3

Earthquake Ground Motion

Nicolas Luco1, Ph.D., P.E., Michael Valley2, S.E.,
C.B. Crouse3, P.E., Ph.D.
Contents
3.1

3.1.1

ASCE 7-05 Seismic Maps ..................................................................................................... 2

3.1.2

MCER Ground Motions in the Provisions and in ASCE 7-10 ............................................... 3

3.1.3

PGA Maps in the Provisions and in ASCE 7-10 ................................................................... 7

3.1.4

Basis of Vertical Ground Motions in the Provisions and in ASCE 7-10 .............................. 7

3.1.5

Summary ................................................................................................................................ 7

3.1.6

References.............................................................................................................................. 8

3.2

DETERMINATION OF GROUND MOTION VALUES AND SPECTRA ................................ 9

3.2.1

ASCE 7-05 Ground Motion Values ....................................................................................... 9

3.2.2

2009 Provisions Ground Motion Values ............................................................................. 10

3.2.3

ASCE 7-10 Ground Motion Values ..................................................................................... 11

3.2.4

Horizontal Response Spectra ............................................................................................... 12

3.2.5

Vertical Response Spectra ................................................................................................... 13

3.2.6

Peak Ground Accelerations ................................................................................................. 14

3.3

BASIS OFEARTHQUAKE GROUND MOTION MAPS ............................................................ 2

SELECTION AND SCALING OF GROUND MOTION RECORDS ....................................... 14

3.3.1

Approach to Ground Motion Selection and Scaling ............................................................ 15

3.3.2

Two-Component Records for Three Dimensional Analysis ............................................... 24

3.3.3

One-Component Records for Two-Dimensional Analysis .................................................. 27

3.3.4

References............................................................................................................................ 28

Author of Section 3.1.

Author of Sections 3.2 and 3.3.
3
Reviewing author.
2

FEMA P-752, NEHRP Recommended Provisions: Design Examples

Most of the effort in seismic design of buildings and other structures is focused on structural design. This
chapter addresses another key aspect of the design processcharacterization of earthquake ground
motion. Section 3.1 describes the basis of the earthquake ground motion maps in the Provisions and in
ASCE 7. Section 3.2 has examples for the determination of ground motion parameters and spectra for use
in design. Section 3.3 discusses and provides an example for the selection and scaling of ground motion
records for use in response history analysis.

3.1 BASIS OFEARTHQUAKE GROUND MOTION MAPS

This section explains the basis of the new Risk-Targeted Maximum Considered Earthquake (MCER)
ground motions specified in the 2009 Provisions and mapped in ASCE 7-10. In doing so, it also explains
the basis for the uniform-hazard ground motion (SSUH and S1UH) maps, risk coefficient (CRS and CR1) maps
and deterministic ground motion (SSD and S1D) maps in the 2009 Provisions. These three sets of maps are
combined to form the Site Class B MCER ground motion (SS and S1) maps in ASCE 7-10. The use of SS
and S1 ground motions in the 2009 Provisions and ASCE 7-10 to derive a design response spectrum
remains the same as it is in ASCE 7-05.
This section also explains the basis for the new Peak Ground Acceleration (PGA) maps in the 2009
Provisions and ASCE 7-10 and the new equations for vertical ground motions. The basis for the longperiod transition (TL) maps in the 2009 Provisions and ASCE 7-10, which are identical to those in ASCE
7-05, is also reviewed. In fact, we start with a review of these maps and the Maximum Considered
Earthquake (MCE) ground motion maps in ASCE 7-05.
3.1.1

ASCE 7-05 Seismic Maps

The bases for the seismic ground motion (MCE) and long-period transition (TL) maps in Chapter 22 of
ASCE 7-05 were established by, respectively, the Building Seismic Safety Council (BSSC) Seismic
Design Procedures Group, also referred to as Project 97, and Technical Subcommittee 1 (TS-1) of the
2003 Provisions Update Committee. They are reviewed briefly in the following two subsections.
3.1.1.1 Maximum Considered Earthquake (MCE) Ground Motion Maps
The MCE ground motion maps in ASCE 7-05 can be described as applications of its site-specific ground
motion hazard analysis procedure in Chapter 21 (Section 21.2), using ground motion values computed by
the USGS National Seismic Hazard Mapping Project (in Golden, CO) for a grid of locations and/or
polygons that covers the US. In particular, the 1996 USGS update of the ground motion values was used
for ASCE 7-98 and ASCE 7-02; the 2002 USGS update was used for ASCE 7-05. The site-specific
procedure in all three editions calculates the MCE ground motion as the lesser of a probabilistic and a
deterministic ground motion. Hence, the USGS computed both types of ground motions, whereas
otherwise it would have only computed probabilistic ground motions. Brief reviews of how the USGS
computed the probabilistic and deterministic ground motions are provided in the next few paragraphs.
For additional information, see Leyendecker et al. (2000).
The USGS computation of the probabilistic ground motions that are part of the basis of the MCE ground
motion maps in ASCE 7-05 is explained in detail in Frankel et al. (2002). In short, the USGS combines
research on potential sources of earthquakes (e.g., faults and locations of past earthquakes), the potential
magnitudes of earthquakes from these sources and their frequencies of occurrence, and the potential
ground motions generated by these earthquakes. Uncertainty and randomness in each of these
components is accounted for in the computation via contemporary Probabilistic Seismic Hazard Analysis
(PSHA), which was originally conceived by Cornell (1968). The primary output of PSHA computations
are so-called hazard curves, for locations on a grid covering the US in the case of the USGS computation.
3-2

Chapter 3: Earthquake Ground Motion

Each hazard curve provides mean annual frequencies of exceeding various user-specified ground motions
amplitudes. From these hazard curves, the ground motion amplitudes for a user-specified mean annual
frequency can be interpolated and then mapped. The results are known as uniform-hazard ground motion
maps, since the mean annual frequency (or corresponding probability) is uniform geographically.
For ASCE 7-05 (as well as ASCE 7-02 and ASCE 7-98), a mean annual exceedance frequency of 1/2475
per year, corresponding to 2% probability of exceedance in 50 years, was specified by the aforementioned
BSSC Project 97. That project also specified that the ground motion parameters be spectral response
accelerations at vibration periods of 0.2 seconds and 1 second, for 5% of critical damping. For the
average shear wave velocity at small shear strains in the upper 100 feet (30 m) of subsurface below each
location (vS,30), the USGS decided on a reference value of 760 m/s. The BSSC subsequently decided to
regard this reference value, which is at the boundary of Site Classes B and C, as corresponding to Site
Class B. Justifications for the decisions summarized in this paragraph are provided in the Commentary of
FEMA 303, FEMA 369 and FEMA 450.
The USGS computation of the deterministic ground motions for ASCE 7-05 is detailed in the FEMA 303
Commentary. As defined by Project 97 and subsequently specified in the site-specific procedure of
ASCE 7-05 (Section 21.2.2), each deterministic ground motion is calculated as 150% of the median
spectral response acceleration for a characteristic earthquake on a known active fault within the region.
The specific characteristic earthquake is that which generates the largest median spectral response
acceleration at the given location. As for the probabilistic ground motions, the spectral response
accelerations are at vibration periods of 0.2 seconds and 1 second, for 5% of critical damping. The same
reference site class (see preceding paragraph) is used as well. Though not applied to probabilistic ground
motions, lower limits of 1.5g and 0.6g are applied to the deterministic ground motions.
As mentioned at the beginning of this section, the lesser of the probabilistic and deterministic ground
motions described above yields the MCE ground motions mapped in ASCE 7-05. Thus, the MCE
spectral response accelerations at 0.2 seconds and 1 second are equal to the corresponding probabilistic
ground motions wherever they are less than the lower limits of the deterministic ground motions (1.5g
and 0.6g, respectively). Where the probabilistic ground motions are greater than the lower limits, the
deterministic ground motions sometimes govern, but only if they are less than their probabilistic
counterparts. On the MCE ground motion maps in ASCE 7-05, the deterministic ground motions govern
mainly near major faults in California (like the San Andreas), in Reno and in parts of the New Madrid
Seismic Zone. The deterministic ground motions that govern are as small as 40% of their probabilistic
counterparts.
3.1.1.2 Long-Period Transition Period (TL) Maps
The details of the procedure and rationale used in developing the TL maps in ASCE 7-05; and now in
ASCE 7-10 and the 2009 Provisions, are found in Crouse et al. (2006). In short, the procedure consisted
of two steps. First, a relationship between TL and earthquake magnitude was established. Second, the
modal magnitude from deaggregation of the USGS 2% in 50-year ground motion hazard at a 2-second
period (1 second for Hawaii) was mapped. The long-period transition period (TL) maps that combined
these two steps delimit the transition of the design response spectrum from a constant velocity (1/T) to a
constant displacement (1/T2) shape.
3.1.2

