You are on page 1of 5

Computers & Geosciences 28 (2002) 851855

Short note
Scat-Cad: a Mathcad 2000 professional package to model
the energy decay due to seismic attenuation
$
Francesca Bianco*, Edoardo Del Pezzo
Osservatorio VesuvianoFIstituto Nazionale di Geosica e Vulcanologia, Via Diocleziano 328, 1-80124 Napoli, Italy
Received 8 January 2001; received in revised form 20 September 2001; accepted 24 September 2001
1. Introduction
An important task in seismology is the estimation of
attenuation parameters for the earth medium, as they
permit the reconstruction of the source spectral shape.
Further, knowledge of the attenuation adds information
on the constitution of the earths interior. In particular,
it allows the modeling at small scale of site effects which
contribute to seismic hazard studies.
Several methods are frequently used to obtain this
information from seismological data. Most of them are
based on the estimation of the spectral decay coefcient
with hypocentral distance, which is proportional to the
inverse of the total quality factor of the earth medium.
The attenuation parameter estimated in this way
contains the combined effects of intrinsic and scattering
attenuation. Scattering attenuation is produced by
interaction of the primary waves with the elastic
heterogeneity (which can be viewed as space uctuations
of the elastic parameters like velocity, for example). This
interaction produces a secondary radiation, which
comes to the receiver later than the primary. In this
way, the apparent energy of the primary waves is lost.
The intrinsic dissipation is the real energy loss produced
by the transformation of the energy into heat.
It is not always easy to interpret the (total) attenua-
tion results in terms of geological structures: a medium
which contains high heterogeneity may in fact produce a
high-energy decay as a function of sourcereceiver
distance, in apparently the same way in which a
uniform, but highly dissipative medium (as a magmatic
reservoir for example) can do. So, it is necessary to
model the seismic propagation in terms of both
scattering and dissipative attenuation. To perform this
task a scattering model is necessary.
Hoshiba (1993) modeled the propagation in an earth
medium lled by randomly but uniformly distributed
heterogeneity by means of numerical simulations. He
hypothesized isotropic scattering and included multiple
scattering, providing as output a set of energydistance
curves evaluated as a function of total attenuation and
seismic albedo (L
1
e
and B
0
; respectively). These two
parameters are dened as a combination of intrinsic and
scattering attenuation coefcients, or what is the same,
of the intrinsic and scattering quality factors. B
0

Q
T
=Q
S
; where Q
T
and Q
S
are, respectively, the total and
the scattering quality factors; L
1
e
Q
1
S
Q
1
I
o=v
where Q
I
is the intrinsic quality factor, o is the angular
frequency and v is the seismic wave velocity. More
recently, Hoshiba et al. (2001) included in the simulation
both velocity and Q
S
as a function+ of depth. Results
obtained using numerical simulations for a half-space
with constant velocity and scattering coefcient are in
agreement with the analytical solution given by Zeng
et al. (1991). Moreover Zeng (1991) shows that the so-
called hybrid-single-scattering-diffusion approximation
of the analytical solution described in the same paper is
good for both weak and strong scattering.
The multiple lapse time window analysis method
(MLTWA) is generally applied in the literature (Sato
and Fehler, 1998) to t the model to the experimental
data. This method consists in considering the energy
integral calculated in three successive time windows, as a
function of hypocentral distance. The shapes of the three
energy curves are simultaneously t to the corresponding
curves deduced by the model, in terms of B
0
and L
1
e
:
Recent studies have been performed applying the
MLTWA method in order to separate intrinsic from
$
URLs: Code on Server at http://www.iamg.org/CGEditor/
index.htm
*Corresponding author. Tel.: +39-81-6108-328; fax: +39-
81-6108-351.
E-mail address: bianco@ov.ing.it (F. Bianco).
0098-3004/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 8 - 3 0 0 4 ( 0 1 ) 0 0 1 0 7 - 8
scattering attenuation coefcients in several areas of the
world. These studies show that scattering attenuation
plays an important role in determining the seismogram
shape. In particular, scattering phenomena seem to be
strong at shallow depth, where most of the seismic
energy release occurs (La Rocca, 1999). An important
task is, for this reason, the re-interpretation of the
observational data of the spectral energy decay with
distance, which allowed the estimation of the sole total
attenuation coefcient. The problem is to nd the
relationship between total and scattering attenuation.
For example, data from active volcanoes seem to
indicate that intrinsic attenuation is not the predomi-
nant mechanism of dissipation. If these results are
conrmed, all the attenuation tomography results must
be revised at least in their interpretation.
The present study is motivated by the previous
considerations. It is important to know the relative
amount of scattering and intrinsic dissipation especially
for shallow geological structures, like those underlying
volcanic areas.
A computer program based on a visual interactive
approach may be useful for observatory practice. This is
the reason why we developed the MLTWA method in
Mathcad
2000
professional
; a language widely used in the
scientic and engineering community for its power and
at the same time ease of application. We develop
spreadsheets with two included routines, the rst dealing
with the model and the second with the best tting
procedures. An example of application is then carried
out using data already published in literature.
2. MLTWA method in Mathcad
2000
professional
2.1. Why Mathcad
2000
professional
Mathcad
2000
professional
is a standard calculation software
for technical professionals, and its friendly use enhances
the diffusion and the understanding of the details of
many algorithms to a wide non-specialist community,
including students. In the seismological community, the
analysts in seismic laboratories may nd advantagous in
its use for solving routine problems, as for example
magnitude evaluation and spectral analysis. In particu-
lar, the application of MLTWA method on a routine
basis may favor the completeness of the data analysis.
Mathcad
2000
professional
allows the use of symbolic routines
(based on MapleV
TM
) together with numerical proce-
dures in the same spreadsheet. It is based on the
Windows 98 operating system and takes advantage of
Microsofts OLE 2 object linking to work with other
Windows 98 applications. Its graphic system is easy to
use and at the same time sufciently powerful to
visualize all the program steps allowing an effective
control of the program functioning. Moreover, it is
connectable to other Windows 98 applications or data
sources through MATHCONNEX, a visual environ-
ment which creates a good control of the data ux
through the different applications.
On the other hand Mathcad
2000
professional
may suffer
crashing problems related to the use of the Windows
98 operating system, and for many tasks may be slower
than other programming languages.
2.2. Theoretical background
Considering a random isotropic scattering medium,
the scattered wave energy E~rr; t may be expressed in
terms of the following integral equation (Zeng, 1991):
E~rr; t E
in
t
j~rr ~rr
o
j
v
_ _
e
Zj~rr~rr
o
j
4pj~rr ~rr
o
j

