Вы находитесь на странице: 1из 34

RECENT DEVELOPMENTS IN THE UNDERSTANDING OF

EARTHQUAKE SITE RESPONSE AND ASSOCIATED SEISMIC


CODE IMPLEMENTATION

Ricardo Dobry
1
and Susumu Iai
2




ABSTRACT

The paper focuses on the compelling evidence on site amplification of earthquake motions accumulated
in the last fifteen years or so, which resulted in recent changes in the way site coefficients are defined by
seismic building codes in the USA. Several studies are reviewed, where basic considerations and theory are
compared with recordings at soft clay sites and other soils during the 1985 Mexico City and 1989 Loma
Prieta, California earthquakes. These and other studies were supplemented with laboratory results and
analyses to account for the effect of soil nonlinearity on the site coefficients at higher levels of rock shaking.
A summary is provided of the new national U.S. seismic provisions for buildings contained in the 1994 and
1997 NEHRP and 1997 UBC. In these new provisions, two site amplification coefficients are defined at
short and long periods respectively (F
a
and F
v
); both coefficients decrease due to soil nonlinearity as the level
of seismic hazard on rock increases; and a representative average shear wave velocity of the top 30m of soil
or rock (V
s
) is used to characterize the site. Special amplification effects associated with large cyclic
straining and liquefaction of the soil are reviewed, as well as continuing efforts to re-evaluate the degree of
nonlinearity of the current site coefficients and the significance of other effects using the growing pool of
earthquake records generated after 1994. The use of microzonation of urban areas to supplement generic
provisions on site coefficients contained in national seismic codes is discussed.


1.0 INTRODUCTION

Soil effects are a very important contributor to human suffering and damage during earthquakes.
There are mainly two types of ground response that give us trouble when the ground starts to shake. In one,
the soil fails (typically by liquefaction), cracks and moves laterally and horizontally; this was an extremely
significant cause of economic damage to ports and other facilities during the 1995 Kobe earthquake in Japan.
In the other, the soil changes the character of the ground shaking by amplifying it and making it more
destructive; this was a main cause of the loss of lives and damage to buildings and bridges in the 1989 Loma
Prieta earthquake in California. While in some cases only one of these two types of soil response is present
(e.g., the 1985 Mexico City earthquake, where amplification of the earthquake waves by the soft Mexico
City clay was the cause of essentially all several thousand deaths and building damage), very often both
liquefaction and soil amplification effects are present in the same earthquake. This was the case in the 1989
Loma Prieta and 1995 Kobe earthquakes, and also in the 1994 Northridge event in southern California. In
fact, the extensive damage in the Marina District of San Francisco in the 1989 earthquake was due to a
combination of amplification and liquefaction. Other major damaging earthquakes in the last 15 years of the
20
th
century where soil effects played a significant role include: 1986 Kalamata, Greece; 1988 Armenia;
1990 Phillippines; 1991 Limon, Costa Rica; and the 1999 Duzce, Turkey and Chichi, Taiwan earthquakes.
This paper focuses on the amplification effect, and especially on recent changes in the way site
amplification factors are defined by seismic building codes in the USA. The effect of local soil conditions
was recognized relatively late in the U.S., especially compared with Japan where soil amplification had been
accepted as a significant effect for many years. While a site coefficient S depending on local soil conditions
was incorporated in the equations to calculate seismic forces for building design in the U.S. starting in 1978
(ATC, 1978), the values of S were relatively low, they amplified only the long period part of the spectrum,

1
Department of Civil Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180-3590, USA
2
Port and Harbour Research Institute, Nagase 3-1-1, Yokosuka 239-0826, Japan
and they were independent of the level of shaking. Accumulation of laboratory, analytical, and instrumental
evidence over the years, and especially the dramatic amplification on soft soils recorded during the 1989
Loma Prieta earthquake - in conjunction with the recognition that earthquakes are a very significant potential
hazard in many areas of the country in addition to California and the Western U.S. - prompted a thorough
review of the assumptions underlying the old S factors. This included a number of new studies, and
culminated with a complete revamping of the site characterization and site coefficient system. The new site
characterization and site coefficients have already been incorporated into the 1994 edition of the seismic
provisions of the National Earthquake Hazard Reduction Program (NEHRP, 1994) and of the 1997 Uniform
Building Code (UBC, 1997). The paper presents some of the basic evidence and results of studies that
preceded these code changes, as well as some that have taken place since 1994, describes the new seismic
provisions, discusses effects which are not quite covered by the new provisions as well as ongoing studies,
and ends with some thoughts on the relation between seismic site provisions and microzonation.

