2nd Reading

June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

Journal of Earthquake and Tsunami

Vol. 8, No. 4 (2014) 1450008 (26 pages)

c World Scientific Publishing Company
DOI: 10.1142/S1793431114500080

CU-PSHA: A MATLAB Software for Probabilistic

Seismic Hazard Analysis

Santi Pailoplee, and Chitti Palasri

Earthquake and Tectonic Geology Research Unit (EATGRU)

Department of Geology, Faculty of Science, Chulalongkorn University

J. Earthquake and Tsunami Downloaded from www.worldscientific.com

Bangkok, 10330, Thailand

Department of Civil Engineering, Faculty of Engineering
Chulalongkorn University, Bangkok, 10330, Thailand
[email protected]
by WSPC on 07/30/14. For personal use only.

Received 17 October 2013

Accepted 10 February 2014
Published 2 July 2014

In this study, an open source MATLAB software, called CU-PSHA, is developed in

order to analyze probabilistic earthquake hazards. This software aims to provide a user
friendly and flexible tool for evaluating reliable earthquake hazard estimates. With the
CU-PSHA, the probability of distances between the earthquake sources and the study
site can be estimated. Two choices for the estimation of earthquake frequencymagnitude
distribution, the exponential magnitude distribution and the characteristic earthquake
models, are provided. Some strong groundmotion attenuation models are available for
both shallow crustal and subduction zone earthquakes. The probability of exceedance
of any individual given ground shaking value can be obtained, allowing the display of a
seismic hazard curve. In addition with the supplementary MATLAB scripts, this CU-
PSHA software can be employed in general seismic hazard mapping, for both ground
shaking level and probability of occurrence, in any specific given time span.

Keywords: Earthquake; probabilistic seismic hazard analysis; MATLAB.

1. Introduction
In earthquake hazard mitigation, seismic hazard analysis (SHA) [Kramer, 1996] is
one of the most widely used scientic methods for determining earthquake ground
shaking based on the known seismic sources and suitable models of the strong
groundmotion attenuation. Conceptually, the SHA widely used nowadays can be
divided into two methods [Kramer, 1996]. The rst method is the deterministic
seismic hazard analysis (DSHA) [Krinitzsky, 2003], which aims to evaluate the
maximum possible ground shaking value, i.e. worst case scenario, that is likely to
occur at any site of interest, whilst the second method is the probabilistic seismic

Corresponding author.

1450008-1
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

hazard analysis (PSHA), which allows quantitative estimation of the probability

of exceeding various ground shaking levels at a site, given all possible earthquake
sources, within a reference time interval [Cornell, 1968].
In contrast to the DSHA, the methodology of PSHA is more complicated and is
dicult to solve manually. In this work, simplied MATLAB codes are, therefore,
developed for the PSHA to allow accurate solutions, which hereafter are called the
CU-PSHA, where CU refers to Chulalongkorn University. The contribution of this
open source MATLAB code aims to mainly provide researchers with a user friendly
set of tools to evaluate the PSHA in any region.

2. PSHA Method
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

PSHA is a well-established method for estimating the hazardous ground shaking

probabilistically in any individual site of interest over a relatively long-term period.
The output ground shaking levels are calculated by considering the possible earth-
by WSPC on 07/30/14. For personal use only.

quake magnitudes, locations, and the attenuation of the ground shaking as a dier-
ent probability of occurrence. Mathematically, the PSHA approach estimates the
probability, , that a particular ground shaking level, A, is equal to or exceeds the
ground shaking level, A0 , as shown in Eq. (1) [Cornell, 1968].

Ns 
(A A0 ) = i fMi (m)fRi (r)P [A(m, r) A0 | m, r]dmdr , (1)
i=1
where fMi (m) is the probability density function of the earthquake magnitude that
describes the probability of an earthquake occurrence having a magnitude in a given
range, fRi (r) is the probability density function for the distance from the earthquake
source to the site, and P [A(m, r) A0 | m, r] is the probability of exceeding the
threshold value A0 , under the condition that an event of magnitude m occurred
at source-to-site distance r. Note that the value of P [A(m, r) A0 | m, r] depends
on the strong groundmotion attenuation model used. The coecient i represents
the activity rate, which implies the average rate of occurrence of earthquakes for
an individual earthquake source i from the total earthquake sources (Ns).

2.1. Probability density function of earthquake magnitude

In general, the probability density function fMi (m) of a given earthquake source
follows the GutenbergRichter (GR) relationship [Gutenberg and Richter, 1944],
as illustrated in Eq. (2).
N (m) = 10abm = e(m) , (2)
where N (m) is the number of events that are equal to or larger than a given
magnitude m, and a and b are constants that depend on the empirical relationship
between m and N (m), = 2.303b and = 2.303a.
However, the GR relationship usually fails for very large earthquakes due to
the instrumental or even historical records covering too short a time span compared

1450008-2
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

to the recurrence interval of large earthquakes. Usually, the recurrence interval of

large earthquakes can be determined from the geological record, for example, from
fault slip rates [Youngs and Coppersmith, 1985]. Currently, there are two candidate
models that are applied to determine the fMi (m) value of an individual earthquake
source, and these are described below.

Model 1: Exponential magnitude distribution model

Youngs and Coppersmith [1985] proposed the exponential magnitude distribution
approach to deal with the possible maximum (mmax ) and minimum (mmin ) mag-
nitudes, as expressed in Eq. (3).


0 for m < mmin ,


e(mmmin)
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

fMi (m) = for mmin m mmax , (3)

1 e(mmax mmin )



0 for m > mmax .
by WSPC on 07/30/14. For personal use only.

These authors also determined an expression for the activity rate (i ) for the
exponential magnitude distribution model, as expressed in Eq. (4).

Af S(c b)[1 e(mmax mmin ) ]

i = , (4)
bM max
0 e(mmax mmin )

where is the rigidity or shear modulus (usually taken to be 3102 newtons/m2 ),

Af is the rupture area (km2 ), S is the slip rate (mm/year), M0max is the seismic
moment for mmax . The value of the constants c and d at 1.5 and 16.1, respectively,
were proposed by Hanks and Kanamori [1979].

Model 2: Characteristic earthquake model

This model is based on the hypothesis that an individual earthquake source, and
in particular the active faults, tend to generate similar-sized or characteristic
earthquakes and that these characteristic earthquakes occur on a fault not to the
exclusion of all other magnitudes, but with a nonexponential frequency distribution
[Youngs and Coppersmith, 1985; Convertito et al., 2006]. When the characteristic
earthquake model is assumed, it is possible to formulate the corresponding fMi (m)
as shown in Eq. (5).


