You are on page 1of 76

ENGINEERING

GEOLOGY
ELSEVIER Engineering Geology 44 (1996) 1-76 ,,

Use of liquefaction-induced features for paleoseismic analysis-


An overview of how seismic liquefaction features can be
distinguished from other features and how their regional
distribution and properties of source sediment can be used to infer
the location and strength of Holocene paleo-earthquakes
Stephen F. Obermeier 1
US Geological Survey, Reston, VA 20192, USA
Received 10 October 1995; accepted 23 February 1996

Abstract

Liquefaction features can be used in many field settings to estimate the recurrence interval and magnitude of strong
earthquakes through much of the Holocene. These features include dikes, craters, vented sand, sills, and laterally
spreading landslides. The relatively high seismic shaking level required for their formation makes them particularly
valuable as records of strong paleo-earthquakes. This state-of-the-art summary for using liquefaction-induced features
for paleoseismic interpretation and analysis takes into account both geological and geotechnical engineering
perspectives.
The driving mechanism for formation of the features is primarily the increased pore-water pressure associated with
liquefaction of sand-rich sediment. The role of this mechanism is often supplemented greatly by the direct action of
seismic shaking at the ground surface, which strains and breaks the clay-rich cap that lies immediately above the
sediment that liquefied. Discussed in the text are the processes involved in formation of the features, as well as their
morphology and characteristics in field settings. Whether liquefaction occurs is controlled mainly by sediment grain
size, sediment packing, depth to the water table, and strength and duration of seismic shaking. Formation of
recognizable features in the field generally requires a low-permeability cap above the sediment that liquefied. Field
manifestations are controlled largely by the severity of liquefaction and the thickness and properties of the low-
permeability cap.
Criteria are presented for determining whether observed sediment deformation in the field originated by seismically
induced liquefaction. These criteria have been developed mainly by observing historic effects of liquefaction in varied
field settings. The most important criterion is that a seismic liquefaction origin requires widespread, regional
development of features around a core area where the effects are most severe. In addition, the features must have a
morphology that is consistent with a very sudden application of a large hydraulic force.
This article discusses case studies in widely separated and different geological settings: coastal South Carolina, the
New Madrid seismic zone, the Wabash Valley seismic zone, and coastal Washington State. These studies encompass
most of the range of settings and the types of liquefaction-induced features likely to be encountered anywhere. The

1Tel.: 703 648 6791; fax; 703 648 6953 or 6717; e-mail: sober@usgs.gov.

0013-7952/96/$15.00 1996 Elsevier Science B.V. All rights reserved


PH S0013-7952 (96) 00040-3
2 S.I( Obermeier/Engineering Geology 44 (1996) 1 76

case studies describe the observed features and the logic for assigning a seismic liquefaction origin to them. Also
discussed are some types of sediment deformations that can be misinterpreted as having a seismic origin.
Two independent methods for estimating prehistoric magnitude are discussed briefly. One method is based on
determination of the maximum distance from the epicenter over which liquefaction-induced effects have formed. The
other method is based on use of geotechnical engineering techniques at sites of marginal liquefaction, in order to
bracket the peak accelerations as a function of epicentral distance: these accelerations can then be compared with
predictions from seismological models.

Key words." Liquefaction-induced features; Paleoseismic analysis; Seismic liquefaction features; Regional distribution:
Source sediment properties; Holocene paleo-earthquakes

1. Introduction the lower limit at which liquefaction effects become


relatively common (Ambraseys, 1988). In this arti-
This text focuses on the methods for determining cle, earthquake magnitude (M) is used rather
whether observed sediment deformation had a loosely as either moment magnitude or surface-
seismic shaking origin or a nonseismic origin. wave magnitude, whichever is larger.
Emphasis is placed on the process of liquefaction, Liquefaction has been severe and widespread in
which is the transformation of a granular material many earthquakes around the world. Effects of
from a solid state into a liquefied state as a liquefaction have many manifestations. Some note-
consequence of increased pore-water pressures worthy reports discuss the effects of the following
(oud, 1973). The discussion encompasses various earthquakes: the 1811-1812 New Madrid,
manifestations of liquefaction-induced deforma- Missouri, earthquakes (Fuller, 1912); the 1886
tion in fluvial and nearshore-marine deposits and Charleston, South Carolina, earthquake (Dutton,
the application of criteria for establishing an earth- 1889); earthquakes in northern California (Youd
quake origin. and Hoose, 1978); the 1964 Alaska earthquake
The systematic study of paleoliquefaction is a (US Geological Survey Professional Papers 542
young discipline. Accordingly, some of the physical through 545; separately authored chapters, which
parameters that control effects of liquefaction in are not cited herein, were published in 1965-1970;
the field are not completely understood. Still, the also, Walsh et al., 1995); various earthquakes in
principles and methods for conducting paleolique- Japan (O'Rourke and Hamada, 1989); and the
faction studies are sufficiently advanced to warrant earthquake of 1897 in India (Oldham, 1899). Only
their routine application in paleoseismic studies. rarely are the deformational effects of liquefaction
The methods in this text for conducting paleolique- illustrated in vertical section, as in this article, even
faction studies were developed largely out of neces- though this view is most useful for paleoliquefac-
sity in the United States, where the historic seismic tion studies. Noteworthy accounts of the vertical
record is particularly short. view were reported by Sieh (1978), Amick et al.
Paleoliquefaction studies are useful to engineers (1990), Audemard and de Santis (1991), Tuttle
and planners because of the high shaking threshold and Seeber (1991), Clague et al. (1992),
required to develop liquefaction features. The Wesnousky and Leffler (1992), Tuttle (1994), Sims
threshold is a horizontal acceleration on the order and Garvin (1995), and Walsh et al. (1995).
of 0.1 x g for strong earthquakes, even in highly Findings of exceptional interest in these reports
susceptible sediment (Ishihara, 1985, p. 352; are discussed herein where appropriate. The princi-
National Research Council, 1985, p. 34). pal bases for this text are my observations of
Worldwide data on historical earthquakes show liquefaction effects in widely diverse geological and
that features having a liquefaction origin can be geographic settings.
developed at earthquake magnitudes as low as The liquefaction effects described in this text are
about 5, but that a magnitude of about 5.5 6 is caused mainly by cyclic shaking of level or nearly
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 3

level ground. Primary seismological factors con- ment. Both types of features can mimic those of
tributing to liquefaction are the amplitude of the earthquake origin. Tests are suggested for inter-
cyclic shear stresses and the number of applications pretation of origin.
of the shear stresses (Seed, 1979). These factors, This text is intended primarily for geologists and
respectively, are related to field conditions of shak- secondarily for geotechnical engineers. A few
ing amplitude (that is, peak acceleration) and terms, though, are used strictly in the geotechnical
duration of strong shaking. Both peak acceleration context because of the lack of adequate geological
and duration generally correlate with the earth- equivalents for semiquantitative description of sed-
quake magnitude. Analytical engineering methods iment properties. Most of these terms are in
for evaluating variable and irregular cyclic stress Table 1, which relates the state of compactness
applications typical of real earthquakes are well (that is, the relative density) of sand to descriptors
developed and yield results acceptable for engineer- such as "very loose" or "very dense." In a similar
ing analysis (Seed et al., 1983), providing that sense, the term "clean sand" refers to a sand with
shaking amplitude-time records can be reasonably no silt or clay or bonding matter. The term "lique-
bracketed. In the text, there is a brief discussion faction susceptibility" refers to the ease with which
of how to set limits for the strength of prehis- a saturated sediment liquefies and is described with
toric shaking. qualifiers such as "very low" or "very high" (Youd
Pseudonodules and other such features caused and Perkins, 1978). To illustrate, a saturated, very
by the plastic deformation or flowage of very soft loose sand always has a very high liquefaction
muds and freshly deposited cohesionless sediments susceptibility. "Liquefaction potential" relates to
(often referred to as "syndepositional features" the likelihood of liquefaction occurring during a
or "soft-sediment deformations") are discussed specific earthquake at a particular strength of
briefly. Liquefaction is not required to deform shaking. Even a saturated, very loose sand has no
muds and extremely loose, freshly deposited cohe- liquefaction potential if the severity of shaking is
sionless sediments, although a high pore-water low enough.
pressure can be involved. The geological literature
is replete with articles attributing an earthquake
origin to deformed muds and convoluted sands. It 2. Overview of the formation of liquefaction-
must be kept in mind, however, that very weak induced features
sediments also are commonly deformed as a result
of other geological processes, such as loading The application of shear stresses causes a
during rapid sedimentation, localized artesian con- buildup of pore-water pressure, which in turn leads
ditions, or slumping. In addition, soft-sediment to liquefaction of saturated cohesionless sediment.
deformations often form at such low levels of For seismically induced liquefaction, these shear
seismic shaking that the shaking poses no hazard. stresses are due primarily to the upward propaga-
Thus the usefulness of these features for hazard tion of cyclic shear waves. Sediment on level
assessment normally is limited, and interpretations ground undergoes loading conditions as depicted
of origin often are equivocal. Interpretation of
origin usually is much more straightforward and Table 1
reliable for suspected liquefaction features than Relativedensityof sand as related to Standard Penetration Test
for soft-sediment deformations. Still, in some field blow counts (from Terzaghi and Peck (1967))
situations where it can be proven that the soft-
No. of blows(N) Relative density
sediment structures formed synchronously at wide-
spread locations, an earthquake origin can be 0-4 Very loose
attributed. 4-10 Loose
This text also contains discussion of features 10-30 Medium or moderate
formed by chemical weathering and of features 30-50 Dense
> 50 Very dense
deformed by processes in a periglacial environ-
4 S.F. Obermeier/Engineering Geology 44 (1996) 1- 76

~]tO() ~h
~ [ ~ <"- - ~ ~ Uncdn;olisidtated
~_:?}_--".' .z.....,,z-.,'.'-,z,~,.' Bedrock

TIME . . . .

Hypocenter Q

Fig. 1. Vertical section showing idealized field loading conditions changing with time as energy propagates upward from hypocenter.
Stresses shown represent preliquefaction cyclic loading conditions, cry, initial vertical effective overburden stress; Zh, earthquake-
induced horizontal cyclic shear stress.

in Fig. 1, with the shear stress applications being commonly originates at a depth ranging from a
somewhat random but nonetheless cyclic. few meters to about 10 m. Liquefaction takes place
Cohesionless sediments that are loosely packed only where the sediment is completely saturated.
tend to become more compact when sheared. The zone of liquefaction during shaking depends
Continued cyclic shearing can cause the pore-water on the relationship between the cyclic shear stresses
pressure to increase suddenly to the static confining generated by the earthquake and the stress required
pressure, leading to large strains and flowage of to initiate liquefaction in the sediment (Fig. 3).
the water and sediment. No appreciable change in Development of liquefaction is increasingly diffi-
volume of the deposit is required for this change cult with depth because the higher static vertical
in state from a solid-like to a viscous, liquid-like effective stress (total overburden pressure minus
material (that is, for liquefaction). The process is pore-water pressure) applied by the overburden
driven by breakdown of the packing arrangement greatly increases the resistance of sediment to
of grains as the grains attempt to move into a shearing and deformation.
more dense packing (Fig. 2). Sediment vented to the ground surface is the
Liquefaction during earthquake shaking most most conspicuous evidence of liquefaction at

A B C

Loose depositional Packing during Resedimentation,


packing liquefaction, with layer of
with excess water at top
pore-water pressure
throughout

Fig. 2. Changes in packing of sediment grains caused by liquefaction.


S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 5

SHEAR STRESS
>
\\

\\

Zone of liquefaction during


earthquake shaking

"1-
I--
IX.
IJ.l
E3 Cyclic stress required to
, ~ initiate liquefaction
in C cycles
%
X
\
\
\
\
\
\
\
\
\
Average cyclic shear stress '
developed for C cycles
by earthquake motions

Fig. 3. Depth of liquefaction induced during earthquake shaking, as affected by the factors of depth below ground surface and seismic
shear stress. C, number of shaking cycles. From Seed and Idriss (1971).

depth. Water from the zone of high pore-water


pressure must escape upward to cause venting. A
water-sediment mixture typically erupts suddenly
and violently to the surface through preexisting
holes or through fractures opened in the capping
material in response to liquefaction. In exceptional
situations, the mixture spouts as high as 6-7 m,
especially where flow is concentrated into holes
and cracks through an overlying finer grained cap
(Dutton, 1889; Fuller, 1912; Housner, 1958).
Water, sand, and silt can continue to flow to the
surface as "sand volcanoes" for hours after earth-
quake shaking has stopped. Sediment is left behind
on the ground surface in the form of cones, often
called sand blows or sand boils (Fig. 4). The cones
of sand can be as much as a meter in height and Fig. 4. Small sand blows near the town of El Centro from the
1979 Imperial Valley, California, earthquake. The sand blows
tens of meters in width.
were produced by a mixture of sand and water that spouted to
The increased pore-water pressure during the surface; they provide evidence of extensive liquefaction at
ground shaking can be manifested in other ways. depth. Length of white bar in foreground is about 15 crn.
The high pore-water pressure decreases the shear Photograph courtesy of R.F. Scott.
6 S. kl Obermeier/lz)tgineering Geoh~gy 44 (1996) 1 76

strength of granular strata at depth. These strata Fine-grained surficial l a y e r


can then fail in shear even where the ground E
~:- .I .'.: .~ :.. " ...'- :':
surface is inclined as gently as 0.1 5% (Youd,
1978: Youd and Bartlett, 1991). Huge masses of
overlying soil can shift horizontally in the form of
laterally moving landslides (called lateral spreads;
Fig. 5). Separation between individual blocks is
commonly as much as 2 3 m where shaking has
been especially strong. This separation tends to be
largest near stream banks or scarps in alluvium,
even if only a few meters in height, because these
breaks in slope are where there is less resistance
to lateral movements. Back-and-forth oscillation
Fig. 6 illustrates how separations in the cap can
form on level ground far from breaks in slope.
Fig. 6. Vertical section showing fissures and blocks resulting
Oscillation of the ground above the liquefied zone from oscillation of level ground far from breaks in slope (figure
causes fissures that separate blocks. These fissures from Youd, 1984a; triangle points to water table). Ground oscil-
are generally smaller than the largest separations lation can originate by fundamentally different mechanisms
formed during lateral spreading and only in excep- driven by surface waves (You& 1984a) or body waves (Pease
and O'Rourke, 1995). ( l ) In the mechanism driven by surface
tional cases are as much as 15 20cm in width
waves, liquefaction in the zone marked by diagonal lines decou-
(T.D. O'Rourke, Cornell University, oral com- pies the surface layer from the underlying liquefied layer. The
mun., 1995). decoupled layer vibrates at a different frequency than the under-
Liquefaction of granular deposits normally leads lying liquefied layer and the laterally surrounding firm ground:
to surface cracking and formation of localized as a result, fissures form between oscillating blocks and adjacent
firm ground. Travelling ground waves and opening and closing
depressions because of densification of liquefied
fissures are commonly seen during ground oscillation. (2) In
sediment after expulsion of water, even where there the mechanism driven by body waves, ground oscillation
is no evidence of venting of sand and water to the becomes severe in response to a resonant frequency effect in
which horizontal displacement in the cap is amplified with
respect to that of the underlying liquefied sand.
Fine-grained surficial layer
v \"
surface. These settlements can be as much as
0.25-0.5 m where thick sands liquefy severely
Initial section (Tokimatsu and Seed, 1987).
On slopes that exceed about 5%, severely lique-
fied sediments can cause huge landslides (some-
times called flow failures) that can flow as much
Generally increasing separation between blocks as tens to hundreds of meters. Ground disruption
~-~. -~ --~ ~ ~ _
can be so severe that it is difficult to establish what
the surface geometry was prior to failure. Strength
properties of materials in the failure zone can
Deform<d section be changed greatly. Flow failures aloug banks
Fig. 5. Vertical section showing lateral spread resulting from of streams and hillsides can also originate by
shaking of level to gently inclined ground near a break in slope. static (nonearthquake) mechanisms, however.
Liquefaction occurs in the zone marked by diagonal lines. The Determination that prehistoric landslide move-
surface layer then moves laterally toward the break and divides ment on a steep slope was seismically triggered
into blocks separated by fissures. Sand is vented to the surface
through some fissures, but other fissures are only partly filled. generally requires complex engineering testing and
The blocks can tilt and settle differentially with respect to one analysis difficult to perform in the best of circum-
another. Triangle points to water table. From Youd (1984a). stances, irrespective of whether liquefaction was
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 7

involved (Jibson and Keefer, 1993). Therefore, densely packed sand, the pore-pressure rise is less
sites on level ground are best for distinguishing a sudden, and the strength loss is less severe; thus,
seismic from a nonseismic origin. any increase in strain is less dramatic and is limited.
Clastic dikes on level to nearly level ground are Very densely packed sands may not develop pore-
the primary source of data used for paleoseismic water pressures high enough to cause liquefaction.
interpretations. Very important factors controlling For the purpose of this text, the condition of
the development and density of dikes include not highly elevated pore-water pressure with a signifi-
only the compactness and thickness of sediment cant loss of strength is referred to simply as
that liquefied but also cap thickness. Dikes almost "liquefaction." When the term "initial liquefac-
certainly form solely in response to hydraulic tion" is used, emphasis is on the elevated pore-
fracturing of the cap in a great many field situa- water pressure rather than loss of strength.
tions, but dike development can be enhanced The cyclic shear strain in sands that is required
greatly by the cap simply being pulled apart by to induce liquefaction usually is very small, being
lateral spreading or by oscillatory shaking at the less than a few percent (Pease and O'Rourke,
surface. 1995). For earthquakes having long durations and
many cycles, the critical shear strain can be
2.1. Process of liquefaction andfluidization extraordinarily small, as l o w as 0.04% (Dobry,
1989). Following liquefaction, densification can
The process of seismically induced liquefaction occur after water is expelled from the sediment.
of saturated granular sediments has been studied The vertical settlement caused by sediment densi-
extensively and is reasonably well understood fication often is quite small, being less than 2-3%
(Seed, 1979; National Research Council, 1985; of the height of the stratum that liquefied (Castro,
Castro, 1987; Dobry, 1989). Fig. 7 illustrates the 1987, p. 175).
typical field situation on level ground. A liquefiable Only minor disturbance to original stratification
sand layer is overlain by a thin, nonliquefiable may take place in that portion of the stratum
stratum, and the ground-water table is shallow. where only initial liquefaction has occurred. The
Earthquake-induced shear stresses propagate effects to bedding are often virtually indistinguish-
through the sand and cause shear strain of the able to the unaided eye. Closer inspection may
sediment structure. (Shear strain, 7, is the angle in show that platy minerals such as mica and clay
radians shown in Fig. 7; shear strain can also be are reoriented from their original flat-lying position
expressed as a percentage of the radians in a (and can form structures known as "dishes,"
complete circle.) Because grains attempt to move shown in Fig. 33), and thin laminations of finest
into a denser packing arrangement relatively constituents are warped. Small, steeply inclined,
quickly during back-and-forth shear straining, flame-like sand-rich structures known as "pillars"
water in the voids does not have time to escape. (see Fig. 34) can form where water collects beneath
The pore-water pressure can thereby increase. In a slightly less permeable lamina and then locally
loose deposits, stresses at the grain contacts can penetrates through the lamina to winnow out silt
approach zero, and concurrently the pore-water and other very fine constituents. This winnowing
pressure alone supports the weight of the overbur- takes place by a process often referred to as
den. The first time this occurs is often referred to "fluidization."
as "initial liquefaction" (National Research Fluidization occurs when flowing water exerts
Council, 1985, p. 42). A large loss of strength can sufficient drag or lift to momentarily suspend
take place once this condition is reached. In very grains of sediment. When a fluid is forced vertically
loosely packed sand, the strain can increase sud- through a layer of cohesionless sediment at a rate
denly, and blocks on slopes can move large dis- sufficiently high to cause fluidization, the layer
tances. The combination of elevated pore pressure expands rapidly, porosity increases, and the sedi-
with large loss of strength is sometimes referred to ment ceases to be grain supported and becomes
as "complete liquefaction" (oud, 1973). In more fluid supported. Fluidized flow typically destroys
8 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

tf/
Nonliquefiable layer
Stratum H 1
,ow

:.' ":"-'"" "".''" : " :'-" ::.' :'."," "" "" 7 " W a t e r :'.' .'":

Stratum H 2

Fig. 7. Vertical section showing typical sediment relations, seismic loading condition, and water flow paths involved in formation of
sand blows; 7, shear strain (angle in radians); a, horizontal acceleration; ~h, shear stress induced by horizontal acceleration. Triangle
points to water table.

original bedding and structures, at least locally. the list of possible causes for formation of the
Fluidization can result from a number of mecha- dikes.
nisms, including seepage caused by compaction of Many variables control the formation of large
underlying sediments, seepage from springs, or fluidization features caused by seismic liquefaction.
seepage from deposits liquefied either from static The influences of some of the most important
or earthquake forces. variables are fairly well understood. The forces
Reoriented minerals and small deformation involved and processes were presented elegantly
structures similar to those caused by liquefaction- by Scott and Zuckerman (1973). The main ele-
induced fluidization occur as syndepositional ments are illustrated in Fig. 7. Assume stratum
features in many environments (see Lowe and H 2 liquefies. Water tends to flow upward by two
LoPiccolo, 1974; Lowe, 1975; and van Loon, mechanisms, relief of the high pore-water pressure
1992), which makes interpretation of a seismic or and reconsolidation. Reconsolidation tends to
nonseismic origin uncertain. In contrast, larger cause densification as it progresses from the
features caused by fluidization from seismic lique- bottom of stratum H 2 upward (Scott, 1986). Water
faction are easier to interpret because, in many expelled from the zone that liquefied during shak-
field situations, sand-filled dikes whose widths ing tends to accumulate beneath a low-permeabil-
exceed several centimeters and whose heights ity capping layer (stratum H2) to form a water-
exceed a meter or so cut weathering horizons or rich zone (Liu and Qiao, 1984; Elgamal et al.,
other strata that are obviously much younger than 1989; Dobry and Liu, 1992; Fiegel and Kutter,
the source zone for the dikes. This relation gen- t994); this zone, in turn, probably supplies much
erally eliminates syndepositional processes from of the water and sand that vent to the surface
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 9

through breaks in the cap. Sediment may also be observations of effects of the 1811-1812 New
vented to the surface from greater depth, where Madrid earthquakes and the 1886 Charleston
liquefaction first developed. Venting can occur earthquake, for these sand sizes, a nonliquefiable
during the time of strong shaking or be delayed cap of only slightly lower permeability can still
by as much as a few minutes following very strong provide adequate confinement to form large sand
shaking (Kawakami and Asada, 1966). Where the blows.
cap is thin (< 1-2 m) is where the increased pore- The threshold magnitude to induce liquefaction
water pressure in the underlying sand-water mix- effects in the most susceptible gravel deposits is
ture can most easily break through to the surface. about 7, whereas the threshold magnitude for
Characteristics of the lower layer (stratum H2) sands is about 5.5 (Valera et al., 1994). Deposits
that enhance the liquefaction-fluidization process of sand and gravel containing more than 30-50%
are (1) a thick, loose sand that, once liquefied, gravel can liquefy, but liquefaction-induced fea-
provides a large volume of water available for tures in such coarse deposits are sparse in compari-
upward flow, and (2) a permeability that is high son to these features in sands. Such diminishment
enough to allow water to flow quickly to the base results from a number of reasons. The high gravel
of the cap but that is not so high as to dissipate content increases the internal friction resistance,
excess pore pressures between seismic cycles of making initial liquefaction more difficult. In addi-
shearing (Castro, 1987, pp. 177-179; Dobry, tion, gravel-rich deposits are generally more
1989). densely packed than sands in field situations. This
This simple model fully explains most field increased resistance to shearing combined with the
observations. However, seemingly contradictory high permeability found in many gravel-rich depos-
manifestations of the liquefaction and fluidization its makes liquefaction much more difficult than
process also are seen in the field. For example, for sands (Wong et al., 1975, p. 582). In deposits
sand dikes cutting gravel layers much more perme- having only a small percentage of fine grains
able than the sand dikes have been reported by (particles smaller than about 0.06 mm), the gravels
Tuttle et al. (1992). are in point-to-point contact rather than being
encased in a matrix of sand and silt. If more than
2.2. Factors affecting liquefaction susceptibility and 80% of the sediment is coarser than 0.7 mm in
effects of fluidization diameter, then the permeability may be too high
to develop a condition of initial liquefaction if
The most important factors controlling develop- there is no cap to inhibit drainage (National
ment of liquefaction-induced dikes and sills are Research Council, 1985, p. 94). However, very
considered in this section. Not discussed here are gravelly sediments can form large liquefaction-
secondary factors affecting liquefaction susceptibil- induced features if the source deposits are loose,
ity, such as grain shape, sediment fabric, weak thick, and capped by a stratum of low
grain-to-grain bonding in sediment, and static hori- permeability.
zontal stress conditions; those factors were summa- For example, extensive liquefaction occurred
rized by Seed (1979) and Mitchell (1976, p. 244). during the 1983 Borah Peak, Idaho, earthquake
(M=7.3) where very coarse sediment (Fig. 8) was
2.2.1. Grain size confined by a low-permeability cap (Andrus et al.,
Both field and laboratory data show that loosely 1991). Large amounts of gravel-bearing sand were
packed, cohesionless sands that are clean (that is, vented onto the ground surface. Significant lateral
they have no clay or silt or bonding matter) can spreading also occurred. The source strata con-
readily liquefy and form great numbers of clastic tained at least 70% gravel. Peak earthquake accel-
dikes, sills, and sand blows. Fig. 8 shows often- erations probably were high at this site, between
cited curves with size boundaries for fine and 0.3 and 0.35 x g (Andrus et al., 1991, p. 257).
medium sands that liquefy most easily and form Corroborating evidence of severe liquefaction in
large fluidization features. On the basis of my field gravel is also provided by a disastrous occurrence
I0 S. 1~ Obermeier/Engineering Geology 44 (1996) 1- 76

UNIFORMLY GRADED (WELL SORTED) DEPOSITS


100
I--
I
(.9
w
75
>-
O3
n-
W 50
z
I,.-
z
w
0
rr
I11
[3.
25
\ \
1', 10 1.0 0.1 0.01

Gravel ] Sand ] Silt ] Clay

PARTICLE SIZE (mm)

WELL GRADED POORLY SORTED) DEPOSITS


100

>- 75 - - \ \ liquefiable --
m \
~r \
w 50--
z
,7- Borah/\
~-
zw Peak Potentially~
25--
o ~ liquefiable--
n" earthquake
ILl

a_ 0
O0 10 1.0 0.1 0.01

Gravel l Sand 1 Silt I Clay

PARTICLE SIZE (mm)

Fig. 8. Gradation curves showing grain sizes and gradations most susceptible to liquefaction; from Tsuchida and Hayashi (1971).
Line for 1983 Borah Peak, Idaho, earthquake (M= 7.3) shows gradation of coarsest natural deposit that is verified to have produced
severe liquefaction effects at ground surface; from Andrus and Youd (1987). +, coarsest grain size suspected to be able to develop
the condition of initial liquefaction without confinement by a cap of low permeability.

