green’s function) are applied to retrieve the source parameters. To this purpose, the data set was divided into pairs of main and empirical Green’s function events. We could select 28 main events and 98 empirical Green’s functions. Six main events are generated in the upper segment of the subducting lithosphere (60 ≤ h < 110 km) and twenty two in the lower one (h ≥ 110 km). In all cases, the signal/noise ratio as recorded by the Romanian seismic stations is acceptable for our purpose. Once the seismic source parameters (seismic moment, source dimension, stress drop) are estimated, we investigate the scaling properties for the Vrancea subcrustal source. A slight deviation from a homogeneous rupture process and significant variations on depth and on magnitude of the stress drop and source size are emphasized. They can be tentatively explained by a more efficient seismic energy release in the deeper segment of the subducting lithosphere and by the role played by fluids at intermediate depth. Key words: subcrustal earthquakes, source scaling, spectral ratios, empirical Green’s function, Vrancea.

1. INTRODUCTION Seismic source study cannot be achieved in most of the cases other way than using the seismic motion effects recorded at the Earth surface, therefore corrections for the propagation effects on the hypocenter – seismic station path and for the site effects induced by the local structure are fundamental to properly understand physics of the source process. The corrections are substantially more difficult to introduce at high frequencies, which are strongly influenced by the structural and process heterogeneities at small scale (much more difficult to understand and control than the large-scale heterogeneities). For this reason, most of the source studies are limited to low frequencies and the large-scale processes and detailed source features are simply ignored and any step forward in extending the frequency range considered to higher values represents for seismologists a real challenge. However, to simulate the strong ground motion, which is of highest interest for seismic

hazard assessment and engineering purposes, we need to introduce the high-frequency contribution since this one plays an important role in characterizing the earthquake motion both as waveform in time and spectra. Looking to the seismic signal at high frequencies is the main objective of this paper. The chance of correctly decoding the source and structure effects in the Vrancea region from the recorded seismograms is greater now because a dense seismic network has recently been installed in Romania: 38 digital stations with high quality recordings of moderate and strong earthquakes (Fig. 1) are operated at present by the National Institute of Research and Development for Earth Physics (www.inp.ro). Starting with 1995 this network has been continuously developed in the framework of Romanian-German program “Strong Earthquakes: A Challenge for Geosciences and Civil Engineering”. We apply relative deconvolution methods (spectral ratios method and empirical Green’s function method) to retrieve the source parameters

Rev. roum. GÉOPHYSIQUE / Rom. GEOPHYS. J., tome 54, p. 3–17, 2010, Bucureşti

1 – The K2 stations distribution. Mircea Radulian 2 from the acceleration records.. if broadband and high signal-to-noise ratio recordings are available (Lindley. This class of methods allows the elimination of the propagation. the spectral ratio can be approximated by the theoretical function: 1 +   R(f) = G Ω 0 1 +   Ω0 P (f/f ) (f/f ) c P c 1/2 G 2γ    2γ 1/2    (1) . 1990). the SR method allows the simultaneous determination of the source parameters for both main and EGF event in a selected pair.4 Emilia Popescu. Basically. In addition to EGF method. The methods are applied to pairs of earthquakes located approximately in the same place and recorded by common seismic stations. Hough et al. For a source model with uniform rupture and spectral fall-off at higher frequencies of ω-2 type. 1994). – the difference between the magnitude of main event and EGF event should be at least one unit. Plăcintă. the methods remove the path and instrument effects by using pairs of earthquakes with close hypocenters recorded by common stations. Fig. DESCRIPTION OF RELATIVE METHODS The relative methods of the spectral ratios (SR) and empirical Green’s function (EGF) deconvolution are widely used in analyzing ground motion recordings at earthquakes. They are very suitable for seismic sequences. 1986. 2.. site and instrument effects for the main event through the deconvolution of the waveform of the smaller co-event. A pair main – EGF events should fulfill the following conditions: – waveforms should be similar (therefore the focal mechanisms should be similar). Frankel et al. Cristian Neagoe. Mori and Frankel. 1989. Felix Borleanu. 1985. Anica O. EGF and SR methods have been developed relatively recently in seismology and have been applied in a series of seismic zones of the World (e. characterized by occurrence in space and time clusters. – the width of the EGF pulse should be sufficiently small relative to the width of main event pulse (to approximate EGF with a deltatype function).. – hypocenters should be close relative to each other.g. Mueller. considered as EGF.