2.0 BASIC CONSIDERATIONS AND RESPONSE OF SOFT CLAY SITES

The Magnitude 7.1 Loma Prieta earthquake of October 17, 1989 was the largest to occur in the San
Francisco Bay Area of California since the great earthquake of 1906 (Housner, 1990). Figure 1 includes a
map of the area. Most of the damage due to both amplification (by soft clays and other soils) and
liquefaction occurred in and around San
Francisco and Oakland, 100km NW of the
epicenter. Much of the recorded instrumental
evidence on amplification was also obtained,
either in those cities or between them and the
epicenter. This is clearly a very important area
in the U.S., containing several cities as well as
the Silicon Valley, which covers the strip along
the Bay passing by San Jose, Palo Alto and San
Francisco. The level of rock shaking in San
Francisco, Oakland and Berkeley was smaller
than that expected during a probable large
magnitude event along the San Andreas Fault,
and thus was not a direct test of the
corresponding building seismic code provisions
for California. However, this level was
comparable to those used in seismic codes in
other regions of lower seismic hazard in the
USA, such as New York and other parts of the
Eastern U.S. Therefore, some of the lessons on
local site amplification from the Loma Prieta
earthquake are directly applicable to those areas
of lower seismic hazard.
Figures 2 to 6 illustrate the
amplification of elastic response spectra
observed at soft clay sites during the Loma
Prieta earthquake. Figure 2 presents the soil
profile at one of the instrumented soft clay sites,
about 70km north of the epicenter, while Fig. 3
includes response spectra recorded at both this
site and at a rock site nearby. The rock
spectrum was amplified by the soil by a factor
of at least two at most periods (RRS greater than 2) and as much as five times at a period around 0.6 seconds
(Fig. 4). Figure 4 defines the two very important concepts of Ratio of Response Spectra, RRS (soil spectrum
divided by the rock spectrum at the same value of period T), and of RRS
max
. In the case of Fig. 4, RRS
max
=
4.9 at T = 0.64 sec. Note that this period is similar and slightly smaller than the fundamental period of the
top soft clay layer in Fig. 2, (4) (18) / 90 = 0.80 sec.


Figure 1 : San Francisco Bay Area in California and
the 1989 Loma Prieta earthquake







Figure 5 presents average horizontal elastic
response spectra recorded on soft clay and rock sites in
San Francisco and Oakland during the Loma Prieta
earthquake, including the records of Fig. 3. Again,
while in general the peak acceleration on rock was
about 0.08 or 0.1g , it was amplified two to three times
to 0.2g or 0.3g at the soft soil sites; the spectral
ordinates at short periods (T 0.2 or 0.3 seconds) were
also amplified on average by factors of 2 or 3. Figure
6 shows the result of a statistical analysis of this
amplification of response spectra between nearby rock
and soil sites, that is of the Ratio of Response Spectra,
versus period, for the soft sites in this earthquake
(Joyner et al., 1994). Figures 5 and 6, as well as Fig. 4,
show that at longer periods between 0.5 and 1.5 or 2
seconds the amplification is even greater than at short
periods, with RRS now ranging from three to six times. A similar amplification behavior but with lower
values of RRS was observed between stiff soil sites and rock sites (Housner, 1990, Chang, 1991, Joyner et
al., 1994). As discussed before for Fig. 2, at some of the soft sites the maximum value of RRS, RRS
max
,
occurred at a period seemingly related to the characteristics of the soil deposit.



Figure 2 : Dynamic soil profile at soft clay site in
Redwood Shores, south of San Francisco
(Dobry, 1991).

Figure 3 : Recorded response spectra at soft clay
site and nearby rock site south of San
Francisco, 1989 Loma Prieta earthquake
(Chang, 1991, Dobry, 1991).



Figure 4 : Soil/rock Ratio of Response Spectra versus
period for acceleration spectra of Fig. 3.

Figure 5: Average spectra recorded during 1989 Loma Prieta earthquake in San Francisco Bay area at
rock sites and soft soil sites (modified after Housner, 1990).


Figure 6: Calculation of average RRS curves for 5 percent damping from records of 1989 Loma Prieta
earthquake on soft soil sites. The middle curve gives the geometric average ratio as function of the period.
The top and bottom curves show the range from plus to minus one standard deviation of the average of the
logarithms of the ratios. The vertical lines show the range from plus to minus standard deviation of the
logarithms of the ratios (Joyner et al., 1994).
An even more extreme case of this type of amplification has been observed in the very soft clay deposits
of Mexico City at periods of the order of 2 or 3 seconds, with RRS
max
being as much as 10 or 20. Figure 7
shows the response spectra recorded in Mexico City during the very destructive 1985 earthquake, on rock
and on different depths of soil from about 40 to 60m. Clearly, again, the motions on soil were much larger
than on rock, with the maximum amplification of the response spectrum occurring at the fundamental period
of the soil profile, which at these three soil sites ranges from about 2 to 4 seconds. Whatever the period of
the site, the spectrum was amplified by about ten times or more. The main reason the peak of the response
spectrum recorded at the SCT site is higher than at the CAO site is because the spectrum on rock is very
small in the 3 to 4 sec period range, and thus at CAO there was not much to amplify. Figure 8 is a graph of
RRS
max
versus the average shear wave velocity of the soil, again for Mexico City. Each data point in Fig. 8
corresponds to one instrumented clay site in the city, and the RRS
max
plotted is an average of the RRS
max
of
the NS and EW components of three to five earthquakes including the 1985 event depicted in Fig. 7. The
value of RRS
max
in Fig. 8 ranges between about 3 and 20, with a clear trend for RRS
max
to decrease as the
shear wave velocity increases.
Fortunately, as discussed later herein, both the extreme softness of the soil as well as other characteristics
of the Mexico City clay inducing these very high amplifications are rather unusual.


Figure 7 : Soil conditions and recorded acceleration spectra for several sites in Mexico City during the
1985 earthquake (modified after Seed et al., 1988).

Figure 8 : Maximum Ratio of Response Spectra, (RRS)
max
, versus V
s
, recorded at Mexico City clay sites
during five earthquakes between 1985 and 1990 (Ordaz and Arciniegas, 1992, Dobry et al., 1999).
Useful insight into the factors controlling the value of RRS
max
and the amplification phenomenon at soft
clay sites revealed by Figs. 1-8 is provided by the 1D amplification model of Fig. 9 and the results in Fig. 10
(Roesset, 1977). Both the horizontal acceleration on rock at point B, a
B
, and the corresponding soil
acceleration at point A, a
A
, are caused by vertically propagating, sinusoidal shear waves of frequency f (cps).
Therefore, both a
A
and a
B
are amplitudes of sinusoidal accelerograms of frequency f. The amplification ratio
a
A
/ a
B
is a function of the ratio of frequencies f / (V
s
/ 4h), of the soil material damping ratio
s
, and of the
rock/soil impedance ratio, I =
r
V
r
/
s
V
s
. In Fig. 9,
r
,
s
are the total unit weights and V
r
, V
s
are the shear
wave velocities of rock and soil. The maximum amplification (a
A
/ a
B
)
max
corresponding to resonance in
shear of the soil layer, occurs at a frequency near V
s
/ 4h, where V
s
/ 4h is the undamped fundamental
frequency of the layer. This maximum amplification is approximately equal to:


a
A
a
B



1
]
1
max

1
1/ I ( )+ / 2 ( )
s
(1)


Figure 9 : Uniform soil layer on elastic rock subjected to vertical shear wave.

