Journal of Applied Geophysics 119 (2015) 128–138

Contents lists available at ScienceDirect

Journal of Applied Geophysics

journal homepage: [Link]/locate/jappgeo

Multi-component joint analysis of surface waves

Giancarlo Dal Moro a,⁎, Rui Miguel Marques Moura b, Sayed S.R. Moustafa c,d
a
Institute of Rock Structure and Mechanics, Academy of Sciences of the Czech Republic, Prague, Czech Republic
b
Faculdade de Ciências da Universidade do Porto (DGAOT), Porto, Portugal
c
Geology and Geophysics Department, Faculty of Sciences, King Saud University, Riyadh, Saudi Arabia
d
Seismology Dept., National Research Institute of Astronomy and Geophysics, Cairo, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: Propagation of surface waves can occur with complex energy distribution amongst the various modes. It is shown
Received 17 October 2014 that even simple VS (shear-wave velocity) profiles can generate velocity spectra that, because of a complex mode
Received in revised form 15 May 2015 excitation, can be quite difficult to interpret in terms of modal dispersion curves.
Accepted 16 May 2015
In some cases, Rayleigh waves show relevant differences depending on the considered component (radial or ver-
Available online 19 May 2015
tical) and the kind of source (vertical impact or explosive). Contrary to several simplistic assumptions often pro-
Keywords:
posed, it is shown, both via synthetic and field datasets, that the fundamental mode of Rayleigh waves can be
Surface waves almost completely absent. This sort of evidence demonstrates the importance of a multi-component analysis ca-
Surface wave dispersion pable of providing the necessary elements to properly interpret the data and adequately constrain the subsurface
Seismic data acquisition model. It is purposely shown, also through the sole use of horizontal geophones, how it can be possible to effi-
Seismic data inversion ciently and quickly acquire both Love and Rayleigh (radial-component) waves.
Velocity spectrum The presented field dataset reports a case where Rayleigh waves (both their vertical and radial components) ap-
Joint inversion pear largely dominated by higher modes with little or no evidence of the fundamental mode.
Full Velocity Spectrum (FVS) analysis
The joint inversion of the radial and vertical components of Rayleigh waves jointly with Love waves is performed
Rayleigh waves
by adopting a multi-objective inversion scheme based on the computation of synthetic seismograms for the three
Love waves
considered components and the minimization of the whole velocity spectra misfits (Full Velocity Spectra – FVS –
inversion).
Such a FVS multi-component joint inversion can better handle complex velocity spectra thus providing a more
robust subsurface model not affected by erroneous velocity spectra interpretations and non-uniqueness of the
solution.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction Since different authors might adopt slightly different terminologies,

a preliminary remark is needed in order to clarify and underline the dif-
Exploitation of surface wave (SW) propagation for retrieving VS pro- ference between the concept of velocity spectrum and the one of disper-
files is a routine technique widely used for a number of applications sion curve. In the present paper, by velocity spectrum we mean the
(Beaty et al., 2002; Lai and Rix, 1998, 2002; Luo et al., 2007; O'Neill correlation matrix obtained by transforming the seismic traces (origi-
et al., 2003; Simons et al., 2002). Nevertheless, some aspects related to nally in the time-offset domain) into the velocity–frequency domain,
SW behaviour are often poorly known and, if not properly considered, where the dispersive properties are clearly more evident. Such a do-
can easily lead to erroneous subsurface reconstruction. The current main transformation can be accomplished according to different ap-
paper attempts to address the problem of complex velocity spectra proaches (Dal Moro et al., 2003; Park et al., 1998) but anyway it does
resulting from non-trivial mode excitement. not imply any interpretation of the data, which are simply brought
Because of some so-to-say historical reasons (vertical-component into a different and more useful domain. In this sense, the background
geophones are commonly used for P-wave refraction and reflection greyscale colours in Fig. 1c represent the velocity spectrum of the (syn-
studies), Rayleigh waves are often considered as the surface waves to thetic) seismic traces reported in Fig. 1b.
consider for the Multi-channel Analysis of Surface Waves (MASW) On the opposite side, by dispersion curve(s) we refer to a successive
and a number of simplistic assumptions regarding their propagation operation where the signals present in the velocity spectrum are
are often adopted, possibly leading to erroneous data interpretation. interpreted (typically in terms of modes) and a set of frequency–velocity
points describing a curve is obtained. In this case, the picked dispersion
⁎ Corresponding author. curve represents an interpretation (thus a subjective entity) of an objective
E-mail address: dalmoro@[Link] (G. Dal Moro). entity (the above-defined velocity spectrum).