0 for m < mmin ,





e(mmmin ) 1

for mmin m mchar

1e (m max m min ) 1+c

= mmax m2 ,
fMi (m) = (5)

e (mmax mmin m1 m2 )
1

for mmax m2

1 e(mmax mmin m2 ) 1 + c



= mchar m mmax ,




0 for m > mmax .

1450008-3
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

Here, the constant c in Eq. (5) is given by Eq. (6):

e(mmax mmin m1 m2 )
c= m2 , (6)
1 e(mmax mmin m2 )
where the parameters , m, mmin , and mmax are the same as above, m1 and m2
represent the intervals below and above the magnitude level mchar , respectively (see
the details in Convertito et al. [2006]). Youngs and Coppersmith [1985] also showed
that the activity rate (i ) is given by Eqs. (7)(9).
m2 e(mmax mmin m1 m2 )
char = NC , (7)
1 e(mmax mmin m2 )
Af S[1 e(mmax mmin m2 ) ]
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

NC = , (8)
KMmax
0 e(mmax mmin m2 )
b10cm2 be m1 (1 10cm2 )
K= . (9)
cb
by WSPC on 07/30/14. For personal use only.

2.2. Probability distribution of the source-to-site distance

The probability distribution for the distance from the site of interest to earthquake
rupture for any source can be computed based on the obtained location and geom-
etry of the earthquake source. At rst, the distances r, from the site of interest to
the earthquake point sources are measured point-by-point. Note that this assumes
that ruptures cannot occur beyond the boundaries of the considered earthquake
source zone, and that ruptures occur with equal likelihood anywhere within the
earthquake sources.

2.3. Probability of exceeding a given threshold value

The threshold value A0 is the prospected value of the ground shaking in terms
of the Peak Ground Acceleration (PGA) or even the Modied Mercalli Intensity
(MMI) scale. According to Kramer [1996], the probability that a target PGA (A0 )
will be exceeded should an earthquake in the given magnitude range occur at the
given distance interval P [A(m, r) A0 | m, r] can be analyzed as follows:
In the rst step, for the individual m and r values obtained from Secs. 2.1 and 2.2
are used to calculate the PGA based on suitable strong groundmotion attenuation
models. The obtained PGA values are then converted to the mean of all possible
PGA values (PHA), which can vary according to the uncertainty represented by
the standard deviation () of the possible ground shaking variations.
Secondly, from the given PHA and , the probability (P [A(m, r) A0 | m, r]) is
then computed from Eq. (10), where is the probability according to the normal
distribution.
 
log(A0 ) log PHA
P [A(m, r) A0 | m, r] = 1 . (10)

1450008-4
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

Finally, given fM (m), fR (r), P [A(m, r) A0 | m, r] and , Eq. (1) is employed

to calculate the mean annual exceedance rate ((A A0 )) for each target A0 in
each area that was analyzed by the PSHA.

3. MATLAB Algorithm
As per the method described above, the data, parameters and models required
for PSHA consist of (1) the location of the study area, (2) the locations and
geometries of all recognized earthquake sources, (3) the earthquake parameters
(e.g. mmax , mmin , a, b, S, Af etc.), indicating the earthquake potential of each indi-
vidual source, and (4) the suitable strong groundmotion attenuation models for
the study region. As a result, the CU-PSHA software developed here requires some
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

inputs which are in a format that is compatible for analyzing the PSHA.
Regarding the algorithm (Fig. 1), CU-PSHA can work with geographic coordi-
nates in decimal degrees using the .txt format for this calculation. All input les
by WSPC on 07/30/14. For personal use only.

are restored under the root Input/ including:

(a) The Study-area.txt le contains the series of point locations (i.e. longitude
(deg) and latitude (deg) in columns 1 and 2, respectively) obtained from the
grid covering of the PSHA study area. It is notable that the more detailed the
grid spacing was then the clearer the mapping of the seismic hazard is.
(b) For the earthquake source les, all point, line, and even area sources can be
applied with this CU-PSHA. However, they all should be gridded similar to
the Study-area.txt le. Individual les of earthquake sources contain three
columns of point sources, the longitude, latitude, and depth from the surface
(km). The nomenclature of each individual earthquake source le should be
numbered accordingly, for instance, 1.txt, 2.txt, and so on.
(c) The EQ-parameters.txt le contains ten columns of earthquake source
parameters, comprised of (1) the earthquake source number (Sn), which should
be identical and compatible with the name of the individual earthquake source
les, and (2) the fault type (Ft ), for when the earthquake sources are an active
fault, where 1 = strike-slip, 2 = oblique, 3 = reverse, and 4 = normal faults,
respectively. As described before, (3) mmax and (4) mmin are the expected max-
imum and minimum magnitudes, respectively, whilst (5) and (6) are the values
of the constants a and b, respectively, of the GR relationship. Column (7) is
the rate of the fault slip S and (8) Af is the rupture area of the fault. (9) Fmds
is the model selected for analyzing the probability density functions, whilst
nally column (10) Att is the strong groundmotion attenuation models. An
example of this EQ-parameters.txt le is shown in Table 1.
At the rst stage in the PSHA calculation (workow 1 in Fig. 1), the script
PSHA.m calls the Study-area.txt and arranges the number of the PSHA-
calculated points (Sec. 1 in Fig. 1). Thereafter, each individual grid containing
a longitude and latitude value is sent continuously to the script CalPSHA.m for

1450008-5
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

J. Earthquake and Tsunami Downloaded from www.worldscientific.com
by WSPC on 07/30/14. For personal use only.

Fig. 1. Schematic flow chart of the CU-PSHA process.

1450008-6
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

Table 1. The earthquake source parameters for three recognized earthquake sources in the case
study site in Sumatra Island, Indonesia.