MCER Ground Motions in the Provisions and in ASCE 7-10

Like the MCE ground motion maps in ASCE 7-05 reviewed in the preceding section, the new RiskTargeted Maximum Considered Earthquake (MCER) ground motions in the 2009 Provisions and ASCE 710 can be described as applications of the site-specific ground motion hazard analysis procedure in
Chapter 21 (Section 21.2) of both documents. For the MCER ground motions, however, the USGS values
(for a grid of site and/or polygons covering the US) that are used in the procedure are from its 2008
update. Still, the site-specific procedure of the Provisions and ASCE 7-10 calculates the MCER ground

3-3

FEMA P-752, NEHRP Recommended Provisions: Design Examples

motion as the lesser of a probabilistic and a deterministic ground motion. The definitions of the
probabilistic and deterministic ground motions in ASCE 7-10, however, are different than in ASCE 7-05.
The definitions were revised for the 2009 Provisions and ASCE 7-10 by the BSSC Seismic Design
Procedures Reassessment Group (SDPRG), also referred to as Project 07. Three revisions were made:
1) The probabilistic ground motions are redefined as so-called risk-targeted ground motions, in lieu
of the uniform-hazard (2% in 50-year) ground motions that underlie the ASCE 7-05 MCE ground
motion maps,
2) the deterministic ground motions are redefined as 84th-percentile ground motions, in lieu of
median ground motions multiplied by 1.5; and
3) the probabilistic and deterministic ground motions are redefined as maximum-direction ground
motions, in lieu of geometric mean ground motions.
In addition to these three BSSC redefinitions of probabilistic and deterministic ground motions, there is a
fourth difference in the ground motion values computed by the USGS for the 2009 Provisions and ASCE
7-10 versus ASCE 7-05:
4) The probabilistic and deterministic ground motions were recomputed using updated earthquake
source and ground motion propagation models, e.g., the Unified California Earthquake Rupture
Forecast (UCERF, Version 2; Field et al., 2008) and the Next Generation Attenuation (NGA)
ground motion models4.
Each of the above four differences between the basis of the MCE ground motions (in ASCE 7-05) and
that of the MCER ground motions (in the 2009 Provisions and ASCE 7-10) is explained in more detail
below. Also explained are the differences in the presentation of MCER ground motions between the 2009
Provisions and ASCE 7-10; the numerical values of the MCER ground motions in the two documents are
otherwise identical.
3.1.2.1 Risk-Targeted Probabilistic Ground Motions
For the MCE ground motion maps in ASCE 7-05, recall (from Section 3.1.1) that the underlying
probabilistic ground motions are specified to be uniform-hazard ground motions that have a 2%
probability of being exceeded in 50 years. It has long been recognized, though, that it really is the
probability of structural failure with resultant casualties that is of concern; and the geographical
distribution of that probability is not necessarily the same as the distribution of the probability of
exceeding some ground motion (p. 296 of ATC 3-06, 1978). The primary reason that the distributions of
the two probabilities are not the same is that there are geographic differences in the shape of the hazard
curves from which uniform-hazard ground motions are read. The Commentary of FEMA 303 (p. 289)
reports that because of these differences, questions were raised concerning whether definition of the
ground motion based on a constant probability for the entire United States would result in similar levels
of seismic safety for all structures.
The changeover to risk-targeted probabilistic ground motions for the 2009 Provisions and ASCE 7-10
takes into account the differences in the shape of hazard curves across the US. Where used in design, the
risk-targeted ground motions are expected to result in buildings with a geographically uniform mean
annual frequency of collapse, or uniform risk. The BSSC, via Project 07, decided on a target risk level
corresponding to 1% probability of collapse in 50 years. This target is based on the average of the mean
annual frequencies of collapse across the Western US (WUS) expected to result from design for the
4

See the February 2008 Earthquake Spectra Special Issue on the Next Generation Attenuation
Project, Volume 24, Number 1.
3-4

Chapter 3: Earthquake Ground Motion

probabilistic ground motions in ASCE 7-05. Consequently, in the WUS the risk-targeted ground motions
in the 2009 Provisions and ASCE 7-10 are generally within 15% of the corresponding uniform-hazard
(2% in 50-year) ground motions. In the Central and Eastern US, where the shapes of hazard curves are
known to differ from those in the WUS, the risk-targeted ground motions generally are smaller. For
instance, in the New Madrid Seismic Zone and near Charleston, South Carolina ratios of risk-targeted to
uniform-hazard ground motions are as small as 0.7.
The computation of risk-targeted probabilistic ground motions for the MCER ground motions in the 2009
Provisions and ASCE 7-10 is detailed in Provisions Part 1 Sections 21.2.1.2 and C21.2.1 and in Luco et
al. (2007). While the computation of the risk-targeted ground motions is different than that of the
uniform-hazard ground motions specified for the MCE ground motions in ASCE 7-05, both begin with
USGS computations of hazard curves. As explained in Section 3.1.1, the uniform-hazard ground motions
simply interpolate the hazard curves for a 2% probability of exceedance in 50 years. In contrast, the risktargeted ground motions make use of entire hazard curves. In either case, the end results are probabilistic
spectral response accelerations at 0.2 seconds and 1 second, for 5% of critical damping and the reference
site class.
3.1.2.2 84th-Percentile Deterministic Ground Motions
For the MCE ground motion maps in ASCE 7-05, recall (from Section 3.1.1) that the underlying
deterministic ground motions are defined as 150% of median spectral response accelerations. As
explained in the FEMA 303 Commentary (p. 296),
Increasing the median ground motion estimates by 50 percent [was] deemed to provide an
appropriate margin and is similar to some deterministic estimates for a large magnitude
characteristic earthquake using ground motion attenuation functions with one standard
deviation. Estimated standard deviations for some active fault sources have been determined to
be higher than 50 percent, but this increase in the median ground motions was considered
reasonable for defining the maximum considered earthquake ground motions for use in design.
For the MCER ground motions in the 2009 Provisions and ASCE 7-10, however, the BSSC decided to
define directly the underlying deterministic ground motions as those at the level of one standard
deviation. More specifically, they are defined as 84th-percentile ground motions (since it has been widely
observed that ground motions follow lognormal probability distributions). The remainder of the
definition of the deterministic ground motions remains the same as that used for the MCE ground motions
maps in ASCE 7-05. For example, the lower limits of 1.5g and 0.6g described in Section 3.1.1 are
retained.
The USGS applied a simplification specified by the BSSC in computing the 84th-percentile deterministic
ground motions for the 2009 Provisions and ASCE 7-10. The 84th-percentile spectral response
accelerations were approximated as 180% of median values. This approximation corresponds to a
logarithmic ground motion standard deviation of approximately 0.6, as demonstrated in the Provisions
Part 1 Section C21.2.2. The computation of deterministic ground motions is further described in
Provisions Part 2 Section C21.2.2.
3.1.2.3 Maximum-Direction Probabilistic and Deterministic Ground Motions
Due to the ground motion attenuation models used by the USGS in computing them5, overall the MCE
spectral response accelerations in ASCE 7-05 represent the geometric mean of two horizontal components
of ground motion. Most users of ASCE 7-05 are unaware of this fact, particularly since the discussion
notes on the MCE ground motion maps incorrectly state that they represent the random horizontal
component of ground motion. For the 2009 Provisions and ASCE 7-10, the BSSC decided that it would
5

See the January/February 1997 Seismological Research Letters Special Issue on Ground Motion
Attenuation Relations, Volume 68, Number 1.
3-5

