SEISMIC HAZARD

ANALYSIS
SEISMIC HAZARD
• Defined as a natural phenomenon (such as ground
shaking, fault rupture, or soil liquefaction) that is
generated by an earthquake, although there are
examples of these phenomena also being produced
by human activities.
• Strong ground motions produced by earthquakes that
could affect engineered structures.
SEISMIC HAZARD
The seismic hazard can be expressed in different
ways: from simple observed macroseismic fields, to
seismostatistical calculations for analyzing earthquake
occurrences in time and space and assessing their
dynamic effects in a certain site or region, to
sophisticated seismogeological approaches for
evaluating the maximum expected earthquake effects
on the Earth's surface.
SEISMIC HAZARD ANALYSIS
• It involve the quantitative estimation of ground-
shaking hazards at a particular site.
• Estimation of earthquake-induced ground motions
having specific probabilities over a given time period.
• Describes the potential for dangerous, earthquake-
related natural phenomena such as ground shaking,
fault rupture, or soil liquefaction.
IDENTIFICATION AND
EVALUATION OF EARTHQUAKE
SOURCES
• To evaluate seismic hazards for a particular site or
region, all possible sources of seismic activity must be
identified and their potential for generating future
strong ground motion evaluated.
IDENTIFICATION AND
EVALUATION OF EARTHQUAKE
SOURCES
Earthquake Sources may be identified on the
basis of the following:
1. Geologic
2. Tectonic
3. Historical
4. Instrumental Evidence
1. Geologic Evidence
• The theory of plate tectonics assures us that the
occurrence of earthquakes is written in the geologic
record, primarily in the form of offset, or relative
displacement of various strata.
• Paleoseismology – study of the geologic record of
past earthquake activity.
1. Geologic Evidence
• Search for geologic evidence of earthquake sources
centers on the identification of faults.
• Tools and techniques used by geologist in identifying
fault:
• Review of published literature
• Interpretation of air photos and remote sensing
• Field reconnaissance
• Test pits and boring
Identification of fault
1. Directly Observable fractures surfaces.
2. Geologically Mappable indicators
4. Secondary geologic features. Includes abrupt
changes in groundwater levels, gradients, and chemical
composition, alignment of springs or volcanic vents
and the presence of hot springs.
5. Lineaments on remote sensing imagery.
6. Geophysical indicators of subsurface faulting.
7. Geodetic indicators. Include fault movement
appearing in geodetic surveys as tilting and changes in
the distance between fixed points.
Fault Activity
•Presence of fault does not indicate the
likelihood of future earthquakes.
•Formal definitions of fault activity are
important because they often trigger legal
requirements for special investigations or
special design provisions.
Fault Activity
•The Specification of fault activity by specific
time intervals is not very realistic(Cluff et al,.
1972; Cluff and Cluff, 1984);
•Faults do not suddenly become inactive on
th th
the 10,000 or 35,000 anniversary of their
last movement.
Active Fault
• California Division of Mines and Geology: one that
produced surface displacement within Holecene time
(approximately the past 10,000 years)
• U.S. Army Corps of Engineers: time period of 35,000
years
• U.S. Bureau of Reclamations: 100,000 years
• U.S. Nuclear Regulatory Commission: Used the term
Capable Fault
Active Fault lines in the Philippines
Fault Lines Affected Areas

Central Philippine Fault Entire Ilocos Norte, Aurora, Quezon, Masbate, Eastren
Leyte, Southern Leyte, Agusan del Norte, Agusan del
Sur, Davao del Norte.
Marikina Valley Fault Montalban, San Mateo, Marikina, Pasig, Taguig,
Muntinlupa, San Pedro, Binan, Carmona, Santa Rosa,
Calamba, Tagaytay, Oriental Mindoro
Western Philippine Fault Luzon Sea, Mindoro Strait, Panay Gulf, Sulu Sea

