SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and...

13
SEISMIC WAVE ATTENUATION IN THE VRANCEA REGION. PART II. CRUSTAL Q-MODELS L. ARDELEANU National Institute for Earth Physics, P.O. Box MG-2, RO-077125 Bucharest – Magurele, Romania, E-mail: [email protected] Received May 24, 2017 Abstract. We investigate the shear wave attenuation in the crust beneath the area from the bending of the Eastern Carpathians by using a procedure based on high- frequency waveform modelling. We determine 1-D local models of the quality factor of the medium Q, adequate for simulating the ground motion (waveforms with maximum frequency 5 Hz) generated by seismic sources of Vrancea region at locations within the extra-Carpathian zone. The results of the study reveal distinct attenuative properties among the structural units lying in the study region; considerably higher attenuation is evidenced beneath the Focşani Basin and the Vrancea region orogen. Key words: Seismic wave attenuation, quality factor of the medium, Q-models, Vrancea region, waveform modelling. 1. INTRODUCTION The procedure proposed by [1] to estimate 1-D models of the quality factor of the medium Q in the Vrancea region and adjacent zone is applied to investigate the crustal attenuation in the extra-Carpathian area. The algorithm is a non-linear inversion in spectral domain: the amplitude spectra of high-frequency local records of low-to-moderate magnitude earthquakes are compared with the synthetic signal spectra, generated for point sources with the same location and mechanism as the recorded events; the theoretical seismograms are computed by multimodal summation in layered inelastic media [2, 3]. The non-linear inversion provides a set of models that satisfies the acceptability criterion – the r.m.s. value of the logarithmic residuals between theoretical and observed spectra smaller than an a priori defined threshold. The drawback is that the choice of the threshold is somewhat subjective; it was determined ad hoc on the basis of synthetic experiments simulating anticipated / expected errors of the available hypocenter locations and earthquake mechanisms. Romanian Journal of Physics 63, 807 (2018)

Transcript of SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and...

Page 1: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

SEISMIC WAVE ATTENUATION IN THE VRANCEA REGION.

PART II. CRUSTAL Q-MODELS

L. ARDELEANU

National Institute for Earth Physics, P.O. Box MG-2, RO-077125 Bucharest – Magurele, Romania,

E-mail: [email protected]

Received May 24, 2017

Abstract. We investigate the shear wave attenuation in the crust beneath the area

from the bending of the Eastern Carpathians by using a procedure based on high-

frequency waveform modelling. We determine 1-D local models of the quality factor

of the medium Q, adequate for simulating the ground motion (waveforms with

maximum frequency 5 Hz) generated by seismic sources of Vrancea region at

locations within the extra-Carpathian zone. The results of the study reveal distinct

attenuative properties among the structural units lying in the study region;

considerably higher attenuation is evidenced beneath the Focşani Basin and the

Vrancea region orogen.

Key words: Seismic wave attenuation, quality factor of the medium, Q-models,

Vrancea region, waveform modelling.

1. INTRODUCTION

The procedure proposed by [1] to estimate 1-D models of the quality factor

of the medium Q in the Vrancea region and adjacent zone is applied to investigate

the crustal attenuation in the extra-Carpathian area.

The algorithm is a non-linear inversion in spectral domain: the amplitude

spectra of high-frequency local records of low-to-moderate magnitude earthquakes

are compared with the synthetic signal spectra, generated for point sources with the

same location and mechanism as the recorded events; the theoretical seismograms

are computed by multimodal summation in layered inelastic media [2, 3].

The non-linear inversion provides a set of models that satisfies the

acceptability criterion – the r.m.s. value of the logarithmic residuals between

theoretical and observed spectra smaller than an a priori defined threshold. The

drawback is that the choice of the threshold is somewhat subjective; it was

determined ad hoc on the basis of synthetic experiments simulating anticipated /

expected errors of the available hypocenter locations and earthquake mechanisms.