FEMA P-752, NEHRP Recommended Provisions: Design Examples

be an improvement if the MCER ground motions represented the maximum direction of horizontal
spectral response acceleration. Reasons for this decision are explained in Provisions Part 1 Section
C21.2.
Since the attenuation models used in computing the 2008 update of the USGS ground motions also
represent (overall) geomean spectral response accelerations, for the 2009 Provisions and ASCE 7-10
the BSSC provided factors to convert approximately to maximum-direction ground motions. Based on
research by Huang et al. (2008) and others, the factors are 1.1 and 1.3 for the spectral response
accelerations at 0.2 seconds and 1.0 second, respectively. The basis for these factors is elaborated upon in
the Provisions Part 1 Section C21.2. They are applied to both the USGS probabilistic hazard curves from
which the risk-targeted ground motions (described in Section 3.1.2.1) are derived and the USGS
deterministic ground motions (described in Section 3.1.2.2).
3.1.2.4 Updated Ground Motions from USGS (2008)
For the MCE ground motion maps in ASCE 7-05, recall (from Section 3.1.1) that the underlying
probabilistic and deterministic ground motions are from the 2002 USGS update. As mentioned above, the
MCER ground motions in the 2009 Provisions and ASCE 7-10 are instead based on the 2008 update of the
USGS ground motion values. This update is documented in Petersen et al. (2008) and supersedes the
1996 and 2002 USGS ground motions values. It involved interactions with hundreds of scientists and
engineers at regional and topical workshops, including advice from working groups, expert panels, state
geological surveys, other federal agencies and hazard experts from industry and academia. Based in large
part on new published studies, the 2008 update incorporated changes in both earthquake source models
(including magnitudes and occurrence frequencies) and models of ground motion propagation. The
UCERF and NGA models mentioned above are just two examples of such changes. The end results are
updated ground motions that represent the best available science as determined by the USGS from an
extensive information-gathering and review process.
It is important to note that the 2008 USGS hazard curves and uniform-hazard maps (posted at
http://earthquake.usgs.gov/hazards/products/conterminous/2008/), like their 2002 counterparts, represent
the geomean ground motions discussed in the preceding subsection. Only the MCER ground motions
and their underlying probabilistic and deterministic ground motions represent the maximum direction of
horizontal spectral response acceleration.
3.1.2.5 Differing Presentation of MCER Ground Motions in the Provisions and in ASCE 7-10
Though their numerical values are identical, the MCER ground motions specified in the Provisions and in
ASCE 7-10 are presented differently. As replacements to the MCE ground motion maps in ASCE 7-05,
ASCE 7-10 presents (in Chapter 22) contour maps of the MCER ground motions for Site Class B, which
are still denoted SS and S1 for the 0.2- and 1.0-second spectral response accelerations, respectively. Like
the MCE ground motions in ASCE 7-05, the MCER ground motions mapped in ASCE 7-10 are accessible
electronically via a USGS web application (see http://earthquake.usgs.gov/designmaps/).
In contrast, Provisions Section 11.4 presents equations to calculate MCER ground motions (SS and S1) for
Site Class B using maps (in Chapter 22) of uniform-hazard 2% in 50-year ground motions (denoted SSUH
and S1UH), so-called risk coefficients (denoted CRS and CR1); and deterministic ground motions (denoted
SSD and S1D, not to be confused with the design ground motions SDS and SD1). The risk coefficient maps
show the ratio of the risk-targeted probabilistic ground motions (described in Section 3.1.2.1) to
corresponding 2% in 50-year ground motions like those used to derive the MCE ground motion maps in
ASCE 7-05. The intent of the equations and three sets of maps presented in the Provisions is
transparency in the derivation of the MCER ground motions. The mapped values of the uniform-hazard
ground motions, risk coefficients and deterministic ground motions are all accessible electronically via
http://earthquake.usgs.gov/designmaps/.

3-6

Chapter 3: Earthquake Ground Motion

3.1.3

PGA Maps in the Provisions and in ASCE 7-10

The basis of the Peak Ground Acceleration (PGA) maps in the Provisions and in ASCE 7-10 nearly
parallels that of the MCE ground motion maps in ASCE 7-05 described in Section 3.1.1.1. More
specifically, the mapped PGA values for Site Class B are calculated as the lesser of uniform-hazard (2%
in 50-year) probabilistic and deterministic PGA values that represent the geometric mean of two
horizontal components of ground motion. Unlike in ASCE 7-05, though, the deterministic values are
defined as 84th-percentile ground motions rather than 150% of median ground motions. This definition of
deterministic ground motions parallels that which is described above for the MCER ground motions in the
2009 Provisions and ASCE 7-10. The deterministic PGA values, though, are stipulated to be no lower
than 0.5g, as opposed to 1.5g and 0.6g (respectively) for the MCER 0.2- and 1.0-second spectral response
accelerations. All of these details of the basis of the PGA maps are provided in ASCE 7-10 Section 21.5;
the Provisions do not contain a site-specific procedure for PGA values.
The USGS-computed PGA values for vS,30 = 760m/s that are mapped, like their MCER ground motion
counterparts in the Provisions and in ASCE 7-10, are from the 2008 USGS update. Also like their MCER
ground motion counterparts, the 84th-percentile PGA values have been approximated as median values
multiplied by 1.8.
While the values on and format of the PGA maps in the Provisions and in ASCE 7-10 are identical, the
terminology used to label the maps (and values) is different in the two documents. In the Provisions, they
are referred to as MCE Geometric Mean PGA maps. In ASCE 7-10, they are labeled Maximum
Considered Earthquake Geometric Mean (MCEG) PGA maps. The MCEG abbreviation is intended to
remind users of the differences between the basis of the PGA maps and the MCE R maps also in ASCE 710, namely that the PGA values represent the geometric mean of two horizontal components of ground
motion, not the maximum direction; and that the probabilistic PGA values are not risk-targeted ground
motions, but rather uniform-hazard (2% in 50-year) ground motions.
3.1.4

Basis of Vertical Ground Motions in the Provisions and in ASCE 7-10

Whereas ASCE 7-05 determines vertical seismic load effects via a single constant fraction of the
horizontal short-period spectral response acceleration SDS, the 2009 Provisions and ASCE 7-10 determine
a vertical design response spectrum, Sav, that is analogous to the horizontal design response spectrum, Sa.
The Sav values are determined via functions (for four different ranges of vertical period of vibration) that
each depend on SDS and a coefficient Cv representing the ratio of vertical to horizontal spectral response
acceleration. This is in contrast to determination of Sa via mapped horizontal spectral response
accelerations. The coefficient Cv, in turn, depends on the amplitude of spectral response acceleration (by
way of SS) and site class. These dependencies, as well as the period dependence of the equations for Sav,
are based on studies by Bozorgnia and Campbell (2004) and others. Those studies observed that the ratio
of vertical to horizontal spectral response acceleration is sensitive to period of vibration, site class,
earthquake magnitude (for relatively soft sites) and distance to the earthquake. The sensitivity to the
latter two characteristics is captured by the dependence of Cv on SS.
The basis of the equations for vertical response spectra in the Provisions and in ASCE 7-10 is explained
in more detail in the commentary to Chapter 23 of each document. Note that for vertical periods of
vibration greater than 2 seconds, Chapter 23 stipulates that the vertical spectral response accelerations be
determined via a site-specific procedure. A site-specific study also may be performed for periods less
than 2 seconds, in lieu of using the equations for vertical response spectra.
3.1.5

Summary

While the new Risk-Targeted Maximum Considered Earthquake (MCER) ground motions in the
Provisions and in ASCE 7-10 are similar to the MCE ground motions in ASCE 7-05, in that they both
represent the lesser of probabilistic and deterministic ground motions, there are many differences in their

3-7

FEMA P-752, NEHRP Recommended Provisions: Design Examples

development. The definitions of the probabilistic and deterministic ground motions that underlie the
MCER ground motions were revised by the BSSC Seismic Design Procedures Reassessment Group
(SDPRG, or Project 07); and the hazard modeling upon which these ground motions are based was
updated by the USGS (in 2008). In particular, the underlying probabilistic ground motions were
redefined as so-called risk-targeted ground motions, which led to the new MCER ground motion
terminology.
The basis of the new Peak Ground Acceleration (PGA) maps in the Provisions and in ASCE 7-10 nearly
parallels that of the 0.2- and 1.0-second MCE spectral response accelerations in ASCE 7-05 (with one
important exception); new equations for vertical ground motion spectra are based on recent studies of the
ratio of vertical to horizontal ground motions. The long-period transition (TL) maps in the new documents
are the same as those in ASCE 7-05.
3.1.6

