TABLE OF CONTENTS

Page

B-2 Seismic Hazard Analysis ..................................................................... B-2-1

B-2.1 Key Concepts ....................................................................... B-2-1
B-2.2 Basics of Probabilistic Seismic Hazard Analysis ................ B-2-4
B-2.2.1 Seismic Source Characterization ....................... B-2-6
B-2.2.2 Seismic Hazard Curves .................................... B-2-14
B-2.2.3 Uniform Hazard Response Spectra .................. B-2-20
B-2.2.4 Ground Motion Time Histories........................ B-2-20
B-2.2.5 Conditional (Mean) Spectra ............................. B-2-25
B-2.2.6 Structural Considerations ................................. B-2-27
B-2.3 Surface Fault Displacement/Permanent Ground
Deformation ................................................................... B-2-28
B-2.3.1 Probabilistic Fault Displacement Hazard
Analysis...................................................... B-2-30
B-2.4 Relevant Case Histories and Examples.............................. B-2-32
B-2.5 References .......................................................................... B-2-32

Tables
Page

B-2-1 General Levels, Intended Use, Staffing, and Scope for

Seismic Hazard Analyses ........................................................... B-2-2
B-2-3 Table of Types of Geologic, Geophysical, and Seismologic
Data that Can Be Considered for Identifying and
Characterizing Seismic Sources ................................................. B-2-7

Figures
Page

B-2-1 Example fault source logic tree for a single fault, showing
fault parameters in boxes above each node, cases
considered for each node in the tree, and assigned weights
for each case ................................................................................. B-2-8
B-2-2 Example seismic source characterization logic tree, showing
GPME used for each seismic source, and subsequent fault
characterization nodes .................................................................. B-2-8

B-2-i
July 2019
Figures (continued)
Page

B-2-3 Map showing simplified traces of faults considered potential

seismic sources of significance to Terminus Dam, and
areal source zones used in background seismicity analysis. ...... B-2-11
B-2-4 Example map of seismotectonic zones in the CEUS and
earthquakes in the historical seismicity catalog ......................... B-2-13
B-2-5 Three recurrence models typically used in PSHA; M (x-axis)
is earthquake magnitude and Log N(m) (y-axis) is number
of events per year. ...................................................................... B-2-15
B-2-6 Example of mean and percentile hazard curves for PHA. ............. B-2-16
B-2-7 Seismic hazard curves for Success Dam in south-central
California, showing contributions to PGA cumulative hazard
from various local and regional fault sources and areal
source zone................................................................................. B-2-17
B-2-8 eismic hazard curves for Hills Creek Dam in central Oregon,
showing Annual Exceedance Probabilities for PGA and
two spectral frequencies............................................................. B-2-18
B-2-9 Seismic hazard de-aggregation plots showing contributions to
mean PGA hazard for each magnitude-distance pair for
Hills Creek Dam in central Oregon, for return periods of:
a) 144 years, b) 1,000 years, and c) 5,000 years. ....................... B-2-19
B-2-10 Mean UHRS from all sources for Wappapello Dam, Missouri,
for return periods of 225 years (20% in 50 yr), 475 years
(10% in 50 yr), 975 years (5% in 50 yr), 2,475 years
(2% in 50 yr), and 5,000 years (1% in 50 yr). ........................... B-2-21
B-2-11 Plots showing differences in V/H ratios with site class and
distance from the seismic source. .............................................. B-2-22
B-2-12 Mean hazard curves for average horizontal component
(a) PGA, (b) spectral acceleration = 0.2 s, (c) spectral
acceleration = 1.0 s, and (d) spectral acceleration = 3.0 s;
before applying site amplification factors.................................. B-2-24
B-2-13 Horizontal UHRS and CMS for T = 0.2 s and T = 1.0 s
for a 2,475-year return period, compared with median and
84th percentile for MW 6.0 and MW 7.5 deterministic spectra ... B-2-27
B-2-14 Example CMS for Green Peter Dam generated using the 2008
USGS de-aggregation tool for probability of exceedance of
2% in 50 yr. ................................................................................ B-2-28

B-2-ii
July 2019
B-2 SEISMIC HAZARD ANALYSIS
B-2.1 Key Concepts
A critical component of risk analyses is the degree of hazard imposed upon the
system of interest, which in this case includes dam and levee facilities that
comprise an infrastructure portfolio. Hazard is distinct from the response of
system components (which are affected by hazard) and from the consequences
(which result from distress or failure of system components); risk is a measure of
how hazard affects system response and possible resultant consequences. For risk
assessments of dam and levee systems, hazard is generally treated in the form of
three types of loading (static, hydrologic, and seismic loading).

This chapter summarizes the Best Practices for characterizing seismic hazard for
dam and levee safety evaluations. Seismic hazard is often 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. These effects may be severe and damaging,
or they may simply be a nuisance, depending on the magnitude of the earthquake,
the distance a site is from the earthquake, local site conditions, and the response
of the system of interest (e.g., dams, levees, buildings, lifelines, power plants).
Also, the term “seismic hazard” in engineering practice can refer specifically to
strong ground motions produced by earthquakes that could affect engineered
structures, such that seismic hazard analysis often refers to the estimation of
earthquake-induced ground motions having specific probabilities over a given
time period. In this section, seismic source characterization, ground shaking, and
the development of ground motions for engineering analyses are summarized. In
addition, where the potential for surface fault displacement through a dam site is a
concern, seismic hazard analysis should include characterization of the location,
amount, style, and recurrence of permanent ground deformation (PGD). While
characterizing surface rupture hazard differs substantially from assessing strong
ground motion hazard, the analytical approaches used to evaluate these two
seismic hazards have some similarities.

Overall, seismic hazard analyses can be grouped into three general levels
depending on the type of study, the risk analysis being performed, and the
intended use of the study. Table B-2-1 provides an outline of the scope and
staffing guidelines for three levels of seismic hazard analysis. Unless a detailed
and up-to-date seismic hazard analysis has been completed for the site(s) of
interest, a first step in developing defensible seismic loading parameters is to
conduct an initial, semi-quantitative screening analysis. For seismic loading, a
valid initial screening effort can use available online tools developed by the
United States Geological Survey (USGS), accessed at http://geohazards.usgs.
gov/hazardtool/application.php.

B-2-1
July 2019
 Chapter B-2 Seismic Hazard Analysis

Table B-2-1.—General Levels, Intended Use, Staffing, and Scope for Seismic
Hazard Analyses
General Staffing
Requirements and
Analysis Level Cost Categories General Scope
Screening or scoping level 1 month Compile and review existing files
(periodic assessment, semi- $ and recent literature
quantitative risk assessment)
Interview local experts and regional
or national subject matter experts

Use existing analyses if judged

applicable

Develop hazard curves based on

existing, readily available seismic
source models and existing
accepted Ground Motion Prediction
Equations (GMPE)
Intermediate level 1 month to 1 year Limited field reconnaissance
(Issue evaluation study) $$
feasibility design Update seismic source model

Produce hazard curves based on

updated seismic source models and
accepted GMPE

Develop Uniform Hazard Response

Spectra (UHRS) and time histories
to assess system response or to
inform design parameters.
Detailed characterization level >1 year Detailed fault-specific seismic
(dam safety modification study, $$$ source characterization
engineering design)
appraisal or conceptual Develop hazard curves, UHRS,
design; corrective action; Conditional Mean Spectra (CMS),
final design and time histories to inform design
parameters; Consider possible
seismic monitoring.

Note that the USGS website tool is not intended to provide ground motions for
use in evaluating critical structures such as dams, nuclear power plants, or
other high-risk facilities. The USGS Web site can be used as a screening tool
for prioritizing facilities on a regional basis, if the user understands and
acknowledges that the tool has limitations related to its regional, non-site-specific
methodology. It must be recognized that using ground motions derived from the
USGS website tool is not recommended for critical decisions involving
engineering design of modifications to a dam structure.

B-2-2
July 2019
 Chapter B-2 Seismic Hazard Analysis

Various screening protocols can be developed based on screening criteria that are
customized to a specific region or location, utilizing information readily available
from the USGS website noted above in conjunction with other regional geologic
and geotechnical information. Initial seismic hazard screening may include an
initial estimate of the geotechnical seismic site-response conditions estimated
based on site-specific information (preferred, if available), by review of available
site geologic maps and aerial photography.

One key parameter in estimating the site response is the average shear wave
velocity in the upper 30 m (100 ft), or Vs30 (table B-2-2). This parameter which
can be estimated using the regional estimated Vs30 mapping tool developed by the
USGS: http://earthquake.usgs.gov/hazards/apps/vs30/.

