Seismic Noise, a tool for evaluating site amplification

1. Seismic Noise, a tool for evaluating site amplification Giuliano Milana, Istituto Nazionale di Geofisica e Vulcanologia Via di Vigna Murata 605, 00143 Roma

2. 1) The chacteristics of seismic noise

3. With the term “Seismic Noise” we describe any kind of vibration recorded by high sensitivity seismometers in absence of seismic events. • Seismic noise has been investigated for longtime both as a disturbing cause in seismic records and as a signal carrying on useful information . • The noise can be related both to natural and antropic sources. • The frequency of noise can range from low frequency (< 0.1 Hz) up to 50-100 Hz.

5. Horizontal seismic noise average specturm recorded at Catania. The average is evaluated on 60 time windows of 60 seconds length. Continous black lines indicate the standard deviation interval.

6. Noise signals are formed by a quite stationary waveform with over imposed short transient phenomena. • Noise amplitude can change during the day according to the level of perturbations.

10. The scientific debate about seismic noise composition is still on. • Initially seismic noise was supposed to be composed mainly by body waves. • More recently researchers believe that surface waves are at the base of noise signals. • This assumption is based on the fact that surface waves show a lower attenuation with respect to body waves.

11. Seismic noise composition according to several authors

12. 2) The local site effect on ground motion

13. Trapped waves Incident wavefront The 1985 Mexico earthquake dramatically showed the effects of soft sedimentary layers on ground motion. The event caused severe damages in Mexico City, about 400 km. away from the epicenter due to the presence of very soft low velocity surface layers. This example clearly show how geological and geometrical surface conditions can modify both amplitude and duration of ground motion.

14. Recorded seismic signals spectra are due to the contribution of three terms. A source term, a propagation term and a site term. Evalute site effects means to compute the term S(f) after the separation of the other terms.

15. The use of spectral ratios between data collected on sites with different geological characteristics allow to achieve the proposed objective. In the spectral ratios source and propagation terms can be eliminated if the distance between the recording stations is much smaller than the distance between the event and the recording stations. In this conditions spectral ratios only depends by the sites term. If one of the sites can be considered free from any amplification effect, spectral ratio can be taken as an experimental measurement of the amplification (transer function). This approach requires a stable “Rock Site” and this hypothesis is not always so easy to be fulfilled.

16. Rock Site Deep Alluvium Site Deep Alluvium Site

17. Surface station • on gravel over • deep clay • Underground • station on deep • clay • Surface station • on deep clay

18. Amplification Factor Frequency (Hz) 5 4 S02 S03 S04 S07 S08 S05 S06 S09 S01 3 2 1 Figure 9 Amplification function obtained through a 2D modeling of L’Aquila basin

19. For simple geometries, amplification function can be obtained analytically. An easy and usefull model is called monodimensional (1D) model. In this hypothesis the share waves velocity only depends on depth. To better understand the origin of amplification let’s consider a 1D model based on two layers with shear waves velocity V1 and V2 in the hypothesis that V1 < V2

20. h At time t=0 and position x=0 (surface) the phase of the incident wave is zero. A trapped wave coming back to the surface will be in phase with the incident wave if its phase will be 2p. That is true at time t=2h/V1 and distance d=2h. Since k=f/V1, the phase will become: V1 V2>V1

21. The obtained formula gives the resonance frequency of surface layers in a 1D model. On an alluvial basin we can use the 1D approach if the ratio between basin aperture and depth is very high. The amplitude of trapped waves depends on the velocity contrast between surface and deep layers (Impedance Contrast). For more complex models (2D and 3D) the amplification function can be evaluated numerically. In any case a good knowledge of velocity distribution with depth is necessary.

23. 3) The use of seismic noise for evaluating local site effect on ground motion

24. A fast and easy way for investigating site amplification is based on seismic noise data. It is possible to obtain site transfer function by the spectral ratio between the horizontal and vertical components of noise recordings (H/V ratio or Nakamura technique). • H/V seems to be able to detect surface layers resonance frequency for high values of impedance contrasts. • Theory on H/V technique is quite poor, recent studies try to justify the method starting from Rayleigh waves ellipticity

25. In a homogeneous half plane Rayleigh waves are characterized by an elliptical retrograde motion. For ellipticity we intend the ratio between maximum displacement in the horizontal and vertical directions. • For stratified media the particle motion is modified according to the characteristics of the propagation media (Impedance Contrast). Ellipticity values change as the direction of motion. • For high impedance contrast the maximum in ellipticity as a function of frequency, the H/V peak and the resonance frequency in the 1D model are coincident.

26. Two layers model (layer 1 = top layer, layer 2 = bottom layer)

27. 4) The H/V technique

28. Stationary windows selection Time Spectrum evaluation and smoothing for the selected stationary windows Spectrum evaluation and smoothing for the selected stationary windows H/V evaluation on the selected stationary windows Evaluation of the average of H/V on the selected stationary windows

29. The public domain software to perform H/V analysis was just released in the framework of the European project SESAME (http://SESAME-FP5.obs.ujf-grenoble.fr) JSESAME

30. 5) H/V application examples

31. Seismic array in Cavola di Toano (Re)

32. H/V ratios moving along the line 3 of the array. P3 station is the reference Rock Site.

33. Contour plot of the amplification frequency along the line 3 of the array. It is possible to notice the shift to lower frequency moving towards highest depth in the soft layer

34. Noise and SASW measurments in Tortora Marina (Cs)

35. H/V ratio for the sites: CIG Upper Left Bar Upper Right SAW Lower Left

36. Velocity model derived by Surface Waves Analysis Shear Waves Velocity (km/s) Depth

37. Comparison between H/V ratio and 1D transfer function

38. Chiesa Scuola Station installed in San Giuliano di Puglia for Site Effect analysis

40. Molise main shock PGA values obtained modeling Site Effects using different amplification function * Il

41. 6) Conclusions

42. H/V technique is easy to apply and cost effective • H/V technique is very suitable in urban areas since it is not invasive • The results are quite satisfactory in simple geometrical (1D) conditions and high impedance contrast • H/V can be considered as a useful tool for discriminating spatial geological variation • In more complex structure H/V does not give useful results and can be misleading • It is always recommended to use data from other techniques to better constrain H/V results