Eastern Philippine Fault Philippine Sea

Southern of Mindanao Fault Moro Gulf, Celebes Sea

Magnitude Indicators
• Geologic Evidence can also be used to estimate the
magnitude of past earthquake.
• Fault rupture length has often been used to estimate
earthquake magnitude.
• Rupture length methods are best suited to cases in
which rupture surface is fairly narrow, typically less
than about 20 km.
2. Tectonic Evidence
• Plate tectonics and elastic redound theory tell us that
earthquakes occur to relieve the strain energy that
accumulates as plates move relative to each other.
𝑀𝑤 = −0.0089𝑇 + 0.134𝑉 + 7.96
T – age in millions of years
V – rate of convergence in cm/year
3. Historical Seismicity
• Historical accounts of ground-shaking effects can be
used to confirm the occurrence of past earthquakes
and to estimate their geographic distribution
intensity.
• A return period is an estimate of the interval of time
between earthquake. It is a statistical measurement
denoting the average recurrence interval over an
extended period of time, and is usually required for
risk analysis.
• It does not mean that 100-year earthquakes will happen
regularly, every 100 years, despite the connotations of the
name "return period”.
• Note also that the estimated return period is a statistic: it is
computed from a set of data (the observations), as distinct
from the theoretical value in an idealized distribution.
• Further, one cannot determine the size of a 1,000-year
event based on such records alone, but instead must use a
statistical model to predict the magnitude of such an
(unobserved) event.
• Note also that the estimated return period is a statistic: it is
computed from a set of data (the observations), as distinct
from the theoretical value in an idealized distribution. One
does not actually know that a certain magnitude or greater
happens with 1% probability, only that it has been observed
exactly once in 100 years.
• Further, one cannot determine the size of a 1,000-year
event based on such records alone, but instead must use a
statistical model to predict the magnitude of such an
(unobserved) event.
4. Instrumental Seismicity
• Instruments capable of recording temblors have been
first deployed since the end the 19th century.
• Modern technology allowed for finer and clearer
identification of seismic events, even of every little
magnitude.
• Modern-era seismometric networks allow to
recognize earthquake location and magnitude.
4. Instrumental Seismicity
• Instrumental recordings represent the best available
information for the identification and evaluation of
earthquake sources.
Deterministic Seismic Hazard Analysis
Earliest approach taken to seismic hazard analysis
Originated in nuclear power industry applications Still
used for some significant structures such as:
1. Nuclear power plants
2. Large dams
3. Large bridges
4. Hazardous waste containment facilities
5. As “cap” for probabilistic analyses
• In Deterministic Seismic Hazard Analysis (DSHA), is done
for a particular earthquake, either assumed or realistic.
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. Usually the earthquakes
are assumed to occur on the portion of the site closest to
the site. The site ground motions are estimated
deterministically, given the magnitude, source-to-site
distance, and site condition.
Deterministic Seismic Hazard Analysis
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
TITLE
Step 1: Identification of all sources capable of
producing significant ground motion at the site
such as Large sources at long distances and
Small sources at short distances.
Characterization includes Definition of source
geometry and Establishment of earthquake
potential.
Estimate maximum magnitude that could be
produced by any source in vicinity of site and
Find value of Rmax - corresponds to Mmax at
threshold value of parameter of interest, Ymin.
TITLE
TITLE
• Point source where there is
constant source- to site
distance. Earthquakes
associated with volcanic
activity, for example, generally
originate in zones near the
volcanoes that are small
enough to allow them to be
characterized as point source.
TITLE
• Linear source in which one
parameter controls distance
example Shallow and distant
fault
TITLE
• Areal source in which two
geometric parameters control
distance example Constant
depth crustal source. Well
defined fault planes, on which
earthquakes can occur at many
different locations, can be
considered as two-dimentional
areal sources.
TITLE
• Areas where earthquake
mechanisms are poorly
defined, or where faulting is so
extensive as to preclude
distinction between individual
faults, can be treated as three-
dimenstional volumentric
sources
TITLE
• Establish earthquake potential - typically Mmax can be
found by the following
1.Empirical correlations
a. Rupture length correlations
b. Rupture area correlations
c. Maximum surface displacement correlations
2. “Theoretical” determination by Slip rate
correlations
Step 2: Selection of source-site distance parameter must be consistent with
predictive relationship and should include finite fault effect (Figs 15.5-15.7)
Step 3: Selection of Controlling
Earthquakes is based on ground
motion parameter(s). 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). An
example is shown in Figure 12.8
below.
Fig 12.8: Selection of Controlling Earthquake (Combination of M2 and R2
produces highest value of Y)
• Step 4: Definition of hazard using controlling earthquake involves
the use of M and R to determine parameters such as Peak
acceleration, spectral acceleration and Duration.

