Sunteți pe pagina 1din 7

Return to the list

Search

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

Application of ambient vibration array techique for site characterization: Angra do Herosmo and S. Sebastio
Teves-Costa P.(1), Veludo I.(1), Almeida J.(1)
(1) University of Lisbon, Centro de Geofisica da Universidade de Lisboa-IDL, FFCUL & DEGGE-FCUL, Campo Grande,Edificio C8,1749-016 Lisboa, ptcosta@fc.ul.pt, idalina.veludo@gmail.com, jdalmeida@fc.ul.pt Abstract

Within the joint research activity JRA4 - Geotechnical Site Characterization performed formed in the framework of Neries, a tool was developed using ambient vibration measurements. One of the main objectives was to estimate Vs30 profiles, usually used on EC8 for engineering purposes. Five sites were selected in Portugal to test this tool. Preliminary results will be present here from data acquired in two different sites in Ilha Terceira (Azores archipelago): one located at a High School, in Angra do Herosmo, and the other at S. Sebastio crater.
Key-words: array techniques, ambient vibrations, Neries, Geopsy, Dinver

1. Introduction This work is being developed within Neries European Project (http://www.neries-eu.org/). One of the main objectives of the Joint research activity (JRA4) Site Geotechnical Characterization - is to develop a reliable low cost tool for geotechnical site characterization which means, estimate Vs 30 profiles, usually used on EC8 for engineering purposes. The methodology was based on ambient vibration measurements using an array configuration. To process the data, the Sesarray package (http://www.geopsy.org/index.html), mainly developed by University Joseph Fourier (in Grenoble) and University of Potsdam, was used. This software includes Geopsy (for ambient vibration analysis) and Dinver (for data inversion) routines. To test the developed tool, five sites were selected, two in Portugal mainland and three in the Azores Archipelago (Terceira Island), to take into account different geological conditions. In Portugal mainland the sites were selected in the Lower Tagus Valley, which is mainly a large plateau, slightly dipping to the river, composed by Miocene and Pliocene formations. This plateau, which is larger in the left margin than in the right margin, is covered by ceno-antropozoic soils. From top to bottom they are composed by recent alluvium, quaternary deposits of fluviatic terraces, Pliocene sandstone and conglomerates, and a thick layer of sandy clays and sandstone (Zbyszewski, 1953-1979). The Azores Islands have a volcanic origin. The main geological formations in Terceira Island are stiff and composed by trachyts and trachy-andesites. However, due to past eruptions, pumice and pyroclastic deposits may exist, with thickness up to 20 m. Preliminary results, of Vs 30 profiles, will be presented here, from data acquired in Ilha Terceira Azores Archipelago, at an High School in Angra do Herosmo (located over a stiff formation), and in S.Sebastio crater (filled with recent deposits).

69

Return to the list

Search

Previous

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

2. Materials and Methods 2.1. Field Measurements The two experiments carried out consisted on measurements of ambient vibrations in circular array configurations. Eight stations were used, with one central station. The arrays were not perfectly circular because of constrains at the field work. Measurements started with the smallest array and increased the radius for the following arrays. The duration of the data acquisition in each array varied from half an hour, for the smallest array, until two hours for the largest array (Cornou et al., 2006). We performed 4 arrays at the High School and 5 arrays in S.Sebastio. In figure1, it is possible to see the position of the stations for the largest array performed in each site.
a b

Figure 1 - Configuration of the largest arrays: a) at the High school; b) in S. Sebastio. In both sites the WA_WAUO1 is the central station

The equipment used was developed in previous European projects (SESAME, HADU) at the University of Potsdam and made available for this experiment (Endrun & Renalier, 2008). It consists of 8 Lennartz 5 second three component seismometers, which are connected to Earth Data digitizers. At each station, a Mesh Cube Linux System manages recording and communication with the other stations. Each station is powered with a 12Volts battery. Data are transmitted in quasi-real time via Wi-Fi. The Wi-Fi connections need line-of-sight between any single station and at least one other station. However no central unit can contact all stations directly, so each station has to act as a transmitter and at that same time as a recorder. Data package can consecutively be transferred between stations until they reach the Wi-Fi receiver which is connected to the field PC. One of the advantages of data transmission by Wi-Fi is the quick and comparatively easy (without laying cables) deployment of quite large arrays, even in an urban environment, as it happened at the high school in Angra do Herosmo: we deployed the stations in two schoolyards that were separated by one building, which was impossible to do it without Wi-Fi system (figure 2).

Figure 2 - One of the schoolyards at Angra do Herosmo.