in a gravel-rich fan deposit near the epicenter of Only in very exceptional circumstances can sedi-
the Fuku, Japan, earthquake of 1948 (Ishihara, ment finer than the lower bound for the "poten-
1985, p. 333). tially liquefiable" sand in Fig. 8 readily liquefy. In
In another example (Meier, 1993), gravels lique- general, natural deposits with a large proportion
fied sufficiently to cause damage to many bridges of fine grains contain a significant amount of clay.
during the 1964 Alaskan earthquake (M=9.2). The presence of even a small amount of cohesion
Ross et al. (1969) found that all these bridge sites from clay- or silt-sized material can impede particle
were relatively near the zone of major energy rearrangement during cyclic straining and greatly
release, being within 80 km. Sites where sand and increase resistance to cyclic loading. As little as
silt caused liquefaction damage extended as far 5% of clay- or silt-sized material in a sand deposit
as 130 km. can make liquefaction significantly less likely than
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 11

for the same sand deposit lacking fine grains 2.2.2. Relative density
(National Research Council, 1985, p. 103). The arrangement or packing of sand grains has
Engineering data show that clay- or silt-bearing a profound effect on a sediment's liquefaction
sediment that liquefies and has large strength loss susceptibility. Susceptibility of the same sand can
generally has less than 15% of the particles finer be changed from very high to nonsusceptible
than 0.005 mm, a liquid limit that is less than 35, simply because of a change in packing. The state
and a water content that is nearly equal to or of looseness or denseness of cohesionless sediments
greater than the liquid limit (Seed et al., 1983,1985, (also called "compactness" or "relative density")
Fig. 6; Youd et al., 1989). Cases where the criteria is generally measured in place by the Standard
above fail for clay-bearing deposits have been Penetration Test (SPT) blow-count method
discussed by Finn et al. (1994). (The liquid limit, (American Society for Testing and Materials,
expressed as a whole number, is the water content 1978). The SPT method is a standard engineering
at which a remolded sample has a soft consistency; test used on unconsolidated deposits. The test
at the liquid limit, the sample is on the semisolid- provides quantitative data for design of bridge and
liquid boundary. Liquid limit is measured in a building foundations. A sampling tube is driven
standardized test described in any elementary soil into the ground by dropping a 140-1b (63.5-kg)
mechanics text.) weight from a height of 30 in. (76 cm). The penet-
Clay- and silt-rich deposits referred to as sensi- ration resistance is measured as the number of
tive or "quick" clays (because they change to a blows required to drive the sampler 1 ft (30.5 cm).
viscous fluid upon remolding) can break down and Table 1 shows the relation between blow count
flow during strong seismic shaking, much as lique- and relative density in terms ranging from "very
loose" to "very dense." (Terms for relative density
fied sediments do (Updike et al., 1988). Field
in the table are used throughout this text as a
situations in which sensitive deposits occur were
semiquantitative indication of liquefaction suscep-
described by Mitchell (1976).
tibility.) Sands that are moderately dense or looser
The ability to form recognizable fluidization
liquefy in many field situations. Dense sands
features generally is greatly diminished in silt and
require exceptionally strong shaking, and very
clayey silt. Even though initial liquefaction may
dense sands probably rarely generate residual pore
occur, the low permeability may restrict water pressures large enough to form fluidization features
from escaping quickly enough to form fluidization (Seed et al., 1983).
features that are large enough to be recognized in SPT blow-count values in granular deposits
the field (Castro, 1987). Similarly, even though a principally reflect states of relative density and
very soft clay-rich deposit may have had high pore static vertical effective stress, which are the factors
pressures, the effects may not be recognized in the mainly controlling liquefaction susceptibility. Even
field. Silty very fine sand is the finest sediment in secondary factors that affect liquefaction suscepti-
which large seismically induced liquefaction fea- bility, such as grain shape, sediment fabric, weak
tures commonly form in the field. However, an sediment bonding, and static horizontal stress, are
exceptionally loose, highly susceptible silt deposit probably correctly reflected by SPT measurements,
can also produce sizable sand blows, providing the at least in a qualitative sense (Seed, 1979, p. 212).
deposit fulfills the engineering criteria described Thus, the SPT method is very useful for evaluation
above (Youd et al., 1989). It has been my experi- of susceptibility.
ence, though, that only very rarely are features The geological environment in which sediment
encountered in paleoseismic searches that can be is deposited generally has a major bearing on
recognized as having originated by liquefaction of relative density. Windblown sand deposited in
thick silt deposits. The influence of silt content on dunes is very loose in many places. Rapidly flowing
liquefaction susceptibility, from an engineering streams tend to form more densely packed sedi-
perspective, has been summarized by Finn et al. ments, but in quiet water even coarse-grained
(1994) and by Singh (1994). sediments can be laid down gently in a loose
12 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

condition. Large quantities of loose sand tend to noes that are much wider and thinner than those
be deposited where small streams empty into much formed under subaerial conditions.
larger streams or lakes. Deltas are especially favor-
able sites for thick, very loose deposits. Near the 2.2.4. Depth and thickness of strata
ocean, the transition zone from a rapidly flowing Liquefaction during earthquake shaking fre-
stream to a tidal estuary presents a setting where quently originates in a zone as shallow as 1.5-2 m
thick, loose, clean sands can be deposited. In brief, below the ground surface (Bennett, 1984;
there are many local and widespread geological Obermeier et al., 1986). In regions of very strong
environments where susceptible sediment occurs. shaking, as for example in the meizoseismal region
The field geologist must reconstruct ancient deposi- of the M 9.2 earthquake of Alaska in 1964, the
tional settings to predict locations of susceptible source beds that liquefied and vented sand were
sediment for a paleoliquefaction study. apparently as shallow as about 0.5 m (Walsh et al.,
1995). Such shallow source zones can give rise to
2.2.3. Depth to water table many large dikes and other features. In special
Liquefaction susceptibility decreases strongly field situations, the top of the source zone can be
with increasing depth to the water table. The extent shallower yet, only a few tenths of a meter (Sims,
of the control is indicated in Table 2, which relates 1973; Audemard and de Santis, 1991); for such
the liquefaction susceptibility in San Fernando shallow depths, though, liquefaction seems to
Valley (California) both to depth of ground-water occur mainly where a thin sand stratum is confined
table and age of the deposit. A high susceptibility between clay-rich layers, and the evidence for
for this example means that a ground motion with liquefaction generally is the development of fea-
10 cycles of acceleration at 0.2 x g (this is equiva- tures such as recumbent folds along the thin sand
lent to shaking by an earthquake having a magni- layer. The sand layer has served as a failure plane.
tude of ~6.5) is required to produce sand blows. Evidence for liquefaction at a depth as shallow as
A moderate susceptibility requires 30 cycles at a few tenths of a meter is not observed often in
0.5 x g ( M ~ 8 ) . Table 2 shows that lowering the paleoseismic studies, perhaps because of destruc-
water table as little as several meters can change tion by weathering. At the other extreme for depth,
the susceptibility from high to low. Such a change the source zone for liquefaction can exceed 20 m
in water-table depth, even seasonally, is common (Seed, 1979). The most common depth is a range
in many field settings. between a few meters and about 10 m. Optimal
Liquefaction with extensive venting can also depth is about 2-4 m in many field settings where
readily occur beneath the water surface. Venting the water table is within a few meters of the surface.
beneath water is characterized by sediment, partic- A 1-m thickness of sand is generally required to
ularly the finer constituents, forming sand volca- form dikes or sills in much abundance. This also
is a minimum thickness to systematically distin-
guish soft-sediment deformation features from
earthquake-induced liquefaction features in many
Table 2
Influence of age of deposit and depth to water table on liquefac- field situations. However, at a few sites in coastal
tion susceptibility in the San Fernando Valley, California Washington and Indiana, I have observed sand
strata as thin as 0.3 m that liquefied and formed
Age of deposit Depth to water table (m) dikes whose heights exceed 1 m. No significant
0-3 3-10 >10 horizontal ground displacements occurred at these
sites. The 0.3-m-thick sand strata graded down
Latest Holocene high low nil abruptly into thick sandy gravel beds that showed
Earlier Holocene moderate low nil no evidence of liquefaction. In an exceptional case,
Late Pleistocene low nil nil
Tuttle and Seeber (1991) reported tracing small
High and moderate susceptibility are defined in 2.2.3, "Depth sand dikes down to a 8- to 10-cm-thick sand source
to water table." From National Research Council ( 1985, p. 91 ). at a depth of 2.5-3 m. Very strong shaking may
S.F Obermeier/Engineering Geology 44 (1996) 1-76 13

be required to form dikes from such thin source of enhancement of ground breakage toward
strata. streams and toward even slight topographic
Liquefaction of a source stratum less than 0.3 depressions (T.L. Youd, Brigham Young Univ.,
m thick can lead to major lateral spreads or flow written commun., 1994). Fig. 10 illustrates break-
failures, providing the stratum has large lateral age patterns commonly observed in the field.
extent (Youd and Perkins, 1987, p. 1377). Indeed, Increased breakage also tends to occur where the
large horizontal movements may be required to base of the cap dips toward a stream or depression
produce recognizable liquefaction effects wherever (Hamada et al., 1986).
the source strata are extremely thin. Strata of fine Another very important factor in breakup of
sand as thin as a few centimeters that liquefied in the cap is oscillatory shaking of the cap above
association with spreading, and also formed scat- liquefied sediment (Youd and Garris, in press).
tered small dikes, have been observed in lacustrine An oscillating cap that is floating on liquefied
deposits in Utah (J.R. Keaton, SHB AGRA, Inc., sediment experiences intensified lateral spreading
Salt Lake City, Utah, written commun., 1993). and breakup of the cap, especially where strong
Lacustrine sediments seem to be most favorable shaking occurs for a long duration. These effects
for development of liquefaction effects in thin are likely to be of increasing importance with an
strata because of the large areal extent of the strata. increasing earthquake magnitude. Breakage of the
Whether liquefaction-induced ground rupturing cap is probably caused by the direct action of
occurs at the surface and whether sand vents seismic shaking at the ground surface, which
through the ruptures commonly depend on the strains and pulls apart a brittle, clay-rich cap that
relation of the thickness of the source stratum lies immediately above the sediment that liquefied.
(stratum H2 in Fig. 7) to the thickness of the Reflected and refracted waves along the ground
overlying cap of nonliquefiable sediment (stratum surface may have important roles (Wang et al.,
H1). Guidelines for critical thicknesses of strata 1983), but documentation is very incomplete.
H1 and H~ for ground rupturing have been devel- Possibly either surface (Rayleigh or Love) waves
oped on the basis of a limited number of field or body waves may be involved. Engineering
observations (Ishihara, 1985). The curves are analysis (Pease and O'Rourke, 1995) indicates that
based on behavior during a few strong earthquakes body waves were the driving mechanism for oscilla-
(M~7.5), and I suspect that they reflect ground tion-induced ground failures during the 1989 Loma
fracturing exclusively by the mechanism of hydrau- Prieta, California, earthquake ( M = 7.1 ). My field
lic fracturing. A cap thickness exceeding 10 m observations indicate that surface waves caused
prevents ground rupture even in the most favorable breakup of the cap, even far beyond the meizoseis-
circumstances for liquefaction, according to mal zone of the great New Madrid earthquakes of
Ishihara (Fig. 9). A cap 1 m thick can be breached 1811-1812.
whenever accelerations exceed about 0.2 x g if the
thickness of liquefied sand exceeds 1 m. Supportive 2.2.6. Geological details of the cap
observations in the meizoseismal region of the Minor characteristics of the cap also can have
1811-1812 New Madrid earthquakes show that a profound influence on the manifestations of flow
the density of dikes in plan view and the amount features. Holes left by decayed roots or dug by
of venting are greatly enhanced where the cap is crayfish can be used as conduits for venting and
very thin, at least for subaerial venting (Obermeier, can remain approximately circular where liquefac-
1989). Field situations where the curves of Fig. 9 tion has not been severe (Audemard and de Santis,
are not applicable, even in a qualitative sense, are 1991 ). Vertical, planar dikes normally form where
noted in the following section. a clay-rich cap is weathered, even slightly, and has
preexisting vertical fractures or planes of weakness
2.2.5. Site effects and nature of seismic shaking (Obermeier et al., 1990). Thick sills tend to form
The curves in Fig. 9 tend to highly overestimate beneath a very flexible, peaty cap (Clague et al.,
accelerations at sites of lateral spreading because 1992), whereas a very strong, brittle, mechanically
14 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

I I I I I

12-

11-

g 10-
-1-

IJJ
9-

5
8-
z
'.'~J
:.:.:+.... Ma,,. acc. .i,i:t
.:'i~ -
orj
7-
LU
.-I
fill
!(-j - -0.2 g .!:::it .!:.iJ -
<
I
LI_
W 6 -
'i E
} :i11
i
:'i':'l
Max.acc.; j
-0.3 9 """[
::.:/
:i:'::.'/
:D -

0
..J
5- o Max. acc. ..'.../ _
0
IJ _
:-.:~:1 .i:J -o.4-o.5 ~ "
o9
o9
:.",.:::1 ..i."7 :"
zv .
4- " !:.:/ ../ :::.57 :"-~/
...(..;, -
.~::::/ . ..:...:: y ...::':: /
-I- 3- .::i::7 ..:,:.....>" . ,...:k::l
b- ,'..:/ .-....:>.:" .. :.'..'.'::-'"/
,''.'" " ".. ~ " i " ".'.. "" "..."'~

2 -
.'..:'.../ ...:'::....
..:.':'~y y y .... .:~55.:> - y
. ......:'7.-."~
)!./:.i~ >" . . : : ~ . > ~
..../....7 ... :.~:>'-
1-

I I t I ~ I I I I
0
0 1 2 3 4 5 6 7 8 9 10
THICKNESS OF SURFACE LAYER, H 1 (m)

Fig. 9. Proposed boundary curves relating thickness of nonliquefiable surface layer to thickness of the liquefiable zone as a function
of peak earthquake accelerations required to induce venting or ground rupturing at the surface. From Ishihara (1985).

isotropic cap, especially if sand rich, tends to favor An apparent contradiction to recurrent lique-
formation of large craters (Obermeier et al., 1990). faction at the same site is the observation that
Other important factors are discussed where case liquefaction commonly densities sediments.
histories are presented. Densification should reduce the liquefaction sus-
ceptibility. Worldwide engineering measurements
2.2.7. P r e v i o u s liquefaction before and after occurrences of liquefaction indi-
Both historical observations and field paleoliq- cate that thick zones (> 10 m) of loosely to moder-
uefaction evidence show that liquefaction has a ately compact sands in the zone that liquefied
strong tendency to recur at the same site, even to during shaking were densified substantially when-
the extent of repeatedly using the same dike ever liquefaction was severe (Koizumi, 1966;
for venting. Such a tendency occurs for earth- Ohsaki, 1970). However, measurable densification
quakes either closely or widely separated in does not always occur, as shown by careful meas-
time (Kuribayashi and Tatsuoka, 1975; Youd urements by T.L. Holzer (written commun., 1994)
and Hoose, 1978; Youd, 1984b; Saucier, 1989; for the Superstition Hills, California, earthquake
Obermeier et al., 1990; Tuttle et al., 1992). ( M = 6.6) of November 1987.
S.F. Oberrneier/EngineeringGeology 44 (1996) 1-76 15

Fig. 10. Commonlyobservedpattern of major groundcrackingproducedby lateral-spreadmovementstoward a topographicdepres-


sion. SituationA showspattern alongstraightand concavebanks. SituationB showspattern alongslightlyconvexbanks. Situation
C showspattern alongmoderatelyconvexbanks. From McCullochand Bonilla(1970).

Even where liquefaction has been severe and there is evidence for some densification caused by
repeated, density increases may not be uniform, repeated liquefaction at the same sites because
and engineering data show that sediments can younger craters are smaller than older craters.
remain quite susceptible. For example, in the mei- Data from studies in Japan show that the pres-
zoseismal region of the 1811-1812 New Madrid ence of a cap of low permeability may not be
earthquakes, where liquefaction was severe, the required for sands to maintain the ability to reliq-
uppermost meter or so of sand directly beneath uefy. The upward flow of water from a liquefied
the impermeable cap is still moderately to highly zone at depth appears to be able to loosen sand
susceptible at many places, and, beneath this thin at shallow depth where sand extends to the surface.
zone, the sand grades to a moderately susceptible This loosening may occur within 2-5 m of the
condition that extends to a much greater depth surface (Kishida, 1966, p. 75). Such a process of
(see engineering boring logs in Obermeier, 1989). loosening has also been observed in the laboratory
The relatively loose zone beneath the cap likely (Scott and Zuckerman, 1973).
originated because water collected there and loos- In contrast to densification of loose sand at
ened the sands prior to venting, as well as emplaced depth, field data suggest that, in special circum-
sills of loose sand along the base of the cap. stances, relatively dense sands may be loosened
Laboratory model studies on layered sand and significantly by an occurrence of liquefaction. Data
clay beds also show formation of a loosened sand from Koizumi (1966) and Kishida (1966) suggest
zone along the base of clay beds (Elgamal et al., that such loosening may occur where the dense
1989; Fiegel and Kutter, 1994). sand is underlain by a much looser sand that
Another well-documented case study demon- liquefied. An upward flow of water through the
strates the possibility of only minor densification dense sand apparently caused fluidization and
by an occurrence of strong liquefaction. In the loosening.
meizoseismal region of the 1886 Charleston, South In the 1979 earthquake of Imperial Valley,
Carolina, earthquake, source sands beneath a low- California, a 3-m thickness of source sand for a
permeability, humate-cemented cap remain in a large lateral spread remained loose beneath the
loose condition at sites of abundant sand blows cap (Youd, 1984b). The large shearing associated
(Martin and Clough, 1990). These South Carolina with the lateral spreading may have maintained
sites experienced at least three previous episodes the source sand in a highly susceptible state.
of strong liquefaction during the Holocene; still, During shear, granular soils can either compact or
16 S.I( Obermeier/Engineering Geology 44 (1996) 1 76

dilate depending on the compactness of the sedi- the depth of ground water and on water chemistry.
ment, the confining pressure, and the amount of Where the water-table depth fluctuates greatly,
shear deformation. Except for very dense packings, bonding can occur as a result of chemical reactions
sands initially tend to become more compact as or an influx of clay. Alternatively, only very minor
shear strain is applied. At large shear strains such changes in bonding may develop in sediment much
as those for lateral spreading, all but very loose older than Holocene if the ground-water table has
sands or sands under high confining pressure tend remained very high.
to dilate toward a relatively loose condition
(Youd, 1984b).
3. Criteria for an earthquake-induced liquefaction
2.2.8. Age and mineralogy of sediments origin
The age of sediment can have a strong influence
on liquefaction susceptibility. At many places in A set of criteria has been developed for deter-
the Western United States, sufficient changes occur mining whether observed sediment deformation
within 500 years to reduce susceptibility from high was caused by seismically induced liquefaction. A
to moderate (Youd and Perkins, 1978). In this seismic origin can be established if the following
region, the susceptibility commonly decreases from conditions are met.
moderate to low as age increases from 500 years ( 1 t The features have sedimentary characteristics
to earliest Holocene. Pleistocene sediment is lique- that are consistent with an earthquake-induced
fiable only in exceptional circumstances, and no liquefaction origin; namely, there is evidence
modern occurrences of liquefaction have been of an upward-directed hydraulic force that
observed in pre-Pleistocene sediments. was suddenly applied and was of short
In the Eastern United States, however, duration.
Pleistocene sediment remains highly susceptible to (2) The features preferably have sedimentary
liquefaction over large regions. Large liquefaction characteristics consistent with historically doc-
features formed in loose sediment as old as umented observations of the earthquake-
200 000-240 000 years during the 1886 Charleston induced liquefaction processes in a similar
earthquake (Obermeier et al., 1990). Apparently, physical setting. In addition, preferably there
a very high ground-water table in combination is more than one type of feature commonly
with weakly acidic (organic acids) ground water caused by seismically induced liquefaction.
has prevented significant bonding in the very Such features include dikes, sills, vented sedi-
quartz rich source sands. ment, lateral spreads, and some types of soft-
In the Central United States, abundant liquefac- sediment deformations.
tion features developed in sediments as old as late (3) The features occur in ground-water settings
Wisconsinan (about 15000-20000 years old) where suddenly applied, strong hydraulic
during the 1811-1812 New Madrid earthquakes. forces of short duration could not be reason-
There also, the ground-water table appears to have ably expected except from earthquake-induced
been very high through the Holocene (Wesnousky liquefaction. In particular, the possibility of
and Leffler, 1992). The source sands in the New an origin from artesian conditions or non-
Madrid earthquake region contain a wide variety seismic landsliding must be eliminated.
of minerals, derived largely from glaciers that (4) Similar features occur at multiple locations,
passed over terranes of igneous and sedimentary preferably at least within a few kilometers of
rocks. Thus, in the Charleston region and the New one another, in similar geological and ground-
Madrid earthquake region, source sands of very water settings. The regional pattern of size
different mineralogies remained quite susceptible and abundance of features should be consis-
to liquefaction for a very long time. tent with a pattern of shaking associable with
Changes in liquefaction susceptibility through an earthquake in which there is a core area
time probably depend strongly on fluctuations in where the effects are most severe.
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 17

(5) The evidence for age of the features supports near Charleston, SC. The estimated magnitude is
the interpretation that they formed in one or about 7.5 (Johnston, 1994). Clear evidence of
more discrete, short episodes that individually seismotectonic conditions in the region is lacking,
affected a large area and that the episodes and this lack prompted searches for prehistoric
were separated by relatively long time periods liquefaction features. Liquefaction evidence has
during which no such features formed. since been found for many strong prehistoric earth-
quakes (Obermeier et al., 1987; Amick et al.,
Determination of the regional pattern of size 1990). Results of the searches were described by
and abundance of the suspected liquefaction fea- Obermeier et al. (1990) and are shown in Fig. 11.
tures may be critical to interpretation of origin. The figure shows the approximate boundary of the
Preferably at least 20-30 km of cumulative fresh 1886 earthquake meizoseismal zone, shows the
exposure is examined. Such a regional approach sites where swarms of liquefaction features
to a paleoliquefaction study helps eliminate the described as "craterlets" were formed in 1886
possibility that nonseismic processes are responsi- (Dutton, 1889; see Fig. 12 herein), and shows the
ble for creating the features, and it helps to develop sites where liquefaction features predating 1886
a sense of the various processes that deform sedi- were found. The prehistoric liquefaction-induced
ments. The next section illustrates application of features are mainly ancient craterlets that are
the five criteria above and presents examples of now filled.
how an understanding of the local geological set- None of the pre-1886 craterlets (called craters
ting is critical to interpretations. in the rest of this article) found has an expression
on the ground surface that is discernible by on-site
surface examination or on aerial photographs.
4. Historical and prehistoric liquefaction - selected These features are seen only in walls of excav-
studies ations. At most sites shown in Fig. 11, at least
three or four pre-1886 craters are exposed within
In this section, I describe earthquake-induced a few hundred meters of one another.
liquefaction features in four geological-geographic The physical setting of the region within 50 km
settings in the United States: coastal South of the South Carolina coast is conducive to wide-
Carolina, the New Madrid seismic zone of spread liquefaction. That region is known locally
Missouri, the Wabash Valley seismic zone of as the "low country" because it has low local relief
Indiana and Illinois, and coastal Washington State. (1-3 m) and low altitude (0-30 m) and because
The two latter areas have not had historical lique- vast expanses are under water much of the year.
faction. The discussion emphasizes the need to Most of the Carolina low country is covered by a
consider the local geological setting in deducing 5- to 15-m-thick blanket of unconsolidated
an earthquake origin. Special consideration is Quaternary marine and fluvial deposits, which lies
given to elimination of artesian springs and non- on semilithified Tertiary sediment. The Quaternary
seismic landsliding as possible sources of observed deposits primarily occur as a series of well-defined,
deformations, because these are the mechanisms temporally discrete, interglacial beaches and
most likely to produce fluidization effects that associated back-barrier and shelf deposits that
mimic those of seismic origin. Where possible, the form belts subparallel to the present shoreline.
magnitudes of the prehistoric earthquakes are esti- Increasingly older beach deposits are progressively
mated by comparison with historical liquefaction- farther inland and at higher altitudes. Most beach
producing earthquakes in the region. deposits consist of clean, medium to fine sand.
Most of the craters discovered were on the
4.1. Coastal South Carolina ancient beach deposits. The search for liquefaction
features was generally restricted to the beach
The strongest historical earthquake in the deposits younger than about 250 000 years and
Southeastern United States took place in 1886 older than about 80 000 years (Fig. 11). Sand in
18 S.F. ObermeieuEngineering Geology 44 (1996) 1-76

+
Nonlm~mott I ~ _ + . . + . . . ~
tctztw) =#e wed~rqt \

Coealtssleareaheaving numerous wldetpreaad


liquefaction felturea probably from
prehistoric eaarthquake=whol4z epicenters
were f i r from Chartelton

~ ---- Approximlzte boundary of


1886 apicentreal region

fL == A r m of eabuncl~mtcr~ler=
in 11186,from 0utton (18811)

Prehistoric liquefaction sit=


,,t, from Obermeier It zd. tlMT)
& from Amic:k t t s1,(1210)

ii
~ro)timaate northern limit of historiclll,
documented =n~lt liquefKtion f e m u ~
caused by the I!116 earthquake

t ....................
IKoproximeateaIouthern limit of 1806 eearthqueake
IklUefzction future=, == determined by
hi=mtoriclldlta tinctextlnlive field studies
81'

~ ++~'-+"r ,. ~ wt.
v . ~ +.

tNl~XI~P ~

o M so ~tEs

Fig. 11+ Coastal portion of South Carolina and southern North Carolina containing liquefaction sites. Unshaded onshore region is
predominantly marine deposits younger than about 240 000 years. Numerous ancient beach ridges lie in this unshaded region. Shading
denotes region of older marine deposits that was not reconnoitered, except locally, Younger fluvial sediments occur locally. All
liquefaction sites along the Edisto River are in fluvial sediments. Almost every liquefaction site shown represents an area where
numerous liquefaction features are exposed in a network of drainage ditches several kilometers in length, Numerous sites in and near
the 1886 meizoseismal (epicenter) zone are not shown because of lack of space, Index map shows coastal region searched.
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 19

Deposits older than about 250 000 years have such


a low susceptibility to liquefaction (due to effects
of chemical weathering) that the likelihood of their
liquefying has been extremely low during the late
Pleistocene and Holocene. Deposits younger than
about 80 000 years generally have such a high
;::.,
.... .. . . . . . . . ~ water table that exposures are very limited.
Tabular sand dikes were discovered in fluvial
terraces and in back-barrier environments where
the cap is rich in clay. Sills were observed only
. . . . . . . . . - "
rarely. The following discussion concentrates on
the craters, because the section below discussing
liquefaction effects in the New Madrid seismic
zone adequately deals with characteristics of tabu-
lar sand dikes and sills, which are the types of
liquefaction-induced features normally formed
(a) where there is a thick clay-rich cap.

4.1.1. Characterization of craters


Fig. 13 is a schematic vertical section through a
typical crater site on an ancient beach ridge on a
barrier bar. According to eyewitness observations
in 1886, "craterlets are found in greatest abun-
dance in belts parallel with (beach) ridges and
along their anticlines" (Peters and Herrmann,
1986, p. 68). Thus, the locations of crater sites we
discovered are consistent with historical
observations.
Almost all craters that predate 1886 have shapes
and sizes similar to those of the 1886 craters of
Fig. 12 except that the craters are now filled with
sediment. Fig. 12 shows moderate to large craters
produced by the 1886 earthquake. Examination of
(b) photographs taken in 1886 shows that surficial
sheets of vented sand around crater rims normally
Fig. 12. Craters produced by the 1886 Charleston, South
Carolina, earthquake. (A) An 1886 crater near the present
had thicknesses of about 15-20 cm. Maximum
Charleston airport. Note that the crater contains sand sloughing reported thickness of vented sand was 1 m, and
toward the lowest parts and that there is a constructional sand the maximum crater diameter was about 6 m
volcano located in the fight part of the crater (at arrow). The (Dutton, 1889).
crater is surrounded by a thin blanket of sand partly veneered The crater sites are located where weathering
with cracked mud. Sketch by H. Livingston from a photograph.
(B) A typical crater produced by the 1886 earthquake. Note
has imposed a strong soil profile on the ancient
the thin blanket of ejected sand around the crater and sand and beaches (Fig. 14). Near the surface, a thin A
clasts of dark soil within the crater. Photograph from the horizon (organic matter and a small percentage of
archives of the Charleston Museum. sand) overlies a thin, very light gray E (eluviated)
horizon. The E horizon overlies a thick, weakly
these deposits is loose at many places. Normal cemented, black Bh horizon (humate-enriched
depth to the water table in these sediments is about sand containing a small percentage o f clay) that
1-2 m, even in topographically elevated regions. grades down rather sharply into a variably thick;
20 S.IE Obermeier/Engineering Geology 44 (1996) 1 76

SE NW

./Wlta't table ~ . / - ~ i l l o d crlNlrll -'r--

t~oom w~ ~, . . . . . . . / \ ~_
n,po,m ,,t , , u , ~ , . , , A " - ~ ' : . - . . " I ~ \ 2 to,,o m

.~ a ~ ..,~.. ;..!.;.~.:.:'.-....;....... . . . . " ' T~V

Fig. 13. Schematic vertical section of representative ancient barrier bar (beach ridge) showing sediment types, ground-water-table
locations, filled craters, and Bh (humate-rich) soil horizons. Modern South Carolina shoreline is located southeastward. Lagoonal
clay deposit at left is younger and lower in elevation than the barrier-bar deposit.

/Nonbedded~U/-~..,.. :.... .-:.i..".


~c_mstter.riehsand//j~/ . . . . .. ; . . ' . ..... : . . , ' / J J , / / / /J/
/////////./~/~,'.": Layer 5: Structureless sand;'. i~/Latmiln: ted sequence
Undisturbed ~/ < / / / / / / / / < / /.." .. . . . . . . . . . .". . . . . .. . . / / o t a t e r n a t n g c e a n
~.."--U//U/ ."..'.:...
' : :'',,."..'' ."-. . ..'. : :';'.":..':.'-"~/and o rgan?_c-_matter-
. . . . . . .
~.~ .... ..~.'. .. . .......
. . . . . . .
. ~...." .
.

Layer 3: Sand and


many small clasts

N%'..." ... -. : . ...... LaYer 2: Sand and scattered. '..'\ .~t.-<.'..


_ _ _ A B-C ""~: .. "..clasts of clean C-horizon sand and . : .- : ~ - / B-C
7 Crater/ ~....: ~.A-.ancl Bh-horizon mate.al ' ".." /
-- --C . . . . t .... Sk
~'-~i ~ = I/~ Graded
zone

"-a--~Y
e~B hS~n rdzal~d/artge'Clla-' t~'~ 1
-- Tubelike dikes containing -
. . . . . . structureless, clean sand -
--
--i
0
i i
15 I N C H E S
i
--
_~--- ~-C-~ -Bedded clean sand~i
__] I I I
0 3~ C E N T I M E T E R S - -
NO VERTICAL EXAGGERATION

Fig. 14. Schematic vertical section of filled liquefaction crater that forms a bowl in three dimensions. Letters correspond to soil
horizons. The filled crater in this figure long predates the 1886 Charleston earthquake, on the basis o f thickness of the Bh horizon
in the filled crater. Insert shows zonation within the crater.

light-colored B - C horizon (transition zone defined zones of sediment. The fill materials are
between B and C horizons). The B - C unit grades fine to medium sand and clasts from the Bh, B - C ,
down into C-horizon sands (parent material). The and C horizons of the host, and there is sand from
craters cut the solum (the horizons above C) and depths much below the exposed C horizon. Walls
the C horizon. Within the filled crater are well- of the crater are commonly smooth and sharply
S.E Obermeier/Engineering Geology 44 (1996) 1-76 21

defined when viewed closely, especially in the from the surface and near-surface (A, E, and Bh)
lower part. soil horizons because the blanket has been incorpo-
The Bh horizon of the laterally adjoining undis- rated within these soil horizons.
turbed host generally is abruptly thicker than the The presence of friable, angular clasts from the
Bh horizon on the filling in the crater. With B-C, C, and Bh horizons in the graded-fill portion
increasing age, the Bh horizon of the filled crater (layers 1-3) is consistent with a short-lived, churn-
is thicker, is more clay rich, and has better devel- ing type of upwelling from the vent. Water com-
oped soil structure. Craters older than about 5000 monly flows for a day or so from vents, as indicated
years have Bh horizons that approach the thickness by worldwide observations. The violent, boiling
and development of those in host sediment enclos- phase is much briefer. Hence, the presence of
ing the craters. friable clasts argues against a long-term artesian
The filled craters are characterized by a sequence spring origin; such a spring would abrade, round,
of five layers. Layer 5 (Fig. 14) is a structureless, and disaggregate the clasts. In addition, springs
gray, humate-enriched and -cemented sand, which are very unlikely to form in the topographic-
overlies a thinly laminated sequence of alternating geological setting at some of the crater sites
light- and dark-colored sands (layer 4). The lami- (Fig. 13).
nae typically are discontinuous and irregular in An earthquake origin for the craters is also
thickness but generally are 2-3 mm thick. The supported by the presence of sand-fiUed tabular
dark color is due to humate staining. The basal fissures, whose overall shape and dimensions
bed sharply overlies a medium-gray structureless strongly suggest that they are "incipient craters."
sand (layer 3), which contains many small clasts Such fissures as shown in Fig. 15 are rather
of Bh material and wood. Layer 3 grades down common in the meizoseismal zone of the 1886
into a structureless sand zone (layer 2) containing earthquake. The figure shows a V-shaped fissure
many intermediate-sized clasts (1-5 cm in diame- connecting to a tubelike dike that transported sand
ter). Layer 2 grades down into layer 1, which upward from depth. The tabular fissures in the
contains densely packed intermediate-sized and V-portion widen with depth until they connect to
large (> 5 cm) clasts of Bh material in a structure- a single, near-vertical, large, sand-filled tube. The
less sand matrix; the large clasts have diameters V-shaped fissures probably represent the early
exceeding 25 cm in many filled craters. Beneath phase of development of craters; upward forces,
the bowl are dikes containing clean sand. The however, were too weak to excavate the overlying
dikes are oval to circular in plan view, material.
The filled craters are interpreted to have formed It is possible that liquefaction produced craters
in the following phases: (1) a large hole is exca- because of a fortuitous combination of sediment
vated at the surface by the violent upward dis- properties. The source beds that liquefied were
charge of the liquefied mixture of sand and water; exceptionally susceptible to liquefaction; they were
(2) a sand rim accumulates around the hole by loose (engineering sense), contained uniformly
continued expulsion of liquefied sand and water; sized fine sand, and were free of clay (Martin and
(3) sand, soil clasts, and water are churned briefly Clough, 1990). These properties would cause the
in the lower part of the bowl, followed by settling source beds to liquefy abruptly; once liquefied, the
of the larger clasts and formation of the graded- sand-water mixture would flow readily. I suspect
fill sequence of sediment (layers 1-3); and (4) the that the liquefied sand strata quickly migrated
crater is intermittently filled by adjacent surface laterally to a hole such as that left by a decayed
materials during the weeks to years after the root. The sudden application of an upward force
eruption. The deeper part of the stratified post- around the hole caused the formation of a
eruption fall is layer 4, which remains laminated. V-shaped crack. The liquefied sand violently
Layer 5 is in the strongly bioturbated zone and vented because of its exceptional ability to flow.
thus has no stratification. The sand blanket ejected The V-shaped cracks developed because overlying
from the crater is indistinguishable in the field sediment is isotropically cemented with humate,
22 S.F Obermeier/Engineering Geology 44 (1996) 1 76

- Shovel

BANK
OVERHANG
, , , ...,

.... . , , ~ . , .,, , /,h~


/,/
~Fiiledcrater / /" / " / ,,'%
~/
" "
./
,Z
,
/ /
, , /
/

//~ much predating / ."/ ," "/'~k~//Y, , //\/


",/ 1886 earthquake , . , / ,
/ / . - . , / /

F-- . . . . . .