[Link]
0926-9851/© 2015 Elsevier B.V. All rights reserved.
 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138 129

Fig. 1. Example of complex Rayleigh-wave velocity spectrum: a) VS model (numbers report the adopted Poisson ratios); b) synthetic traces (vertical component); c) phase velocity spec-
trum with, overlaying, the modal dispersion curves of the first four modes (M#0 is the fundamental mode while M#1, M#2 and M#3 are the first, second and third higher modes, respec-
tively). See text for comments.

Actually, it is also possible to refer to a second kind of dispersion curve in vaguer phase velocity spectra (this happens because the range of veloc-
as that set of frequency–velocity values that solve the surface-wave ei- ities that produce some correlation is wider).
genvalue problem (e.g., Dunkin, 1965; Herrmann, 2003) for a given The importance of joint analysis of different datasets or components
model. The curves reported in Fig. 1c over the background velocity spec- of the wavefield (Rayleigh and Love waves, refraction and/or reflection
trum represent the dispersion curves of the first four modes of Rayleigh events and Horizontal-to-Vertical Spectral Ratio) has been previously de-
waves for the model reported in Fig. 1a. scribed as a solution both to non-uniqueness and interpretative issues
The phase velocity spectrum presented in Fig. 1c actually represents (e.g., Dal Moro, 2008, 2010, 2011, 2015; Picozzi and Albarello, 2007).
a clear example for which the velocity spectrum cannot be easily More specifically, the importance of analyses based also on Love
interpreted in terms of modal dispersion curves. In fact, by comparing waves, has been treated by Safani et al. (2005), Winsborrow et al.
the velocity spectrum and the overlaying modal dispersion curves, it is (2003), Dal Moro and Ferigo (2011) and Dal Moro (2014).
apparent that the continuous signal that dominates the velocity spec- In the present paper a series of synthetic and field datasets are pre-
trum for frequencies higher than 20 Hz does not pertain to a single sented in order to highlight some issues related to complex mode exci-
mode but to two different modes: from 20 to 40 Hz the signal is in tation in SW propagation. The possible complexity of the velocity
fact related to the first higher mode, while for higher frequencies the spectra suggests the acquisition of multi-component data whose joint
signal is related to the fundamental one. That means that the continuity analysis can be proficiently adopted to solve ambiguities and interpreta-
of a signal in the velocity spectrum does not necessarily mean that that tive issues, thus eventually leading to robust and accurate VS profiles.
signal belongs to a single mode (for the mathematical description of the
so-called effective or apparent dispersion curves see Lai and Rix, 1998; 2. Components and modal energy: a synthetic dataset
Tokimatsu et al., 1992). In short, sometimes the way surface-wave ener-
gy unfolds cannot be interpreted in terms of modal dispersion. In order to put in evidence some fundamental aspects related to
This kind of problems is often (but not solely) caused by abrupt VS non-trivial mode excitement in surface wave propagation, a series of
variations in the shallowest layers (e.g., Dal Moro, 2011; O'Neill and synthetic seismograms was computed according to the modal summa-
Matsuoka, 2005). In fact, when the site is characterized by a complex tion approach (Herrmann, 2003). Although approximate (compare for
stratification with large VS variations, mode excitation can be quite instance with Lai and Rix, 2002), such approach is definitely sufficient
complex and the velocity spectra can consequently appear quite hard to place in evidence a few important aspects related to surface wave
to interpret in terms of modal dispersion curves. Some authors have propagation (see also e.g., Malagnini et al., 1994, 1997) that will be
sometimes pointed out the role of the geophone-array length with eventually important to fully comprehend the case study presented
respect to the appearing and disappearing of higher modes (e.g., Foti later on.
et al., 2000), but practical and logistical problems (limited space and/or Since surface waves can be generated and recorded by using differ-
natural or man-made obstacles) often impose short arrays which further- ent source–receiver combinations (vertical and/or horizontal geo-
more necessarily produce blurred velocity spectra. In fact, in the case of phones and vertical/horizontal impact force as well as explosive
short arrays, the limited time shift between the traces necessarily mirrors sources), in Table 1 we summarize the meaning of the coding system

Table 1
The five components considered for surface wave analysis: first letter indicates the type and orientation of the geophones (Z, R or T), while the second and third letters relate to the ori-
entation and nature of the adopted seismic source (VF, EX or HF).