EQ source Sn Ft mmax mmin a b S Af Fmds Att

Interplate SASZ 1 3 9.2 4 3.28 0.48 46.7 67608.3 2 4
Intraslab SASZ 2 4 9.2 4 4.13 0.65 46.7 32359.37 2 4
Intraplate SFZ 3 1 8.8 4 4.17 0.68 27 31622.78 2 6

the PSHA calculation (Sec. 2 in Fig. 1). During the PSHA process, the PSHA.m
displays the interface of progress for the PSHA running. For each individual PSHA
calculating point, the script CalPSHA.m calls the EQ-parameters.txt le and
calculates the PSHA as follows:
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

(a) The possible magnitude distributions (fM (m)) are calculated by the script
ProbM.m (Sec. 3 in Fig. 1). This CU-PSHA software allows for selecting the
suitable model by quoting the number in column 9 of the EQ-parameters.txt
by WSPC on 07/30/14. For personal use only.

le using (1) for the exponential magnitude distribution model and (2) for the
characteristic earthquake model. For CU-PSHA, 10 cases between mmin and
mmax are separated. The probability density functions, i.e. Prob-magnitude.txt,
are then determined according to the selected model.
(b) At the same time, the possible source-to-site distances and their probability
(fR (r)) are also calculated by the script ProbR.m (Sec. 4 in Fig. 1). All point-
generated earthquake sources, e.g. 1.txt, are determined for their distance from
source to the site. The probability density function fR (r), i.e. Prob-distance.txt,
is approximated by dividing the distance range into 50 equal intervals between
the shortest and longest distances and then determining how many of the dis-
tances fall within each distance interval, normalized by the total number of
distances possible for the entire earthquake source.
(c) Next, the probability P [A(m, r) A0 | m, r] is calculated by the script CalP-
SHA.m using strong groundmotion models for the required attenuation
(Sec. 5 in Fig. 1). Five attenuation models are available for PSHA analysis in the
CU-PSHA software, selected by quoting the number of the desired model in col-
umn 10 of the EQ-parameters.txt, being the (1) Esteva and Villaverde [1973],
(2) Idriss [1993], (3) Sadigh et al. [1997], (4) Boore et al. [1997], (5) Kobayashi
et al. [2000], (6) Crouse [1991], and (7) Petersen et al. [2004] or modied Youngs
et al. [1997] models. The (4) and (5) models are proposed for use incases of
shallow crustal earthquakes whereas the last two are for subduction zone earth-
quakes. After determining the mean of the possible PGA values (PHA), the
probability P [A(m, r) A0 | m, r] is analyzed according to the obtained PHA
and . For the CU-PSHA software, 200 cases of A0 , starting with 0.005 g and
increasing every 0.01 g up to 1.995 g, are regarded.
(d) The output le Hazard-curve.txt is then generated. This le contains the
annual probability that individual target accelerations A0 (i.e. 0.0051.995 g)
will be exceeded. At any site of interest, this data is useful to clarify and

1450008-7
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

illustrate the seismic hazard curve which represents the probability of exceed-
ing dierent groundmotion levels at a given site [Cornell, 1968; Kramer, 1996].

In PSHA mapping (workow 2 in Fig. 1), the most useful way of presenting the
PSHA results are in the format of a nite time period of interest, depending on the
lifetime of the infrastructure of interest [Cornell, 1968]. Furthermore, in the nite
time period, Kramer [1996] suggested two representative methods to present the
PSHA mapping, as follows.
(1) The ground shaking map describes the ground shaking value (i.e. PGA) that
can be exceeded with any given % probability (Prob) that is of interest (selected)
in any specic time span (T ). Thus, the probability of exceeding the seismic hazard
curve (Prob hazard ) can be evaluated according to Eq. (11) [Kramer, 1996], and
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

then converted to the ground shaking level using the seismic hazard curve in each
individual area.
ln(1 Prob)
Prob hazard =
by WSPC on 07/30/14. For personal use only.

. (11)
T
This procedure is performed by the script Shakemap.m (Sec. 6 in Fig. 1), with
the output le released in terms of a .txt le with the name related to the given
Prob and time period (T ), such as P2Y50.txt.
(2) The probability map reveals the probabilities (Prob) that the ground shaking
may equal or exceed that of the constant ground shaking level of interest (in the
MMI scale) in a given time (T ) period. This procedure is evaluated by the script
Percentmap.m (Sec. 7 in Fig. 1), and the output is released as a .txt le with
the name related to the given ground shaking level of interest and T , such as
MMI6Y50.txt. At rst, the ground shaking of interest is converted from the MMI
scale to the PGA unit (g), according to the PGAMMI relationship, for each specic
area. Based on the literature review, there are some PGAMMI relationships that
have been proposed previously, and are available in this CU-PSHA model, such
as the models of Okamoto [1973], Shabestari and Yamazaki [2001] and others.
Thereafter, the obtained PGA is converted into the probability of exceedance in
the seismic hazard curve (Prob hazard ). Finally, the probabilities of exceedance
(Prob) can be evaluated according to Eq. (12) [Kramer, 1996].

Prob = 1 e((Prob hazard)(T ))

. (12)

4. A Case Study
As a test of the validity and to illustrate the capability of the CU-PSHA software, a
PSHA case study using the CU-PSHA software for the Sumatra island of Indonesia
was performed (Fig. 2). Geologically, there are three major seismotectonic sources
occupied in this region consisting of (i) interplate and (ii) intraslab of the Sumatra
Andaman Subduction Zone (SASZ) including (iii) the intraplate Sumatra Fault
Zone (SFZ).

1450008-8
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

J. Earthquake and Tsunami Downloaded from www.worldscientific.com
by WSPC on 07/30/14. For personal use only.

Fig. 2. Map of Sumatra island and the adjacent areas showing the epicentral distribution of
earthquakes generated by various earthquake sources during 19642010. The SFZ (black lines) is
the intraplate strike-slip faults, lined in a NWSE direction along the Sumatra island.

Due to the interaction between the Indo-Australian and Eurasian plates, the
earthquake activities along these seismic sources are still active, as represented
by the present day seismicity (Fig. 2). For each seismic source mentioned above,
the earthquake parameters were determined from both the instrumental-based
earthquake records and from the paleo-seismological data (Table 1). Based on the
completeness earthquake catalogue recorded instrumentally by the Incorporated
Research Institutions for Seismology during 19642010, the GR relationships of
individual source zone are contributed (Fig. 3). The value of a and b were, then,
determined from the GR relationships as expressed in Table 1.
Regarding to the values of mmax , there are a number of signicant earthquakes
posed along the SASZ such as Mw 8.4 (1779), Mw 8.3 (1861), Mw 9.2 (1833),
Mw 9.2 (2004), Mw 8.4 (2007), and Mw 7.6 (2009) earthquakes. Therefore,
the devastating Mw 9.2 earthquake posed on December 26, 2004 including the
geological records of Mw 9.2 earthquake in A.D. 1883 [Putra et al., 2012] are
applied to be the mmax of the SASZ (Table 1). In case of the SFZ, the length of
SFZ is converted to magnitude around 8.8 Mw using the Wells and Coppersmiths

1450008-9
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

J. Earthquake and Tsunami Downloaded from www.worldscientific.com
by WSPC on 07/30/14. For personal use only.

(a) Interplate SASZ (b) Intraslab SASZ

(c) Intraplate SFZ

Fig. 3. Plots of GR relationships using the completeness mainshock catalogue recorded during
19642010. Triangles reveal the number of earthquakes in individual magnitude level whereas
squares mean the cumulative number of earthquakes of a magnitude equal to or larger than each
magnitude. The solid line is the line of best fit, according to Woessner and Wiemers [2005]
assumption. The Mc is defined as the magnitude of completeness.