References

American Society of Civil Engineers. 1998. Minimum Design Loads for Buildings and Other Structures,
ASCE/SEI 7-98. ASCE, Reston, Virginia.
American Society of Civil Engineers. 2002. Minimum Design Loads for Buildings and Other Structures,
ASCE/SEI 7-02. ASCE, Reston, Virginia.
American Society of Civil Engineers. 2006. Minimum Design Loads for Buildings and Other Structures,
ASCE/SEI 7-05. ASCE, Reston, Virginia.
American Society of Civil Engineers. 2010. Minimum Design Loads for Buildings and Other Structures,
ASCE/SEI 7-10. ASCE, Reston, Virginia.
Applied Technology Council. 1978. Tentative Provisions for the Development of Seismic Regulations for
Buildings, ATC 3-06. ATC, Palo Alto, California.
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, 8:175-207.
Building Seismic Safety Council. 1997. NEHRP Recommended Provisions for Seismic Regulations for
New Buildings and Other Structures, Part 2: Commentary, FEMA 303. FEMA, Washington, D.C.
Building Seismic Safety Council. 2000. NEHRP Recommended Provisions for Seismic Regulations for
New Buildings and Other Structures, Part 2: Commentary, FEMA 369. FEMA, Washington, D.C.
Building Seismic Safety Council. 2003. NEHRP Recommended Provisions and Commentary for Seismic
Regulations for New Buildings and Other Structures, FEMA 450. FEMA, Washington, D.C.
Building Seismic Safety Council. 2009. NEHRP Recommended Seismic Provisions for New Buildings
and Other Structures, FEMA P-750. FEMA, Washington, D.C.
Cornell, C.A. 1968. Engineering Seismic Risk Analysis, Bulletin of the Seismological Society of
America, 58(5):1583-1606.
Crouse C.B., E.V. Leyendecker, P.G. Somerville, M. Power and W.J. Silva. 2006. Development of
Seismic Ground-Motion Criteria for the ASCE 7 Standard, in Proceedings of the 8th US National
Conference on Earthquake Engineering. Earthquake Engineering Research Institute, Oakland,
California.
Field, E.H., T.E. Dawson, K.R. Felzer, A.D. Frankel, V. Gupta, T.H. Jordan, T. Parsons, M.D. Petersen,
R.S. Stein, R.J. Weldon and C.J. Wills. 2008. The Uniform California Earthquake Rupture Forecast,
Version 2 (UCERF 2), USGS Open File Report 2007-1437 (http://pubs.usgs.gov/of/2007/1437/).
USGS, Golden, Colorado.

3-8

Chapter 3: Earthquake Ground Motion

Frankel, A.D., M.D. Petersen, C.S. Mueller, K.M. Haller, R.L. Wheeler, E.V. Leyendecker, R.L. Wesson,
S.C. Harmsen, C.H. Cramer, D.M. Perkins and K.S. Rukstales. 2002. Documentation for the 2002
Update of the United States National Seismic Hazard Maps, USGS Open File Report 02-420
(http://pubs.usgs.gov/of/2002/ofr-02-420/). USGS, Golden, Colorado.
Huang, Y.-N., A.S. Whittaker and N. Luco, 2008. Maximum Spectral Demands in the Near-Fault
Region, Earthquake Spectra, 24(1):319-341.
Leyendecker, E.V., R.J. Hunt, A.D. Frankel and K.S. Rukstales. 2000. Development of Maximum
Considered Earthquake Ground Motion Maps, Earthquake Spectra, 16(1):21-40.
Luco, N. B.R. Ellingwood, R.O. Hamburger, J.D. Hooper, J.K. Kimball and C.A. Kircher. 2007. RiskTargeted versus Current Seismic Design Maps for the Conterminous United States, in Proceedings
of the SEAOC 76th Annual Convention. Structural Engineers Association of California, Sacramento,
California.
Petersen, M.D., A.D. Frankel, S.C. Harmsen, C.S. Mueller, K.M. Haller, R.L. Wheeler, R.L. Wesson, Y.
Zeng, O.S. Boyd, D.M. Perkins, N. Luco, E.H. Field, C.J. Wills and K.S. Rukstales. 2008.
Documentation for the 2008 Update of the United States National Seismic Hazard Maps, USGS Open
File Report 2008-1128 (http://pubs.usgs.gov/of/2008/1128/). USGS, Golden, Colorado.

3.2 DETERMINATION OF GROUND MOTION VALUES AND SPECTRA

This example illustrates the determination of seismic design parameters for a site in Seattle, Washington.
The site is located at 47.65N latitude, 122.3W longitude. Using the results of a site-specific
geotechnical investigation and the procedure specified in Standard Chapter 20, the site is classified as Site
Class C. (This is the same site used in Design Example 6.3.)
In the sections that follow design ground motion parameters are determined using ASCE 7-05, the 2009
Provisions and ASCE 7-10. Using the 2009 Provisions, horizontal response spectra, vertical response
spectra and peak ground accelerations are computed for both design and maximum considered earthquake
ground motions.
3.2.1

ASCE 7-05 Ground Motion Values

ASCE 7-05 Section 11.4.1 requires that spectral response acceleration parameters SS and S1 be determined
using the maps in Chapter 22. Those maps are too small to permit reading values to a sufficient degree of
precision for most sites, so in practice the mapped parameters are determined using a software application
available at www.earthquake.usgs.gov/designmaps. That application requires that longitude be entered in
degrees east of the prime meridian; negative values are used for degrees west. Given the site location, the
following values may be determined using the online application (or read from Figures 22-1 and 22-2).
SS = 1.306
S1 = 0.444
Using these mapped spectral response acceleration values and the site class, site coefficients Fa and Fv are
determined in accordance with Section 11.4.3 using Tables 11.4-1 and 11.4-2. Using Table 11.4-1, for SS
= 1.306 > 1.25, Fa = 1.0 for Site Class C. Using Table 11.4-2, read Fv = 1.4 for S1 = 0.4 and Fv = 1.3 for
S1 0.5 for Site Class C. Using linear interpolation for S1 = 0.444,

3-9

FEMA P-752, NEHRP Recommended Provisions: Design Examples

Fv = 1.4 +

0.444 0.4
(1.3 1.4) = 1.356
0.5 0.4

Using Equations 11.4-1 and 11.4-2 to determine the adjusted maximum considered earthquake spectral
response acceleration parameters,

SMS = Fa SS = 1.0(1.306) = 1.306

SM 1 = Fv S1 = 1.356(0.444) = 0.602
Using Equations 11.4-3 and 11.4-4 to determine the design earthquake spectral response acceleration
parameters,

2
2
S DS = S MS = (1.306) = 0.870
3
3
2
2
S D1 = S M 1 = (0.602) = 0.401
3
3
Given the site location read Figure 22-15 for the long-period transition period, TL = 6 seconds.
3.2.2

2009 Provisions Ground Motion Values

Part 1 of the 2009 Provisions modifies Chapter 11 of ASCE 7-05 to update the seismic design ground
motion parameters and procedures as described in Section 3.1.2 above. Given the site location, the
following values may be determined using the online application (or read from Provisions Figures 22-1
through 22-6).
SSUH = 1.305
S1UH = 0.522
CRS = 0.988
CR1 = 0.955
SSD = 1.5
S1D = 0.6
UH and D appear, respectively, in the subscripts to indicate uniform hazard and deterministic values
of the spectral response acceleration parameters at short periods and at a period of 1 second, SS and S1.
CRS and CR1 are the mapped risk coefficients at short periods and at a period of 1 second. S1D should not
be confused with SD1, which is computed below.
The spectral response acceleration parameter at short periods, SS, is taken as the lesser of the values
computed using Provisions Equations 11.4-1 and 11.4-2.
SS = CRS SSUH = 0.988(1.305) = 1.289
SS = SSD = 1.5
Therefore, SS = 1.289.
The spectral response acceleration parameter at a period of 1 second, S1, is taken as the lesser of the
values computed using Provisions Equations 11.4-3 and 11.4-4.

3-10

Chapter 3: Earthquake Ground Motion

S1 = CR1 S1UH = 0.955(0.522) = 0.498
S1 = S1D = 0.6
Therefore, S1 = 0.498.
Using these spectral response acceleration values and the site class, site coefficients Fa and Fv are
determined in accordance with Section 11.4.3 using Tables 11.4-1 and 11.4-2 (which are identical to the
Tables in ASCE 7-05). Using Table 11.4-1, for SS = 1.289 > 1.25, Fa = 1.0 for Site Class C. Using Table
11.4-2, read Fv = 1.4 for S1 = 0.4 and Fv = 1.3 for S1 0.5 for Site Class C. Using linear interpolation for
S1 = 0.498,

Fv = 1.4 +

0.498 0.4
(1.3 1.4) = 1.302
0.5 0.4

Using Provisions Equations 11.4-5 and 11.4-6 to determine the MCER spectral response acceleration
parameters,

SMS = Fa SS = 1.0(1.289) = 1.289

SM 1 = Fv S1 = 1.302(0.498) = 0.649
Using Provisions Equations 11.4-7 and 11.4-8 to determine the design earthquake spectral response
acceleration parameters,

2
2
S DS = S MS = (1.289) = 0.859
3
3
2
2
S D1 = S M 1 = (0.649) = 0.433
3
3
Given the site location read Provisions Figure 22-7 for the long-period transition period, TL = 6 seconds.
3.2.3