_
V
Z
s
E ~rr
1
; t
j~rr
1
~rrj
v
_ _
e
Zj~rr
1
~rrj
4pj~rr
1
~rrj
2
d~rr
1
; 1
where E
in
is the incident wave energy, ~rr the position of
the receiver, ~rr
1
the position of the scatterer, Z the
coefcient of total attenuation and Z
s
the coefcient of
scattering attenuation.
The rst term on the right-hand side of Eq. (1)
represents the incident wave energy from the source
located at the original point, whereas the second term is
a linear superposition of scattered waves energies
contributed from all possible scatterer points ~rr
1
to the
receiver ~rr: The exact solution of Eq. (1) agrees well with
Monte Carlo simulation results (Hoshiba, 1993) for the
scattering order n up to 7. Moreover, in case of weak
scattering, both the single scattering model and the
energy ux model (Sato and Fehler, 1998) are in good
agreement with Eq. (1), whereas the diffusion model has
a better agreement only in the situation of strong
scattering. Gathering these observations up, Zeng (1991)
modied Eq. (1) in order to obtain the following
approximate solution for multiple scattered wave
energy:
Er; t CE
0
e
Zvt
d
t r=v
4pvr
2
Z
s
Ht r=v
4pvrt
ln
1 r=vt
1 r=vt
_ _
cH t
r
v
_ _
3Z
s
4pvt
_ _
3=2
e
3Z
s
r
2
=4vtZ
i
vt
2
with
c E
0
1 1 Z
s
vte
Z
s
vt

4=

p
_ _

3Z
s
vt
_
=2
0
e
a
2
a
2
da
;
where E
0
is the energy at t 0; H is the Heaviside
function, a vt=r and Z
i
and Z
s
are expressed in terms of
F. Bianco and E. D. Pezzo / Computers & Geosciences 28 (2002) 851855 852
L
1
e
and B
0
as follows:
L
1
e
Z
s
Z
i
;
B
0