[1994] assumption. The Af of all seismic sources were, then, determined from the
obtained Mw values, according to the Af Mw relationship proposed by Wells and
Coppersmith [1994]. The value of mmin is taken to be 4.0 Mw for all earthquake
sources, which assumes that there is no signicant earthquake hazard on engineering
structures [Kramer, 1996].

1450008-10
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

Note that the values of S are cited from previous publications. For instance, the
rate of SASZ slip, Sieh et al. [1999] and Petersen et al. [2008] determined roughly the
slip rates along this subduction zone around 4045 and 2040 mm/yr, respectively.
However, according to the most up-to-date investigation, Megawati and Pan [2009]
and Putra et al. [2012] proposed in detail that the SASZ along the Sumatra island
slips with the average rate 46.7 mm/yr, which applied for the SASZ in this study
(Table 1). In case of the SFZ, Petersen et al. [2004] reveals the rate of fault slip
around 627 mm/yr. Thus, the maximum rate of 27 mm/yr is recognized for the
SFZ in this PSHA investigation.
The Fmds of characteristic earthquake models [Youngs and Coppersmith, 1985],
including the attenuation models (Att) of shallow crustal SFZ [Boore et al., 1997]
and the SASZ [modied Youngs et al., 1997] were selected for this study according
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

to the suggestions of Petersen et al. [2004], Irsyam et al. [2008], and Delfebriyadi
[2011]. However, as mentioned above, the other choices of attenuation models for
the Mainland Southeast Asia are also available in the CU-PSHA software. For
by WSPC on 07/30/14. For personal use only.

instance, the models of Sadigh et al. [1997] and Crouse [1991] proposed by Chin-
tanapakdee et al. [2008] are suitable for Thailand region and the neighborhood
areas.
In the computation procedure, three earthquake sources including the study
area are converted systematically to each individual 0.1 0.1 point. The centers
of grid cells are applied for representing the multi-points of PSHA investigation. For
instance, the probabilities density functions of the earthquake magnitudes along the
SFZ are calculated with the characteristic earthquake model and using the earth-
quake parameters in Table 1. The probability of occurrence decreases exponentially
from 1.8 103 at a Mw of 4.2 down to 4 106 at a Mw of 8.1. After that, the
probabilities of occurrence illustrate a more stable characteristic earthquake being
1.5 105 for a Mw of 8.6 (Fig. 4(a)). In addition, Fig. 4(b) shows the probability
of possible distances measured from Padang city to the SFZ. Due to the location of
Padang city is quite close to the SFZ, the most measured source-to-site distances
are, therefore, around 32 km showing 0.99 probability of occurrence. Meanwhile
for the other, the probabilities of the possible distances during 50.6928 km varies
between 105 and 106 (Fig. 4(b)).
Besides both demonstrated probabilities of possible magnitude and distance, the
seismic hazard curves for the major cities in the study area are also plotted between
the individual PGA against the probability of exceedance (Fig. 4(c)). It is notable
that most cities in the Sumatra island illustrate high seismic hazard. Meanwhile
for Singapore, George Town, and Kuala Lumpur are posed by the comparative low
hazardous levels. For instance at PGA 0.1 g, the highest hazardous Padang city has
0.01 probability of exceedance whereas the lowest hazardous Singapore city is 1/50
time of Padang city (Fig. 4(c)).
In addition, two kinds of PSHA maps are also presented here (Fig. 5) to illus-
trate the analysis. The ground shaking maps are contributed illustrating the 2% and
10% probability of exceeding the given values in a 50-year time period (Figs. 5(a)

1450008-11
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

J. Earthquake and Tsunami Downloaded from www.worldscientific.com

(a) (b)
by WSPC on 07/30/14. For personal use only.

(c)
Fig. 4. (a) Probability density function of the earthquakes might be generated along the SFZ
according to characteristic earthquake model. (b) Probability of the source-to-site distance mea-
sured from the SFZ to Padang city. (c) Probabilistic seismic hazard curve of the major cities in
the study areas derived from considering three earthquake sources.

and 5(b)). Taking a 2% probability of exceedance (Fig. 5(a)), the PGA values
indicate a high seismic hazard (up to 0.75 g) at small islands located oshore of
the Sumatra Island that, however, decreases gradually eastward down to 0.15 g at
Palembang city, Southeastern part of Sumatra island. In case of Singapore, George
Town, and Kuala Lumpur cities, PSHA reveal a low hazard (i.e. 00.2 g) compared
with the Sumatra island (Fig. 5(a)). In addition to the map considering 10% prob-
ability of exceedance, the spatial distribution of the ground shaking levels conform

1450008-12
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

J. Earthquake and Tsunami Downloaded from www.worldscientific.com
by WSPC on 07/30/14. For personal use only.

Fig. 5. Probabilistic seismic hazard maps of Sumatra island and the adjacent areas showing the
PGA distribution that exceeds (a) 2% or (b) 10% probability for return period of 50 years. (c) and
(d) Illustrating the probabilities (%) that ground shaking will be equal to or greater than the MMI
levels VI and VII for return periods of 50 years, respectively.

to those recognizing 2% probability of exceedance. However, in individual site of

interest, the seismic hazards of 10% are lower than 2% around 1.5 time.
Comparing with the previously available PSHA maps, the seismic hazard lev-
els evaluated from this CU-PSHA-based study are lower than that proposed by
Petersen et al. [2004] which the PGA for 10% probability of exceedance in 50-year
time is up to 1 g. However after improving both seismic sources and attenuation
models, the obtained PSHA maps of Petersen et al. [2008] including Irsyam et al.
[2008] illustrate quite similar to those analyzed from the CU-PSHA in terms of

1450008-13
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

both PGA level and spatial distribution, i.e. 0.10.9 g and 0.050.6 g for 2% and
10% probability of exceedance in 50 years. This implies that the CU-PSHA applied
here is eective for analyzing the probabilistic seismic hazard.
For the probability maps, the probabilities of the MMI level VI and VII in 50
years are demonstrated (Figs. 5(c) and 5(d)). Using the PGAMMI relationship
suggested lately by Shabestari and Yamazaki [2001], the Sumatra island is reveled
to have the potential to be aected by the MMI levels VI and VII with 4080%
and 020% probabilities. Meanwhile for the Singapore, George Town, and Kuala
Lumpur cities, the obtained probabilities are less than 15% indicating the area to
be quiescent with respect to ground shaking, at least according to this CU-PSHA
evaluation.
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