C-horizon sand ....


- - : : - : -: Tubelike di

0 20 INCHES

FLOOR OF DITCH 0 30 CENTIMETERS


NOVERTICALEXAGGERATION

Fig. 15. Schematic vertical section showing V-shaped, sand-filled fissures that probably represent the early phase of crater development.
The fissures are interpreted as resulting from liquefaction during the 1886 Charleston earthquake. Sand in the fissures came from a
depth of 6 m, on the basis of grain size and mineralogy. Fissures cut soil horizon developed on filled crater predating 1886.

has no pronounced planes of weakness, and is very craters) through a clay-rich cap; some "craters"
brittle; the process is similar to formation of a (actually surface depressions) as much as 10 m in
conchoidal fracture in an isotropic, brittle medium diameter and extremely shallow (25-40 cm) also
such as glass, when struck by a rock. formed (Reimnitz and Marshall, 1965). These very
Liquefaction-induced craters are common large "craters" apparently developed as sand
during earthquakes worldwide. Good examples for vented to the surface and undermined a clay-rich
various earthquakes in Japan are shown in articles cap, thereby making a swale.
by Kawakami and Asada (1966) and by Iwasaki In Argentina, Youd and Keefer (1994,
(1986). Craters were especially prevalent during pp. 227-229) have shown that preexisting holes
the Niigata earthquake (M=7.5) of 1964; soils through a clay-rich cap later led to erosion of large
around Niigata typically are sand rich all the way craters in response to liquefaction during an earth-
to the ground surface (Katayama et al., 1966), quake having M=7.4. Craters also formed in a
much as for the Charleston crater region. For the clay-rich cap in the Nile River Valley during the
Northeastern United States, sketches of prehistoric Dahshure, Egypt, earthquake (M=5.9) of 1992
seismically induced craters in host sand deposits (Elgamal et al., 1993). No mechanism for forma-
are in Tuttle (1994). tion of the craters in Egypt can be demonstrated
Craters can also form in a clay-rich cap. Craters because of the lack of geological and geotechnical
commonly develop in the clay-rich cap along the data at the crater sites. However, the regularity of
length of cracks induced by seismic shaking (for alignment of the craters suggests that manmade
example, Swenson, 1964, p. 162). In Alaska during holes led to formation of the craters.
the great earthquake of 1964, craters having a Another field situation that may lead to forma-
relatively small diameter (many ~ 1 m) formed at tion of a crater occurs where liquefaction applies
sites of violent venting that eroded deep holes (the a large hydraulic force to the base of a thin, clay-
S.F Obermeier/Engineering Geology 44 (1996) 1-76 23

rich layer that is bounded above and below by as far as the craters produced by the earthquake
sand (Elgamal et al., 1989, p. 242); the small of 1886 that had a magnitude of about 7.5.
resistance to both shearing and tension in the host Interpretations of prehistoric earthquake magni-
sediment above the base of the clay layer could tudes must account for liquefaction susceptibility.
explain development of a crater in preference to Principal variables are water-table depth and the
other modes of ground failure. I have observed compactness of the source sands. The water table
where this mechanism probably led to formation is presently very shallow, being less than 1-2 m
of small craters in the meizoseismal region of the below the ground surface. Almost certainly the
1811-1812 New Madrid earthquakes. water table has been essentially unchanged for the
past few thousand years at many of the crater sites
(Amick et al., 1990). Just prior to the 1886 earth-
4.1.2. Prehistoric seismicity quake, the Charleston area was experiencing an
Some filled craters in South Carolina contained extraordinarily wet period, so water-table condi-
small twigs and bark from trees. This woody tions were optimal for production of liquefaction
matter is concentrated along the contact between features (Taber, 1914, p. 126). Standard
layers 3 and 4 (Fig. 14) and obviously fell into the Penetration Test (SPT) data also show that the
open pits soon after they formed. Radiocarbon source sands are so loose as to readily liquefy.
data on such matter in scattered craters yield Sand deposits at some liquefaction sites have SPT
approximate ages of 600, 1250, 3200, 5150, and blow counts as low as 10 or less (Martin and
older than 5150 years, documenting five prehistoric Clough, 1990). It is difficult to conceive of any
earthquakes in the Charleston area (Amick and mechanism that would have made the sands much
Gelinas, 1991). more compact when the prehistoric earthquakes
Estimates of the magnitudes of the South occurred. In summary, the liquefaction susceptibil-
Carolina prehistoric earthquakes are provided by ity was high at many places when the 1886 earth-
comparison of their liquefaction effects with quake struck.
worldwide observations and also by comparison As noted above, craters having ages of 600 and
with observations of liquefaction in South Carolina 1250 years extend along the coast at least as far
in 1886. Data from the 1886 earthquake furnish a as craters of the 1886 earthquake extend. A com-
data base for the regional development of craters parison of the size (diameter) of the craters shows
as well as their size and abundance. Worldwide that those formed 600 years and 1250 years ago
data show that features having a liquefaction are larger than the 1886 craters, both in the vicinity
origin can be developed at magnitudes as low as of Charleston and far away. Consideration of all
about 5 but that a magnitude of about 5.5 is the factors suggests that these prehistoric earthquakes
lower limit at which liquefaction effects are rela- were at least as strong as the 1886 event that had
tively common (Ambraseys, 1988). The source a magnitude of about 7.5.
sands that produced craters in coastal South Paleoliquefaction evidence for the event that
Carolina commonly are highly susceptible to lique- took place 3200 years ago has been found only in
faction and flow; because of this susceptibility, one the vicinity of Charleston. The existence of abun-
might suggest that a low-magnitude earthquake dant, exceptionally large craters for this event
could have produced the prehistoric craters. might suggest that the earthquake was exception-
However, numerous prehistoric craters in the ally large, but the limited size of the affected area
Charleston area, many having diameters in excess suggests otherwise. The absence of craters far from
of 3 m, clearly are too large to have been the result Charleston might be explained alternatively by a
of marginal liquefaction. Such large diameters lower sea level and thus a lower water table, or by
suggest that the earthquake that produced them a generally drier climate earlier during this part of
had a magnitude much greater than 5-5.5. In the Holocene (Amick et al., 1990). Absence of the
addition, the prehistoric craters that formed 600 3200-year-old craters far from Charleston might
and 1250 years ago extend along the coast at least also be explained by an exceptionally shallow
24 S. E Obermeier/Engineering Geology 44 (1996) 1-76

earthquake, in which energy attenuated rapidly 94" 92 90~ 88" 86"

within a short distance.


For craters 5000 years old or older, there is a
J IOWA .~..
greatly diminished chance for preservation of i r'., i HIGAN/
organic material that can be dated with accuracy, J ) ILLINOIS : -4
so it is difficult to evaluate the magnitude of such i .j I I

old events. i : I
40" i- I } q
Some of the craters far to the north of the "X WabashRiver .INDIANA /
Charleston area, in the vicinity of the South
"\ V~bash F ' ~ !
Carolina-North Carolina border (Fig. 11), have ~ Valley [ ~ / I:)
ages different from those of craters to the south. i St. LoUiSe,) seismic --./. ..,' / I
This difference suggests that another meizoseismal I|L MISSOURI " zone / ,./ / ;
region may be located in the vicinity of the State ! Madrid \-: I /"-/Oh}oRiver t
boundary. I'
seismic / ,--" ...... I
:,/~_ . r~P_NIUCKY
zone -- v.. ,~ '" "-~. ,-t
4.2. N e w Madrid seismic zone

The 1811-1812 sequence of earthquakes in the


Central United States consisted of four very strong f ARKANSAS TENNESSEE !
)e M e m p h s 'i
earthquakes (having magnitudes of about 7.8-8.3)
within a 3-month period. Six aftershocks had J o so loo~ .~.-~ J'/ :
I I

magnitudes of 6-7 (Hamilton and Johnston, 3 ,~ .~.,.,. j i


1990). Epicenters of the strongest 1811-1812 ' ,(.~ MISSISSIPPI i ALABAMA i
earthquakes probably were distributed along a t : I I
L ~ : 't
fault zone exceeding 100 km in length (McKeown
et al., 1990). These epicenters, in combination with Fig. 16. Approximate limits of the New Madrid seismic zone
continuing seismicity, define the New Madrid and the Wabash Valley seismic zone. The New Madrid seismic
seismic zone (Fig. 16). zone is the source area of the 1811-1812 earthquakes and con-
The meizoseismal region for the 1811-1812 tinues to have many small earthquakes and a few slightly dam-
aging earthquakes. The Wabash Valley seismic zone is a weakly
earthquakes was centered in a huge area of alluvial defined zone of seismicity having infrequent small to slightly
lowlands. Prominent effects of liquefaction damaging earthquakes.
extended over an area of 10 000 km; in the low-
lands and are plainly visible on the ground today
(Fig. 17 and Fig. 18). Large areas have more than the braid-bar terraces are large areas of Mississippi
25% of the surface covered with vented sand more River meander-belt deposits that were laid down
than 1 m thick (Obermeier, 1989). during Holocene time. Most of the meander belt
The alluvial lowlands is an area of very low consists of point-bar accretion topography of arcu-
relief, a very high water table, and a clay-rich cap ate ridges and swales, abandoned channels, and
over thick strata of fine and medium sand at natural levees. Some abandoned channels are filled
shallow depth. The sand strata generally are mod- with as much as 30 m of soft clay and silt. A cap
erately compact. The lowland is made up largely of montmorillonite-rich clay at least a few meters
of late Wisconsinan braid-bar terraces that formed thick lies on meander-belt sediments at most
in floods of glacial meltwater carrying great quanti- places. Overall, the alluvial lowland region is quite
ties of sand. Thickness of sand beneath the terraces susceptible to the formation of earthquake-induced
generally exceeds 30 m, and, at most places, the liquefaction features during strong shaking.
sands are capped with clay-rich strata interbedded Reports made shortly after the 1811-1812 earth-
with thin sand and silt strata having a total cap quakes noted great multitudes of sand blows,
thickness of a few meters. Inset slightly lower into linear fissures as deep as 6 m and hundreds of
S.F. Obermeier/EngineeringGeology44 (1996) 1-76 25

91" 90"
readily visible on the ground surface. Great num-
bers of intruded dikes and sills can be seen in walls
of deep ( > 3-4 m) drainage ditches.
37*
Also within the meizoseismal region of the
1811-1812 earthquakes are many sedimentary fea-
tures of unknown or nonseismic origin. Mainly,
the features formed as nonseismic sand boils, mima
mounds, or deformed mud.
The discussion following is almost exclusively a
description of liquefaction-induced features in the
meizoseismal region of the 1811-1812 earthquakes,
where modified Mercalli intensity values typically
ranged between X (severe damage to man-made
36*
structures) and XII (total destruction).

4.2.1. Characterization of venting and fracturing at


the ground surface
Even though sand that was vented to the surface
by the earthquakes of 1811-1812 is still visible
today, most evidence of venting has been obliter-
ated by agricultural practices. However, even small
features were abundant at the time of a field study
by Fuller (1912). Individual sand blows induced
EXPLANA~ON by the 1811-1812 earthquakes typically are dome-
like accumulations of clean sand on the ground
Upland area== Alluvium with mora than
25 pement of ground surface
surface. Fuller (p. 79) noted that "the normal blow
[ I COVered by sand-bfow deposits is a patch of sand nearly circular in shape, from 8
Alluvium with few or i
to 15 feet across, and 3 to 6 inches high." Such
no observed sand blows "t =
Eplcentsrs for three largest small sand blows as Fuller described can rarely be
1811-1812 earthqullum
according to Nutffi (1979) found at present. The sand blows now obvious
Alluvium w;th recognizable
sand-blow deposits (>1 per- range from about 0.3 to 0.7 m in height at the
cent ground coverage), but
not major ground coveroge ~ Fault .11' Fault zone center, and they thin to a featheredge at a distance
/ I
o f 5-20 m from the center.
Sand vented to the surface by the 1811-1812
Fig. 17. Regions having abundant vented sand, excluding
modem flood plains, in the New Madrid seismic zone (from earthquakes is obvious on aerial photographs of
Obermeier et al., 1990). Sand was presumably vented in areas where the sand vented onto the dark clay-
response to 1811-1812 earthquakes. Severe liquefaction also rich cap. The vented sand dries more rapidly than
occurred locallybeyond the areas shown on the map, especially clay during seasonal drying, making a tonal con-
along streams west of Crowleys Ridge (Fuller, 1912). Also
shown are the approximate epicenters for the three strongest trast on the photograph. Fig. 18A and B illustrates
1811-1812 earthquakes and major faults and fault zones the contrast and also illustrate how venting has
(Nuttli, 1979). taken place at irregular and somewhat erratically
spaced intervals. The photographs also show that
extensive venting took place through approxi-
meters long, craters many meters in diameter, and mately linear dikes that are more or less parallel.
lateral spreads as long as hundreds of meters These patterns of erratic spacing and parallelism
(Penick, 1976). Individual and coalesced sand generally reflect small differences in site character-
blows and some long linear fissures through which istics. One o f the most obvious is cap thickness.
sand vented are the only features that are still Typically, venting in point-bar deposits has taken
26 S. E Obermeier/Engineermg Geology 44 (1996) 1 70

(? 1 MILE
~- - - - I I
0 ! KILOMETER

(a) Location

Fig. 18. Aerial photographs that show long fissures (dikesj through which sand vented (light-colored linear features) and also show
individual sand blows (light-colored spots) formed by liquefaction during the 1811 1812 New Madrid earthquakes. (A) Part of the
Manila, Arkansas, 7.5-minute orthophotographic quadrangle (US Geological Survey, 19741. Fissuring and venting took place in
braid-bar deposits of latest Pleistocene age and in younger Holocene point-bar sediments. Note how fissures formed parallel to the
scrolls of point-bar deposits. Note also the abundance of fissures in the area shown in the upper right side of the photograph. These
fissures have formed near a break in slope, where the terrace of braid-bar deposits is adjacent to the slightly lower flood-plain level
of point-bar deposits. (B) Aerial photograph taken in 1959 about 40 km north of the area shown in (A). Extensive fissuring and
venting took place in braid-bar deposits of latest Pleistocene age. Cap thickness (about 6 m) is relatively uniform. Topographic relief
is only on the order of 1 m throughout the region of extensive fissuring. Note severe fissuring over a width of at least 3 km.
S.F. Obermeier/EngineeringGeology44 (1996) 1-76 27

0 1 MILE

(b) o , KIt.OMETER

place along the highest, thinnest part of the mean- of meters away. Fuller (1912, p. 49) for example
der scroll. Where the cap is thicker, sand blows stated,
tend to be less abundant but larger. The wider
spacing between dikes apparently causes more "In the sand-blow districts the spacing of [lateral
concentrated flow to the surface. spreading] fissures varies from several hundred feet
Long, linear dikes, commonly with exceptionally down to less than 10 feet . . . In the case of the
large quantities of vented sand, also tend to large [several meters wide] fault-block [lateral
develop parallel to topographic declivities along spread] fissures the spacing is greater, several
streams and scarps. Dikes here have formed in hundred feet often intervening between the cracks,
response to lateral spreading movements, which while the space between them may be half a mile
generally take place more readily near the declivi- or more. Isolated cracks of this type are not
ties (Fig. 18A). Wide (> 1 m) dikes having lengths uncommon."
of many hundreds of meters are not unusual. The direction of shaking during the earthquake
Although the widest dikes tend to be close to the probably had a secondary influence on orientation
declivities, they also may develop many hundreds of the widest dikes at most places in the meizo-
28 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

seismal region of the 1811-1812 earthquakes sively in the lowermost quarter of vented material.
(Obermeier, 1989). The local geological-topo- Above this in the larger sand blows, away from
graphic setting is the predominant influence. Cap the vent, is a much thicker zone (tens of centime-
thickness and proximity to stream banks and aban- ters) of very clean, generally medium to coarse
doned meanders are most important. Such impor- sand that is nearly structureless except for sugges-
tant influence of the local setting has also been tions of laminations of sediment; grain size of this
shown in a report about the 1964 Alaskan earth- sand corresponds very closely to that of the fluid-
quake (M= 9.2) by McCulloch and Bonilla (1970, ized source beds at depth. Higher yet in the sand
Fig. 46) and was emphasized by Oldham (1899) blow, the sand grades upward to a mainly medium
for the great earthquake of 1897 in India, so the sand. Here there are weakly to moderately devel-
observed effects in the meizoseismal region of the oped planar to wavy laminations of silt and sand
1811-1812 earthquakes seem typical. Still, local- generally a few millimeters in thickness, which
ized extension and compression of the ground gently dip down and away from the central part
surface, which may relate to the direction of strong of the sand volcano (Fig. 19A).
surface shaking, may be relatively common Where sediment vented into swales on the
(Oldham, 1899, p. 99). ground surface, this sand-blow sequence may be
The aerial photographs in Fig. 18 indicate that capped by a silt-rich stratum containing organic
there is enough randomness in dike orientation debris and ranging from 0.5 cm to several centime-
that most vertical exposures will intersect many ters in thickness; this cap may also contain multiple
dikes. This is especially relevant because searches
very thin (1 mm thick) clay-rich layers (Saucier,
for paleoliquefaction features are often made in
1989). The organic matter in the cap consists of
banks of ditches or rivers, which may not be
small pieces of charcoal and wood. The organic
oriented optimally to intersect dikes.
debris and thin clay-rich layers were deposited in
swales located both above the vent (Fig. 19A) and
4.2.2. Characterization of sand blows and dikes in
in depressions far from the mound of vented
sectional view
sediment.
Most sand blows of the 1811-1812 earthquake
Closest to where dikes vented onto the ground
have a well-defined set of internal relations and
surface, there can be well-defined strata that dip
stratigraphy, shown in a somewhat idealized ver-
sion in Fig. 19. The main feeder dike is beneath steeply into the dike (Fig. 19A). These strata typi-
the central part of the dome. The basal few centi- cally contain the coarsest sediment vented. Next
meters of sediment that vented onto the original to these strata, in the lower and central part of
ground generally is a fine to medium sand with a the sand blow, there may be evidence of shearing
slight to moderate amount of silt, containing scat- and disruption caused by the forceful expulsion
tered centimeter-long, round to irregular clasts and boiling of sediment and water.
derived from the underlying clay-rich strata cut by The overall fining-upward sequence of vented
the dike. The basal part of the sand-blow deposit sediment, from the basal clast-bearing sand to the
also contains numerous 1-3-cm-long, rounded lig- uppermost organic-matter-bearing stratum, repre-
nite fragments and other low-density materials sents the transition from the turbulent violent
vented to the surface. These low-density materials eruption very shortly after initial venting to the
originated from the source sand. final ebbing flow to the ground surface. The planar
Sediment above the basal few centimeters grades and wavy laminations probably represent weak
up within a few centimeters to coarser sand with variations in flow from the dike.
minor silt containing numerous irregular, 1-5-cm- Dikes that formed in the clay-rich cap of the
long, clay-rich clasts. The clasts are encased in 1811-1812 earthquakes' meizoseismal region typi-
clean, medium to coarse sand. The clasts are cally are sand-filled fissures that are steeply dipping
largest and most plentiful near the feeder dike. (60-90 ) and are mainly planar. In vertical section,
Clasts derived from the cap occur almost exclu- dikes having widths exceeding several centimeters
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 29

lOto40m
Stratum of silt
with organic debris \ Fining-up sequence
sand with minor silt
~ - Laminations
0.3 to ~ " ' " " ' "" ~" - " " " ~ . . . ,' - . . . ~

8 .~" . . '

Basal zone of sand containing


o
many clasts of sidewall material
and wood and coal fragments Ground
/ cracking
~ ~ ~ : ( / ~ ~'\ Bifurcatingdikes
"6 Tabular sand-filled fissure - ' ~.,.: caused by weathering in weathered zone
~_~ (dike), commonly 0.15 to 0.3 m : .
,E ~ wide; may contain clasts of !'-'
=~ ~ sidewall material "'i
pinch together upward
2~ 'I.
:'.'.
A
ii. B c
_J ....

::'heddinghes ::i
: " ....." . '......,: .. '. /...: : ....:.. -..':.-.......:: ::: v.'.:........:: ...:... '. -...:..:.' . . . c...: ~*...:.'~. ~ ,~.::;
..~.:...~ ... .... . .. :. ~........,~ . . . ~.......,~. :..~.......~ . .:.....:

:::::::::::::::::::::::::::::::::::::::::::::::
:.:::::::.:: ::::::::::::::::::::::
.:.:::G"-:::.::::::: .'~::):::?~:
? :::. ::::: ;;,;}:~:':':~:?
~: ::: :.~':;'~:~';:::;:
?.:::.
:;.ZS:'::::>i
:.~.::~'.-::::';~
"::::
:'.:.~. o ,-~:'i

Fig. 19. Schematic vertical section showing dikes cutting through overbank silt and clay strata and showing the overlying sand-blow
deposits. (A) Stratigraphic details of sediment vented to the surface. (B) Dikes that pinch together as they ascend. (C) Characteristics
of dikes in fractured zone of weathering, in highly plastic clays. Situations shown are found in many places in the meizoseismal zone
of the 1811-1812 New Madrid earthquakes.

commonly are spaced from several meters to hun- common. Dikes in this width range normally
dreds of meters apart. become more narrow upward as illustrated in
Dike widths range from millimeters to several Fig. 19A. The tapering may represent downwarp-
meters. Many of the widest "dikes" are sand-filled ing of the ground surface in response to sand at
fissures that were almost certainly caused by lateral depth having been vented to the surface. At almost
spreading. Fig. 20 shows a feature probably having all places, even isolated sand blows have vented
such an origin; the lignitic silty clay stratum in the through small, vertically planar dikes. The smallest
sand indicates the location of the top of the sand dikes pinch together as illustrated in Fig. 19B.
($1) that flowed into the opening of the lateral Within the uppermost meter or so, dikes cutting
spread. The upper sand ($2) probably was vented through the weathered portion of the clay-rich cap
later during a following earthquake. (In contrast, may branch irregularly upward into many smaller
often only little or no sand vents to the ground segments (Fig. 19C). Possibly preexisting weather-
surface from the large space between blocks ing planes of weakness cause a single large dike to
opened by lateral spreading.) Sidewalls of many branch into many small members. The clay cap
of the larger sand dikes throughout the 1811-1812 typically contains a large proportion of montmoril-
earthquakes' meizoseismal zone tend to be parallel lonite. During dry years, desiccation cracks extend
to one another in vertical section, which also a meter or more in depth. Pedogenesis has also
indicates a pulling-apart origin during lateral developed a strong soil structure (pedons) in a
spreading. thick B horizon near the surface.
Dikes about 15 cm or less in width are very Where larger dikes branch extensively
30 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

..... :. :.: ::!: :. i:ii! i!. " :: : ' .....


:i:.." .!:i.7,7."...:
i,;21;";

0 1 2 METERS
I I I
NO VERTICALEXAGGERATION

Fig. 20. Sketch of vertical exposure in a ditch in the meizoseismal zone of the 1811 1812 New Madrid earthquakes showing evidence
for about 1.5 m of lateral spreading. The $1 sand was emplaced during lateral spreading. The lignitic silty clay layer next was laid
down on Sx sand. In a second earthquake, S 2 sand was vented to the surface, burying the lignitic silty clay and $1 sand. Note that
the sidewalls are approximately parallel. From Wesnousky and Leffler (1992).

(Fig. 19C), there may be only minor evidence of in width. Locally, winnowing extends several
venting onto the ground surface. Apparently this meters down into the dike. This winnowing proba-
network is effective in dissipating the energy of bly took place by water flowing up through the
flow. Locally though, venting has excavated the dike during final phases of expulsion of water,
highly fractured zone, leaving behind a widened following initial emplacement of the sand in the
dike at the top. This excavated zone may contain dike.
many clay clasts mixed irregularly in a matrix Many dikes that pinch together upward have a
of sand. large proportion of silt and clay mixed with the
Dikes that cut through the fine-grained cap sand near the top. Often it is unclear whether the
generally are filled with a loose mixture of fine silt and clay have invaded the dike by pedogenesis
and medium sand and a minor amount of silt. or whether the silt and clay were constituents of
Clasts of clay, some as long as 20 cm, may also the originally intruded sediment. Dikes that taper
occur but generally are not abundant. Elongate upward are generally of limited usefulness for
clasts tend to be parallel to sidewalls. The clasts paleoliquefaction studies, owing to difficulties in
were derived from the sidewalls and transported determining when the dikes formed.
up the dike. The mixture of sand, silt, and clasts Many variations of the relations shown in
has a sharply defined contact with the sidewalls. Fig. 19 exist. One of the most common is a large
Weak laminations within the sand and silt may amount of downwarping of the cap toward the
parallel the sidewalls. Crosscutting, vertically ori- dike. This downwarping tends to be most pro-
ented zones of sand and silt within the dike are nounced where a large amount of sand has been
also commonplace. These probably represent epi- vented to the surface. It is not unusual that the
sodes of venting during separate pulses or venting cap be downdropped by more than 0.5 m on one
from different source zones at depth. or both sides of the dike, and that the cap be
The finest constituents (fine sand and silt) have otherwise faulted or severely deformed near the
been winnowed from dike fillings at some sites of dike (Wesnousky and Leffler, 1992, Figs. 9, 12,
venting. Winnowed zones within the dike are com- and 14).
monly tubular and as much as several centimeters Also relatively common in the meizoseismal
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 31

region of the 1811-1812 earthquakes are dikes also are commonly observed conduits for sand
having a morphology that differs substantially venting during the 1811-1812 earthquakes.
from the type shown in Fig. 19, where a feeder Another field example, in Venezuela, of where
dike at the base of the cap is a single, planar break venting was localized in tubular (crab) burrows
through the cap. The substantially different dikes was reported by Audemard and de Santis (1991);
are often irregular, tabular, sand-filled breaks that this venting occurred in response to limited lique-
are discontinuous and short (< 1-2 m) in length, faction during moderate earthquakes.
irregular in width, and anastomose in all orienta- The preceding discussion has focused on charac-
tions. Such dikes appear to have formed only teristics of dikes that cut through a clay-rich cap
where the cap was so weak as to have deformed that ranges in consistency from very soft to brittle.
very plastically. Original bedding in the host sedi- Excellent descriptions and detailed drawings of
ments within the cap can be strongly warped. Not tabular dikes that cut interbedded clay and sand
unusual are dikes having walls that have squeezed strata during the 1989 Loma Prieta, California,
together except for a width of only a few sand earthquake (M= 7) were given by Sims and Garvin
grains. Even though plastic deformation has (1995). The dikes and sand blows that they
allowed considerable warping of silt and clay beds described generally are much smaller and represent
in the host, most dikes that terminate within the less forceful venting than those of the 1811-1812
host do so by pinching together as illustrated in earthquakes, yet the sediment relations in the dikes
Fig. 19B. Because the dikes narrow upward and and sand blows in both regions are quite similar.
terminate at a point, hydraulic fracturing still Locally, the cap in the 1811-1812 earthquakes'
appears to have been the dominant mechanism meizoseismal region is a very weakly cemented
whereby the dikes formed, rather than intrusion sand containing only a minor amount of silt and
of liquefied sediment causing plastic deformation clay, with slight bonding imparted from oxides of
of the host. The anastomosing, discontinuous dikes iron and manganese. The liquefaction features in
often have developed only near the base of the the areas of weakly cemented sand caps appear to
cap, where sediments are softest. Such dikes may have been mainly large open craters, similar to
be relatively common in earthquakes where seismic those of the 1886 Charleston, SC, earthquake. In
shaking has been very strong and the cap was firm contrast to filled craters in South Carolina, craters
yet quite plastic. For example, Walsh et al. (1995) from the 1811-1812 earthquakes generally lack
noted that this is the dominant morphology that very large dasts in their lowermost deposits (see
they observed in the meizoseismal region (modified sketches in Tuttle, 1994). Apparently the h o s t
Mercalli intensity values of IX-X) of the M 9.2 sands are too friable to form large clasts.
Alaska earthquake of 1964; many excellent photo-
graphs and line drawings are shown in their report. 4.2.3. Characterization of sills in sectional view
Some scattered small tubular dikes can also be Combinations of dikes and sills are also common
found in clay caps of the 1811-1812 earthquakes' within the nonliquefiable cap. Where sills are abun-
meizoseismal region. Holes that originated from dant, dikes generally are also plentiful. Sills form
decay of tree roots or from excavation by crayfish preferentially at three locales: (1) along the base
are ubiquitous and doubtless were used as the of the cap, (2) along bedding planes and other
conduits for small tubular dikes. These tubular horizontal planes of weakness in the cap, and (3)
holes through the cap had a very minor role as beneath exceptionally dense, strong root mats.
conduits for venting, though, as compared with These three locales are illustrated in Fig. 21. All
the role of steeply dipping planar dikes. However, examples (unless otherwise noted) of dikes and
these holes possibly were preferred paths during sills in the New Madrid area discussed in this
the early phase of venting and thereby controlled section resulted from liquefaction caused by
where hydraulic fracturing developed planar dikes. seismic activity.
Small holes whose walls are defined by angular Laterally extensive sills as thick as 0.5 m are
breaks, and which have a tortuous upward path, commonplace along the base of the cap. An intrud-
32 S.F Obermeier/'Engineering Geology 44 (1996) 1-76

, . , .. .