Figure 10 : Amplification ratios soil/rock for h = 100 ft (30.5m), V
s
/ 4h = 1.88 cps, I = 6.7 (Roesset, 1977).
In Fig. 10, V
s
/ 4h = 1.88 cps, I = 6.7, and for
s
= 0.05, (a
A
/ a
B
)
max
4.4. A plot such as Fig. 10 provides
the transfer function of the site, which is constant and independent of the rock motion for the assumed linear
soil and vertically propagating shear waves. While the transfer function of a layered soil profile such as that
of Fig. 2 may be more complex than Fig. 10, every soil profile on stiffer rock has a transfer function which is
constant and independent of the rock motion, under the assumed linear soil and vertically propagating shear
waves. The transfer function can be properly estimated by dividing Fourier Spectra and obtaining the Ratio
of Fourier Spectra of recorded horizontal accelerations on nearby soil and rock sites. On the other hand,
plots of Ratio of Response Spectra, RRS, such as discussed before and plotted in Figs. 4 and 6 for the Loma
Prieta earthquake records, or of the maximum value of RRS, RRS
max
such as plotted in Fig. 8 for Mexico
City, are more appropriate for engineering evaluations involving response spectra and for determination of
site coefficients in seismic codes. However, RRS and RRS
max
are not independent of the rock motion.
Fortunately, analyses and comparisons at actual soft clay sites on much stiffer rock or soil, suggest that: (i)
both RRS
max
and (a
A
/ a
B
)
max
occur at about the same frequency, (ii) Equation 1 often predicts reasonably well
the value of RRS
max
(Dobry, 1991), and (iii) average values of RRS and (a
A
/ a
B
) for the same period range
are within 30% of each other (Joyner et al., 1994).
Therefore, based on Eq. 1 it could be expected that the value of RRS
max
at soft clay sites is controlled by
two main factors: the impedance ratio I = (
r
/
s
) ( V
r
/ V
s
) = (
r
/
s
) ( V
r
/ V
s
) and the internal damping
s

of the soil. (In the previous expression for I, is the mass density of the soil and is its total unit weight,
with = g and g = acceleration due to gravity.). Some main reasons why the Mexico City clay exhibits
unusually high amplifications are the unusually low values of V
s
,
s
and
s
of this soil, in addition to its
extremely linear stress-strain response in shear. For most soft clay sites such as those typically encountered
in the U.S., the values of
s
and
s
are higher (Dobry, 1991); in fact the ratio
r
/
s
1.1 to 1.4 is quite
constant for most sites, while
s
depends mostly on the intensity of the rock motions (due to soil
nonlinearity) and on the plasticity index of the clay (Vucetic and Dobry, 1991). Thus, for a given type of
rock outcrop, I is about proportional to 1/V
s
, and for a relatively narrow range of plasticity indices, it should
be expected that RRS
max
at a site would depend mainly on V
s
and on the intensity of the rock motions. This
use of Eq. 1 also predicts that for a given clay deposit of very high plasticity and low soil nonlinearity (and
correspondingly small value of
s
), RRS
max
should depend mainly on V
s
. This has been confirmed by the
Mexico City experience, as shown by Fig. 8.
Let us go back to Eq. 1. It gives the maximum amplification, (a
A
/

a
B
)
max
, as a function of the impedance
ratio I between the rock and the soil above it, and of the damping ratio of the soil,
s
. As discussed before,
in first approximation and for many practical applications, RRS
max
(a
A
/

a
B
)
max
. Therefore, Eq. 1 can be
used as simplified analytical framework for an approximate prediction of RRS
max
at clay sites on different
cities, for earthquake excitation having a peak rock acceleration of the order of 0.1g. This is done in Fig. 11,
where the curves are the graphical representation of Eq. 1, and the data points correspond to soft clay sites in
San Francisco, Mexico City, Boston and New York (Dobry, 1991). When applying the crude model of Eq. 1
to predicting (a
A
/

a
B
)
max
, and thus also RRS
max
for these
sites, the impedance ratio I was calculated between the
soft clay and the rock, neglecting the effect of the
intermediate soil layers; and the shear wave velocity
used for the rock corresponds to a measured or
estimated value for the top 20 or 30 m below the soil-
rock interface. Other simplifications introduced and
possible complications implicit in this kind of
prediction are discussed by Dobry (1991). Despite the
simplicity of the model, Eq. 1 and Fig. 11 predict
reasonably well RRS
max
in the two cases for which
instrumental information exists from two destructive
earthquakes: Mexico City 1985 and San Francisco
(Loma Prieta) 1989. For San Francisco, RRS
max
= 4 is
predicted in Fig. 11, which compares well with the
range of RRS
max
3 to 6 recorded in 1989 (Figs. 4-5);
and for Mexico City, RRS
max
= 9 is predicted in Fig. 11,
which is within the range of RRS
max
8 to 20 recorded
both in 1985 and in subsequent earthquakes at the
softest clay sites in that city (Figs. 7-8). The prediction
of high values of RRS
max
for soft soil sites in Boston
and New York City, for which no records are available,
including an amplification almost as high as Mexico
City at a site in New York City (Flushing Meadows), is
related to the fact that in those two Eastern U.S. cities
the rock underlying the soft soil is much harder than
either in San Francisco or Mexico City, and thus the
value of I in Eq. 1 is high.