Component Geophone Source Use

ZVF Vertical (Z) (Fig. 2a) Vertical force (e.g., sledgehammer, Vibroseis or weight drop) Vertical component of Rayleigh waves
ZEX Vertical (Z) (Fig. 2a) Explosive Vertical component of Rayleigh waves
RVF Radial (R) — axis parallel to the array (Fig. 2b) Vertical force (e.g., sledgehammer, Vibroseis or weight drop) Radial component of Rayleigh waves
REX Radial (R) — axis parallel to the array (Fig. 2b) Explosive Radial component of Rayleigh waves
THF Transversal (T) — axis perpendicular to the array (Fig. 2c) Horizontal force (shear source) Love waves
130 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138

by 90° the axis of the horizontal geophones according to the scheme re-
ported in Fig. 2.
A series of synthetic seismograms were computed considering 48
traces (geophone spacing 1 m, minimum offset 5 m) and a sampling in-
terval of 1 ms. The considered model is shown in Fig. 3 and can be rep-
resentative of a typical sequence that can be found in a mountain area
where soil and debris lie over a hard rock (set at a depth of about
11 m). Although no Low-Velocity Layer (LVL) or stiff layer is present,
it will be shown how even such a seemingly-simple stratigraphic se-
quence can generate non-trivial velocity spectra.
The analysis of the seismic quality factors (Q) has been considered
by a number of authors (e.g., Barton, 2007; Tonn, 1991; White, 1992;
Xia et al., 2002). In general terms, Q factors are proportional to the cor-
responding velocities and the quality factors here considered represent
standard average values as commonly reported by several authors
(e.g., Kudo and Shima, 1970; Prasad and Meissner, 1992; Xia et al.,
2002; Xu and Stewart, 2006: Barton, 2007; Barton — personal
communication).
Regarding the explosive source (EX), in order to simulate common
near-surface acquisitions (typically done with a rifle gun), the explosive
source for the REX and ZEX components was set at a depth of 0.3 m from
the surface.
The velocity spectra reported in Figs. 4 and 5 (and computed from
the respective synthetic traces) put in evidence that the four compo-
nents related to Rayleigh waves show different energy distribution
amongst the modes. This should not come as a surprise since the func-
tions that express the effective dispersion for the vertical and radial
components are different. For the active and passive cases see Lai and
Rix (1998) and Tokimatsu et al. (1992), respectively. Overlaid on the ve-
locity spectra (Figs. 4 and 5) are the modal dispersion curves computed
according to Dunkin (1965).
While Love waves (THF component) appear largely dominated by
the fundamental mode (even if the first higher mode is also clearly vis-
ible), Rayleigh waves show complex velocity spectra and very different
Fig. 2. Map view of the source and geophone orientation to adopt in order to obtain the mode excitement for the different components (mode jumps occur at
five considered components (see Table 1): a) vertical component of Rayleigh waves different frequencies).
(ZVF or ZEX); b) radial component of Rayleigh waves (RVF or REX); c) Love waves
In particular, the ZVF component [i.e., the most common way to ac-
(THF). By using horizontal geophones only, it is possible to acquire both Love waves
(axis of the geophones set perpendicular to the array) and Rayleigh waves (the radial quire Rayleigh waves — vertical geophones and vertical-impact force] is
component) (axis of the geophones parallel to the array). fundamentally dominated by the first higher mode without any clear
evidence of the fundamental one (Fig. 4a). On the other side, the radial
component (RVF) shows evidence of several modes (fundamental
adopted by Herrmann (2003) and used in the present work (see Fig. 2 mode included).
and Dal Moro, 2014). It can be noted that by using horizontal geophones Compared to the VF case (Fig. 4a and b), the use of an explosive
only, it is possible to quickly acquire both Love and Rayleigh (radial source (Fig. 5) creates a different distribution of the energy amongst
component) waves. In order to do that it is simply necessary to rotate the different modes.

Fig. 3. Considered velocity (VP and VS) and quality factor model.
 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138 131

Fig. 4. Synthetic datasets for the model reported in Fig. 3: ZVF (a) and RVF (b) are the vertical and radial components of Rayleigh wave while considering a vertical-impact force, while THF
(c) represents the Love waves. The corresponding velocity spectra are overlaid by modal dispersion curves with indication of the modes.