~ ' i ' - ' . Loose silty.


l'e~'" ,, : - i " . , . - . : - sand with,
{ ' / " :]:" .'" i: '.;::.traceclay
v ~ ~ . -- ~

' "" " " , " " " ." " , " " -i ." ', "
."Siltysand /~Cap
Bowl-shaped intrusion L Laminations in sill Soft clay

-. Sand with
~ ' : : i siltstratum
~Sill Soft, friable
silt stratum
clayey silt

Fig. 21. Schematic vertical section showing where sills form preferentially at the base of a thick nonliquefiable cap overlying source
sand, along bedding planes within the cap, and beneath a dense root mat. Thin sills can extend great distances horizontally, especially
where the overlying cap is thin. Such severe sill development as illustrated in the figure is typically accompanied by large sand blows,
as in the meizoseismal zone of the 1811-t812 New Madrid earthquakes. Bowl-shaped intrusions often are associated with sills; a
bowl commonly contains a fining-upward sequence of clasts.

ing sill can dome up an overlying flexible, clay- angular, suggests a brittle or shattering mode of
rich cap having a thickness of as much as a few fracturing of the stratum from which the clasts
meters. Sills are also common within the cap where were derived. Despite this pattern of fracturing,
original horizontal sedimentary structures and the clay-rich clasts and their source beds commonly
planes of weakness have not been destroyed by have such a soft consistency as to permit very easy
weathering. Sills especially tend to form irregularly penetration of several centimeters by thumb
within thin beds of silt or sand sandwiched between pressure.
clay-rich beds. Small branches from the main sill Fig. 21 also shows intrusions that are bowl
commonly intrude into an overlying more clayey shaped in three dimensions. The bowl-shaped
stratum, forming more sills and dikes. intrusion typically has developed within a thin
Clay-rich clasts can abound within a sill. The ( 5-20 cm) stratum of sand. Connected to the base
clasts have been transported many meters horizon- of the bowl-shaped feature is a thin (a few millime-
tally at many places, but clasps that have simply ters to centimeters) tabular dike that has served as
been detached and foundered vertically several the conduit for the fluidized mixture of sand and
centimeters are also very common (see Fig. 21). water to travel upward. Within the bowl portion,
The shape of clay-rich clasts in the sills, generally there is generally a fining-upwards sequence of
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 33

angular clasts derived from the directly underlying


fine-grained stratum. It is not unusual, though,
that some of the largest clasts be at the top of the
bowl-shaped feature, just beneath the overlying
fine-grained stratum. Bowl-shaped features are
commonplace, especially where a fine-grained stra-
tum as thin as a few centimeters to 10 cm is
underlain by a thick, liquefied sand stratum.
Sills at a depth of about a meter or less in a
clay-rich cap can be quite wavy in vertical section.
These sills can thin and thicken dramatically within
a horizontal distance of a meter or less and produce
blister-like bulges on the surface. In plan view,
these bulges can range from circular to very elon- 50 INCHES
J
gate. Sills can be as thick as 0.7 m near the surface. I
100CENTIMETERS
Such large thicknesses appear to be less c o m m o n
at greater depths, except locally along the base of
a thick cap. Sills very close to the ground surface
generally seem to have formed beneath root mats
of grass and trees.
Fig. 22A shows a commonly observed field
example in which near-vertical dikes are connected
to a laterally extensive sill. The sill has formed
within a clay cap that has only incipient horizons
of weakness. The sill is at least 25 m long but is
only 10 cm thick. The sill more or less follows a
single horizon in sectional view. The type of
internal layering seen in this sill is commonplace.
Individual laminae are composed of small pieces
of lignite or fine sand and silt (Fig. 22B,C). Also
c o m m o n at this site and elsewhere are structureless
sills containing m a n y clay clasts in a sand matrix,
as well as sand with graded bedding of clay clasts
concentrated along the base.
Laterally extensive sills also tend to form in the
upslope direction along the base of a cap, where
the base dips appreciably. Such sills especi?.lly

Fig. 22. Sand dikes and sills exposed in a nonliquefiable cap of


silt and clay; the dikes and sills are interpreted as having origi-
nated by liquefaction during the 1811-1812 New Madrid earth-
quakes. Outcrop is the bank of a ditch in the meizoseismalarea
of the 1811-1812 earthquakes. (A) Typical small dikes and
small sills. The most continuous sill extends far beyond the area
2 INCHES
shown in the photograph and is at least 25 m long. Rectangle I
shows area of (B). (B) View showing details of layering in a I
sill. (C) Veryclose view showing details of layering in a sill. Sill 5 CENTIMETERS
is made up of fine and medium sand with some silt- and sand-
sized lignite. Black bands consist of small pieces of lignite.
34 S.F Obermeier/Engineering Geology 44 (1996) 1-76

occur along the base of a clay-filled abandoned The exposure illustrated in Fig. 23 shows
channel; here a sill typically extends upslope to another common type of field example in which
where the cap is thinner, and the sill feeds into a sand dikes and sills cut the upper liquefied sands
steeply dipping dike that has vented onto the and the lower portion of the nonliquefiable cap.
surface (M.P. Tuttle, Univ. of Maryland, written In the upper part of the liquefied sand (bed A,
commun., 1993). Fig. 23A), small dikes branch out from a large

GROUND SURFACE BED

Clay, blocky; altered by


soil processes

~i : i ~.~ : i ~I i
' : : ": ' . : : . -. ' : : . i." ' . ' . " 1H Sand, fine-tomedium-
... . . : - : . . . . . - . .. ' ... ' ... - . ' : . . . . . grained, clean,
'.".".". :"i : i: ( 4 ""~. '- ~ , / ~.". / ~' 'C l a y c l a s t s - :"i :.i: '. structureless
: ,:, : , ....
-- I -~--Mediurn s a n d - t ~ / ~ . - - I -- Clay, blocky,
_ t _ ~ ~ -__~__~c~ - I~ ~ __ G micro-fractured

~ F Silty sand, fine-grained


bedded
E Silty sand, fine-grained
J
-~ - ~ ~ _ ~ 7~- CIay, highly plastic,
D grading up to
silty sand

Sandy clay, coarsening


upward to sand,
weakly laminated
Sandy silt,
i laminated at base

Sand, medium-grained,
clean, crossbedded,
contains wood
(dated at 12,000 yrs
before present)

0 3 FEET
I r
0 1 METER

NO VERTICAL EXAGGERATION

Fig. 23. Schematic vertical section and photographs showing Holocene sediments (nonliquefiable cap) and underlying Wisconsinan
braid-bar source sand that liquefied. Exposure is in a ditch in the meizoseismal region of the 1811-1812 New Madrid earthquakes.
Earthquake-induced intrusions cut section at many places. (A) Schematic diagram of stratigraphic relations and liquefaction-induced
features (numbered 1-6). Feature 1: dikes of medium sand that cut cap and source sand. Features 2 and 3: intruded dikes and sills
of structureless, clean, medium sand. Feature 4: dike and flame structures of medium sand containing large clasts from bed G. Feature
5: dikes of medium sand, truncated unconformably. Feature 6: pseudonodules collapsed in bed D. (B) Photograph of beds A-D,
showing sand intrusion (feature 3). Knife shown in the four photographs is 12 cm long. (C) Photograph of dikes (feature 1) cutting
source (bed A) and bed B. (D) Photograph of plan view of bed B, showing intrusion structures caused by feature 1; the plan view
is in the area of the knife that is oriented vertically in (B). (E) Photograph of part of feature 4, showing clay clasts in sand matrix
intruded into bed H.
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 35

clay
clasts

dike (feature 1), cut through the basal bed of the


cap (bed B) at horizontal intervals of 0.5-1 m,
I
the very soft clay with sharp angular turns and
breaks that follow a haphazard path (features 2
and extend upward about 0.5-1 m. A few dikes through 3). Edges of intrusions are generally less
and sills intrude to much higher levels (features 2, distinct in beds of clean, permeable sand. Dikes
3, and 4). At exposures nearby (not shown), dikes commonly widen and terminate upward as flame-
extend to the surface. Sand has vented to the shaped structures (feature 4) in a permeable sand
surface to produce many large sand blows in the bed (bed H). Clay clasts may be present in lower
field adjoining the outcrop of Fig. 23. portions of the structures.
The dikes and sills shown in Fig. 23A as features The sill at the base of bed B, at the extreme
1 through 4 contain clean medium sand. The edges right side of the figure, has an irregular contact
of the intrusions are sharp in clay-rich beds such along the top. Here the sill has corraded the base
as beds C, D, and G. Locally, intrusions puncture of the friable bed B. Small intact pieces of bed B
36 S. I~ Obermeier/ Engineermg Geology 44 (1996) 1-76

have sunk into the sill, probably attesting to a the cap contains a convoluted mixture of severely
very water-rich condition in the sill at the time of disturbed, plastically deformed silt and clay,
its intrusion. Such destruction of friable beds by intruded by sand. Such convolutions probably take
sills is commonplace throughout the meizoseismal place only where extremely soft silt and clay lie
region, especially along the base of the cap. In directly on liquefied sand.
regions of only marginal development of small The preceding discussion has focused on sill
dikes, thin (< 10 cm) sills along the base of the cap development in the meizoseismal zone of the
and within the cap are still relatively common. 1811-1812 New Madrid earthquakes. Beyond the
Features labeled 1 through 4 in Fig. 23 are meizoseismal zone, sills within the cap are much
interpreted to be earthquake induced because: (1) sparser than those along the base of the cap. Sills
they are widely distributed over tens of kilometers; within the cap are also sparse in comparison to
(2) many of these dikes and sills are as wide as the number of dikes beyond the meizoseismal zone.
15 crn, and this width suggests intrusion by large Still, along the base of the cap, sills having thick-
volumes of water-saturated sediment; (3) they nesses of 3-10 cm occur even where dikes have
consist of clean medium sand containing large, widths of only several centimeters. I have observed
angular clay clasts (which are evidence of forceful in other tectonic regions that such sill development
intrusion); and (4) artesian conditions are unlikely can be commonplace along the base of the cap,
at these sites. even where dikes are small and widely spaced.
The dike shown as feature 5 has an uncertain Sill development in the meizoseismal zone of the
origin (Fig. 23A). Three small dikes that were 1811 - 1812 New Madrid earthquakes is extraordi-
truncated at the contact between beds C and D nary in comparison to what I have seen in tectonic
were exposed in a 25-m section along the ditch regions elsewhere. Such extensive sill development,
but were not found in other nearby exposures of especially within the cap, may require somewhat
beds C and D. The dikes may represent special conditions. I suspect (1) that earthquake
pre- 1811-1812 seismic liquefaction. However, they shaking must be very strong, (2) that the water
contain a large proportion of silty fine sand, which table must extend far up into the cap, and (3) that
does not suggest forceful intrusion at this site. the cap must have abundant horizontal zones of
Possibly they resulted from springs that formed weakness.
near the base of a stream bank, or they may be The possible seismic liquefaction origin of sill-
slump-related features that formed soon after ini- like lenses can be difficult to deduce in many field
tial deposition of the host sediments. situations. The problem of origin is especially
Pseudonodules (feature 6) are discussed in a difficult for some very thin ( 1-2 cm) sill-like lenses
following section, "Load Structures in Muds." of sand found in the transition zone where a
Flame-shaped structures such as those of feature deposit of thick, clean sand fines upward to a
4 have developed at many places within thick sand deposit containing more silt and clay. Within the
beds in the meizoseismal region of the 1811 - 1812 transition zone are thin sand deposits intercalated
earthquakes. Flame-shaped structures commonly with thin lenses of muddy sediment, in many field
have widths ranging from millimeters to about a settings. Sequences of this nature may be repeated
third of a meter. Apparently the upper sand beds vertically and laterally many times. It is not unu-
do not liquefy, and, because of their relatively high sual in this scenario to also observe a thin, sill-like
permeability, perhaps in combination with an sand intrusion going laterally and up into overlying
unsaturated condition, they permit the energy of muddy sediment. Such an intrusion can originate
the pressurized water from beneath to dissipate simply by compaction of very soft sediment caus-
within the large volume. ing buildup of pore-water pressure, or alternatively
A rather uncommon deformation feature (not can originate by seismic liquefaction. Only where
shown) probably related to seismic liquefaction liquefaction-induced sills have a thickness exceed-
involves plastic deformation of silt and clay along ing a few centimeters are there likely to be found
the base of the cap. The lowermost 10-20 cm of characteristics that indicate a seismic origin. Here
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 37

there can also be steeply dipping dikes cutting 4.3. Wabash Valley seismic zone
through a fine-grained stratum along the base of
the sill. Dikes of seismic origin commonly have Many small to slightly damaging earthquakes
fractured irregularly across the finer grained stra- have occurred throughout the region of the
tum and have formed clasts with sharp, angular Wabash River valley of Indiana and Illinois in the
shapes. Rounded clasts are sparse to nonexistent. past 200 years. Seismologists have long suspected
The bowl-shaped features shown in Fig. 21 have that the weakly defined Wabash Valley seismic
also developed. These relations occur plentifully zone (Fig. 16) could be capable of producing
in the New Madrid seismic zone as well as most earthquakes stronger than the largest of record
other seismically active regions that I have (M~5.5; Hamilton and Johnston, 1990). A search
examined. for paleoliquefaction features has resulted in dis-
covery of evidence for very large, prehistoric,
4.2.4. Paleoliquefaction studies Holocene earthquakes.
Systematic paleoliquefaction studies were under- The main search was in the Wabash River valley
taken only recently throughout the meizoseismal along the central axis of the Wabash Valley seismic
region of the 1811-1812 earthquakes. This region zone. Alluvium generally as thick as 10-30 m lies
has large terraces of late Wisconsinan and early on bedrock. The valley contains expanses of low
Holocene age where the water table appears to glaciofluvial terraces of late Pleistocene age. These
have been very shallow since the terraces were terraces are mainly braid-bar deposits of gravel
formed (Wesnousky and Leffler, 1992). Thus, if and gravelly sand. Inset into the Pleistocene ter-
very strong earthquakes occurred since early races are slightly lower Holocene flood plains of
Holocene time, liquefaction features should be finer point-bar sediment. The sand and gravel
present in the geological record. Definitive evi- deposits of both braid bars and point bars typically
dence for liquefaction predating 1811-1812 has are overlain by a 2-5-m-thick alluvial cap of sandy
been found in the northern part of the New Madrid silt to clayey silt. Bordering the valley are extensive
seismic zone (see Figs. 16 and 17) near Reelfoot plains of silt and clay that contain patches of clean
Lake (Russ, 1979) and at a site about 30km sand; the plains were laid down in slack-water
northeast of Reelfoot Lake (Saucier, 1991b). Russ areas during glaciofluvial alluviation. The water
found that three earthquakes have induced lique- table is presently shallow (< 3 m) and appears to
faction during the past 2000 years; on that basis, have been shallow over large areas much of the
he suggested a recurrence interval of 600 years for time following glaciofluvial alluviation, on the
liquefaction-producing events. Saucier (1991b) basis of depth of carbonate leaching and B-horizon
estimated an average recurrence interval of 470 soil development in sandy and silty alluvium. This
years for liquefaction-producing events in the past combination of a relatively shallow water table
1300 years. Further south, in the southern half of and widespread sand-rich deposits with an overly-
the seismic zone, field work has discovered evi- ing fine-grained cap has provided an excellent
dence of three prehistoric liquefaction-producing opportunity for liquefaction features to form
events in the past 2000 years (Tuttle and Schweig, throughout much of the Holocene.
1995). Whether these more southern episodes are
coeval with one another or with those further 4.3.1. Characterization of features
north has not yet been proven. Therefore, the Hundreds of dikes have been found. Sills within
upper limits for the magnitudes of these prehistoric the cap are sparse, even where thick sills occur
earthquakes cannot yet be determined. Only the along the base of the cap. Dikes and sills occur
lower limits can be estimated, and these are the both in Holocene point-bar deposits and in the
threshold values. The threshold magnitude is about late Pleistocene glacial outwash and slack-water
5.5 on the basis of historical observations of lique- deposits. Fig. 24 shows the area searched, sites
faction-producing events in the New Madrid where dikes were discovered, and sites of excep-
seismic zone. tionally wide dikes. Nearly all sites shown on the
ILLI~INNOIDISANA~
38 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

Limit of liquefaction
/ 6,100200 yr B.P.
,~ earthquake

Indianapolis

/
/

Limit of
/ /12,000:
.k.//// eartt

~ J ~ 0

EXPLANATION
50 100 KILOMETERS

- - Surveyed stream
[] City
Paleoliquefaction site
ylr 12,0002,000 yr B.P.
6,100200 yr B.P., dikes < 0.7m wide
C) 6,100200 yr B.P., dikes > 0.7m wide
A 3,0001,000 yr B.P.
Other events in Indiana
Undated events in Illinois
X Possibly caused by 1811-12
New Madrid earthquakes

Fig. 24. Area in Indiana and Illinois searched for liquefaction features, sites where paleoliquefaction features (mainly dikes) were
discovered, and regional limits of liquefaction for different earthquakes. A b o u t 10% of the length of the rivers searched has freshly
eroded exposures. Only exceptionally are there no fresh exposures of mid-Holocene or older sediments within a 20-km length of
river; no suitable exposures are in the region of shallow bedrock shown on the map. Liquefaction sites plotted on the m a p generally
have numerous dikes. Undated events in Illinois are almost all more than a few hundred years in age, yet Holocene. Some sites in
the southern part of the study area have unweathered dikes near the surface, probably induced by the 1811 1812 New Madrid
earthquakes. Dike width was measured at least 1 m above the base of the dike; sites with dikes having a width >0.7 m are shown
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 39

figure have more than 1 dike, and many sites have from the dikes onto the ground surface rather than
more than 10. Almost all the liquefaction sites intruded as sills is based on several lines of evi-
found are in actively eroding banks of rivers, which dence. Sills would cut irregularly across sedi-
were about the only areas searched. mentary horizons at some places (see Figs. 21 and
The dikes are steeply dipping, are tabular, and 22), rather than always being confined to a single
connect to a sediment source at depth. Many horizontal layer as in the Wabash Valley. Sills also
smaller dikes pinch together as they go up. Where would tend to follow the contact between sand
dikes cut through a thick clay cap without pro- and clay strata rather than lie on a paleosol, and
nounced horizontal planes of weakness, it is not sills probably would have some dikes branching
unusual for the dike width to be less than a up.
centimeter throughout a height of as much as 4-5
m. Widths as much as 15 cm or more throughout
4.3.2. Ages of dikes
the height of the dike are very common and
Radiocarbon dating and archeological evidence
widespread in the region. Locally, though, in what
at widespread sites show that most of the liquefac-
is interpreted as meizoseismal regions of the prehis-
tion features in Fig. 24 resulted from a single
toric earthquakes, dikes have widths ranging from
earthquake 6100___200 years ago (Munson et al.,
0.7 to 2.5 m.
1992,Munson et al., 1994). Very large dikes with
Dikes filled with sand containing some gravel
extensive sand blows from this earthquake are
and silt are very common. Vented sediment was
centered about Vincennes, implying a nearby epi-
mainly sand that commonly contained large quan-
center. The evidence for two pulses of venting
tifies of gravel (Fig. 25). Within many dikes is a
admits the possibility of more than one strong
fining-upward sequence of coarsest material. Also
earthquake spaced very closely in time. Such close
within many dikes are sharply defined, vertically
spacing of strong earthquakes happened in the
intertwined and intersecting zones containing dis-
nearby 1811-1812 New Madrid earthquakes.
tinctly different grain sizes, which in places can be
The next strongest earthquake in the Wabash
traced to different source strata at depth.
Valley took place about 12 000 years ago. Again,
Sediment vented to the surface extends as much
the largest dikes appear approximately central to
as 40 m in width. Thicknesses of vented sediment
the regional distribution. A yet smaller event
of 0.15-0.2 m are not unusual. Most vented sedi-
occurred about 3000 years ago, and a few other
ment fines upward and laterally, especially if gravel
scattered small events are represented. Other earth-
was vented. The vented sediment at sites bordering
quakes are represented by a few widespread dikes
the Wabash River generally lies on a paleosol and
of late Pleistocene age, but the paucity of such old
is buried beneath a 1-3-m thickness of overbank
exposures precludes determining a regional pattern
deposits, as illustrated in Fig. 25. On slightly ele-
of sizes.
vated terraces, where flooding has been rare, the
Bracketing the time when the dikes west of the
vented sediment has been incorporated into the
Wabash River formed has not yet been done.
surface soil.
However, the severity of weathering of vented
At some sites, there is good evidence for more
material and in dikes shows that almost all the
than one pulse of venting. This evidence is best
dikes have ages of thousands of years.
shown at sites near Vincennes (Fig. 24), where
both sand blows and their dikes are largest for the
event of 6100 year B.P. The vented sediment 4.3.3. Evidence for seismic origin
consists of two fining-upward sequences. The lower All aspects of the Wabash Valley dikes can also
pulse of vented sand fines upward to a thin silt be observed in the 1811-1812 New Madrid earth-
layer; the upper pulse has abundant sand and quakes' meizoseismal region, which has a physical
gravel that also fines upward. There is no evidence setting generally similar to that of the Wabash
of a significant hiatus between pulses. Valley. An earthquake-induced liquefaction origin
The interpretation that sediments were vented is interpreted for the dikes of the Wabash Valley
40 S.F Obermeier/Engineering Geology 44 (1996) 1-76

Ap

Gray silty clay

C/B

. ,
~ ~ ~ ~ --o-7...
Ped..~-mix zone . . " - ~ . . - - : ' - . - - . ~ . . . , :~-'--,.-----;

F-"-= .... .o o'" ." . . . . . '~ . .... ~. ~'.-~*"-:" '-~~'-I


1.i ."/ o -.:.o?~ o.~o. ~: ~ ..:.. ~. .". . ".~"~o~" : "~"- ;.~ "? " :/ Sand and
L'~'~O0"c~" LJ'.~'o'~. "" -" "- ~ - " " ~ - - o O@'U'~.'O'.L)O0." / gravel
~ - ' ~ "' : : "". . . . " : . o. ".' " . . ' . . 0 ~ , ~ , . ."-,-:-=~. ~ ' . I
LI ' I' l-~ '~ : /~J I --r--~ 2SIAb
L-____ J ~I j ---".'"' ~o.': ".'.( Blocky soil s t r u c t u r e ~ - ~

Yellowish-brown i ~ ': ""


.'." " 0 - ' ' '

silty clay loam I" ~'' ' " " "i


.. "o'" . ' : ' 2Bb
;. "'.~. ' ~ - ' ~ Clasts of (iron oxide
" 8" ?L.'.:~. sidewall material and clay
-I- ~ ~Qo'.o.'.'~ if;.~ enrichment

:0 ": Q..

i:,~ ." ~:~."X


t.4.;,--:~.o- \

Yellowish-brownsiltyloam/<f":%'.'. ~ . ::~~)"~..'..~.

I/ "" C:~." " . : . . : : . . " . . .' "." "{~:.. : : . . " . sand "".~. :~P'~:".~o P'~." *"o" ' :
Lo.:. ''. :.~.":. ~.O:-".":.~._". :~''.".1
I. ". "~'~. "..~'~. "." :" : .'.":" .':. "." :.-.'."'. "."~ ":~.~.<." :. ":" : ~ Sand
~.. ' . .. - . . . . . . ~/.,.: . . . .. . .. . ...o. ...~. ~ . ...~
. . . /. ~. :.. :.1
. ~ ~ - - - ..... and
L .~. :..' . - j . ~ . ....~.~...-.. ... : . . : . . . . " . . L - ; . . ~ gravel
I .- "" ,~. - ,. "' . ." " ~ . ~ ' " O ' .' . .c-~ ..I
o o " o - . " ' * ' ~ ~ " '
~. :: e~" . ' -- ,' .", . ~ - : '. -C") .~. . '. ."c~~ . . . . . : . . , . . t:'D.." ~. .. . .' . ". ...o. . . o '. ...:. .- . ' ~ .0' ~ . ' .. -1, .

t 1 meter t

Fig. 25. Schematic vertical section showing general characteristics of buried sand- and gravel-filled dikes and their vented sediments
along the Wabash River. Source beds are Holocene point-bar deposits or late Wisconsinan braid-bar deposits overlain by m u c h finer
overbank sediment. Sediment in source beds beneath dikes shows evidence of flow into dikes. Gravel content and size decrease
upward in dikes at m a n y places. Grey silty clay of overbank origin overlies sand and gravel that was vented onto the ground surface.
The column on the right side of the figure contains pedological descriptions of host materials.
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 41

for the following reasons, considered in com- induced by the 1949 earthquake of Olympia,
bination: Washington ( M = 7.1) (Obermeier, 1995). The field
(1) the dikes widen downward or have walls that setting is a thick (6-8 m) cap of stiff clay without
are parallel (agreeing with a lateral spreading pronounced horizontal planes of weakness that
origin); overlies medium sand. The dikes along the
(2) the dikes are approximately linear in plan view Chehalis River have a tabular morphology that
and exhibit strong parallel alignment in local strongly resembles that of many dikes in the
areas; Wabash Valley, because very thin (1 cm) dikes
(3) the dikes vented large quantities of sandy commonly extend upward several meters or more
sediment to the surface; before pinching together.
(4) material in the dikes fines upward and was
transported upward; 4.3.4. Paleoseismic implications
(5) bedding in some source beds is homogeneous, Historical earthquakes in the Wabash Valley
and the contact of source beds with overlying have had magnitudes as high as ~5.5 and have
fine-grained sediment is highly disturbed in not caused liquefaction. The paleo-earthquakes in
some places; the Wabash Valley far exceeded the magnitude of
(6) flow structures project upward from the source any historic events because of the large areal
zones into the bottoms of the dikes; distribution of liquefaction effects and the large
(7) many dikes are in flat and topographically size of some of the dikes. Magnitude of the event
elevated landforms, located at least several 6100 years ago is estimated to be ~ 7.5 (Obermeier
kilometers from any high, steep slopes that et al., 1993).
might have existed at the time of formation
of the dikes, and therefore could not have 4.4. Coastal Washington State
been induced by nonseismic landsliding;
(8) other nonseismic mechanisms such as artesian Tidal marshes buried in coastal Washington and
springs that could produce similar features are nearby coastal Oregon record episodes of sudden
not plausible at many dike sites because of the downdropping of land accompanied by tsunamis
lack of topographic relief and the local geolo- during late Holocene time; these episodes have
gical setting; been ascribed to great earthquakes ( M ~ 8-9) on
(9) the size and abundance of the dikes along the the basis of the large region along the coast that
Wabash River, the area where data are most appears to have dropped all at once (Atwater,
complete, generally decrease with increasing 1987,1992; Darienzo and Peterson, 1990; Nelson,
distance from a core region of largest dikes 1992). No direct evidence of strong shaking had
(Fig. 24); been discovered to corroborate that seismic shak-
(10) large regions that have liquefiable sediment ing accompanied the episodes of dropping before
in the same geological setting have been the work described below (Obermeier, 1995).
searched far north of the Wabash Valley These inferred earthquakes are interpreted to have
seismic zone and have no dikes (Munson originated by rupture along the thrust fault where
et al., 1994). an oceanic (Juan de Fuca) plate is being subducted
beneath the continental (North America) plate -
Detailed discussion of these factors is in the Cascadia subduction zone (Fig. 26). However,
Obermeier et al. (1993). no strong Cascadia earthquake has occurred
Sills within the cap such as those illustrated in during the time of written history in the Pacific
Fig. 22 are sparse in the Wabash Valley region. I Northwest, some 200 years. Modem seismicity
have observed elsewhere a similar lack of sill near the subduction zone is limited to scattered,
development in the cap. For example, in the banks small earthquakes, none with thrust mechanisms
of the Chehalis River, near Centralia, Washington, in the region of sudden submergence.
there are many dikes exposed that doubtless were Atwater (1992) inferred at least two occurrences
42 S. F Obermeier/Engineering Geology 44 (1996) 1 76

124

50

0
48

46
C~~gvie~.

44
- ,....... i
) lS . . . . idedikes -- dS;~emsWide
Bonneville
I Dam _

Portiad ~ ~ ~ 42

100 KILOMETERS 0 100 KILOMETERS


I I _ l__ I

Fig. 26. The part of the Columbia River where banks of islands were searched for paleoliquefaction features. These searched islands
have ages between 600 and 1000 years at most places. Sands beneath islands are fine to medium grained and generally are at least
moderately susceptible to liquefaction. M a x i m u m dike width is measured at least 1 m above the base of the dike; it generally decreases
inland in the islands as shown. The density of small and medium-sized dikes in the islands also generally decreases inland. Index
m a p shows seaward edge of Cascadia subduction zone, which dives beneath the continent.

