Department of Seismotectonics, Institute of Rock Structure and Mechanics, Prague, Czech Republic 2

Transcription

1 less Is more: From van der rohe to the 4-channel system For the efficient and holistic acquisition of surface waves: an urban case study G. Dal Moro 1*, S.R. Moustafa 2,3, N. Al-Arifi 2 1 Department of Seismotectonics, Institute of Rock Structure and Mechanics, Prague, Czech Republic 2 Geology and Geophysics Department, Faculty of Sciences, King Saud University, Riyadh, Saudi Arabia 3 Seismology Dept., National Research Institute of Astronomy and Geophysics, Cairo, Egypt Introduction. Nowadays, the analysis of surface wave propagation for depicting the shearwave velocity (V S ) vertical profile is routinely performed for a large number of applications ranging from crustal studies to Non-Destructive Testing (e.g. Ryden et al., 2004; Prodehl et al., 134

2 2013). In the last decade, several methods have been actually proposed in order to acquire data to analyze according to different (active and passive) procedures. The critical point that must be considered in order to retrieve a reliable V S profile, is that because of the intrinsic non-uniqueness of the solution that inevitably characterizes any methodology, the use of a single component cannot provide a fully constrained subsurface model (e.g. Scales et al., 2001; Dal Moro, 2014). For this reason, over the years, a number of solutions capable of handling and jointly analyze various datasets and components have been proposed (e.g. Arai and Tokimatsu, 2005; Picozzi and Albarello, 2007; Dal Moro, 2014; Dal Moro et al., 2015a). As a matter of facts, the design of acquisition and analysis procedures which are efficient (limited equipment and straightforward field procedures) and capable of providing a sufficiently large amount of data necessary for a robust inversion process, represents a challenging task which can be seen also from a so-to-speak aesthetic point of view. The work and philosophical approach of the well-known German architect and designer Mies van der Rohe, who adopted minimalistic elements with multiple functional purposes, inspired the holistic acquisition and analysis system here briefly introduced. The pursued goal is to obtain the maximum possible outcome out of an extremely-light equipment and field efforts. The proposed system consists of a simple 4-channel seismograph, one 3-component (3C) geophone and four vertical-component geophones. The acquired (active and passive) data, are used to determine up to six independent (but mutually related and complementary) objective functions that, altogether, fully describe the surface-wave propagation. It is important to point out that, in general terms, an acquisition system is composed by the combination of several elements: the A/D (Analog-to-Digital) conversion unit (usually referred to as seismograph), a certain number of sensors (geophones or accelerometers), the seismic cables that transmit the signal to the A/D unit and the acquisition software that handles the data also possibly computing some quality check to ensure that no major problems occur during the acquisition procedures. The overall efficiency of the adopted acquisition system is determined by the combination of all these elements and actually depends on the quality of the weakest component so that, for instance, very high-quality geophones risk to be quite useless when a poorly designed and engineered A/D unit is used to convert and acquire the signals. Data acquired while considering the considered system, are eventually jointly used to set up an extremely-constrained inversion, thus giving no room to ambiguities in the obtained V S subsurface model. After briefly introducing the fundamentals of the designed system and approach, we also present the analyses performed on a dataset acquired in a NW-Italy urban area. The designed system. The design of the acquisition and analysis system described in the present paper represents the concrete realization of the paradigmatic van der Rohe s motto: less is more. Differently than the standard multi-channel approach (24 vertical geophones actually provide a single component), here, the mindful use of a simple 4-channel seismograph, one 3C geophone and four vertical geophones, allows the determination of four components from active acquisitions and two from passive seismics. In order to properly appreciate the value of the four components extracted from the active acquisitions, a clear understanding of the concept of component (with respect to the one of number of geophones) is crucial, and Fig. 1 will help clarifying the point. If we consider the data that can be acquired by a 3C geophone in the framework of an active survey, we can define up to four components (for further details see Dal Moro et al., 2015b): - the THF (Love waves) group velocity spectrum; - the group-velocity spectra of the Radial (RVF) and Vertical (ZVF) components of Rayleigh waves; 135