5. Summary
by WSPC on 07/30/14. For personal use only.

The CU-PSHA algorithms in MATLAB scripts presented herein provide a relatively

easy-to-understand, exible package for PSHA. With these MATLAB scripts, we
can calculate the probability of the (1) possible source-to-site distances, (2) earth-
quake frequencymagnitude distributions, and (3) exceedance of threshold ground
shaking values, which are useful to contribute to the formation of seismic hazard
curves. Furthermore, for facilitating the users decision, these scripts allow evalua-
tion of (4) the ground shaking level when the probabilities of occurrences are xed
and (5) determining the probability of any ground shaking level of interest, in an
individual specic time span. A particular strength of this system is its output is
simple and friendly to use. As a result, these provide a simple and very useful data
set and analysis for the production of PSHA maps of both the ground shaking and
probability aspects.
The principal weaknesses of the CU-PSHA package are the limited choices of
attenuation models that can be selected in this version and the menu-driven user
interface for input data preparation. Further CU-PSHA development, therefore,
requires a basic level of knowledge in contribution of the user interface and more
choices of the strong groundmotion attenuation relationship that are appropriate
to any region.

Acknowledgments
This work was sponsored by the Research Endowment Fund, Chulalongkorn Uni-
versity. Thanks are also extended to T. Pailoplee for the preparation of the draft
manuscript. We thank the Publication Counseling Unit (PCU), Faculty of Sci-
ence, Chulalongkorn University, for a critical review and improved English. We
acknowledge thoughtful comments and suggestions by Prof. Dr. Fook-Hou Lee, the
Managing Editors and anonymous reviewers which enhanced the quality of this
manuscript signicantly.

1450008-14
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

Appendices: CU-PSHA Script

Appendix A: PSHA.m
clearall; clc; format longg;
% CU-PSHA Software
% PSHA Interface

% Step 1: Load study area (Require)

Ar1 = fopen('Input/Study-area.txt','r');
[Ar2,Arcount] = fscanf(Ar1,'%f',[2 inf]);
Arpoint = Arcount/2;
Ar = sortrows(Ar2');
fclose(Ar1);
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

% Step 2: Print out the hazard-curve file

PGA = 0.005:0.01:1.995;
Hc1 = fopen('Output/Hazard-curve.txt','wt');
fprintf(Hc1,'%f %f %f' ,[0,0,]);
by WSPC on 07/30/14. For personal use only.

fprintf(Hc1, '%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f\n',[PGA]);

% Step 3: Show status (no. of calculating point) of PSHA calculation

fori = 1:Arpoint
Pcal = i;
disp(['CU-PSHA'])
disp(['Calculating PSHA point no.' num2str(Pcal) ' from total ' num2str(Arpoint) ' points'])
disp([' '])

% Step 4: PSHA calculation

HZC = CalPSHA(Ar(i,1),Ar(i,2));
fprintf(Hc1,'%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f %f
%f %f %f %f %f %f %f %f %f %f\n',[HZC']);
end

% Step 5: Show status (finish) of PSHA calculation

fclose(Hc1);
disp([' '])
disp(['Complete !!! Probabilistic Seismic Hazard Analysis by CU-PSHA'])
disp([' '])

1450008-15
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

Appendix B: CalPSHA.m

functionPacc = CalPSHA(long,lat)
formatlongg;
% CU-PSHA Software
% Calculate Hazard Curve

% Step 1: Load earthquake source parameters(Require)

QP1 = fopen('Input/EQ-parameters.txt','r');
[QP2,QPcount] = fscanf(QP1,'%f',[10 inf]);
Fnum = QPcount/10;
QP = sortrows(QP2');
fclose(QP1);
HZZ = zeros(Fnum,200);
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

% Step 2: Determine Probability of magnitude and distance

for f = 1:Fnum
PM = ProbM(QP(f,1),QP(f,3),QP(f,4),QP(f,5),QP(f,6),QP(f,7),QP(f,8),
by WSPC on 07/30/14. For personal use only.

QP(f,9));
PR = ProbR(long,lat,QP(f,1));

% Step 3: Determine mean and standard deviation of Peaks Ground Acceleration

lnPHA = zeros(10,50);
SDPHA = zeros(10,50);
fori = 1:10
for j = 1:50
if QP(f,10) == 1
PGA = att1(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 2
PGA = att2(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 3
PGA = att3(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 4
PGA = att4(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 5
PGA = Att5(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 6
PGA = Att6(PM(i,1),PR(1,j),QP(f,2));
elseif QP(f,10) == 7
PGA = Att7(PM(i,1),PR(1,j),QP(f,2));
end
end
end
end
end
end
end
lnPHA(i,j) = PGA(1); % mean PGA in unit "ln(gal)" (1 gal=0.01 m/s2)
SDPHA(i,j) = PGA(2); % standard deviation
end
end

1450008-16
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

% Step 4: Evaluate probability of exceedance in each PGA level, M, and R

PHA = 0.005:0.01:1.995; % in g unit (1g = 9.81 m/s2)
for k = 1:200
lnacc(k) = log(PHA(k)*981); % Change g unit to be ln(gal)unit
fori = 1:10
for j = 1:50
probPHA(k,i,j) = 1-normcdf(lnacc(k),lnPHA(i,j),SDPHA(i,j));
end
end
end

% Step 5: Evaluate the average rate of threshold magnitude exceedance

delM2 = 0.5;
MC = QP(f,3)-delM2;
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

for k = 1:200
fori = 1:10
for j = 1:50
if PM(i,1) >= MC
by WSPC on 07/30/14. For personal use only.

lamda(k,i,j) = PM(11,1)*PM(i,2)*PR(2,j)*probPHA(k,i,j);
else
lamda(k,i,j) = PM(11,2)*PM(i,2)*PR(2,j)*probPHA(k,i,j);
end
end
lamdasum1(k,i) = sum(lamda(k,i,:));
end
lamdasum2(k) = sum(lamdasum1(k,:));
end
HZZ(f,:) = lamdasum2(1,:)';
end
HZS = sum (HZZ,1);

% Step 6: Sum the probability of exceedance from every EQ sources

Pat = zeros(1,202);
Pat(1,1) = long;
Pat(1,2) = lat;
fori = 1:200
Pat(1,i+2) = HZS(i);
end
Pacc = Pat;

1450008-17
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

Appendix C: ProbM.m
function PM = ProbM(Fnum,Mmax,Mmin,a,b,SR,Af,FMD)
formatlongg;
% CU-PSHA Software
% Calculate probability of earthquake magnitude

% Step 1: Determine the required earthquake parameters

c = 1.5; % Hanks and Kanamori(1979)
d = 16.1; % Hanks and Kanamori(1979)
Mshear = 3*(10^11); % Rigidity or shear modulus
alpha = 2.3026*a; % Gutenberg and Richter (1944)
beta = 2.3026*b; % Gutenberg and Richter (1944)
delM1 = 1; % Young and Coppersmith (1985)
delM2 = 0.5; % Young and Coppersmith (1985)
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

MC = Mmax-delM2; % Characteristic EQ
Momax = 10^((c*Mmax)+d); % Maximum seismic moment
fm = zeros(10,1);
by WSPC on 07/30/14. For personal use only.