2. By inverse FFT application we obtain the source time function of the main event. 98 events). From the total number of events. For the present study. Note at the same time that an event may act as EGF for different main events. The ratio of the low-frequency asymptotes. Subsequently we estimate the source radius on the basis of rise time value (eq. the seismograms are first corrected for instrument response. Generally. we can estimate in time domain the width of the source pulse. 28 events) and the lower segment (110 ≤ h <230 km. α is the P-wave velocity in source. Ω0G are low-frequency levels of displacement amplitude spectra of the principal event and Green’s event. v is the rupture velocity (taken as a fraction of the shear wave velocity β at the source depth).2. These operations were repeated for each pair of earthquakes and for each . The Fourier spectra of the main and associated EGF event and their spectral ratio are computed using FFT algorithm. In this case. 1970): ∆σ B = 7M 0 16r 3 (3) Alternatively. the velocity of shear waves at source depth. using the calibration technique proposed by Trifu and Radulian (1991). the radius of the source was determined using the formula of Boatwright (1980): r = (τ1/2v)/(1-v/αsinθ) (4) where τ1/2 is the rise time (approximately half of the source duration). and fcP. applying the EGF technique. DESCRIPTION OF DATA SET AND DETERMINATION OF SOURCE PARAMETERS The data set selected to estimate the source parameters consists of 126 intermediate-depth earthquakes (62 ≤ h ≤ 166 km) with magnitudes range 3. An example of a pair main event – EGF is presented in Fig. fcG. Once we know the seismic moment and source radius values. The source radius estimate for a given main event is the average of the radii obtained for each individual pair (in case there are several EGFs for the main event) and for all the considered stations. we can apply (2) to estimate the source dimension (average radius). θ is the angle between the normal to the fault and P-wave emergent direction.28 β/fc (2) r representing the equivalent radius of the source and β.3 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 5 where Ω0P. The ratio of the low-frequency spectral levels is equivalent to the ratio of the seismic moments. We consider the earthquakes separated in two segments of the subducted lithosphere. To apply the EGF deconvolution procedure. we can evaluate the source stress drop (Brune. which is a measure of the duration of the rupture process. Thus. respectively. only P-wave recordings are considered in the relative deconvolution procedures. are estimated by approximating the theoretical function of the relationship (1) with the observed ratios. a. segment A. In parallel we applied the spectral ratios technique to the same event pairs. then P-wave windows are selected and cosine tapering process is applied to the window edges. the corner frequency is directly related to the size of the rupture area: r = 0. the upper segment (60 ≤ h <110 km. 3. The magnitudes are calculated from the duration measured on the seismograms recorded by Vrâncioaia (VRI) and Muntele Roşu (MLR) stations. According to Brune (1970). fcG are the corner frequencies of the two events. segment B. For each of them. This technique assures a homogeneous scale for magnitude. we can find at least one co-located event (EGF). knowing the corner frequencies for the main and EGF events. the source time function is like a simple pulse with the width measuring the source duration and half-width time (τ1/2) measuring the rise time. 28 are considered as main events.2 ≤ MD ≤ 6. They are the spectral free parameters which are determined by a procedure of non-linear regression which find the best fit of the observed spectral ratios with a function of type (1). the corner frequency of the main earthquake fcP as well as the corner frequency of the EGF event. 4).

5). Note that when we use as co-located event an aftershock (an event occurred in a small time interval after the main shock occurrence) with similar focal mechanism as the main event.05. 2004).05. which generally is better determined from independent methods. The method is relative and does not allow simultaneous estimation of absolute values of seismic moments for both earthquakes.15 0 5 10 15 20 25 30 t(s) VRI2z-2002. 2003b. Radulian et al.05.03 20 25 30 0 5 10 15 GHR2Z-2002. The values of parameters were estimated as the average of all determinations available. . The source parameters estimated separately for each station and as average are presented in Table 2.. VRI2Z-2002. and his co-located aftershock of 15 May 2002). It is necessary to have independently a reference moment value for one of the two earthquakes in a pair. We present in detail the results obtained by the two methods for the pair of events given in Table 1 (3 May 2002.. the resulted pulse from deconvolution is well constrained. Contrary to EGF procedure.15 0 5 10 15 GHR2Z-2002. 4.6 Emilia Popescu.03 0 -8000 1600 0 -1600 400 0 -400 40 0 -40 200000 0 -200000 0 10000 0 -10000 0 5 10 15 20 25 30 5 10 15 SIR2Z-2002.. Mircea Radulian 4 seismic station. The source parameters obtained by the EGF deconvolution and SR technique together with the associated errors are given in Tables 3 (for the main events) and 4 (for the EGF events). We adopted the seismic moments estimated from the spectral analysis of the moderate earthquakes from Vrancea region (Popescu et al. 2002 (Fig.05. the SR 8000 amplitude approach allows the estimation of the corner frequency (source radius) for the EGF event in addition to that of the main event. Anica O.15 20 25 30 0 5 10 15 SIR2Z-2002.05.03 20 25 30 20 25 30 Fig. 3 and Fig. We can observe also the perfect shape of the average of the source time function in this case. 2003a. SIR). The spectral ratios and the source time functions are shown in Fig.05. Cristian Neagoe. Felix Borleanu. Popescu et al. We preferred to use the seismic moment of the main event. GHR. 2 – Waveforms of the main event of 3 May 2002 and the EGF event of 15 May 2002 recorded at three stations of the K2 network (VRI. suggesting a simple uniform rupture process for the event of May 3. respectively. Plăcintă.