Several studies describe the role of attenuation in surface wave the higher modes with respect to the fundamental one as a consequence
propagation (Lee and Solomon, 1978; Liu and de Boer, 1997; Pecorari, of the attenuation is very well-known in seismological studies (Panza,
2000; Yang et al., 2007). In particular, the increase in the amplitude of 1989; 2012 — personal communication) but is often neglected in

Fig. 5. Synthetic datasets for the model reported in Fig. 3: ZEX (a) and REX (b) are the vertical and radial components of Rayleigh wave while considering an explosive source. The cor-
responding velocity spectra are overlaid by modal dispersion curves with indication of the modes.
132 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138

Fig. 6. ZVF (a), RVF (b) and THF (c) velocity spectra for the same velocity model reported in Fig. 3 but now considering the elastic case (infinite quality factors). The corresponding velocity
spectra are overlaid by modal dispersion curves with indication of the modes.

near-surface applications. In order to quickly put in evidence this aspect, only, while the Love waves give evidence of the fundamental mode
we adopted the same velocity model presented in Fig. 3 but now consid- only), thus showing how the attenuation tends to emphasize higher
ering the elastic case (simply substituting the Q values reported in Fig. 3 modes (see also Panza, 1989).
with infinite values). The obtained velocity spectra for the ZVF, RVF and The considered synthetics show that different components can put
THF components are shown in Fig. 6. Compared to the data reported in in evidence different portions of the modal dispersion curves. This evi-
Fig. 4 and 5, higher modes are now less prominent (even if the ZVF com- dence, along with the velocity spectrum presented in Fig. 1, clearly sug-
ponent is still definitely dominated by the first higher mode, the RVF ve- gests that a multi-component approach can be capable of solving
locity spectrum is dominated by the fundamental and first higher mode possible ambiguities in the velocity spectra interpretation. It must be

Fig. 7. Location of the site where multi-component (ZVF, RVF and THF) data were acquired (a highly industrialized area close to Prato, Tuscany — Italy).
 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138 133

underlined that it is not possible to generalize how different the energy The dataset was acquired on a valley in Tuscany (Italy — Fig. 7) while
distribution amongst the modes can be for the various components and, using both horizontal and vertical geophones (ZVF, RVF and THF com-
as shown by the intentionally-simple considered model (Fig. 3), the ponents; 24 channels, geophone spacing 3 m and minimum offset
simplicity of the VS model (i.e., of the stratigraphic conditions) does 4 m) and an 8-kg sledgehammer applied to a base plate (VF) and to a
not necessarily mean that the related velocity spectra will be equally horizontal wooden beam (HF).
simple and easily interpretable in terms of modal dispersion curves The acquired data (Love waves and radial and vertical components
(see for instance the ZVF component presented in Fig. 4a as well as of Rayleigh waves) are reported in Fig. 8 and, not surprisingly (see pre-
the data presented in Fig. 1). vious paragraph), show that the velocity spectra of the ZVF and RVF
Similarly, a number of further synthetics (not shown for the sake of components are different.
brevity) indicate that it is not possible to identify the best Rayleigh- The data were jointly inverted considering the simple isotropic case
wave component that invariably shows easily-interpretable velocity (VSV = VSH) and according to the Full Velocity Spectrum (FVS) approach
spectra. The only clear evidence is that, although less sensitive to subtle (Dal Moro, 2014; Dal Moro et al., 2014), thus not considering
stratigraphic features, Love waves necessarily show an easier behaviour interpreted modal dispersion curves as in the standard approach. Fun-
(see the presented synthetics, the following field dataset and Dal Moro, damentally, the FVS analysis is based on the computation of the
2014). synthetic traces for the considered components and on their successive
transformation into the frequency–velocity domain to obtain the veloc-
3. Field dataset ity spectra to compare with the ones of the field data. The related cost
function is then minimized in order to obtain the subsurface model
In line with the facts illustrated in the previous sections, the follow- whose velocity spectrum matches as close as possible the one of the
ing field dataset is presented to further show the relevance of the joint field dataset. It is clear that such a process takes place without any inter-
analysis of various components in surface-wave analysis. pretation of the spectra in terms of dispersion curve(s) and while