of coseismic subsidence when great earthquakes I thought that the inferred earthquakes probably
struck the coast of Washington, including the should have caused strong shaking and abundant
region around the Columbia River valley, during liquefaction features near the coast, even in sedi-
the past 2000 years. Strong evidence indicates that ments having moderate to low susceptibility. To
one event was about 300 years ago, and less verify occurrence of the earthquakes, I initiated a
widespread evidence indicates that another event search in 1992 for liquefaction features in cutbanks
occurred between 1400 and 1900 years ago. The of islands in the Columbia River.
portion of the thrust fault that ruptured and
provided energy for seismic shaking was most 4.4.1. Columbia River features
likely a small distance offshore (a few tens of Many large islands were searched between the
kilometers), on the basis of the location of the towns of Astoria and Portland (Fig. 26). These
subsided zone (Atwater, 1987), heat-flow data islands originated as braid bars on a grand scale.
(Hyndman and Wang, 1993), and strain data The islands are flat, poorly drained, and swampy.
(Savage and Lisowski, 1991). Large portions are submerged during the highest
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 43

tides. Strong currents and wave pounding are fossil marsh plants found in growth position, which
severely eroding many islands and have sculpted now are located just above the level of low tide.
clean, vertical banks as high as 2 m, which extend Regional stratigraphic control of sediments
from water level to the top of the banks. Significant exposed on the islands is excellent. About 1.5 m
areas are also being cleaned in plan view by tides. below the top of the banks is a tan horizon with
The banks of the islands between Astoria and a thickness of a few centimeters. This horizon is
Longview expose mainly soft, clay-rich silt deposits exceptionally rich in volcanic ash. About 10-15 cm
(Fig. 27). Age at the base of the exposed clay-rich lower is a blue-gray horizon, generally several
cap is less than 1000 and more than 600 years on centimeters thick, also rich in ash. Very locally,
most islands, on the basis of radiocarbon ages of rounded pumice clasts as large as 5 cm in diameter

fO ym old

d~ol~KI
moll

~,'Im ~Ind
~tr rich

~r rich
pumice
,,, 1 - 2 m
Cop

._~ eiUdn

Fig. 27. Typical field relations at liquefaction sites in Columbia River islands. A sand-filled dike cuts through a 1-2-m thickness of
soft silt and clay with a weakly developed paleosol at the top. The dike connects to a thin sheet of vented sand that lies on the
paleosol; the sand sheet is buried by a 1-m thickness of silt and clay. Tubers at widespread sites collected in their growth position
near the base of the stratum cut by dikes have radiocarbon ages ranging between 600 and 1000 years. Widths of tops of dikes are
only several millimeters at most places.
44 S.F Oberrneier/Engineering Geology 44 (1996) 1 76

occur in the lower ash horizon. Chemical analysis All dikes are interpreted to have been caused by
shows that the ash and pumice have minerals and the coastal downdropping earthquake about 300
elements identical with those of an eruption from years ago for the following reasons: (1) the radio-
Mount St. Helens in A.D. 1480-1482 (C.D. carbon ages of sticks along the surface of venting
Peterson, Portland State Univ., written commun., agree with an earthquake 300 years ago; (2) trees
1992). Therefore, the radiocarbon ages on fossil rooted in sediment above vented sand have maxi-
marsh plants and the ash data show that the mum ages (about 230 years; determined from tree
sediments are old enough to record liquefaction rings) that are younger than 300 years; (3) dikes
associated with the 300-year-old downdropping generally increase in abundance toward the coast:
event, but they probably were deposited after the and (4) maximum dike widths increase toward the
event of 1400-1900 years ago. coast. The 1-m thickness of silt and clay above
At many places, sand is exposed immediately vented sand is interpreted to have been deposited
beneath the clay cap. Alluvium that makes up the following the regional downdropping from the
islands, generally thick, fine to medium sand, prob- subduction-zone earthquake 300 years ago. This
ably typically exceeds 100 m in thickness. 1-m thickness agrees well with the estimates of
Conditions on many islands are nearly ideal for coastal tectonic submergence (Atwater, 1987, 1992:
formation of large liquefaction-induced features. Darienzo and Peterson, 1990).
Not only is the cap thin, but the ground-water The 1-4-cm-thick sheet of vented sand may be
table has almost certainly been within a meter or exceptionally thin because of the tidewater action
so of the ground surface since the islands formed. or because of subaqueous venting. The surface of
The tidal range at these islands is about 2 2.5 m, venting is submerged at high tide. Tidewater flows
and high tides inundate parts of the islands and relatively fast in this area, so any large cones of
have done so for at least several hundred years. sand initially vented to the surface could have been
Hundreds of dikes have been found along 9 km beveled off and the sand scattered over a large area.
of vertical banks in scattered islands upstream as Many dikes are so narrow at the top as to be
far as Deer Island (Fig. 26). At some places, the hardly distinguishable. This same relation can also
tabular nature of the dikes is exposed in plan view. be observed in a few dikes in the Wabash Valley
Maximum dike widths and abundance of dikes and in dikes in the 1811-1812 New Madrid earth-
tend to decrease in the upstream direction. Fig. 27 quakes' meizoseismal region. The cause for this
illustrates relations observed at many islands. A pinching is not known but may be venting of a
thin sheet of vented sand lies on a weakly devel- very water rich mixture of sand and water through
oped, very soft soil that is about 1 m below the a very soft cap that closed partially after venting.
present surface. Locally, the upper few centimeters The widest dikes (as much as 30cm) on the
of the soil is contorted by small (centimeter-sized) Columbia River islands are interpreted to have
folds and other soft-sediment deformations. The formed by lateral spreading because the sidewalls
sand sheet is 1-4 cm thick and is as wide as 10 m. appear to be parallel.
The sheet connects to a nearly vertical, narrow, Others have suggested that the properties of the
tabular dike that widens downward markedly and soft silt cap on the islands may have prevented
connects to sand beneath the clayey cap. Width of dikes from forming far inland; they suggested,
the uppermost 5-15 cm of almost all dikes is only instead, that only sills formed as a result of severe
several millimeters or less. Width near the base of liquefaction over large regions. The basis for this
the cap is generally less than a few centimeters. suggestion is the presence of numerous thin sills
Where pits were dug along the bottom of the cap, and sill-like features along the base of the cap,
flow structures in the sand could be observed going which have been observed in samples collected in
into the base of the dikes. Pits also exposed sills 8-cm-diameter tubes. I believe that such an inter-
as thick as 0.1 m running along the base of the pretation is unlikely for two reasons: (1) I have
cap. Rarely were sills observed to have intruded never observed in other geographic-tectonic
into the cap. regions where extensive sill formation was not
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 45

accompanied by dikes, and (2) the mechanics of 1811-1812 New Madrid earthquakes or in the
forming sills by severe liquefaction over a large Wabash Valley.) Another plausible model consis-
region without also forming dikes does not seem tent with observed liquefaction effects in the
plausible. In order to form sills over a large area, Columbia River islands is a subduction earthquake
the cap must be essentially floating to provide having a long duration of shaking at moderate to
space for the intrusion. The force to float the cap low peak accelerations and at an exceptionally low
must equal the weight of the cap. The simplest of dominant vibration frequency (perhaps exceeding
calculations shows that the uplift hydraulic pres- a few hertz). The fine sands that underlie the
sure to counteract the weight of a 1-2-m cap is Columbia River islands are so thick and have such
much higher than the tensile strength of a soft a low permeability that pore pressures would not
cap. Therefore, dikes should develop. (If the cap dissipate between cycles of shaking, even at an
had been extraordinarily soft at the time of the extremely low frequency. In addition, the thick
earthquake, plastic deformation might have devel- alluvial deposits of the Columbia River valley
oped along the base of the cap, but none was probably strongly amplify bedrock accelerations,
observed.) I can envision other mechanisms irrespective of vibration frequency (Dickenson
whereby small sills and other sedimentary struc- et al., 1994). Therefore, the sand deposits of the
tures could form without liquefaction, however, as Columbia River should be especially susceptible
a result of seismic shaking. to forming liquefaction features during a long
duration of shaking, even at low to moderate
4.4.2. Strength of prehistoric shaking accelerations and exceptionally low frequencies.
Subduction earthquakes can have very large
variations in shaking characteristics, offering the 4.4.3. Ancient marine-terracefeatures
possibility of an especially long duration of shaking Many ancient fluidization features have been
at very low frequencies. Such uncertainties cause identified in late Pleistocene marine-terrace depos-
difficulty in interpretations of the strength of pre- its in cliffs along the coast for 500 km from central
historic shaking. Still, significant conclusions can Washington to near the California-Oregon bound-
be drawn for the earthquake of 300 years ago. ary (Peterson et al., 1991; Peterson, 1992). The
Small dikes possibly with venting in the Columbia features are of particular interest because they
River islands appear to go inland as far as 90 km. admit the possibility of a long continuing record
These dikes very likely formed at an acceleration of subduction-zone earthquake shaking near the
level on the order of 0.1-0.2 x g (Obermeier, 1995). coast. The method for interpretation of an earth-
These accelerations accord with both theoretically quake liquefaction origin is instructive.
and statistically derived accelerations from seismo- Source beds for the fluidization features include
logical models for a subduction earthquake slightly beach sands and sandy gravels and lagoonal sands.
offshore that had a magnitude of ~8-8.5 Clastic dikes are as much as 5 m in height. Dikes
(Geomatrix Consultants, 1995), although the are filled with clean sand or gravelly sand at almost
uncertainties in the models are relatively large. all places. Dikes are as wide as a meter in scattered
The lack of abundant very wide dikes through- locales. Some dikes have penetrated upward into
out islands of the lower Columbia River valley dune sands or have cut through lagoonal muds
also supports the seismological models noted and peat. Sills are particularly abundant. Sills
above, which predict that exceptionally strong commonly extend beneath lagoonal muds and
shaking should not extend very far onshore. Even peats; small, steeply dipping dikes branch off from
though severe erosion of some islands has probably these sills at many places and cut up into thin
removed evidence of dikes wider than 30 cm at (< 0.5 m) strata of low permeability at the surface.
some places, there are many locales where erosion The largest sills are as much as a meter in thickness.
probably has been slight. (The widest observed Even thin sills can extend far laterally.
dikes, 30 cm, are much narrower than dikes com- The possibility that the fluidization features were
monly found in the meizoseismal region of the caused by wave action that induced liquefaction
46 S. F Obermeier/Engineering Geology 44 (1996) 1- 76

must be considered because the terrace deposits geotechnical engineering perspective is often help-
were laid down under shallow-marine or shoreline ful for interpreting origin.
conditions. Storm waves can impose significant
shear stresses on the ocean-bottom sediments, even 5.1.1. Nonseismic sand boils'
where the water depth exceeds 60-70 m (E.C. Features suspected to have a seismic origin
Clukey, geotechnical engineer, Exxon Corp., writ- should be examined first with the perspective of
ten commun., 1992). Wave-induced cyclic shear determining whether artesian conditions were plau-
stresses are thought to cause liquefaction in sands sible at the site. This is the basic approach that
and granular deposits in a manner analogous to was taken for evaluation of the craters in coastal
seismically induced liquefaction (Nataraja and South Carolina, where it was shown that non-
Gill, 1983). The action of storm waves pounding seismic sand boils were implausible because the
on beaches also seems plausible as a mechanism craters had formed along the crests of ridges.
for forming fluidization features. For the fluidiza- Nonseismic sand boils that abound in parts of
tion features in the marine terraces of coastal the meizoseismal region of the 1811-1812 New
Oregon and Washington, though, the mechanism Madrid earthquakes are especially instructive.
of wave-related liquefaction probably can be elimi- Flooding causes strong artesian flow beneath man-
nated at some sites because dikes extend up into made levees that parallel the Mississippi River.
dune sands where wave action seems unlikely. These artesian conditions cause abundant sand
Additionally, some dikes and sills cut lagoonal boils to form in nearby lowlands every few years
deposits at places that probably would have been (Kolb, 1976). Most appear to have formed under
protected from wave action. Significant artesian subaerial conditions. The sand boils are generally
pressures at these lagoonal sites also are implausi- restricted to within 0.5-1 km of the levees.
ble. Thus, a seismic liquefaction origin seems prob- (Natural levees should not cause artesian condi-
able for some of the features along the coast. tions because of the many breaks in the levees.)
Detailed characteristics of the sand boils are
described by Li et al. (1996).
The cone-shaped external form of the nonseismic
5. Features generally of nonseismic or unknown sand boils along the levees is similar to that of
origin solitary sand blows of earthquake-induced lique-
faction. Internally though, differences commonly
I have surveyed the voluminous literature of occur because artesian flow normally starts slowly
soft-sediment deformation, making special note of and then steadily increases, in contrast to the much
common features that might be confused with more sudden, violent flow that typically accompa-
those resulting from seismic shaking, Assignment nies early phases of venting due to seismic liquefac-
of unambiguous origin to isolated features may be tion, especially where large sand blows have been
impossible unless the regional pattern of sizes and vented. Evidence for this large difference in flow
abundance is determined. Nonseismic features that rates can be in the sedimentary record.
form in a wide variety of environments are consid- The basal portion of sediment vented by non-
ered in this section. The discussion includes both seismic sand boils along the Mississippi River
chemical processes related to weathering and phys- levees typically is significantly finer than that of
ical processes as diverse as slumping, freezing the large seismic sand blows. The sand boils gen-
behavior, and flowing water. erally have a discrete basal zone of silt to silty
very fine sand that ranges in thickness from a few
5.1. Terrestrial disturbance features millimeters to a few centimeters. As shown in
Fig. 28, going upward in sand boils are rhythmic,
A wide variety of ground-disturbance features thin, planar strata of silty sand to clean sand that
similar to those caused by seismic liquefaction dip away from the central part of the sand boil.
form above the water table. A hydrological or a The individual strata are much more sharply
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 47

Fig. 28. Schematic vertical section showing typical nonseismic sand boil that formed under subaerial conditions (no vertical exaggera-
tion). The sand boil formed adjacent to a high manmade levee along the Mississippi River during a period of severe flooding. Vented
sediment has been laid down on the dark-colored ground surface. View shows bedding details characteristic of venting from artesian
conditions. Note the well-defined rhythmic planar bedding. Darker bands are slightly finer grained (silty sand) than lighter bands
(clean sand). Dip of dark bands typically exceeds 30. Feeder dike (vent) has a form that is irregularly tubular. Very steeply dipping
bedding planes are in upper part of the conduit. Source sand lies at very shallow depth. Alternating layers of fine (dark) and coarse
(light) vented sediment indicate a variable discharge rate through the feeder dike. Sketch courtesy of Yong Li.

defined, are more uniform in thickness (typically of artesian flow beneath levees along the
a few millimeters to 1 cm), and have a narrower Mississippi River, which forms sand boils that
range of grain sizes than sediment vented in typically cut through the thinnest part (generally
response to liquefaction caused by the 1811-1812 < 1 m) of the clay-rich cap, much as sand blows
New Madrid earthquakes. Probably the individual do. In contrast to sand blows, though, sand boils
layers in sand boils represent slight variations in have approximately circular vents. Their diameters
the rate of venting. generally range from a few centimeters to greater
The dip angle of individual layers also typically than 1 m. I emphasize that in contrast to the
much exceeds that found in a seismically induced circular dikes for artesian sand boils, earthquake-
sand blow (see Fig. 19). induced liquefaction beneath a clay-rich cap nor-
Not always, though, are there readily diagnostic mally develops tabular fissures in the cap.
differences in bedding characteristics between non- Therefore, tabular sand-filled fissures that widen
seismic sand boils and seismic sand blows. Bedding downward and connect to a sand unit that extends
planes within a sand blow can be about as sharply horizontally for tens to hundreds of meters are
defined as in a nonseismic sand boil, especially for probably dikes associable with earthquake-induced
a small sand blow. (For example, see the photo- liquefaction, although the possibility of a non-
graph on the cover of the April 7, 1989, vol. 244, seismic landsliding origin (discussed below) must
issue of Science.) Whether the dip angle of bedding be considered.
planes is diagnostic of origin is not yet known. Where the cap in the meizoseismal region of the
Thus bedding characteristics alone may be inade- 1811-1812 New Madrid earthquakes is very sand
quate for distinguishing between sand volcanoes rich and friable, both sand blows and nonseismic
formed by artesian flow and those resulting from sand boils typically form approximately circular
seismic liquefaction. craters at the surface. Craters from both can be as
Fortunately though, characteristics other than much as a meter in diameter. I am unaware of
bedding generally suffice for determining origin. definitive sediment relations to always ascertain a
These characteristics are well illustrated by effects seismic or nonseismic origin for individual craters.
48 XF. Obermeier/Engineering Geology 44 (1996) 1 76

Therefore, the regional pattern of occurrence is where the erosion surface on the shale dips toward
critical for interpretations. Such a pattern becomes a deep river. Good examples of this type of failure
more evident as earthquake magnitude increases. are located along the Ohio River, where shales
have internal friction angles as low as 7-8 (Mesri
5.1.2. Streambank landslides and Gibala, 1971). In contrast to the process of
Slumps (of largely rotational movement) are earthquake-induced liquefaction, the large blocks
commonplace along streambanks for both seismic seem to move slowly, and there is no evidence of
and nonseismic conditions. Sliding blocks (of sudden or forceful venting of sediment (Obermeier
dominantly horizontal movement) can develop in et al., 1993).
special nonseismic field settings and can resemble Nonseismic sliding of blocks can form steeply
seismically induced lateral spreads. I judge that dipping fissures through a clay-rich cap. These
generally it is not worthwhile to attempt to assign breaks conceivably could be filled with cohesion-
seismic or nonseismic origin to prehistoric slumps less sediment that flows into the fissure from
that may have occurred along actively eroding beneath the cap, which could resemble dikes
streambanks; there are too many uncertainties formed by earthquake-induced liquefaction. A test
about the physical setting when such an ancient for seismic origin is provided, though, if dikes can
slump formed. For example, questions crucial to be found on alluvial plains far from any significant
interpretation include the initial ground slope and slopes. Another criterion for origin is to place
the possibility of a streambank having been limits on the maximum artesian head; if this head
undercut by an eroding stream. These questions is sufficiently low, a seismic origin is likely
may be virtually impossible to resolve. Slumps in (Obermeier et al., 1993).
loose, fine sands are especially difficult to interpret One more test for landslides of suspected seismic
because they may be prone to static liquefaction origin is the regional pattern of sizes and ages
even on relatively gentle slopes (Lade, 1992). (Jibson and Keefer, 1988,1989). Landslides far
Very thin sand layers bounded above and below from active streambanks, in more stable settings,
by low-permeability materials can slump whenever also can be analyzed using geotechnical engineer-
river levels drop rapidly (Springer et al., 1985). ing procedures to interpret the likelihood of seismic
This same sediment may move as a lateral spread triggering (Jibson, this volume).
during earthquake shaking. For example, large
blocks of clay bedded with thin, laterally con- 5.1.3. Ground disturbance by trees
tinuous sand lenses ruptured as far back as 200 m Many Quaternary beach sediments of the
from a free face during the 1964 Alaska earthquake Southeastern United States contain evidence of
(M=9.2); Seed (1968) interpreted the failures to ground disturbance by trees. Tree roots commonly
have initiated in the thin sand lenses. penetrate the sand-rich cap to a depth of 2-3 m.
Recent studies show instead that failure at some When the roots decay, large holes develop that
of the Alaskan sites initiated during the earthquake later fill with sand. Many now display poorly
through sensitive ("quick") clay layers in the defined layers of clay-mineral segregation that
blocks (Updike et al., 1988). Large slablike blocks result from weathering; however, the fill sediment
of sensitive clay also commonly fail under static has neither the well-defined laminar bedding nor
conditions, however (Karlsrud et al., 1984). the underlying graded zone of sediment that char-
Proving a paleoseismic origin for ancient block acterizes filled seismic craters.
failures along stream banks in sensitive clays or in The beach deposits also contain fairly wide-
clays with thin sand beds has never been done to spread evidence of pits that were excavated by
my knowledge, and probably would be a formida- root mats of trees that were thrown down (perhaps
ble task. by hurricanes). Sediment in these pits typically
Large blocks of alluvium that slip nonseismically does not have the orderly progression of clast sizes
along buried erosional surfaces on clay-rich shales present in the lower part of the liquefaction-
are not unusual. This type of failure is enhanced induced craters (Fig. 14). Some pits excavated by
S.F Obermeier/Engineering Geology 44 (1996) 1-76 49

thrown trees can be distinguished from liquefac- deposits, with Fig. 29, showing mima mounds,
tion-induced craters by the absence of feeder dikes illustrates the contrast.
or by the absence of sediment from depth in the
pit. Verification that sediment has been introduced 5.2. Features formed in subaqueous environments:
from depth may require analysis of mineralogical,
weathering, and grain-size data (Gelinas, 1986). It is not possible in this article to describe and
Roots from thrown trees can also form tabular discuss the great number of features, mainly synde-
breaks in a fine-grained cap. Locally the breaks positional in origin, that can also be triggered by
can be filled with sand, gravel, and other sediment seismic shaking. Only some of the more common
dragged into the break as the root is pulled from types, mainly formed in fluvial, lacustrine, and
the ground. However, the dragged sediment tends nearshore marine situations, are discussed herein.
to be arranged much more haphazardly and have For a more complete overview of the myriad
a much larger range of grain sizes than an intrusion possibilities, the reader is referred to Lowe and
formed by seismic liquefaction. LoPiccolo (1974), Lowe (1975), Sims (1978),
Reineck and Singh (1980), Allen (1982), Mills
5.1.4. Mima mounds (1983), Jones and Preston (1987), van Loon and
Mima mounds (sometimes called prairie or Brodzikowski (1987), Einsele et al. (1991), and
pimple mounds) have been interpreted as having van Loon (1992).
a seismic origin in many geographic areas (Berg, Four types of features are considered below: (1)
1990); however, seismicity has never been observed load structures in muds; (2) water-escape struc-
to produce features closely resembling mima tures in granular sediments; (3) sheet slumps,
mounds (Saucier, 1991a). Mima mounds occur warped beds, and recumbent folds; and (4) turbid-
throughout a large part of the meizoseismal region ites. It is difficult in most field situations to assem-
of the 1811-1812 New Madrid earthquakes. Here ble compelling evidence for a seismic versus
the mounds are domes of silt- or sand-rich materi- nonseismic origin for these features. However,
als that are generally less than 30 m in diameter deformation features in some lake sediments can
and 1 m high. Exceptional domes are as much as be attributed to: seismic events, and examples
1.7 m high but have diameters of only 10-12 m. follow descriptions of the four types of features.
The mounds have essentially the same texture as
the underlying sediment. Soil development in the 5.2.1. Load structures in muds
mounds is strong, indicating an age of thousands Any plastic deformation of mud almost always
of years. The origin(s) of the mima mounds in the occurs soon after deposition. The deformation
vicinity of the 1811-1812 earthquakes is (are) not generally is triggered because the rapid deposition
known. In many upland areas, though, the mima of sand above very soft, clay-rich sediments causes
mounds are formed on nonliquefiable deposits and buildup of pore-water pressure and gravitational
therefore are not of earthquake origin. Mounds instability (Dzuiynski and Walton, 1965; Allen,
on alluvial lowlands can be identified as not result- 1982; Brodzikowski et al., 1987). Plastic deform-
ing from earthquake liquefaction if excavation ations seem to be most abundant where rivers
shows an absence of vents to connect the mounds carry much sand, silt, and clay; settings where
to underlying source beds (Fuller, 1912, p. 80). rivers debouch into lakes are also favorable.
Mima mounds can usually be distinguished from Montmorillonite-rich clays seem to enhance the
sand blows on aerial photographs. Mounds typi- deformations because of the exceptional ability of
cally show regularity of spacing, alignment, and montmorillonite to absorb water and become
size (Fig. 29). In contrast, sand blows are typically extremely soft.
more irregularly spaced, have widely differing sizes Especially plentiful in the geological record are
(heights and diameters), and are not aligned along soft-sediment deformational features known as
such precisely defined curves. Comparison of "load-casted" or simply "load" structures. Load
Fig. 18, showing sand blows formed on point-bar structures are bulbous bodies of sandy or silty
50 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

o 1 MILE
I J I
o 1 KILOMETER

Fig. 29. Aerial photograph (1940) of mima mounds (white circular spots) near Sikeston, Scott County, MO, in the northern part of
the meizoseismal area of 1811-1812 New Madrid earthquakes. Note regularity of spacing and alignment of mounds.
S.F.. Obermeier/Engineering Geology 44 (1996) 1-76 51

material that intrude downward into underlying


weaker, finer grained muddy sediment. The sever-
ity of deformation is controlled by two factors: ,4
the difference in densities between the two adjacent
layers and the weakness of the underlying layer.
Sand deposited rapidly over water-saturated,
extremely soft mud is ideal for formation of load
features. Two types of load structures are preva-
lent: pseudonodules and load-casted ripples. B
Pseudonodules form when overlying sandy or
silty sediment becomes detached and sinks to
become isolated kidney-shaped bodies encased in
the underlying mud. Pseudonodules generally
occur laterally adjacent to other undetached
load structures (Allen, 1982, pp. 359-360). C
Pseudonodules have been produced experimentally
in the laboratory by hammering a container in
which sand overlay water-saturated, very soft clay
(Kuenen, 1958; see Fig. 30 herein). Sims (1975)
correlated pseudonodules found in modern lake
sediments, similar to those produced by Kuenen,
with known earthquake events.
Load-casted ripples are sandy or silty intrusions
that form because of the unequal loading of D
migrating ripples of sand on a mud substratum.
Load-casted ripples show progressively deformed
radial internal lamination caused by the rotation
of the ripple cross laminations as the ripples sink
(Dzulynski and Walton, 1965, pp. 146-149). The
asymmetry reflects current direction, as illustrated
in Fig. 31. Although load-casted ripples can result
from earthquake shaking, they generally develop
synchronously with deposition of overlying sand
and do not require earthquake shaking. E
0 5 INCHES
Load structures produced by rapid sedimenta- j ~ t
tion are illustrated in a ditch exposure in the 0 10 CENTIMETERS
meizoseismal region of the 1811-1812 New Madrid
earthquakes. The section in Fig. 32 shows clays Fig. 30. Pseudonodules formed by shaking (from Kuenen,
1958). A layer of sand (unit 1) overlies very soft clay (unit 2)
interbedded with layers of sand that are apparently in (A). With shaking, pseudonodules developed ((B) and (C))
related to intermittent deposition in a water-filled and became completely enclosed in clay ((D) and (E)). Note
swale. Small crevasse deltas formed convex-up the destruction of layering from (C) to (D). This deformation
sand lenses containing climbing ripple cross lamin- is due to sinking of the entire sand layer into the very soft clay
ations. The lenses are coarser upward, fine toward and also to localized sinking caused by uneven loading.
the edges, and are laterally adjacent to layers of
pseudonodules (25-50 cm long and 5-25 cm wide) percentage of each sand lens as it thins toward the
made up of similar grain sizes. The base of each edge (see north-south view in Fig. 32). Detailed
sand lens has load structures, including load-casted discussion of these sedimentary structures is in
ripples. The load structures make up a greater Obermeier et al. (1990, pp. 35-38).
52 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

mation. Although load structures such as these


may be formed by earthquakes, their vertical repe-
tition in this sequence (as shown also by Allen,
1982, Figs. 9, and 10; or Coleman and Prior, 1980,
Figs. 15, 21, 22, and 32) within what was obviously
a short time span argues for a depositional origin.

5.2.2. Water-escape structures in granular


sediments
The water-escape features discussed below, listed
in order of generally increasing size, include dish
structures, pillars, and convolute bedding
(Pettijohn and Potter, 1964; Lowe and LoPiccolo,
1974). Also included in this class are ball-and-
pillow structures. These water-escape features can
develop either shortly after or long after initial
deposition, in response to a generally upward flow
of water through granular deposits (Lowe and
LoPiccolo, 1974), although most form very shortly
after deposition. Laminar flow by seepage may be
adequate to form small features such as dishes,
but turbulent flow by fluidization is required for
many others (Lowe and LoPiccolo, 1974). All
these features can form in seismic or nonseismic
conditions.
In vertical section, dish structures appear as thin,
dark-colored, subhorizontal, flat to concave-up
laminations (Fig. 33). Individual dishes range
from clean sharp lines as thin as a featheredge to
Fig. 31. Developmentof load-casted ripples, caused by ripple diffuse zones up to 2 mm thick. Widths can be as
crests sinking into soft mud. Note the progressive tilt of the
internal cross-lamination, which causes the downflow portion much as 50cm (Lowe and LoPiccolo, 1974).
to be more steeplyinclined than the upflow portion (modified Dishes appear to form as upward moving water
from Dzulynskiand Kotlarczyk, 1962). Load-castedripplescan locally flows beneath laminations of lower perme-
also have an opposite sense of rotation (upflow more steeply ability and escapes by horizontal flow. The finer
inclined than downflow). Length of the ripples is commonly grains, such as clay, mica, and organic grains, are
2 5cm.
filtered out and become trapped within the lower
permeability zone.
The load structures in Fig. 32 are interpreted as Pillars are circular columns or sheetlike zones
synsedimentary in origin and not as earthquake of structureless or swirled sand, in places bounded
induced on the basis of the following criteria: (1) by dark laminations, that cut steeply through sand
both the pseudonodules and load-casted ripples ranging from structureless to laminated (Fig. 34).
are gradational into the sand lenses, with the sense These features range from tiny vertical tubes less
of development related to flow direction; (2) sev- than a millimeter in diameter to large flame struc-
eral layers of pseudonodules are laterally equiva- tures more than a meter across and several meters
lent to largely undeformed sand lenses; and (3) in height (Lowe and LoPiccolo, 1974). These
the recurring conditions of sedimentation (rapid features form from a localized fluidization of the
progradational deposition of sand over soft clay) sediment.
were conducive to this type of soft-sediment defor- Convolute bedding is the name given to a later-
S.F. Obermeier/EngineeringGeology44 (1996) 1-76 53

w~
~c~th

-In

,_/,S I

EXPLANATION
Medium sand

U Very fine 1o fine sand

" ] Coarse silt to very fine sand

m Laminated silt with minor very fine sand

m Clay with Ittir,~ts and laminae of silt

Fig. 32. Load structures produced by rapid sedimentation. Block diagram shows spatial relations inferred for crevassedeltas extending
into a shallow, water-filled swale. The deltas form convex-up lenses having ripple cross laminations (white, with diagonal lines). The
edges of the delta are rippled sands that completely foundered into the soft clay (gray), forming pseudonodules (detached white
bodies with diagonal lines), whereas the main body exhibits local load-casted ripples and a broad sagging. Arrows indicate direction
of paleocurrents. The pseudonodules and load-casted ripples are load structures resulting from rapid sedimentation. Diagram from
Obermeier et al. (1990).

ally extensive series of more-or-less regular folds and LoPiccolo, 1974). The origins of convolute
(Fig. 35), which are either developed throughout bedding are unclear but probably relate to a variety
a single sedimentary unit or confined to its upper of mechanisms (Lowe, 1975). Lowe noted that
part (Allen, 1982). As Allen described, folds convoluted sediment is typically somewhat finer
increase in wavelength with increasing thickness grained and more clay rich than subjacent sand;
of the bed or deformed zone. Strong asymmetry formation o f convolutions in sand beds may be
and recumbent folding are rare. Convolute bedding caused more by instability of the subjacent sand
can develop at the top of sand units as thin as than by fluidization o f the convolute-bedding zone.
15 cm (Lowe, 1975) and thicker than 5 m (Lowe Fig. 35 shows field relations commonly observed
54 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

Fig. 34. Pillar structures in fine-grained sandstone. Pillars are


light-colored washed sand between the upturned margins of
strongly concave dishes. Photograph from Lowe and
LoPiccolo (1974).