• DSHA calculations are relatively simple, but implementation of

procedure in practice involves numerous difficult judgments. The
lack of explicit consideration of uncertainties should not be taken
to imply that those uncertainties do not exist.

• Typical results obtained from DSHA analysis is shown in Figure 12.9

Fig 12.9: Typical spectral curve and hazard plot
from DSHA analysis
Probabilistic Seismic Hazard Analysis (PSHA)
• It provides a framework in which these uncertainties
can be identified, quantified, and combined in a
rational manner to provide a more complete picture
of the seismic hazard.
• Understanding the concept and mechanics of PSHA
requires familiarity with some of the terminology and
basic concept of probability theory.
Four Step of PSHA
1. Identification and
Characterization of earthquake
sources and its potential rupture.
Uniform probability distributions
are assigned to each source zone.
These distribution are then
combined to source geometry to
obtain corresponding probability
distribution of source-to-site
distance
Four Step of PSHA
2. Characterization of Seismicity or
temporal distribution of
earthquake recurrence. A
recurrence law, which specifies
the average rate at which an
earthquake of some size will
exceed, is used to characterize
the seismicity of each source
zone.
Four Step of PSHA
3. The ground motion produced at
the site by earthquakes of any
possible size occurring at any
point in each source zone musty
be determined with the use of
predictive relationship.
Four Step of PSHA
4. The uncertainties in earthquake
location, earthquake size, and
ground motion parameter
prediction are combined to
obtain the probability that the
ground motion will be exceed
during particular period of time.
Earthquake Source Characterization
Spatial Uncertainty
Earthquake Source Characterization
Source-to-site distance
Earthquake Source Characterization
Spatial Uncertainty
The distance R is known to be
𝑟𝑠 , consequently, the probability
that 𝑅 = 𝑟𝑠 is assumed to be 1 and
the probability that 𝑅 ≠ 𝑟𝑠 is zero.
Earthquake Source Characterization
Spatial Uncertainty

Where 𝑓𝑙 and 𝑓𝑟 are the probability

density for the variables L and R
respectively.
If earthquakes are assumed to
uniform distributed over the length
of fault.
Earthquake Source Characterization
Spatial Uncertainty
For source zone with more complex
geometry, it is easier to evaluate 𝑓𝑟 𝑟
by numerical than analytical.
Earthquake Source Characterization
Size Uncertainty
The source zone will produce earthquakes
of different sizes up to maximum
earthquake, with smaller earthquake
occurring more frequently.
The distribution of earthquake sizes in a
given period of time is described by a
recurrence law.
Earthquake Source Characterization
Size Uncertainty
Gutenburg – Richter Recurrence Law

where:
is the mean annual rate of exceedance of
magnitude m.
10𝑎 is the mean yearly number of earthquake of
magnitude greater than or equal to zero.
b is describes the relative likelihood of large or small
earthquakes.
Earthquake Source Characterization
Size Uncertainty Bounded Gutenburg – Richter Recurrence Law

Where 𝛼 = 2.303𝑎 and 𝛽 = 2.303𝑏

Where 𝑣 = exp( 𝛼 − 𝛽𝑚𝑜 )

Earthquake Source Characterization
Size Uncertainty Cumulative Distribution Function

Probability Density Function

Earthquake Source Characterization
Size Uncertainty
Characteristics Earthquake Recurrence Law

Paleoseismic Studies indicates that

individual points of faults and fault segments
tend to move by approximately the same
distance in each earthquake. This has been
interpreted to suggest that individual fault
repeatedly generate (within about one-half
magnitude unit) size, known as characteristic
earthquake, at or near maximum magnitude.
Earthquake Source Characterization
Size Uncertainty
Available evidence is insufficient to
determine whether the Gutenburg-Richter or
Characteristic Earthquake law is correct.
Evaluation of each model is most appropriate
for a given source is hampered by the brevity
of historical and/or instrumental record.
Predictive Relationship
Predictive Relationship of Campbell and Bozorgnia