02 2.5 3 2 1 0 -1 -1.5 TUD log (sr) 1 CFR GHR SIR station TUD station SIR VRI VAR 0 1 0 -1 -1.5 log f (Hz) 1.45± 0.5 1.5 0. 3 – (a) The spectral ratios for the main earthquake of May 3.40 1.0 -0.36 1.0 -0.20 2.02 6.5 (a) (b) Fig.05. 2002 and the EGF of May 15.15.128 0.0 0.5 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 7 Table 1 Hypocentral parameters of main shock of May 3.27 2.0 0.48 1. M=4.62 2.093 0. h=162 km Green's function:2002. M=5.55 Table 2 Main event of 2002/05/03 and associated earthquake of 2002/05/15 Station CFR GHR VAR VRI TUD SIR Average a 1.33 26.5 VAR station VRI station -1.5 1. .42± 0.5 1.0 0.0 1.0 1. 2002 and the associated empirical Green’s function of May 15.5 -1.5 1.5 log f (Hz) 3 2 1 0 -1 -1.15 fcP (Hz) 3.36 h (km) 162 153 MD 5.0 -0.0 -0.5 0.3.5 0.3 main event: 2002.05 fcG (Hz) 7. h=153 km 3 3 CFR station 2 log (sr) 1 0 -1 -1.0 1.17 5. 2002 at different stations.32 5.112±0.5 -1. (b) spectral ratio (dashed line) obtained by averaging the individual spectral ratios for the 6 available stations.03.0 0.62 2.5 -1.34 2.0 0.40 4.2.5 3 2 2 1 0 -1 GHR station 3 2 -1.5 -1.023 regf (m) 604 640 744 832 594 973 731± 150 Lon (oE) 26.5 1.5 -1.0 -0.114 0.5 0.42 1.5 0.0 1.0 -0.32 5.0 0.0 -0.5 0.49± 1.45 rspG (m) 214 352 306 251 352 316 299±56 rspP (m) 484 784 720 605 714 672 663±106 τ1/2P (s) 0.5 -1.0 1.2 4.57 45.05.5 3 2 1 0 -1 1.5 -1.0 0.091 0.149 0.5 -1 -1.0 1.098 0.40 1. 2002 (bold line is the main event) Data M EGF 2002/05/03 2002/05/15 hh:mm 18:31 04:26 Lat (oN) 45.5 0.48 1.50 6.0 1.

49 4.030 0.0 -0.83 rsr (m) 571±136 742±82 536±99 440±84 795±158 712±270 687±130 1051±344 443±78 Nsr 23 10 8 8 11 11 12 57 5 τ1/2 (s) 0.0 1.5 3.0 1.5 0.0 0.0 0. 2002 Green's function: May 15.8 Emilia Popescu.5 0.036 0.0 VAR 0.5 1.0 1.0 1.041 regf (m) 729±177 952±148 595±129 564±34 793±201 846±117 787±95 1015±233 653±270 Negf 12 8 6 4 10 6 4 26 6 .020 0.027 0.5 1.0 -0.0 t (s) 1. No.0 0.5 1.5 -1.0 2. Negf – number of values used to obtain the average estimation by EGF technique.40±0.35±0. Anica O.5 2.83±0. 2002 resulted from the deconvolution with event Green function of May 15. Felix Borleanu.5 -1.0 CFR 1.0 1.121±0.0 -0.5 0.0 0.0 2.5 -1.0 -0.0 (a) (b) Fig.0 1.5 3.5 1.5 -1. empirical Green's function.5 0.0 2. 4 – (a) Apparent source time functions of the main earthquake of May 3.0 2.0 0.0 -0.086±0.5 0.0 2.0 0.0 0.27 3.5 1.0 2.015 0.0 0. dashed line represent standard deviation 0.0 0.5 3.0 0.0 GHR 6 relative amplitude 0.5 3.0 -0.40 2.0 1. Plăcintă.5 TES spectral amplitude 0.59 4.5 3.56±0. 1 2 3 4 5 6 7 8 9 Data 1997/10/11 1997/11/18 1997/12/30 1998/01/19 1998/03/13 1998/07/27 1999/03/22 1999/04/28 1999/04/29 fc (Hz) 3.84±0.0 0.5 0.30±0.5 2.5 1.0 SIR 1.0 t (s) 1.82 2.0 VRI TUD 1.25±0.150±0. Cristian Neagoe.0 2.0 0. Nsr – number of values used to obtain the average estimation by SR technique.0 0.5 2.110±0.5 -1.0 0.0 1.120±0.81 2.0 0. Mircea Radulian Main event: May 3.0 0.0 1.020 0. 2002 deconvolved with May 15.0 1.5 1.5 2. Table 3 Source parameters and associated errors for the considered main events.5 1. (b) Average source time function in the case of the same earthquake (continuous line) and the standard error (dashed line).5 2.0 -0.156±0. 2002 1.0 0.5 2.79±0. 2002.5 2.5 0.5 2.0 0.02 0.53±0. 2002 for the common stations available.130±0.5 -1.5 3.0 0.005 0.090±0.0 -0.5 3.5 -1.0 0.0 Average source time function of the main event (continous line) on May 3.100±0.5 0.69 2.51 1.