Fig. 33. Strong development of dish structures in fine-grained Mudstone


sandstone Dish structures are the very thin laminations. Two :.'.~.~.'." "..'- "..".'.: . . ~ . i..;
generations of dish structures are present Older dishes ( 1) are
strongly concave, are discontinuous, and have commonly been
- ...):'... Convolute bedding
blurred by the upward flow of fluidized sediment. Dishes formed
later (2) are flatter, more continuous, and separated vertically
by sand showing a well-defined substructure - the dark dish
lamination, an underlying zone of white clay-poor sand, and Pillars
an underlying zone of gray argillaceous sand. Photograph from
Lowe and LoPiccolo (1974). I have observed such strong devel-
"~'---~" " -<..H_." " ' ""
opment of dish structures in sediments exposed in banks of the
Edisto River (Fig. 11) near the meizoseismal zone of the 1886 -..~--~..'."..:~ .. ,.SL,..--,a--'.. : .'.
earthquake at Charleston. Dishes

a m o n g d i s h structures, pillars, a n d c o n v o l u t e b e d -
f ~ P
ding, I h a v e o b s e r v e d all these features a n d rela- ~ f L-,- Flat laminations
t i o n s in p r o x i m i t y to dikes at sites o f e a r t h q u a k e -
induced liquefaction.
Structureless sand
I n b a l l - a n d - p i l l o w structures, k i d n e y - s h a p e d
b o d i e s o f fine o r silty s a n d ( p s e u d o n o d u l e s ) h a v e
foundered into a cleaner sand. A photograph of
these s t r u c t u r e s was p r o v i d e d b y P e t t i j o h n a n d Fig. 35. Commonly observed vertical sequence of the sedi-
P o t t e r (1964, plate 100A). B a l l - a n d - p i l l o w struc- mentary structures of convolute bedding, pillars, and dishes.
Example shown (from Lowe and LoPiccolo, 1974) is in thick
tures are a b u n d a n t in m a n y glaciofluvial deltas sandstone beds. I have observed this sequence of structures in
a n d h a v e l e n g t h s r a n g i n g f r o m c e n t i m e t e r s to tens sediments exposed in banks of the Edisto River (Fig. 11) near
o f centimeters. C l e a r l y the features r e a d i l y f o r m the meizoseismal zone of the 1886 earthquake at Charleston.
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 55

very soon after deposition of the glaciofluvial host model sequence of deposits going upward is (1) a
sediments. Seismic shaking has also been docu- thick bed of structureless sand possibly with some
mented as a cause of ball-and-pillow features gravel that fines upward, (2) beds of sand with
(Ringrose, 1989). parallel and ripple laminations, and possibly beds
with convolute laminations, (3) thin beds of silt
5.2.3. Sheet slumps, warped beds, and recumbent and sand with parallel laminations, and (4) a bed
folds of mud. The many variations found in nature were
Features such as sheet slumps and warped beds discussed by Allen (1982).
(Allen, 1982) and recumbent folds (Allen and Widespread turbidites are very common in
Banks, 1972; Sims, 1973; Owen, 1987) are well marine sediments (see numerous articles in the
represented in the geological record. Sheet slumps journal special issue edited by Cita and Lucchi,
tend to form in thin, laterally extensive deposits 1984). It is often tacitly assumed that such a
of silt and sand interbedded with mud. The mode distribution strongly suggests a seismic origin for
of failure is mainly translational movement of a turbidites and warrants the name "seismites."
sheetlike body down a very gentle slope. Slight to Although a widespread synchronous distribution
severe warping and local thrusting may develop in is suggestive of seismicity, other mechanisms such
the failed sheet. Sheet slumps range in thickness as storm-induced wave trains or tsunamis can be
from a few tenths of a meter to greater than 100 responsible. Probably widespread distribution of
m (Allen, 1982). Lacustrine and marine deposits thick turbidites is indicative of a seismic origin,
normally host the slumps. Studies in modern envi- though. An excellent example of how turbidites
rouments show that huge sheet slumps can develop can be used in conjunction with other evidence for
on subaqueous slopes of 1% or lower (Allen, regional tectonic downdropping was given by
1982). Oversteepening or earthquake shocks are Adams (1990). (See Jibson, this volume, for discus-
often invoked to explain sheet slumps. sion of seismic versus nonseismic origin of
Thin beds of alternating mud and cohesionless turbidites.)
sediment can also be warped by earthquakes,
especially in strata near the surface. Audemard 5.2.5. Paleoseismie examples distinguished from
and de Santis (1991) observed that during a mod- nonseismic subaqueous features
erate earthquake, warping developed within the Data showing the regional pattern of types,
uppermost 0.2 m where mud overlay a thin sand sizes, and abundance of features generally are
stratum. Sims (1973) reported the occurrence of required to develop compelling evidence for a
low-amplitude folds in the uppermost 4-5 cm of seismic origin for the types of subaqueous features
lake sediments (mud with very fine sandy laminae described in this section. Proving widespread devel-
and partings) and interpreted the folds to have opment of deformed sediments at the same time
developed in response to a nearby earthquake in fluvial deposits is extremely difficult; proving
having a magnitude of 6.5. Load structures and such development in lakes, especially deep lakes,
recumbent folds were also developed in the muds. can be more feasible.
Sims found similar features in buried strata and A good example was given by Ringrose (1989).
showed that the buried features were synchronous Features that probably resulted from seismic shak-
with previous historic earthquakes. Similarity of ing were found at widespread exposures in sedi-
morphology in the same types of source sediment ment in a late Quaternary glacial lake in Scotland.
played an important role in the arguments of Sims. The sediment consists mostly of laminated silt and
sand. Deformation features developed in two epi-
5.2.4. Turbidites sodes about 10 000 years ago. The features are
A turbidite is a sequence of deposits produced restricted to well-defined horizons of very large
by the flow of a sediment-water mixture known as areal extent. Fig. 36 shows how the styles range
a turbidity current. Many turbidites are thought from most severe (Fig. 36A) to marginal
to be caused by huge submarine landslides. The (Fig. 36D). The most severe deformations are
56 S.F Obermeier/Engineering Geology 44 (1996) 1-76

Seismicity can also cause a short increase in the


rate of sedimentation. Doig (1986) suggested that
undeformed silt horizons in widespread lakes were
caused by prehistoric earthquakes in eastern
Canada. An important element in interpretation
was the presence of the silt horizons containing
distinctive chemical imprints within normally
organic-matter-rich deposits; these same relations
were documented for a historic earthquake.

5.3. Featuresformed by weathering

Distinguishing liquefaction features from those


of weathering origin can be difficult where weather-
ing has been severe. Many varied features pro-
duced by chemical weathering in the Southeastern
United States mimic those caused by earthquake
liquefaction (Obermeier et al., 1990). The discus-
sion below concerns features that are regional in
Fig. 36. Schematic depiction of sediment deformation structures occurrence, but the concepts are applicable to
observed by Ringrose (1989) in late Quaternary lake sediments;
deformation probably resulted from seismic shaking. Sediments
sediments in many subtropical areas. The discus-
consist of laminated silt (white) and sand (stippled). (A-D) sion focuses on the following: (1) E-horizon
Change from most severe deformations to slightly distinguisha- tongues and (2) BE or fragipan soil horizons.
ble deformations. Dimension of each figure is about 2 m. A loose, clean, white sand blanket covers large
(Reproduced with permission of Blackwell Science Ltd.) regions of the Southeastern United States where
the ground surface is underlain by sandy sediments
centrally located in a geographic sense and are of barrier-beach and nearshore marine origin that
characterized with the "fault-grading stratigraphy" are Pleistocene to Tertiary in age. The white sand
of Seilacher (1969). This stratigraphy refers to a underlies a 0-15-cm-thick, dark-gray A horizon.
four-fold sequence, going from the top down: (1) This white sand resembles vented sand but, instead,
completely fluidized sediment, (2) partially coher- is a pedogenic E horizon that has formed by severe
ent sediment layers, (3) faulted layers, and (4) weathering and leaching of the sandy sediment
undeformed sediment. Evidence of deformation in (Gamble, 1965). Clays and labile minerals have
Fig. 36D is so slight that beds are only gently been removed from the E horizon, and weathering
warped, and there is very minor evidence for products have been deposited in the underlying
fluidization. B horizon.
Adams (1982) reported the presence of thin The boundary between the E and B horizons is
contorted zones within silt and clay varves of early commonly abrupt and irregular and is charac-
Holocene age in eastern Canada. The contorted terized by narrow, near-vertical tongues of the
zones appeared to be coeval at sites in widely white E horizon that penetrate downward into the
scattered glacial lakes, and there was a crude B horizon. Locally, tongues of E-horizon sand
regional pattern with greatest deformations at the extend more than a meter into a thick, red to
center. Thus, a paleoseismic origin was interpreted. brown, clayey B horizon. Tongues of this size and
M6rner (1985) reported a similar type of finding shape can give the impression of fractured and
in glacial lake varves of early Holocene age in brecciated ground and might be mistaken for
Sweden; association of deformed varves with tec- liquefaction features unless examined carefully.
tonic faulting, fracturing, and slumping led to a Pedogenic tongues can range in morphology from
seismic interpretation. tubular (Gamble, 1965) to planar (defining
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 57

B-horizon polygons; Nettleton et al., 1968a). (Daniels et al., 1966; Nettleton et al., 1968a, b;
Hence, tongue morphology is tenuous evidence for Steele et al., 1969). Relations illustrated in Fig. 38
either pedogenic or earthquake origin. tend to occur in the area (plan view) between red,
A very common type of pedogenic tonguing is oxidized spodosols on beach ridge crests and the
sketched in Fig. 37. The tongue of white sand is area where there are black, organic-matter-rich
10-30 cm thick and dips gently downslope. The spodosols in interridge depressions. For the exam-
tongue is surrounded by a rind of black to dark- ple shown, the BE' horizon is a zone of pale-
orange, humate- and iron-oxide-stained sand. brown, clayey sand and white sand. Both above
Some of the nearly horizontal parts of the tongue and below this soil horizon are zones of white,
can be traced as far as 7 m. At the upslope end of well-sorted, fine to medium quartz sand. This sand
the tongue, the feature abruptly turns up, breaks cuts vertically across the clay layer at many places
through the overlying Bh soil horizon, and flares and creates irregular masses of the brown clayey
toward the surface. Just below the surface, the sand. The contacts between the white sand and
tongue is continuous with white sand of the E the brown clayey sand are typically sharp with
horizon. The downslope end of the tongue (not regard to both color and texture. The sand grain
shown) is terminated by a black- and orange- sizes in both the leached sand and the clayey sand
stained rind that is continuous with the Bh horizon; masses are essentially the same. The sharp contacts
in other places, the tongue-rind contact becomes may suggest to the uninitiated that this feature
diffuse downslope. formed by some type of ground disruption, possi-
Another category of pedogenic feature that bly ancient liquefaction.
might be confused with earthquake-induced lique- The examples above are but two of many types
faction features is the BE' horizon. The horizon is of confusing features caused by weathering. Other
surrounded by leached, clean sand (Fig. 38). common types resemble shear zones and joints
Studies of this weathering profile indicate that the caused by faulting.
clean sand forms from the progressive chemical Distinguishing liquefaction features from those
destruction of a clay-rich Bt (argillic) horizon of weathering origin cannot always be done, but

~ ~ . ... S//~hdy
~/op/ngOroundsu~ce
.-".'" .' : ~ : - - ~ - - , L ~ ".White, manive, w e l l - l o r t e d u n d ,'~'~'='
~y~'

~ ~ . . . U ~ F / -~ . ,. _~/ /, / ./ /. 7 ./ /./ /, / / / / -' . /. : : . . . / ~/ .j f. . .. . ~- ~ . _ ~ .. ..~.....


..... / / . [ /'.. --~--~-Z / / / / J ~ / .Grayish-black, m a s s i v e B h - h o r i z o n M n d , / ' ~ ' - - ~

" ~- > -,. ~ "~" Dark-reddish-brown,


m--hrhoon
~ m ~ ~ " " \ " ~ 7 ~ sand veryf rm
am -,, ~ .... . horizontc;,;gi~el.. "..'.. . .

> Q.

R~ o ~ ~ . ,;~'~,~. E,~,~=~

o
NOVERTICALEXAGGERATION //

Fig. 37. Block diagram showing a white, pedogenic, E-horizon sand tongue exposed in a drainage ditch near Charleston, SC. Prime
designations on E and Bh horizons indicate the lower horizons of a vertically repeated weathering profile. Host deposit is a structureless
(massive) sand. Bhir is iron-enriched Bh horizon. Note that the E-horizon tongue dips in the same direction as the ground surface.
58 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

Ii 0 20 iNCHES
! Bank overhang I II II
O 50 CENTIMETERS
NO VERTICAL EXAGGERATION

/ / , . / /
/ / / / ,, / / / /
/ / / / , ~ / , , /
/ /~ , / / /
/ -
/ / .... , / , ," . /
, / / , /"

," .- / / , , ,/ ,/ Massive Bh horizon s a n d ; / / /


/ / / , / /" /" /"
/
/
///
// , /
//," / .grayish-black grading down" , /
//,///,". .... /, / / to dark-reddish-brown / ", ///
,-'/ ./ , , / / / / / / / . /- /

" ../ z/ / / , // , - / " ./ , " " //,

L
! ~: !..J~ i . ...~
i: W h te, mass ve, well-SOrted Sand

~r

L .. . -

White, well-sorted sano,


locally stained pale brown
\

Not exposed --- ~ - / Not scraped

Fig. 38. Schematic vertical section through BE' weathering horizon near Charleston, SC. White, structureless (massive), well-sorted
sand is interpreted to have had clays removed by chemical weathering.

some clues include the following: (1) comparison casts are downward-pinching, planar, nearly verti-
of mineralogy and grain sizes of stable minerals cal features that originated by thermal contraction
(especially quartz) in the feature with those in the of frozen ground.
host sediment, and (2) regional pattern of size and Not discussed are sediment deformations caused
abundance of the features. by the weight or movement of glaciers. Such
deformations, called "glacitectonic," have been
5.4. Features formed in a periglacial environment extensively referenced by van Loon and
Brodzikowski (1987).
Two classes of features produced in a periglacial
(freezing-melting) environment, loosely defined as 5.4.1. Involutions
( 1) involutions and (2) ice-wedge casts, can resem- Involutions are level-ground perturbations of
ble features that have an earthquake-induced surficial sediments caused by freezing or thawing
origin. Both classes are discussed below. action (see examples in Fig. 39). This action can
Involutions are surficial manifestations of frost- also cause localized development of excess pore-
related stirring (cryoturbation) and are generally water pressure. Involutions frequently show effects
characterized by distortion and mixing of the of fluidization (Fig. 39B). Involutions very com-
uppermost meter or so of the ground. Ice-wedge monly have boundaries and convolute bedding
S.F. Obermeier/EngineeringGeology 44 (1996) 1-76 59

Fig. 39. Examples from the Northeastern United States of ground disturbances associated with involutions. (A) Very large involution
of eohan sand penetrating down into glaciofluvial gavel. Involution extends about 1 m in depth. (B) Fluidized sand and gravel
penetrating upward into eolian silty sand. Note the similarity between these frost-induced involutions and dikes of seismic liquefaction
origin. Shovel handle is 50 cm long. Photographs courtesy of B.D. Stone.
60 S. F Obermeier/Engineering Geology 44 (1996) 1-76

that point to a predominantly plastic mode of In summary, many involutions are so similar
deformation. Involutions can be especially abun- to earthquake-induced features that assigning
dant where an eolian silty or fine sand deposit is an unequivocal origin may not be possible.
at the surface. Beneath the uppermost deposit, Involutions commonly are associated with ground
sediment may range from gravel sized to silt sized patterned by freezing ("patterned ground") and
(see Fig. 39 herein; Van Vliet-Lanoe, 1988, ice-wedge casts, however, and this association
Fig. 11). helps in interpretation.
The morphological range of involutions is large.
Vandenberghe (1988, p. 182) listed six types in 5.4.2. Ice-wedge casts
terms of symmetry, amplitude-wavelength ratio, The types of features that are included in this
and pattern of occurrence. The features range from category are variously described as ice-wedge casts
individual folds of small amplitude and large wave- (Black, 1976), ground-wedge pseudomorphs
lengths (resembling slightly warped bedding caused (Harry and Gozdzik, 1988), or other names based
by earthquake shaking) to intensely convoluted on morphological variations. All result from
forms having amplitudes generally between 0.6 ground cracking caused by contraction of frozen
and 2 m (also resembling earthquake-induced con- ground. These contractions often fracture the
voluted bedding); load structures, diapirs, and ground in a polygonal network, in plan view.
dikes are also common. The genesis of involutions Wedges of ice then form in the cracked ground.
is probably related to three main process cate- The ice wedges grow wider through time. Later,
gories: load casting during melting, pressures in when permafrost melts, host sediment slowly
water trapped between freezing fronts, and pres- replaces the melting ice. Eolian sediment may also
sures and heaving caused during freezing be blown into the depression.
(Vandenberghe, 1988). Fig. 40 shows a vertical section of an ice-wedge
Features caused by load casting during melting cast of a type commonly seen in the Northeastern
(mainly bulbous downward intrusions and detach- United States. The cast is the downward-pinching
ments of sediment going into a host) are similar body of sediment. It contains slump structures
in form to soft-sediment deformation features that formed as sediment fell into or moved down
described in a previous section. One important the fissure as ice melted. Upturned edges of strata
difference, though, is that frost-related load fea- in the upper part of the host were formed by
tures are independent of the vertical lithological horizontal pressures as the ice wedge grew in
composition of the sediment. Even lithologically width. The cast can have a width much exceeding
homogeneous sand can host load structures. Low- a meter and a depth of as much as 10 m (Flint,
density, organic-matter-rich sediments can also 1971). Commonly, though, in the Central and
sink into denser sand (Harry and Gozdzik, 1988, Eastem United States, widths at the top are on
pp. 48-49). the order of 0.5-1 m and depths are 3-4 m (see
Permanently frozen ground beneath a zone of sketches in Johnson, 1990 and Stone and Ashley,
thawing commonly leads to formation of involu- 1992). Ice-wedge casts are common and wide-
tions along the contact. Involutions are also spread in clean granular deposits such as glaci-
thought to form as a result of high pressure in ofluvial sands and gravelly sands that have been
water trapped between freezing fronts. Such a subjected to permafrost conditions (Svensson,
pressure may develop between two impermeable 1988; an excellent photograph is in Flint, 1971,
layers, in which the basal layer is permafrost and p. 279). Casts are also widespread in glacial till
the upper layer is the seasonal freezing front. Small (Svensson, 1988; Johnson, 1990) but tend to be
mud volcanoes (Shilts, 1978), as large as 3 m in more faintly delineated than those in clean granu-
diameter, are thought to be caused by this process. lar deposits.
Different susceptibilities to frost heaving and ice An earthquake-induced origin can be incorrectly
lensing may also explain some involutions. Small deduced for ice-wedge casts, especially in clean
differences in grain size may be significant. granular deposits. However, with careful field
S. F. Obermeier /Engineering Geology 44 (1996) 1- 76 61

o 1 2m

liii!i!i!i !:ii EOUAN MANTLE


(cryoturbated)

~ ~ : : i~:: ~~i~i~ ~).~, .... FLUVIAL TERRACE


DEPOSITS

~,fine sand
and silt

1-1 ~ . "' \ . ~ - A ' ~ . " V " ".'.../"

.- ._h . ' . ' . ' . , " . .


. . . . . . . . . . , --i..c_~
, ".'.'.'.'.''.."..'.'''..

o a " . ". " x ' . ' ~


.~/-/.".~ ~:S, <"" ". :" ~ - gravel

. -. .- ., ? _ , , _ ~ ~: '' ~" ". "".o" 2 . ,, ~t


! ~!~:
oo . s. ." "." ~ ". " . . ' " ' ' . " . ".',

"..-7.
4.---.'-: .... ." '-'" '.'.. ". '-
. .~. .~....-7T7"-.
.... ..'.." ,:. "-=-~
. . ' . ~ , ~...-
. . . . - .-..,o
;.~
medium-coarse sand
(cross-bedded)
... .?.::." : : . . . . . '.... " - ~S-

:ii:;i.i.)il-".:"i.!:.:./.ii:~!).!-
...............~.....~?..: .....-.

:~..... ~.-.._~....
~!
" . :', .".: ' . .: . . . ; : i. - i "
PIT FLOOR
Fig. 40. Schematic vertical section of an ice-wedge cast showing upturned strata in the upper part of the host. Downturned strata
occur at greater depth. Within the cast are clasts from the sidewalls, and strata sag toward the middle. Small normal faults border
the walls of the host. Schematic drawing from Stone and Ashley (1992) represents relations commonly observed in the Northeastern
United States.

study, an earthquake hypothesis can be tested. the Northeastern United States show little or no
Field studies (Stone and Ashley, 1992) for the evidence of upturning of strata in host deposits.
example of Fig. 40 show that the uppermost meter The upturned deformation is probably destroyed
of the cast is composed of fine sand and silt, rather whenever the uppermost host material moves into
than sediment from depth vented to the surface; the void left by melting ice. Host material moves
the cast filling includes pebbles polished by wind by sloughing, creep, and localized normal faulting.
abrasion. Stone and Ashley also showed that air Strong vertical alignment of infilling sediment com-
photographs reveal a well-developed polygonal monly takes place, especially in narrow casts. This
network of the wedge structures in plan view. vertical alignment superficially resembles effects of
Therefore, evidence is very strong that the wedge upward flowing water, causing winnowing out of
structures were the result of permafrost. fine grains and fluidization. However, a liquefac-
Many ice-wedge casts in granular materials in tion origin is almost certainly eliminated if there
62 SF. Obermeier/Engineering Geology 44 (1996) 1 76

are no feeder dikes going from the host to sediment donodules and small recumbent folds. A second
in the cast. approach (magnitude-bound method), applicable
Features in glaciofluvial deposits described by to field situations where the regional extent of
Black (1976) as pseudo-ice-wedge casts (Fig. 41) liquefaction from a paleo-earthquake can be esti-
can very closely resemble ice-wedge casts. The mated, determines probable minimum magnitude.
features probably develop as nearly vertical, open A third method (engineering-based method of Seed
tension fractures, which form in response to et al. (1983,1985)) places limits on the accelera-
any number of nonfrost-related mechanisms. tions that formed liquefaction features in many
Possibilities include loss of lateral support or field situations; magnitude is then assessed by
differential settlement of underlying sediments. comparing these accelerations with estimates from
Such mechanisms can occur thousands of years seismological models or from statistical analysis
after the host sediment was laid down. Simply of historical earthquakes. All three of these meth-
lowering the water table can cause large differential ods are discussed in the following sections.
settlements and crack the ground at places where Yet another engineering-based method that
the glaciofluvial deposits are underlain by thick, shows great promise for paleoseismic analysis uses
soft silts or clays. an energy approach for assessing the likelihood of
liquefaction at a site (Law et al., 1990). Although
Pond (1996, in press) has used the energy approach
6. Estimation of strength of paleo-earthquakes for analyzing paleoliquefaction effects, the results
of his analysis have not yet been published. Thus,
The strength of paleo-earthquakes can be inter- discussion of this method will be limited. The
preted by using several independent techniques. energy method will likely be used frequently in the
One such technique uses the very crude association future because it is inherently more accurate than
that exists between severity of shaking, as mea- the method of Seed et al. (1983,1985) and, in
sured by the modified Mercalli intensity scale, and addition, circumvents some of the problems
the threshold for formation of liquefaction effects implicit in that method.
and soft-sediment deformations such as pseu- A promising method based primarily on geology
would use dike width as a measure of the severity
of lateral spreading and thereby of the strength of
z/tension fra___cture shaking. Such a method would be a variation of
i ~\ /" L. . . . . . . . . . the Liquefaction Severity Index (Youd and
....
. . . . .
~ ~\lJ I I t - . . . . . . . . . .
-'N \ / I
Perkins, 1987) and its updated version (Bartlett
]\ ~tk. ..........
I \ "~'1 and Youd, 1992), which provides a quantitative
,~,, \ , f - - - - --~- estimate of the amount of lateral spreading as a
I\\ [l
-~ ,\,~- . . . . . . . . . . .
function of earthquake magnitude and distance
t\ "~I
-.2.\d~-----___ ~--
from the seismic energy source.
_.~,__~-,,Y.,.~-- -- _ _
The methods discussed below for estimating
e e o e .,,,..~',//,I~ ~ e e~ prehistoric earthquake magnitude all require locat-
ing the meizoseismal region. I suggest using a
regional pattern of maximum dike widths to
approximate this region. Where field data are
adequate, a preferable measure is the sum of dike
-iN ............
widths normalized to the amount of outcrop
(P.J. Munson, Indiana Univ., oral commun.,
1994). An excellent example of using dike widths
Fig. 41. Schematicverticalsection of pseudo-ice-wedgecast of to approximate the location of the meizoseismal
Black(1976), probablycausedby tensionfracturingof ground. region has been given by Munson et al. (1995).
Note downturnedbeddingin proximityto fracture. Maximum dike width is used rather than dike
S.F Obermeier/EngineeringGeology 44 (1996) 1-76 63

density because density generally is highly sensitive magnitude by associating intensity values with the
to cap thickness (Ishihara, 1985; Obermeier, 1989). regional extent of ground failure (Galli and
The amount of lateral spreading, alternatively, is Ferreli, 1995).
so insensitive to cap thickness that it is not even
used as a parameter for the Liquefaction Severity 6.2. Method associating magnitude with farthest
Index (Bartlett and Youd, 1992). Caution is appro- occurrence of liquefaction
priate when dike width is used as a measure of
shaking severity at places where prehistoric shak- Fig. 42 shows the distance from the epicenter to
ing has been relatively weak and only relatively the farthest observed liquefaction effect at the
small dikes (on the order of 0.3 m or less in width) ground surface (plan view), such as venting of
have formed. Such small dikes are likely to have sand or ground fissuring, for earthquakes of
formed within 100 m of a stream bank or free different magnitudes. The data are taken from
face. Along many major rivers, the lateral erosion worldwide earthquakes in a wide variety of tectonic
subsequent to a paleo-earthquake exceeds 100 m and sedimentary settings. Liquefaction sites are
within a few hundred to a thousand years and in thin to thick unconsolidated material.
removes many of the larger dikes that were near Amplification of bedrock accelerations is probably
the free face when the earthquake occurred. small where unconsolidated deposits are thin,
Therefore, it may be necessary to sample enough whereas amplification is probably as high as 2-2.5
sites where erosion was probably not severe in at many places where deposits are thick (Idriss,
order to develop meaningful relations concerning 1990). Liquefaction susceptibility is doubtless very
dike widths. high at many of the farthest sites. For earthquakes
having focal depths of less than 50 kin, the farthest
6.1. Method associating modified Mercalli intensity occurrences of liquefaction are bounded by a rea-
with liquefaction sonably well defined curve (Fig. 42). Therefore,
using the curve in the figure should provide a
The modified Mercalli intensity (MMI) value is minimum estimate of paleo-earthquake magnitude,
a qualitative measure of earthquake-induced especially in regions where conditions are less than
damage (Wood and Neumann, 1931). The scale optimal for the formation of liquefaction effects.
ranges from I to XII, where I represents the level Using this figure is the magnitude-bound method.
at which shaking may be felt slightly and XII The farthest sites of liquefaction effects on the
represents total destruction. An MMI of about VI ground surface (in plan view) usually cannot be
is the threshold for widespread development of found in a paleoliquefaction search (normally in
small-scale soft-sediment deformation features sectional view). However, if enough exposure is
such as folds, pseudonodules, contorted lamin- searched in sectional view, it has been my experi-
ations, and recumbent folds (Sims, 1975). ence that this farthest distance can be reasonably
Although liquefaction effects have occurred at approximated. A lower limit magnitude thus can
MMI values as low as V and VI (Keefer, 1984), be established by using the plot in Fig. 42.
the lowest intensity at which liquefaction-induced I cite two examples on the basis on my field
features can become common is VII, provided that experience, one from the Charleston earthquake
highly susceptible deposits are present (National of 1886 and the other from the New Madrid
Research Council, 1985, p. 34). Values of VIII-IX earthquakes of 1811-1812. The regions of both
are generally required before liquefaction-induced these earthquakes exhibit widespread moderate to
ground failure becomes severe enough to cause high liquefaction susceptibility and probable mod-
damage to buildings. A serious shortcoming of erate amplification of bedrock accelerations. Using
using MMI values for paleoseismic interpretations the farthest liquefaction effects observed
is the very crude association of ground-failure (90-100 km) at the time of the Charleston earth-
effects with shaking severity. Still, very approxi- quake as a point on the curve of Fig. 42 yields an
mate limits can probably be placed on earthquake estimated minimum moment magnitude (M) of
64 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

10 10

9 ] 9


O i
,'- 8 e / 8

E
. . ; o

"." _..i_ .
e- 0
t~
,,, ~

6
/ 6
g" ~'g" r

..2
/
1 2 5 10 20 50 100 200 500

Epicentral distance (radius) to farthest liquefaction f e a t u r e (km)

EXPLANATION

Farthest liquefaction effects (sand


blows, fissures, and such) for
shallow-focus earthquakes
o Farthest liquefaction effect for
deep-focus earthquakes (>50 km)
/k Farthest sand blows reported for
1811-12 New Madrid earthquakes

Fig. 42. Relationship between earthquake moment magnitude (M) and distance from earthquake epicenter to the farthest liquefaction
effect (venting to the surface or ground fracturing), with bound suggested by Ambraseys (1988). Curve bounds data for earthquakes
worldwide that had focal depths < 50 krn. Also shown are data for earthquakes having depths > 50 km.