Where:
M is local magnitude
R is the closest distance to the fault rupture in Kilometers

Schematic illustration of conditional probability of exceeding F is zero for strike slip and normal faulting; and 1 for reverse, reverse-
a particular value of a ground motion parameter for a given oblique, and trust faulting.
magnitude and distance.
𝑆𝑆𝑅 = 1 for soft-rock site
𝑆𝐻𝑅 = 1for hard-rock site
𝑆𝑆𝑅 = 𝑆𝐻𝑅 = 0 for alluvium site
TEMPORAL UNCERTAINTY
• To calculate the probabilities of various hazards
occurring in a given time period, the distribution of
earthquake occurrence with respect to time must be
considered.
• The assumption of random occurrence allows the use
of simple probability models, but is inconsistent with
the implications of elastic rebound theory
Poisson Model
The Poisson model provides a simple frame work
for evaluating probabilities of events that follow a
Poisson process, one that yields values of a random
variable describing the number of occurrences of a
particular event during it given time interval or in a
specified spatial region.
Since PSHAs deal with temporal uncertainty, the spatial
applications of the Poisson model will not be considered further.
Poisson processes possess the following properties:

1. The number of occurrences in one time interval are

independent of the number that occur in any other time
interval.
2. The probability of occurrence during a very short time
interval is proportional to the length of the time interval.
3. The probability of more than one occurrence during a
very short time interval is negligible.
 • These properties indicate that the
events of a Poisson process occur
randomly, with no "memory" of the time,
size, or location of any preceding event.
For a Poisson process, the probability of a
random variable N, representing the
number of occurrences of a particular
event during a given time interval is given
by (EQ. 1)
• Where ụ is the average number of
occurrences of the event in that time
interval. The time between events in a
EQ. 1 Poisson process can be shown to be
exponentially distributed.
 • To characterize the temporal
distribution of earthquake
recurrence for PSHA purposes, the
Poisson probability is usually
expressed as (EQ. 2). Where is the
EQ. 2 average rate of occurrence of the
event and t is the time period of
interest
• Note that the probability of
occurrence of at least one event in a
period of time t is given by (EQ.3)

EQ. 3
 When the event of interest is the
exceedance of a particular
earthquake magnitude, the
Poisson model can be combined
with a suitable recurrence law to
predict the probability of at least
EQ.4 one exceedance in a period of t
years by the expression (EQ. 4)
Other Models
Elastic rebound theory suggests that the occurrence of
earthquakes on a particular fault or fault segment should
not be independent of past seismicity. If earthquakes occur
to release strain energy that builds up over extended
periods of time, the occurrence of a large earthquake
should substantially reduce the chances of another
independent, large earthquake (from the same source)
occurring shortly thereafter. If earthquakes are triggered
when the stress on a fault reaches some limiting value, the
chances of occurrence should depend on the times, sizes,
and locations of preceding events.
A number of models that account for prior seismicity
have been proposed :