Z
s
Z
s
Z
i
:
Zeng (1991) called the solution of Eq. (2) the hybrid-
single-scattering-diffusion solution.
We adopted this approximation to model absorption
and scattering in the crust. The approximate solutions of
Eq. (2) are in good agreement with the general solution
of Eq. (1) for both strong and weak scattering at
different source and receiver congurations.
2.3. The MLTWA technique
The seismogram can be separated into three different
parts: the rst starting from the onset of the direct S-
wave and stopping at the end of the early coda; the
second including the middle part of the coda; and the
third the last part. It is experimentally well established
that the rst window contains a signicant contribution
of the direct S-wave energy, whereas the last two contain
a predominant contribution of the scattered wave
energy. Eq. (2) is used to calculate the time integral of
the energy density for each window, obtaining three
energy curves as a function of the hypocentral distance,
as follows:
E
i
*
r
_
t
i1
r
t
i
r
Er; t dt; i 1; y; 3: 3
The time values t
i
r reconstruct three successive time
windows of appropriate length, starting from the S wave
onset r=v: These energy curves are parameterized in term
of the seismic albedo B
0
Z
s
=Z
s
Z
i
and of the total
attenuation L
1
e
; where L
e
1=Z
s
Z
i
is the extinction
distance over which the primary S-wave energy is
decreased by e
1
: The integrals in Eq. (3) are calculated
for various B
0
and L
1
e
; normalized to take into account
different source power and different site amplication.
Then, the coda normalization method (Aki, 1980) is
applied. This method consists in dividing E
i
*
r by the
average coda energy calculated at a xed reference time
interval t
*
, evaluated at later lapse times (at least twice
the S wave time travel). Additionally, E
1
*
r; E
2
*
r and
E
3
*
r are corrected for geometrical spreading, multi-
plying by the squared hypocentral distance. The
corrected integrals are consequently written as
E
i
r log
E
i
*
=t
i1
r t
i
r
Er; t
*

_
_
_
_
_
_
_
_
4pr
2
_

_
_

_
; i 1; y; 3:
4
Scat-Cad package calculates E
i
r; B
0
; L
1
e
for i
1; y; 3 after setting the appropriate range of variability
for the seismic albedo and the extinction length. The
input values are: the duration of the time window over
which to calculate the energy integrals (e.g. 15 s); the S
wave velocity (e.g. 4 km=s) and the energy at the source
(e.g. E
0
1).
2.4. Experimental data t
Once that experimentally evaluated energies in three
successive windows are available for a set of earthquakes
spanning a large hypocentral distance interval, the data
can be t to the model. The method for obtaining the
energy integrals as a function of distance from the
experimental data is described in Hoshiba (1993). In our
program, following Hoshiba et al. (1991) we use the
mist function between data and model given by
MB
0
; Le
1

