How Does the Ground Shake?

1. Arthur D. Frankel*[HN17]
See all authors and affiliations
Science 26 Mar 1999:
Vol. 283, Issue 5410, pp. 2032-2033
DOI: 10.1126/science.283.5410.2032

When a multistory building is subjected to ground shaking from an

earthquake, elastic waves travel up the structure, with some of the energy
reflected at each floor and the remainder reflected from the top of the building
[HN1]. As the shaking continues, the structure begins to vibrate at various
frequencies. Earthquakes generate ground motions over a wide range of
frequencies, from static offsets to tens of cycles per second [hertz (Hz)]. Most
man-made structures have natural frequencies of vibration between about 0.1
and 20 Hz; a typical 10-story building has a natural frequency around 1 Hz.
Each structure is most sensitive to ground motions with frequencies near its
natural frequency. Damage to a building thus depends on its properties and
on the character of the earthquake ground motions, such as peak acceleration
and velocity, duration, frequency content, kinetic energy, and phasing and
coherence.

One of the major goals of modern seismology is the prediction of the time
series (or “time histories”) of the ground motions [HN2] at specific locations
when a large earthquake occurs on a particular fault. These artificial time
histories are then used to model [HN3] the response and improve the
resistance of structures such as buildings, bridges, or power plants to damage
from ground shaking. Many time histories have been recorded at sites near
earthquakes in the western United States and elsewhere, although these are
mostly for earthquakes with magnitudes (M) £ 7. In producing artificial time
histories for engineering applications, the most attention has been paid to
simulating horizontal shear-wave motion (where ground motion is
perpendicular to the direction of wave propagation), because it is most
damaging to structures. On page 2045, O'Connell (1) [HN4] considers an
inherent property of Earth's crust that is often overlooked when ground
motions are simulated: its randomness.

Predicting earthquake ground motions requires a detailed description of the

source, that is, the slip between opposite sides of the fault during the
earthquake rupture process[HN5], and of the path along which the seismic
waves [HN6] propagate from the fault to the site of interest. Variations of
material properties and stress in the crust occur over a wide range of spatial
scales, from the small-scale variety of minerals in a rock to the large-scale
complexity of a geologic map, affecting both source and path of the seismic
waves.

Earthquakes nucleate when the slip between sides of a fault accelerates in a

small patch. The interaction of the resulting propagating rupture with the
stress and strength variations on the fault generates seismic waves over a
broad frequency range. A fractal distribution of stress release on a fault, with
stress release independent of scale, can explain the white spectrum of ground
accelerations commonly observed above about 1 Hz for large earthquakes
(2).

Once generated, seismic waves are refracted and reflected [HN7] by

approximately horizontal boundaries in the crust. At shallow depth, seismic
velocities under a site can often be approximated by horizontal layers of soil
over bedrock. Multiple reflections within soil layers can cause resonances of
ground motions. Superimposed on this structure variation with depth are
lateral variations in rock type and composition, fractures, and fluid pressure.
Variations in seismic velocity and density on a scale of tens to hundreds of
meters scatter seismic waves with frequencies above about 0.5 Hz, producing
much of the tail of energy observed after the arrival of shear waves traveling
directly from the source. Such scattering can lower the coherence of the
shear-wave motion over the dimensions of the foundation of a building, which
can affect building response.

At frequencies below about 1 Hz, relatively simple deterministic models (3)

can reproduce the strong pulse of coherent ground motion observed at
locations in the direction of rupture propagation. At high frequencies above
about 5 Hz, the small-scale variation of stress on the fault and scatterers in
the crust affects the generation and propagation of seismic waves
substantially, and stochastic approaches (4) are used to match the duration or
envelope (or both) of the ground motion and its spectrum, rather than
attempting a “wiggle for wiggle” match to observed seismograms [HN8].

In the important frequency range of 0.5 to 5 Hz, where many buildings have
their natural frequencies, the deterministic and stochastic approaches need to
be combined to achieve a reasonable match to observed seismograms and to
produce synthetic time histories suitable for use in engineering design (5). It is
this problematic frequency range that O'Connell addresses in his research
article.