Table B-2-2.—Site Class Definitions

Site Class Generalized Description Vs100 (ft/s) Vs30 (m/s)
A Hard rock >5,000 ft/s >1,520 m/s
B Rock 2,500 to 5,000 ft/s 760 to 1520 m/s
C Very dense soil/soft rock 1,200 to 2,500 ft/s 360 to 760 m/s
D Stiff soil 600 to 1,200 ft/s 180 to 360 m/s
E Soft clay soil <600 ft/s <180 m/s
F Requires site response analysis

Using the estimated site class, the USGS online ground-motion hazard tool can be
used to obtain an estimated peak ground acceleration (PGA) with a mean annual
frequency of 1/2,475 (2 percent (%) probability of exceedance in 50 years).
Using pre-selected criteria to screen facilities in various site classes and
PGA values, a decision can be made to proceed with additional seismic hazard
analyses, such as utilizing existing recent seismic hazard analyses from nearby
sites or completing a site-specific seismic hazard analysis. Using the USGS tools,
ground motions can be estimated for sites across the nation assuming a site class
B/C boundary (Vs30 = 760 m/s); the USGS website allows estimation of shaking
levels for other site classes at locations where appropriate ground motions
prediction equations are available. Also, the values at the B/C boundary can be
estimated for other site classes using the correction procedures outlined in
ASCE 7.

The following sections provide an overview of the probabilistic and deterministic

seismic hazard analyses, followed by summaries of analytical outputs and
parameters needed for evaluating seismic loading. As noted above, seismic
hazards also include surface fault rupture; if applicable to the subject site, a

B-2-3
July 2019
 Chapter B-2 Seismic Hazard Analysis

Probabilistic Fault Displacement Hazard Analysis (PFDHA) may also be

necessary. Latter sections of this chapter summarize the basic elements of a
PFDHA.

B-2.2 Basics of Probabilistic Seismic Hazard

Analysis
A Probabilistic Seismic Hazard Analysis (PSHA) collectively considers the
contributions from all known potential sources of earthquake shaking. Most
importantly, a PSHA considers the likelihood of various earthquakes from
multiple potential seismic sources, each having a range of uncertainty in source
characteristics (e.g., rupture length, distance, fault dip, maximum magnitude, slip
rate). Uncertainty is treated explicitly, and the annual probability of exceeding
specified ground motions (commonly expressed as spectral acceleration at periods
of interest) are computed. A PSHA thus provides seismic loading parameters
over the full range of potential loading time intervals, and provides a complete
consideration of site seismic hazard from multiple sources and for appropriate
recurrence intervals.

Prior to the acceptance and incorporation of PSHA into standard hazard

assessment methodologies, most seismic hazard assessments were completed
using scenario-based, “deterministic” analyses. A Deterministic Seismic Hazard
Analysis (DSHA) typically assigns a maximum earthquake magnitude for a
particular seismic source, often referred to as the Maximum Credible Earthquake
(MCE). Based on the minimum distance from the site to the fault source, the
level of ground shaking at the site is estimated. Outdated DSHA methodologies
typically involved a simple judgment by experts as to the largest earthquake that a
significant source could produce. The earthquake identified as the MCE for a
given source was acknowledged as being a judgment based on the level of
information available to the hazard assessment team; the MCE sometimes is
simply (and incorrectly) chosen as the largest magnitude on the nearest major
fault.

In current practice, the MCE magnitude for a given seismic source is still a
judgment, but the ground motion estimation is informed by empirical
relationships between magnitude and distance based on local or worldwide
historical earthquakes, which allows a better assessment of uncertainty in the
ground motion values. Also, multiple seismogenic sources near and far from the
site can be considered. Nearby faults may be less active and limited to generating
lower magnitude events than more distant but larger faults, although the motions
from those events may yield higher expected PGA and higher short period/high
frequency energy content. Because of the attenuation of energy as waves
travel through Earth’s crust, distant faults that can produce larger magnitude
earthquakes may result in lower expected PGA values and lower high-frequency

B-2-4
July 2019
 Chapter B-2 Seismic Hazard Analysis

energy at a given site than nearby fault sources. However, large-magnitude

earthquakes on distant faults may still result in substantial damage because of
long durations of strong ground shaking and, in some cases, long-period
wavelengths. This depends in large degree on the type of structure being
analyzed. Very stiff structures such as concrete dams may be controlled by
acceleration (high frequency content), while very flexible structures such as
buildings (depending on the height and construction) may be controlled by
displacement (low frequency content). Intermediate structures will be
controlled by intermediate frequency content, or ground velocity. The different
ground-motion periods produced by different seismic sources emphasizes the
need for the hazard evaluation to include collaboration among geologists,
seismologists and structural engineers. In many tectonic settings, several different
fault sources will be considered as a source of the MCE ground motions required
for analysis, to account for variations in PGA, spectral content and duration
effects. For example, areas of plate convergence (such as the Pacific Northwest
of North America) may be affected by distant large-magnitude subduction-zone
earthquakes, as well as nearby moderate-magnitude earthquakes generated by
faults in the shallow part of Earth’s crust.

Note that the MCE (as defined above) classically refers to the largest earthquake
magnitude judged to be possible from a given seismic source, although
present-day structural engineering and building codes utilize the term Maximum
Considered Earthquake (also MCE), which represents a ground motion
acceleration (not an earthquake magnitude) with a certain probability of
exceedance in a given time period (usually 2 percent in 50 years; see Luco et al.
2007). When most seismic engineering applications and building codes refer to
the “MCE,” they are referring to the maximum considered earthquake ground
motion value rather than the earthquake magnitude.

The accepted and now commonly utilized PSHA approach is based on the model
developed by Cornell (1968), refined by McGuire (1974, 1978), summarized by
the Senior Seismic Hazard Analysis Committee (SSHAC 2007) and the
U.S. Nuclear Regulatory Commission (USNRC 2007), and utilized by many
subsequent analysts. The objective of a PSHA is to estimate the probability in a
given future time period that a specified level of some ground motion parameter
will be exceeded at a site of interest (SSHAC 2007). The analysis requires
characterization of all known earthquake sources that could affect the site,
including faults (“line sources”) and areas of seismicity (“areal sources”). The
analysis also considers the attenuation of seismic energy as it emanates from the
earthquake hypocenter to the site of interest, which is evaluated through the use of
empirical Ground Motion Prediction Equations (GMPE). Analysis of hazard
must also consider the geotechnical response of the site to ground motion input,
known as “site response.”

Products from a seismic hazard analysis typically consist of four basic

components related to strong ground motions and will be discussed below:

B-2-5
July 2019
 Chapter B-2 Seismic Hazard Analysis

1) seismic source characterization; 2) development of hazard curves (including

GMPE and site response); 3) development of uniform hazard spectra; and
4) development of acceleration time-histories. Lastly, this section provides
summary comments on structural considerations that may be relevant because of
seismic strong ground motions.

B-2.2.1 Seismic Source Characterization

Seismic source characterization involves the identification and documentation of
relevant possible earthquake sources in a region, and forms the knowledge base
for a site-specific or regional PSHA. As noted by SSHAC (2007) and the Central
and Eastern United States Seismic Source Characterization (CEUS-SSC 2012), a
proper and complete seismic source characterization includes an evaluation of the
complete set of data, models and methods that are relevant to the hazard analysis,
and the integration of the center, body, and range of technically defensible
interpretations. At present, the CEUS-SSC (2012) effort provides a
comprehensive seismic source characterization of the Central and Eastern
United States (CEUS, east of longitude 105W degrees), which was developed for
seismic hazard analyses of nuclear facilities. This effort provides a high level of
confidence in the data, models and methods of the technical community for CEUS
PSHA, and establishes guidance for seismic source characterizations in the
Western United States (WUS; west of longitude 105W degrees) and elsewhere
throughout the world. Similarly, a recent comprehensive seismic source
characterization incorporating a broad range of technical interpretations and
uncertainty has been completed for the entire State of California (USGS 2013).

The heart of any seismic source characterization for a PSHA is the description of
the future spatial and temporal distribution of earthquakes (CEUS-SSC 2012).
The seismic source characterization results in a seismic source model, which is a
description of the magnitude, location, and timing of all earthquakes (usually
limited to those that pose a significant threat); source models consist of several
earthquake scenarios, each with its own magnitude, location, and rate. These
scenarios usually represent earthquakes on particular faults or within a region
(i.e., volume of earth’s crust) that have a distribution of events in time and space.
Thus, earthquake source models typically consist of possible earthquakes that can
occur on either: 1) particular faults; or 2) areal (i.e., background) seismic source
zones. These two types of seismic sources are summarized below.