Figure 11 : Maximum amplification computed
with Eq. 1 (lines) and predicted
values of maximum amplification
at clay sites in four cities (data
points) (Dobry, 1991).
3.0 RESPONSE OF OTHER SOILS AND DEVELOPMENT OF SITE COEFFICIENTS FOR
SEISMIC CODES

The theoretical framework and instrumental observations of RRS and RRS
max
just described, are a good
starting point for the development of scientifically sound and realistic site coefficients for seismic codes.
Seismic code provisions based directly on the theoretical framework discussed around Eq. 1 and Figs. 9-11,
that is on calculations of the site period and of RSS
max
, may be appropriate for a specific area consisting
mainly of soft clays of known depth on much stiffer soil or rock, and for expected earthquakes of long
duration which induce soil resonance without much soil nonlinearity. This is the case of Mexico City
(Reinoso and Ordaz, 1999) and the Mexico City seismic code (Simn and Suarez, 1994). However, national
seismic codes including those in the USA typically must consider a much wider variety of site conditions and
earthquake motions of various levels of intensity, durations and frequency contents, for which a direct
application of the model of Fig. 9 either is not relevant or is impractical. In many soft clay sites there is
often the complication of stiffer soil layers between the soft clay and the rock (Fig. 2). Also, in most soft
clay sites the values of site period and RRS
max
will typically depend on the earthquake. Therefore, a more
practical approach for the evaluation of site coefficients for seismic codes is the calculation of an average (or
average plus one standard deviation) value of RRS at a given period or range of periods for a number of
stations having generally similar soil conditions, as done by Joyner et al. (1994) for the Loma Prieta
earthquake (Fig. 6). Similar considerations are also valid if Ratios of Fourier Spectra are used to measure
amplification. Based on the assumption that the energy of the wave is preserved, Joyner et al. (1981) and
Aki (1988) have suggested that site amplification at a given period should be approximately proportional to
(V
s
)
-0.5
, where V
s
is again the shear wave velocity of the soil at shallow depth.
Empirical calculations done by Roger Borcherdt at the U.S. Geological Survey, show that average
amplification factors calculated using Ratios of Fourier Spectra between soil sites (including both clays and
other soil types) and nearby rock sites, are proportional to the mean shear wave velocity of the top 30 m, V
s,
raised to a (typically negative) exponent; with this exponent depending on the period band and on the
amplitude of the rock ground motion (Borcherdt, 1994a). These results are consistent with those derived
from other data sets by Borcherdt and Glassmoyer (1994), Midorikawa et al. (1994) and Boore et al. (1994,
1997), using either RRS or Ratio of Fourier Spectra. For relatively low amplitude rock motions, 0.1g and
less, these results by Borcherdt show that the corresponding amplification factors are approximately
proportional to (V
s
)
-0.4
for short periods, and to (V
s
)
-0.6
for periods about 1 second and longer.
Corresponding regression curves fit to the average Ratio of Fourier Spectra computed for a short-period band
(0.1 - 0.5 sec) and for a longer-period band (0.4 - 2.0 sec) are shown in Figure 12 for the 1989 Loma Prieta
earthquake. Figure 12 indicates values of short - and long-period amplification factors, F
a
and F
v
, decreasing
toward 1.0 as V
s


approaches a value of about 1000 m/sec or greater corresponding to the reference rock
sites. A plot such as Fig. 12 allows defining average values of Ratio of Fourier Spectra or RRS at both short
(F
a
) and long (F
v
) periods for many soil conditions ranging from soft clay to rock sites, thus extending the
framework developed before mostly for soft clay sites, and providing the basis for defining site amplification
coefficients such as F
a
and F
v
to be used in seismic codes.

4.0 NONLINEAR SOIL EFFECTS

As illustrated by Figs. 8, 11 and 12, for relatively small levels of rock shaking (which after
amplification may not be small at all, and in fact may be quite destructive on soft soil), the most important
soil property controlling the amount of site amplification is the shear wave velocity of the soil, V
s
. Small
nonlinear effects are manifested mostly through the value of the soil damping ratio, which is
s
= 0.05 to
0.10 in most clays, being usually small (
s
= 0.03) in Mexico City (Fig. 11). This significance of V
s
is
shown by the instrumental evidence and is explained theoretically by the key role of V
s
in defining the
impedance ratio soil/rock of the site, I (e.g., see Eq. 1). As discussed later, this fact is at the basis of the site
classification scheme included in the new generation of seismic building codes in the U.S., where the
average shear wave velocity of shallow soil layers is used to define the site amplification factors.



Figure 12 : Short-period F
a
and mid-period (termed long-period in code provisions) F
v
amplification factors
with respect to firm to hard rock (B) plotted as a continuous function of mean shear-wave velocity,
s
V ,using
the regression equations derived from the Loma Prieta earthquake from Ratios of Fourier Spectra of nearby
soil and rock records. The 95% confidence intervals for the ordinate to the true population regression line
and the limits for +2 standard error of estimate are shown. Corresponding amplification factors predicted for
the simplified site classes with respect to firm to hard rock also are shown (Borcherdt, 1994b).