It is of particular importance to building codes and engineering design [HN9]

to establish what happens to seismic waves as they propagate through the
unconsolidated material at shallow depths beneath a site. Soil sites amplify
ground motions [HN10] more than rock sites, at least at frequencies of about
2 Hz and lower, because of the lower rigidity of soils. However, laboratory
experiments have demonstrated that soil behavior becomes nonlinear [HN11]
at the high strains achieved near large earthquakes. Such nonlinear behavior
would reduce the amplitude of seismic waves at frequencies above about 2
Hz and lower resonant frequencies caused by the soil. Modern building codes
such as the Uniform Building Code in the United States include nonlinear soil
behavior in their amplification factors.

The question of nonlinear behavior during earthquakes remains controversial,

particularly for the stiff soils that are common in most urban areas. Recent
studies have claimed evidence for a nonlinear response of stiff soils during
the M 6.7 Northridge earthquake [HN12] of 1994 (see the figure), where the
amplification between soil and rock sites was less for the mainshock shaking
than for the weaker aftershock motions (6). O'Connell contends that this
difference could be a direct result of seismic-wave scattering (1). If scattering
is more intense under soil sites than rock sites, it could cause more
incoherence of seismic energy at soil sites in the frequency range of 1 to 5
Hz. The seismic waves from the mainshock sample a wider zone of the crust
than those from an aftershock, and thus scattering effects under the sites
could cause different amplification factors for mainshock and aftershock
motions. O'Connell simulates wave propagation through a three-dimensional
random medium to show that scattering in a linear medium can, at least in
some cases, mimic one of the observations cited as evidence for soil
nonlinearity.

• Download high-res image

• Open in new tab
• Download Powerpoint
Factors affecting earthquake ground motions.

Time histories of ground acceleration at a soil site (top left) and a rock site (top
right) for the 1994 Northridge earthquake (M 6.7) show a brief but powerful set of
cycles caused by rupture directivity toward the sites. Note the higher peak
acceleration for the soil site. The map shows the distribution of peak velocity for this
earthquake derived from strong motion recordings (10). High peak velocities north of
the epicenter [HN16] were caused by the rupture propagating upward and northward
along the fault plane. The plot below the map shows the slip on the fault, the initial
rupture point (hypocenter), and the direction of rupture propagation (white arrow)
(10). Map and fault plane diagram from (11).
To resolve this question, we require a more thorough investigation of
scattering effects. We must determine the statistical properties of velocity and
density variations under typical rock and soil sites using borehole information
and recordings from closely spaced seismometers. It is important to
discriminate between nonlinear soil response and scattering. Nonlinear
response would cause a maximum limit to ground accelerations, whereas
scattering would produce occasional accelerations much larger than average.

Understanding the amplification of local soil sites is only one aspect of

predicting ground shaking in urban areas near active faults. Many high-
seismic risk areas are situated on deep sedimentary basins [HN13] with
relatively unconsolidated deposits of soil and sedimentary rock that can be
several kilometers thick. The three-dimensional geometry of these
sedimentary basins focuses seismic waves at certain locations, resulting in
narrow zones with destructive ground motions, as illustrated by the
1995 M 6.9 Kobe earthquake (7) [HN14]. Seismic waves can be trapped in
these basins as surface waves, causing low-frequency shaking over a long
time period, which causes particular damage to tall buildings. Three-
dimensional simulations of low-frequency ground motions (usually less than
1.0 Hz) in these basins have been performed (8), but our limited knowledge of
the deep basin configuration precludes more accurate prediction of strong
shaking in high-seismic risk areas such as Los Angeles, San Jose, and
Seattle.

Prediction of ground shaking will always involve deterministic and random

elements. The exact slip distribution and rupture directivity of future events on
a specific fault are unpredictable; time histories need to be generated for a
wide variety of rupture models. Focusing points in sedimentary basins are
likely to vary with the location of the earthquake. However, with detailed
knowledge of the basin structure, seismologists can simulate different
earthquake scenarios to develop maps of mean values of ground motion
parameters and their uncertainties. Probabilistic seismic hazard maps (9)
[HN15] that incorporate basin focusing effects and rupture directivity can be
developed by combining simulation results with the recurrence rates of large
earthquakes on specific faults derived from geological investigations.