Fault sources are usually modeled as planar surfaces and are described with a set
= of parameters such as activity (expressed by either slip rate or recurrence
interval), geometry (location, length, dip, and down-dip extent), amount and
sense of coseismic slip, maximum magnitude (Mmax), and recurrence model
(characteristic earthquake or maximum magnitude and associated frequencies).
These parameters are generally defined on the basis of geologic, geomorphologic,
and paleoseismologic information, as summarized on table B-2-3. The level of

B-2-6
July 2019
 Chapter B-2 Seismic Hazard Analysis

Table B-2-3.—Table of Types of Geologic, Geophysical, and Seismologic Data that can be
Considered for Identifying and Characterizing Seismic Sources (CEUS-SSC 2012)
Seismic Source
Area/Volume
Individual Faults Sources
Data Type Location Activity Length1 Dip Depth Style Area Depth
Geological/Remote Sensing
Detailed mapping X X X X X
Geomorphic data X X X X X
Quaternary surface rupture X X X X
Fault trenching data X X X X
Paleoliquefaction data X X X
Borehole data X X X X
Aerial photography – X X X
Low sun-angle photography X X X
Satellite imagery X X X
Regional structure X X X X
Balanced cross section X X X X
Geophysical/Geodetic
Regional potential field data X X X X
Local potential field data X X X X X
High resolution reflection data X X X X
Standard reflection data X X X
Deep crustal reflection data X X X X X
Tectonic geodetic/stain data X X X X X X X
Regional stress data X X
Seismological
Reflected crustal phase data X
Pre-instrumental earthquake data X X X X X
Teleseismic earthquake data X
Regional network seismicity data X X X X X X X
Local network seismicity data X X X X X X
Focal mechanism data X X
1 Fault length includes both total fault length and information on segmentation.

information for each fault as a possible seismic source varies according to the
degree of past investigation and regional or local field conditions, such that the
uncertainties in seismic source parameters needed for a PSHA can vary
substantially. To capture and document the range in uncertainties, the data
and interpretations of a seismic source model are depicted on a logic tree.
Figures B-2-1 and B-2-2 show examples of logic trees for seismic source
characterizations from the WUS (for United States Army Corps of Engineers
(USACE) Terminus Dam, California) and from the CEUS (Meers fault in
south-central Oklahoma). In the more tectonically active WUS, the recognition of

B-2-7
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-1.—Example fault source logic tree for a single fault,

showing fault parameters in boxes above each node, cases considered
for each node in the tree, and assigned weights (in parentheses) for
each case. Nodes are repeated for entire logic tree, but shown only
once on figure (example is Meers fault, Oklahoma [CEUS-SSC 2012]).

Figure B-2-2.—Example seismic source characterization logic tree,

showing GPME used for each seismic source (listed on chart without a
nodal circle), and subsequent fault characterization nodes. The source
geometry node consists of several individual nodes similar to logic tree
on figure B-2-1.

B-2-8
July 2019
 Chapter B-2 Seismic Hazard Analysis

active faults and the collection of detailed paleoseismologic data have allowed
inclusion of fault-specific sources and background seismicity sources zones in the
seismic source models. In contrast, the relative dearth of identified fault sources
in the CEUS has resulted in the use of spatial and temporal patterns of historical
small- to moderate-magnitude earthquakes to provide information about the
spatial and temporal pattern of future large-magnitude earthquakes.

Many definitions exist for what constitutes an “active” or “potentially active”

fault and hence what faults should be considered potential seismic sources. In a
PSHA, most faults with evidence of late Quaternary activity (i.e., faults with
documented or suspected evidence of displacement during roughly the past
1.6 million years are considered to be active; this criteria has been used in the
development of the USGS Quaternary Fault and Fold Database, available here:
http://earthquake.usgs.gov/hazards/qfaults/.

For most seismic hazard analyses in the WUS, faults within approximately 50 km
of the site are characterized, but for some sites that do not have nearby active or
potentially active source, the faults as far as 100 km or more are included if they
may have a significant impact on the hazard. In the WUS, large earthquakes
associated with the Cascadia Subduction zone in the region offshore Washington,
Oregon, and northern California will probably produce strong long-period ground
motions (and perhaps trigger secondary surface fault rupture) within a large part
of the Pacific Northwest. However, local earthquakes generated along faults in
the shallow crust above the subduction zone may also generate strong ground
motions at sites in the Pacific Northwest, although these ground motions may
consist of high-frequency motions and be shorter in overall duration. Similarly, a
large magnitude earthquake on the San Andreas fault in California will likely
generate strong long-period ground motions in areas more than 50 km from the
actual fault trace. In summary, a distant seismic source may be the primary
source of long-period spectral accelerations, whereas local fault sources or
background seismicity may be primary contributors to short-period ground
motions; as a result, both PSHA and DSHA should consider multiple distant
sources so that the evaluation includes all possible contributors to hazard.

Developing a complete and defensible seismic source model requires considerable

resources and continued efforts to maintain an updated database. During the
past decade or so, the USGS has compiled an extensive database of known
or suspected Quaternary faults and folds, which can be accessed at
http://earthquake.usgs.gov/hazards/qfaults/.

This compilation included contributions from USGS scientists, other Government

agencies, State geological surveys, academic institutions, and private industry

B-2-9
July 2019
 Chapter B-2 Seismic Hazard Analysis

experts. Most of the faults and folds in this database are located in the WUS,
primarily because of the difficulty in characterizing individual seismic sources in
the CEUS.

The USGS quaternary fault and fold database was compiled to help develop the
national USGS National Seismic Hazard Map Web site and does not include
many faults with low slip rates or those that have been more recently identified or
characterized. Also, the quality of data and level of study varies greatly for faults
in the database, particularly for faults with low slip rates or outside densely
populated areas. This is relevant because seismic hazard for dams and other
important facilities may be strongly influenced by nearby faults with fairly low
slip rates (~0.01 mm/year). As a result, seismic source characterizations for
specific critical facilities should review available geologic information of the site
region and strive to develop a complete source model that may include faults
presently not in the USGS Quaternary Fault and Fold Database for the United
States. Figure B-2-3 is an example map showing a regional earthquake source
model, including both fault-specific and areal sources.

For the State of California, a recent comprehensive compilation of fault sources

has been developed specifically for the Unified California Earthquake Rupture
Forecast, version 3 (UCERF3 USGS 2013). This effort is summarized in
UCERF3: A New Earthquake Forecast for California’s Complex Fault System
(USGS).

The UCERF3 fault database and supplemental material can be accessed at USGS
Quaternary Fault and Fold Database for the United States (USGS).

The UCERF3 effort provides the most comprehensive and up-to-date compilation
of fault characteristics and earthquake occurrence models for California, and can
be readily incorporated into PSHA and DSHA. A key difference between the
UCERF3 approach and previous regional seismic source characterizations is the
removal of fault segmentation as a condition on earthquake rupture lengths, which
was warranted by an anomalous rate of moderate to large earthquakes predicted
by detailed fault segmentation models. The newly minted California source
model depicted by UCERF3 will most likely form the basis for future DSHA and
PSHA throughout the State, and may help spawn similar data compilation efforts
in nearby WUS.

For areas outside of California and west of longitude 105W degrees, the USGS
fault and fold database can be utilized as an initial database for fault sources for
regional PSHA and DSHA. However, site-specific seismic hazard analyses at this
time still require development of seismic source models that are customized to the
site of interest and the level of investigation warranted by specific project goals.
In this portion of the WUS (including all or part of Washington, Oregon, Idaho,
Montana, Wyoming, Nevada, Utah, Colorado, Arizona, and New Mexico), there

B-2-10
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-3.—Map showing simplified traces (bold red lines) of faults considered
potential seismic sources of significance to Terminus Dam, and areal source zones
(outlined in green) used in background seismicity analysis (USGS).

have been many detailed and valuable seismic source characterizations (and
PSHA) completed by or for Reclamation, US Department of Energy, Bonneville
Power Authority, other Federal agencies, State agencies, and private utilities.
Some of these analyses are available publicly and may be valuable resources for
completing seismic hazard analyses at a given site. Several private consulting
companies also hold proprietary seismic source models compiled through years of
experience, and in certain areas these may be the most comprehensive data sets

B-2-11
July 2019
 Chapter B-2 Seismic Hazard Analysis

available for site-specific analyses. It is prudent that initial phases of a seismic

hazard analysis (see table B-2-1) to obtain any potential existing sources of
up-to-date seismic source models and seismic hazard analyses.

In the CEUS, there are far fewer known or suspected fault-specific source zones
than in the WUS, but considerable effort has also recently been completed to
develop a defensible database. CEUS-SSC (2012) summarizes the fault-source
model for the United States east of longitude 105W degrees.