However, as the level of rock shaking
increases, a more intense nonlinear
stress-strain behavior develops in the
soil, that generally tends to decrease both
RRS and RRS
max
. The situation now
becomes more complicated, as both V
s

and the nonlinear soil properties must be
considered. The field evidence on site
nonlinearity from earthquake records is
much scarcer than that on the influence
of V
s
discussed in the previous section;
some of this field evidence is discussed
later in this section. However, much is
known about the relevant cyclic
nonlinear stress-strain soil response from
cyclic shear laboratory tests using a
variety of experimental techniques (e.g.,
cyclic simple shear, cyclic triaxial and
cyclic torsional tests). Figures 13-14
summarize some of this laboratory work,
while Figs. 15-16 present results of one-
dimensional equivalent linear site
response analyses using program
SHAKE (Schnabel et al., 1972), which used laboratory results to predict RRS and RRS
max
of soft clays at
higher levels of rock shaking.



Figure 13 : Typical shape of backbone curve and of hysteresis
loop for soil subjected to cyclic shear loading
0
1
2
3
4
5
6
7
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400
Mean Shear-Wave Velocity to 30 m (100 ft)
m/s) , V (
s
A
v
e
r
a
g
e

S
p
e
c
t
r
a
l

A
m
p
l
i
f
i
c
a
t
i
o
n

F
a
c
t
o
r
s
,
F
a
a
n
d
F
v
Loma Prieta Strong-Motion Data
Fa - 95%
Fa = (997 /
s
V ) ^0.36
Fa +95%
Fv - 95%
Fv = (1067 /
s
V ) ^0.64
Fv + 95%
Fa for Site Class Intervals
Fv for Site Class Intervals
E
D
C
B
Firm to Hard rocks
Gravelly soils and Soft rocks
Stiff clays and
Sandy soils
Soft soils



Figure 14 : Influence of plasticity index, PI, on curves of G/G
mas
and
s
versus strain for soils (Vucetic
and Dobry, 1991).

Figure 13 sketches a typical laboratory stress-strain hysteresis loop developed in the soil when subjected
to cyclic shear loading similar to that existing in the field during vertical shear wave propagation such as
depicted in Fig. 9 (Seed and Idriss, 1970, Hardin and Drnevich, 1972a,b). The tips of the loop are located on
the backbone curve, which in first approximation is the same as the monotonic stress-strain curve of the soil
loaded at a comparable strain rate. The tangent shear modulus at the origin is G
max
, also called "small-strain
shear modulus," which in first approximation is related to the shear wave velocity, V
s
, by the basic
expression, G
max
= V
s
2
, where = mass density of soil. Therefore, G
max
serves to anchor the backbone
curve and hysteresis loops in the laboratory, and to link the laboratory results to V
s
measured using
geophysical methods in the field (e.g., Hoar and Stokoe, 1978, Woods and Stokoe, 1985, Stokoe et al.,
1994). There are two main aspects of the loop that are most important. One is the evolution of the secant
modulus G with cyclic strain
c
, which is an indication of how nonlinear is the backbone curve. The other is
the value of the area W inside the loop, representing the energy dissipated per cycle, which is used to define
the internal or hysteretic damping ratio of the soil,
s
= W / (2G
c
2
). This is the damping ratio
s
included
in Eq. 1 and Fig. 11 that influences directly the site amplification. As the cyclic strain strain
c

increases, G
decreases and
s
increases. This is shown by Fig. 14, which is a summary of cyclic laboratory results from
many soils, and which shows that clays of high plasticity (PI = 200), such as those of Mexico City, exhibit
small nonlinearity (high values of G/G
max
and small values of
s
), while nonplastic clays, silts and sands (PI
= 0) have high stress-strain nonlinearity. Many of the soft clays in San Francisco which produced the high
spectral amplifications recorded during the Loma Prieta earthquake (Figs. 2-6) have PI 50. During the
1985 Mexico City and 1989 Loma Prieta earthquakes, it is estimated that the maximum values of cyclic
shear strain developed in the clay layers at both sites, were of the order of
c
0.1%. The analytically
predicted influence of nonlinearity on site amplification depends in a complicated manner on a number of
factors including: level, frequency content and duration of the rock motions, stratigraphy of the site including
impedance ratio with the rock below, and especially the shear wave velocity V
s
and the G/G
max
and
s
versus

c
curves of the soft layer. However, generally speaking site amplification becomes smaller as
c
increases,
and for a given
c
amplification is smaller when PI decreases (more nonlinear soil).
This is illustrated by Figs. 15 and 16. In Fig. 15, Idriss (1990, 1991) conducted site response calculations
at soft clay sites to determine the amplification of peak acceleration (that is, of RRS at period, T = 0 sec).
For low rock accelerations of the order of 0.05g to 0.10g, the corresponding predicted soft soil accelerations
are 2 to 3 times greater than the rock accelerations, consistent with the measurements at soft sites during the
Loma Prieta and Mexico City earthquakes. The predicted amplification factor decreases as the rock
acceleration increases, and it becomes approximately unity (RRS 1) for a rock acceleration of 0.4g, with a
tendency for deamplification to occur at larger rock accelerations. Figure 16 summarizes other results of site
response analyses, where the effect of the plasticity index, PI, is explicitly introduced to evaluate the
influence of the nonlinear character of the soil as defined by the locations of the laboratory curves in Fig. 14.
Figure 16 presents values of peak amplification of response spectra at long periods for soft sites, RRS
max
,
calculated using the equivalent linear approach as a function of the PI of the soil, and of the rock wave
velocity V
r
, for both weak (peak rock acceleration, a
pr
= 0.1g) and strong (a
pr
= 0.4g) input rock shaking. For
a
pr
= 0.1g, V
r
= 4,000 ft/sec (1220 m/s) and PI = 50, roughly representative of San Francisco Bay area soft
sites in the Loma Prieta earthquake, RRS
max
= 4.4, which for a soil shear wave velocity of 150 m/sec is
within the recorded range for this earthquake (Figs. 4 to 6). Note the reduction of this value of RRS
max
from
4.4 to about 3.3 in Fig. 16 when a
pr
= 0.4g due to soil nonlinearity. Note also the tendency for RRS
max
to
decrease as the PI of the soil decreases.