ASCE 7-10 Ground Motion Values

The seismic design ground motion parameters and procedures in Chapter 11 of ASCE 7-10 are consistent
with those in the 2009 Provisions. Given the site location, the following values may be determined using
the online application (or read from ASCE 7-10 Figures 22-1 and 22-2).
SS = 1.289
S1 = 0.498
Using these spectral response acceleration values and the site class, site coefficients Fa and Fv are
determined in accordance with Section 11.4.3 using Tables 11.4-1 and 11.4-2 (which are identical to the
Tables in ASCE 7-05 and in the 2009 Provisions). Using Table 11.4-1, for SS = 1.289 > 1.25, Fa = 1.0 for
Site Class C. Using Table 11.4-2, read Fv = 1.4 for S1 = 0.4 and Fv = 1.3 for S1 0.5 for Site Class C.
Using linear interpolation for S1 = 0.498,

Fv = 1.4 +

0.498 0.4
(1.3 1.4) = 1.302
0.5 0.4

3-11

FEMA P-752, NEHRP Recommended Provisions: Design Examples

Using Equations 11.4-1 and 11.4-2 to determine the MCER spectral response acceleration parameters,

SMS = Fa SS = 1.0(1.289) = 1.289

SM 1 = Fv S1 = 1.302(0.498) = 0.649
Using Equations 11.4-3 and 11.4-4 to determine the design earthquake spectral response acceleration
parameters,

2
2
S DS = S MS = (1.289) = 0.859
3
3
2
2
S D1 = S M 1 = (0.649) = 0.433
3
3
Given the site location read ASCE 7-10 Figure 22-12 for the long-period transition period, TL = 6
seconds.
The procedure specified in ASCE 7-10 produces seismic design ground motion parameters that are
identical to those produced using the 2009 Provisionsbut in fewer steps.
3.2.4

Horizontal Response Spectra

The design spectrum is constructed in accordance with Provisions Section 11.4.5 using Provisions Figure
11.4-1 and Provisions Equations 11.4-9, 11.4-10 and 11.4-11. The design spectral response acceleration
ordinates, Sa, may be divided into four regions based on period, T, as described below.

S D1
0.433
= 0.2
= 0.101 seconds, Sa varies linearly from 0.4SDS to SDS.
S DS
0.859
S
0.433
From T0 to TS = D1 =
= 0.504 seconds, Sa is constant at SDS.
S DS 0.859
From TS to TL, Sa is inversely proportional to T, being anchored to SD1 at T = 1 second.
S
At periods greater than TL, Sa is inversely proportional to the square of T, being anchored to D1 at TL.
TL

From T = 0 to T0 = 0.2

As prescribed in Provisions Section 11.4.6, the MCER response spectrum is determined by multiplying
the design response spectrum ordinates by 1.5. Figure 3-1 shows the design and MCER response spectra
determined using the ground motion parameters computed in Section 3.2.3.

3-12

Chapter 3: Earthquake Ground Motion

1.2
)
g(
a
S
,
n
iot
ar
el
ec
c
A
la
rt
ce
p
S

1.0

MCER spectrum

0.8
0.6
0.4
design
spectrum

0.2
T0

0.0
0

TL

TS
1

Period, T (s)

Figure 3-1 Horizontal Response Spectra for Design and MCER Ground Motions
3.2.5

Vertical Response Spectra

Part 1 of the 2009 Provisions adds a new chapter (Chapter 23) to ASCE 7-05 to define vertical ground
motions for seismic design. The design vertical response spectrum is constructed in accordance with
Provisions Section 23.1 using Provisions Equations 23.1-1, 23.1-2, 23.1-3 and 23.1-4. Vertical ground
motion values are related to horizontal ground motion values by a vertical coefficient, Cv, which is
determined as a function of site class and the MCER spectral response parameter at short periods, SS. The
design vertical spectral response acceleration ordinates, Sav, may be divided into four regions based on
vertical period, Tv, as described below.
Using Provisions Table 23.1-1, read Cv = 1.3 for SS 2.0 and Cv = 1.1 for SS = 1.0 for Site Class C. Using
linear interpolation for SS = 1.289,

Cv = 1.1 +

1.289 1
(1.3 1.1) = 1.158
2 1

From Tv = 0 to 0.025 seconds, Sav is constant at 0.3CvSDS = 0.3(1.158)(0.859) = 0.298. From Tv = 0.025 to
0.05 seconds, Sav varies linearly from 0.3CvSDS = 0.298 to 0.8CvSDS = 0.8(1.158)(0.859) = 0.796. From Tv
= 0.05 to 0.15 seconds, Sav is constant at 0.8CvSDS = 0.796. From Tv = 0.15 to 2.0 seconds, Sav is inversely
proportional to Tv0.75, being anchored to 0.8CvSDS = 0.796 at Tv = 0.15 seconds. For vertical periods
greater than 2.0 seconds, the vertical response spectral acceleration must be determined using site-specific
procedures.
As prescribed in Provisions Section 23.2, the MCER vertical response spectrum is determined by
multiplying the design vertical response spectrum ordinates by 1.5. Figure 3-2 shows the design and
MCER vertical response spectra determined using the ground motion parameters computed in Section
3.2.3.

3-13

FEMA P-752, NEHRP Recommended Provisions: Design Examples

1.2

)
g(

av

S
,
n
oi
t
rae
le
cc
A
la
rt
ce
p
S
la
ci
tr
e
V

MCER spectrum

1.0
0.8
0.6

design
spectrum

0.4
0.2
0.0
0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Vertical Period, Tv (s)

Figure 3-2 Vertical Response Spectra for Design and MCER Ground Motions
3.2.6

Peak Ground Accelerations

Part 1 of the 2009 Provisions modifies Section 11.8.3 of the ASCE 7-05 to update the calculation of peak
ground accelerations used for assessment of the potential for liquefaction and soil strength loss and for
determination of lateral earth pressures for design of basement and retaining walls. Given the site
location, the following value of maximum considered earthquake geometric mean peak ground
acceleration may be determined using the online application (or read from Provisions Figure 22-8).
PGA = 0.521 g
Using this mapped peak ground acceleration value and the site class, site coefficient FPGA is determined in
accordance with Section 11.8.3 using Table 11.8-1. Using Table 11.8-1, for PGA = 0.521 > 0.5, FPGA =
1.0 for Site Class C. Using Provisions Equation 11.8-1 to determine the maximum considered earthquake
geometric mean peak ground acceleration adjusted for site class effects,
PGAM = FPGA PGA = 1.0(0.521) = 0.521 g
This value is used directly to assess the potential for liquefaction or for soil strength loss. The design
peak ground acceleration used to determine dynamic seismic lateral earth pressures for design of
basement and retaining walls is computed as 2 3 PGAM = 2 3 (0.521) = 0.347 g.

3.3 SELECTION AND SCALING OF GROUND MOTION RECORDS

Response history analysis (whether linear or nonlinear) consists of the step-wise application of timevarying ground accelerations to a mathematical model of the subject structure. The selection and scaling
of appropriate horizontal ground motion acceleration time histories is essential to produce meaningful

3-14

Chapter 3: Earthquake Ground Motion

results. For two-dimensional or three-dimensional structural analysis, single-component or twocomponent records are used, respectively. The sections that follow discuss the approach to selection and
scaling of ground motion records as prescribed in the Provisions (and ASCE 7), illustrate the selection
and scaling of two-component ground motions for the structure analyzed in Design Example 6.3 located
at the site considered in Section 3.2 and discuss differences in the process for single-component ground
motions.
3.3.1

Approach to Ground Motion Selection and Scaling

In the simplest terms the goal of ground motion selection and scaling is to produce acceleration histories
that are consistent with the ground shaking hazard anticipated for the subject structure at the site in
question. As difficult as it is to forecast the occurrence of an earthquake, it is even more difficult to
predict the precise waveform and phasing of the resulting accelerations at a given site. Instead it is
necessary to approximate (somewhat crudely) what ground motions can be expected based on past
observations (and, possibly, geologic modeling). Provisions Section 16.1.3 prescribes the most
commonly applied approach to this process. While some aspects of the process are quite prescriptive,
others permit considerable latitude in application.
The Pacific Earthquake Engineering Research Center makes available a database of ground motions (at
http://peer.berkeley.edu/peer_ground_motion_database/) and a web application for the selection and
scaling of ground motions (PEER, 2010). As useful as that data and application are, they do not provide a
comprehensive solution to the challenge of ground motion selection and scaling in accordance with the
Provisions for all U.S. sites. Pertinent limitations include the following.