Improved prediction of shaking and mitigation of damage in high-seismic risk

areas will require urban arrays of seismometers spaced about 1 km apart and
geophysical surveys to determine the detailed basin structure under these
areas. Specialized arrays with even closer spacing and borehole
seismometers are needed to resolve the scattering versus nonlinearity
problem. By recording small earthquakes, urban arrays can be used to verify
our simulations of ground motions before large events occur. When the large
shocks inevitably happen, these arrays of urban seismometers will be
essential in documenting the relation between ground motions and building
performance.

HyperNotes Related Resources on the World

Wide Web
General Hypernotes
Surfing the Internet for Earthquake Data is a collection of links to global
Web resources maintained by S. Malone, Geophysics Program, University of
Washington, Seattle.

The Montana Bureau of Mines and Geology offers an earthquake

glossary.

GeoRisk, a company that models natural hazards for risk management and
insurance purposes, provides a glossary of earthquake terminology.

The U.S. Geological Survey (USGS) offers an introduction to

earthquakes by K. Shedlock and L. Pakiser and provides a page with recent
worldwide earthquake listings and links to other USGS earthquake resources.

The USGS National Earthquake Information Center, Golden, CO, collects

and disseminates earthquake information; it presents a page of general
earthquake information.

SeismoLinks is a topical collection of links to seismology Web resources

provided by the USGS Pasadena, CA, office.

IRIS (Incorporated Research Institutions for Seismology), a consortium of

U.S. research programs in seismology, is dedicated to the collection and
distribution of seismographic data. It offers earthquake monitoring in near real
time, ground motion records, and links to seismic stations around the world.

The Southern California Earthquake Center presents an educational

module titled “Investigating earthquakes through regional seismicity.”

C. Ammon, Department of Earth and Atmospheric Sciences, St. Louis

University, provides lecture notes for a course on earthquakes.

R. Phinney, Department of Geosciences, Princeton University, offers lecture

notes on earthquakes for a course on earthquakes, volcanoes, and other
hazards. A glossary is provided.

T. Lay, Department of Earth Sciences, University of California, Santa Cruz,

provides lecture notes for a course on Earth catastrophes.

For a course on global change J. Park, Department of Geology and

Geophysics, Yale University, presents lecture notes on earthquakes.

The Reducing Earthquake Hazards Web site from the USGS Western
Region provides a series of fact sheets on how a wide range of loss
mitigation actions have reduced earthquake-induced losses.

The National Information Service for Earthquake Engineering (NISEE) at

the University of California, Berkeley, provides access to technical research
and development information in earthquake engineering and related fields.
An introduction to earthquake engineering principles by V. Bertero,
illustrated with photographs from W. Godden's structural engineering slide
library, is presented. Also provided are links to the earthquake research
centers sponsored by the National Science Foundation and other Web
information resources.

The Multidisciplinary Center for Earthquake Engineering

Research (MCEER), at the State University of New York at Buffalo, provides
an introduction to earthquakes and earthquake engineering in a collection of
FAQs, maintains a collection of links to other Internet resources, and offers
the Quakeline database. MCEER was originally established in 1986 by the
National Science Foundation as the National Center for Earthquake
Engineering Research (NCEER).

The Electronic Desktop Project, California State University, Los Angeles,

presents the Virtual Earthquake, an interactive computer program designed
as an introduction to how the epicenter and magnitude of an earthquake is
determined.

Understanding Earthquakes is an educational Web site presented by

the Institute for Crustal Studies, University of California, Santa Barbara.

New Scientist offers a special earthquakes report with links to articles and
Web resources about earthquakes.

The 1996 report titled “Strategy for national earthquake loss reduction”
prepared by the National Earthquake Strategy Working Group is made
available by the U.S. National Science and Technology Council.

Numbered Hypernotes
1. The Earthquake Maps and Information page from the Association of
Bay Area Governments provides descriptions of the effects of
earthquakes on different types of buildings. NISEE makes available
a paper by C. Arnold titled “The Nature of ground motion and its effect on
buildings.” “How Do Earthquakes Affect Buildings?” and “How Do
Buildings Respond to Earthquakes?” are presentations available
from MCEER. C. Ammon, Department of Earth and Atmospheric Sciences,
St. Louis University, discusses shaking effects of earthquakes on buildings in
his lecture notes on earthquake effects. The October 1996 NCEER
Bulletin had an article by A. Reinhorn, R. Valles-Mattox, and S. Kunnath
titled “Seismic damageability evaluation of a typical r/c building in the central
U.S.”