Figure 15 : Relationship between maximum acceleration on rock and on soft soil sites (Idriss, 1990, 1991).
Let us go back to the field instrumental evidence available on the influence of soil nonlinearity on site
amplification. A number of studies have been performed which use either of three main approaches :
1) a statistical approach, using a number of rock and soil records from the same earthquake or several
earthquakes, often trying to extract evidence of site nonlinearity from separate peak acceleration or
spectral attenuation relations performed independently for rock and soil sites (e.g., Chin and Aki,
1996, Crouse and McGuire, 1996, Boore et al., 1997);
2) a pairs-of-records approach, where pairs of rock and soil sites which recorded the same earthquake
are selected and empirical amplification factors such as RRS or Ratios of Fourier Spectra are
obtained by simple division of the corresponding measured spectra. In this approach only nearby
pairs of sites, or at least sites located along a similar azimuth with respect to the source are used,
with appropriate correction for distance to the source (e.g., Borcherdt, 1996b, Joyner, 1998,
Cultrera et al., 1999, Dobry et al., 1999); and
3) a borehole-array approach, where simultaneous records at the ground surface and at some depth in
a soil site are processed, typically using one or several System Identification techniques, to provide
evidence of site nonlinearity and backfigure the corresponding nonlinear soil cyclic stress-strain
properties (e.g., Chang et al., 1989, 1996, Dobry et al., 1991, Elgamal et al., 1995a, Zeghal et al.,
1995, Iai et al., 1995, Sato et al., 1996, Ghayamghamian and Kawakami, 2000).


Figure 16 : Variation of RRS
max
of uniform layer of soft clay on rock from equivalent linear site response
analyses (Dobry et al., 1994, Dobry et al., 2000)

Approaches (1) and (2) use the ground surface records which are of most direct interest to practical
engineering applications. These two approaches are especially relevant to the problems of: (a) validating
curves such as that of Fig. 15 for soft sites and high rock accelerations, and (b) validating site coeficients for
seismic codes and microzonation. However, these two approaches have difficulties of their own. They are:
the difficulty of finding pairs of recording soil and rock sites which are truly nearby, the current scarcity of
soil (and especially soft soil) records associated with very strong rock shaking, the influence of other factors
such as rapidly changing radiation pattern/directivity effects on both soil and rock in the case of strong near-
field records, and generally the scatter of the recorded data. Some of these problems will probably be solved
in the next few years as an increasing number of soil and rock sites are instrumented and the instruments are
triggered by weak and strong earthquakes. So far, some of the studies using approaches (1) and (2) have
provided more evidence about the significance of nonlinearity than others, and research continues. On the
other hand, a number of the available borehole records and the studies using approach (3), correspond to sites
having water-saturated shallow sandy layers that have liquefied during the shaking. In these cases the effect
of soil stress-strain nonlinearity in decreasing secant modulus G with cyclic shear strain, which is known
from laboratory tests (Figs. 13-14), has been dwarfed in those shallow layers by an even greater effect of
liquefaction on G
max
and G. However, careful analysis of 1995 Kobe borehole records at sites that liquefied
show clear effect of stress-strain nonlinearity in deeper soil strata that did not liquefy (Sato et al., 1996). The
effect of liquefaction on site response as inferred from borehole records is discussed in more detail in a later
section of this paper.
The rest of this section discusses borehole records obtained at the site of the SMART-1 experiment in
Lotung, Taiwan (Fig. 17), which have provided some of the most compelling field evidence of stress-strain
nonlinearity in the absence of liquefaction. Accelerographs were installed in the profile of Fig. 17 at the
ground surface and at depths of 6m, 11m, 17m and 47m. Detailed results for seismic event LSST16 reported
by Zeghal et al. (1995) and Chang et al. (1996) are reproduced in Figs. 18-20. Figure 21 compares G/G
max
at
various values of shear strain, backfigured from this and other two events, with the corresponding G/G
max
curves measured in the laboratory on soil samples from the site.
Event LSST16 had a Magnitude 7 and occurred at a distance of about 80km from the site. The recorded
peak ground surface accelerations at the site were 0.13g and 0.17g, respectively, in the EW and NS
directions. Figures 18 and 19 correspond to the EW components of the records at various depths. Figure 19
is especially interesting. The records at the ground surface and at 6m were divided in time windows, each
having a 5-second duration. These time windows are associated with different levels of ground surface peak
acceleration which induced different levels of shear stress and strain in the soil. Figure 19 shows the Ratios
of Fourier Spectra between the records at ground surface and at 6m depth, versus frequency, for these 5-
second time windows. According to one-dimensional theory, the peak of each curve in Fig. 19 gives the
fundamental frequency of the top 6m of soil. Therefore, for the time window between 5 and 10 seconds,
corresponding to a low shear strain in the soil, this fundamental frequency is about 5.3Hz, which corresponds
to an average shear wave velocity, V
s
= (4) (6m) (5.3 cycles/sec) = 127 m/sec, in agreement with V
s
measured geophysically at the site in the top 6m. As the level of shaking increases, the fundamental
frequency at which the peak occurs in Fig. 19 decreases, and for the window between 15 and 20 sec, having
a much greater acceleration and hence inducing larger shear strain in the soil, the frequency is about 4.2Hz,
suggesting a value of G/G
max
(4.2/5.3)
2
0.63. Further decreases in the frequency after about 20 seconds
of shaking seem to have been affected by pore pressure buildup in the sand layer, possibly with a change in
the value of G
max
, and thus cannot be used as evidence for the G/G
max
curve of the intact material (Zeghal et
al, 1995).
Figure 20 presents average stress-strain loops backfigured for this same event between ground surface
and 6m for these various time windows using System Identification; these loops show clear evidence of soil
nonlinearity. The summary in Fig. 21, where the instrumental data from the field is plotted as data points
and various ranges of laboratory results are plotted as curves, reveals good agreement between soil
nonlinearity obtained from the records and from laboratory tests; this figure can be compared with Fig. 14
which is based only on laboratory results from a number of soils.





















