Ground Motion

1
 Strong Ground Motion
Evaluation of the effects of earthquakes requires the
study of ground motion
Engineering Seismology deals with vibrations related
to earthquakes, which are strong enough to cause
damage to people and environment

Time (sec)

2
 Components of Ground Motion

The ground motions

produced by earthquakes at
any particular point have 3
translational and 3 rotational
components.
In practice, generally
translational components of
ground motion are measured
and the rotational
components are ignored.

3
 Measuring ground motion
Seismographs are used to measure the
ground motion. Seismogram is the
output from the seismograph. It is a
record of the ground motion at a
measuring station as a function of time.
Seismograms typically record motions
in three cartesian axes (x, y, and z),
with the z axis perpendicular to the
Earth's surface and the x- and y- axes
parallel to the surface. Three inertial
seismometers are commonly used in
one instrument housing to measure
up-down, east-west, north-south
motions simultaneously. 4
Seismogram interpretation
Seismograms can provide
information on
location of epicenter
magnitude of
earthquake
source properties
Most seismograms record P, S
& surface waves
First arrival is P wave - surface waves follow and may
After a pause of several continue for tens of seconds
seconds/10s of seconds - surface waves are slower but
persist to greater distances than
the higher amplitude S
P & S waves
wave arrives
5
Wave terminology
Wave amplitude

displacement, y
height of a wave above its
zero position T= 1/f
Wave period
time taken to complete one A
cycle of motion time, t
Frequency
number of cycles per second
(Hertz)
— human ear can detect
frequencies in range
15-20000 Hz A= amplitude
— felt shaking during
quake has frequencies T= Period
from 20 down to 1
Hertz f = frequency

6
 Ground Motion Recording

Seismometers are instruments that measure motions of the ground.

Early seismometers used optical levers or mechanical linkages to amplify
the small motions involved, recording on photographic paper.
Modern instruments use electronic sensors, amplifiers, and recording
devices. Most are broadband covering a wide range of frequencies. These
seismometers can measure motions with frequencies from 500 Hz to
1/850 Hz.
Seismometers spaced in an array can also be used to precisely locate, in
three dimensions, the source of an earthquake, using the time it takes for
seismic waves to propagate away from the hypocenter, the initiating
point of fault rupture

7
Vertical Benioff Seismometer

•Consists of Black aluminium

cylinder cover: Diameter 26 cm;
Height 49.5 cm.
•The outside case is latched to
prevent air currents .
•Internal mass is supported on a
central spring.
•Works on the principle of magnet-
coil relative motion to generate
electric current used to displace a
galvanometer (ink or light source)
writing to a drum recorder for
earthquake recording (in the
traditional way) or digitized to a file
today.
8
 Ground Motion Recording
The actual ground motion at a given
location is derived from instrumentally
recorded motions. The most commonly
used instruments for engineering
purposes are strong motion
accelerographs/ accelerometers. These
instruments record the acceleration time
history of ground motion at a site, called
an accelerogram.
Digital strong motion accelerograph

By proper analysis of a recorded accelerogram to account for instrument

distortion and base line correction, the resulting corrected acceleration record can
be used by engineers to obtain ground velocity and ground displacement by
appropriate integration.
9
Types of Accelerometers
Types of Accelerometers:

Electronic : transducers produce voltage output

Servo controlled: use suspended mass with
displacement transducer
Piezoelectric: Mass attached to a piezoelectric
material, which develops electric charge on surface.

Generally accelerometers are placed in three orthogonal

directions to measure accelerations in three directions at any
time. Sometimes geophones (velocity transducers) are
attached to accelerometers to measure the seismic wave
velocities. 10
Servo-controlled Accelerometer

Principle: An acceleration a will cause the mass to

be displaced by ma/k
or alternatively, if we observe a displacement of x,
we know that the mass has undergone an
acceleration of kx/m.

11
Piezoelectric Accelerometer

Principle: piezoelectric accelerometers convert one form of energy into

another and provide an electrical signal in response to a quantity that is
being measured. Acceleration acts upon a seismic mass that is restrained by a
spring or suspended on a cantilever beam, and converts a physical force into
an electrical signal. Before the acceleration can be converted into an
electrical quantity it must first be converted into either a force or
displacement. This conversion is done via the mass spring system shown in
the figure. 12
Electronic Accelerometer
Three-direction electronic accelerometer uses the latest
integrated accelerometer chip. It features small volume,
low power consumption, zero fine toning, continuous,
stable and reliable operation

13
 Geophone
Geophones (velocity transducers) are used to measure the
seismic wave velocities. They transform the mechanical wave
energy to electrical voltage and the response is captured with
time to obtain the velocity of seismic waves

14
 Geophone
Traditional geophones consist of a magnetized mass fixed to
the container and a surrounding coil suspended by springs.
Natural frequency is controlled by the mass and spring
constant.
Response curve is the plot of the electric voltage output from
a geophone as it changes with different wave frequencies.

15
 Array of Geophones

Vertical component geophones

Take-out points where a

geophone can be connected to
the pair of conductors

An array of geophones connected to conductors are used for

seismic surveying 16
Micro Electronic Mechanical Systems
(MEMS)
Micro Electronic Mechanical Systems (MEMS) are a recently
developed device providing Broad-Band Sensing. MEMS are
micro fabricated non-laser passive components. These are about
1 to 10 microns in size, which is smaller than the width of a human
hair. Most of the traditional moving coil type geophones are now
being replaced by MEMS.
Velocity Sensitive

Coil

MEMS

Acceleration sensitive 17
Hydrophones
Instead of recording displacement, hydrophones record changes
in pressure. A piezo-electric crystal senses the pressure change
and creates a voltage. This sensor is used for ocean floor seismic
sensing.

18
Broadband Seismometer
Broadband seismometers can detect motion over a wide range
(or band) of frequencies and usually over a large range of
amplitudes (the dynamic range). Broadband sensors respond to
most frequencies from 0.01 Hz to 50 Hz. For regional seismology,
the frequency range of interest is from 0.05 to 20 Hz therefore;
broadband sensors are most useful for recording regional
earthquakes .

19
Strong motion seismographs

The Global Seismographic Network (GSN) is a 150+ station, globally distributed,

state-of-the-art digital seismic network providing free, realtime, open access. The
map above shows the distribution of the current 150 station network with respect
to network operations.
20
 National Network (RSNC)

21
http://200.119.88.135/RSNC/index.php/joomla-overview/red-sismologica-nacional-de-colombia
 Recommended Reading
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.

 http://www.iris.edu/stations/seisWorkshop04/iris_sensor_ws_9.19.05.pdf

 Havskov, J. (2006) Instrumentation in earthquake seismology, Springer

 John M Reynolds (2011). An Introduction to Applied and Environmental

Geophysics. Wiley Blackwell.

 http://en.wikipedia.org/wiki/Accelerometer

 Norton, Harry N.(1989). Handbook of Transducers. Prentice Hall.

 Patrick, Walter L. (2006)The History of the Accelerometer 1920’s-1996

Prologue and Epilogue, Available at
http://www.sandv.com/downloads/0701walt.pdf

22
 Ground Motion Parameters
An earthquake history can be described using Amplitude, Frequency
content, and Duration.
Amplitude: The most common measures of amplitude are
PGA: Peak ground acceleration (Horizontal- PHA & Vertical- PVA)
EPA: Effective peak acceleration
PGV: Peak ground velocity (PHV & PVV)
EPV: Effective peak velocity
PGD: Peak ground displacement

Frequency Content: The frequency content of an earthquake history is often

described using Fourier Spectra, Power spectra and response spectra.

Duration: The duration of an earthquake history is somewhat dependent on

the magnitude of the earthquake.

23
 Ground Motion Parameters

PGA: The absolute value of the largest horizontal

acceleration at a location.

It is the most important strong motion parameter. In a

stiff structure, the maximum value of the induced
force is equal to PGA  Mass

24
 Measurement of ground acceleration
A seismograph can be illustrated by a mass-spring-dashpot
single degree of freedom system.

The response of such system for shaking is given by

 2u u
m 2  c  k u  m ug  0
t t
Where u is the trace displacement (relative displacement
between seismograph and ground), ug is the ground
displacement, c is the damping coefficient, k is the stiffness
coefficient. 25
 Measurement of ground acceleration

If the ground displacement is simple harmonic at a circular

frequency g , the ground acceleration amplitude is
calculated from the trace displacement amplitude using
the equation: u 1

 2u g  2
0 1   
2 2
 4 2  2
t 2

Where 0 is the undamped natural circular frequency

 is tuning ratio, given by g/ 0

c
 Is damping ratio, given by
2 km
26
Amplitude Parameters

Source: Kramer (1996)

From the time histories of acceleration, velocity and
displacement are obtained by integrating the acceleration
records. All other amplitude parameters are calculated from
these time histories. 27
Amplitude Parameters

Source: Kramer (1996)

28
 PGA = 0.106 g

PGA = 0.08 g

PGA = 0.07 g

Strong motion accelerograms recorded on the ground floor of the Passport Office Building at
29
Ahmedabad during 2001 Bhuj earthquake.
 Amplitude Parameters

Effective Acceleration: The acceleration which is effective

in causing structural damage. This depends on size of
loaded area, weight, damping and stiffness properties of
structure and its location with respect to epicenter.