2. Time history is defined in the glossary from the Montana Bureau of Mines
and Geology. A paper titled “Attenuation and excitation of three-component
ground motion in southern California” by M. Raoof, R. B. Herrmann, and L.
Malagnini is made available by R. Herrmann, Department of Earth and
Atmospheric Sciences, St. Louis University. MCEER provides a guide to
obtaining earthquake strong motion/time history information. D. Wald, USGS
Pasadena office, provides a guide titled “Surfing the Internet for strong motion
data.” The United States Strong Motion Program maintains a national
cooperative instrumentation network, a national data center, and a data
analyses and research center in support of its mission.

3. The Hazards Mitigation Center at Lawrence Livermore National

Laboratory (LLNL) provides a report of its research titled “Calculating strong
ground motion due to earthquakes.” The Ground Motion Modeling group at
the Southern California Earthquake Center presents an introduction to their
research. An article titled “Synthetic strong ground motions for engineering
design utilizing empirical Green's functions” by L. Hutchings et al. is available
from the LLNL Hazards Mitigation Center. K. Olsen, Institute for Crustal
Studies, University of California, Santa Barbara, presents a collection of
reports on his ground motion modeling research. D. Wald, USGS Pasadena
office, makes available a conference paper titled “Estimation of uniformly
spaced, near-source, broadband ground motions for the 1994 Northridge
earthquake from forward and inverse modeling.” The Seismological
Laboratory, University of California, Berkeley, offers movies of simulated
seismic wave propagation using 3-D models; a press release about this
research is available.

4. D. O'Connell is in the Seismotectonics and Geophysics Group, U.S.

Bureau of Reclamation.

5. Slip is defined in the glossary of the Montana Bureau of Mines and

Geology. R. Phinney, Department of Geosciences, Princeton University,
provides an introduction to earthquake faulting. C. Ammon, Department of
Earth and Planetary Sciences, St. Louis University, provides lecture notes
on faults and faulting.

6. Seismic waves and related terms are defined in the glossary provided by
the USGS Cascades Volcano Observatory. The MichSeis Web site
defines S-waves and presents an introduction to seismic waves. T. Lay
presents lecture notes titled “Seismic waves: Faulting direction and earth
structure.” J. Louie, Seismological Laboratory, University of Nevada, Reno,
presents lecture notes on seismic waves for a course on earthquakes and
Earth structures. C. Ammon provides lecture notes on waves, seismograms,
and seismometers.

7. Reflection and refraction are defined in the glossary from the Montana
Bureau of Mines and Geology.

8. The GeoRisk glossary defines deterministic models and probabilistic

models. Guidelines for Evaluation and Mitigation of Seismic Hazards in
California from the California Division of Mines and Geology includes
a chapter titled “Estimation of earthquake ground-motion parameters” that
discusses deterministic and probabilistic seismic hazard analysis.

9. NISEE's Online Resources for Earthquake Engineering provides links to

organizations concerned with building standards and codes. General
Lessons in Earthquake Engineering is a collection of papers available from
NISEE; it includes a conference paper titled “Energy concepts and damage
indices” by A. Teran-Gilmore, Universidad Autonoma Metropolitana, Mexico
City. The USGS Western Region presents a fact sheet titled “Saving lives
through better design standards.” A 1994 report titled “Earthquake site
response and seismic code provisions” by G. Martin and R. Dobry is available
from MCEER. The National Earthquake Hazards Program of Canada
provides information on seismic hazard calculations. The John Blume
Earthquake Engineering Center, Stanford University, provides a collection
of overviews of their earthquake engineering research projects, many of
which could lead to improved building standards and building designs for
earthquake protection.

10. A section on the amplification of ground motion is included in

the report titled “Preliminary seismic microzonation assessment for British
Columbia,” which is available from the Resources Inventory Committee of
British Columbia. The USGS Western Region provides a feature titled “Soil
type and shaking hazard in the San Francisco Bay area.” A paper by J.
Prevost, G. Deodatis, and R. Popescu, Department of Civil Engineering and
Operations Research, Princeton University, titled “Integrated procedure for
dynamic analysis of fault-soil-structure systems” is available on the Soil
Dynamics and Geotechnical Earthquake Engineering Web page.