,,~6.8; actual magnitude was M ~ 7.5. I have found spaced regions as far as 250-275 km from the
small liquefaction features (unweathered dikes epicenter of the New Madrid M~8.3 event.
extending almost to the surface) as far as 100 km Farthest surficial effects observed in these widely
from the epicenter of the 1886 earthquake and spaced regions in 1811-1812 were at the same
almost certainly caused by that earthquake. I have distances. Using Fig. 42 yields M = 7.5-7.7. These
also found liquefaction features (unweathered findings suggest that there should be an abundance
dikes extending nearly to the surface) in widely of dikes that pinch together only slightly below
S.F Obermeier/Engineering Geology 44 (1996) 1-76 65

the ground surface in the region where effects of 0.5 i i I

liquefaction were observed at the surface at the


time of the earthquake and that some of these
dikes could be found in a search for liquefaction
features. (Small, pinching-up dikes that do not
approach the surface may extend much further
from the meizoseismal region than do dikes that
_O

LU
n"

~"
0.4

0.3
///III
./ / t

approach the surface.) In summary, my fieldwork 0


>-
in the areas of the two earthquakes above, which 0

was based on farthest discovered paleoliquefaction 9 0.2


uJ
,'-r
features that approach the ground surface (within
1-2 m), indicates that the estimates of magnitude
obtained from Fig. 42 are not outrageously lower
than actual values, if field conditions are favorable
for liquefaction. I I I
The curve in Fig. 42 sometimes can be adjusted 0 10 20 3O 40
in regions of historical liquefaction to account for MODIFIED PENETRATION RESISTANCE N 1,
the influence of local setting on the farthest devel- iN BLOWS PER FOOT

opment of liquefaction effects, as Obermeier et al.


(1993) have done. The prehistoric magnitude can
EXPLANATION
probably be assessed more accurately if the caus-
1;h avg Average earthquake-induced horizontal
ative fault can be located. Curves relating farthest cyclic shear stress.
liquefaction effects to distance from the fault rup-
ture zone have been given by Bartlett and Youd t~6 Initial vertical effective overburden stress.
(1992, Section 4). N1 Standard Penetration Test blow count
measured in field, modified to blow count
6.3. Method of Seed et al. associating acceleration resistance at vertical effective stress
of 1 ton/ft2 (97.5 k Pa).
with liquefaction
Fig. 43. Curves for the method of Seed et al. (1983) used to
Use of the Seed et al. (1983,1985) method avoids evaluate the potential occurrence of liquefaction with accompa-
nying venting of sand or appreciable ground cracks at a site on
the problem with the magnitude-bound method of
level ground. Curves are for dean sand deposits (average grain
not being able to locate the distal effects of lique- diameter >0.25 ram) and for different earthquake magnitudes
faction. The Seed et al. method was developed to (5.25-8.5); M is Richter magnitude ML or surface-wave magni-
provide engineers with a way of estimating the tude M,, whichever is larger. Points above and to the left of
shaking threshold required to produce surface curves show conditions having high potential for liquefaction.
manifestations of liquefaction during future earth-
quakes (Fig. 43). The method can be adapted to tions only. Sites of large lateral spreads, especially
paleoseismic studies. The method is based on near free faces, should be excluded because of the
worldwide observations following many earth- possibility of enhanced ground breakage accompa-
quakes, supplemented by extensive laboratory nied by copious venting.
studies. The curves in Fig. 43 are not intended to To use Fig. 43, the relative density of source
indicate whether any very minor liquefaction at deposits must be evaluated in place by means of
depth occurs (such as the liquefaction of an iso- the Standard Penetration Test (SPT) blow-count
lated thin sand bed); rather, the curves predict method (Table 1). An adjusted SPT blow-count
where liquefaction is severe enough to produce value (the N1 value of Fig. 43, which is the value
scattered occurrences of limited venting or small corrected for site conditions of depth to the water
ground openings (several centimeters) at the sur- table and static stress conditions of the source
face. The method applies to level ground condi- stratum) is then related to the shaking required to
66 S.F. Obermeier/Engineering Geology 44 (1996) 1 76

cause venting. Fig. 43 shows the boundary curves depth to the water table. If all factors are favorable,
for different earthquake magnitudes. The curves the Seed et al. curves probably provide a reason-
relate the N1 value in clean sands to the field cyclic able assessment for paleoseismic calculations at
stress ratio, which, for an element in the sand sites where only minor venting or ground breakage
layer, is the ratio of the average earthquake- occurred. Favorable conditions would include a
induced horizontal cyclic shear stress ('l~h avg) to cap thickness of less than a few meters (Ishihara,
the vertical static effective stress (eye). The field 1985; Obermeier, 1989; Youd and Garris, 1995),
cyclic stress ratio due to earthquake shaking is a water table within a few meters of the surface,
computed from the following equation (Seed et al., and a cap that is not greatly strengthened by a
1983): mat of roots. At sites where liquefaction is mar-
ginal, it is likely that the blow-count N-values of
Zh avg/tr'o = ( 0.65amaxaor d)/( tr'og) the source bed have not been changed substantially
where amax is the peak horizontal acceleration by the occurrence of liquefaction. Paleoseismic
at the ground surface, c~o is the total overburden calculations should not be made at sites where
stress on the sand under consideration, g~ is the severe paleoliquefaction took place, because these
initial effective overburden stress (total stress sites can yield only estimates of minimum shaking
minus pore-water pressure) on the sand layer under levels. Severe liquefaction can be indicated by large
consideration, r d is the stress reduction factor sand blows or a strong warping of the cap. Sites
(ranging from a value of 1 at the ground surface of wide lateral spreads, especially near free faces,
to a value near 0.9 at a depth of about l0 m), and should not be used with the curves in Fig. 43,
g is the acceleration due to gravity. The various because these sites can yield acceleration values
magnitude curves in Fig. 43 are based on the that are too low.
premise that the expected duration (that is, the The curves in Figs. 43 and 44 do not account
number of cycles) of strong shaking becomes fully for field conditions that are not conducive to
longer as magnitude increases. venting, such as the presence of a very thick cap
The locations of the curves in Fig. 43 were or a mat of peat in the cap. Eliminating sites
picked partly by judgment. Liao et al. (1988) have having these unfavorable conditions would likely
used statistical regression methods to assess the increase the probability of liquefaction. A miscon-
probability of liquefaction for the curves. Their ception among many geologists is that the strength
results are shown in Fig. 44, normalized in terms of clay in the cap has a major bearing on whether
of magnitude. Although Liao et al. results basically venting occurs. However, the pore-water pressure
agree with the trend of the Seed et al. curves, that develops because of liquefaction along the
Fig. 44 shows some noteworthy differences. In base of the cap normally greatly exceeds the tensile
clean sand deposits, the Liao et al. 50% probability strength of the sediment in the cap; any extensive
curve lies almost everywhere above the Seed et al. liquefaction thus should be adequate for hydraulic
curve. This difference is understandable in terms fracturing of the cap, and dikes should form. Even
of the original intent of Seed et al. to use the curve where the cap is very soft, field observations show
for engineering design. If, however, the intent is that brittle fracturing almost always occurs rather
to reverse the process to calculate the maximum than only formation of plastic intrusions.
ground acceleration that probably occurred at a Fig. 43 can be modified to account for the fact
paleotiquefaction site, then the engineering conser- that a high content of silt and clay in source
vatism in the Seed et al. curve could lead to an sediment diminishes the effects of liquefaction.
acceleration value that may be too low. Fig. 44B shows the effect of silt and clay according
Whether acceleration values obtained from the to the statistical analysis of Liao et al. (1988).
Seed et al. curves are too low depends on the Liao et al. noted that their results do not indicate
constraints on liquefaction and venting at the site, a major difference between sediments containing
and these constraints depend very much on factors 15% silt and clay and those containing 35%, a
such as cap thickness, strength of the cap, and difference that Seed et al. (1985) would imply. The
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 67

I I I I I I I I
_o A B
Prbability\.9 5 .1 35%
n 0.5 .9 15%5 .1
u~
.
.Q6,~efe/,
._~ 0.4 ~'.,"~,oF,?~. /
IIi;. IO~ .

q / :'~'~;o; ~ ,o;~ ,% /
w
,"r 8.3

/
~ 0.2 //
0
z
Seed et al. 119851,
I 11985)
~ o.1
~ c lUy cNontfr nS~It ~o d curves o;ailt anci
z clay content = 15%
end 35%
<
/
/
0.0 i I I I I I I I
0 10 20 30 4O 10 20 30 40
NORMALIZED (N116o, BLOWS PER FOOT NORMALIZED (N1160, BLOWS PER FOOT

Fig. 44. Probability of liquefaction with accompanying venting or appreciable ground cracking, normalized to blow-count value,
magnitude, and field cyclic stress ratio. (A) Curves for sediment having silt and clay content of < 5%. (B) Curves for sediment having
silt and clay contents of 15 and 35%. From Liao et al. (1988). (N1)6o blow-count value corresponds to a rod energy value of 60%
being transmitted from the hammer to the end of the rod (National Research Council, 1985, pp. 98-99). Earthquake magnitudes
are the same as those in Fig. 43.

reliability of the curves in Fig. 44B is not high, been dissipated into granular, permeable sediment
however, because of the paucity of good data, along the base of the cap.
especially at high values of cyclic stress ratio Earthquake shaking can increase the SPT blow
(Youd, 1996a, in press). count (N1 value) of a source stratum (generally by
Liao et al. have provided a sound mathematical a small amount) even if the shaking is not strong
method for evaluating Seed et al. procedure in enough to cause liquefaction (Castro, 1987). An
terms of probability. A larger data base having occurrence of severe liquefaction may significantly
better estimates of earthquake magnitudes was increase the blow count of the source stratum (see
used by Loertscher and Youd (1994) to update data of Kawakami and Asada, 1966, Fig. 4;
the mathematical analysis. Results of a very recent Chameau et al., 1991, Figs. 4, 6); it is not known
conference show that the procedures of both Seed how severe liquefaction must be to cause a signifi-
et al. and Liao et al. need to be adjusted to better cant change or what field settings are conducive
account for the influence of earthquake magnitude to such a change. Alternatively, sites showing field
at magnitudes of less than about 7.5 (Youd, 1996b, evidence of strong liquefaction may not show
in press). evidence of measurable changes in the source
To use the method of Seed et al. for estimating stratum (T.L. Holzer, US Geological Survey, writ-
the strength of paleoseismic shaking, bounds must ten commun., 1994). At sites where liquefaction
be placed on depth to the water table at the time effects have been marginal, however, changes
of an earthquake. Maximum depth to the water caused by an occurrence of liquefaction are likely
table sometimes can be bracketed by observing the to be small. Even this hypothesis should be tested
highest regional level at which dikes cut the base where possible. In some geographic regions, a
of the cap. The water level likely was at least that sense of likely changes may be obtained by statis-
high; if the water level had been much lower, tically comparing blow counts in sediments that
excess pore-water pressure probably would have predate an earthquake with those in sediments
68 S.F. Obermeier/Engmeering Geology 44 (1996) 1-76

that postdate it. Liquefaction during a paleo- cal data. Comparisons should be made at various
earthquake is likely to have been most severe at epicentral distances for a prehistoric earthquake.
sites where blow counts are lowest. Large differ- Wherever possible, an estimate of magnitude
ences between the lowest blow counts in pre- should also be based on the observed areal distri-
earthquake sediments and those in post-earth- bution of paleoliquefaction features (i.e., the mag-
quake sediments could indicate changes caused by nitude-bound method).
liquefaction. Factors that control the regional extent (span)
A common practice among research engineers and sizes of liquefaction features include liquefac-
is to make penetration (SPT) borings near a site tion susceptibility and amplification of bedrock
of venting following an earthquake; data from shaking, as well as seismological factors such as
these borings are used to estimate the properties focal depth and shaking frequency of the earth-
of the sediment that liquefied, and these properties quake and stress change (drop) in the rock of the
are then used as data for Seed et al. method (for rupture zone at the time of the earthquake.
example, Fear and McRoberts, 1995). This prac- Possible shaking characteristics are myriad. A shal-
tice fails to recognize that there are many field low hypocentral depth can cause more severe
situations where venting could have developed shaking in the meizoseismal area than a deeper
from a sill that extended laterally for tens of meters earthquake. Away from the meizoseismal area,
from the source sediment. (Examples of such however, shaking may be less severe for a shallower
sills (Fig. 22) were discussed in the section earthquake than for a deeper one. A high stress
"Characterization of Sills in Sectional View" with drop should cause accelerations to be high both
regard to the 1811-1812 New Madrid earth- in the meizoseismal area and far away. Clearly,
quakes.) As a result of this engineering practice, any interpretation of prehistoric magnitude needs
the penetration resistance of a source bed may be to be calibrated as much as possible to the local
overestimated. Geological examination of tube tectonic setting in order to select the most reason-
samples near a venting site in combination with able parameters.
more widespread penetration testing around the Neither the magnitude-bound method nor the
vent might indicate whether venting occurred from method of Seed et al. (1983,1985) automatically
a sill or a steeply dipping dike and thereby indicate accounts for some factors that control the regional
the location of the source bed. extent of liquefaction. Peak accelerations, for
The discussion above has focused on using SPT example, are strongly related to stress drop (Hanks
blow counts for evaluating liquefaction potential. and Johnston, 1992, Eq. 6). An exceptionally high
Use of the SPT procedure for this purpose is stress drop, however, implies strong shaking of
extremely common but in the near future will short duration. Liquefaction is controlled mainly
likely be supplanted by use of the cone penetrome- by acceleration and duration of strong shaking.
ter, which can determine sediment properties con- Where liquefaction is concerned, a high stress drop
tinuously through depth. A continuous record is likely to be largely offset by strong shaking of
would also allow better estimates for paleoseismic low duration, and vice-versa. This offset probably
analysis. prevents estimates of prehistoric magnitude from
being greatly in error, although this theory has
6.4. Overview of methods for estimating magnitude not yet been evaluated rigorously.
Some inherent shortcomings in Seed et al.
The curves shown in Figs. 43 and 44 for the method for evaluating liquefaction susceptibility
method of Seed et al. (1983,1985) do not yield a have been discussed by Law et al. (1990). The
unique solution for both acceleration and magni- shortcomings are due mainly to the fact that the
tude; only possible combinations can be deter- peak accelerations of Seed et al. method do not
mined. Any estimate of magnitude made by using always correspond with the earthquake energy that
the curves should be compared with estimates of causes shaking at a site, which Law et al. have
accelerations from seismological models or histori- shown to be a more fundamental and accurate
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 69

parameter for evaluating whether liquefaction from the epicenter. These peak accelerations are
should occur. Specific deficiencies of Seed et al. then compared with seismological estimates of
method are that (1) the peak horizontal accelera- peak acceleration versus epicentral distance for an
tion in one direction does not always reflect the earthquake of that magnitude. It should be noted
amount of energy acting on a point, and (2) the that Pond examined data from six historical
peak horizontal acceleration does not reflect what worldwide earthquakes and found that his method
is happening in all vibration frequencies. In addi- works well for predicting sites that exhibit no
tion, the peak horizontal acceleration may relate liquefaction effects.
poorly to shear strain, which is the physical mecha- The analysis of prehistoric liquefaction features
nism that breaks down grain-to-grain contacts and centered about Charleston, S.C., previously pre-
leads to liquefaction. This poor correspondence sented, implicitly accounts for all the factors con-
may be important in situations where surface trolling liquefaction noted above (earthquake
shaking is dominated by very low vibration fre- shaking characteristics, local tectonic situation,
quencies and large shear strains but low accelera- sediment propreties, etc.). I concluded that some
tions. This situation could possibly occur where prehistoric earthquakes near Charleston were at
very thick (on the order of a kilometer?), unconsol- least equal in strength to the 1886 earthquake
idated sediments dampen out high-frequency (M~7.5). Seismological factors were indirectly
vibrations as they pass upward. Alternatively, considered by showing that the size and span of
severe dampening seems unlikely for lower-fre- features predating 1886 exceeded those of the 1886
quency vibrations (less than about 10 cycles features, both in the meizoseismal region of the
per second) that largely control field occurrences 1886 earthquake and far away. In addition, the
of liquefaction (Martin Chapman, seismologist,
fact that features that predate 1886 are more
Virginia Polytechnical Institute, oral commun.,
abundant and larger near Charleston than they
1996).
are elsewhere helped to define the region of strong-
A problem occurs in trying to use Seed et al.
est shaking.
method for paleoseismic analysis because it
The South Carolina example, where one can
requires selection of marginal liquefaction sites.
compare historical and prehistoric liquefaction
Any selection process is by nature rather sub-
jective. Also, identifying the source bed that lique- effects to estimate past earthquake magnitudes, is
fied is sometimes difficult or impossible. These exceptional. In most places, such extensive histori-
issues as they relate to paleoseismic analysis have cal liquefaction data from a very strong earthquake
been discussed in more depth by Obermeier (1995). do not exist for prehistoric earthquakes. Estimates
A procedure for using Law et al. energy method of magnitudes elsewhere must be based on other
of for paleoseismic analysis has been developed by approaches. One such approach was used for
Pond (1996, in press). Pond's approach is first to analyzing liquefaction features in the Wabash
assume the magnitude of a prehistoric earthquake Valley (Obermeier et al., 1993). Historical liquefac-
and then to apply the energy method to that tion-producing earthquakes in 1811-1812 and
magnitude to select the most compact sediment 1895 occurred in the New Madrid seismic zone,
(that is, highest N1 value) that should have lique- about 100 km southwest of the Wabash Valley.
fied at a specific site. Field geotechnical data (SPT Because the seismotectonic settings of the two
data, for example) are then examined at a paleoliq- areas are likely similar, the factors of hypocentral
uefaction site in order to determine whether lique- depth, stress drop, and frequency characteristics
faction should have occurred there. This procedure can be assumed to be similar as a first approxima-
is repeated at paleoliquefaction sites located at tion. The geological and ground-water settings
various distances from the epicenter. Highest NI (which control liquefaction) are also similar at the
values at the various sites are then used with the sites of historical liquefaction and in the Wabash
Seed et al. curves, such as those shown in Fig. 43, Valley. Accounting for differences in the amplifi-
to estimate peak accelerations at different distances cation of bedrock shaking permitted definition of
70 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

a curve parallel to the bound in Fig. 42 and thereby turing or surface oscillations can be greatly
an estimate of magnitude. restricted where cap thickness exceeds 5 m, even
in sediments of moderate susceptibility within the
6.5. Conduct of field search meizoseismal region of a very large earthquake.
Optimal cap thickness is generally about 2-4 m.
The absence of liquefaction features (negative Development of even very small dikes may be
evidence) also plays an important role in assessing suppressed along the base of a cap that extends to
prehistoric earthquakes. If depth to the water table too great a depth. In areas of suppressed dike
can be established through time and the suscepti- development, however, sills as thick as 10 cm may
bility of potential source deposits can be estimated, form along the base of the cap and provide evi-
the lack of liquefaction effects can be used to place dence of seismic liquefaction.
limits on the maximum levels of prehistoric ground Where source sediment may have been only
shaking. Still, there is no well-defined procedure moderately susceptible to liquefaction through
for determining the amount of outcrop that must time, I suggest that it may be necessary to search
be searched for liquefaction features in order to 10 km or more of exposure in order to find effects
support a conclusion based on negative evidence. of accelerations as low as 0.1-0.2xg. The field
Some uncertainty arises because strong shaking setting should also be evaluated to ascertain
probably attenuates from the energy source as a whether the exposures were favorably situated for
somewhat variable (stochastic) function rather the development of dikes through time. A field
than in a smooth manner. Many observations of search ideally should permit the observation of
ground-failure effects and damage to buildings many places where the base of a fine-grained cap
show that large variations in ground response lies on thick liquefiable sand. Such a search is best
probably are commonplace within both short and done when the water table is very low. Even small
long distances from the meizoseismal area. These dikes and sills can be discovered in this way.
variations often have no readily identifiable basis. In searching for paleoliquefaction features, one
Similarly, both the size and the abundance of should keep in mind that correlations often exist
liquefaction-induced features can vary widely between maximum dike width, abundance of dikes,
within a local area. For example, in the meizoseis- severity of venting, and severity of ground warping
mal area of the 1811-1812 New Madrid earth- in a localized area (Youd and Perkins, 1987,
quakes, a given length of outcrop may contain Table 4). All these parameters can be of value in
hundreds of dikes and sills of all sizes. Yet, in an defining the epicentral region of a prehistoric earth-
outcrop of the same length nearby, where condi- quake. In my experience, however, there are some
tions for liquefaction appear to be about the same, field situations where the relationships described
features may be fewer by an order of magnitude. by Youd and Perkins do not seem to apply.
Any search for paleoliquefaction features should Whether their relationships apply likely depends
take into account the fact that dikes generally on the existence of a very shallow ground-water
develop in response to lateral spreading much table. In addition, the lack of consistent relation-
more readily than they do from hydraulic fractur- ships may result from many other independent
ing and surface oscillations. In many areas where factors controlling liquefaction effects. These
seismic shaking levels have been as high as MMI factors vary greatly in origin and include the nature
VIII or IX, almost all dikes probably formed by of seismic shaking as well as sediment and site
lateral spreading. Lateral spreading at these inten- characteristics.
sity levels can be restricted to within a few hundred Walsh et al. (1995, p. 1) have advocated using
meters of a stream bank or topographic depression a statistical procedure to determine whether the
in field settings where the liquefaction susceptibility widest dikes have been located in a search for
is not high. Dikes may not develop at all or be paleoliquefaction effects. The procedure is
rare at greater distances. intended to determine whether the widest dikes
Dike development as a result of hydraulic frac- have been discovered. (Walsh et al. (1995) also
S.F Obermeier/Engineering Geology 44 (1996) 1-76 71

mistakenly cite me as advocating that m a x i m u m American Society for Testing and Materials, 1978. Designation
dike widths can be relied on to determine earth- D 1586-67 (reapproved 1974), Standard method for penetra-
tion test and split barrel sampling of soils, part 19 of Annual
quake magnitude.) A l t h o u g h their statistical pro-
book of ASTM standards: Philadelphia, PA, pp. 235-237.
cedure is m o s t likely valid in m a n y places, I have Amick, D. and Gelinas, R., 1991. The search for evidence of
observed so m a n y exceptions that I have no great large prehistoric earthquakes along the Atlantic seaboard.
confidence in using their m e t h o d alone. Instead, I Science, 251(4994): 655-658.
suggest using other approaches in addition to their Amick, D., Gelinas, R., Maurath, G., Cannon, R., Moore, D.,
Billington, E. and Kemppinen, H., 1990. Paleoliquefaction
method. I r e c o m m e n d searching as wide an area
features along the Atlantic seaboard. US Nuclear Regulatory
as possible and observing the severity o f venting Commission Report NUREG CR-5613, 146 pp.
and g r o u n d warping a n d the thickness o f cap Andrus, R.D., Stokoe, K.H. and Roesset, J.M., 1991. Liquefac-
penetrated by dikes at locales far f r o m free faces, tion of gravelly soil at Pence Ranch during the 1983 Borah
as well as measuring dike widths (e.g., Obermeier, Peak, Idaho earthquake. Proceedings of the Fifth Interna-
1995). Estimates o f m a x i m u m dike widths should tional Conference on Soil Dynamics and Earthquake Engi-
neering, Karlsruhe, Germany, pp. 251-262.
also depend o n whether the exposures are in field Andrus, R.D. and Youd, T.L., 1987. Subsurface investigation
settings that were conducive to the f o r m a t i o n o f of a liquefaction-induced lateral spread, Thousand Springs
wide dikes at the time o f the earthquake. Valley, Idaho. US Army Engineer Waterways Experiment
Station Miscellaneous Paper GL-87-8, 132 pp.
Atwater, B.F., 1987. Evidence for great Holocene earthquakes
along the outer coast of Washington State. Science,
Acknowledgment 236(4804): 942-944.
Atwater, B.F., 1992. Geologic evidence for earthquakes during
F u n d i n g for the research required for this article the past 2,000 years along the Copalis River, southern coastal
was provided by grants f r o m the U S Nuclear Washington. J. Geophys. Res., 97(B2): 1901-1919.
R e g u l a t o r y C o m m i s s i o n and the N a t i o n a l H a z a r d s Audemard, F.A. and de Santis, F., 1991. Survey of liquefaction
structures induced by recent moderate earthquakes. Bull. Int.
Reduction P r o g r a m o f the U S Geological Survey.
Assoc. Eng. Geol., 44: 5-16.
Various parts o f the manuscript received con- Bartlett, S.F. and Youd, T.L., 1992. Empirical analysis of hori-
structive reviews f r o m J o h n Sims, A l a n Nelson, zontal ground displacement generated by liquefaction-
T o m Holzer, and Eugene Schweig o f the U S induced lateral spreads. State University of New York at
Geological Survey, f r o m T. Leslie Y o u d o f Buffalo, National Center for Earthquake Engineering
Brigham Y o u n g University, f r o m Eric P o n d and Research, Technical Report NCEER-92-0021, variously
paged.
M a r t i n C h a p m a n o f Virginia Polytechnical Bennett, M.J., 1984. Quaternary geology of ground failure sites
Institute, and f r o m S.S.C. Liao o f Parsons in Long Valley and Little Antelope Valley, in Sint.z, J.
Brinkerhoff, Inc. Careful editing by Elizabeth (Editor), Western geological excursions - (Guidebook for
G o o d m u c h sharpened the clarity o f thought. field trips at the) 1984 Annual Meeting of the Geological
Society of America, Reno, Vol. 2: pp. 67-72.
Berg, A.W., 1990. Formation of Mima mounds; A seismic
hypothesis. Geology, 18(3): 281-284.
References Black, R.F., 1976. Perigiacial features indicative of permafrost;
ice and soil wedges. Quaternary Res., 6: 3-26.
Adams, J., 1982. Deformed lake sediments record prehistoric Brodzikowski, K., Haluszczak, A., Krzyszkowski, D. and van
earthquakes during the degiaciation of the Canadian Shield Loon, A.J., 1987. Genesis and diagnostic value of large-scale
(abstr.) Eos, 63(18): 436. gravity-induced penecontemporaneous deformation hori-
Adams, J., 1990. Paleoseismicity of the Cascadia subduction zons in Quaternary sediments of the Kleszczow Graben
zone; Evidence from turbidites off the Oregon-Washington (central Poland). In M.E. Jones and M.F. Preston (Editors),
margin. Tectonics, 9(4): 569-583. Deformation of sediments and sedimentary rocks. Geologi-
Allen, J.R.L., 1982. Sedimentary Structures - Their Character cal Society (of London) Special Publication 29, pp. 287-298.
and Physical Basis. Amsterdam, Elsevier, Vol. 2:663 pp. Castro, G., 1987. On the behavior of soils during earthquakes
Allen, J.R.L. and Banks, N.L., 1972. An interpretation and - Liquefaction. In A.S. Cakmak (Editor), Soil dynamics and
analysis of recumbent folded deformed cross-bedding. Sedi- liquefaction. New York, Elsevier, pp. 169-204.
mentology, 19: 257-283. Chameau, J.L., Clough, G.W., Reyna, F. and Frost, J.D., 1991.
Ambraseys, N.N., 1988. Engineering seismology. Earthquake Liquefaction response of San Francisco bayshore fills.
Eng. Structural Dynamics, 17: 1-105. Seismol. Soc. Am. Bull., 81 (5): 1998-2018.
72 S.F. Oberrneier/Engineering Geology 44 (1996) 1-76