3 - the Radial-to-Vertical Spectral Ratio (RVSR) describing (clearly in relative terms) the amplitude ratio of the two components as a function of the frequency (thus enabling us to go behind the analysis of the velocities alone). Since these four components are capable of fully describe the surface wave propagation, such a methodology [which can be used as a purely-active stand-alone application - see Dal Moro et al. (2015a, 2015b)] is here indicated as Holistic analysis of Surface waves (for the sake of brevity hereafter HS). It must be clearly underlined that the radial and vertical components of Rayleigh waves are, in general terms, different, and such a difference is a crucial point that enables more-constrained analyses based on the multi-component Full Velocity Spectrum (FVS) approach (Dal Moro et al., 2015b, 2015c). On the other side, while considering the multi-channel data recorded for instance by 24 or 48 vertical-component geophones for MASW (Multichannel Analysis of Surface Waves - see e.g. Dal Moro, 2014) or ESAC (Extended Spatial Auto-Correlation - Ohori et al., 2002), we are actually dealing with less information with respect to the described HS approach, since such classical multi-channel techniques actually provide information on just one component (in this case the phase velocities of the vertical component of Rayleigh waves). Being an active methodology, the depth investigated through the HS approach will clearly depend on the adopted offset (Dal Moro et al., 2015a, 2015b). From the logistical point of view it must be anyway considered that moving a single geophone is definitely faster and easier than moving an array of geophones. Considering for instance an urban environment, it is clearly possible to set the source on one side of a road and the geophone on the opposite, without significantly interfering with the circulation of traffic. Fig. 1 Active seismics (HS multi-component approach): the components acquired in case a single 3C geophone is used to record the signals produced while considering both Vertical and Horizontal forces (Rayleigh and Love waves, respectively). Further details in Dal Moro (2014) and Dal Moro et al. (2015b). In addition to such a multi-component active dataset, for the present work we also considered two passive datasets/components acquired with the same 4-channel acquisition system: the HVSR curve (Arai and Tokimatsu, 2004) and the Rayleigh-wave (vertical component) effective dispersion curve (phase velocities) retrieved via Miniature Array Analysis of Microtremors (MAAM - see Fig. 2) (Cho et al., 2013). This latter technique relies on the passive data collected by an uneven number of verticalcomponent geophones deployed along a small-radius circular array (the triangle is the easiest and quickest solution), with an additional central geophone used for the noise compensation (Cho et al., 2013). The frequency range where the retrieved phase velocities are properly defined is a function of the adopted radius (Fig. 2). For the most common near-surface applications that require the investigation of some tens of meters, the typically radii range from 0.5 up to about 5 m. 136

4 By fully and efficiently exploiting the data that can be collected through a simple 4-channel (triggerable) acquisition system, we are then able to collect a considerable amount of active and passive data then used to perform a joint analysis based on the following objective functions: 1) Horizontal-to-Vertical Spectra Ratio (HVSR) [3-channel passive seismics]; 2) phase-velocity Rayleigh-wave dispersion curve from Miniature Array Analysis of Microtremors (MAAM) [4-channel passive seismics]; 3) Active HS data recorded by means of the 3-component geophone and used for the holistic analysis of the group-velocity spectra of Rayleigh and Love waves (defined according to the Multiple Filter Analysis - e.g. Bhattacharya, 1983) also jointly with the RVSR (i.e., altogether, four components). Since the phenomenology of Love waves is typically by far simpler than that of Rayleigh waves (Dal Moro, 2014; Dal Moro et al., 2015c) but their analysis via miniature array techniques does not provide highly accurate results (Tada et al., 2009), we decided to analyze their propagation from the active dataset (THF component depicted in Fig. 1). By properly combining the data acquired by means of these three procedures (for which we just need a 4-channel acquisition system), we can potentially jointly analyze up to 6 components. Since this would result in a so-to-speak extreme inversion procedure, here we decided to limit the presented analyses to just three components, chosen with the goal of anyway ensuring the most robust solution. The here-considered components are the HVSR curve, the effective dispersion curve of the vertical component of Rayleigh waves as defined via MAAM and the Love-wave group velocities (from the active acquisitions) here analyzed according to the FVS approach. Fig. 2 MAAM (passive seismics): a) acquisition setting (in this case the four vertical-component geophones are deployed along a 2-m radius circle); b) Rayleigh-wave effective dispersion curve (vertical component) determined while considering the data acquired for the case study illustrated in the present paper. The 3.5-λ line reports the upper frequency limit determined by spatial aliasing effects. An urban case study. The site considered for this pilot study is located in the city center of La Spezia (NW Italy) where, as common for any urban area, logistical problems (related to the lack of available spaces) and a considerable amount of noise (related to several human, commercial and industrial activities), make the acquisition procedures quite challenging. Local stratigraphy is superficially dominated by a mixture of soft alluvial and marine sediments that, at deeper layers, turn into gravel-like materials. The active (HS) and passive (MAAM) data to consider for the previously-described joint analyses were acquired according to the parameters reported in Tabs. 1 and 2. Passive data used 137