The database is limited to shallow crustal earthquakes recorded in active tectonic regimes, like
parts of the western U.S. It does not include records from subduction zone earthquakes, deep
intraplate events, or events in less active tectonic regimes (such as the central and eastern U.S.).

The web application allows use of a code design spectrum (from the Provisions or ASCE 7) as a
target and includes powerful selection and scaling methods. However, the set of selected and
scaled records produced would still require minor adjustment (scaling up) to satisfy the
requirements in Provisions Section 16.1.3 over the period range of interest.

3.3.1.1 Number of ground motions. In recognition of the impossibility of predicting the actual ground
motion history that should be expected, Section 16.1.3 requires the use of at least three ground motions in
any response history analysis. Where at least seven ground motions are used, Sections 16.1.4 and 16.2.4
permit the use of average response quantities for design. The difference is not one of statistical
significance; in either case mean response is approximated, but an incentive is given for the use of more
records, which could identify a potential sensitivity in the response. The objective of the response history
analyses is not to evaluate the response of the building for each record (since none of the records used
will actually occur), but to determine the expected (average) response quantities for use in design
calculations. If the analysis predicts collapse for one or more ground motions, the average cannot be
computed; the structure is deemed inadequate and must be redesigned.
3.3.1.2 Recorded or synthetic ground motions. Horizontal ground motion acceleration records should
be selected as single components (for two-dimensional analysis) or as orthogonal pairs (for threedimensional analysis) from actual recorded events. Where the number of appropriate recorded ground
motions is insufficient, use of simulated records is permitted. While generation of completely artificial
records is not directly prohibited, the intent (as expressed in Provisions Section C16.1.3) is that such
simulation is limited to modification for site distance and soil conditions.

3-15

FEMA P-752, NEHRP Recommended Provisions: Design Examples

3.3.1.3 Appropriate ground motions. The measure of appropriate applied to ground motions by the
Provisions is consistency with the magnitude, fault distance and source mechanism that control the
maximum considered earthquake. (Other characteristics of ground motion, such as duration, may
influence response, but are not addressed by the Provisions.) While it is good practice to select ground
motions with these characteristics in mind, the available data are quite limited. And even where the
available records are very carefully binned and match the target characteristics quite closely, they are far
from homogeneous.
As discussed in Section 3.1 the mapped ground motion parameters reflect the likelihood that a certain
level of spectral acceleration will be exceeded in a selected period, considering numerous sources of
earthquake ground shaking. While the mapping process does not sum accelerations from various sources
it does aggregate the probabilities of occurrence from those sources. As a result, it is impossible to
determine the controlling source characteristics using only the mapped acceleration parameters. In order
to identify the magnitude, fault distance and source mechanism that control the maximum considered
earthquake at a specific spectral period, it is necessary to deaggregate the hazard, which requires
reviewing the underlying calculations to note the relative contribution of each source. The USGS
provides tools to deaggregate hazard, providing results in three formats: a text tabulation, a graphic
presentation binned by distance and magnitude and a graphic presentation projected on a map. Figure 3-3
shows the two graphic formats for the 2-second period spectral acceleration with a 2% probability of
exceedance in 50 years (the maximum considered earthquake) at the site considered in Section 3.2.

3-16

Chapter 3: Earthquake Ground Motion

Figure 3-3 Graphic results of deaggregation

At most sites deaggregation of hazard reveals that a single source controls the maximum considered
earthquake ground motions for all spectral periods. However, at some sites different sources control the
maximum considered earthquake ground motions at different spectral periods. Figure 3-4 shows, for one
such site, the maximum considered earthquake response spectrum generated from mapped ground motion

3-17

FEMA P-752, NEHRP Recommended Provisions: Design Examples

parameters as well as median acceleration response spectra for two of the contributing sources. Since the
shape of the uniform hazard spectrum, upon which the design spectrum is based, is artificial (arising from
the probabilistic seismic hazard analysis rather than the characteristics of recorded ground motions), there
may be conservatisms involved in providing an aggregate match for design purposes (PEER, 2010).
However, that aggregate match is exactly what the Provisions requires, so it is prudent to consider how
that conservatism may best be balanced.

mapped
a

S
,
n
oi
t
ar
el
ec
c
A
l
ar
tc
e
p
S

source 1

source 2

period range
of interest

T1

Period, T

Figure 3-4 Response spectra for a site with multiple controlling sources
In this example, source 1 can generate moderate magnitude events close to the site; Source 2 can generate
very large magnitude events far from the site. Due to differing source and attenuation characteristics,
each source can control a portion of the MCER response spectrum. The response of short period
structures or very long period structures will be governed by source 1 or source 2, respectively. However,
the controlling source is less clear for a structure with a fundamental period shown as T1 in the figure.
Source 2 appears to control at period T1, but as discussed in Section 3.3.1.5, the Provisions defines a
wider period range of interest over which the selected ground motions must be appropriate. As outlined
below, three approaches are readily apparent.

3-18

First (and arguably most technically correct), select two full sets of (seven or more) ground
motion records conditionallyone set for each source, enveloping the MCER spectrum for the
portion of the period range of interest controlled by that source. Since the corresponding
portions of the actual and target spectral shapes would be similar, scale factors would be modest.
In this case, an independent series of analyses would be performed for each set of ground motion
records. Mean response parameters of interest could be computed for each set of analyses and the
more conservative of the two mean values for each response parameter could be used for design
verification. Although this approach has technical appeal, the Provisions do not outline such a

Chapter 3: Earthquake Ground Motion

procedure that makes use of two sets of ground motions, instead requiring use of a single set that
on average envelops the entire period range of interest of the target spectrum.

Second, select a full set of ground motions consistent with Source 2 and then scale the set to
envelop the much differently shaped MCER spectrum over the specified period range of interest.
While permitted by the Provisions, this approach would require large scale factors that
unrealistically exaggerate the long period response. It may seem that this set of ground motions
has a desired degree of homogeneity, but that comes at the expense of a very poor fit for the
average.

Third, select a set with some ground motions for each controlling source type. Select individual
scale factors so that the average of their linear elastic spectra envelops the target spectrum (as
required by the Provisions) and is shaped similarly to the target. As a result of this process,
records consistent with Source 1 will control short periods and those consistent with Source 2 will
control long periods. The scale factors will be somewhat larger than those required by the first
(conditional) approach, but not excessively large like those in the second approach. Although the
record set is less homogeneous than that used in the second approach, the average is much closer
to the target. Where used for linear response history analysis, this approach will produce average
response quantities consistent with the average linear response spectrum used in the scaling
process. Where used for nonlinear response history analysis, this approach (which uses scale
factors that are larger than those for the conditional approach) will bias the average response
quantities to be slightly more conservative and may increase the prediction of response extremes
(collapse). This third method is commonly employed by seismological consultants where
multiple source types may govern.

3.3.1.4 Scale factors. The most commonly employed ground motion scaling method involves
multiplying all of the acceleration values of the time-acceleration pairs by a scalar value. This timedomain scaling modifies the amplitude of the accelerations (to approximate changes in source magnitude
and/or distance) without affecting frequency content or phasing. Although not limited by the Provisions,
the scale factors applied to recorded ground motions should be modest (usually falling between 1/3 and
3); if very small or very large scale factors are needed, some aspect of the event that produced the source
motion likely is inconsistent with the maximum considered earthquake being modeled. An identical scale
factor is applied to both components of a given ground motion to avoid unrealistically biasing one
direction of response. Since the response spectra for time-domain scaled ground motions retain their
natural jaggedness, the acceptance criterion compares their average to the target spectrum, without
imposing limits on the scaling of individual ground motions. That means that there is no single set of
scale factors that may be applied to the selected ground motions (as discussed further in Provisions Part 2
Section C16.1.3.2)
Another ground motion scaling method involves transforming the time-acceleration data into the
frequency domain (such as by means of the fast Fourier transform), making adjustments (to match exactly
the target spectrum at multiple, specific frequencies) and transforming back into the time domain. This
method affects amplitude, frequency content and phasing (and tends to increase the total input energy).
This method makes it possible to estimate mean response with fewer ground motions, but may obscure
somewhat the potential variability of response. Use of this method is permitted by the Provisions, but the
same number of records is required as for time-domain scaling. Given the jaggedness of individual
response spectra, the process of spectral matching (which produces smoother spectra) requires scale
factors that can be considerably smaller or larger than those used in time-domain scaling. Since this
method applies numerous scale factors to differing frequencies of each ground motion component in order
to match spectral ordinates, there is no requirement that the two components be scaled identically. As the
spectral ordinates of frequency-domain scaled records may fall below the target spectrum at frequencies
3-19