(a) (b)

Figure 17 : Site of SMART-1 field experiment in Taiwan (Elgamal et al., 1995b)



Figure 18 : Recorded acceleration time histories at site of Fig. 17, EW components, Event LSST16
(Chang et al., 1996).



Figure 19 : Ratios of Fourier Spectra ground surface/6 m depth, for 5-second time windows, Event
LSST16 (Chang et al., 1996).


Figure 20 : Shear stress-strain hysteresis loops obtained by System Identification from the records of
Event LSST16 (Zeghal et al., 1995)



Figure 21 : Comparison between G/G
max
and
s
data obtained from three recorded
events at site of Fig. 17 (data points) and laboratory ranges (curves) (Zeghal et al., 1995).


5.0 SITE COEFFICIENTS AND SITE CLASSIFICATION IN U.S. BUILDING CODES BEFORE
1994

This section considers the basis for the regulations on local site conditions contained in building seismic
codes in the USA prior to 1994, such as the Uniform Building Code. Similar code site classifications and
site coefficients (S1, S2, S3 and S4) have been adopted in other countries. The next section discusses the
new site categories and site coefficients for seismic design of buildings, incorporated into the U.S. codes
after 1994.
Seed et al. (1976a), and more recently Idriss (1990, 1991) studied the relation between peak acceleration
recorded on soil and that obtained on a nearby rock outcrop. Seed and his coworkers had at their disposal
mostly records on stiff and deep cohesionless soil and found little difference between rock and soil
accelerations for those sites. On the other hand, Fig. 15 already discussed includes the more recent curve
developed by Idriss for soft sites; this figure indicates substantial amplification of small rock accelerations of
the order of 0.05 to 0.10g at soft clay sites.



Figure 22 : Average acceleration spectra for different site conditions (Seed et al., 1976a,b).

Figure 23 : Spectral shapes proposed by ATC-3 (1978) for Soil Profile Types S1, S2 and S3.
A second step in the development of seismic regulations prior to 1994 was the study of the shape of the
response spectrum and its correlation with the site conditions. Average spectral shapes for various soil
conditions were developed by Seed et al. (1976a, b), on the basis of a statistical study of more than 100
records from twenty-one, mostly California earthquakes available at the time. These spectral shapes are
shown in Fig. 22. As they are anchored at T = 0 to a value of 1.0, their use requires knowing the peak
acceleration (or "effective peak acceleration") on rock or soil, to be obtained from seismic hazard maps. The
spectral shapes in Fig. 22 are fairly constant and independent of site conditions at short periods, but they are
very different at longer periods, T > 0.5 sec. Therefore, based on these and other similar studies (e.g.
Mohraz, 1976), the ATC-3 Project (Applied Technology Council, 1978) developed the simplified response
spectrum shapes shown in Fig. 23. These simplified spectral shapes were later incorporated into the U.S.
Uniform Building Code. It can be noted in Fig. 23 that the Ratios of Response Spectra, RRS, relative to rock
or shallow stiff soil (i.e., corresponding to soil profile type S1) are approximately equal to 1.5 for soil type
S2 (deep firm soils) and 2.2 for soil type S3 (soft soils 20 to 40 feet thick). The response spectral shapes in
Fig. 23 are good idealizations of the findings of Seed et al. (1976a, b) shown in Fig. 22. Response spectral
shapes such as those in Fig. 23, or, alternatively, long-period amplification site coefficients S1, S2, S3, and
S4 associated with the corresponding soil profiles, were specified in the seismic codes prior to 1994, to be
used for building design in conjunction with mapped measures of the seismic hazard in rock which reflected
the level of shaking in rock. Soil profile type S4, not included in Fig. 23, corresponds to deep soft clays, and
was added later in 1998 with an even larger amplification incorporating the experience of the 1985 Mexico
City earthquake.
In principle, the spectral shapes of Fig. 23 or long-period site coefficients S used before 1994 could have
been applied in conjunction with amplification curves for the peak acceleration such as shown in Fig. 15.
This would have amplified the short period spectra, and would have also introduced the effect of soil
nonlinearity. However, this was not done, because the studies by Seed et al. (1976a) had found little effect
of site conditions on the acceleration. As a result, in the U.S. building codes before 1994 the soil
acceleration was assumed to be equal or close to the rock acceleration. This similarity between soil and rock
acceleration seems to be about right on the average for soil sites, including soft clay deposits in highly
seismic parts of California, as soil presumably does not amplify or amplifies very little the rock peak
acceleration of about 0.4g obtained from the hazard maps (e.g., Fig. 15). However, this similarity is
certainly not true in the Eastern United States and other parts of the U.S. where a level of rock acceleration
of 0.1g or 0.2g is obtained from the hazard maps, and where an amplification of the peak acceleration by soft
soil of as much as 2 or 3 (Fig. 15), and corresponding amplification of the short period part of the spectrum,
should be considered (Dobry et al., 2000).