Sustained Maximum Acceleration: The absolute values of

highest accelerations that sustained for 3 and 5 cycles in
acceleration time history are defined as 3-cycle sustained
and 5-cycle sustained accelerations respectively.

30
 Design ground motion parameters
Complete reliance on peak ground acceleration for
design proved to give wrong results.
This is because, more often the peak acceleration
corresponds to very high frequencies which are out of the
range of the natural frequencies of most structures.
Therefore, large values of peak ground acceleration alone can
seldom initiate either resonance in the elastic range or be
responsible for large scale damage in the inelastic range.
Therefore, different parameters are required to characterize
the severity and the damage potential of the earthquake
ground motion. In general, these parameters are frequency
and duration of strong ground motion
31
Frequency Content Parameters
The frequency content
describes clearly how the
amplitude of ground motion is
distributed among different
frequencies. The frequency
content of a ground motion
can be obtained by
transforming the ground
motion from time domain to a
frequency domain through a Source: Kramer (1996)
Fourier transform.

32
Frequency Content Parameters
The frequency content of an earthquake history is often
described using Fourier Spectra, Power spectra and
response spectra.
Fourier Spectra
A periodic function (for which an earthquake history is an
approximation) can be written as


x(t )  c0   cn sin( n t   n )
n 1

where cn and n are the amplitude and phase angle

respectively of the nth harmonic in the Fourier series.
33
 Fourier Spectra
A spectrum is, first of all, a function of
frequency.
For our purposes, it is determined from a
single time series, such as a record of the
ground motion.
The spectrum in general shows some
frequency-dependent characteristic of the
ground motion.

34
 Fourier Spectra
The Fourier amplitude spectrum is a plot of cn versus n
Shows how the amplitude of the motion varies with
frequency.
Expresses the frequency content of a motion
The Fourier phase spectrum is a plot of n versus n
 Phase angles control the times at which the peaks of
harmonic motion occur.
Fourier phase spectrum is influenced by the variation of
ground motion with time.

35
Fourier Spectra

36
 Fourier Amplitude Spectrum

Source: Kramer (1996)

Fourier amplitude spectra for Gilroy No. 1 and Gilroy No.

2 strong motion records in Gilroy, California during 1989
Loma Preita Earthquake
37
 Fourier Amplitude Spectrum
The Fourier amplitude spectra of actual
earthquakes are often plotted on

Fourier Amplitude (log)

logarithmic scales, so that their
characteristic shapes can be clearly
distinguished from the smoothed
curves.
Two frequencies that mark the range of
frequencies for largest Fourier fc fmax
acceleration amplitude are corner Frequency (log)
frequency (fc) and cutoff frequency
(fmax)

fc is a very important parameter because it is inversely proportional

to the cube root of seismic moment, thus indicating that large
earthquakes produce greater low-frequency motions. 38
Frequency Content Parameters
Power Spectra
The power spectrum is a plot of G() versus  n . The power
spectrum density (PSD) function is defined by the following
equation and is closely related to the Fourier amplitude
spectrum:
1 2
G ( )  cn
π Td

where G( ) is the PSD, Td is the duration of the ground

motion, and cn is the amplitude of the nth harmonic in the
Fourier series. PSD function is used to characterize an
earthquake history as a random process.
39
 Response Spectra
A response spectrum is used to provide
the most descriptive representation of
the influence of a given earthquake on a
structure or machine.
Response spectra are widely used in
earthquake engineering.
A response spectrum is a plot of the
maximum response amplitude
(displacement, velocity or acceleration)
versus time period of a system to a
given component of ground motion.
Using the response spectrum, peak
response of buildings to earthquakes can
be assessed and their natural frequency
40
can be determined.
 Response Spectra
The response spectrum
describes the maximum
response of a structure or a
machine to a particular input
motion as a function of
frequency and damping ratio.
The seismic response spectra
from two sites (one rock and
the other soil) for the same
earthquake motion are
shown. From the figure, it is
very clear that the response
of soil and rock is quite
different for the same Source: Kramer (1996)
vibration.
Response spectra with 5% damping
41
 Predominant Period

Fourier Amplitude
Predominant Period (Tp): Period of GM1
vibration corresponding to the maximum GM2
value of the Fourier amplitude spectrum.
This parameter represents the frequency
content of the motion.
Tp
The predominant period for two Period
different ground motions (GM1 & GM2) Tp is same for the two
with different frequency contents can be ground motions, though
the frequency content is
same, making the estimation of
different
frequency content crude.

42
 Predominant Period

Source: Kramer (1996)

Variation Of Predominant Period At Rock Outcrops With

Magnitude And Distance

43
 Duration
Duration of an earthquake is a very important parameter that
influences the amount of damage due to an earthquake. A
strong motion of very high amplitude of short duration may
not cause as much damage to a structure as a motion with
moderate amplitude and long duration. This is because ground
motions with long durations cause more load reversals, which
is important in the degradation of stiffness of the structures
and in building up pore pressures in loose saturated soils.

Relative duration does not depend on the peak values. It is

the time interval between the points at which 0.05% and
0.95% of the total energy has been recorded.

44
 Duration
Source: Kramer (1996)
The duration of an
earthquake history is
somewhat dependent
on the magnitude of
the earthquake. Figure
below shows
accelerograms from six
earthquakes of the
Pacific coast of Mexico.
The epicentral distance
was the same for all six
earthquakes.

45
 Duration
Bracketed duration is the measure of time between the first
and last exceedence of a threshold acceleration 0.05 g.

46
Source: Kramer (1996)
 Bracketed Duration

Source: Kramer (1996)

Variation of Bracketed Duration (0.05 G Threshold) With Magnitude

And Epicentral Distance: (A) Rock Sites; (B) Soil Sites

47
 Recommended Reading

 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.

 Stein, S. and Wysession, M. (2003). An introduction to seismology,

earthquakes, and earth structure, Blackwell Science.

 http://webshaker.ucsd.edu/homework/Ground_motion_parameters.pdf

 Download free software for the estimation of ground motion parameters :

http://www.seismosoft.com/SeismoSignal-2016-Release-1

48
 Other Spectral Parameters
RMS acceleration : This is the parameter that includes the
effects of amplitude and frequency, defined as

Where a(t) is the acceleration over the time domain and Td

is the duration of strong motion
AI - The Arias Intensity is a measure of the total energy at
the recording station and is proportional to the sum of the
squared acceleration. It is defined as

49
 Other Spectral Parameters
SI - The Spectrum Intensity is defined as the integral of the
psuedo-spectral velocity curve (also known as the velocity
response spectrum), integrated between periods of 0.1 - 2.5
seconds. These quantities are motivated by the need to
examine the response of structures to ground motion, as
many structures have fundamental periods between 0.1 and
2.5 sec. The SI can be calculated for any structural damping
ratio.
Dominant frequency of ground motion (Fd) is defined as the
frequency corresponding to the peak value in the amplitude
spectrum. Thus, Fd indicates the frequency for which the
ground motion has the most energy. The amplitude
spectrum has to be smoothed before determining Fd.
50
 Other Spectral Parameters
Bandwidth BW - of the dominant frequency; measured
where the amplitude falls to 0.707 (1 /2) of the amplitude
of the dominant frequency. Again, this is based on a
smoothed amplitude spectrum.
Fourier Amplitude

Band

Period

51
Spatial variability of ground motions
The ground motion parameters at any site depend upon the
magnitude of earthquake and the distance of the site from
epicenter.
The ground motion parameters measured at a site have been
used to develop empirical relationships to predict the
parameters as functions of earthquake magnitude and source-
to-site distance. But these predictions are not accurate.
For structures that extend over considerable distance (such as
bridges and pipelines), the ground motion parameters will be
different at different part of the structure, causing differential
movement of the supports. Local variation of ground motion
parameters need to be considered for the design of such
structures.
52
 Ground Motion Parameters
One of the most important decisions in carrying out proper
design is to select a design earthquake that adequately
represents the ground motion expected at a particular site and in
particular the motion that would drive the structure to its critical
response, resulting in the highest damage potential.
The quantification of such ground motion is not easy. It requires
a good understanding of the ground motion parameters that
characterize the severity and the damage potential of the
earthquake ground motion and the seismological, geological, and
topographic factors that affect them.

53
 Why is PHA important?

PHA is the most important strong motion

parameter.

In a stiff structure, the maximum value of

the induced force is equal to PHA  Mass

Earthquake force is directly related to the

magnitude and intensity.

54
 Why vertical accelerations are not very
important?
Structures have gravity acting against vertical
accelerations due to earthquake. Static forced
induced due to gravity provide adequate resistance
to dynamic forces induced due to vertical
accelerations during earthquake.
PVA  2/3 PHA
PVA > 2/3 PHA near epicenter
PVA < 2/3 PHA at distances far from epicenter

55
 Why high peak accelerations
cause little damage?

Many times, very high peak

accelerations occur at high
frequencies If these frequencies are
higher than the natural frequency
of the structure, then the damage
to the structure is little.