0 -0.5 0.60±0.077±0.0 0.122±0.120±0.5 2.0 0.92±0.0 VAR 1.69 1.112±0.0 0.0 1. 2002 with a theoretical function of the type sin x/x.57 635±149 639±33 658±216 756±185 1044±245 581±121 1196±292 1131±270 1063±302 649±130 663±106 501±137 933±340 586±114 590±73 661±138 882±153 1222±360 1190±358 GHR 11 11 14 17 58 13 18 37 46 7 6 2 39 14 8 19 39 74 59 0.040 0.68 1.036 0.5 1. .5 0.0 -0.46 3.020 0.5 2.76±1.0 -0.020 0.038 0.39 1.5 1.040 0.63 2.5 0.5 2.5 1.5 2.90±0.0 Fig.16±0.020 0.161±0.030 0.150±0.5 1.5 0.0 0.72±0.56 3.0 9 3.15 3.140±0.73±0.0 0.170±0. 5 – Approximation of source time function for the main earthquake of May 3.03 1.87±0.0 1.0 0.5 0.131±0.015 0.5 0.005 0.91±0.117±0.140±0.124±0.5 2.5 0.114±0.0 0.5 1.101±0.79±0.0 1.0 -0.0 0.0 relative amplitude 0.5 2.0 t (s) TUD 1.5 0.0 0.0 -0.46 1.0 1.45± 0.08±0.166±0.115±0.076±0.0 1.021 0.76 2.0 0.004 0.32±0.0 -0.40 2.023 0.060 0.028 0.5 0.03±0.0 0.0 VRI 1.36 1.019 0.5 0.7 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths Table 3 (continued) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 1999/06/29 1999/11/08 1999/11/14 2000/03/08 2000/04/06 2000/05/10 2001/03/04 2001/05/24 2001/07/20 2001/10/17 2002/05/03 2002/09/06 2002/11/30 2003/10/05 2004/02/07 2004/07/10 2004/09/27 2004/10/27 2005/05/14 CFR 1.88±0.5 0.47 2.047 0.5 0.17±0.5 1.045 659±160 766±129 745±140 755 ±36 1047±189 504±99 1065±154 967±237 918±210 808±31 731±150 498±121 1087±308 795 ±238 848±342 878±241 782±142/ 1140±242 1052±292 TES 10 12 4 4 36 16 13 26 34 4 6 3 14 13 7 14 38 47 41 1.0 1.160±0.92±0.55±0.45 3.58 2.0 1.025 0.41 1.27±0.160±0.71 2.72 3.

85 r (m) 290±72 443 224±38 350 234±52 258±66 397±42 308±1.41 8.23 6.02±0.67 6. Cristian Neagoe.29 5.74±1. Felix Borleanu.41 269 160 264±68 252±41 361±118 189±57 197±38 162±13 215±56 259±64 261±15 286±54 491±89 233±39 200±20 277±116 277±0.96 5.15 6.10 Emilia Popescu.12±2.22 5.13±1.65±1.01 11.47±2.52±0.06±2.14±0.16±2.96 7.11±0. Plăcintă.40 6.33±2.24±1.91±2.25 9.29 7.71±0.22±2.72±0.01±2.74±1.21 5.83 7.64±2.00 7.94±3.16 4.91±1.27 7.47±2.90 11.40 6.70±3.26 7. Mircea Radulian Table 4 The source parameters and associated errors for the EGF events estimated through spectral ratios method N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 Data 1997/03/19 1997/07/14 1997/11/11 1997/12/18 1998/01/14 1998/01/31 1998/02/19 1998/03/06 1998/06/06 1998/08/24 1998/09/21 1998/11/14 1998/12/12 1998/12/17 1998/12/28 1999/01/06 1999/01/09 1999/01/23 1999/03/09 1999/03/17 1999/03/23 1999/04/04 1999/04/15 1999/04/30 1999/05/05 1999/06/06 1999/06/22 1999/07/15 1999/10/12 1999/11/24 1999/12/17 2000/05/13 2000/05/28 2000/07/01 2000/07/27 2000/08/06 2000/10/12 2000/12/19 2000/12/28 2001/01/17 2001/02/03 2001/02/27 2001/03/18 2001/03/28 2001/05/20 2001/07/06 2001/07/23 2001/07/29 fc (Hz) 6.19 8.91±0.93 10.52 4.30 9.70±2. Anica O.71 165 310±113 353 374±228 186±67 242±78 259±97 166±17 283±90 302±199 299±80 217±50 489±99 466±33 407±83 204±79 237±116 452±224 315±76 390±126 320±93 422±128 240±61 8 .82±2.94±0.05 7.79 10.53±1.69 7.53 9.61±1.62±0.10±1.14±3.34±2.49±1.96 8.29 3.02 10.83 3.57±2.03 6.34±2.34 5.69 11.71 8.40 4.12 5.65±1.88±0.56 3.85 4.40±1.88±1.51 5.57±1.94 9.93 4.34 10.84 8.60 7.95 6.