FEMA P-752, NEHRP Recommended Provisions: Design Examples

other than those used for matching, a second round of (minor) scaling is needed to satisfy the Provisions
requirements.
Where single-component records are being selected for two-dimensional analysis the design response
spectrum is used as a target; and Provisions Section 16.1.3.1 requires that the average of the response
spectra not fall below the target over the period range of interest. A different approach is needed where
two-component records are being selected for three-dimensional analysis. The code writers selected the
square root of the sum of the squares (SRSS) of the response spectra for the two components as a measure
of the ground motion amplitude for each record. However, the SRSS of two spectra is always larger than
the average (and larger than the maximum). In practical terms for ground motion, it is reasonable to
expect that the SRSS is larger than the average by a factor of 1.4 to 1.5 and is larger than the maximum
(resultant) by a factor of about 1.2.
The code writers decided that it is sufficiently conservative to scale two-component records such that the
average of the SRSS spectra does not fall below the target over the period range of interest. Given the
relationship between SRSS and average, that means that scale factors for ground motions used in threedimensional analysis are only 2/3 of those for ground motions used in two-dimensional analysis. The
rationale is that a three-dimensional analysis (using two-component ground motions) subjects the
structure to the maximum (resultant) acceleration in some direction due to the interaction of ground
motion components, while that is not possible in two-dimensional analysis. Considering other
conservative criteria, such as fitting over the entire period range of interest, code writers accepted that the
resultant acceleration could be about 20 percent less than the design acceleration at some periods. Note
that Provisions Section 16.1.3.2 erroneously requires that the average of the SRSS spectra not fall below
the MCER spectrum over the period range of interest. ASCE 7-10 corrects this error by requiring that the
average of the SRSS spectra be compared to the response spectrum used in the design (rather than to
the MCER response spectrum).
For the special case described in Section 3.3.1.7 below, both ASCE 7-10 and the Provisions require
scaling so that the maximum acceleration exceeds the MCER response spectrum. Apparently, this is an
error carried forward from the Provisions to ASCE 7-10. Like the rest of Section 16.1.3.2, the target
spectrum used for scaling should be the response spectrum used in the design rather than the MCE R
response spectrum (which is 1.5 times the design response spectrum).
3.3.1.5 Period range of interest. The smooth spectral acceleration response spectrum constructed using
mapped acceleration parameters (and site response coefficients) is a location-specific estimate of the
ground shaking hazard. No matter how carefully recorded ground motions are selected and scaled, it is
unrealistic to expect a close match to the smooth target spectrum over all periods. On the other hand,
selecting and scaling ground motions to match the target spectrum at the natural period for the
fundamental mode of vibration of a structure is not enough to produce reasonable estimates of response;
important aspects of structural response (including collapse) are affected by both higher modes of
response and period elongation due to yielding. To balance these realities, code writers have established a
period range of interest (with respect to the fundamental period, T) that extends from 0.2T (to capture
higher mode effects) to 1.5T (to include period elongation). Although yielding and period elongation
cannot occur in linear response history analysis, for simplicity of application ground motions are selected
and scaled considering the same period range of interest as for nonlinear response history analysis.
3.3.1.6 Orientation of ground motion components. Accelerometers record earthquake ground shaking
along the vertical axis and two horizontal (orthogonal) instrument axes. Acceleration records can be used
in the as-recorded orientation, but orientation in the directions normal to and parallel to the strike of the
causative fault (termed the fault-normal and fault-parallel directions, respectively) by means of a simple
trigonometric transformation permits greater seismological insights, since some ground motions recorded
3-20

Chapter 3: Earthquake Ground Motion

very close to the causative fault contain rupture directivity effects. The differences may be meaningful
for selection and scaling, application in analysis, or both. Since the orientation of instrument axes is
arbitrary and reorientation along the fault-normal and fault-parallel directions can provide additional
insight, it has become common (but not universal) to reorient all horizontal ground motion records in that
manner.
In the very common condition where a site is not within several miles of the controlling source, the
orientation of ground shaking is inconsequential, so the Provisions contain no general requirement to
consider orientation. As discussed in Section 3.3.1.7 below, there is a selection and scaling orientation
requirement (but no application orientation requirement) for sites close to active controlling faults.
Figure 3-5 shows the time series of two components of ground acceleration. Component 1 is faultnormal; component 2 is fault-parallel. What is not apparent in such traces is the interaction of the
components. Figure 3-6 shows an orbit plot of ground acceleration pairs (effectively zero-period
response) for the same recording. The maximum resultant acceleration occurs along a diagonal direction.
0.8
)g
(
1
t 0.4
en
n
o
p
o
m
0
oc
0
n
oi
ta
re -0.4
le
cc
A
-0.8

10

20

30

Time (s)

40

10

20

30

Time (s)

40

0.8
)g
(
2t 0.4
n
e
n
o
p
m
0
oc
n
0
oi
ta
re -0.4
le
cc
A
-0.8

Figure 3-5 Horizontal acceleration components for the 1989 Loma Prieta earthquake
(Saratoga Aloha Avenue recording station)

3-21

FEMA P-752, NEHRP Recommended Provisions: Design Examples

0.8

)
g(
2
p
m
o
C

0.4

Comp 1 (g)

0
-0.8

-0.4

0.4

0.8

-0.4

-0.8

Figure 3-6 Horizontal acceleration orbit plot for the 1989 Loma Prieta earthquake
(Saratoga Aloha Avenue recording station)
Unfortunately, the direction of maximum ground acceleration may or may not correspond to the direction
of maximum acceleration response at any other period and the direction of maximum response generally
differs at various periods. If bilinear oscillators with various fundamental periods are subjected to the
two-component acceleration record, response spectra like those in Figure 3-7 result. The uniaxial
response spectra in that figure are identified by component. The resultant response spectrum indicates
the maximum acceleration along any direction. The SRSS response spectrum is obtained by taking the
square root of the sum of the squares of the corresponding component response spectrum ordinates. The
case illustrated reflects a possible near-source condition: for long periods, the fault-normal component
(component 1) is much larger than the fault-parallel component and is very close to the maximum
(resultant) response.
The Provisions do not require application of ground motions in multiple possible orientations. Whether
using three, seven, or more pairs, it is acceptable to consider a single, arbitrary orientation of a given twocomponent pair. For example, analysis can be performed with Component 1 applied in the +X
direction without considering the implications of applying that component in the -X, +Y, -Y, or other
directions. Since the objective of the analyses is to estimate average response quantities, it may be
advisable (but is not required) to consider whether there is an unwanted directional bias in the selected
and scaled ground motions. For instance, in the common case where the controlling source should not
produce strongly directional response, records could be oriented when applied so that the average of the
component 1 spectra is similar to the average of the component 2 spectra. The much-less-common case,
where strongly directional response is expected, is discussed in Section 3.3.1.7. Section 12.4.4 of these
Design Examples outlines a more involved approach that is recommended for seismically isolated
structures.

3-22

Chapter 3: Earthquake Ground Motion

2.0

1.5
)
g(
a
S
,
n 1.0
oi
t
ar
el
ec
c
a
la 0.5
rt
ce
p
S

SRSS
Resultant
Component 1
Component 2

0.0
0.0

0.5

1.0

1.5

2.0
Period, T (s)

2.5

3.0

3.5

4.0

Figure 3-7 Horizontal acceleration response spectra for the 1989 Loma Prieta earthquake
(Saratoga Aloha Avenue recording station)
3.3.1.7 Sites close to controlling active faults. Ground motions at sites close to a causative fault can be
strongly directional. At such sites, the maximum long period ground motion often occurs in the faultnormal direction. The last paragraph of Provisions Section 16.1.3.2 addresses this case, where code
writers have judged that scaling should be more conservative than that achieved using the SRSS-based
method. Although this requirement is well intentioned, the specific language provides a degree of
additional conservatism that can vary greatly. The intent is that the maximum spectral acceleration for
the scaled motions exceeds the target response spectrum.
While it is often true that the fault-normal component is dominant at long periods, some near-field ground
motions show no directional bias and some are dominant in the fault-parallel direction. For instance, of
the 3182 records in the PEER Ground Motion Database (for shallow crustal earthquakes), only 109 have
pulse-like directional effects. Of those, 60 have pulses only in the fault-normal direction, 19 have pulses
only in the fault-parallel direction, and 30 have pulses in both directions. As discussed above, it is
acceptable to reorient all horizontal ground motion records to the fault-normal and fault-parallel
directions. However, that does not assure that the fault-normal component will coincide with the
maximum. Figure 3-8 shows response spectra for a ground motion where the fault-parallel direction
(component 2) dominates for long periods. Scaling such that the fault-normal component exceeds the
target response spectrum, as required in the last paragraph of Section 16.1.3.2, would force the maximum
well above the target response spectrum. To obtain the intended result, ground motions should be scaled
so that the average of the fault-normal dominant components is not less than the MCER response spectrum
used in the design for the period range from 0.2T to 1.5T. (Section 3.3.1.4 above explains why all of
Section 16.1.3.2 should refer to the response spectrum used in the design rather than to the MCER
response spectrum.)
While the Provisions set forth orientation requirements for the selection and scaling of ground motions at
sites close to controlling active faults, the orientation of ground motion components as applied in analysis
is not prescribed. After going to the effort of orienting records in the fault-normal and fault-parallel
directions and applying special rules for scaling in recognition of near-source effects, it would be prudent