% Step 2: Determine Probability Density Functions

subM = (Mmin+((Mmax-Mmin)/20):(Mmax-Mmin)/10:Mmax-((Mmax-
Mmin)/20))';

% 2.1. Exponential magnitude distribution model (Youngs and Coppersmith, 1985

if FMD == 1
fori = 1:10
fm(i) = (((beta*exp((-beta)*(subM(i)-0)))/(1-exp((-beta)*(Mmax-
0)))))*((Mmax-0)/10);
end
Mull1 = (Mshear*Af*SR*(c-b)*(1-exp((-beta)*(Mmax-
Mmin))))/(b*Momax*exp((-beta)*(Mmax-Mmin)));
Mull2 = exp(alpha-(beta*Mmin));
% 2.2. Characteristic earthquake model (Youngs and Coppersmith, 1985)
elseif FMD == 2
Cat = ((beta*exp((-1)*(Mmax-0-delM1-delM2)))/(1-exp((-
beta)*(Mmax-0-delM2))))*delM2;
fori = 1:10
ifsubM(i) < Mmin
fm(i) = 0;
elseifsubM(i) >= Mmin&subM(i)<=MC
fm(i) = (((beta*exp((-beta)*(subM(i)-0)))/(1-exp((-beta)*(Mmax-0-
delM2))))*(1/(1+Cat)))*((Mmax-0)/10);
elseifsubM(i) >= MC &subM(i)<=Mmax
fm(i) = (((beta*exp((-beta)*(Mmax-0-delM1-delM2)))/(1-exp((-
beta)*(Mmax-0-delM2))))*(1/(1+Cat)))*((Mmax-0)/10);
elseifsubM(i) > Mmax
fm(i) = 0;
end
end
end
end
end
K = ((b*10^((-1)*c*delM2))/(c-b))+((b*exp(beta*delM1)*(1-(10^((-
1)*c*delM2))))/c);

1450008-18
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

alphaNC = (Mshear*Af*SR*(1-exp((-beta)*(Mmax-Mmin-
delM2))))/(K*Momax*exp((-beta)*(Mmax-Mmin-delM2)));
Mull1 = (alphaNC*(beta*delM2*exp((-beta)*(Mmax-Mmin-delM1-
delM2))))/(1-exp((-beta)*(Mmax-Mmin-delM2))) ;
Mull2 = exp(alpha-(beta*Mmin));
end
end
PMM = zeros(11,2);
fori = 1:10
PMM(i,1) = subM(i);
if PMM(i,1) <= Mmin
PMM(i,2) = 0;
else
PMM(i,2) = fm(i);
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

end
end
PMM(11,1) = Mull1;
PMM(11,2) = Mull2;
by WSPC on 07/30/14. For personal use only.

PM = PMM;

% Remarks: Supplementary scripts for print out the probability of magnitude

% Pmag = PMM(1:10,:)';
% Magname = strcat('Output/Prob-magnitude (Fault no. ',num2str(Fnum),').txt');
% PD = fopen(Magname,'wt');
% fprintf(PD,'%s\t%s\t' ,['Magnitude',' Probability', ]);
% fprintf(PD,'\n%f %f',[Pmag]);
% fclose(PD);

1450008-19
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

Appendix D: ProbR.m

function PR = ProbR(long,lat,Fnum)
formatlongg;
% CU-PSHA Software
% Calculate probability of distance from earthquake source to site

% Step 1: Call EQ source

Qn1 = strcat('Input/',num2str(Fnum),'.txt');
Qn2 = fopen(Qn1,'r');
[Qn3,Qncount] = fscanf(Qn2,'%f',[3 inf]);
Qnpoint = Qncount/3;
Qn = sortrows(Qn3');
fclose(Qn2);
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

% Step 2: Determine probability of distance distribution

Rdist = zeros(Qnpoint);
fori = 1:Qnpoint
by WSPC on 07/30/14. For personal use only.

Rdist(i) = sqrt((((Qn(i,1)-long)*100)^2)+(((Qn(i,2)-lat)*100)^2)+
(Qn(i,3)^2));
end
distmax = max(Rdist);
distmin = min(Rdist);
subR = distmin+((distmax-distmin)/100):(distmax-distmin)/50:distmax-
((distmax-distmin)/100);
DistR = hist(Rdist,subR);
probdistR = DistR/sum(DistR);
ProbR = zeros(2,50);
ProbR(1,:) = subR;
ProbR(2,:) = probdistR;
PR = ProbR;

% Remarks: Supplementary scripts for print out the probability of distance

% Distname = strcat('Output/Prob-distance (Long ',num2str(long),' Lat
',num2str(lat),' Fault no. ',num2str(Fnum),').txt');
% PD = fopen(Distname,'wt');
% fprintf(PD,'%s\t%s\t' ,['Distance',' Probability', ]);
% fprintf(PD,'\n%f %f',[PR]);
% fclose(PD);

1450008-20
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

Appendix E: Attenuation Relationship (Example: att3.m)

function PGASD = att3(M,R,Ft)

% CU-PSHA Software
% Strong Ground-Motion Attenuation Relationship (PGA)
% Reference: Sadigh et al. (1993) &Sadigh et al. (1997)

% Step 1: Set Coefficient

c11 = -0.624;
c12 = 1;
c13 = 0;
c14 = -2.1;
c15 = 1.29649;
c16 = 0.25;
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

c17 = 0;
c21 = -1.274;
c22 = 1.1;
c23 = 0;
by WSPC on 07/30/14. For personal use only.

c24 = -2.1;
c25 = -0.48451;
c26 = 0.524;
c27 = 0;
h = 0;

% Step 2: Calculate mean PGA and Standard Deviation

R1 = sqrt(R^2+h^2);
if M <= 6.5
PGA = log(981*exp(c11+c12*M+c13*((8.5-M)^2.5)+c14*log(R1+
exp(c15+c16*M))+c17*log(R1+2)));
else if M >= 6.5
PGA = log(981*exp(c21+c22*M+c23*((8.5-M)^2.5)+c24*log(R1+
exp(c25+c26*M))+c27*log(R1+2)));
end
if M < 7.21
SD = 1.39-(0.14*M);
else
SD = 0.38;
end
PGASD = [PGA SD];

1450008-21
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

Appendix F: Shakemap.m
clearall; clc; format longg;
% CU-PSHA Software
% Evaluate the ground shaking of exceedance

% Step 1: Load necessary data (Require)

Pc = 2; % Percent of
exceedance
Yr = 50; % Year
Lamda = log10((-log(1-(Pc/100)))/Yr);

% Step 2: Call the hazard-curve file

HZC1 = fopen('Output/Hazard-curve.txt','r');
[HZC2,HZCcount] = fscanf(HZC1,'%f',[202 inf]);
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

HZCpoint = HZCcount/202;
HZC = HZC2';
fclose(HZC1);
Acc = 0.005:0.01:1.995;
by WSPC on 07/30/14. For personal use only.