56
 Why can’t peak amplitudes quantify
destruction?

Although peak amplitude is a very useful parameter, it

provides no information on the frequency content or
duration. Sometimes acceleration time histories are
characterized by single cycle peak amplitudes, that are
much greater than amplitudes of other cycles

time time

Ground motion 1 Ground motion 2

57
Why can’t peak amplitudes quantify
destruction?

58
 Peak acceleration and effective
acceleration
Peak accelerations correspond to the highest acceleration
value recorded during earthquake at a specific location.
Effective acceleration is the acceleration which is most
effective in causing damage to structure.
Peak acceleration is independent of structural response.
Effective acceleration is based on structural response
In a free field, effective and peak accelerations are equal
Near the structure, effective acceleration is generally less
than the peak acceleration.
59
 Effective
acceleration

Effective acceleration is a function of

size of loaded area, frequency content,
weight, embedment, damping
characteristics of the structure and its
foundation

60
 PHA and Earthquake
Intensity
Many researchers tried to correlate PHA to
earthquake intensity and magnitude.
Though, this procedure is not accurate.
Many times, these relations are useful to
obtain PHA from Mercalli’s intensity,
especially for earthquakes which have
occurred prior to strong motion instruments
were available.
61
 PHA and PHV
Peak horizontal acceleration (PHA) and peak horizontal
velocity (PHV) do not occur at same frequencies. Peak
accelerations and peak velocities are associated with higher
and intermediate frequencies, respectively.

Thus, PHA ( ) and PHV ( ) can be related to the

frequency content of the motion. For a simple harmonic
motion (SHM) with period T:

62
 PHA and PHV
Recall that T is the period of vibration of an equivalent
harmonic wave.
This period provides information on what frequencies are
more significant dominant in a given ground motion. Thus,
this is called the predominant period.

Seed and Idriss (1982)

proposed different values of
𝑣 ⁄𝑎 for different sites
within 50km of the source

63
Estimation of Ground Motion Parameters

Ground motion parameters are usually

estimated through predictive
relationships. These relationships
express ground motion parameter in
terms of the quantities that affect it
strongly.
Predictive relationships play important
role in seismic hazard analysis.
64
Estimation of Ground Motion Parameters
The functional form of the predictive relationship is usually
selected to reflect the mechanism of the ground motion
process as closely as possible
This minimizes the number of empirical coefficients and
allows to apply the relationship with greater confidence to
ground conditions that are poorly represented in the
database.

65
Estimation of Ground Motion Parameters
Predictive relationships are often arrived at by
regression analysis of the available strong motion
data
These relationships have to be updated time to
time after major earthquakes in the region.

66
Estimation of Ground Motion Parameters
Common forms of predictive relations are based on the following
observations:
1. Peak values of strong motion parameters are usually lognormally
distributed.
2. Earthquake magnitude M is typically defined as the logarithm of some peak
ground motion parameter Y. Thus M proportional to ln Y
3. The spreading of stress waves as they travel away from the earthquake
source attenuates the body wave and surface wave amplitudes
4. The area over which the fault rupture occurs, increases with increase in the
earthquake magnitude. Thus the effective distance R increases with the
earthquake magnitude
5. The material damping decreases the ground motion amplitudes with
distance exponentially
6. Ground motion parameters may be influenced by source characteristics like
type of fault or the site characteristics like the type of soil and topography
of the region. 67
 Typical Predictive relationship
Campbell (1981) used worldwide data to obtain a relationship for the
mean PHA for sites within 50 km of fault rupture in magnitude 5.0 to 7.7
earthquakes as:

Note 1 gal = 1 cm/sec2 then, 1g = 981 gal

Where M is the local magnitude for magnitude less than 6 or surface

wave magnitude for magnitude less than 7 and R is the closest distance
to the fault rupture in kilometers.
Several other predictive relationships are available in literature, which
are developed for different regions
68
Estimation of Ground Motion Parameters

69
 Predictive (Attenuation) Relationships
Ground Motion Parameter

Distance from the source

70
 Predictive (Attenuation) Relationships
Ground Motion Parameter

Distance from the source

71
 Attenuation Relationships

Modern attenuation relationships take into account many parameters,

which increases their complexity:

1) Basin Effects
2) Site Response Effects
3) Fault Type
4) Fault geometry
5) Hanging Wall/ Foot Wall
6) Main and after shock effects

72
 Example Problem
The strong motion data of an earthquake recorded for 3 seconds near epicenter with a
sampling interval of 0.1 sec is given in the table below. Plot acceleration-time history of the
earthquake. Find out PHA, PVA, vmax, maximum displacement, bracketed duration,
predominant period, sustained maximum acceleration (peaks need not be successive)
and local magnitude of the earthquake [ML = log10 A (A in micrometers)]. Comment on the
intensity of this earthquake and expected hazards.

No. Acceleration Velocity Displacement No. Acceleration Velocity Displacement

(g) (cm/sec) (cm) (g) (cm/sec) (cm)
1 -0.01 0.00 0.00 17 -0.26 -8.60 -18.52
2 -0.02 -15.59 -0.05 18 -0.01 -14.63 -18.23
3 0.01 -19.08 -0.21 19 0.03 -16.49 -17.49
4 -0.05 -21.44 -0.52 20 -0.20 -31.36 -16.79
5 0.03 -22.48 -0.99 21 -0.16 -48.31 -17.20
6 0.02 -23.57 -1.63 22 0.49 -27.86 -18.59
7 -0.01 -19.73 -2.44 23 -0.33 -3.09 -19.61
8 0.18 -9.83 -3.42 24 -0.02 -4.11 -20.08
9 0.08 -0.01 -4.65 25 0.12 -16.42 -20.52
10 0.02 3.48 -6.14 26 0.12 -17.17 -21.97
11 -0.02 10.43 -7.90 27 -0.11 -12.13 -23.64
12 -0.14 1.63 -9.84 28 -0.06 -5.24 -25.83
13 -0.16 -10.44 -12.11 29 0.01 -0.12 -29.81
14 0.04 -12.78 -14.37 30 0.01 1.25 -34.22
15 0.10 -5.41 -16.48 31 -0.04 4.66 -35.04
16 0.01 -6.36 -18.14

73
 Solution
0.60
PHA = 0.49 g
PVA = 2/3 PHA = 0.327 g
0.50 Bracketed duration = 2.7-0.3 = 2.4 sec
vmax = 48.31cm/sec
predominant period = 2 p vmax/amax = 0.63 sec
0.40 (amax = 0.49 g = 480.69 cm/sec2)
sustained max. acceleration ( 3cycle) = 0.2 g (3rd
highest)
0.30 maximum displacement = 35.04 cm

0.20

0.10
0.05 g

0.00
0 0.5 1 1.5 2 2.5 3 0.05 g 3.5

-0.10
2nd exceedence of
1st exceedence of
-0.20

-0.30 2

1 cycle
-0.40
3

74
 Exercise Problems
1. If an earthquake is recorded at four equally spaced locations along a
straight line, show the variation of all the following parameters along
the straight line if the epicenter lies on the straight line, dividing it into
two equal parts
(a) Magnitude (b) PHA (c) Duration (d) Predominant period
2. Obtain the time-acceleration response for any earthquake.
Determine the bracketed duration, predominant period and Arias
intensity
3. Download the software Seismosignal, which is a freeware. Analyze
any earthquake record in the program and obtain all possible ground
motion parameters.

75
 Recommended Reading
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.
 Iyengar, R.N. and Raghu kanth, S.T.G. (2004) Attenuation of Strong Ground
Motion in Peninsular India. Seismological Research Letters, 75(4), 530-540
 Sharma, M. L. (2000) Attenuation relationship for estimation of peak ground
vertical acceleration using data from strong motion arrays in India. 12th World
conference on Earthquake Engineering, New Zealand, 1964/ 4 / A, pp1-8.
 Stein, S. and Wysession, M. (2003). An introduction to seismology, earthquakes,
and earth structure, Blackwell Science.
 http://webshaker.ucsd.edu/homework/Ground_motion_parameters.pdf
 Download free software for the estimation of ground motion parameters :
http://www.seismosoft.com/SeismoSignal-2016-Release-1

76
Seismic Hazard Analysis

77
 Seismic Hazard Analysis
Seismic hazard analysis is concerned with getting an
estimate of the strong-motion parameters at a site for
the purpose of earthquake resistant design or seismic
safety assessment.
For generalized applications, seismic hazard analysis can
also be used to prepare macro or micro zoning maps
of an area by estimating the strong-motion parameters
for a closely spaced grid of sites.
Seismic hazard analysis can be done in two ways
1. Deterministic seismic hazard Analysis (DSHA)
2. Probabilistic seismic hazard analysis (PSHA)

78
 Seismic Hazard Analysis
DSHA
In the deterministic approach, the strong-motion
parameters are estimated for the maximum
credible earthquake, assumed to occur at the
closest possible distance from the site of interest,
without considering the likelihood of its
occurrence during a specified exposure period.
PSHA
Probabilistic approach integrates the effects of all the
earthquakes expected to occur at different
locations during a specified life period, with the
associated uncertainties and randomness taken
into account.
79
 Deterministic seismic hazard analysis
The DSHA approach uses the known seismic sources
sufficiently near the site and available historical seismic
and geological data to generate discrete, single-valued
events or models of ground motion at the site.

• Typically one or more earthquakes are specified by magnitude and

location with respect to the site of interest.