•Nonhomogeneous Poisson models allow the annual rate of

exceedance to vary with time.
• Renewal models use arrival-time distributions other than
exponential (implied by the homogeneous Poisson model) to
allow the hazard rate to increase with time since the last
event. Gamma and Weibull distributions are most common.
• Time-predictable models specify a distribution of the time
to the next earthquake that depends on the magnitude of
the most recent earthquake.
• Slip-predictable models consider the distribution of
earthquake magnitude to depend on the time since the
most recent earthquake.
•Markov models incorporate a type of memory that
describes the chances that a process moves from some past
"state" to a particular future state. The time for which the
process stays in a particular state before moving to another
state is exponentially distributed.
• semi-Markov models are not restricted to the
exponential distribution. Both Markov models and
semi-Markov models have been used in seismic
hazard analysis
•The semi-Markov models for example, relate the
probability of future earthquakes of various sizes to
the size of the most recent event and the elapsed time
since its occurrence. Trigger models can account for
clusters of events (aftershocks) that occur after
triggering events.
Model Applicability
• Investigations of the applicability of Poisson and non-
Poissonian models have shown that the Poisson model is useful
for practical seismic risk analysis except when the seismic hazard
is dominated by a single source for which the time interval since
the previous significant event is greater than the average
interevent time and when the source displays strong
"characteristic-time" behavior. For this and other reasons related
to simplicity, ease of use, and lack of sufficient data to support
more sophisticated models, the Poisson model is the most
widely used in contemporary PSHA.
PROBABILITY COMPUTATION
• The results of a PSHA can be expressed in many different
ways. All involve some level of probabilistic computations to
combine the uncertainties in earthquake size, location,
frequency, and effects to estimate seismic hazards. A
common approach involves the development of seismic
hazard curves, which indicate the annual probability of
exceedance of different values of a selected ground motion
parameter. The seismic hazard curves can then be used to
compute the probability of exceeding the selected ground
motion parameter in a specified period of time.
Seismic Hazard Curves
• Seismic hazard curves can be obtained for individual source
zones and combined to express the aggregate hazard at a
particular site. The basic concept of the computations required
for development of seismic hazard curves is fairly simple. The
probability of exceeding a particular value, y*, of a ground
motion parameter, Y, is calculated for one possible earthquake at
one possible source location and then multiplied by the
probability that that particular magnitude earthquake would
occur at that particular location. The process is then repeated for
all possible magnitudes and locations with the probabilities of
each summed.
 • For a given earthquake occurrence,
the probability that a ground
motion parameter Y will exceed a
particular value y* can be
computed using the total
probability theorem, that is, (EQ.5)
where X is a vector of random
variables that influence Y. In most
cases the quantities in X are limited
EQ.5 to the magnitude, M, and distance,
R.
 • Assuming that M and R are
independent, the probability of
exceedance can be written as
(EQ.6).
• Where P[Y>y*1m,r] is obtained
from the predictive relationship
and fM(m) and fR(r) are the
probability density functions for
EQ.6 magnitude and distance,
respectively.
 • If the site of interest is in a region of Ns
potential earthquake sources, each of
which has an average rate of threshold
magnitude exceedance, the total average
exceedance rate for the region will be
given by EQ. 7. The individual
components of EQ.7 are, for virtually all
realistic PSHAs, sufficiently complicated
that the integrals cannot be evaluated
analytically. Numerical integration, which
can be performed by a variety of
EQ.7 different techniques, is therefore
required.
 • One approach, used here for
simplicity rather than
efficiency, is to divide the
possible ranges of magnitude
and distance into NM and NR
segments, respectively. The
average exceedance rate can
EQ.8 then be estimated by (EQ.8)
 • EQ.9 is then equivalent to EQ.
10.
• The accuracy of the crude
numerical integration
procedure described above
increases with increasing NM
and NR. More refined methods
EQ.10 of numerical integration will
provide greater accuracy at the
same values of NM and NR.
Example
The basic procedures of a typical PSHA can be illustrated for the
site shown in Figure 1-3 if the recurrence relationships for each of
the source zones is known. Assuming that the seismicity of the
respective source zones are described by:
the PSHA can be performed in the four previously described steps:
• (1) The problem statement provides the location, geometry, and maximum
magnitude of each source zone. The distribution of source-to-site distance
must also be characterized. To limit the number of computations involved in
this simple example, we will characterize the distribution of source-to-site
distance by a relatively coarse histogram. Consider first source zone I.
• It is a simple matter to show that the shortest possible source-to-site distance
will be 23.72 km and that the longest will be 90.12 km. We can divide this total
range into 10 distance intervals of length (90.12 km - 23.72 km)/IO = 6.64 km. If
we divide the source zone into a large number of segments of equal length, we
can characterize the distribution of source-to-site distance by determining how
many of the segments fall within each distance interval.
• For 1000 segments, the normalized histogram of source-to-site distance
is shown in Figure 1.a. The ordinates of the normalized histogram
represent the relative frequency that would be equal to the probability if
an infinite number of segments were used, but which is an approximation
to the probability in this case.
• The probability that the source-to site distance is between 23.72 and
30.36 km (or about equal to the midpoint of that range, 27.04 km) is
approximately 0.336. For source zone 2, the source-to-site range 25 to
125 km can be divided into 10 intervals of IO-km length; dividing the areal
source zone into 2500 elements of equal area, the normalized histogram
of Figure 1.b is obtained. Since there is only one possible source-to-site
distance, obtaining the normalized histogram of Figure 1.c for source zone
3 is trivial matter.
Figure 1 a—c : Approximations to source-to-site probability distributions for source zones
(a)l, (b)2, and (c)3.
• (2) The temporal distribution of earthquake recurrence can be
characterized using the recurrence relationships provided in the
problem statement. Assuming that earthquakes of magnitude less
than 4.0 do not contribute to the seismic hazard, the mean rates of
exceedance of magnitude 4.0 events from each of the source zones
are