Cita, M.B. and Lucchi, F.R. (Editors), 1984. Special Issue, ation of liquefaction in sandy soils. J. Geotech. Eng., Am.
Seismicity and sedimentation. Marine Geol., 55(1 and 2): Soc. Civil Engineers, 121(3): 249 261.
151. Fiegel, G.L. and Kutter, B.L., 1994. Liquefaction mechanism
Clague, J.J., Naesgaard, E. and Sy, A., 1992. Liquefaction fea- for layered soils. J. Geotech. Eng., Am. Soc. Civil Engineers,
tures on the Fraser delta; Evidence for prehistoric earth- 120(4): 737 755.
quakes? Can. J. Earth Sci., 29(8): 1734-1745. Finn, W.D.L., Ledbetter, R.H. and Wu, G., 1994. Liquefaction
Coleman, J.M. and Prior, D.B., 1980. Deltaic sand bodies. in silty soils. In S. Prakash and P. Dakoulas (Editors),
American Association of Petroleum Geologists Continuing Ground failures under seismic conditions. Am. Soc. Civil
Education Course Note Series no. 15, 171 pp. Engineers Geotechnical Special Publication 44, pp. 51-76.
Daniels, R.B., Nettteton, W.D., McCracken, R.J. and Gamble, Flint, R.F., 1971. Glacial and Quaternary geology. Wiley and
E.E., 1966. Morphology of soils with fragipans in parts of Sons, New York, 892 pp.
Wilson County, North Carolina. Soil Sci. Soc. Am. Proc., Fuller, M.L., 1912. The New Madrid earthquake. US Geol.
30: 376-380. Surv. Bull., 494: 119.
Darienzo, M.E. and Peterson, C.D., 1990. Episodic tectonic Galli, P. and Ferreli, L., 1995. A methodological approach for
subsidence of late Holocene salt marshes, northern Oregon historical liquefaction research. In L. Serva and D.B. Slem-
central Cascadia margin. Tectonics, 9( 1): 1-22. mons (Editors), Perspectives in Paleoseismology (for Italy).
Dickenson, S.E., Obermeier, S.F. and Roberts, T.H., 1994. Assoc. Eng. Geologists Special Publication 6, pp. 35 49.
Constraints on earthquake shaking in the lower Columbia Gamble, E.E., 1965. Origin and morphogenetic relations of
River region of Washington and Oregon, during late Holo- sandy surficial horizons of upper Coastal Plain soils of North
cene time, in Proceedings of the Fifth US National Confer- Carolina. Raleigh, NC, North Carolina State University,
ence on Earthquake Engineering at Chicago, Illinois. Ph.D. thesis, 254 pp.
Earthquake Eng. Res. Inst. El Cerrito, CA, 3: 313-322. Gelinas, R.L., 1986. Mineral alterations as a guide to the age
Dobry, R., 1989, Some basic aspects of soil liquefaction during of sediments vented by prehistoric earthquakes in the vicinity
earthquakes. In K.H. Jacob and C.J. Turkstra (Editors), of Charleston, South Carolina. Chapel Hill, NC, Univer-
Earthquake Hazards and the Design of Constructed Facili- sity of North Carolina, M.S. thesis, 304 pp.
ties in the Eastern United States. Ann. NY Acad. Sci., Geomatrix Consultants, 1995. Seismic design mapping, State of
558, 172-182. Oregon; report prepared for Oregon Department of Trans-
Dobry, R. and Liu, L., 1992. Centrifuge modelling of soil lique- portation under contract 11688. San Francisco, variously
faction. Proceedings of the Tenth World Conference on paged.
Earthquake Engineering, Madrid, Vol. 11: pp. 6801-6809, Hamada, M., Yasuda, S., Isoyama, R. and Emoto, K., 1986.
Doig, R., 1986. A method for determining the frequency of Study on liquefaction induced permanent ground displace-
large-magnitude earthquakes using lake sediments. Can. ments. Tokyo, Association for the Development of Earth-
J. Earth Sci., 23(7): 930-937. quake Prediction in Japan.
Dutton, C.E., 1889. The Charleston earthquake of August Hamilton, R.M. and Johnston, A.C. (Editors), 1990. Tecum-
31(1886): US Geological Survey Ninth Annual Report seh's prophecy - Preparing for the next New Madrid earth-
1887-88, pp. 203-528. quake. US Geological Survey Circular 1066: 30.
Dzulynski, S. and Walton, E.K., 1965. Sedimentary Features Hanks, T.C. and Johnston, A.C., 1992. Common features of
of Flysch and Greywackes. Elsevier, New York, 274 pp. the excitation and propogation of strong ground motion for
Einsele, G., Ricken, W. and Seilacher, A. (Editors), 1991. North American earthquakes. Seismol. Soc. Am. Bull.,
Cycles and events in stratigraphy. Springer-Verlag, Berlin, 82(1):1 23.
955 pp. Harry, D.G. and Gozdzik, J.S., 1988. Ice wedges; growth, thaw
Elgamal, A-W.M., Amer, M. and Adalier, K., 1993. Liquefac- transformation, and palaeoenvironmental significance.
tion during the October 12, 1992 Egyptian Dahshure earth- J. Quaternary Sci., 3(1): 39 55.
quake. In S. Prakash (Editor), Third International Housner, G.W., 1958. The mechanism of sandblows. Seismol.
Conference on Case Histories in Geotechnical Engineering, Soc. Am. Bull., 48: 155-161.
St. Louis, Mo., June 1-6, 1993. Rolla, University of Missouri Hyndman, R.D. and Wang, K., 1993. Thermal constraints on
at Rolla, Vol. 3: p. 8. the zone of major thrust earthquake failure; The Cascadia
Elgamal, A.-W.M., Dobry, R. and Adalier, K., 1989. Study of subduction zone. J. Geophys. Res., 98(B2): 2039-2060.
effect of clay layers on liquefaction of sand using small-scale ldriss, I.M., 1990. Response of soft soil sites during earth-
models. In T.D. O'Rourke and M. Hamada (Editors), Pro- quakes, in H. Bolton Seed Memorial Symposium Proceed-
ceedings from the Second US-Japan Workshop on Liquefac- ings. Vancouver, B.C., Canada, BiTech Publishers, Ltd.,
tion, Large Ground Deformation and Their Effects on Vol. 2: pp. 273-289.
Lifelines. State University of New York at Buffalo, National Ishihara, K., 1985. Stability of natural soil deposits during
Center for Earthquake Engineering Research, Technical earthquakes. Proceedings of the Eleventh International Con-
Report NCEER-89-0032, pp. 233-245. ference on Soil Mechanics and Foundation Engineering, San
Fear, C.E. and McRoberts, E.C., 1995. Reconsideration of initi- Francisco, Vol. 1: pp. 321-376.
S.F. Obermeier/Engineering Geology 44 (1996) 1-76 73

Iwasaki, T., 1986, Soil liquefaction studies in Japan, State of Japanese Society of Soil Mechanics and Foundation Engi-
the art. Soil Dynamics and Earthquake Eng., 5 (1): 2-68. neering, Vol. 15, No. 4: 81-91.
Jibson, R.W. and Keefer, D.K., 1988. Landslides triggered by Lade, P.V., 1992. Static instability and liquefaction of loose fine
earthquakes in the Central Mississippi Valley, Tennessee and sandy slopes. J. Geotech. Eng., Am. Soc. Civil Engineers,
Kentucky. US Geological Survey Professional Paper 118(1): 51-71.
1336-C, 24 pp. Law, K.T., Cat, Y.L. and He, G.N., 1990. An energy approach
Jibson, R.W. and Keefer, D.K., 1989. Statistical analysis of for assessing seismic liquefaction potential. Can. Geotech.
factors affecting landslide distribution in the New Madrid J., 27(3): 320-329.
seismic zone, Tennessee and Kentucky. Eng. Geol., 27: Li, Y., Craven, J., Schweig, E.S. and Obermeier, S.F., 1996.
509-542. Sand boils induced by the 1993 Mississippi River flood:
Jibson, R.W. and Keefer, D.K., 1993. Analysis of the seismic Could they one day be misinterpreted as earthquake-induced
origin of landslides; Examples from the New Madrid seismic liquefaction? Geology, 24(2): 171-174.
zone. Geol. Soc. Am. Bull., 105(4): 521-536. Liao, S.S.C., Veneziano, D. and Whitman, R.V., 1988. Regres-
Johnson, W.H., 1990. Ice-wedge casts and relict patterned sion models for evaluating liquefaction probability.
ground in central Illinois and their environmental signifi- J. Geotech. Eng., Am. Soc. Civil Engineers, 114(4):
cance. Quaternary Res., 33(1): 51-72. 389-411.
Johnston, A.C., 1994. The stable continental region earthquake Loertscher, T.W. and Youd, T.L., 1994. Magnitude scaling
data base. In J.F. Schneider (Editor), The Earthquakes of factors for analysis of liquefaction, in O'Rourke, T.D. and
Stable Continental Regions: Assessment of Large Earth- Hamada, M. (Editors), Proceedings from the Fifth
quake Potential. Palo Alto, CA, Electric Power Research US-Japan Workshop on Earthquake Resistant Design of
Institute, Report TR-102261, ch. 3, pp. 1-80. Lifeline Facilities and Countermeasures Against Soil Lique-
Jones, M.E. and Preston, M.F. (Editors), 1987. Deformation faction. State University of New York at Buffalo, National
of sediments and sedimentary rocks: Geological Society (of Center for Earthquake Engineering Research, Technical
London) Special Publication 29, 350 pp. Report NCEER-94-0026, pp. 703-711.
Karlsrud, K., Aas, G. and Gregersen, O., 1984. Can we predict Liu, H., and Qiao, T., 1984, Liquefaction potential of saturated
landslide hazards in soft sensitive clays? Summary of Norwe- sand deposits underlying foundation of structure: Proceed-
gian practices and experiences. Fourth International Sympo-
ings of the 8th World Conference on Earthquake Engineer-
sium on Landslides, Toronto (Proceedings), Vol. 1:
ing, San Francisco, Vol. 3: pp. 199-206.
pp. 107-130.
Lowe, D.R., 1975. Water escape structures in coarse-grained
Katayama, S., Fujii, T. and Takahashi, Y., 1966. Damage
sediments. Sedimentology, 22(2): 157-204.
caused by the Niigata earthquake and the geological features
Lowe, D.R. and LoPiccolo, R.D., 1974. The characteristics and
of National Highway in the suburbs of Niigata City. Soils
origins of dish and pillar structures. J. Sediment. Petrol.,
and Foundations, Japanese Society of Soil Mechanics and
44(2): 484-501.
Foundation Engineering, Vol. 6, No. 1: 54-70.
Kawakami, F. and Asada, A., 1966. Damage to the ground and Martin, J.R., II and Clough, G.W., 1990. Implications from a
earth structures by the Niigata earthquake of June 16, 1964. geotechnical investigation of liquefaction phenomena associ-
Soils and Foundations, Japanese Society of Soil Mechanics ated with seismic events in the Charleston, SC area. Virginia
and Foundation Engineering, Vol. 6, No. 1: 14-30. Polytechnic Institute and State University, Blacksburg,
Keefer, D.K., 1984. Landslides caused by earthquakes. Geol. VA, 414 pp.
Soc. Am. Bull., 95(4): 406-421. McCulloch, D.S. and Bonilla, M.G., 1970. Effects of the earth-
Kishida, H., 1966. Damage to reinforced concrete buildings in quake of March 27, 1964, on the Alaska Railroad. US Geo-
Niigata City with special reference to foundation engineer- logical Survey Professional Paper 545-D, 161 pp.
ing. Soils and Foundations, Japanese Society of Soil McKeown, F.A., Hamilton, R.M., Diehl, S.F. and Glick, E.E.,
Mechanics and Foundation Engineering, Vol. 6, No. 1: 1990. Diapiric origin of the BlytheviUe and Pascola arches
71-88. in the Reelfoot Rift, east-central United States; Relation to
Koizumi, Y., 1966. Changes in density of sand subsoil caused New Madrid seismicity. Geology, 18(11): 1158-1162.
by the Niigata earthquake. Soils and Foundations, Japanese Meier, L.S., 1993. The susceptibility of a gravelly soil site to
Society of Soil Mechanics and Foundation Engineering, liquefaction. Virginia Polytechnic Institute and State Univer-
Vol. 6, No. 2: 38-44. sity, Blacksburg, VA, M.S. thesis, 71 pp.
Kolb, C.R., 1976. Geologic control of sand boils along Missis- Mesri, G. and Gibala, R., 1971. Engineering properties of a
sippi River levees. In D.R. Coates (Editors), Geomorphol- Pennsylvanianshale. Proceedings of the 13th Symposium on
ogy and engineering. Dowden, Hutchinson, and Ross, Inc., Rock Mechanics, Urbana, IL, pp. 55-75.
Stroudsburg, PA, pp. 99-114. Mills, P.C., 1983. Genesis and diagnostic value of soft-sediment
Kuenen, P.H., 1958. Experiments in geology. Trans. Geol. Soc. deformation structures - A review. Sedimentary Geol.,
Glasgow, 23: 1-28. 35(2): 83-104.
Kuribayashi, E. and Tatsuoka, F., 1975. Brief review of lique- Mitchell, J.K., 1976. Fundamentals of soil behavior. John Wiley
faction during earthquakes in Japan. Soils and Foundations, and Sons, Inc., New York, 422 pp.
74 S.F Obermeier/Engineering Geology 44 (1996) 1 76

Mrmer, N.-A., 1985. Paleoseismicity and geodynamics in South Carolina and in the fluvial setting of the New Madrid
Sweden. Tectonophysics, 117: 139-153. seismic zone. US Geological Survey Professional Paper
Munson, P.J., Munson, C.A. and Bleuer, N.K., 1994. Late 1504, 44 pp.
Pleistocene and Holocene earthquake-induced liquefaction Obermeier, S.F., Martin, J.R., Frankel, A.D., Youd, T.L.,
in the Wabash Valley of southern Indiana. In: M.L. Jacobson Munson, P.J., Munson, C.A. and Pond, E.C., 1993. Lique-
(Comp.), National Earthquake Hazards Reduction Pro- faction evidence for one or more strong Holocene earth-
gram, summaries of technical reports, Vol. XXXV. US Geo- quakes in the Wabash Valley of southern Indiana and
logical Survey Open-File Report 94-176, pp. 553-557. Illinois, with a preliminary estimate of magnitude. US Geolo-
Munson, P.J., Munson, C,A., Bleuer, N.K. and Labitzke, M.D., gical Survey Professional Paper 1536, 27 pp.
1992. Distribution and dating of prehistoric liquefaction in Obermeier, S.F., Weems, R.E. and Jacobson, R.B., 1987. Earth-
the Wabash Valley of the Central US. Seismol. Res. Lett.. quake-induced liquefaction features in the coastal South
63(3): 337-342. Carolina region. US Geological Survey Open-File Report
Munson, P.J., Munson, C.A. and Pond, E.C., 1995. Paleolique- 87-504, 20 pp.
faction evidence for a strong Holocene earthquake in south- Ohsaki, Y., 1970. Effects of sand compaction on liquefaction
central Indiana. Geology, 23(4): 325 328. during the Tokachioki earthquake. Soils and Foundations,
Nataraja, M.S. and Gill, H.S., 1983. Ocean wave-induced lique- Japanese Soc. Soil Mech. Foundation Eng., Vol. 10, No. 2:
faction analysis. J. Geotech. Eng., Am. Soc. Civil Engineers, 112-128.
109(4): 573-590. Oldham, R.D., 1899. Report on the great earthquake of 12th
National Research Council, 1985. Liquefaction of soils during June 1897. Geological Survey of India Memoirs, Vol. 29:
earthquakes. National Academy Press, Washington, DC, 379.
240 pp. O'Rourke, T.D. and Hamada, M. (Editors), 1989. Proceedings
Nelson, A.R., 1992. Holocene tidal-marsh stratigraphy in from the Second US-Japan Workshop on Liquefaction,
south-central Oregon Evidence of localized sudden submer- Large Ground Deformation and Their Effects on Lifelines.
gence in the Cascadia subduction zone. In C.H. Fletcher, III State University of New York at Buffalo, National Center
and J.F. Wehrniller (Editors), Quaternary coasts of the for Earthquake Engineering Research, Technical Report
United States - Marine and lacustrine systems. SEPM (Soci- NCEER-89-0032, 498 pp.
ety for Sedimentary Geology) Special Publication 48, Owen, G., 1987. Deformation processes in unconsolidated
pp. 287 301. sands, in Jones, M.E., and Preston, M.F. (Editors), Defor-
Nettleton, W.D., Daniels, R.B. and McCracken, R.J., 1968a. mation of Sediments and Sedimentary Rocks. Geological
Two North Carolina Coastal Plain catenas ~I. Morphology Society (of London) Special Publication 29, pp. 11-24.
and fragipan development. Soil Sci. Soc. Am. Proc., 32: Pease, J.W. and O'Rourke, T.D., 1995. Liquefaction hazards
577--582. in the San Francisco Bay region: site investigation, modeling,
Nettleton, W.D., McCracken, R.J. and Daniels, R.B., 1968b. and hazard assessment at areas most seriously affected by
Two North Carolina Coastal Plain catenas II. Micromor- the 1989 Loma Prieta earthquake. Cornell University, School
phology, composition, and fragipan genesis. Soil Sci. Soc. of Civil and Environmental Engineering, Ithaca, NY,
Am, Proc., 32: 582-587. 176 pp.
Nuttli, O.W., 1979. Seismicity of the central United States. Penick, J., Jr., 1976. The New Madrid earthquakes of 1811-12.
Reviews in Engineering Geology, Geol. Soc. Am., IV: University of Missouri Press, Columbia, MO, 181 pp.
67-93. Peters, K.E. and Herrmann, R.B. (Compilers and Editors),
Obermeier, S.F., 1989. The New Madrid earthquakes; An engi- 1986. First-hand observations of the Charleston earthquake
neering-geologic interpretation of relict liquefaction features. of August 31, 1886, and other earthquake materials. South
US Geological Survey Professional Paper 1336-B, 114 pp. Carolina Geol. Survey Bull., 41:116.
Obermeier, S.F., 1995. Preliminary limits for the strength of Peterson, C.D., 1992. Variation in form and scale of paleolique-
shaking in the Columbia River valley and the southern half faction structures in late Pleistocene deposits of the central
of coastal Washington, with emphasis for a Cascadia subduc- Cascadia margin (abstr.). Geol. Soc. Am. Abstracts with
tion earthquake about 300 years ago. US Geological Survey Programs, 24(5): 74.
Open-File Report 94-589, 46 pp. Peterson, C,D., Hansen, M. and Jones, D., 1991. Widespread
Obermeier, S.F., Jacobson, R.B., Powars, D.S., Weems, R.E., evidence of paleoliquefaction in late-Pleistocene marine ter-
Hallbick, D.C., Gohn, G.S. and Markewich, H.W., 1986. races from the Oregon and Washington margins of the Cas-
Holocene and late Pleistocene(?) earthquake-induced sand cadia subduction zone (abstr.) Eos, 72(44): 313.
blows in coastal South Carolina: Proceedings of the Third Pettijohn, F.J. and Potter, P.E., 1964. Atlas and Glossary of
US National Conference on Earthquake Engineering, Primary Sedimentary Structures. Springer-Verlag, New
Charleston, South Carolina, August 24-28, 1986. Earth- York, 370 pp.
quake Eng. Res. Inst., E1 Cerrito, CA, Vol. 1: pp. 197-208. Pond, E.C., 1996, in press. Seismic parameters for eastern North
Obermeier, S.F., Jacobson, R.B., Smoot, J.P., Weems, R.E., America based on paleoliquefaction evidence in the Wabash
Gohn, G.S., Monroe, J.E. and Powars, D.S., 1990. Earth- Valley. Virginia Polytechnic Institute, Blacksburg, VA,
quake-induced liquefaction features in the coastal setting of Ph.D. thesis.
S.F Obermeier/Engineering Geology 44 (1996) 1-76 75

Reimnitz, E. and Marshall, N.F., 1965. Effects of the Alaska of Van Norman Lake, San Fernando, California. Science,
earthquake and tsunami on recent deltaic sediments. J. Geo- 182(4108): 161-163.
phys. Res., 70(10): 2363-2376. Sims, J.D., 1975. Determining earthquake recurrence intervals
Reineck, H.E. and Sing,h, I.B., 1980. Depositional sedimentary from deformational structures in young lacustrine sediments.
environments with reference to terrigenous elastics (2d edn.). Tectonophysics, 29: 141-t52.
Springer-Verlag, Berlin, 542 pp. Sims, J.D., 1978. Annotated bibliography of penecontempora-
Ringrose, P.S., 1989. Paleoseismic(?) liquefaction event in late neous deformational structures in sediments. US Geological
Quaternary lake sediments at Glen Roy, Scotland. Terra Survey Open-File Report 78-510, 158 pp.
Nova, 1(1): 57-62. Sims, J.D. and Garvin, C.D., 1995. Recurrent liquefaction
Ross, G.A., Seed, H.B. and Migliaccio, R.R., 1969. Bridge induced by the 1989 Loma Prieta earthquake and 1990 and
foundation behavior in Alaska earthquake. J. Soil Mech. 1991 aftershocks; Implications for paleoseismicity studies.
Found. Eng. Div., Am. Soc. Civil Engineers, 95(SM4): Seismol. Soc. Am. Bull., 85(1): 51-65.
1007-1036. Singh, S., 1994. Liquefaction characteristics of silts. In S. Pra-
Russ, D.P., 1979. Late Holocene faulting and earthquake recur- kash and P. Dakoulas (Editors), Ground failures under
rence in the Reelfoot Lake area, northwestern Tennessee. seismic conditions. Am. Soc. Civil Engineers Geotech.
Geol. Soc. Am. Bull., 90(11): 1013-1018. Special Publication 44:105-116.
Saucier, R.T., 1989. Evidence for episodic sand-blow activity Springer, F.M. Jr., UUrich, C.R. and Hagerty, D.J., 1985.
during the 1811-12 New Madrid (Missouri) earthquake Streambank stability. J. Geotech. Eng., Am. Soc. Civil Engi-
series. Geology, 17(2): 103-106. neers, 111(5): 624-640.
Saucier, R.T., 1991a. Comment on "Formation of Mima Steele, F., Daniels, R.B., Gamble, E.E. and Nelson, L.A., 1969.
mounds; a seismic hypothesis". Geology, 19(3): 284. Fragipan horizons and Be masses in the middle coastal plain
Saucier, R.T., 1991b. Geoarchaeological evidence of strong pre- of north central North Carolina. Soil Sci. Soc. Am., Proc.,
historic earthquakes in the New Madrid (Missouri) seismic 33: 752-755.
zone. Geology, 19(4): 296-298. Stone, J.R. and Ashley, G.M., 1992. Ice-wedge casts, pingo
Savage, J.C. and Lisowski, M., 1991. Strain measurements and scars, and the drainage of glacial Lake Hitehcock. In P.
the potential for a great subduction zone earthquake off the Robinson and J.B. Brady (Editors), Guidebook for Field
coast of Washington. Science, 252: 101-103. Trips in the Connecticut Valley Region of Massachusetts and
Scott, R.F., 1986. Solidification and consolidation of a liquefied Adjacent States, Vol. 2. University of Massachusetts,
sand column. Soils Found. Jpn. Soc. Soil Mech. Found. Eng., Department of Geology and Geography, Contribution 66,
26(4): 23-31. pp. 305-331.
Scott, R.F. and Zuekerman, K.A., 1973. Sand blows and lique- Svensson, H., 1988. Ice-wedge casts and relict polygonal pat-
faction, in The Great Alaska Earthquake of 1964. National terns in Scandinavia. J. Quaternary Sci., 3(1): 57-67.
Academy of Sciences, Washington, DC, pp. 179-189. Swenson, F.A., 1964. Ground-water phenomena associated
Seed, H.B., 1968. Landslides during earthquakes due to soil with the Hebgen Lake earthquake. US Geological Survey
liquefaction. J. Soil Mech. Found. Eng. Div., Am. Soc. Civil Professional Paper 435-N, pp. 159-165.
Engineers, 94, SM5: 1055-1122. Taber, S., 1914. Seismic activity in the Atlantic Coastal Plain
Seed, H.B., 1979. Soil liquefaction and cyclic mobility evalua- near Charleston, S.C. Seismol. Soc. Am. Bull., 4: 108-160.
tion for level ground during earthquakes. J. Geotech. Eng., Terzaghi, K. and Peck, R.B., 1967. Soil Mechanics in Engineer-
Am. Soc. Civil Engineers, 105, GT2: 201-255. ing Practice (2d edn.). New York, John Wiley, 729 pp.
Seed, H.B. and Idriss, I.M., 1971. A simplified procedure for Tokimatsu, K. and Seed, H.B., 1987. Evaluation of settlements
evaluating soil liquefaction potential. J. Soil Mech. Found. in sands due to earthquake shaking. J. Geotech. Eng., Am.
Eng. Div., Am. Soc. Civil Engineers, 97, SM9: 1249-1274. Soc. Civil Engineers, 113(8): 861-878.
Seed, H.B., Idriss, I.M. and Arango, I., 1983. Evaluation of Tsuchida, H. and Hayashi, S., 1971. Estimation of liquefaction
liquefaction potential using field performance data. potential of sandy soils. Proceedings of the Third Joint Meet-
J Geotech. Eng., Am. Soc. Civil Engineers, 109(3): 458-482. ing, US-Japan Panel on Wind and Seismic Effects, UJNR,
Seed, H.B., Tokimatsu, K., Harder, L.F. and Chung, R.M., Tokyo, May 1971, pp. 91-109.
1985. Influence of SPT procedures in soil liquefaction resis- Tuttle, M.P., 1994. The liquefaction method for assessing paleo-
tance evaluations. J. Geotech. Eng., Am. Soc. Civil Engi- seismicity. US Nuclear Regulatory Commission Report
neers, 111(12): 1425-1445. NUREG/CR-6258, 38 pp.
Seilacher, A., 1969. Fault-graded beds interpreted as seismites. Tuttle, M.P., Cowie, P. and Wolf, L., 1992. Liquefaction
Sedimentology, 13: 155-159. induced by modern earthquakes as a key to paleoseismicity;
Shllts, W.W., 1978. Nature and genesis of mudboils, central A case study of the 1988 Saguenay event. In A.J. Weiss
Keewatin, Canada. Can. J. Earth Sci., 15(7): 1053-1068. (Comp.), Proceedings of the Nineteenth Water Reactor
Sieh, K.E., 1978. Prehistoric large earthquakes produced by slip Safety Information Meeting. US Nuclear Regulatory Com-
on the San Andreas fault at PaUett Creek, California. J. Geo- mission Report NUREG/CP-0119, Vol. 3, pp. 437-462.
phys. Res., 83(B8): 3907-3939. Tuttle, M.P. and Schweig, E.C., 1995. Archeological and pedo-
Sims, J.D., 1973. Earthquake-induced structures in sediments logical evidence for large prehistoric earthquakes in the New
76 S.F. Obermeier/Engineering Geology 44 (1996) 1-76

Madrid seismic zone, central United States. Geology, Youd, T.L., 1978. Major cause of earthquake damage in ground
23(3): 253-256. failure. Civil Eng., 48(4): 47-51.
Tuttle, M.P. and Seeber, L., 1991. Historic and prehistoric Youd, T.L., 1984a. Geologic effects Liquefaction and associ-
earthquake-induced liquefaction in Newbury, Massachu- ated ground failure. In: Williams, M.E., Proceedings of the
setts. Geology, 19(6): 594-597. Geological and Hydrologic Hazards Training Program. US
Updike, R.G., Egan, J.A., Moriwaki, Y., Idriss, I.M. and Geological Survey Open-File Report 84-760, pp. 210-232.
Moses, T.L., 1988. A model for earthquake-induced transla- Youd, T.L., 1984b. Recurrence of liquefaction at the same site.
Proceedings of the Eighth World Conference on Earthquake
tory landslides in Quaternary sediments. Geol. Soc. Am.
Engineering, Vol. 3: pp. 231-238.
Bull., 100(5): 783-792.
Youd, T.L., 1996a. in press, Liquefaction criteria based on
Valera, J.E., Traubenik, M.L., Egan, J.A. and Kaneshiro, J.Y..
probabilistic analyses of cyclic stress ratio and (N1)6o data.
1994. A practical perspective on liquefaction of gravels. In In: T.L. Youd (Editor), NCEER Workshop on Evaluation
S. Prakash and P. Dakoulas (Editors), Ground Failures of Liquefaction Resistance, January, 1996. State University
Under Seismic Conditions. Am. Soc. Civil Engineers of New York at Buffalo, National Center for Earthquake
Geotech. Special Publication 44: pp. 241-257. Engineering Research.
van Loon, A.J., 1992. The recognition of soft-sediment deform- Youd, T.L., 1996b in press, Magnitude scaling factors. In T.L.
ations as early-diagenetic features - A literature review. In Youd (Editor), NCEER Workshop on Evaluation of Lique-
K.H. Wolf and G.V. Chilingarian (Editors), Diagenesis III. faction Resistance, January, 1996. State University of New
Elsevier, New York, pp. 135-189. York at New York, National Center for Earthquake Engi-
van Loon, A.J. and Brodzikowski, K., 1987. Problems and pro- neering Research.
gress in the research on soft-sediment deformations. Sedi- Youd, T.L. and Bartlett, S.F., 1991. Case histories of lateral
mentary Geol., 50: 167-193. spreads from the 1964 Alaska earthquake. In T.D. O'Rourke
Van Vliet-Lanoe, B., 1988, The significance of cryoturbation and M. Hamada (Editors), Proceedings from the Third
phenomena in environmental reconstruction: J. Quaternary Japan-US Workshop on Earthquake Resistant Design of
Lifeline Facilities and Countermeasures for Soil Liquefac-
Sci., 3(I): 85-96.
tion. State University of New York at Buffalo, National
Vandenberghe, J., 1988. Cryoturbations. In M.J. Clark
Center for Earthquake Engineering Research, Technical
(Editor), Advances in Perigiacial Geomorphology. John Report NCEER-91-0001, pp. 175-189.
Wiley and Sons, New York, pp. 179-198. Youd, T.L. and Garris, C.T., 1995. Liquefaction-induced
Walsh, T.J., Combellick, R.A. and Black, G.L., 1995. Liquefac- ground surface disruption. J. Geotech. Eng., Am. Soc. Civil
tion features from a subduction earthquake; Preserved exam- Engineers, 121 ( 11 ): 805-809.
ples from the 1964 Alaska earthquake. Washington Division Youd, T.L., Holzer, T.L. and Bennett, M.J., 1989. Liquefaction
of Geology and Earth Resources, Report of Investigations lessons learned from Imperial Valley, California. Twelfth
32, 80 pp. International Conference on Soil Mechanics and Foundation
Wang, Zhong-qi, Fang, Hong-qi and Zhao, Shu-dong, 1983. Engineering, Proceedings of Discussion Session on Influence
Macroscopic features of earthquake induced soil liquefaction of Local Conditions on Seismic Response, Rio de Janeiro,
and its influence on ground damage. Can. Geotech. J., pp. 47-54.
20(1): 61-68. Youd, T.L. and Hoose, S.N., 1978. Historic ground failures in
Wesnousky, S.G. and Letiter, L.M., 1992. The repeat time of northern California triggered by earthquakes. US Geological
the 1811 and 1812 New Madrid earthquakes; a geological Survey Professional Paper 993, 177 pp.
Youd, T.L. and Keefer, D.K., 1994. Liquefaction during the
perspective. Seismol. Soc. Am. Bull., 82(4): 1756-1784.
1977 San Jose Province earthquake (M~ 7.4). Eng. Geol.,
Wong, R.T., Seed, H.B. and Chan, C.K., 1975. Cyclic loading 37:211-233.
liquefaction of gravelly soils. J. Geotech. Eng., Am. Soc. Youd, T.L. and Perkins, D.M., 1978. Mapping liquefaction-
Civil Engineers, 101(GT6): 571-582. induced ground failure potential. J. Geotech. Eng., Am. Soc.
Wood, H.O. and Neumann, F., 1931. Modified Mercalli inten- Civil Engineers, 104(GT4): 433-446.
sity scale of 1931. Seismolog. Soc. Am. Bull., 21: 277-283. Youd, T.L. and Perkins, D.M., 1987. Mapping of liquefaction
Youd, T.L., 1973. Liquefaction, flow, and associated ground severity index. J. Geotech. Eng., Am. Soc. Civil Engineers,
failure. US Geological Survey Circular 688, 12 pp. 113(11): 1374-1392.

You might also like