• Usually the earthquakes are assumed to occur on portions of the

source closest to the site.

• The site ground motions are estimated deterministically, given the

magnitude, source-to-site distance, and site condition.
80
 Deterministic seismic hazard analysis

DSHA consists of four primary steps:

1. Identification and characterization of all sources

2. Selection of source-site distance parameter

3. Selection of “controlling earthquake”

4. Definition of hazard using controlling earthquake

81
 Deterministic seismic hazard analysis
Identification and characterization of all sources
Identification
All sources capable of producing significant ground motion
at the site
Large sources at long distances
Small sources at short distances

Characterization
Definition of source geometry
Establishment of earthquake potential
82
Deterministic seismic hazard analysis

Identification and characterization of all sources

Characterize geometry Source
Point source
Constant source-site distance Site
Volcanoes, distant short faults
Linear source Source
One parameter controls distance
Shallow, distant fault

Site 83
Deterministic seismic hazard analysis

Identification and characterization of all sources

Characterize geometry
Three-dimensional Source
Three parameters control distance
Close sources of large dimensions
Insufficient data for accurately determining the
source geometry

Site
Source
84
Deterministic seismic hazard analysis

Identification and characterization of all sources

Which sources are capable of producing significant motion at the
site of interest?

What is significant motion?

Parametric definition
Peak acceleration - usually ~0.05g
Spectral acceleration - at fundamental period, if known
Other parameters
Use predictive (attenuation) relationship to determine distance of
interest
85
Deterministic seismic hazard analysis

Determination of source-site distance

Which distance to consider?

86
 Deterministic seismic hazard analysis

Determination of source-site distance

Typically assume shortest source-site distance (“worst
case” scenario)

87
Determination of source-site distance

M1 M3

M2

88
Selection of controlling earthquake

Establish earthquake potential - typically Mmax

Empirical correlations
Rupture length correlations
Rupture area correlations
Maximum surface displacement correlations

“Theoretical” determination
Slip rate correlations
89
 Active Fault

The definition of active fault requires:

Movement at or near the ground surface at least once
within the past 35,000 yrs or movement of a recurring
nature within the past 500,000 yrs
Instrumentally determined Macroseismicity showing a
direct relationship with the fault;
or
A structural relationship to a capable fault, such that
movement on one could reasonably be expected to be
accompanied by movement on the other

90
 Magnitude Indicators

Magnitude of earthquakes can be estimated by

correlating observed deformation characteristics with
known magnitudes of recorded earthquakes.
Faults do not rupture over their entire lengths.
Individual fault segments with physically controlled
boundaries rupture repeatedly.
Rupture length, area and displacement can be estimated
by post earthquake field geological investigations.
Correlating earthquake magnitude with rupture
parameters involves regression on limited data sets and
consequently produces an estimate of expected value of
earthquake magnitude.
91
 Magnitude Indicators
Empirical relationships between Mw, Surface rupture
length L (km), rupture area A (km2) and maximum
surface displacement D (m)

92
 Magnitude Indicators
Scatter inherent in databases used by Wells and
Coppersmith (1994) in developing the correlations.

93
 Magnitude Indicators
Comparison of several empirical relationships used to
find the maximum magnitude from
(a) the fault rupture length and (b) the fault rupture
area

Source: Kramer (1996) 94

 Tectonic Evidence
Ruff and Kanamori (1984) related earthquake
magnitude to rate of convergence of moving slabs and
the age of the subducted slab

Mw = -0.0089T +0.34V +7.96

T = age in millions of
years
V= rate of convergence
in cm/year

Source: Kramer (1996)

Diagonal lines in figure correspond to the equation. Data points represent actual
earthquakes 95
Historic Seismicity

Major tectonic
features and the
distribution of
epicenters of
available data on
past earthquakes
in a typical
segment of the
Himalayan region

96
“Theoretical” Determination of Mmax

Slip rate approach

Seismic moment Mo = µ A D
where
µ = shear modulus of rock; A = rupture area;
D = average displacement over rupture area
If average displacement D relieves stress/strain built
up by movement of the plates over some period, T,
then
D=SxT
where S is the slip rate
97
Theoretical Determination of Mmax

Slip rate approach

Mo = µ A S T (S is the slip rate)
and the “moment rate” can be defined as
Mr = Mo/T = µ A S
Knowing the slip rate and knowing (assuming) values
of m, A, and T, the moment rate can be used to
estimate the seismic moment as
Mo = Mr T
Then Mw = log Mo/1.5 - 10.7

98
 Selection of controlling earthquake
Decision should be based on ground motion parameters of
greatest interest
• Consider all sources
• Assume Mmax occurs at Rmin for each source
• Compute ground motion parameter(s) based on Mmax
and Rmin
• Determine critical value(s) of ground motion
parameter(s)

99
Selection of controlling earthquake

R3
R1
M1 M3
R2

M2

100
Selection of controlling earthquake

Estimate maximum magnitude that could be produced

by any source in vicinity of site.
Find value of Rmax that corresponds to Mmax at threshold
value of parameter of interest, Ymin. (Y is ground motion
parameter)

101
 Selection of controlling earthquake
Predictive relationships are used to estimate the ground motion
parameter (Y).
Most commonly used ground motion parameter is PHA
Typical predictive relationship for PHA (Cornell, 1979)

• Plot the variation of Y with R for known magnitude M

• Plot these variations for all the sources
• Mark the distances from the sources
• Measure Y for the distance for different sources from the plots

102
Selection of controlling earthquake source

Select the controlling source for which the combination of

magnitude and distance produces highest value of Y.

103
Hazard Estimation from Mmax and R

Use M and R to determine such parameters as:

Peak acceleration
Spectral accelerations
Duration

104
 Steps in DSHA: Summary
R3
R1
M3 M1 M3
M1
R2

M2 M2

(i)
(ii)

 Y1 
Y 
 2 
Y  . 
 . 
 
Yn 
105

(iii) (iv)
Deterministic seismic hazard analysis

DSHA produces a “scenario” earthquake for design

DSHA is based on worst-case scenario
DSHA provides no information about how likely design
earthquake is to occur during life of structure
Design earthquakes may occur every 200 yrs in some
places, every 10,000 yrs in others
DSHA may require subjective opinions on some input
parameters
Variability in effects not rationally accounted

106
Example:
Theoretical Determination of Mmax for
Himalayas
From west to east, the entire Himalaya has a length of
about 2500 km, and the width of the associated
seismic source is about 100 km.
The source of major earthquakes along the Himalaya
has been postulated as a gently dipping detachment
plane, north of the main boundary fault (MBF), at a
depth of about 20 to 30 km.
Thus, the total rupture plane of the Himalaya has an
area A of about 2.5 x 105 km2.
The shear modulus, µ for the Himalayan rocks can be
taken as 34 GPa (3.4 x 1011 dyne/cm2).
107
Theoretical Determination of Mmax for Himalayas

After accounting for the trans-Himalayan deformations, the long-

term average of the slip rate, s , along the Himalayan detachment
plane is corroborated to be about 15 mm/year.

This gives the moment rate

Assuming that the recurrence period (T)for largest earthquakes

with magnitude 8(+) anywhere in the Himalaya is about 40 years,

108
 Example Problem
The site shown in the Figure below is located near four active faults.
Fault A is a normal fault, faults B and C are strike-slips faults and
fault D is a reverse fault. The coordinates of the site and faults given
in the figure are in km. Assuming that only linear segment can
rupture in an individual event, determine the anticipated peak
acceleration at the site.
 Magnitude Indicators
Empirical relationships between Mw, Surface rupture
length L (km), rupture area A (km2) and maximum
surface displacement D (m)

110
 Example Problem

The site shown in the

Figure below is located
near four active faults.
Fault A is a normal fault,
faults B and C are strike- ln 𝑃𝐻𝐴 = 6.74 + 0.859𝑀 − 1.80 ln 𝑅 + 25 , 𝑃𝐻𝐴 𝑖𝑛 𝑔𝑎𝑙𝑠 & 𝑅 𝑖𝑛 𝑘𝑚
slips faults and fault D is a
reverse fault. The
coordinates of the site
and faults given in the
figure are in km.
Assuming that only linear
segment can rupture in an
individual event,
determine the anticipated
peak acceleration at the
site.
 Solution
1. There are total 6 fault segments to be considered. Fault B has
three segments (B1, B2 and B3)
2. The closest distance of the fault segment from the site (R) is
computed.
3. Length of each fault segment is calculated
4. Maximum magnitude associated with each fault is obtained from
the empirical relationships given by wells and coppersmith
(1994) for the particular fault type given in the problem.
5. Peak acceleration is obtained by predictive relationship of
Cornell et al. (1979)

Results are shown.

 Solution
Fault Distance to Length Mmax PHA
site (g)
R (km)
A 33.0 70.7 7.3 0.305

B1 42.7 22.4 6.7 0.138

B2 18.0 30.0 6.8 0.340

B3 17.5 22.4 6.7 0.319

C 5.0 12.0 6.4 0.467

D 25.0 31.6 6.8 0.260

Calculations indicate that the peak acceleration site is

0.467 g produced by an earthquake of magnitude 6.4 at
site C.
 Exercise Problem

Using DSHA compute the PHA for the site below. Use the
following attenuation relationship.