This database is the most comprehensive and up-to-date compilation of fault

characteristics and earthquake occurrence models for the CEUS, and can be
readily incorporated into PSHA and DSHA. In addition to fault-specific sources,
seismic source characterization for PSHA must include areal seismic sources, or
possible seismogenic features for which there is insufficient knowledge to
attribute earthquakes to a specific fault source. In regions such as the CEUS
where the past occurrence of large earthquakes is not directly attributable to
specific causative faults in Earth’s crust, the science of seismology has relied
heavily on the locations, magnitudes, and rates of historical seismicity to aid
seismic hazard assessment. A key criterion is the degree of “uniform” earthquake
potential or characteristics, essentially using areas (actually, volumes) of Earth’s
crust that have had consistent rates, locations and sizes of past earthquakes to
interpret the characteristics of future earthquakes. BC Hydro (2008), for example,
defines a seismic source as “a volume of the earth’s crust that has the same
earthquake potential as defined by the size of events that may be generated.” For
the purposes of assessing seismic loading for USACE dams and levees in the
WUS, the consideration of areal seismic sources should be consistent with the
identification of areal sources by the USGS website (noted above) based on
the historical rate and pattern of seismicity and the location of regional
seismotectonic zones. In the CEUS, areal source zones can be those identified by
the USGS Web site or CEUS-SSC (2012). In this way, the hazard assessment
includes possible sources of strong ground motions that are not clearly known to
be associated with specific faults or other sources of repeatable large-magnitude
earthquakes.

For areal source zones, the parameters of interest are maximum magnitude,
associated rate of activity, and recurrence model (typically exponentially
truncated Gutenberg-Richter relationship, or just truncated exponential). The
areal (or “background seismicity”) source zones are characterized using the
locations, magnitudes, and rates of historical seismicity in a region. The analysis
of background seismicity considers several characteristics of the historical
earthquake catalog, and requires processing of the raw historical data on
earthquake magnitude, depth, timing, and epicentral locations. A key part of this
analysis is the identification of regional or local seismotectonic domains that,
based on geologic, geophysical, and geodetic information, may be interpreted to
have relatively consistent spatial and temporal variations in historical seismicity.

B-2-12
July 2019
 Chapter B-2 Seismic Hazard Analysis

Areal source zones are best depicted on a regional seismotectonic map figure; an
example map of regional areal source zones in the WUS is given on figure B-2-3.
In the WUS, the maximum size of the background earthquake is usually M 6 to 7
because events larger than this usually produce recognizable fault or fold-related
features at Earth’s surface. The regional seismotectonic domains identified
throughout the CEUS are plotted on figure B-2-4, which also demonstrates that
the historical seismicity is not uniformly distributed within the seismotectonic
zones (CEUS-SSC 2012). The characterization of seismicity patterns and rates in
the CEUS is an important component of seismic hazard analysis, primarily
because extensive searches have yielded very little definitive evidence of active,
causative faults. Even in the WUS where local fault sources may dominate
seismic hazard, the contribution of background seismicity to total seismic hazard
can be substantial or dominant.

Figure B-2-4.—Example map of seismotectonic zones in the CEUS (black polygons) and
earthquakes in the historical seismicity catalog (grey circles) (CEUS-SSC 2012).

Key parameters within a historical seismicity catalog that need to be addressed for
a PSHA include:

• Catalog temporal and spatial completeness (has varied by region through

history)

• Magnitude detection thresholds (have varied by region through history)

B-2-13
July 2019
 Chapter B-2 Seismic Hazard Analysis

• Location accuracy (both epicentral and hypocentral)

• Consistent magnitude scale (Moment magnitude now preferred)

• Declustering to remove foreshocks and aftershocks

• Removal of non-tectonic events (e.g., human-induced seismicity,

explosions)

• Overall rate of seismicity, as defined by the y-intercept of the recurrence

line (“a-value”)

• Magnitude-frequency decay, as defined by the slope of the recurrence line

( “b-value”)

• Spatial smoothing of b-values within a region

In general, historical seismicity constrains the low-magnitude, high-frequency

part of recurrence relationships for given areal source zones. The part of the
recurrence relationships derived from historical seismicity are generally viewed as
a Gutenburg-Richter process, which is then truncated at higher magnitudes
(figure B-2-5); the high-magnitude part of the recurrence relationship is often
constrained by paleoseismic information (most often in the WUS) and may take
the form of a “characteristic earthquake model” or a “maximum magnitude
model” (figure B-2-5). However, in cases where a large earthquake has occurred
within an areal source zone during historical time and is not attributed to a
specific seismic source, then the historical record provides critical data for the
low-frequency recurrence relationships. The best example of this condition is the
Charleston, South Carolina, seismic zone, which produced an ML7.3 earthquake
in 1886 but has yet to be associated with a causative fault despite decades of
geologic and geophysical investigations.

B-2.2.2 Seismic Hazard Curves

The basic product of a PSHA is a series of seismic hazard curves that show the
annual rate or probability at which a specific ground motion level will be
exceeded at the site of interest. Hazard curves typically have “annual probability
of exceedence” or its reciprocal, “return period”, on the vertical axis on a
logarithmic scale, and PGA (usually expressed in terms gravity, or “g”) on the
horizontal axis on an arithmetic scale. Hazard curves may also depict another
measure of seismic loading, such as the response spectra acceleration at a given
period of vibration, on the horizontal axis. The seismic hazard curve is the most
important and most commonly used screening tool in analyzing hazard and risk.

B-2-14
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-5.—Three recurrence models typically used in PSHA; M (x-axis)

is earthquake magnitude and Log N(m) (y-axis) is number of events per
year.

A hazard curve is developed for each seismic source, and these individual curves
are added to develop the cumulative hazard curve for the given site. The total rate
at which a given ground motion level is exceeded is the sum of the rates for these
individual sources. Seismic hazard curves are produced using one of several
available computer programs, including the USGS Unified Hazard Tool.

The 2014 version of the USGS hazard application improves upon the previous
2002 and 2008 versions and has been revised to include the new Next
Generation Attenuation (NGA) models for the CEUS (NGA-East) and the WUS
(NGA-West2). These GMPE were developed as part of the NGA Project
sponsored by the Pacific Earthquake Engineering Research (PEER) Center
(Lokke 2013).

The seismic hazard curves are calculated using ground motion attenuation
relations that relate PGA or spectral acceleration (SA) to the distance between the
seismic source and site, and the magnitude of the earthquake associated with the
source. Site conditions are very important, and include the footwall verses
hanging wall location, rupture direction, and foundation characteristics (i.e., “soil”
verses “rock”). Most of the new NGA GMPE incorporate the input parameter
VS30.

Most dam and levee safety studies include several types of hazard curves,
depending on the level of study and the type of dam (embankment versus
concrete). Thus, it is important to communicate with the seismic hazard analysis
team so that the hazard evaluation includes the parameters most likely to affect

B-2-15
July 2019
 Chapter B-2 Seismic Hazard Analysis

the dynamic response of the dam. A common curve is one showing mean and
percentile values for PGA (or interchangeably, peak horizontal acceleration
(PHA), also in g), as exemplified by figure B-2-6.

Figure B-2-6.—Example of mean and percentile hazard curves for PHA.

Hazard curves can be developed that show the contribution from various
earthquake sources (figures B-2-7 and B-2-8), which help select appropriate time
histories for use in the engineering analyses because they help interpret which
type of seismic source controls hazard at various spectral accelerations. For
example, hazard curves from a nearby moderate magnitude, strike-slip earthquake
source will differ from those of a fairly distant, large-magnitude subduction zone
earthquake. Also, the identification of the source or sources controlling the
hazard is important because this knowledge can focus additional studies on the
sources with the most significance to the dam. For example, a nearby fault
thought to have a large contribution to hazard (i.e., based on an estimated high
slip rate) may be a good target for additional investigation. Subsequent detailed
geologic mapping or paleoseismic trenching may demonstrate a lower slip rate
and thus indicate that the source contributes less to the total hazard.

A similar tool for identifying likely major contributors to seismic hazard is a

seismic hazard de-aggregation plot, which helps identify the magnitudes and
distances of controlling seismic sources. A seismic hazard curve combines all
sources (magnitudes and distances) to define the probability of exceeding a given

B-2-16
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-7.—Seismic hazard curves for Success Dam in south-central California,

showing contributions to PGA cumulative hazard from various local and regional
fault sources and areal source zone.

B-2-17
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-8.—Seismic hazard curves for Hills Creek Dam in central Oregon, showing Annual
Exceedance Probabilities (AEP) for PGA and two spectral frequencies. The plots show the
contributions to various spectral acceleration cumulative hazard from the Cascadia Subduction
Zone (CSZ) source, the Juan de Fuca intra-slab source, and shallow crustal fault sources. Note
that the CSZ dominates hazard for all three spectral accelerations, but that relative contributions
from other sources vary for different spectral accelerations.

ground motion level. Because all of the sources, magnitudes, and distances are
combined, it can be difficult to develop an intuitive understanding about the
sources that control the hazard based on just the hazard curve (Hashash et al.
2015). De-aggregation provides the relative contributions to seismic hazard from
seismic sources in terms of earthquake magnitude, source-to-site distance, and
epsilon (ε, a measure of ground motion uncertainty) (Bazzurro and Cornell 1999).
Figure B-2-9 shows three de-aggregation plots for the same site given different
return periods. For this site, the primary contributors to hazard at an Annual
Exceedance Probability (AEP) of 1/144 are moderate-magnitude sources less than
about 100 km away and a large-magnitude source more than about 100 km away.
For AEP of 1/1,000 and 1/5,000, the primary hazard contributor is the
large-magnitude source more than about 100 km from the site. These plots
suggest that additional information on the local, low slip-rate (but
large-magnitude) faults close the site may serve to improve the hazard
characterization.