50±3.93±1.30±1.21±2.84 5.33±1.10 4.30±1.27±1.51 9.85 4.43 4.83±1.77±1.61±0.94±1.90±4.55 5.97 8.61±0.06 14.86±1.62 4.33±1.24 9.94 3.94±0.79 4.08±1.44 7.31±2.87 8.27±2.76±0.45±1.38 7.42±1.56 6.48±3.72±0.15±0.76 9.9 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths Table 4 (continued) 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 2001/09/25 2001/09/28 2001/10/17 2001/12/14 2002/01/25 2002/03/16 2002/05/15 2002/05/26 2002/06/14 2002/07/14 2002/08/04 2002/08/05 2002/08/16 2002/08/27 2002/09/10 2002/11/03 2002/11/27 2002/12/15 2002/12/23 2002/12/30 2003/01/03 2003/01/05 2003/04/06 2003/05/02 2003/05/19 2003/05/26 2003/08/02 2003/08/27 2004/01/21 2004/02/13 2004/03/17 2004/04/02 2004/04/04 2004/04/06 2004/04/15 2004/04/22 2004/06/03 2004/07/02 2004/09/12 2004/10/24 2004/11/17 2005/01/10 2005/01/10 2005/01/29 2005/02/17 2005/03/06 2005/03/07 2005/04/15 2005/05/09 2005/05/14 3.59 8.75 12.39±0.31±0.26 4.13 10.41 7.61 4.91 8.66±2.73±1.01 7.11 8.04±1.80±1.58 4.63 10.34 8.46 7.23 6.04±3.5±1.40±1.25 3.75 6.41±2.21±087 8.61 14.93±0.35±1.24±2.72 7.93±2.20 10.25 7.83±0.91 5.31 2.50 9.62±1.81±1.5 11.85 7.15 562±134 178±34 400±273 162±22 406±97 615±242 299±47 176±8 301±112 235±48 406±49 262±44 216±13 225±79 474±190 244±41 258±41 178±26 123±1 292±136 269±48 200±37 207±56 288±105 423±104 191±12 455±66 410±145 254±30 246±82 509±249 239±48 435±81 211±55 179±57 215±27 125 219±87 409±194 651 270±49 189±67 202±17 128±15 229±56 493±108 144 337±96 286±77 405±108 11 .14 9.37 3.10±2.26 10.56±1.77±1.62 14.52 10.27±1.82 6.

Felix Borleanu.12 Emilia Popescu. Plăcintă.89. Specifically for our data set. 7.92.53 ± 0. rupture duration.15)Mw + (8. 110 < h ≤ 180 km.88. plotted in Fig. Enescu.80 ± 0.89. Similarly. 2000): zone A. 1979 – to 2. see Hanks and Kanamori. Notably. source radius).07)Mw + (6.83 ± 0. 28 earthquakes (6 main events) occurred in zone A.45 ± 0.51± 0. According to the previous investigations.. obtained for shallow earthquakes.0).46 Zone B: lg Mo = (2.34) R = 0.07)MD + (8. Mircea Radulian 10 4. The equations of the regression lines are as follows: Zone A: lg M0 = (1. σ = 0. 60 ≤ h ≤ 110 km and zone B. Radulian.5 – which is standard value.40) R = 0. σ = 0. The tendency of the slope of the regression line to increase with increasing depth is probably due to the (5) (6) (7) increase of the stress drop and the more efficient release of the seismic energy at depth. 6 (a).38 Zone A+B: lg M0 = (1. The analysis of the source parameters scaling is carried out over the entire depth domain and separately on two segments on depth. characteristic for the Vrancea subcrustal seismogenic zone (Trifu.47 Zone B: lg M0 = (1. which is close to the theoretical value (1. σ = 0. We determined subsequently the scaling of the seismic moment with the corner frequency (respectively. SCALING OF SOURCE PARAMETERS On the basis of the source parameters obtained as presented in the previous section (corner frequency or source radius and seismic moment) we estimate all the other parameters of interest (seismic moment. σ = 0.61) R = 0.14)MD + (8. stress drop) using the conversion formula (2) – (4) and the independent estimations of the magnitude and seismic moments for the main events.5). the seismic rate is approximately five times higher in the lower segment (B) than in the upper segment (A). The regression lines approximating the observation data are given by: . The duration magnitude was estimated using the total duration of the seismogram and the difference between S-wave arrival time and P-wave arrival time.44 Again the slope of the regression line tends to increase with increasing depth (from 1. while 98 earthquakes (22 main events) occurred in zone B.39 ± 0.90. 1994.09)Mw + (7. Cristian Neagoe.43 The slope has values between 1. Anica O.58. σ = 0.26 ± 0. source radius.88.05 ± 0. plotted in Fig.40 Zone A+B: lg Mo = (1. 6 (b). 1998.43 ± 0. rise time. Popescu et al. we determined the relationship between the seismic moment and the moment magnitude (Mw).57 ± 0. First we considered the scaling between the seismic moment and duration magnitude (MD).58) R = 0.73± 0.07)MD + (7. σ = 0. The equations of the regression lines are given by: (8) (9) (10) Zone A: lg Mo = (1.56 ± 0.30) R = 0. Enescu. represented in Fig. the two segments on depth are assumed to be characterized by different seismicity patterns and triggering mechanisms.32) R = 0.40 and 1.