N
i1

3
k1
E
k
r
%
E
k
r
2
; 5
where
%
E
k
r are the energies measured from the
experimental data. The minimum of MB
0
; L
1
e
corre-
sponds to the best estimate of B
0
and L
1
e
in the sense of
least squares.
In Mathcad
2000
professional
we graphically search the mini-
mum in a grid of possible values of the parameters, using
the same approach utilized by many authors (see e.g.
Hoshiba, 1997). Errors can be estimated using the F
distribution at 90% level of condence. For this
distribution the ratio of two random variables, each
having a given number, Nd; of degrees of freedom, is
tabulated. If we plot the mist function normalized by
its maximum, we can dene the range of the possible
solutions, accepting those lower than on the tabled
values. An example of application of this procedure is
reported in Mayeda et al. (1992).
2.4.1. The program
The program is developed in Mathcad
2000
professional
data
sheets, in which two different routines have been carried
out. The rst produces the model output and the second
performs the t between data and model. Due to
the ease in writing symbolic mathematics in
Mathcad
2000
professional
the 3-D formulas can be easily
modied and a 2-D multiple scattering model can be
easily adapted to the experimental data.
2.4.2. Test of the program and an example of application
A test of the program is carried out generating
E
1
r; E
2
r and E
3
r (see Eq. (4)) for the same values of
B
0
and L
1
e
used by Hoshiba et al. (1991), who simulated
the energy envelope in case of multiple scattering (up to
the 8th order) using a Monte-Carlo approach. The
results show a substantial superposition of the curves
obtained with the present program with those obtained
by numerical simulation. The maximum discrepancy (of
few percent) can be observed for values of distance lower
than 10 km for the rst energy integral (E
1
r). This test
indicates that the approximation used in the present
F. Bianco and E. D. Pezzo / Computers & Geosciences 28 (2002) 851855 853
approach (Eq. (2)) can be considered sufciently good
except for earthquakes located close to the receiver.
Those data must be disregarded by the analysis if made
using the present computer program.
Finally, an application, reported in a word le on
the IAMG server at http://www.iamg.org/CGEditor/
index.htm, is carried out on data from Southern Spain
described by Akinci et al. (1995). We use their data with
hypocentral distance between 10 and 80 km: The best t
carried out using the Mathcad
2000
professional
program yelds
values of the parameters B
0
and L
1
e
comparable with
those obtained by Akinci et al. (1995), taking into
account the respective error bounds. In Fig. 1, the
theoretical normalized energy curves obtained by means
of Scat-Cad package as well as the theoretical normal-
ized energy curves with the experimental data from the
paper by Akinci et al. (1995) are reported.
3. Concluding remarks
Most of the papers dealing with the problem of
separating intrinsic from scattering attenuation through
MLTWA use a set of energy curves, numerically
generated in the hypothesis of isotropic multiple
scattering. In the present paper, we use the approxima-
tion developed by Zeng (1991), which assumes single
scattering for the early S-coda and a diffusion regime for
the late coda. Such an approximation has been
successfully used by Feustel and Young (1994) who
examined a set of seismic data from an underground
research laboratory in granite. We observed a slight
discrepancy of the approximate model with the non-
approximate numerical simulation at short distances
(less than 10 km) and this mainly for E
1
r; the energy
integral relative to the early S-wave coda (Fig. 1). The
differences between the numerical approximation by
Akinci et al. (1995) and the Scat-Cad output are
generated by the hybrid-single-scattering-diffusion solu-
tion (Zeng, 1991) we adopted.
We adapted this method to a Mathcad
2000
professional
work-
sheet. Using this worksheet, non-expert programmers
can easily modify the formulas in order to use models
different from the 3-D multiple scattering adopted in the
present report. We hope that the present worksheet will
help in routine application of the MLTWA methodology
in seismological observatory practice.
Acknowledgements
The authors thank J. Iba* nez and Aybige Akinci for
useful discussions. This study has been supported by
Italian MURST (Ministery of Scientic Research)
Grant # 2.13.10.01.
References
Aki, K., 1980. Scattering and attenuation of shear waves in the
lithosphere. Journal of Geophysical Research 85, 64966504.
Fig. 1. Normalized energy curves (solid lines) with super-
position of experimental data by Akinci et al. (1995) and
theoretical normalized energy curves (dashed lines) obtained by
means of Scat-Cad using same values of seismic albedo and
total attenuation. Values used are. L
1
e
0:042 and B
0
0:66
for the 12 Hz frequency band; L
1
e
0:042 and B
0
0:54 for
24 Hz frequency band; L
1
e
0:038 and B
0
0:36 for 8
10 Hz frequency band; L
1
e
0:038 and B
0
0:36 for 12
14 Hz frequency band. Slight differences among solid and
dashed lines are due to hybrid-single-scattering diffusion
solution (Zeng, 1991) that we adopted in Scat-Cad package.
F. Bianco and E. D. Pezzo / Computers & Geosciences 28 (2002) 851855 854
Akinci, A., Del Pezzo, E., Iban ez, J., 1995. Separation of
scattering and intrinsic attenuation in southern Spain and
Western Anatolia (Turkey). Geophysical Journal Interna-
tional 121, 337353.
Feustel, A.J., Young, R.P., 1994. Q
b
estimates from
spectral ratios and multiple lapse time window analysis:
results from an underground research laboratory in
granite. Geophysical Research Letters 21 (14),
15031506.
Hoshiba, M., 1993. Separation of scattering attenuation and
intrinsic absorption in Japan using the multiple lapse time
window analysis of full seismogram envelope. Journal of
Geophysical Research 98, 1580915824.
Hoshiba, M., 1997. Seismic coda wave envelope in depth-
dependent S wave velocity structure. Physics of the Earth
and Planetary Interiors 104, 1522.
Hoshiba, M., Rietbrock, A., Scherbaum, F., Vakahara, H.,
Haberland, C., 2001. Scattering attenuation and
intrinsic absorption using uniform and depth dependent
modelFapplication to full seismogram envelope
recorded in Northern Chile. Journal of Seismology 5,
157179.
Hoshiba, M., Sato, H., Fehler, M., 1991. Numerical basis of the
separation of scattering and intrinsic absorption from full
seismogram envelopeFa Montecarlo simulation of multiple
isotropic scattering. Papers in meteorology and Geophysics
42 (2), 6591.
La Rocca, M., 1999. Multichannel array tecniques applied to
volcanic hazard. Ph.D. Dissertation. University of Salerno-
Faculty of Sciences, Salerno, 158pp.
Mayeda, K., Koyanagi, S., Hoshiba, M., Aki, K., Zeng, Y.,
1992. A comparative study of scattering, intrinsic, and coda
Q
1
for Hawaii, Long Valley and Central California
between 1.5 and 15 Hz: Journal of Geophysical Research
97, 66436659.
Sato, H., Fehler, M.C., 1998. Seismic Wave Propagation and
Scattering in the Heterogeneous Earth. Springer, New
York, 308pp.
Zeng, Y., 1991. Compact solutions for multiple scattered wave
energy in time domain. Bulletin of the Seismological Society
of America 81, 10221029.
Zeng, Y., Su, F., Aki, K., 1991. Scattering wave energy
propagation in a random isotropic scattering medium. 1.
Theory. Journal of Geophysical Research 96, B1, 607620.
F. Bianco and E. D. Pezzo / Computers & Geosciences 28 (2002) 851855 855

You might also like