The de-aggregation plots also serve to inform subsequent seismic analyses, in

particular the development of deterministic hazard results (i.e., DSHA) and of
Conditional Mean Spectra (CMS, addressed in a later section). If deterministic

B-2-18
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-9.—Seismic hazard de-aggregation plots

showing contributions to mean PGA hazard for each
magnitude-distance pair for Hills Creek Dam in
central Oregon, for return periods of: a) 144 years,
b) 1,000 years, and c) 5,000 years.

B-2-19
July 2019
 Chapter B-2 Seismic Hazard Analysis

ground-motion parameters are needed for risk-evaluation or design purposes,

these can be developed using scenario earthquakes that are shown to be
representative based on the modal values of magnitude and distance
from de-aggregation results. The modal values of specific scenario
earthquakes are generated by the 2008 USGS deaggregation Web site
(https://earthquake.usgs.gov/hazards/interactive/ for certain return periods
(including AEP of 1/975 and 1/2,475). For example, Hashash et al. (2015) show
that the median deterministic spectrum for the local earthquake scenario is similar
to the 2,475-year UHRS at short spectral periods, and the 84th percentile
deterministic spectrum for the New Madrid Seismic Zone (NMSZ) earthquake
scenario is similar to the 2,475-year UHRS at long spectral periods.

B-2.2.3 Uniform Hazard Response Spectra

UHRS are developed during the PSHA by developing hazard curves (i.e., spectral
acceleration versus exceedance probability) for several vibration periods, to
define the response spectra. Then, for a given exceedance probability or return
period, the ordinates are taken from the hazard curves for each spectral
acceleration, and an “equal hazard” response spectrum is generated for the same
return period. The response spectra curves are generated for specified AEP of
interest, as chosen by the analyst according to engineering or design needs for the
project. The USGS National Seismic Hazard Map website provides UHRS using
the regional seismic source model.

An example of the UHRS generated using this online application is given on

figure B-2-10 for Wappapello Dam in eastern Missouri.

B-2.2.4 Ground Motion Time Histories

Conventional design practice uses the UHRS obtained from PSHA at selected
probabilities or return periods. For cases in which time series analysis is needed,
ground motion time histories matched to the target UHRS are used. When
selecting these motions, it is necessary to associate them with a source and event
magnitude to properly represent the duration, amplitude, and frequency content of
the anticipated ground motion. In other words, the ground motion time histories
most appropriate for analysis of structural response are those that represent
the ground motions that are likely to impact the site from primary hazard
contributors. If the ground motions (for a selected return period) are dominated
from, say, a specific fault source (e.g., the Reelfoot fault in the NMSZ), then the
analysis should utilize seed time histories that best represent the hazard-dominant
earthquake from that source. The corresponding earthquake magnitude is also
required for liquefaction evaluation.

For facilities considered to be exposed to high seismic hazard or likely to impose

substantial seismic risk, the analyst may choose to develop acceleration time
histories that represent seismic hazards at specific AEP of interest. For some

B-2-20
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-10.—Mean UHRS from all sources for Wappapello Dam, Missouri, for
return periods of 225 years (20% in 50 years), 475 years (10% in 50 years),
975 years (5% in 50 years), 2,475 years (2% in 50 years), and 5,000 years (1% in
50 years) (from USGS Unified Hazard Tool).

dams where the seismic hazard or probability of failure under certain loading
conditions is high, relatively short return periods (e.g., 5,000 or 10,000 years) are
considered. Where consequences are high, consideration can be given to longer
return periods, and associated higher ground motions that are significant (but less
likely on an annualized basis). Expected ground motions at other return periods
should be considered when significant changes in failure likelihood are expected.
The range in ground motion spectral frequencies at each return period are usually
selected to span the likely variability in spectral responses at different periods,
and to account for differences in distance, magnitude, and site conditions. The
selected ground motions may then be used for dynamic analyses using accepted
software tools (e.g., FLAC, SHAKE, LS-DYNA). A wavelet-based method can
be used to produce acceleration time-histories through spectral matching to the
5-percent damping mean UHRS at the return period of interest. Because a UHRS
calculated from PSHA hazard curves are usually only available for horizontal
ground motions, the vertical components of the full response spectra used for
spectral matching is developed by scaling the UHRS via a ratio between the
vertical and horizontal components observed during historical events (McGuire
et al. 2001; Bozorgnia and Campbell 2004; Gülerce and Abrahamson 2011). The
vertical/horizontal (V/H) ratios are the regression results of near-field earthquake
records. These ratios vary with PGA and frequency, such that higher PGA values
typically are associated with larger V/H ratios. At low frequencies, V/H ratios are

B-2-21
July 2019
 Chapter B-2 Seismic Hazard Analysis

always less than one. Around 14 Hz, V/H ratios are large and sometimes greater
than one. The V/H ratio also depends on the distance between the source and the
site (figure B-2-11).

Figure B-2-11.—Plots showing differences in V/H

ratios with site class and distance from the seismic
source.

Ground motion time histories for use in engineering analyses are developed via
several steps:

• From the results of a PSHA, select the magnitude and source locations that
contribute the largest to the hazard for the site at the AEP of interest, with
consideration to the structure being analyzed. The PSHA de-aggregation
plots provide information to interpret the magnitude and distance for
sources that contribute significantly to the site hazard, which can be used
to guide selection of sources for time history development. The seismic
responses of many facilities are sensitive to different frequencies of

B-2-22
July 2019
 Chapter B-2 Seismic Hazard Analysis

ground motions, and PSHA results can be used to develop facility-specific

ground motion characteristics to improve assessment of seismic demand.
Some facilities are sensitive to long-period ground motions (i.e., >0.8 sec),
while the seismic response of other structures may be controlled by short-
period, high-frequency shaking (i.e., <0.4 sec). Different seismic sources
(with different earthquake distances, magnitudes, and return periods) may
control these different ground-motion parameters for a given site. For
example, sites in the Pacific Northwest of North America typically have
long-period ground motions controlled by the Cascadia Subduction Zone
(CSZ), which can produce potentially very large earthquakes (M 8 to 9) at
large distances (more than 100 km). In contrast, short-period ground
motions may be contributed by nearby, moderate-magnitude (but less
frequent) crustal fault sources. In the CEUS near the NMSZ, short-period
(about 0.2 sec) spectral accelerations typically are dominated by the
background seismicity from areal sources zones; but for periods of about
1 sec, the NMSZ typically contributes slightly more hazard than the
background seismicity for return periods of 975 years (5 percent
probability of exceedance in 50 years) to 2,475 years (2 percent in
50 years), but the NMSZ dominates hazard for longer spectral periods
(> 3 sec) (Hashash et al. 2015). The different hazard contributions
from the NMSZ and background seismicity sources are illustrated on
figure B-2-12.

• Determine a target spectrum for each source based on the magnitude,

distance, and site conditions by using a range of GMPE. Alternatively, the
UHRS for the site could be used, which will generally be conservative.
The CMS can also be used as a target spectrum (see next section).

• Locate at least one historic earthquake recording for each source, to be

used as a “seed time history.” A suitable recording from an historic
earthquake is used as the initial time history to provide the required
characteristics for spectral matching of ground motion accelerations from
the seed time history. Because the amplitude and frequency content will
be matched to the site, the most important property of the acceleration
recording is the duration. Correlations exist to estimate the target
duration given the magnitude and site-to-source distance. There are
many websites containing historic earthquake records available for
downloading, and the analyst should inquire with local and regional
experts for the most appropriate data sources; for example, the following
website can be used to search based on site to source distance and
magnitude: http://www.strongmotioncenter.org/.

B-2-23
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-12.—Mean hazard curves for average horizontal component

(a) PGA, (b) spectral acceleration = 0.2 s, (c) spectral acceleration = 1.0 s,
and (d) spectral acceleration = 3.0 s; before applying site amplification
factors. Two returns periods of 975 and 2,475 years are marked on the
right y-axis (from Hashash et al. 2015).

• Perform spectral matching of selected time histories to the target spectrum

using available software such as RSPMATCH. This can also be done
manually by modifying the amplitude spectra of the Fourier transform
such that the geometric mean of the two-component ground motion fits the
target spectrum. Care should be taken to ensure that the velocity and
displacement time series are compatible with the altered time history.
Additionally, synthetic time histories can be developed using specialized
programs that will generate a random time history that matches the input
response spectrum. These can be used directly without acquiring a
historic record.