0 5.0 6.0 18 17 16 15 14 13 12 3.0 4. The regression lines are given by: .11 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 13 Zone A: lg Mo = -(3.39) lg fc + (17.0 6. 6 – Seismic moment – magnitude dependence: (a) magnitude from duration. which is the characteristic slope for a uniform rupture process in the source.0 3.0 13 MD 2.0 5.0 5. (b) magnitude from seismic moment.0 18 4.28) lg fc + (17.0 5. σ = 0.58 Zone A+B: lg Mo = -(3.0 7.0 4.0 18 4.0 7. σ = 0. σ = 0. 8.49 Zone B: lg Mo = -(3.24 ± 0.0 5. This deviation suggests a slightly complex process of breaking for the earthquakes in the Vrancea subcrustal domain.0 6.0 (a) (b) Fig.0 5.0 M W 17 16 15 A+B zone 17 16 A+B zone 15 14 14 13 13 12 3.23) lg fc + (17.57 18 (11) (12) (13) 18 17 A zone log M o 17 log M (Nm) A zone 16 16 o 15 15 14 14 13 3.0 MW B zone B zone 15 14 13 12 3.0 6.32± 0.0 7.80.50± 0.18) R = 0.59 ± 0. The scaling of the seismic moment and stress drop is represented in Fig.79.36) R = 0.0 7.87.0 12 2.0 4.0 18 17 16 3.18 ± 0.21) R = 0.0 6. In all cases the slope is larger than 3 (absolute value).0 4.51± 0.0 6.

8 0.5 12 13 14 zone 2.41 18 17 (14) (15) (16) 2. σ = 0.0 -0.0 12 13 14 15 16 17 18 Fig.0 1.5 -1.0 1.5 0. Cristian Neagoe.1 1.33 ± 0.59) R = 0.5 0.07) lg Mo .7 0.57) R = 0.14 Emilia Popescu. Plăcintă.73.0 0.9 1.5 -1.0 1.5 0. Felix Borleanu.(3.04) lg Mo .0 zone A+ 12 13 14 15 16 17 18 2.1 1.83 ± 1.2 zone A+B 0.5 0.8 0.2 o log fc (Hz) 18 17 16 15 14 13 12 0. Mircea Radulian 12 Zone A: lg ∆σ = (0.2 0.6 0.2 -0.4 0.5 2.49 Zone B: lg ∆σ = (0.3 0.5 0.9 1. Fig.7 0.0 log ∆σ(MPa) 1.37 Zone A+B: lg ∆σ = (0. σ = 0.9 1. Anica O.5 1.3 0.49 ± 0.0 -0.5 0.4 0.(5.5 zone A zone A 2.6 0.0 1.3 0.2 0.6 0.84 ± 0.41 ± 0.4 0.(4.8 0.7 0.38 ± 0. 7 – Seismic moment-corner frequency dependence for the two segments of subducted lithosphere and for the whole area.5 1. σ = 0.0 0.1 0.2 0.1 0.0 log M (Nm) 16 15 14 13 0.1 1.66.5 1.0 1.04) lg Mo .0 log Mo (Nm) 15 16 17 18 zone B 18 17 16 15 14 13 12 0. .1 0.5 2.66. 8 – Stress drop – seismic moment dependence.07) R = 0.