Romanian Journal of Physics 63, 807 (2018)

Page 2: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 2

Specifically, the adopted threshold allows the recovery of the real Q-

structure for simulated depth mislocations of 5 km, or significant deviations of the

focal mechanisms. Since the hypocenter mislocation, and consequently the

incorrect fault plane solution are both effects of using of an inexact velocity model,

the uncertainty of the velocity structure – the main source of discrepancy between

computed and observed waveforms – is implicitly taken into account in the choice

of the threshold level.

To define the representative attenuation structure from the set of acceptable

solutions, the ‘Median Model’ is chosen (following [4]); the half-width of the

corridor of acceptable Q-values provides, at each depth, the model uncertainty.

2. DATA AND VELOCITY MODELS

The waveform data used in the present study are short period velocity

records of low-magnitude crustal earthquakes of Vrancea region, collected by

seismic stations situated at epicentral distances less than 200 km, and equipped

with vertical S13 seismometers (1s free period and damping of 0.7).

The location of the earthquakes (Table 1) and the focal mechanisms are

taken from [5].

Table 1

Earthquake locations

Event

no. Date Origin time

Latitude

[oN]

Longitude

[oE]

Depth

[km]

Local

magnitude

1 30 June 1982 21 : 38 : 47 45.42 26.43 13 3.5

2 21 February 1983 18 : 21 : 05 45.33 26.97 16 3.6

3 21 February 1983 18 : 33: 37 45. 37 27.02 17 3.0

The 1-D models of the elastic parameters of the medium – density and

seismic wave velocities – are compiled by integrating and harmonizing published

results, after a thorough selection and evaluation.

The models consist of several layers for the sedimentary cover and two

layers for the crystalline crust (the upper and the lower crust).

To model the sedimentary cover we used the series of local structures

constructed by [6] for the seismic station sites. For the crystalline crust we

compiled average velocity models along the seismic source – recording station

paths, using the information available from recent literature [7–13].

Page 3: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

3 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

The density models are based on the work of [6].

We note that the selection of the data for this first application of the

proposed procedure to investigate the crustal attenuation in the area of interest took

into account the availability of detailed velocity models for the structure beneath

the seismic station sites, as well as the availability of fault plane solutions of small

local earthquakes.

3. 1-D Q-MODEL ESTIMATION

In our study Q ≡ QS – the shear wave attenuation factor, and the relation

SP QQ 2.2 [2] between the quality factors of P- and S-waves is considered in the

generation of the synthetics.

To construct the input Q-models we use the layer configuration of the

velocity and density structures. The quality factor is allowed to vary in the range 0

to 1500 units; we adopted a step of 25 units for Q-values in the range 0 to 100, a

step of 50 units for Q in the range 100 to 400, a step of 100 for the range 400 to

600, a step of 200 between 600 and 1200, and a step of 300 for Q greater than

1200.

The response of the input structures to point sources having the same

location and mechanism as the recorded events are computed by modal summation

for the maximum frequency of 10 Hz.

Both records and synthetics are then filtered by low-pass with cut-off

frequency at 5Hz, and the synthetics are convolved with the instrument response.

To determine the best-fitting Q-models, the amplitude spectra of observed

and theoretical waveforms – calculated for time windows centered on the most

energetic part of the signals – are smoothed by averaging the amplitudes, for a

bandwidth of 1 Hz. The data misfit is evaluated only over the frequency range

where the signal-to-noise spectral amplitude ratio is greater than 2, in order to

avoid the effect of data noise.

The region from the bending of the Eastern Carpathians has a complex

tectonic structure. Consequently, for particular source-to-station paths the

mismodelling of the real medium by simplified 1-D structures may be severe, and

the synthetic waveforms constructed with average 1-D models may be far from the

true ones, in the considered frequency range.

If, for a specified path, the best-fitting Q-model leads to a r.m.s. value of the

residuals larger than the acceptance threshold, we assume that the overall error

Page 4: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 4

– mainly result of the poorly modelled velocity structure – is large enough to bias

the solution of the inversion significantly, and discard the result.