3-23

FEMA P-752, NEHRP Recommended Provisions: Design Examples

(but not required) to apply the records in the analyses consistent with the fault-normal and fault-parallel
directions at the actual site.
2.0

)
(g

1.5

S
,
n 1.0
oi
t
ar
el
ec
c
a
la 0.5
rt
ce
p
S

SRSS
Resultant
Component 1
Component 2

0.0
0.0

0.5

1.0

1.5

2.0
Period, T (s)

2.5

3.0

3.5

4.0

Figure 3-8 Horizontal acceleration response spectra for 1999 Duzce, Turkey earthquake
(Duzce recording station)

3.3.2

Two-Component Records for Three Dimensional Analysis

Design Example 6.3 is a buckling restrained braced frame structure located at the Seattle, Washington site
considered in Section 3.2. Some aspects of the design are based on results from three-dimensional
nonlinear response history analysis performed in accordance with Provisions Section 16.2. This section
illustrates application of the procedures described in Section 3.3.1 for the selection and scaling of twocomponent ground motion records. Pertinent information from Sections 3.2 and 6.3.6.1 is summarized as
follows.

Location: 47.65N, 122.3W

Site Class C
SMS = 1.289
SM1 = 0.649
TL = 6 seconds
Tx = Ty = 2.3 seconds

The period range of interest is from 0.2 2.3 = 0.46 seconds to 1.5 2.3 = 3.45 seconds. If the two
fundamental translational periods differed, the period range of interest would extend from 0.2 times the
shorter period to 1.5 times the longer period.
The next step is to deaggregate the hazard, as discussed in Section 3.3.1.3, over the period range of
interest. Figure 3-9 shows the MCER (target) response spectrum and the relative contributions of three
important sources to spectral acceleration at periods between 0 and 4 seconds. For periods greater than
about 1.5 seconds, ground shaking hazard is controlled by very large, but distant, subduction zone events.
At shorter periods, hazard is controlled by deep intraplate events, with substantial contributions from
shallow crustal events. It is necessary to identify not only the magnitude of the controlling event, but also

3-24

Chapter 3: Earthquake Ground Motion

the distance and source type. Short and intermediate period response may be more important than long
period response (depending on the period of the structure).

1.5
)
g(
a
S
,
n
iot 1
ar
el
ec
c
a
la
rt
ce 0.5
p
S

0
0

d
r
az
a
h
ot
n
oi
t
u
ib
rt
n
oc
ev
ti
al
e
R

2
Period, T (s)

Cascadia subduction zone

M8.5 - 9.0
d = 110 - 134 km

Deep intraplate
M6.6 - 7.0
d = 65 - 74 km

Shallow crustal
M6.8 - 7.2
d = 8 - 13 km

Figure 3-9 MCER response spectrum and corresponding hazard contributions

Since the MCER response spectrum over the period range of interest is controlled by multiple sources
with substantially different spectral shapes, the procedure recommended in Section 3.3.1.3 is used. Table
3-1 provides key information for the selected ground motion records. Few large magnitude subduction
zone records are available. Records 1, 2 and 3 are for slightly smaller events than those that control the
long period hazard, but at closer distances. These differences are partially offsetting so the required scale
factors are acceptable. Record 4 is nearly a perfect match for the hazard that controls short period
response; and it is from a past occurrence of a similar event in the same region. Records 5, 6 and 7 are
from shallow crustal events with magnitude and distance appropriate for this site. Two of those records
include near-source velocity pulses. In a manner similar to that illustrated in Figure 3-4, the actual
spectra for the selected ground motion records control different periods of response. Figure 3-10 shows
the SRSS spectra for Records 1 and 4, along with the target (MCER) spectrum. The subduction zone
event (Record No. 1) dominates long period response; the deep intraplate event (Record No. 4) dominates
short period response.

3-25

FEMA P-752, NEHRP Recommended Provisions: Design Examples

Table 3-1 Selected and Scaled Ground Motions for Example Site
Record
No.
1
2
3
4
5
6
7

Year
2003
2003
1968
1949
1989
1999
1995

Earthquake name
Tokachi-oki, Japan
Tokachi-oki, Japan
Tokachi-oki, Japan
Western Washington
Loma Prieta
Duzce, Turkey
Kobe, Japan

M
8.3
8.3
8.2
7.1
6.9
7.1
6.9

Source type
Subduction zone
Subduction zone
Subduction zone
Deep intraplate
Shallow crustal
Shallow crustal
Shallow crustal

Recording station
HKA 094
HKD 092
Hachinohe (S-252)
Olympia
Saratoga -- Aloha Ave
Duzce
Nishi-Akashi

Distance
(km)
67
46
71
75
9
7
7

Scale
factor
2.99
0.96
1.28
1.92
1.28
0.85
1.18

Record no. 4

)
g( 1.5
a
S
,
n
oi
t
ar
el 1
ec
c
a
la
rt
ce
p0.5
S

Record no. 1

0
0

2
Period, T (s)

Figure 3-10 SRSS response spectra from different source types

Figure 3-11 compares the average of the SRSS spectra for the selected ground motions with the target
(MCER) response spectrum. It also shows the period range of interest for ground motion selection for this
structure. In an average sense the suite of ground motions provides a very good fit to the target. Since
seven records are used, average response quantities may be used in design. This suite so well matches the
target spectrum that it could be used with no modification for periods from 0.18 to 4.95 seconds, a range
much wider than the period range of interest defined in Provisions Section 16.1.3.2. Since this suite of
ground motions has been selected and scaled to match the MCER response spectrum, an additional scale
factor of 2/3 must be applied when the records are used in an analysis to represent design-level
conditions.

3-26

Chapter 3: Earthquake Ground Motion

1.5

Average of SRSS spectra

)g
(
a
S
,
n1.0
iot
ar
el
ec
ca
l
ar
tc
e0.5
p
S

MCER
spectrum

For fundamental structural period of 2.30 s,

the period range of interest is 0.46 to 3.45 s.
0.0
0.0

1.0

2.0

3.0

4.0

Period, T (s)

Figure 3-11 Fit of the selected suite of ground motion records to the target spectrum
(for three-dimensional analysis)
3.3.3

One-Component Records for Two-Dimensional Analysis

As discussed in Section 3.3.1.4, one-component records (for use in two-dimensional analysis) are selected
and scaled such that their average fits the design response spectrum, which is two-thirds of the MCER
response spectrum. Figure 3-12 compares the average of the 14 component spectra (for the records
selected and scaled in Section 3.3.2) to the design response spectrum. These records provide an excellent
fit to the target spectrum. The suite of 14 records could be used without modification. If a subset of
seven records were selected, some minor adjustment to scale factors might be required. Figure 3-12 also
shows the average of the SRSS spectra for those 14 scaled records. As observed in Section 3.3.1.4, the
average of the SRSS spectra is about 1.5 times the average of the component spectra. Therefore, if the
same suite of records was used for three-dimensional analysis, the scale factors required would be about
2/3 of those required for two-dimensional analysis, due to the difference between average and SRSS
spectra (and not due to the purely coincidental 2/3 relationship between design and MCE R response
spectra).

3-27

FEMA P-752, NEHRP Recommended Provisions: Design Examples

Average of
SRSS spectra

Spectral acceleration, Sa (g)

1.5

1.0

Average of
component spectra

0.5

Design
spectrum

0.0

For fundamental structural period of 2.30 s,

the period range of interest is 0.46 to 3.45 s.

0.0

1.0

2.0

3.0

Period, T (s)

Figure 3-12 Fit of the selected suite of ground motion records to the target spectrum
(for two-dimensional analysis)

3.3.4

References

PEER. 2010. Technical Report for the PEER Ground Motion Database Web Application, Pacific
Earthquake Engineering Research Center, Berkeley, California.

3-28

4.0