Source 3, Mmax= 6

Source 2, Mmax= 6.5

site

40 km

Source 1, Mmax= 7.5

40 km
 Recommended Reading

 Reiter, L. (1990) Earthquake Hazard Analysis: Issues and insights,

Columbia university press.
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.
 Anderson, J.G. and J.N. Brune (1999). Probabilistic hazard analysis without the
ergodic assumption, Seism. Res. Lett. 70, 19-23.
 Cornell, C.A. (1968). Engineering seismic risk analysis, Bull. Seism. Soc. Am., 58,
1583-1606.
 http://geo.cv.nctu.edu.tw/prob/download/SeismicHazardAnalysis.pdf
 https://www.usbr.gov/ssle/damsafety/risk/BestPractices/Chapters/II-3-
20150620.pdf
Probabilistic Seismic Hazard
Analysis (PSHA) – Part I

116
Probabilities
Supongamos que la estadística
muestra que hay una correlación
entre que los partidos ganados de
la Selección Colombia y los goles
que metió el Tino cuando jugaba
de titular. Sí el Tino mete un gol,
Colombia tiene 50% de
“probabilidad” de ganar. Sí hace
dos goles, 60%; y si hace tres
goles 70%. Sí al Tino lo echan por
cochino, la probabilidad baja a
25%. Sí no hace goles, de 50%.

117
Además, las estadísticas futboleras muestran que
cuando el Tino se enfrenta a Chilavert (Mostro
Arquero de Paraguay), tiene las siguientes
“probabilidades”:
Echan = 20%
Sin goles = 20%
Un gol = 30%
Dos Goles = 15%
Tres Goles = 10%

¿Cuál es la probabilidad de que Colombia gane el

próximo partido con Paraguay teniendo en cuenta
que el Tino y Chilavert van a jugar?
118
Probabilistic Seismic Hazard Analysis

DSHA PSHA
Assumes single scenario Assumes many scenarios
Selects single magnitude for Considers all magnitudes
each seismic source associated with all seismic
sources
Selects the closest distance Considers all possible
between the source and site distances between source
and site
Assumes effects due to Considers all effects
magnitude and distance
Probabilistic Seismic Hazard Analysis
Why PSHA?
Because there is lot of
uncertainty involved in the
time of occurrence,
magnitude and size of the
earthquakes

Assumes many scenarios

Considers all magnitudes associated with all

seismic sources
Considers all possible distances between source
and site
Considers all effects
Probabilistic Seismic Hazard Analysis

PSHA characterizes uncertainties in the

location, size, frequency and effects of
earthquake and combines all of them to
compute the probability of different
levels of ground shaking
Probabilistic Seismic Hazard Analysis

PSHA consists of four primary steps:

1. Identification and characterization of all sources
2. Characterization of seismicity of each source
3. Determination of ground motions from each source
4. Probabilistic calculations
Probabilistic Seismic Hazard Analysis

Uncertainty in source to site distance:

Where on fault is the rupture most likely to occur?

Source

Site

Source to site
distance depends on
where rupture occurs
Probabilistic Seismic Hazard Analysis

Uncertainty in source to site distance:

Where on fault is the rupture most likely to occur? Lot of possibilities

Source

Site

Source to site
distance depends on
where rupture occurs
 Probability Distribution Function
l
rmin dl Lf
r source
For linear source shown, the r+dr
probability of earthquake to site
occur on the small segment of
the fault between L=l and L=l+dl
is same as the probability that it
occurs between R=r and R=r+dr
fR(r)

Where fL (l) and fR (r) are the

probability distribution functions
for the variables L and R. r
rmin rmax

125
Probability Distribution Function
l
rmin
r dl Lf source
r+dr
site

If earthquakes are assumed

to uniformly distributed over
the length of the fault
fR(r)

r
rmin rmax

126
 Probability Distribution Function

Uncertainty in source to site distance:

Two practical ways to determine fR(r)

Draw series of concentric

rmin circles with equal radius
increment
Measure length of Li
between each pair of
concentric circles
rmax
Assign weight equal to Li/L
to each corresponding
distance
 Probability Distribution Function

Uncertainty in source to site distance:

Two practical ways to determine fR(r)

Divide entire fault into

rmin equal length segments
Compute the distance of
site to center for each
segment

rmax
Create histogram of source
to site distance. Accuracy
increases with increasing
number of segments
 Probability Distribution
For example if we roll a single die, the probability we roll a 2 = 1/6
We usually write this in "short-hand" as Pr (2) = 1/6
So, Pr (even number) = 3/6 or 1/2
If we toss a single coin, Pr(heads) = 1/2
If we toss two coins, Pr (exactly 2 heads) = 1/4
For any event like this, we can draw a probability histogram, which
is simply a picture that shows all the possible outcomes and their
corresponding probabilities.
For a single roll of a die, the probability histogram looks like:
1/6

1 2 3 4 5 6 129
 Probability Histogram
For the number of heads in a toss of two coins, the
probability histogram is:

1/2

1/4

No heads 0 1 2

130
 Probability Histogram
Uncertainty in source to site distance:

Divide source into equal

volume increments
Compute the distance of
site to center for each
element
Create histogram of source
to site distance. Accuracy
increases with increasing
number of segments
131
Probabilistic Seismic Hazard Analysis
Uncertainty in earthquake magnitude:

Determination of Mmax is same as for DSHA

Empirical Correlations
• Rupture length correlations
• Rupture area correlations
• Maximum surface displacement correlations
Theoretical determination using slip rate correlations

Knowledge of the distribution of magnitude is important

Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes

NM log N M

M M
Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes

log M

Mean annual rate of

Exceedance
λM = NM/T
M
Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes

log M

0.01 100 years

0.001 1000 years

Return period log TR
(recurrence interval)
M
TR=1/λM
Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes

log M
10a

Gutenberg-Ritcher log TR
Recurrence Law 0
M
log λM = a-bM
Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes

Gutenberg –Ritcher Recurrence Law

log λM = a-bM
Implies that earthquake magnitudes are
exponentially distributed (exponential pdf)
Can also be written as
llln λM = α – βM
Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes
For worldwide data (Circumpacific belt)
logλm= 7.93-0.96M
M=6 λm=148/yr TR=0.0067yr
M=7 λm=16.2/yr TR=0.062yr
M=8 λm=1.78/yr TR=0.562yr

M=12 λm=0.437/yr TR=2.29yr

M>12 every two years?

Probabilistic Seismic Hazard Analysis

Gutenberg –Ritcher Recurrence Law

log λM = a-bM
Implies that earthquake magnitudes are
exponentially distributed
The equation can also be written as
lln λM = α – βM
Then,
λM = 10a-bM = exp[α – βM]
where α=2.303a and β=2.303b
Probabilistic Seismic Hazard Analysis

Gutenberg –Ritcher Recurrence Law

log λM = a - bM
If a lower threshold value is sent of the Magnitude as
m0,
Probability of an earthquake of magnitude <m and
>m0 is given as
m0  m
P[M  m | M  m 0 ]   1  e   ( m  m0 )
m0
This is called Bounded Gutenberg –Ritcher
Recurrence Law
 Probabilistic Seismic Hazard Analysis
Neglecting events below minimum magnitude, mo
m = nexp[- (m - mo)] for m > mo (McGuire and Arabasz, 1990)
where n= exp[a- mo].
FM (m) = P[M<m| M>m0] = (m0 – m)/ m0 = 1-exp (-(m-m0))
Then,
fM(m) = d/dm (FM(m)) = e-m- mo)

Every source has some maximum magnitude

Distribution must be modified to account for Mmax
Bounded G-R recurrence law
exp[   (m  m0 )]  exp[   (mmax  m0 )]
m  v
1  exp[   (mmax  m0 )]

Where n = exp [a –  m0) and a a ln (10 )and  b ln (10 ) 141
Probabilistic Seismic Hazard Analysis

Bounded G-R
logλm Recurrence Law

Mmax M
Probabilistic Seismic Hazard Analysis

Distribution of earthquake magnitudes

Characteristic Earthquake Recurrence Law

Paleoseismic investigations
Show similar displacements in each earthquake
Individual faults produce characteristic earthquakes
Characteristic earthquake occur at or near Mmax
Could be caused by geologic constraints
More research, field observations needed

143
 Probabilistic Seismic Hazard Analysis
Distribution of earthquake magnitudes
Characteristic Earthquake Recurrence Law

Seismicity data

Geologic data

Mmax M
Probabilistic Seismic Hazard Analysis
Predictive relationships
Standard error –use to evaluate conditional probability

ln Y
P[Y > Y*l M=M*, R=R*]
Y=Y*
lnY

M=M*

R=R* log R

145
Probabilistic Seismic Hazard Analysis
Predictive relationships
Standard error –use to evaluate conditional probability

ln Y

P[Y > Y*l M=M*, R=R*]

lnY
Y=Y*
M=M*

R=R* log R

146
 Recommended Reading

 Reiter, L. (1990) Earthquake Hazard Analysis: Issues and insights,

Columbia university press.
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.
 Anderson, J.G. and J.N. Brune (1999). Probabilistic hazard analysis without the
ergodic assumption, Seism. Res. Lett. 70, 19-23.
 Cornell, C.A. (1968). Engineering seismic risk analysis, Bull. Seism. Soc. Am., 58,
1583-1606.
 http://geo.cv.nctu.edu.tw/prob/download/SeismicHazardAnalysis.pdf
 https://www.usbr.gov/ssle/damsafety/risk/BestPractices/Chapters/II-3-
20150620.pdf
Probabilistic Seismic Hazard
Analysis (PSHA) – Part II

148
 Probabilistic Seismic Hazard Analysis

Temporal uncertainty
Poisson process- describes number of occurrences of an
event during a given time interval or spatial region.
1. The number of occurrences in one time interval are
independent of the number that occur in any other time
interval.
2. Probability of occurrence in a very short time interval
is proportional to length of interval.
3. Probability of more than one occurrence in a very
short time interval is negligible.