In Figs. 1–5 we illustrate the successful application of the inversion

procedure, for different source-station paths of the study region. Figure 6

summarizes the results, showing the attenuation models together with the velocity

structures used in their estimation; the Q-model determined in the first part of the

study [1] is also displayed (path #1).

The earthquakes that provided the observed data are located at depths of

13–17 km, hence the estimates of the attenuation factor are reliable for the

crystalline upper crust and the layers above it [1]. For this depth range

(sedimentary cover and upper crust) Fig. 7 gives another representation of the

results of Fig. 6 – the Q-factor versus the seismic wave velocity.

4. DISCUSSION

Generally, the seismic wave attenuation provides a direct measure of the

Earth’s inelastic properties.

Lateral variations in attenuation, as well as a qualitative relationship between

tectonic province and magnitude of attenuation have been pointed out by a number

of studies. Specifically, the stable continental zones were found to be considerably

less attenuating than the tectonically active regions (e.g. [14–19]).

Our area of interest comprises several juxtaposed tectonic units: the

Carpathian Orogen, the sedimentary Focşani Basin (component of the Carpathian

Foredeep), the stable Moesian and Scythian Platforms.

The ray paths displayed in Fig. 6 span all these crustal provinces: paths #1

and #3 – the Focşani Basin and Scythian Platform; path #2 – the Carpathian

Orogen, Focşani Basin and Scythian Platform; path #4 – the Focşani Basin and

Moesian Platform; path #5 – the Focşani Basin and Carpathian Orogen; path #6 –

the Carpathian Foredeep (Focşani Basin).

The large-scale velocity structure in the region is heterogeneous, and this

heterogeneity is reflected by the 1-D velocity models of Fig. 6.

The seismic attenuation and seismic velocity are in general controlled by the

same factors, such as temperature, composition, water content, partial melt

(e.g.[20–29]). However, the sensitivity of attenuation to these factors is different

from that of velocity (e.g.[20–25]), consequently the elastic and inelastic Earth

models are not identical (e.g. [30, 31]).

Page 5: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

5 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

Fig. 1 – (a) Location of epicenter (star) and seismic station (triangle), and earthquake focal

mechanism (event #1 of Table 1). (b) The 1-D velocity and density models. (c) The acceptable

Q-structures; thick grey line – best-fitting model, thick black line – last accepted model. (d)

Normalised amplitude spectra (smoothed by averaging the spectral amplitudes over a bandwidth of

1 Hz) calculated for the observed data and for the theoretical signals generated with the best-fitting

and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the

average spectral amplitudes are plotted against the highest frequencies of the 1 Hz smoothing

intervals, the vertical bars indicate the frequency range considered in data misfit evaluation. (e)

Selection of the frequency range for data misfit evaluation: analyzed time windows for signal and

noise (up), and the corresponding amplitude spectra (down).

Page 6: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 6

Fig. 2 – (a) Location of epicenter (star) and seismic station (triangle), and earthquake focal

mechanism (event #1 of Table 1). (b) The 1-D velocity and density models. (c) The acceptable

Q-structures. (d) Selection of time window for spectrum calculation: observed data (up) and

theoretical seismograms computed using the best-fitting and the last accepted Q-model, respectively;

synthetics are calculated for a scalar moment M0 = 4·1012 Nm. (e) Normalised amplitude spectra

(smoothed over a bandwidth of 1 Hz) calculated for observed data and theoretical signals generated

with the best-fitting and the last accepted Q-model, respectively (up), and with all acceptable

Q-models (down). (f) Selection of the frequency range for data misfit evaluation: analyzed time

windows for signal and noise (up), and the corresponding amplitude spectra (down).