5 for determining the H/V spectral ratio were acquired (recording time 20 min; sampling rate 64 Hz) with the same 3C geophone used for the active HS acquisitions. The source considered for the active data was a standard 8-kg sledgehammer and for the HF acquisition we used a simple and classical wooden beam. As previously stated, for the sake of brevity here we will consider just three components: the HVSR, the Rayleigh-wave effective dispersion curve (i.e. the phase velocities defined via MAAM) and the Love wave group velocities (from the active dataset). As also pointed out by Cho et al. (2013), during the analyses (being a pilot project data were acquired using various seismographs) we verified that the Signal-to-Noise (S/N) ratio determined by the adopted acquisition system, reveals as crucial in the acquisition of the data then used for the Miniature Array Analysis of Microtremors. The noise inevitably introduced in the data by non-optimized A/D units, can in fact represent a serious problem that cause incorrect (lower) phase velocities in particular in the low-frequency range (see the theoretical description in Cho et al., 2013), being the noise compensation procedure capable of compensating the noise only when its level does not reach extreme values. Following the procedure suggested by Cho et al. (2013), we divided the data into small windows and, in order to remove the possibly-pernicious effects of large-amplitude transients, removed all the segments having an average amplitude larger than a threshold fixed on the basis of the mean root-mean-square (rms) values of the acquired traces. While the highest frequency which can be analyzed is limited by mere spatial aliasing effects and is then related to the adopted radius of the array [its value can be estimated in about 3.5 times the wavelength λ as defined in Cho et al. (2013): see the 3.5-λ line reported in Fig. 2b], the lowest frequency that can be soundly determined depends also on the quality of the adopted acquisition system (that determines the amount of electronic noise that can significantly pollute the data and analyses, especially in the low-frequency range). For the present pilot survey, we actually acquired 2 MAAM data sets (according to two different radii - see acquisition parameters reported in Tab. 1): while the 2-m radius allowed to retrieve the dispersion curve in the 4 16 Hz frequency range (see Fig. 2b), the 5-m radius array provided reliable phase velocities approximately in the 2 10 Hz range. Tab. 1 - MAAM acquisition parameters (see also Fig. 2a). sampling rate acquisition length radius sensors 4 ms (Nyquist frequency 125 Hz) 30 min 2 m; 5 m four vertical 2Hz geophones Tab. 2 - HS acquisition parameters. 138 sampling rate acquisition length offset sensor 1 ms (1000 Hz) 1 s 40 m one 3-component 2Hz geophone (used also for the acquisition of the microtremor data used to define the HVSR) stack 4 For validation purposes, MAAM analyses were compared with the results obtained from standard passive multi-channel bidimensional array (L-shaped 18-channel configuration), processed according to the ESAC methodology and the obtained dispersion curves resulted perfectly consistent.

6 HVSR curve was modeled according to Arai and Tokimatsu (2004), thus considering the amount of Love waves as one of the variables (for details see Dal Moro, 2014). The Rayleighwave effective dispersion curve (retrieved from MAAM and confirmed by the ESAC) was modelled according to Tokimatsu et al. (1992) and Love waves (from the active acquisition) via FVS analysis. The outcomes of the performed joint analysis are summarized in Fig. 3, where the overall good agreement for all the three considered components is quite apparent (the identified V S model is summarized in Tab. 3). Fig. 3 Joint analysis of the three considered components (the retrieved model is reported in Tab. 3): a) acquired active traces (vertical Z, radial R and transversal T components); b) field and synthetic HVSR curves; c) field (from MAAM) and synthetic Rayleigh-wave (vertical component) effective dispersion curves; d) field (background colors) and synthetic (overlain black contour lines) Love-wave group velocity spectra from the active acquisition (FVS analysis). Tab. 3 - Retrieved V S model down to a depth of 50 m. thickness (m) V S (m/s)