• Develop the vertical target spectrum using V/H ratios based on currently
accepted methodology (McGuire et al. 2001, Bozorgnia and Campbell,
2004, Gülerce and Abrahamson 2011). Similarly, match the vertical time
history to the vertical spectrum.

B-2-24
July 2019
 Chapter B-2 Seismic Hazard Analysis

For each AEP of interest (e.g., 1/2,475 versus 1/9,975), accelerations for multiple
records can be used to develop the three-component, free-field acceleration time
histories. The time history ground motions can be generated synthetically. Time
histories reproduce major characteristics of earthquake recordings, such as
duration, envelope (defining the build-up of shaking amplitude – duration of
strong shaking – and decay of shaking), and change in waveform frequency with
time. For some structures where main contributors to hazard are different
(Cascadia versus background seismicity), the mean UHRS and acceleration time
histories can be developed that correspond to the specific hazard from individual
controlling sources. If all source zones are included in the hazard analysis, then
each time history is given equal weight. However, if the hazard for the source
zones is evaluated separately, then the hazard from each of the sources is additive.

B-2.2.5 Conditional (Mean) Spectra

As noted above, ground-motion time histories expected at a site, for various return
periods, may be needed for seismic analysis of dam structures. These histories
are developed to match selected target UHRS. However, in many tectonic
settings, critical ground motion parameters differ substantially depending on the
controlling seismic source (at different spectral accelerations and earthquake
return periods). When developing time histories based on the UHRS, the
association of the ground motions with source characteristics can be difficult and
may result in ground motions that do not represent the contributing seismic
sources. In other words, the UHRS combines the hazard from different sources
and does not reflect a realistic spectrum that can be expected to occur during a
single earthquake. The UHRS combines large amplitude short- and long-period
accelerations generated by two (or more) different sources. Additionally, the
response spectra of a given earthquake will tend to have peaks at some periods
and valleys at others. Because UHRS are generated by averaging many
earthquake recordings, these peaks and valleys are smoothed through the
spectrum. This will tend to overestimate the energy through large period ranges
of the response spectrum. Also, for less frequent events, the UHRS envelopes
many historic response spectra, such that the target spectrum will be equally
above average for all periods, which is unrealistic compared to the actual response
spectra. Baker and Cornell (2006) propose the use of a CMS for generating
ground motion time histories to address this limitation, so that the response
spectra are consistent with controlling seismic sources for the site and do not
overstate the structural response over the whole spectrum. Developing a CMS
should be considered if the site is located in an area of multiple distinct seismic
sources (shown by de-aggregation plots) or if it is thought that the UHRS is
overly conservative.

As developed by Baker and Cornell (2006), a CMS matches the UHRS level only
at the period of interest, which can be the expected fundamental period of the
structure. Development of a CMS for specific facilities should follow the

B-2-25
July 2019
 Chapter B-2 Seismic Hazard Analysis

guidance provided by Baker and Cornell (2006) and subsequent improvements.

As shown by example on figure B-2-13, two target periods were selected to
represent short and long natural periods of relevance to structural response
(Hashash et al. 2015). The plots show that the CMS at 0.2 sec resembles the
deterministic spectra from background sources and therefore can be used to
estimate seismic demand imposed from the background sources on the structure.
In contrast, the CMS at 1.0 sec resembles the deterministic spectra for NMSZ
large, distant earthquakes and can be used to estimate seismic demand associated
with these events. The high-frequency content of this spectrum is significantly
less than that of the UHRS, suggesting that the standard UHRS-based analysis
may yield conservatively high seismic demands for structural components
affected by high-frequency ground motions. Because seismic hazard is bimodal
at many sites across the country, use of the more detailed CMS analysis may
better represent the expected ground motions for selected system components and
may be worth the additional analytical effort. The 2008 USGS de-aggregation
tool allows the calculation of the CMS for the calculated mean and modal site to
source distance and magnitude combinations. This may be helpful as a first
estimate and to determine if more work should be completed. An example of the
USGS calculated CMS is shown on figure B-2-14 for Green Peter Dam located in
the Pacific Northwest. On this figure, the M 9.0 event is an interface earthquake
on the Cascadia Subduction Zone, while the M 7.15 event corresponds to an
intraslab earthquake generated by the subduction zone. As an example, the period
used is 0.3 seconds. Using the USGS tool, only a few periods can be selected.
The figure shows the mean spectrum; however, the USGS tool allows plotting
individual GMPE outputs to determine the range of spectra.

The primary difficulty in using the CMS lies in the fact that it is generated by
anchoring the spectrum to the UHRS at the fundamental period of the structure of
interest. This means that an estimate of the structural period must be made prior
to the detailed analysis of the structure. For concrete gravity dams, the
fundamental period is a function of the height of the dam, the ratio of the
foundation to structure stiffness, and the reservoir level at the time of the
earthquake. Further complicating matters is the fact that the fundamental period
will increase during the earthquake if damage to the structure is sustained.
Therefore, a reasonable period must be selected that will be representative of a
range of possible pools and potential damage to the structure. If a large range is
expected, it may be necessary to develop multiple CMS. For concrete gravity
dams, an initial estimate of the fundamental period can be made using the
methods outlined in Lokke and Chopra (2013) for non-overflow monoliths and
Chopra and Tan (1989).

B-2-26
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-13.—Horizontal UHRS (with 5% damping) and CMS for

T = 0.2 s and T = 1.0 s for a 2,475-year return period, compared
with median and 84th percentile for MW 6.0 and MW 7.5 deterministic
spectra (DS). Also shown is the 2008 USGS-derived UHRS
data for a 2,475-year return period and “rock” site conditions
(VS = 2,000 meters per second). Note similarity between the
1.0-sec CMS (dotted line) and the spectra derived from the large
(MW7.5) NMSZ earthquake; and the similarity between the 0.2-sec
CMS and the spectra derived from the smaller (MW6.0)
background-source earthquake.

B-2.2.6 Structural Considerations

It is important for engineers doing fragility evaluations to coordinate with seismic
hazard specialists on generating the hazard curves, uniform hazard spectra, and
time-history accelerograms. These products should reflect the parameters that
control the structural response of the dam and/or appurtenant structures, such that
seismic loading of smaller exceedance probabilities produce higher structural
response. This is not always peak horizontal ground acceleration, but may instead
be spectral acceleration at the predominant period of a structure, or perhaps the
area under a response spectrum curve covering more than one structural vibration
period if several modes contribute to the structural response. It may be necessary
to develop different hazard curves for different structures within the system
components at a given site. In some cases, certain combinations of acceleration
and velocity may be critical to the structural response, and hazard curves should
be developed that relate to simultaneous exceedance of given acceleration and
velocity levels.

B-2-27
July 2019
 Chapter B-2 Seismic Hazard Analysis

Figure B-2-14.—Example CMS for Green Peter Dam (Oregon) generated using the
2008 USGS de-aggregation tool for probability of exceedance of 2% in 50 yr.

B-2.3 Surface Fault Displacement/Permanent

Ground Deformation
For some dam sites, the potential for surface fault displacement (or the more
general case, PGD) through the site or foundation is a major concern (Allen and
Cluff, 2000). Basically, two types of surface faulting are generally recognized:
principal (or primary) and distributed (or secondary) (Stepp et al. 2001). By
definition, principal faulting occurs along the main fault plane (or planes) along
which the release of seismic energy occurs. PGD can be generated by moderate
to large earthquakes, as exemplified in many recent events (e.g., Kelson et al.
2001, Kelson et al. 2012, Langridge et al. 2018). At the ground surface,
displacement may be confined to a single narrow fault or a relatively narrow zone
that is a few to several meters wide, or it can occur over a broad zone and on
several fault strands. Distributed (or secondary) faulting is displacement that
occurs on a fault or fracture away from the primary rupture and can be quite
spatially discontinuous. It may occur in response to an earthquake on another
nearby fault (possibly due to ground shaking) or it may be due to a structural
connection or linkage between the causative fault and the fault upon which the

B-2-28
July 2019
 Chapter B-2 Seismic Hazard Analysis

secondary displacement occurs. In either case, secondary faulting can occur a few
to several kilometers away from the causative fault. Distributed faulting can
occur on a fault that is considered capable of principal surface rupture; however,
distributed faulting displacement is typically less than that observed by primary
faulting. For most dams, primary and secondary fault displacement should be
considered, as well as distributed deformation adjacent to an active primary fault
if the dam is within a zone of potential distributed deformation.

Although many existing dam sites contain faults, most of these do not contain
active faults that could produce surface rupture of the dam foundation and
consequent displacement of embankment material. Known examples of active
faults beneath dams include Coyote Dam across the active Calaveras fault in
Santa Clara County, California (Werner et al. 2017), and Isabella Auxiliary Dam
across the active Kern Canyon fault in Kern County, California (Serafini et al.
2015).