149
Probabilistic Seismic Hazard Analysis

Temporal uncertainty
Poisson process
m ne m
P[N  n] 
n!
where n is the number of occurrences and μ is the
average number of occurrences in the time interval of
interest.

150
Probabilistic Seismic Hazard Analysis

Temporal uncertainty
Poisson process
Letting μ= λt

P[N  n] 
t n e  t
n!
Then
P[ N  0]  P[ N  1]  P[ N  2]  P[ N  3]  .....P[ N  ]

=1- P[N=0]
=1- e-λt

151
Probabilistic Seismic Hazard Analysis

152
Probabilistic Seismic Hazard Analysis

ln(1  P)

t

153
 Probabilistic Seismic Hazard Analysis

Temporal uncertainty
Then, the annual rate of exceedance for an event with a 10%
probability of exceedance in 50 yrs is

ln(1  0.1)
  0.0021
50

The corresponding return period is TR= 1/λ = 475 yrs.

For 2% in 50 yrs, λ = 0.000404/yr TR = 2475 yrs.

154
Probabilistic Seismic Hazard Analysis

Summary of uncertainties

Location fR(r) Source-site distance pdf

Magnitude pdf
Size fM(m)

Effects P[Y>Y* l M=M*, R=R*] Attenuation

relationship
including standard
error
Timing P=1-e-λt
Poisson model

155
 Probabilistic Seismic Hazard Analysis

Combining uncertainties – probability computations

P[A] = P[A∩B1] + P[A ∩ B2] +…….+ P[A ∩ BN]

P[A] = P[A|B1]P[B1] + P[A|B2] P[B2] +…….+ P[A|BN] P[BN]

Total Probability
B2 B3 Theorem
B1
A BB
B4 B5

156
Probabilistic Seismic Hazard Analysis

Combining uncertainties – probability computations

Applying total probability theorem,

where X is a vector of parameters.

We assume that M and R are the most important parameters

and that they are independent. Then,

157
 Probabilistic Seismic Hazard Analysis

Combining uncertainties-probability computations

Above equation gives the probability that y* will be

exceeded if an earthquake occurs. Can convert probability to
annual rate of exceedance by multiplying probability by
annual rate of occurrence of earthquakes.

where
158
 Probabilistic Seismic Hazard Analysis

Combining uncertainties-probability computations

If the site of interest is subjected to shaking from more than one

site (say Ns sites), then
Ns
y   v i
*
 P [Y  y* | m, r] f Mi (m)f Ri (r)dmdr
i 1

For realistic cases, pdfs for M and Rare too complicated to

integrate analytically. Therefore, we do it numerically.

159
 Probabilistic Seismic Hazard Analysis

Combining uncertainties-probability computations

Dividing the range of possible magnitudes and distances into NM

and NR increments, respectively
Ns N M N R
y   v i
*
 P [Y  y* | m , r ] f
j k Mi (m j )f Ri (rk )mr
i 1 j1 k 1

This expression can be written, equivalently as

Ns N M N R
y   v i
*
 P [Y  y* | m , r ] P[M  m ]P[R  r ]
j k j k
i 1 j1 k 1

160
 Probabilistic Seismic Hazard Analysis

Combining uncertainties-probability computations

What does it mean?

All possible distances are

considered- contribution
of each is weighted by its
probability of occurrence
Ns N M N R
y   v i
*
 P [Y  y* | m , r ] P[M  m ]P[R  r ]
j k j k
i 1 j1 k 1

All sites are All possible magnitudes are

considered considered- contribution of
All possible effects are
considered- each weighted each is weighted by its
by its conditional probability of occurrence
probability of occurrence
161
Probabilistic Seismic Hazard Analysis
Combining uncertainties-probability computations
Compute conditional probability for each element on grid
Enter in matrix (spreadsheet cell)

P[Y>Y*| M=m2, R=r1]

ln Y
P[Y>Y*| M=m2, R=r2]
P[Y>Y*| M=m2, R=r3]

Y = Y*
ln Y

r1 r2 log R
r3
rN
162
 Probabilistic Seismic Hazard Analysis

Combining uncertainties-
probability computations Seismic hazard curve shows the
mean annual rate of exceedance of
a particular ground motion
parameter. A seismic hazard curve
is the ultimate result of a PSHA.
log λamax

log TR

amax
163
 Probabilistic Seismic Hazard Analysis
Using seismic hazard curves
Combining uncertainties-probability
computations
Contribution of sources
Can break λ values down into
contributions from each source.
Plot seismic hazard curves for each
Total
source and total seismic hazard
curve (equal to sum of source
log λamax

curves)
log TR

2 Curves may not be parallel, may

cross
1 Shows which source(s) most
important.
3

amax
164
 Probabilistic Seismic Hazard Analysis

Using seismic hazard curves

Contribution of sources
Combining uncertainties- Can develop seismic hazard curves
probability computations for different ground motion
parameters
Total Peak acceleration
log λamax

Special accelerations
log TR

Other
Choose desired λ value
Read corresponding parameter
values from seismic hazard curves

amax, Sa
165
 Probabilistic Seismic Hazard Analysis

Using seismic hazard curves

Contribution of sources
Can develop seismic hazard curves
for different ground motion
parameters
Peak acceleration
log λamax

Special accelerations
log TR

Other
Choose desired λ value
Read corresponding parameter
values from seismic hazard curves

amax, Sa
166
 Probabilistic Seismic Hazard Analysis
Disaggregation (De-aggregation)
Common question:
What magnitude & distance does that amax value correspond
to?
5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 Total hazard
25km 0.01 0.01 0.02 0.03 0.03 0.02 0.01 0.01 includes contributions
from all combinations
50km 0.02 0.03 0.04 0.04 0.05 0.04 0.03 0.02 of
M&R
75km 0.03 0.03 0.05 0.06 0.09 0.06 0.05 0.02

100km 0.03 0.03 0.05 0.05 0.08 0.05 0.05 0.02

125km 0.02 0.02 0.03 0.04 0.05 0.03 0.02 0.01

150km 0.01 0.01 0.02 0.03 0.05 0.02 0.01 0.00

175km 0.00 0.00 0.01 0.01 0.03 0.01 0.01 0.00

200km 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00

167
 Probabilistic Seismic Hazard Analysis
Disaggregation (De-aggregation)
Common question

What magnitude & distance does that amax value correspond to?
5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5
Total hazard includes
25km 0.01 0.01 0.02 0.03 0.03 0.02 0.01 0.01 contributions from all
combinations of M & R.
50km 0.02 0.03 0.04 0.04 0.05 0.04 0.03 0.02
Break hazard down into
75km 0.03 0.03 0.05 0.06 0.09 0.06 0.05 0.02 contributions to “see
where hazard is coming
100km 0.03 0.03 0.05 0.05 0.08 0.05 0.05 0.02 from.”
125km 0.02 0.02 0.03 0.04 0.05 0.03 0.02 0.01 M=7.0 at R=75km

150km 0.01 0.01 0.02 0.03 0.05 0.02 0.01 0.00

175km 0.00 0.00 0.01 0.01 0.03 0.01 0.01 0.00

200km 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00

168
Probabilistic Seismic Hazard Analysis
Disaggregation (De-aggregation)

169
Probabilistic Seismic Hazard Analysis

Logic tree methods

Not all uncertainty can be described by probability distributions

Most appropriate model may not be clear

Attenuation relationship
Magnitude distribution
etc
Experts may disagree on model parameters
Fault segmentation
Maximum magnitude
etc

170
Probabilistic Seismic Hazard Analysis
Logic tree methods
Attenuation Magnitude
m Model Distribution Mmax

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
BJF
7.0 (0.2)
(0.5) Char 7.2 (0.6)
(0.3) 7.5 (0.2)

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
A&S
7.0 (0.2)
(0.5)
Char 7.2 (0.6)
(0.3) 7.5 (0.2)

171
Probabilistic Seismic Hazard Analysis
Logic tree methods
Attenuation Magnitude
Model Distribution Mmax
m

7.0 (0.2)
Sum of weighing factors G-R 7.2 (0.6)
coming out of each
(0.7) 7.5 (0.2)
node must equal 1.0
BJF
7.0 (0.2)
(0.5) Char 7.2 (0.6)
(0.3) 7.5 (0.2)

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
A&S
7.0 (0.2)
(0.5)
Char 7.2 (0.6)
(0.3) 7.5 (0.2)