70

Return to the list

Search

Previous

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

Recordings were done with a sampling rate of 100Hz and a sensitivity of 2.5 V/bit. GPS receivers were included in each station to synchronise the clock of the recorders. Precise position determination, however, as demanded for array analysis methods, needs the use of different GPS measurements, carried out with a Leica GPS 1200 system as reference station and another Leica GPS 1200 as a Rover.This combinations leads to a relative accuracy in the centimetre-range, for stations locations. On the field PC, all arriving data are archived via seed link and can be processed in near-real time, allowing for first evaluation of the data (Wathelet, 2005; Konno & Ohmachi, 1998; Ohrnberger et al., 2004). 2.2. Methodology The ideal array would contain an infinite number of sensors perfectly sampling the whole space. Unfortunately, for economical and practical reasons we use discrete sampling. As the number of sensors decrease, resolution and aliasing restrict the ability of the array to correctly measure the velocity. At both sites 8 stations per array were used. The quality of the velocity profile estimation is a function of the sampling in space. In Angra do Herosmo, High school data was acquired using four arrays. For the smallest array the distance between stations was around 5m and for the largest array was around 100m. In S. Sebastio, the distances between stations was, for the smallest array, around 5m and, for the largest array, around 200m. The first step consisted on the verification of the nature of the ambient vibration, to check if it was natural or artificial, and on the estimation of H/V. The H/V peak obtained gives the resonance frequency of the soil layers beneath the sensors (figure 3). For estimating the array validity range, we did our approach using theoretical array response from which the resolution and aliasing ranges were calculated in the wave number space (figure 4). Three methods can be used to extract the dispersion curves from ambient vibration signals (Wathelet et al., 2004; Wathelet, 2005; 2006): Frequency Wave-number (FK) High Resolution Frequency (HRFK) Spatial autocorrelation method (SPAC)

All of them are dedicated to vertical component according to their current implementation in Geopsy (http://www.geopsy.org). High Resolution FK is sometimes less affected and allows extending a bit the valid frequency range towards low frequencies. Some times in both, FK and HRFK, it is possible to see that there is more than one mode. It was used a probability function, that estimates over the raw output of FK or HRFK the fundamental mode (Capon, 1969). In the absence of other means to confirm the velocity at low frequency and to avoid the inclusion of biased information into the inversion, the dispersion should be cut close to the lower limit of the valid wave number range of the array. For this reason we used HRFK (figure 5). The autocorrelation computation was performed with SPAC routine (Kudo et al., 2001). The results of are given versus frequency. A dispersion curve can be clearly delineated for each array. They are directly compared to the dispersion curve obtained with FK and HRFK methods. SPAC method allows to successfully checking the dispersion curve obtained with FK methods, even at low frequency. SPAC results carry almost the same information as the dispersion curve obtained with FK methods (figure 6). Both results, from HRFK and SPAC, can be directly included in the inversion, using the Dinver routine (Wathelet, 2008). It allows the inversion of FK and SPAC results either alone or together. A bi-modal dispersion curves determined by FK methods and crosschecked with SPAC methods is inverted to retrieve the shear wave velocity profile (Vs). A simple ground model is usually used in the first approach (Figures 7 and 8).

71

Return to the list

Search

Previous

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

3. Results and Discussion The results are presented in the following figures for the High school in Angra do Herosmo and for S. Sebastio crater: (i) H/V curve for the central station WA_WAUO1 (figure 3); (ii) Theoretical array response for the smallest array High school (figure 4); (iii) Dispersion curve HRFK (figure 5); (iv) Spatial autocorrelation curve plus HRFK dispersion curve (figure 6); (v) Vs profile (figures 7 and 8).
A B

Figure 3 Average H/V curve with standard deviation. The grey area indicates the uncertainty of the resonance frequency. A)High school at Angra do Heroismo; B) S.Sebatio crater.

Kmin/2 =0.2507 Kmax =1.5161

Figure 4 Results from the smallest array at the High school: A - Theoretical array response in wavenumber space. The small and large black circles are located at Kmin/2 and Kmax, respectively. B - Azimuthal cross section of figure (A). C - Resolution and aliasing limits.

72

Return to the list

Search

Previous

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

B A

Figure 5 - Dispersion curve for HRFK: A - High school; B - S. Sebastio crater.

Figure 6 - Spatial autocorrelation curve plus HRFK dispersion curve: A - High school; B - S. Sebastio crater.

Figure 7 - Dispersion Curve fit: A - High school; B - S. Sebastio crater.