In order to assess the possibility of fault rupture in a dam foundation, several steps
should be taken in a sequence of progressively more detailed analyses. The
results from each phase in the sequence should be used in deciding whether or
not additional data are required to confidently evaluate rupture potential and
hazard at the site. The steps range from relatively simple collection of existing
evidence for or against fault activity, to a complex and comprehensive PFDHA,
below.

Any fault or fold that, if active, would pose a surface deformation hazard at the
dam site should be evaluated. Faults that have slipped and geologic structures
that have deformed during the late Quaternary period should be considered
potentially active and evaluated to assess rupture characteristics. Faults and
folds of the site region should be evaluated in the context of their structural
development with primary attention given to their tertiary-quaternary evolution
and relationship to the contemporary tectonic regime. All seismic sources with a
potential to cause PGD that may affect performance of dam facilities at a site shall
be identified and characterized as described in this section.

A general outline of important phases in assessing surface rupture displacement

includes:

• Determine presence/absence of faulting through the dam site, usually

based on review of technical literature and geologic maps, and analysis of
the current tectonic setting

B-2-29
July 2019
 Chapter B-2 Seismic Hazard Analysis

• Estimate fault activity or inactivity, typically through basic geologic or

geomorphic analyses (e.g., analysis of existing imagery, field
reconnaissance; limited age-dating analyses; Angell et al. 1999); if
inactivity can be reasonably shown according to applicable criteria, no
further action is needed

• If fault activity is demonstrated or indeterminate, the assessment can

utilize existing empirical relationships from historical worldwide
earthquake ruptures (e.g., Wells and Coppersmi 1994; Wesnousky 2008),
or utilize seismologic principles (Hanks and Kanamori 1979), to estimate
an expected amount of displacement for specific deterministic scenarios

• If better assessment of uncertainty is needed to constrain the seismic

loading, detailed, site-specific geologic or paleoseismic data may be
collected (e.g., detailed site geologic mapping, detailed geomorphic
analyses, subsurface trenching or drilling, geophysical surveying,
age-dating); site-specific data on the location, amount, and timing of
ruptures may provide adequate information for the risk assessment

• If loading conditions or design needs warrant, a PFDHA may be required

to develop the likelihood and range of rupture beneath the dam foundation,
based on site-specific data on the location, amount, and timing of surface
rupture

As a general rule, for most dam portfolios, the need to complete a full PFDHA to
assess fault rupture characteristics will be limited. However, where potential
failure modes related to surface fault rupture are plausible, the information
generated by a PFDHA is critical input for assessing the seismic (rupture) loading
over time periods of interest in risk assessments. For these cases, the following
section provides background and guidance for completing a PFDHA for dam
safety purposes.

B-2.3.1 Probabilistic Fault Displacement Hazard Analysis

PFDHA have been done in only a limited number of cases. A recent example
includes the analysis conducted for Isabella Auxiliary Dam near Bakersfield,
California (Serafini 2015). This analysis basically followed the methodology set
forth by Stepp et al. (2001) for the proposed Yucca Mountain, Nevada waste
repository.

The hazard calculations done for PFDHA are analogous to the probabilistic
ground motion methodology (Cornell 1968; McGuire 1978), and the results are
represented by a hazard curve similar to those for PSHA. The curve depicts

B-2-30
July 2019
 Chapter B-2 Seismic Hazard Analysis

annual occurrence of fault displacement values (i.e., the annual frequency of

exceeding a specified amount of displacement). The frequency of displacement
v (d) can be computed from the expression:

v(d) = λDE P(D > d) Equation B-2-1

Where:

λDE = The frequency of displacement events at the point where the fault
intersects the dam foundation

P(D > d) = A displacement attenuation function, the conditional probability

that displacement, D, will exceed d, given the occurrence of an
earthquake.

The conditional probability of exceeding a displacement, d, P(D>d), is based upon

empirical data and thus contains event to event variability as well as the
variability of displacement along strike of the fault.

Following development of the PFDHA approach noted above, recent efforts

have defined the data requirements and refined the analytical approaches, as
summarized in American Nuclear Society Standard 2.30 (ANS 2014). Similar
to PSHA, a PFDHA consists of calculating the exceedance probabilities of
fault-related surface displacements and capturing the range of natural variability
and knowledge uncertainty. Youngs et al. (2003) developed two approaches for
conducting site-specific PFDHA, an earthquake approach and a displacement
approach. The methodology for the earthquake approach is similar to the PSHA
methodology, with the ground motion prediction model replaced by a
displacement attenuation function. Petersen et al. (2011) extended the Youngs
et al. (2003) earthquake approach to include dependence of fault displacement
hazard on the accuracy of surface fault mapping and the complexity of the
mapped fault trace, and Moss and Ross (2011) also provide clarifications of the
PFDHA methodology. In the displacement approach, the characteristics of
surface fault displacement (amount of displacement and frequency of occurrence)
observed at the site of interest are used to quantify the hazard without invoking a
specific mechanism for their cause (i.e., rupture of a fault as for the earthquake
approach). The approaches and the basis for selection of the appropriate approach
to conduct a PFDHA are described in ANS (2014).

In cases where the potential for surface fault rupture may affect a dam or dam site
and additional characterization is needed to address potential failure modes,
assessment of surface fault rupture can be completed via site-specific studies or
through a PFDHA. An example of a site-specific analysis is summarized by
Serafini et al. (2015), in which the expected maximum coseismic slip was
evaluated based on (1) site-specific paleoseismic data and worldwide empirical
data on event-to-event slip variability, and (2) scenario-based fault displacements

B-2-31
July 2019
 Chapter B-2 Seismic Hazard Analysis

using empirical relationships between earthquake magnitude and surface

displacement. Earthquake rupture scenarios developed from fault-specific
paleoseismic and geologic analyses provided a range of expected earthquake
magnitudes and associated minimum, mean, and maximum coseismic
displacements, with annual exceedance probabilities developed from earthquake
magnitude-frequency relations. From these analyses, a design displacement value
was chosen for sizing filter and drain zones for the downstream buttress
modifications.

Where site-specific data are not available, a PFDHA may be needed to assist
PFMA identification and/or mitigation design. For a PFDHA, the geological,
seismological, and geotechnical characteristics of a site and its environs should be
investigated in sufficient scope and detail necessary to support the evaluation of
PGD. Site characterization of PGD should include a review of pertinent literature
and field investigations and subject matter experts with experience in PFDHA
should help define the program of investigations. Any fault or fold that, if active,
would pose a surface deformation hazard at the site should be evaluated.
Faults that have slipped and geologic structures that have deformed during the
Quaternary period should be considered potentially active and evaluated to assess
rupture characteristics. Faults and folds of the site region should be evaluated in
the context of their structural development with primary attention given to their
geologic development and relationship to the contemporary tectonic setting. All
seismic sources with a potential to cause PGD that may affect performance of
facilities at a site should be identified and characterized. The characterization of
sources of PGD should be evaluated in a probabilistic framework as described in
ANS (2014), including fault zone activity, location, width, orientation, sense of
movement, and expected amount of coseismic PGD.

B-2.4 Relevant Case Histories and Examples

Several PSHA have been conducted for the USACE, Reclamation, and Tennessee
Valley Authority (TVA) during the past few years.

For recent examples completed by or for Reclamation and TVA, the readers
are encouraged to contact respective national technical centers for available
up-to-date PSHA examples.

B-2.5 References
Allen, C.R. and L.S. Cluff. 2000. “Active Faults in Dam Foundations: An
Update.” Proceedings of the 12th World Conference on Earthquake
Engineering, Paper No. 2490, 8 p. Auckland, New Zealand.

B-2-32
July 2019
 Chapter B-2 Seismic Hazard Analysis

American Nuclear Society. 2014. Criteria for Assessing Tectonic Surface Fault
Rupture and Deformation at Nuclear Facilities, American National
Standards Institute/ANS 2.30, La Grange Park, Illinois.

American Society of Civil Engineers. Continually Updated. Minimum Design

Loads for Buildings and Other Structures, ASCE 7.

Angell M.A., K.L. Hanson, K.I. Kelson, W.R. Lettis, and E. Zurflueh. 1999.
Techniques for Identifying Faults and Determining Their Origins,
NUREG/CR-5503. Nuclear Regulatory Commission, Washington DC.

Baker, J.W. and C.A. Cornell. 2006. “Spectral Shape, Epsilon and Record
Selection,” Earthquake Engineering and Structural Dynamics, Vol. 35,
pp. 1077–1095.

Bazzurro, P. and C.A. Cornell. 1999. “Disaggregation of Seismic Hazard,”

Bulletin of the Seismological Society of America, Vol. 89, pp. 501–520.

BC Hydro. 2008. Seismic Source Characterization for the Probabilistic Seismic

Hazard Analysis for the BC Hydro Service Area. BC Hydro and Power
Authority, Canada.