172
Probabilistic Seismic Hazard Analysis
Logic tree methods
Attenuation Magnitude
Model Distribution Mmax
m

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
BJF
7.0 (0.2)
(0.5) Char 7.2 (0.6)
(0.3) 7.5 (0.2)

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
A&S 0.5x0.7x0.2 = 0.07
7.0 (0.2)
(0.5)
Char 7.2 (0.6)
(0.3) 7.5 (0.2)

173
Probabilistic Seismic Hazard Analysis
Logic tree methods
m
Attenuation Magnitude
Model Distribution Mmax

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
BJF
7.0 (0.2)
(0.5) Char 7.2 (0.6)
(0.3) 7.5 (0.2)

7.0 (0.2)
G-R 7.2 (0.6)
(0.7) 7.5 (0.2)
A&S 0.5x0.7x0.2 = 0.07
7.0 (0.2)
(0.5)
Char 7.2 (0.6)
(0.3) 7.5 (0.2)

174
Probabilistic Seismic Hazard Analysis
Logic tree methods

Recent PSHA logic tree included:

Crustal areal source zones

7 source zones
2 attenuation relationships
3 maximum magnitudes
2 recurrence models
3 source depths
252 terminal branches
Total PSHA required analysis of 612 combinations

175
 Recommended Reading

 Reiter, L. (1990) Earthquake Hazard Analysis: Issues and insights,

Columbia university press.
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.
 Anderson, J.G. and J.N. Brune (1999). Probabilistic hazard analysis without the
ergodic assumption, Seism. Res. Lett. 70, 19-23.
 Cornell, C.A. (1968). Engineering seismic risk analysis, Bull. Seism. Soc. Am., 58,
1583-1606.
 http://geo.cv.nctu.edu.tw/prob/download/SeismicHazardAnalysis.pdf
 http://www.ce.memphis.edu/7119/PDFs/FEAM_Notes/Topic05a-
SeismicHazardAnalysisNotes.pdf

176
Exercise Problems in PSHA

177
 Problem 1
The hypothetical vertical fault segment shown in Fig.1 is represented as a quarter-
circle. On the same graph, plot histograms of expected epicentral distance for
motions at site A and site B assuming:
(a) Earthquakes are equally likely to occur at any point on the fault segment.
(b) Earthquakes are twice as likely to occur at the midpoint of the fault segment as
at either end and the likelihood is linearly distributed between the midpoint
and the ends.

Site B
12km

Fault
Fig 1 N
30km

Site A
30km 12km 178
 Problem 2

In a hypothetical seismically active region, earthquakes have been recorded

over an 80-year period. Part of the record is instrumental, but part is not. Combining
all available data, it appears that the earthquakes have been distributed as follows:

MOMENT MAGNITUDE NUMBER OF EARTHQUAKES

3-4 1800
4-5 150
5-6 11
>6 1

(a) Estimate the Gutenberg-Richter parameters for the region.

(b) Neglecting earthquakes of magnitude less than 3, compute the probability that
an earthquake in the region will have a moment magnitude between 5.5 and 6.5.
(c) Repeat Part (b) assuming that paleoseismic evidence indicates that the region is
not capable of producing earthquakes of moment magnitude greater than 6.5.

179
 Problem 3
The seismicity of a particular region is described by the Gutenberg-Ritcher
recurrence law:

log λm=4.0-0.7M

(a) What is the probability that at least one earthquake of magnitude

greater than 7.0 will occur in a 10-year period? In a 250-year period?

(b) What is the probability that exactly one earthquake of magnitude greater
than 7.0 will occur in a 10-year period? In a 50 year period? In a 250-year
period?

(c)Determine the earthquake magitude that would have a 10% probability of

being exceeded at least once in a 50-year period?

180
 Problem 4
Using given seismic hazard curve, estimate the probability of exceeding
amax = 0.3g in a 50 year period and in 500 years period

Combining uncertainties-probability computations

log λamax

log TR
0.001

amax
181
 Problem 5
Using seismic hazard curve, estimate the peak acceleration that has 10%
probability of being exceeded in a 50 yr period.
Combining uncertainties-probability computations

log λamax

0.0021 log TR

amax =0.21g 182

 Lecture-19

Wave Propagation Theories

183
 Wave Propagation Theories

The propagation of seismic waves is most easily

understood by first considering an unbounded or
infinite medium. A simple, 1D idealization of an
unbounded medium is that of an infinitely long rod or
bar. Using basic requirements of equilibrium of forces
and compatibility of displacements, and using strain-
displacement and stress-strain relationships, a one-
dimensional wave equation can be derived and
solved.
Three different types of vibration can occur in a thin
rod: longitudinal vibration; torsional vibration; and
flexural vibration.
 Wave Propagation Theories

One dimensional Wave propagation:

Longitudinal vibration (P waves) Torsional Vibration (S waves)

Wave
Particle
Particle movemen
movement t

Flexural Vibration
(Surface waves)

Particle
movemen
t
 Longitudinal waves in infinitely long rod

Stress at the left end of the

dx
element = x0
Stress at the right end of the
element =   ( / x)dx
x0 x

Dynamic equilibrium of
element requires

  x   2u
  x0  dx  A   x0 A  Adx
 x  t 2
Longitudinal waves in infinitely long rod

dx

Cross-sectional Area = A
Young’s modulus = E
Poisson’s ratio = n
Density = 
Wave propagation velocity = vp
Vp = sqrt (M/)
M = Constrained modulus
Longitudinal waves in infinitely long rod

 x 0
 x0  dx
xo = x (x0 , x
t)

M
vp  p wave velocity


u
x  Mx , x 
x

Stress-strain Strain-
relationship displacement
relationship
 Longitudinal waves in infinitely long rod

, E n, A

dx

 x 0
xo = x (x0 , t)  x0  dx
x

u= u (x0 , t) u
u  dx
x
Longitudinal waves in infinitely long rod

 x 0
 x0  dx
xo = x (x0 , t) x

  x  u 2
 x0  dx  A   x 0 A   A dx 2
 x  t

 x  2u
 2
x t

stresses displacements
 Longitudinal waves in infinitely long rod

  x  u 2
 x  dx  A   x 0 A  Adx 2
 x  t
Simplifying and expressing stress in terms of
displacement, we get one dimensional wave equation
as
  x   2u
 
 x  t 2

This is one-dimensional equation of motion

 2u 2  2
u
 vp
t 2 x 2
 Constrained Modulus M

dx
Cross-sectional Area = A
xx =1/E [xx – nyy – n zz ]
Young’s modulus = E
Poisson’s ratio = n
yy =1/E [yy – nxx – n zz ] Density = 
Wave propagation velocity = vp
zz =1/E [zz – nxx – n yy ] vp = sqrt (M/ )
M = Constrained modulus
yy and zz are zero for constrained conditions

And constrained modulus M = xx/ xx

1 n
Solve and get M E
1 n (1  2n )
 Longitudinal waves in infinitely long rod

M = xx/ xx
xx = M xx = M u/ x
  x   2u
 
 x  t 2
  2u   2u  2u M   2u 
M  ; 
 x 2   t 2 t 2   x 2 
 

 2u 2  2
u
 vp
t 2 x 2 M
vp 

Longitudinal waves in infinitely long rod

 x 0
 x0  dx
xo = x (x0 , t) x

M
vp 


How fast does the particle in the

rod move when the wave passes
by?
 Longitudinal waves in infinitely long rod

Particle velocity v = u/ t

u = x x (from strain-displacement relationship)
v = u/ t = x x/ t
x = vp t (from the definition of wave propagation velocity)
x = x/ M
x x x
v vp  vp 
M v 2p v p
Particle velocity is proportional to axial stress. vp is
the specific impedance of the material.
Longitudinal waves in infinitely long rod

 x 0
 x0  dx
xo = x (x0 , t) x

 x  2u
 2
x t
u
x  Mx , x 
x
1 n
M E
1 n (1  2n )

 2u M  2u  2u 2  u
2
 or  vp 2
t 2
 x 2 t 2
x
 Torsional waves in an infinitely long rod

Tx0 d
Tx0+( T/ x) dx
dx

Waves are propagating in only one direction

Particle movement is in planes perpendicular to the
direction of wave propagation
But the rod is subjected to angular deformation due to
S wave propagation.
Particle movement
 Torsional waves in an infinitely long rod

Torsional waves involve Tx0 d

rotation of the rod about its dx

Tx0+( T/ x)
dx

own axis.
The particle motion is  T   2
 Tx0  dx   Tx0   J dx
 x  t 2
constrained to planes
perpendicular to the direction Where J = Polar moment of
of wave propagation. inertia of the rod about its axis


For a bar of constant cross- T GJ
section, the equation of motion x
here is T  2
J
 2 G  2  2
 x t 2
  v s2
t 2  x 2 x 2
Torsional waves in an infinitely long rod

 T   2
 Tx 0  dx   Tx0   J dx
 x  t 2

T  2
J
x t 2
Torque Rotation
Torsional waves in an infinitely long rod

 T   2
 Tx 0  dx   Tx 0   J dx
 x  t 2

T  2
J
x t 2

T GJ
x

 2 G  2 2
2 
  vs
t 2  x 2 x 2

This is also 1-D wave equation (wave propagation in one

direction), but particle movement is in two planes.
 Solution of 1D wave equation