9). independently of source size (slope zero). The increase of stress drop with increasing source size shows a more efficient and rapid release of strain for the larger earthquakes than the smaller ones. . Perhaps.(0.20 zone B 0. Finally. we analyzed the scaling of source duration (more precisely. (2003a) for a data set of 28 Vrancea events occurred in the time interval 1997–2000.91.029 ± 0. 1991).018 Zone A+B: τ1/2 = (0. like in a percolation process (Trifu.13 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 15 The standard scaling implies a constant stress-drop.06) lg Mo .2 ≤ MD ≤ 6. 9 – Scaling of rise time vs.0). this result indicates the role of fluids in triggering the larger events. σ = 0.20 zone A 0.10 0.12) R = 0.030 ± 0.05 15 0.9 ≤ MW ≤ 6.36 ± 0.05) lg Mo . Radulian.017 0.15 τ1/2(s) 0. σ = 0.08) R = 0.10 0.(3.05 15 16 17 18 Fig.77.(0.20 zone A+B 0. A similar result was obtained by Popescu et al.15 16 17 18 0.08) lg Mo .33 ± 0.15 16 log M o (Nm) 17 18 0. seismic moment.05 15 0.38 ± 0. The linear dependencies are given by: (17) (18) (19) Zone A: τ1/2 = (0.030 ± 0.2 (2. the scaling indicates an increasing stress drop with increasing earthquake size. σ = 0.10) R = 0. half of the source duration which approximates the rise time) with source size (Fig.71. For our data set and for the magnitude range involved.02 Zone B: τ1/2 = (0.10 0. 3.

1 log r RS Fig. 3. associated to the triggering of Vrancea major events) indicate a tendency of increasing stress drop with depth and respectively of a more efficient seismic energy release in the deeper part of the Vrancea subducting slab. 1970. half duration (rise time) and seismic moment are first estimated for the 28 earthquakes selected as main events.6 2.6 2. . Felix Borleanu. Anica O. The errors in the case of corner frequencies and the source radius of empirical Green functions are contained in the range of [0. Cristian Neagoe. Radulian.5 2. Plăcintă. 76%]. The variation of scaling properties on depth (in this study we simply considered two active segments on depth. lg regf = (0.7 2.7 2. Mircea Radulian 14 The increase of τ1/2 with source size is practically the same all along the Vrancea subducting lithosphere.05 3. Then the source radius and stress drop are computed using standard relationships (Brune. like in a percolation process (Trifu.2.84.2). The seismic source parameters are retrieved applying to alternate deconvolution methods: spectral ratios and empirical Green’s function. 1991).8 2.08) lg rsr + (1. Fig.2 ≤ MD ≤ 6. 10 – The relationship between the source radius of the main earthquakes as obtained from deconvolution with empirical Green’s function and from spectral ratios. (20) R = 0. Perhaps.9 3. The errors of determination were also estimated and in the case of corner frequency and source radius of 98 earthquakes used as empirical Green’s functions using spectral ratios method. 2001). Boatwright.0 2. In this case the errors are larger and are presented in Table 4. CONCLUSIONS The main objective of the present work is to analyze the Vrancea seismic source characteristics from high-frequency recordings of moderate size earthquakes using the improvements of the digital stations operated by the National Institute of Research and Development for Earth Physics (Bucharest) recently installed on the Romania territory (Fig. This study benefited of the data recorded by the K2 network installed in the framework of Collaborative Research Center 461 ‘Strong Earthquakes’ log r RT 2. as shown by relations (17) – (19).8 2. Finally.11 ± 0.1 3. The RS method and the EGF deconvolution provide two independent ways to estimate the source radius – formulae (2) and (4).5 2. In support of this hypothesis comes and form a complex function of time the source obtained in the case of empirical Green function deconvolution. different scaling relationships are determined on the basis of the source parameter values. 1980). σ = 0. this result indicates the role of fluids in triggering the larger events. The scaling of the seismic source with corner frequency (or source duration) suggests the presence of a certain complexity in the rupture process. Note that the radius obtained from the rise time tends to be higher than that obtained from spectral ratios at smaller magnitudes and becomes smaller at higher magnitudes. The analysis is made on a set of 126 intermediate-depth earthquakes recorded between 1997 and 2005 (62 ≤ h ≤ 166 km. Acknowledgements.22).9 5. 10 shows dependence between the two radius estimates for the 28 main events considered. These are relative techniques which very efficient in removing the path and instrument effects if adequate pairs of earthquakes with close hypocenters recorded by common stations are available.16 Emilia Popescu. The increase of stress drop with increasing source size shows a more efficient and rapid release of strain for the larger earthquakes than the smaller ones. This behavior correlates also with the decrease noticed in the slope of the frequency-magnitude distribution (Popa. Radulian. 1).63 ± 0.0 3. The source corner frequency.