giving Vtotal= 4.515. For each source zone, the probability that the magnitude
will be within an interval between a lower bound ml and an upper bound mu is
given by
Where fM(m) is given in equation (4.12). If NM= 10, the lowest magnitude
interval for source zone I will be from M= 4.0 to M= 4.33. The probability that
the magnitude would fall within that interval would be:

The probabilities of various magnitudes for each source zone are as shown in
Figure 2.d-f.
• (3) To compare the results of this PSHA with those from the DSHA
example, we will use the same predictive relationship: that is, the
Cornell et al. (1979) relationship

Uncertainty in this relationship is expressed by the standard deviation

• (4) Finally, we compute the total seismic hazard as the sum of the
contributions from each possible combination of source-to-site distance and
earthquake magnitude on each of the three source zones. First, we consider
source zone 1. For the lowest magnitude interval (j =1),

as computed in step 2. For the lowest distance interval (k = 1),

Figure 2.d-f: Approximations to magnitude probability
distributions for source zones (d)1 , (e)2, and (f)3
• as computed in step 1.This combination of magnitude and
distance indicates an expected value of In PHA of

•Now we can calculate the probabilities that various target peak

acceleration levels will be exceeded. For a* = O.Olg(9.81 gals), the
corresponding standard normal variable is

•Then the probability that the peak acceleration is greater than a.01g, using
Table 4-2, is
• Annual rate of exceedance of a peak acceleration of 0.01 an
earthquake of magnitude 4.165 at a distance of 27.04 km on source
zone 1 (given that an earthquake of M>m0 occurs on source zone 1)
will be :

• If the preceding calculations are repeated for the 99 other possible

combinations of magnitude and distance for source zone 1, the
contributions of each will be:
• Summing all of these contributions indicates that the mean annual
rate at which an acceleration of 0.01g will be exceeded by an
earthquake on source zone 1 will be 1.923. Repeating all of these
calculations for the other source zones yields equivalent exceedance
rates of 1.016 for source zone 2 and 0.005 for source zone 3.
Consequently, the probability that a target acceleration of 0.01g will
be exceeded by an earthquake of M>m0 on any of the three source
zones will be 1.923 + 1.016 +0.005 =2.944. This implies a return
period of 0.34 year for this low acceleration. By repeating this
process for different target accelerations, the seismic hazard curves
of Figure 3.g can be developed..
Figure 3.g Seismic hazard curves for source zones 1,2, and 3 and
total seismic hazard curve for all sources
Finite Time Periods
• The seismic hazard curve can easily be combined with the
Poisson model to estimate probabilities of exceedance in
finite time intervals. From equation, the probability of
exceedance of y* in a time period T is
Example
• Returning to Example 1, the probability that an acceleration of
0.10g would be exceeded in a 30-year period would be

• It is often necessary to compute the value of a ground motion parameter

corresponding to a particular probability of exceedance in a given time period.
For example, the acceleration level that has a 10% probability of exceedance in
a 50-year period would be that with an annual rate of exceedance, obtained by
rearranging equation (4.23), of