Bozorgnia, Y. and K.W. Campbell. 2004. “The Vertical-to-Horizontal Response

Spectral Ratio and Tentative Procedures for Developing Simplified V/H
and Vertical Design Spectra.” Journal of Earthquake Engineering, Vol. 8,
pp. 175-207.

Central and Eastern United States Seismic Source Characterization (CEUS-SSC).

2012. Technical Report: Central and Eastern United States Seismic
Source Characterization for Nuclear Facilities. CEUS-SSC, Electrical
Power Research Institute, Palo Alto, California.

Chopra, A.K. and H. Tan. 1989. Simplified Earthquake Analysis of Gated

Spillway Monoliths of Concrete Gravity Dams, USACE Technical Report
SL-89-4. United States Army Corps of Engineers, Washington DC.

Cornell, C.A. 1968. “Engineering Seismic Risk Analysis.” Bulletin of the

Seismological Society of America, Vol. 58, pp. 1583-1606.

Gülerce, Z. and N. Abrahamson. 2011. “Site-Specific Design Spectra for

Vertical Ground Motion,” Earthquake Spectra, Vol. 27, pp. 1023-1047.

Hanks, T.C. and H. Kanamori. 1979. “A Moment Magnitude Scale.” Journal of

Geophysical Research, Vol. 84, pp. 2348-2350.

Hashash, Y.M.A., N.A. Abrahamson, S.M. Olson, S. Hague, and B. Kim. 2015.
“Conditional Mean Spectra in Site-Specific Seismic Hazard Evaluation for

B-2-33
July 2019
 Chapter B-2 Seismic Hazard Analysis

a Major River Crossing in the Central United States.” Earthquake Spectra,

Vol. 31, No. 1, pp. 47-69.

Kelson, K.I., K.H. Kang, W.D. Page, and L.S. Cluff. 2001. “Representative
Styles of Deformation along the Chelungpu Fault from the 1999 Chi-Chi
(Taiwan) Earthquake: Geomorphic Characteristics and Responses of
Man-Made Structures.” Bulletin of the Seismological Society of America,
Vol. 91, No. 5, pp. 930-952.

Kelson, K.I., R.C. Witter, A. Tassara, I. Ryder, C. Ledezma, G. Montalva,

D. Frost, N. Sitar, R. Moss and L. Johnson. 2012. “ Coseismic Tectonic
Surface Deformation during the 2010 Maule, Chile, Mw 8.8 Earthquake,”
Earthquake Spectra, Vol. 28, No. S1, pp. S39-S54.

Langridge, R.M., J. Rowland, P. Villamor, J. Mountjoy, D.B. Townsend,

E. Nissen, C. Madugo, W.F. Ries, C. Gasston, A. Canva, A.E. Hatem, and
I. Hamling. 2018. “Coseismic Rupture and Preliminary Slip Estimates for
the Papatea Fault and its Role in the 2016 Mw 7.8 Kaikōura, New Zealand,
Earthquake.” Bulletin of the Seismological Society of America, Vol. 108,
No. 3B, pp. 1596-1622.

Lokke, A. and A.K. Chopra. 2013. Response Spectrum Analysis of Concrete

Gravity Dams Including Dam-Water-Foundation Interaction. Pacific
Earthquake Engineering Research Center, Berkeley, California.

Luco, N., B.R. Ellingwood, R.O. Hamburger, J.D. Hooper, J.K. Kimball, and
C.A. Kircher. 2007. “Risk-Targeted Versus Current Seismic Design
Maps for the Conterminous United States,” Proceedings for the Structural
Engineers Association of California 2007 Convention.

McGuire, R.K. 1974. Seismic Structural Response Risk Analysis Incorporating

Peak Response Regressions on Earthquake Magnitude and Distance,
Research Report R74-51. Massachusetts Institute of Technology,
Department of Civil Engineering, Cambridge, Massachusetts.

_____. 1995. “Probabilistic Seismic Hazard Analysis and Design Earthquakes:

Closing the Loop.” Bulletin of the Seismological Society of America,
Vol. 85, pp. 1275-1284.

_____. 1978. “Seismic Ground Motion Parameter Relations,” Journal of the

Geotechnical Engineering Division, Vol. 104, Issue 4, pp. 481-490.

B-2-34
July 2019
 Chapter B-2 Seismic Hazard Analysis

McGuire, R.K., W.J. Silva, and C.J. Costantino. 2001. Technical Basis
for Revision of Regulatory Guidance on Design Ground Motions:
Hazard- and Risk-Consistent Ground Motion Spectra Guidelines,
NUREG/CR-6728. Nuclear Regulatory Commission, Washington DC.

Moss, R. and Z. Ross. 2011. “Probabilistic Fault Displacement Hazard Analysis

for Reverse Faults.” Bulletin of the Seismological Society of America,
Vol. 101, No. 4, pp.1542-1553.

Petersen, M., T.E. Dawson, R. Chen, T. Cao, C.J. Wills, D.P. Schwartz, and
A.D. Frankel. 2011. “Fault Displacement Hazard for Strike-Slip Faults,”
Bulletin of the Seismological Society of America, Vol. 101, No. 2,
pp. 805–825.

Senior Seismic Hazard Analysis Committee. 2007. Recommendations for

Probabilistic Seismic Hazard Analysis: Guidance on Uncertainty and Use
of Experts, NUREG/CR-6372, Nuclear Regulatory Commission, SSHAC,
Washington D.C.

Serafini, D. 2015. Isabella Dam PED Design Fault Rupture Parameters for the
Right Abutment of the Auxiliary Dam; Memorandum of Record. United
States Army Corps of Engineers, Sacramento District Engineering, Dam
Safety Production Center, Sacramento, California.

Serafini, D.C., K.I. Kelson, R.S. Rose, and A.T. Lutz. 2015. Characterization
and Mitigation of Fault Rupture Hazard: Engineering Basis of Design for
Isabella Auxiliary Dam, California. Association of Engineering and
Environmental Geologists Annual Meeting, Pittsburgh, Pennsylvania.

Stepp, J.C., I. Wong, J. Whitney, R. Quittmeyer, N. Abrahamson, G. Toro, R.

Youngs, K. Coppersmith, J. Savy, and T. Sullivan. 2001. “Probabilistic
Seismic Hazard Analysis for Ground Motions and Fault Displacement
at Yucca Mountain, Nevada,” Earthquake Spectra, Vol. 17, No. 1,
pp. 113-151.

U.S. Geological Survey. Continuously updated. Quaternary Fault and Fold

Database for the United States.

_____. Continuously updated. Seismic Hazard Maps and Site-Specific Data.

_____. Continuously updated. Unified Hazard Tool.

_____. Continuously updated. Uniform California Earthquake Rupture Forecast

(UCERF)3: A New Earthquake Forecast for California’s Complex Fault
System.

B-2-35
July 2019
 Chapter B-2 Seismic Hazard Analysis

U.S. Nuclear Regulatory Commission. 2007. A Performance-Based Approach

to Define the Site-Specific Earthquake Ground Motion, USNRC
Guide 1.208.

Wells, D.L and K.J. Coppersmith. 1994. “New Empirical Relationships among
Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface
Displacement,” Bulletin of the Seismological Society of America, vol. 84,
pp. 974-1002.

Werner, C., B. Baker, U. Wegmuller, R. Cassotto, and M. Fahnestock. 2017.

“Three-Dimensional Deformation of Coyote Dam by the Calaveras Fault
Obtained from Multi-Aspect Ku-Band Terrestrial Radar Interferometry,”
Advances in the Science and Applications of SAR Interferometry and
Sentinel-1 InSAR, Helsinki, Finland.

Wesnousky, S.G. 2008. “Displacement and Geometrical Characteristics of

Earthquake Surface Ruptures,” Bulletin of the Seismological Society of
America, vol. 98, pp. 1609-1632.

Youngs, R., W.J. Arabasz, R.E. Anderson, A.R. Ramelli, J.P. Ake,
D.B. Slemmons, J.P. McCalpin, D.I. Doser, C.J. Fridrich, F.H. Swan,
A.M. Rogers, J.C. Yount, L.W. Anderson, K.D. Smith, R.L. Bruhn,
P.L.K. Knuepfer, R.B. Smith, C.M. dePolo, D.W. O’Leary,
K.J. Coppersmith, S.K. Pezzopane, D.P. Schwartz, J.W. Whitney,
S.S. Olig, and G.R. Toro. 2003. “A Methodology for Probabilistic Fault
Displacement Hazard Analysis (PFDHA),” Earthquake Spectra, Vol. 19,
No. 1, pp. 191-219.

B-2.5.1 Other References

Building Seismic Safety Council. 2003. National Earthquake Hazards Reduction

Program Recommended Provisions for Seismic Regulations for New
Buildings and Other Structures, Part 1: Provisions. FEMA 368, Federal
Emergency Management Agency, Washington, DC.

B-2-36
July 2019