 2u 2  2u T = 2p/
 vp
1 D wave equation: t 2 x 2
u
Solution for the above equation can be expressed as t
u (x, t) = A cos (t – k x) + B cos (t + k x)
First and seconds terms represent harmonic waves
u
propagating in positive and negative x directions respectively
k = wave number = /v x
 = 2p/k
The same solution can be written using complex notation

u ( x , t )  C e i ωt  kx   D e i wt  kx 
 Recommended Reading

 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.

http://geo.cv.nctu.edu.tw/EngGeo/download/Seismic_Wav
e_Propagation1D.pdf
http://octavia.ce.washington.edu/drlayer/theory/T-Basic-
Concepts.html

203
 Lecture-20

Seismic Wave Propagation

204
 Solution of 1D wave equation

 2u 2
2  u T = 2p/
1 D wave equation: 2
 vp
2
t x u
Solution for the above equation can be expressed as t
u (x, t) = A cos (t – k x) + B cos (t + k x)
First and seconds terms represent harmonic waves
propagating in positive and negative x directions u
respectively x
 = 2p/k
k = wave number = /v
The same solution can be written using complex
notation

u ( x , t )  C e i ωt  kx   D e i wt  kx 
 Waves in semi-infinite body
For waves whose motion is concentrated in a shallow zone near the
free surface (surface waves), earth is idealized as semi-infinite body
with planar free surface.
Plane wave

These plane waves in homogeneous

elastic half-space represent Rayleigh y
waves. Particle motion occurs in the x
x-z plane in figure.
If the semi-infinite half-space is
overlain by a layer of material with
lower body wave velocity, Love waves
can develop.
z
i.e. When ( G1/ 1 < G2/ 2 ) in figure x
Surficial
H z  1, G 1 layer
2, G2 Half-
space
Three-Dimensional wave propagation

y

xy
yz
xy

zy
x
zx xz
z

Three-dimensional stress on an element.

 Equations of Equilibrium

 x  xy  xz
   Xb  0
x y z
 xy  y  yz
   Yb  0
x y z
 xz  yz  z
   Zb  0
x y z
 Strain Displacement

(u,v,w) are the x, y and z components of

displacement.
u u  v
x   xy  
x y x
v u  w
y   xz  
y z x
w w  v
z   yz  
z y z
 Stress-Strain Relationships

x  1  n n n 0 0 0  x 
   n 1 n n 0 0 0   
 y   y 
z  E  n n 1 n 0 0 0  z 
   (12n)  
xy  (1  n)(1  2n)  0 0 0 2 0 0  xy 
yz   0 0 0 0
(12n)
2 0  yz 
   (12 n)  
zx   0 0 0 0 0 2  zx 

E
Note :G 
2(1  n )
 Stress-Strain Relationships
For isotropic, linear elastic material,

x  2m   0 0 0x 

     
 y    2m  0 0 0 y 
z     2m 0 0 0 z 
   
xy  0 0 0 m 0 0 xy
 yz  0 0 0 0 m 0 yz
    
zx  0 0 0 0 0 mzx

Where  and m and Lame’s constants

 Elastic constants
For isotropic, linear elastic material,

3  2 m
E  m
  m
2m
K   
3
G  m

n 
2   m 

Where  and m and Lame’s constants

 Three-Dimensional wave propagation

Equations of motion for 3D case are:

 Three-Dimensional wave propagation

Equations of motion for an elastic solid for 3D case

are: 2  u 
    m   m 2u
t 2 x
 2v 
    m   m 2 v
t 2 x
 2w 
    m   m 2 w
t 2 x

Where  is the volumetric strain = x+y+z

& m are Lame constants and
2 = Laplacian operator = 2 2 2
 
x 2 y 2 z 2
 Solution of the 3D equations of motion

Solving 3 D equations of motion for an isotropic, linear, elastic

solid produce equations for two types of waves, which can travel
through unbounded solid. These are called body waves (P and S
waves).
Longitudinal waves are P waves, which involve no rotation
Transverse waves in two mutually perpendicular directions are S
waves, which involve no volume change
SH waves: S- waves in which particle motion is in horizontal
plane
SV waves: S- waves in which particle motion is in vertical plane
Waves in layered body
Incident P-wave: P-Reflection, P-transmitted, S-reflection, S-
transmitted

When a P-wave hits a velocity boundary, it makes three P-waves: a refracted, a

transmitted, and a reflected wave. But, it also forms a reflected and
transmitted S-wave too. Snell’s law still works, except when the outgoing ray is
a S-wave, the S-wave velocity must be used in Snell’s Law.
 Refraction of a (P or S) ray due to velocity change: Snell’s Law

reflecte When a ray traverses a velocity

inciden
d contrast (change), the ray MUST
t refract. Because otherwise the
wavefront would ‘tear’ apart
which the physics does not
permit. Also, in most all cases a
transmit
reflected wave is made.
ted
All waves refract: e.g., seismic,
light, EM, water waves.
The refraction law: Snell’s Law. Note that For a wave reflected from a flat
all angles between the ray and interface interface, the angle of incidence
are reckoned with respect to the surface EQUALS the angle or reflection.
normal.
sin(1 ) sin( 2 ) incidence   reflection

v1 v2
 Snell’s Law derivation

An explosion makes a wavefield at (S). The waves travel outwards as

represented by the wavefront at different times. Where the wave hits the
velocity interface, the waves refracts (and reflects).

What determines the angle of the refraction into the rock-2 layer ?
 T
Derivation of
0 λ1 T Snell’s Law
1
Ѳ Ѳ
v1 1A
Ѳ B
v2 Ѳ1 Ѳ1 Ѳ
2
2 2
sin(1 ) v1 1
λ2  
T
sin( 2 ) v2 2
T
0
 1
 sin(1 ) sin( 2 ) sin(1 ) sin( 2 )
sin(1 )  1  AB  1   
AB sin(1 ) 1 2 v1 v2

sin( 2 )  2

 AB  2
AB sin( ) 2 v1  f * 1 v2  f * 2

The ratios of the velocities and wavelengths and sin(angles) are all equal.
Otherwise, the wavefield would ‘tear’ apart.
Tracing rays using Snell’s Law in multiple layer medium

Tracing a raypath through multiple layers is

simple. It is just the process of using Snell’s
Law sequentially each successive interface.

Note that the angle (i1) at the top and the

bottom (i1’) of a layer is the same.

If the lower layer’s velocity increases

( v2 > v1 ), the ray refracts AWAY from the
interface normal. In the converse, the ray
refracts TOWARDS the normal.
sin ( 1 ) v
 1
sin ( 2 ) v2
If the deeper layers have a monotonic
increase in velocity, the ray will continue to
flatten out with depth. Eventually, the ray
will reach its turning depth where it goes
exactly horizontal and will start to go up.
Using Snell’s Law for a spherical geometry (not Cartesian)

r1 sin(1 ) r2 sin(2 )

v1 v2
Note that the angle at the top and
the bottom of a spherical shell are
NOT the same. This is because the
‘layers’ are curved.

But, if one is just calculating the

angles on either side of an
interface, then the two radius
values (r1 and r2 ) are the same and
the Cartesian form of Snell’s Law is
operative (i.e., the radius scaling
cancels out in the spherical Snell’s
Law.
 Waves in layered body

Waves perpendicular to boundaries

Transmitted SV
wave

Reflected SV
wave
Incident SV
wave
SV wave: S-waves polarized in the
vertical plane
 Waves in layered body

Waves perpendicular to boundaries

Transmitted SH
wave

Reflected SH
wave
Incident SH
wave
SH wave: S-waves polarized in the
horizontal plane
 Waves in layered body

Inclined waves

Refracted SV wave
Refracted P wave
Reflected P wave
Incident P wave Reflected SV wave
 Waves in layered body

Inclined waves

Refracted SV wave
Refracted P wave
Reflected P wave
Incident SV Reflected SV wave
wave
 Waves in layered body

Inclined waves

Refracted SH wave

Incident SH Reflected SH wave

wave
Waves in layered body

Vs= 150 m/sec

Vs= 300 m/sec
Vs= 450 m/sec
Vs= 600 m/sec
Vs= 750 m/sec
 Attenuation of stress waves

Attenuation of stress waves

The amplitude of stress waves in real

materials decrease or attenuate with
distance

Two primary sources for the attenuation

Material Damping
Radiation Damping
 Attenuation of stress waves

Material Damping

A portion of the elastic energy of stress

waves is lost due to heat generation
Specific energy decreases as the waves
travel through the material
Consequently the amplitude of the stress
waves decreases with the distance
 Attenuation of stress waves

Radiation Damping

The specific energy could also decrease

due to geometric spreading
Consequently the amplitude of the stress
waves decreases with the distance even
though the total energy remains constant
 Recommended Reading

 Chapman, C.H. (2004) Fundamentals of Seismic Propagation, Cambridge

University Press.
 Sato, H., and M. Fehler (1998) Seismic wave propagation and scattering
in the heterogeneous earth, Springer.
 Kramer (1996) Geotechnical Earthquake Engineering, Prentice Hall.
 http://geo.cv.nctu.edu.tw/EngGeo/download/Seismic_Wave_Propagation1D.p
df
 http://octavia.ce.washington.edu/drlayer/theory/T-Basic-Concepts.html

232