11. The LLNL Hazards Mitigation Center provides a brief introduction

to nonlinear response of geologic formations to strong ground motion.
A workshop report on nonlinear site response is provided by the online
edition of the SCEC Quarterly Newsletter. A conference paper by E. Jones
and K. Olsen titled “Three-dimensional finite-difference modeling of non-linear
ground motion” is made available by K. Olsen.

12. The Global Earthquake Response Center provides a collection

of links to Web resources on the Northridge earthquake. The USGS
Pasadena office maintains a Web page about research on the Northridge
earthquake. NISIEE provides a paper by S. Mahin titled “Lessons from steel
buildings damaged by the Northridge earthquake.” The Southern California
Earthquake Center Data Center offers an illustrated presentation on
the Northridge earthquake. “The local effects of strong ground shaking” is
a section of the report titled “USGS Response to an Urban Earthquake —
Northridge '94” available from the USGS Central Region.

13. The Quake Project is working on developing the capability for predicting,
by computer simulation, the ground motion of large basins during strong
earthquakes.

14. The Global Earthquake Response Center provides a collection of links

to Kobe earthquake information on the Web. The Geotechnical
Engineering Program, University of California, Berkeley, presents a
collection of research reports about the Kobe earthquake. NISEE provides a
collection of reports on engineering lessons from the Kobe earthquake.

15. The USGS Western Earthquake Hazards Team provides a fact

sheet titled “Hazard maps help save lives and property.” The National
Seismic Mapping Project, USGS Central Region, includes a brief
introduction to hazard maps in their collection of FAQs.
16. Epicenter and hypocenter are defined in the GeoRisk glossary.

17. A. D. Frankel is with the USGS National Seismic Hazard Mapping

Project.

References and Notes

1. ↵

1. D. R. H. O'Connell
, Science 283, 2045 (1999).
Abstract/FREE Full TextGoogle Scholar

2. ↵

1. A. Frankel
, J. Geophys. Res. 96, 6291 (1991);
Google Scholar

2. D. J. Andrews
, ibid. 86, 10821 (1981);
Google Scholar

3. T. C. Hanks
, ibid. 84, 2235 (1979);
Google Scholar

4. C. H. Scholz
, Bull. Seismol. Soc. Am. 88, 1325 (1998).
Abstract/FREE Full TextGoogle Scholar

3. ↵

1. D. J. Wald,
2. D. V. Helmberger,
3. T. H. Heaton
, Bull. Seismol. Soc. Am 81, 1540 (1991) P. G. Somerville, M. K. Sen,
B. P. Cohee, ibid., p. 1.
Abstract/FREE Full TextGoogle Scholar

4. ↵

1. T. C. Hanks,
2. R. K. McGuire
, ibid. 71, 2071 (1981) W. J. Silva et al. in Proceedings of the 4th U.S.
National Conference on Earthquake Engineering (Earthquake
Engineering Research Institute, Oakland, CA, 1990), vol 1, pp. 487–
494.
 Google Scholar

5. ↵

1. K. Kamae,
2. K. Irikura,
3. A. Pitarka
, Bull. Seismol. Soc. Am 88, 357 (1998).
Abstract/FREE Full TextGoogle Scholar

6. ↵

1. E. H. Field
2. et al.
, J. Geophys. Res. 103, 26869 (1998).
Google Scholar

7. ↵

1. H. Kawase
, Seismol. Res. Lett 67, 25 (1996).
Google Scholar

8. ↵

1. A. Frankel
, Bull. Seismol. Soc. Am 83, 1020 (1993);
Abstract/FREE Full TextGoogle Scholar

2. R. W. Graves
, ibid., p. 1042;
Google Scholar

3. K. B. Olsen,
4. R. J. Archuleta,
5. J. R. Matarese
, Science 270, 1628 (1995).
Abstract/FREE Full TextGoogle Scholar

9. ↵

USGS national seismic hazard maps, available

at http://geohazards.cr.usgs.gov/eq/.
Google Scholar

10. ↵

1. D. J. Wald,
 2. T. H. Heaton,
3. K. W. Hudnut
, Bull. Seismol. Soc. Am 86, S49 (1996).
Abstract/FREE Full TextGoogle Scholar

11. ↵

U. S. Geological Survey Open-File Report 96–263 (1996) (available

at http://geohazards.cr.usgs.gov/northridge).
Google Scholar