% Step 3: Print out the ground shaking map file

Ske1 = strcat('Output/P',num2str(Pc),'Y',num2str(Yr),'.txt');
Ske2 = fopen(Ske1,'wt');
fprintf(Ske2,'%s\t%s\t%s\t\n' ,['longitude ','latitude ','PGA(g)']);

% Step 4: Calculate the ground shaking map

Rate = zeros(HZCpoint-1,200);
fori = 1:HZCpoint-1
for j = 1:200
if HZC(i+1,j+2) > 0
Rate(i,j) = log10(HZC(i+1,j+2));
else
Rate(i,j) = -19.3920890976347;
end
end
for k = 2:200
ifLamda > Rate(i,k) &Lamda< Rate(i,k-1)
PGA(i) = interp1(Rate(i,k-1:k),Acc(k-1:k),Lamda,'spline');
elseifLamda == Rate(i,k)
PGA(i) = Acc(k);
elseifLamda == Rate(i,1)
PGA(i) = 0.005;
elseifLamda > Rate(i,1)
PGA(i) = 0;
elseifLamda < Rate(i,200)
PGA(i) = 2.000;
end
end
end
end
end
end
fprintf(Ske2,'\n%f %f %f',[HZC(i+1,1);HZC(i+1,2);PGA(i)]);
end

1450008-22
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

fclose(Ske2);

% Step 5: Show status (finish) of ground shaking map calculation

disp([' '])
disp([' Finish calculation of ground shaking'])
disp([' ' num2str(Pc) ' % probability of exceedance in ' num2str(Yr) ' year'])
disp([' '])

Appendix G: Percentmap.m
clearall; clc; format longg;
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

% CU-PSHA Software
% Evaluate percent of occurrence of ground shaking (MMI) of interest

% Step 1: Load necessary data (Require)

by WSPC on 07/30/14. For personal use only.

MMI = 7; % Percent of exceedance

Yr = 50; % Year
MG = 13;
% Remarks for MG is to determine MMI vs PGA(unit g) relationship
% 1. Cancani (1904) Log(PGA) = 0.33MMI-1.17
% 2. Ishimoto (1932) Log(PGA) = 0.4228MMI-1.4178
% 3. Kawasumi (1951) Log(PGA) = 0.3409MMI-0.5705
% 4. Neumann (1954) Log(PGA) = 0.3041MMI-0.0096
% 5. Savarensky and Kirnos (1955) Log(PGA) = 0.2773MMI-0.5915
% 6. Hershberger (1956) Log(PGA) = 0.4329MMI-0.921
% 7. Richter (1958) Log(PGA) = 0.3297MMI-0.499
% 8. Medvedev and Sponheuer (1969) Log(PGA) = 0.3019MMI-0.2394
% 9. Okamoto (1973) Log(PGA) = 0.32MMI-0.3985
% 10. Trifunac and Brady (1975)-Vert. Log(PGA) = 0.304MMI-0.2115
% 11. Trifunac and Brady (1975)-Horiz. Log(PGA) = 0.2949MMI+0.046
% 12. Theodulides and Papazachos (1992) Log(PGA) = 0.291MMI+0.304
% 13. Shabestari and Yamazaki (2001) Log(PGA) = 0.2545MMI+0.2977

% Convert from m/sec2 to g

if MG == 1
PGA = (10^((0.33*MMI)-1.17))/981;
elseif MG == 2
PGA = (10^((0.4228*MMI)-1.4178))/981;
elseif MG == 3
PGA = (10^((0.3409*MMI)-0.5705))/981;
elseif MG == 4
PGA = (10^((0.3041*MMI)-0.0096))/981;
elseif MG == 5
PGA = (10^((0.2773*MMI)-0.5915))/981;
elseif MG == 6
PGA = (10^((0.4329*MMI)-0.921))/981;
elseif MG == 7
PGA = (10^((0.3297*MMI)-0.499))/981;
elseif MG == 8

1450008-23
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

PGA = (10^((0.3019*MMI)-0.2394))/981;
elseif MG == 9
PGA = (10^((0.32*MMI)-0.3985))/981;
elseif MG == 10
PGA = (10^((0.304*MMI)-0.2115))/981;
elseif MG == 11
PGA = (10^((0.29493*MMI)+0.046))/981;
elseif MG == 12
PGA = (10^((0.291*MMI)+0.304))/981;
elseif MG == 13
PGA = (10^((0.2545*MMI)+0.2977))/981;
end
end
end
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

end
end
end
end
by WSPC on 07/30/14. For personal use only.

end
end
end
end
end
end

% Step 2: Call the hazard-curve file

HZC1 = fopen('Output/Hazard-curve.txt','r');
[HZC2,HZCcount] = fscanf(HZC1,'%f',[202 inf]);
HZCpoint = HZCcount/202;
HZC = HZC2';
fclose(HZC1);
Acc = 0.005:0.01:1.995; % in g unit (1g = 9.81 m/s2)

% Step 3: Print out the percent map file

Ske1 = strcat('Output/MMI',num2str(MMI),'Y', num2str(Yr),'.txt');
Ske2 = fopen(Ske1,'wt');
fprintf(Ske2,'%s\t%s\t%s\t\n' ,['longitude ','latitude ','Probability(%)']);

% Step 4: Calculate the percent map

Rate1 = zeros(HZCpoint-1,200);
fori = 1:HZCpoint-1
for j = 1:200
Rate1(i,j) = HZC(i+1,j+2);
end
end
fori = 1:HZCpoint-1
Rate = Rate1(i,:);
for k = 2:200
if PGA < Acc(k) & PGA>Acc(k-1)
MM(i) = interp1(Acc(k-1:k),Rate(k-1:k),PGA,'spline');
elseif PGA == Acc(k)
MM(i) = Rate(k);
elseif PGA == Acc(1)