7 Some conclusions. In spite of the limited adopted equipment (one 4-channel seismograph, one 3C geophone and four vertical-component geophones) and the simple and straightforward field procedures, the proposed acquisition and analysis procedures can provide a significant amount of data to jointly analyze in order to fully constrain an inversion procedure eventually capable of providing a robust subsurface model free from ambiguities and non-uniqueness of the solution. The adopted procedures appear particularly relevant especially in case of surveys to accomplish in urban areas (for instance for seismic-hazard studies), characterized by heavy logistical problems and limited room to deploy and move the sensors. While the determination of the HVSR curve and the active acquisitions are not significantly influenced by the electronic noise produced by the acquisition system, the analyses performed according to the Miniature Array Analysis of Microtremors result instead strongly affected by this sort of noise and the use of high-quality A/D units results then mandatory. Furthermore, an improper geophone coupling (in urban areas planting a geophone can be quite tricky) can determine low-quality data that would then mirror in poor analyses. In fact, because of the mathematics intrinsically involved in the Miniature Array analyses, even relatively-small differences in the amplitudes of the acquired traces can produce major problems and prevent from the determination of fully reliable dispersion curves. Data quality should be then checked already on the field even by simply computing the rms values for all the acquired traces. If the rms amplitude of one trace significantly deviates from the median value, that is the evidence that some problem occurred during the acquisition of that trace. Common causes are represented by a bad coupling, problems with the geophone-cable connections or, finally (worst case), by the fact that the coil of that geophone is not consistent with that of the other geophones (in this case the geophone must be substituted). From the practical point of view is important to underline that while the HVSR and MAAM data necessarily refer to very local conditions, the active data refer instead to the average subsurface conditions between the source and the receiver. In case of significant superficial lateral variations, at the high frequencies (i.e. the shallowest layers) it is then possible a relatively poor match for the three objective functions here considered. For this reasons, HVSR and MAAM data should be considered only in the low-frequency range (incidentally often poorly defined in the active data). Acknowledgements. The present work was partially supported by the Visiting Professor Program of the King Saud University (Riyadh, Saudi Arabia) and the authors are grateful to the Deanship of the Scientific Research of the King Saud University (Riyadh, Saudi Arabia) for the financial support (PRG Research Grant). References Arai H. and Tokimatsu K.; 2004: S-wave velocity profiling by inversion of microtremor H/V spectrum. Bull. Seism. Soc. Am, 94, Arai H. and Tokimatsu K.; 2005: S-Wave velocity profiling by joint inversion of microtremor dispersion curve and horizontal-to-vertical (H/V) spectrum. Bull. Seism. Soc. Am., 95: Bhattacharya S.N.; 1983: Higher order accuracy in multiple filter technique. Bull. Seismol. Soc. Am., 73: Cho I., Senna S. and Fujiwara H.; 2013: Miniature array analysis of microtremors. Geophysics, 78, KS13 KS23. Dal Moro G., Moura R.M. and Moustafa S.R.; 2015c: Multi-component Joint Analysis of Surface Waves. J. Appl. Geophysics, 119, Dal Moro G., Moustafa S.R. and Al-Arifi N.; 2015b: Efficient acquisition and holistic analysis of Rayleigh waves, Proceedings of the Near-Surface EAGE 2015 (Turin - Italy) Dal Moro G., Keller L. and Poggi V.; 2015a: A Comprehensive Seismic Characterization via Multi-Component Analysis of Active and Passive Data. First Break, 33, Dal Moro G.; 2014: Surface Wave Analysis for Near Surface Applications. Elsevier, ISBN , 252pp Ohori M., Nobata, A. and Wakamatsu K.; 2002: A comparison of ESAC and FK methods of estimating phase velocity using arbitrarily shaped microtremor analysis. Bull. Seism. Soc. Am., 92, Picozzi M. and Albarello D.; 2007: Combining genetic and linearized algorithms for a two-step joint inversion of Rayleigh wave dispersion and H/V spectral ratio curves. Geophys. J. Int., 169,

8 Prodehl C., Kennett B., Artemieva I.M. and Thybo H.; 2013: 100 years of seismic research on the Moho. Tectonophysics, 609, Ryden N., Park C.B., Ulriksen P. and Miller R.D.; 2004: Multimodal Approach to Seismic Pavement Testing. Journal of Geotechnical and Geoenvironmental Engineering, 130, Scales J.A., Smith M.L. and Treitel S.; 2001: Introductory Geophysical Inverse Theory. Open file, Samizdat Press, 193 pp. Tada T., Cho I. and Shinozaki Y.; 2009: New Circular-Array Microtremor Techniques to Infer Love-Wave Phase Velocities. Bull. Seism. Soc. Am., 99, Tokimatsu K., Tamura S. and Kojima H.; 1992: Effects of Multiple Modes on Rayleigh Wave Dispersion Characteristics. Journal of Geotechnical Engineering, 118,