Fig. 8. Acquired traces and phase velocity spectra for the three considered components (ZVF, RVF and THF).
134 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138

considering the velocity spectrum as a whole (i.e., the entire frequency– and the evaluation of the inversion process must be performed while
velocity matrix — please notice that in the following analyses no disper- considering the overall trend and, most of all, the symmetry of the
sion curves are considered). For details see Dal Moro et al. (2014) and Pareto front models with respect to the cloud of models (see Dal Moro
Dal Moro (2014). and Ferigo, 2011). The Pareto models representing the final outcome
A three-objective optimization (i.e., inversion) scheme was imple- of the inversion process also express the uncertainty of the final VS
mented according to the Pareto-dominance criterion (Dal Moro, 2010, model which can be obtained as their mean value.
2014; Van Veldhuizen and Lamont, 1998; Zitzler and Thiele, 1999) According to the graphical representation proposed in Dal Moro
with the results shown in Figs. 9, 10 and 11. et al. (2014), Fig. 10 reports the velocity spectra of the field dataset in
Fig. 9 reports the three-objective space with the cloud of models the background and, superimposed, the contour lines of the synthetic
pointing towards the [0, 0, 0] utopia point (thus giving evidence of the velocity spectra of the identified model (shown in Fig. 9b). The overall
overall consistency of the inversion procedure) along with the final VS good match is apparent.
profile and the Rp/RL values (point resistance/side friction — Begemann, For clarity, in Fig. 11 we also report the phase velocity spectra of the
1965) of the Cone Penetration Test (CPT) data, which appear in apparent field and synthetic data, separately together with some indication about
agreement with the retrieved shear-wave velocity profile. It is also no- the energy in terms of modes.
ticeable that the CPT stopped at about 7.5 m because of a stiff (coarse It is apparent that, in spite of the simplicity of the identified
sand/gravel) layer that mirrors the value of almost 340 m/s in the VS model (Fig. 9b), Rayleigh waves (both the vertical and radial compo-
profile reported in Fig. 9b. nents) are widely dominated by the first higher mode (with little or
For a number of reasons (different components can be differently no evidence of the fundamental one) while Love waves, as very fre-
sensitive to lateral variations and some amount of anisotropy is typically quently observed, appear entirely dominated by the fundamental
always present), the obtained cloud of model is rarely perfectly pointy mode.

Fig. 9. Main results of the joint inversion compared to the available CPT (Cone Penetration Test) data: a) model distribution in the three-objective space; b) identified VS model; c) CPT data
expressed in terms of ratio between the point resistance (Rp) and side friction (RL) and showing the increase in the measured soil parameters down to a depth of about 7.5 m (here the CPT
stopped because of a stiff layer — see text for further comments). The overall consistency between the CPT data and the retrieved VS profile is apparent.
G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138 135
136 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138

Fig. 11. ZVF, RVF and THF components: same data as for Fig. 10 while keeping separate the field (left column) and synthetic (right column) phase velocity spectra.

It is also noteworthy the evidence that the energy can pass from one similarity of the four Rayleigh-wave components appears clear, being
mode to the other in an abrupt way (see the jump from the fundamental the vertical and radial components quite similar, independently from
to the first higher mode in the ZVF component at about 8 Hz), but can the adopted source (for a wider collection of case studies see also Dal
also move from one mode to another in a smoother (so-to-speak inter- Moro, 2014).
polated) way, as shown by the RVF component (Figs. 8, 10 and 11). One of the fundamental points of the proposed multi-component
approach is actually represented by the possibility of analyzing the radi-
4. Discussion and conclusions al and vertical components of Rayleigh waves (thus, so-to-speak, their
relationships). Their analysis (which should be possibly performed
The presented synthetic and field datasets show that, especially for jointly with Love waves) is a key point that strongly reduces the ambi-
Rayleigh waves, the distribution of energy amongst different modes guity of the solution (i.e., the subsurface VS model). Actually, consider-
can be quite complex and different for the various components ing the simpler behaviour that Love waves invariably show (compared
(i.e., source–receiver combinations) and that in some cases surface to Rayleigh's), it might be argued that Love waves should be preferred
waves (Rayleigh's in particular) can propagate without clear evidence and considered on their own. Actually, since Love waves appear less
of the fundamental mode. In general terms, it is not possible to identify sensitive to small stratigraphic details and, differently than Rayleigh
the best component to consider. Depending on the local stratigraphy, waves, are completely independent on VP, in order to retrieve a detailed
the radial and vertical components of Rayleigh waves can be very simi- and robust subsurface model it is recommended their joint analysis
lar or completely different: sometimes the easiest velocity spectrum to with Rayleigh waves.
understand in terms of modal energy is the radial one, sometimes the This point can be clarified through the synthetic data reported in
vertical one and sometimes (especially when compared with Love Fig. 13, where two VS models are presented together with their respec-
waves) both of them can result quite tricky. tive ZVF and THF phase velocity spectra.
Fig. 12 reports the four components (related to Rayleigh waves) for a While the Love-wave phase velocity spectra appear quite similar,
field dataset acquired while considering both vertical (Z) and radial some definitely larger differences are apparent for the velocity spectra
(R) geophones and both explosive (EX) and vertical-impact force of the Rayleigh waves (ZVF component). The two mode jumps that
(VF). Differently than in the previously-illustrated cases, here the characterize these latter spectra (Figs. 13b and 13e) occur at different