1450008-24
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

CU-PSHA: A MATLAB Software for Probabilistic Seismic Hazard Analysis

MM(i) = Rate(1);
elseif PGA > Acc(200)
MM(i) = 0;
elseif PGA < Acc(1)
MM(i) = 1;
end
end
end
end
end
end
MMGG(i) = (1-exp((-1)*MM(i)*Yr))*100;
fprintf(Ske2,'\n%f %f %f',[HZC(i+1,1);HZC(i+1,2);MMGG(i)]);
end
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

fclose(Ske2);

% Step 5: Show status (finish) of percent map calculation

disp([' '])
by WSPC on 07/30/14. For personal use only.

disp([' Finish calculation of percent of groundshaking (MMI) occurence'])

disp([' Percent of ground shaking equal or greater than MMI ' num2str(MMI) ' in ' num2str(Yr) '
year'])
disp([' '])

References
Boore, D. M., Joyner, W. B. and Fumal, T. E. [1997] Equations for estimating horizontal
response spectra and peak acceleration from Western North American earthquakes:
A summary of recent work, Seismol. Res. Lett. 68, 128153.
Chintanapakdee, C., Naguit, M. E. and Charoenyuth, M. [2008] Suitable attenuation
model for Thailand, The 14th World Conf. Earthquake Engineering, 1217 October,
Beijing, China, 2008, pp. 18.
Convertito, V., Emolo, A. and Zollo, A. [2006] Seismic-hazard assessment for a character-
istic earthquake scenario: An integrated probabilistic-deterministic method, Bullet.
Seismol. Soc. Amer. 96(2), 377391.
Cornell, C. A. [1968] Engineering seismic risk analysis, Bullet. Seismol. Soc. Amer. 58,
15831606.
Crouse, C. B. [1991] Ground-motion attenuation equations for earthquakes on the Cas-
cadia subduction zones, Earthq. Spectra 7(2), 201236.
Delfebriyadi [2011] Seismic hazard assessment of Padang city using probabilistic method,
Asian Trans. Eng. (ATE ISSN : 2221-4267) 01(05), 5662.
Esteva, L. and Villaverde, R. [1973] Seismic risk, design spectra and structural reliability,
Proc. 5th World Conf. Earthquake Engineering, Rome, Italy, pp. 25862597.
Gutenberg, B. and Richter, C. F. [1944] Frequency of earthquakes in California, Bullet.
Seismol. Soc. Amer. 34, 185188.
Hanks, T. C. and Kanamori, H. [1979] A moment-magnitude scale, J. Geophys. Res.
84, 23482350.
Idriss, I. M. [1993] Procedures for selecting earthquake ground motions at rock sites,
Report-NIST GCR 93-625, U.S. Department of Commerce, National Institute of Stan-
dards and Technology, Gaithersburg, Maryland, U.S.A.
Irsyam, M., Dangkua, D. T., Hoedajanto, D., Hutapea, B. M., Kertapati, E. K., Boen, T.
and Petersen, M. D. [2008] Proposed seismic hazard maps of Sumatra and Java

1450008-25
 2nd Reading
June 30, 2014 8:33 WSPC/S1793-4311/238-JET 1450008

S. Pailoplee & C. Palasri

islands and microzonation study of Jakarta city, Indonesia, J. Earth Syst. Sci.
117(S2), 865878.
Kobayashi, S., Takahashi, T., Matsuzaki, S., Mori, M., Fukushima, Y., Zhao, J. X. and
Somerville, P. G. [2000] A spectral attenuation model for Japan using digital strong
motion records of JMA87 type, Proc. 12th World Conf. Earthquake Engineering,
Auckland, New Zealand, pp. 146150.
Kramer, S. L. [1996] Geotechnical Earthquake Engineering (Prentice Hall, Inc., Upper
Saddle River, New Jersey).
Krinitzsky, E. L. [2003] How to combine deterministic and probabilistic methods for
assessing earthquake hazards, Eng. Geol. 70, 157163.
Megawati, K. and Pan, T.-C. [2009] Regional seismic hazard posed by the Mentawai
segment of the Sumatran megathrust, Bullet. Seismol. Soc. Amer. 99(2A), 566584.
Okamoto, S. [1973] Introduction to Earthquake Engineering (University of Tokyo Press,
J. Earthquake and Tsunami Downloaded from www.worldscientific.com

Tokyo).
Petersen, M., Dewey, J., Hartzell, S., Mueller, C., Harmsen, S., Frankel, A. D. and Ruk-
stales, K. [2004] Probabilistic seismic hazard analysis for Sumatra, Indonesia and
across the Southern Malaysian Peninsula, Tectonophys. 390, 141158.
by WSPC on 07/30/14. For personal use only.

Petersen, M. D., Harmsen, S., Mueller, C., Haller, K., Dewey, J., Luco, N., Crone, A.,
Rukstales, K. and Lidke, D. [2008] New USGS Southeast Asia seismic hazard maps,
The 14th World Conf. Earthquake Engineering, 1217 October 2008, Beijing, China.
Putra, R. R., Kiyono, J., Ono, Y. and Parajuli, H. R. [2012] Seismic hazard analysis for
Indonesia, J. Nat. Disas. Sci. 33(2), 5970.
Sadigh, K., Chang, C.-Y., Egan, J. A., Makdisi, F. and Youngs, R. R. [1997] Attenuation
relationships for shallow crustal earthquakes based on California strong motion data,
Seismol. Res. Lett. 68, 180189.
Shabestari, K. T. and Yamazaki, F. [2001] A proposal of instrumental seismic intensity
scale compatible with MMI evaluated from three-component acceleration records,
Earthquake 17(4), 711723.
Sieh, K., Ward, S. N., Natawidjaja, D. and Suwargadi, B. W. [1999] Crustal deformation
at the Sumatra subduction zone revealed by coral data, Geophys. Res. Lett. 26(20),
31413144.
Wells, D. L. and Coppersmith, K. J. [1994] Updated empirical relationships among mag-
nitude, rupture length, rupture area, and surface displacement, Bullet. Seismol. Soc.
Amer. 84, 9741002.
Youngs, R. R., Chiou, S. J. and Silva, W. J. [1997] Humphrey strong ground motion
attenuation relationships for subduction zone earthquakes, Seismol. Res. Lett. 68,
5873.
Youngs, R. R. and Coppersmith, K. J. [1985] Implications of fault slip rates and earth-
quake recurrence models to probabilistic seismic hazard estimates, Bullet. Seismol.
Soc. Amer. 75, 939964.

1450008-26