Page 7: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

7 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

Fig. 3 – (a) Location of epicenter (star) and seismic station (triangle), and earthquake focal

mechanism (event #3 of Table 1). (b) The 1-D velocity and density models. (c) The acceptable

Q-structures. (d) Selection of time window for spectrum calculation: observed data (up) and

theoretical seismograms computed using the best-fitting and the last accepted Q-model, respectively;

synthetics are calculated for a scalar moment M0 = 2·1012 Nm. (e) Normalised amplitude spectra

(smoothed over a bandwidth of 1 Hz) calculated for observed data and theoretical signals generated

with the best-fitting and the last accepted Q-model, respectively (up), and with all acceptable

Q-models (down). (f) Selection of the frequency range for data misfit evaluation: analyzed time

windows for signal and noise (up), and the corresponding amplitude spectra (down).

Page 8: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 8

Fig. 4 – For captions see Fig. 3.

Page 9: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

9 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

Fig. 5 – For captions see Fig. 3.

Page 10: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 10

Fig. 6 – (a) Location of investigated ray paths. (b) – (g) Left – the velocity models;

right – the attenuation models: thin lines – the acceptable models; thick lines – the representative

Q-structures; grey area – the estimated uncertainty.

Figure 7 points out noticeable differences between attenuation variations and

velocity variations along the analyzed source-station paths.

The most attenuating ray paths – #5 and #6 – lie predominantly within a

single crustal province – the Focşani Basin. The less attenuating rays travel across

Page 11: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

11 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

several provinces, passing through stable platforms: Moesia – path #4, or Scythia –

paths #1, #2 and #3.

Fig. 7 – Shear wave attenuation versus velocity, for different paths of the study region.

The vertical bars show the uncertainty of the estimated Q-values.

Lateral variations of the crustal attenuation in the region from the bending of

the Eastern Carpathians were previously pointed out by [32], who noticed higher

attenuation in the foredeep area (Focşani Basin) and Vrancea orogen,

comparatively to the adjacent Scythian and eastern Moesian Platforms.

Our analysis evidences that the attenuation structure in the shallow crust

differs considerably among the tectonic units.

The results indicate high shear wave attenuation beneath the southwestern

portion of the sedimentary Focşani Basin, as well as beneath the nappes of the

Vrancea region orogen, with Q-values of 100–150 in the crystalline upper crust,

and less than 100 in the shallower layers.

In contrast, the rays crossing portions of stable platforms display optimal

1-D attenuation models with Q-values of 600 to 800 in the crystalline upper crust,

and between 100 and 400 in the layers above it.

Page 12: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

Article no. 807 L. Ardeleanu 12

5. CONCLUSIONS

We have used a procedure based on high-frequency waveform modelling to

investigate the shear wave attenuation in the crust beneath the Vrancea region and

adjacent extra-Carpathian zone.

The estimated 1-D (depth dependent) optimal attenuation structures are

adequate for the computation of synthetic seismograms (maximum frequency 5 Hz)

simulating the ground motion generated by seismic sources of Vrancea region, at

locations within the extra-Carpathian area.

Our results evidence that the structural units lying in the study region are

characterized by distinct attenuative properties.

Acknowledgments. The theoretical seismograms were calculated using the computational

packages developed at the Department of Earth Sciences of the University of Trieste, Italy.

The research was supported by the Ministry of National Education, Contract PN09 30, Project

PN09-01 03, and by the Ministry of Research and Innovation, Contract 21N, Project PN 16 35 01 02.

REFERENCES

1. L. Ardeleanu, Seismic wave attenuation in the Vrancea region. Part I. The approach for 1-D

Q-model estimation, Rom. Journ. Phys. 62, 815 (2017).

2. G.F. Panza, J. Geophys. 58, 125–145 (1985).

3. G.F. Panza and P. Suhadolc, Seismic strong motion synthetics (A. B. Bolt, ed.), Academic Press,

Orlando, Florida, 135–204 (1987).

4. N. M. Shapiro and M. H. Ritzwoller, Geophys. J. Int. 151, 88–105 (2002).

5. L. Ardeleanu, M. Radulian, J. Šílený, and G.F. Panza G.F., Pure Appl. Geoph. 162, 495–513

(2005).

6. V. Răileanu, C. Diaconescu, D. Mateciuc, and M. Diaconescu, Rom. Rep. Phys. 50, 123–141

(1998).