Fig. 10. ZVF, RVF and THF components. In the background are reported the velocity spectra of the field dataset while the overlaying black contour lines represent the synthetic velocity
spectra of the identified model (reported in Fig. 9b).
 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138 137

Fig. 12. Rayleigh-wave dispersion on a further test site: in this case, the four components (radial and vertical components acquired while considering both explosive and vertical-impact
source) show almost identical phase-velocity spectra.

frequencies: for the model#1 the mode jumps occur at about 15 and In these cases the joint acquisition and analysis of the RVF and ZVF com-
20 Hz, while for the model#2 at about 12 and 25 Hz. Furthermore, the ponents might represent the unique viable solution, also clearly consid-
phase velocities at the lower frequencies (A and B boxes) are also signif- ering the opportunities of using passive techniques such as the Extended
icantly different (definitely larger for the model#2). Spatial AutoCorrelation (ESAC — Ohori et al., 2002), which result
Furthermore, from a practical point of view, it should also be consid- particularly effective especially while working in complex urban
ered that the acquisition of Love waves in noisy (e.g., urban) environ- environments.
ments and/or with asphalt covers (that can cause some problems/ A practical consequence is also implicitly clear from the illustrated
challenges to the source coupling) might sometimes result problematic. field and synthetic data: even by using only the horizontal geophones,

Fig. 13. Rayleigh and Love waves for two different models. Model#1 (upper panel): a) VS model (logarithmic scale adopted in order to emphasize the shallow layers); b) phase velocity
spectrum of the vertical component of Rayleigh waves (ZVF); c) phase velocity spectrum of Love waves (THF). Model#2 (lower panel): VS model (logarithmic scale); b) phase velocity
spectrum of the vertical component of Rayleigh waves; c) phase velocity spectrum of Love waves.
138 G. Dal Moro et al. / Journal of Applied Geophysics 119 (2015) 128–138