• From the total seismic hazard curve of Figure 4.18, that acceleration level
would be approximately 0.63g.
 • These types of analyses have been
performed for a variety of seismically
active areas within the United States.
As the exposure time, T, increases, the
probability of exceeding a particular
ground motion parameter value also
increases. Similarly, the value of a
ground motion parameter with a
particular probability of exceedance
increases with increasing exposure
time. Figure 4.1 illustrates the peak
acceleration with a 10% probability of
exceedance for a number of
metropolitan areas within the United
States. Seismic hazard maps, such as
Figure 4.1. Peak horizontal bedrock accelerations that shown in Figure 4.2, have been
with 10% probability of exceedance over various developed to express the relative
exposure times for 14 areas in North America. seismicity of different regions in
(After NEHRP,1993.) building codes
Deaggregation
• The PSHA procedures described in the preceding sections allow computation
of the mean annual rate of exceedance at a particular site based on the
aggregate risk from potential earthquakes of many different magnitudes
occurring at many different source-site distances. The rate of exceedance
computed in a PSHA, therefore, is not associated with any particular
earthquake magnitude or source-site distance.
• In some cases, however, it may be useful to estimate the most likely
earthquake magnitude and/or the most likely source-site distance. These
quantities may be used, for example, to select existing ground motion records
(recorded in earthquakes of similar magnitude at similar source-site distance)
for response analyses
• This process of deaggregation requires that the mean annual rate of
exceedance be expressed as a function of magnitude and/or distance.
Computationally, this simply involves the removal of terms from the
summations of Equation . For example, the mean annual rate of
exceedance can be expressed as a function of magnitude by

• Similarly, the mean annual rate of exceedance can be expressed as a function of

source-site distance by
Figure 4.2 Contours of mean horizontal acceleration on rock (expressed as a percentage of gravity) with I0%
probability of exceedance in 50 years.
• Finally, it is possible to compute the mean annual rate of exceedance as
functions of both earthquake magnitude and source-site distance, i.e.
Logic Tree Methods
• The probability computations described previously allow systematic
consideration of uncertainty in the values of the parameters of a
particular seismic hazard model. In some cases, however, the best
choices for elements of the seismic hazard model itself may not be
clear.
•The logic tree approach allows the use of alternative models, each of
which is assigned a weighting factor that is interpreted as the relative
likelihood of that model being correct. It consists of a series of nodes,
representing points at which models are specified and branches that
represent the different models specified at each node.
• The sum of the probabilities of all branches connected to a given node
must be 1. The simple logic tree shown in Figure 4.17 allows uncertainty
in selection of models for attenuation, magnitude distribution, and
maximum magnitude to be considered. In this logic tree, attenuation
according to the models of Campbell and Bozorgnia (1994) and Boore et
al. (1993) are considered equally likely to be correct, hence each is
assigned a relative likelihood of 0.5. Proceeding to the next level of nodes,
the Gutenberg-Richter magnitude distribution is considered to be 50%
more likely to be correct than the characteristic earthquake distribution.
• At the final level of nodes, different relative likelihoods are assigned to
the maximum magnitude. This logic tree terminates with a total of 2 x 2 x
3 = 12 (no. of attenuation models x no. of magnitude distributions x no. of
maximum magnitudes) branches
• The relative likelihood of the combination of models and/or
parameters implied by each terminal branch is given by the product of the
relative likelihood of the terminal branch and all prior branches leading to
it. Hence the relative likelihood of the combination of the Campbell
attenuation model, Gutenberg-Richter magnitude distribution, and
maximum magnitude of 7.5 is 0.5 x 0.6 x 0.3 = 0.09. The sum of the
relative likelihoods of the terminal branches, or of those at any prior level,
is equal to 1.
• To use the logic tree, a seismic hazard analysis is carried out for the
combination of models and/or parameters associated with each terminal
branch. The result of each analysis is weighted by the relative likelihood of
its combination of branches, with the final result taken as the sum of the
weighted individual results.
• It is easy to see that the required computational effort increases quickly
with increasing numbers of nodes and branches. Parameters best
characterized by continuous distributions (e.g., the maximum magnitude
in the example of Figure 5) are difficult to treat in the logic tree without
resorting to large numbers of branches. Nevertheless, the logic tree is a
very useful tool for the analysis of seismic hazards.
Figure 5. Simple logic tree for incorporation of model uncertainty.