73

Return to the list

Search

Previous

Next

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

Figure 8 Vs profiles: A - High school; B - S. Sebastio crater.

For the High School (figure 8A) the ground profile shows a lack of resolution for the first 5m and below 30m. Velocities start around 200m/s at the surface, and reach about 450m/s at 30m. The model shows a LVZ below 50m, where velocities are close to 200m/s, but we are not sure about the validity of these values because we are bellow of the resolution range. However, if these values are true, this could be explained by the heterogeneity of the volcanic formations (that could include holes or pyroclastic deposits, at higher depths, due to ancient volcanic eruptions). For S. Sebastio the results we obtained are not satisfactory. It is possible to observe (figure 7B) the bad fit between the dispersion curves obtained from the recording data and from the output models. However, former studies (for instance, Lopes, 2005) present velocities below 200 m/s in the upper 20m. More work is needed in order to understand this behavior. 4. Conclusions The presented results are still preliminary. Crosscheck with available information and tests to the consistency of the output, need to be performed. According to previous studies (Endrun & Renalier, 2008), a good agreement was found between the results of the ambient vibration array measurements and the results obtained by other seismic methods (for instance, active seismic measurements) in the frequency range covered. However, it is necessary to start the inversion with a simple model of the site but close to the real structure. From our results it seems that the structure beneath S. Sebastio is very complex and it is needed complementary information to build the starter model. In Angra do Herosmo, the structure is simpler and the results seem satisfactory. Acknowledgements The authors would like to thank to the Azores Civil Protection (SRPCBA Proteco Civil dos Aores) and Cmara Municipal de Angra do Herosmo. Thanks are also due to Brigitte Endrun and Daniel Vollmer, from University of Potsdam, for their valuable help. This work was financed by the NERIES European project, I3, FP6, Contract No RII3-CT-2006-026130.

74

Return to the list

Search

Previous

Print

Full screen

Quit

8th International Workshop on Seismic Microzoning and Risk Reduction 15-18 March 2009 Almera, Spain

References Cornou C., Ohrnberger M., Boore D.M., Kudo K. & Bard P.-Y. (2006). Using ambient noise array techniques for site characterisation: results from an international benchmark, in: Bard, P.Y., Chaljub, E., Cornou, C., Cotton, F. and Guguen, P. (Editors), Proc. 3rd Int. Symp. on the Effects of Surface Geology on Seismic Motion, Grenoble, 29 August - 01 September, 2006, LCPC Editions. Capon J. (1969). High-resolution frequency-wavenumber spectrum analysis, Proc. IEEE 57, 14081418. Endrun B. & Renalier F. (2008). Report on in-situ measurements at the 20 selected sites. JRA4 D2, http://www.neries-eu.org/ Konno, K. & Ohmachi, T. (1998). Ground-motion characteristics estimated from spectral ratio between horizontal and vertical components of microtremors, Bull. Seism. Soc. Am., 88, 288-241. Kudo, K., Tsuno, S., Sasatani, T., Horike, M., Higashi, S., Maeda, T., Bettig, B. & Bard, P.-Y. (2001). Estimation of the S-wave velocity structure in the Grenoble area using the SPAC method for array data of microtremors, Program and Abstracts Seismological Society of Japan, 166. Lopes I. (2005). Caracterizao geotcnica de solos no domnio das pequenas deformaes. Aplicao do mtodo das ondas superficias. PhD Thesis (in Portuguese), University of Lisbon. Ohrnberger et al. (2004). User manual for software package CAP a continuous array processing toolkit for ambient vibration array analysis, SESAME Deliverable D18.06, EVG1-CT-2000-00026, http://sesame-fp5.obs.ujf-grenoble.fr/Deliverables/Del-D18-Wp06.pdf. Wathelet, M (2005). Array recordings of ambient vibrations: surface-wave inversion, Ph.D. Thesis, University of Lige, Belgium. Wathelet M. (2006) Noise Blind test: Retrieving dispersion curves and inversion with the conditional neighbourhood Algorithm, ESG 2006, Grenoble Sptember2006. Wathelet M. (2008). An improved neighbourhood algorithm: parameter conditions and dynamic scaling. Geophysical Research Letters, 35, L09301, doi:10.1029/2008GL033256. Wathelet, M., Jongmans, D. & Ohrnberger, M. (2004). Surface-wave inversion using a direct search algorithm and its application to ambient vibration measurements, Near Surf. Geophys., 2, 211-221. Zbyszewski, G., (1953-1979). Explanation notes of the geological maps at the scale 1:50 000 (in Portuguese). Servios Geolgicos de Portugal, Lisbon.

75