it is in fact possible to record both Love and Rayleigh (radial component) Engineering Geology for Society and Territory, Volume 5. Springer International Pub-
lishing, pp. 1177–1182.
waves, thus easily allowing a joint analysis capable of solving possible Dunkin, J.W., 1965. Computation of modal solutions in layered, elastic media at high fre-
interpretative issues (i.e., ambiguities) related to complex velocity spec- quencies. Bull. Seismol. Soc. Am. 55, 335–358.
tra. From a practical point of view, this kind of “double” acquisition is Foti, S., Lancellotta, R., Sambuelli, L., Socco, L.V., 2000. Notes on fk analysis of surface
waves. Ann. Geofis. 43, 1199–1210.
easily and quickly accomplished by rotating the horizontal geophones Herrmann, R.B., 2003. Computer programs in seismology. Open files, [Link]
by 90° (see Fig. 2) and considering the appropriate source (vertical edu/People/RBHerrmann/[Link] (accessed March 2015).
force or explosive for Rayleigh waves and horizontal force for Love Kudo, K., Shima, E., 1970. Attenuation of shear wave in soil. Bull. Earthquake Res. Inst. 48,
145–158.
waves).
Lai, C.G., Rix, G.J., 1998. Simultaneous Inversion of Rayleigh Phase Velocity and Attenua-
Needless to say that, on a higher level, the analyses would highly tion for Near-Surface Site Characterization Contract Report Prepared for the National
benefit from the use of 3-component geophones that would allow the Science Foundation under Grant no. CMS-9402358, and the U.S. Geological Survey,
Department of Interior, Award No. 1434-95-G-2634, Report No. GIT-CEE/GEO-98-2.
joint analysis of all the three components (radial and vertical compo-
p. 258.
nents of Rayleigh waves plus Love waves). Lai, C.G., Rix, G.J., 2002. Solution of the Rayleigh eigenproblem in viscoelastic media. Bull.
It is important to underline that the presence of higher modes does Seismol. Soc. Am. 92, 2297–2309.
not have to be considered as a problem: the critical point is represented Lee, W.B., Solomon, S.C., 1978. Simultaneous inversion of surface wave phase velocity and
attenuation: Love waves in Western North America. J. Geophys. Res. 83 (B7),
by the correct understanding and processing of the velocity spectra. Ac- 3389–3400.
tually, when correctly considered, due to their higher velocities, higher Liu, Z., de Boer, R., 1997. Dispersion and attenuation of surface waves in a fluid-saturated
modes are extremely useful for better constraining the model, in partic- porous medium. Transp. Porous Media 29, 207–223.
Luo, Y., Xia, J., Liu, J., Liu, Q., Xu, S., 2007. Joint inversion of high-frequency surface waves
ular for the deepest layers. with fundamental and higher modes. J. Appl. Geophys. 62 378–38.
When the energy related to specific modes unfolds following an Malagnini, L., Herrmann, R.B., Biella, G., de Franco, R., 1994. Rayleigh waves in Quaternary
easily-interpretable trend, a so-to-speak standard inversion based on alluvium from explosive sources: determination of shear-wave velocity and Q struc-
ture. Bull. Seismol. Soc. Am. 85, 900–922.
the modal dispersion curves can be performed (Dal Moro and Ferigo, Malagnini, L., Herrmann, R.B., Mercuri, A., Opice, S., Biella, G., de Franco, R., 1997. Shear-
2011). On the other side, when the velocity spectra are particularly wave velocity structure of sediments from the inversion of explosion-induced
complex and intricate, then the analysis can benefit from the Full Veloc- Rayleigh waves: comparison with cross-hole measurements. Bull. Seismol. Soc. Am.
87, 1413–1421.
ity Spectrum approach adopted in the present work, considering the ve- Ohori, M., Nobata, A., Wakamatsu, K., 2002. A comparison of ESAC and FK methods of es-
locity spectra as a whole and thus potentially better handling their timating phase velocity using arbitrarily shaped microtremor analysis. Bull. Seismol.
complexities. Soc. Am. 92, 2323–2332.
O'Neill, A., Matsuoka, T., 2005. Dominant higher surface-wave modes and possible inver-
sion pitfalls. J. Environ. Eng. Geophys. 10, 185–201.
Acknowledgements O'Neill, A., Dentith, M., List, R., 2003. Full-waveform P-SV reflectivity inversion of surface
waves for shallow engineering applications. Explor. Geophys. 34, 158–173.
The authors are deeply grateful to the Deanship of the Scientific Re- Panza, G.F., 1989. Attenuation measurements by multimode synthetic seismograms. In:
Cassinis, R., Nolet, G., Panza, G.F. (Eds.), Digital Seismology and Fine Modeling of
search of the King Saud University (Riyadh, Saudi Arabia) for generously the Lithospehere. Plenum Publishing Corporation, pp. 79–115.
supporting the present work (PRG-1436-06 Research Grant). Park, C.B., Xia, J., Miller, R.D., 1998. Imaging dispersion curves of surface waves on multi-