Am. 1. K. Cerc.. in “Nonlinear Dynamics and Predictability of Geophysical Phenomena”. M. POPESCU. Portugal. M. 2348–2350. RADULIAN. T. Southern California. 53. Source. 96. Ground motions from a M = 3.. Rep. BRUNE. 2011 . E. M. BRUNE.. BUSBY. Rizescu.. (2002). Nos.. Rupture characteristics and tomography source imaging of ML = 3 earthquakes near Anza. RADULIAN. Geofiz. 2004 Vrancea (Romania) earthquake.. (1980). 84. POPESCU.. C. nr. Book of Abstracts and Papers. T. Source parameters for small events associated with the 1986 North Palm Springs. Seismotectonic model regarding the genesis and the occurrence of the Vrancea (Romania) earthquakes. Clustering properties of the Vrancea (Romania) intermediate depth seismicity... HAAR. Bull. (2007). RADULIAN. (1979). M. A moment-magnitude scale. K. 12... 80. B. R. SSA-2-13-O. (2000). Rom.. dynamic stress drop and radiated energy.E.T.. 2002). Bonjer et al. Soc.. (1991). K. M.. (1994). Res. St. Received: October 12. ORFEUS Newsletter. M. MUELLER.. GRECU. Karlsruhe– Bucharest. Physics. 2000).. v3-files. California earthquake determined using empirical Green functions. D. C.D. J. 75. Soc. B. January 2007. RADULIAN. p. AGU. REFERENCES BOATWRIGHT. Res. Geophys. KANAMORI. D.... JACOB. M. Gabrielov and W. M..15 Scaling properties of the Vrancea (Romania) seismic source at intermediate depths 17 of Karlsruhe University in collaboration with the National Institute for Earth Physics. California earthquake. (2001). 2002. B. Bull. 1– 2. A spectral theory for circular seismic source: simple estimates of source dimensions. Publishing House of the Romanian Academy. Bull. Report in Physics. G. 1999 intermediate depth Vrancea earthquake: First results from the new K2 network in Romania.. 55.C. J. E. 41–51. IONESCU. 2000. 83. Res. Bucharest–Karlsruhe. H.-P.. M. T. The October 27th. Geophysical Monograph. J. Soc. Depth-dependent scaling relations of the source parameters for the Vrancea subcrustal earthquakes.N. POPA. M. July 15. v2-.M. 46. J. 60.. POPESCU. C. ENESCU. A. Am. Newman. MORI. BAZACLIU. Efficiency of the spectral ratio method to constrain the source scaling properties of the Vrancea (Romania) subcrustal earthquakes.I. HANKS. (2000). Seism. J. 7. Seism. Res. October 24. vol. Geophys. M.. M.. FRANKEL. (2003a). Source pulse enhancement by deconvolutions with empirical Green’s function.. 18. 91. C. J.. Lisbon. K. Six CDs with evt-files and KMI v1-. POPA. (1989). Geophys. Seism. RADULIAN. A.-I. L. FRANKEL. ONCESCU. Romanian Reports in Physics. Geophys. Seismic source properties: Indications of lithosphere irregular structure on depth beneath Vrancea region. M. Data Release 1996– 1999 of the Vrancea K2 Seismic Network. BONJER. MOLDOVEANU. 95–99. Source parameters of the 23 April 1992 Joshua Tree. GRECU. (1986). 4301–4311.. C. POPESCU. Res.. 84.-P. TRIFU. HOUGH. E. Am. 50. BONJER... 163–181. J. Lett. IONESCU.... S. 12633–12650. C-I. 1–28. M. Bucharest–Măgurele (Bonjer et al. A. The data recorded for each year are available on CD’s for the partners (Bonjer. Geophys. Res. Phys. 38. Lett. 2000. (1994).. O. L. 43–53. 70. (2003b).. (2000). RIZESCU. 97–122. XXVII General Assembly of the European Seismological Commission. GRECU. LINDLEY.-P. New York: Evidence for the poor resolution of corner frequency from small events.. L. POPA. RIZESCU.... RIZESCU. ENESCU. J. M. ARDELEANU.and site-parameters of the April 28.. 278–285. Two CD’s (evt-files). RADULIAN. (1998). J. 55. 4997–5009. (1990). J. 33–36. M. DRIAD. BONJER. Frequency – magnitude distribution of earthquakes in Vrancea: relevance for a discrete model. K. M. RADULIAN. vol.. B. BONJER... (1985). BERGER. F. VERNON... 458–509. IUUG. E. TRIFU. HANKS. Data Release 2000–2001 of the Vrancea K2 Seismic Network.S... 2010 Accepted for publication: March 14. POPA. Rom. Tectonic stress and the spectra of seismic shear waves from earthquakes. 499–513. RIZESCU.5 earthquake near Massena. FLECHTER.-P. Dynamics of a seismic regime: Vrancea – a case history. (1970). its largest foreshock and aftershocks. ENESCU.. J. Seismol. eds.. 1051–1057. RADULIAN.