7. F. Hauser, V. Răileanu, W. Fielitz, A. Bălă, C. Prodehl, G. Polonic and A. Schultze,

Tectonophysics 340, 233–256 (2001).

8. F. Hauser, V. Răileanu, W. Fielitz, C. Dinu, M. Landes, A. Bălă and C. Prodehl, Tectonophysics

430, 1–25 (2007).

9. V. Răileanu, A. Bălă, F. Hauser, C. Prodehl and W. Fielitz, Tectonophysics 410, 251–272 (2005).

10. M. Grad, T. Tiira, and ESC Working Group, Geophys. J. Int. 176 (1), 279–292 (2009).

11. I. Molinari and A. Morelli, Geophys. J. Int. 185 (1), 352–364 (2011).

12. V. Răileanu, D. Tătaru and B. Grecu, Rom. Rep. Phys. 64 (2), 539–554 (2012).

13. V. Răileanu, D. Tătaru, B. Grecu and A. Bălă, Rom. Journ. Phys. 57 (9–10), 1438–1454 (2012).

14. S.A. Sipkin and J. Revenaugh, J. Geophys. Res. 99, 2687–2699 (1994).

15. W.W. Chan and Z.A. Der, Geophys. J. Int. 92, 303–314 (1988).

16. T. Lay and T. C. Wallace, Bull. Seismol. Soc. Am. 78, 2041–2061 (1988).

17. J.F. Lawrence, P.M. Shearer and G. Masters, Geophys. Res. Lett. 33, L07315,

doi:10.1029/2006GL025813 (2006).

18. A. Venkataraman, A.A. Nyblade and J. Ritsema, Geophys. Res. Lett. 31, L15611, doi:10.1029/

2004GL020351 (2004).

19. A. Thirunavukarasu, A. Kumar and S. Mitra, Geophys. J. Int. 208, 257–268 (2017).

20. D.L. Anderson, Geophys. J. R. Astron. Soc. 14, 135–164 (1967).

21. W.C. Hammond and E.D. Humphreys, J. Geophys. Res. 105, 10975–10986 (2000).

Page 13: SEISMIC WAVE ATTENUATION IN THE VRANCEA ...and the last accepted Q-model, respectively (up), and with all acceptable Q-models (down); the average spectral amplitudes are plotted against

13 Seismic wave attenuation in the Vrancea region. Part II. Crustal Q-models Article no. 807

22. W.C. Hammond and E.D. Humphreys, J. Geophys. Res. 105, 10987–10999 (2000).

23. I. Jackson, J.D. Fitz Gerald, U.H. Faul and B.H. Tan, J. Geophys. Res. 107, B12, 2360,

doi:10.1029/2001JB001225 (2002).

24. I. Jackson, U.H. Faul, J.D. Fitz Gerald and B.H. Tan, J. Geophys. Res. 109, B06201,

doi:10.1029/2003JB002406 (2004).

25. U.H. Faul and I. Jackson, Earth Planet. Sci. Lett. 234, 119–134 (2005).

26. K. Priestley and D. McKenzie, Earth Planet. Sci. Lett. 381 78–91 (2013).

27. Z.C. Eilon and G.A. Abers, Sci. Adv 3, e1602829 (2017).

28. Z. Wang, D. Zhao, X. Liu, X. Li, J. Geophys. Res. 122 (4), 2988–3007 (2017).

29. Z. Wang, D. Zhao, X. Liu, Ch. Chen, X. Li, Geochem. Geophys. Geosyst. 18 (4), 1688–1710

(2017).

30. C.A. Dalton, G. Ekström and A.M Dziewonski., J. Geophys. Res. 113, B09303,

doi:10.1029/2007JB005429 (2008).

31. H. Zhua, E. Bozdag, T.S. Duffy, J. Trompa, Earth Planet. Sci. Lett. 381 1–11 (2013).

32. L. Ardeleanu and V. Răileanu, Rom. Journ. Phys., 54 (9–10), 973–983 (2009).