Authors also wish to express their gratitude to Prof. Shtivelman and channel record. Proceedings 68th Ann. Internat. Mtg., Soc. Explor. Geophys. Expanded
Abstracts, pp. 1377–1380.
to a further anonymous reviewer whose comments significantly helped Pecorari, C., 2000. Attenuation and dispersion of Rayleigh waves propagating on a cracked
in improving the manuscript. surface: an effective field approach. Ultrasonics 38, 754–760.
Picozzi, M., Albarello, D., 2007. Combining genetic and linearized algorithms for a two-
step joint inversion of Rayleigh wave dispersion and H/V spectral ratio curves.
References Geophys. J. Int. 169, 189–200.
Prasad, M., Meissner, R., 1992. Attenuation mechanisms in sands; laboratory versus theo-
Barton, N., 2007. Rock Quality, Seismic Velocity, Attenuation and Anisotropy. Taylor &
retical (Biot) data. Geophysics 57, 710–719.
Francis (729 pp.).
Safani, J., O'Neill, A., Matsuoka, T., Sanada, Y., 2005. Applications of Love wave dispersion
Beaty, K.S., Schmitt, D.R., Sacchi, M., 2002. Simulated annealing inversion of multimode
for improved shear-wave velocity imaging. J. Environ. Eng. Geophys. 10, 135–150.
Rayleigh wave dispersion curves for geological structure. Geophys. J. Int. 151,
Simons, F.J., van der Hilst, R.D., Montagner, J.P., Zielhuis, A., 2002. Multimode Rayleigh
622–631.
wave inversion for heterogeneity and azimuthal anisotropy of the Australian upper
Begemann, H.K.S., 1965. The friction jacket cone as an aid in determining the soil profile.
mantle. Geophys. J. Int. 151, 738–754.
Proceedings of the 6th International Conference on Soil Mechanics and Foundation
Tokimatsu, K., Tamura, S., Kojima, H., 1992. Effects of multiple modes on Rayleigh wave
Engineering, ICSMFE, Montreal, September 8–15 1965, Volume 2, pp. 17–20.
dispersion characteristics. J. Geotech. Eng. ASCE 118 (10), 1529–1543.
Dal Moro, G., 2008. Vs and Vp vertical profiling via joint inversion of Rayleigh waves and
Tonn, R., 1991. The determination of the seismic quality factor Q from VSP data: a com-
refraction travel times by means of bi-objective evolutionary algorithm. J. Appl.
parison of different computational methods. Geophys. Prospect. 39, 1–27.
Geophys. 66, 15–24.
Van Veldhuizen, D.A., Lamont, G.B., 1998. Evolutionary computation and convergence to a
Dal Moro, G., 2010. Insights on surface-wave dispersion curves and HVSR: joint analysis
Pareto front. In: Koza, John R. (Ed.), Late Breaking Papers at the Genetic Programming
via Pareto optimality. J. Appl. Geophys. 72, 29–140.
1998 Conference. Stanford University, pp. 221–228.
Dal Moro, G., 2011. Some aspects about surface wave and HVSR analyses: a short over-
White, R.E., 1992. The accuracy of estimating Q from seismic data. Geophysics 57,
view and a case study. Boll. Geofis. Teor. Appl. 52, 241–259.
1508–1511.
Dal Moro, G., 2014. Surface Wave Analysis for Near Surface Applications. Elsevier 978-0-
Winsborrow, G., Huwsa, D.G., Muyzertb, E., 2003. Acquisition and inversion of Love wave
12-800770-9 (252 pp.).
data to measure the lateral variability of geo-acoustic properties of marine sediments.
Dal Moro, G., 2015. Joint inversion of Rayleigh-wave dispersion and HVSR of lunar seismic
J. Appl. Geophys. 54, 71–84.
data from the Apollo 14 and 16 sites. Icarus 254, 338–349.
Xia, J., Miller, R.D., Park, C.B., Tian, G., 2002. Determining Q of near-surface materials from
Dal Moro, G., Ferigo, F., 2011. Joint analysis of Rayleigh and Love wave dispersion for near-
Rayleigh waves. J. Appl. Geophys. 51, 121–129.
surface studies: issues, criteria and improvements. J. Appl. Geophys. 75, 573–589.
Xu, C., Stewart, R.R., 2006. Estimating seismic attenuation (Q) from VSP data. CSEG Re-
Dal Moro, G., Pipan, M., Forte, E., Finetti, I., 2003. Determination of Rayleigh wave disper-
corder 31, 57–61.
sion curves for near surface applications in unconsolidated sediments. Proceedings
Yang, Y., Forsyth, D., Weeraratne, D., 2007. Seismic attenuation near the East Pacific Rise
73rd Ann. Internat. Mtg., Soc. Explor. Geophys. Expanded Abstracts, pp. 1247–1250.
and the origin of the low-velocity zone. Earth Planet. Sci. Lett. 258, 260–268.
Dal Moro, G., Coviello, V., Del Carlo, G., 2014. Shear-wave velocity reconstruction via un-
Zitzler, E., Thiele, L., 1999. Multiobjective evolutionary algorithms: a comparative case
conventional joint analysis of seismic data: a case study in the light of some theoret-
study and the strength Pareto approach. IEEE Trans. Evol. Comput. 3, 257–271.
ical aspects. Proceedings of the IAEG (International Association for Engineering
Geology and the Environment) XII CONGRESS — Turin, September 15–19, 2014, in