6.0 NEW SITE COEFFICIENTS AND SITE CLASSIFICATION IN CURRENT BUILDING
CODES IN THE USA

A new system of site classification and site amplification coefficients has been incorporated into the U.S.
seismic building codes starting in 1994, partially based on the instrumental evidence, laboratory results and
analyses summarized in previous sections of this paper. This system is now part of the NEHRP and Uniform
Building Code (UBC) seismic provisions (Bachman and Bonneville, 2000, Dobry et al., 2000, Holmes,
2000). It is based on a consensus developed by the U.S. geotechnical engineering and earth science
communities on the meaning and practical implications for seismic safety of the recorded evidence and
theory.
The new procedure specifies two site coefficients, F
a
and F
v
, corresponding to the short-period and long-
period ranges, respectively, which replace the single long-period site factor S used before. The two site
coefficients depend on both site category and mapped intensity of rock motions (defined by "effective peak
accelerations" A
a
or A
v
in the 1994 NEHRP Provisions, and by other parameters in other provisions). In
addition, each Site Class is unambiguously defined by a representative average V
s
of the top 30m of the
profile at the site, labeled V
s
. Therefore, the old site categories, spectral shapes and site factors (e.g., Fig.
23) have become obsolete and are replaced by the procedure summarized in Fig. 24 and Table 1. This table
includes the approximate correspondence between the new Site Classes A to E and the old site categories S1
to S4. The values of the new site coefficients F
a
and F
v
are plotted in Fig. 25. For Site Class B (rock), which
serves as the reference for mapping of the seismic hazard, F
a
= F
v
= 1.0 in all cases, and for Site Class A
(hard rock), F
a
= F
v
= 0.8 in all cases.

Figure 24 : Two-factor approach to site response incorporated into new seismic building codes in the
U.S., starting in 1994
It is useful to review briefly the main changes and some of the implications of these new site response
provisions in the U.S. seismic building codes. These changes include the switch from a one- to a two-factor
approach for site effects through the introduction of the the short-period site coefficient F
a
; the dependence
of both site factors F
a
and F
v
on the level of rock motions to consider the effect of soil nonlinearity; and the
use of V
s
of the top 30m to characterize the site.

TABLE 1 : Summary of site categories in new U.S. seismic building codes (from NEHRP, 1994, 1997;
UBC, 1997, Dobry et al., 2000), including approximate correspondence with old site categories S1 to S4.
Site Class
Or
Soil
Profile Type
Description Shear Wave
Velocity
s
V
Top 30m
(m/sec)
Standard
Pen.
Resistance
N or N
ch

(blows/ft)
Undrained
Shear
Strength
u
S
(kP
a
)

S1
A
B

'




S1
and
S2
C
D

'






S3
and
S4
E
F

'





Hard Rock

Rock


Very dense
soil/soft rock

Stiff soil


Soft soil


Special soils
requiring site-
specific
evaluation

> 1500

760 - 1500


360 - 760


180 - 360


< 180









> 50


15 - 50


< 15









> 100


50 - 100


< 50




.
Figure 25 : Influence of site conditions and level of shaking on site coefficients F
a
and F
v
in new U.S.
seismic building codes (NEHRP, 1994, 1997).
The sketch of spectral acceleration, S
a
, in Fig. 24 was prepared using A
a
and A
v
as measures of seismic
hazard on rock. As shown by the figure, this spectral acceleration S
a
, (which is used to define the seismic
force for the design of a building located at a soil site), is proportional in the codes to the product F
a
A
a
at
short periods, and to the product F
v
A
v
at long periods. The site coefficients F
a
and F
v
are given in Fig. 25.
As F
a
and F
v
are greater in areas of small or medium seismic hazard corresponding to relatively small A
a
and
A
v
, the products F
a
A
a
and F
v
A
v
are not so different between comparable soft soil sites located in areas of
low and high seismic hazard. This conclusion can be easily verified by comparing two soft sites classified as
E, one in a highly seismic hazard area (A
a
= 0.4g) and the other in a moderate hazard area (A
a
= 0.2g). The
products F
a
A
a
for Site Class E are (0.4) (0.9) = 0.36g and (0.2) (1.7) = 0.34g, that is they are essentially
identical. Therefore, the dramatic amplification of the level of seismic ground motions by soft soil sites in
low or moderate seismicity areas tends to erase the traditional concept of seismic hazard focused exclusively
on motions on rock or firm ground. This large amplification of weaker rock motions is of course a direct
consequence of the nonlinear response of soil illustrated by Figs. 15-16, which tends to amplify more the
rock motions when they are smaller.
The site characterization (Site Class) is now based almost exclusively on the top 30m (100ft) of soil
(Table 1), disregarding the depth of soil to rock if greater than 30m, the soil properties below 30m, and the
properties of the rock underlying the soil. The Site Class is in general determined solely and unambiguously
by one parameter (the representative average shear wave velocity to a depth of 30m). This representative
average V
s
to 30m, denoted in the provisions as V
s
, is not computed as the arithmetic average of values of
V
s
down to that depth. Rather, V
s
is calculated from the time taken by the shear wave to travel from a depth
of 30m to the ground surface. The use of soil properties sometimes more readily available than V
s
, such as
the Standard Penetration Resistance (
N
or
N
ch in Table 1), or undrained shear strength (
s
u in Table 1), is
also conservatively allowed in the new provisions to characterize the top 30m of soil. Therefore, an
important change in site characterization as compared to previous codes is that essentially no information on
soil and rock stratigraphy and properties is required below a depth of 30m. While this is not entirely correct
from a theoretical viewpoint, it was felt that the stiffness of the shallow soil as measured by V
s
is in general
the most important single site parameter determining site amplification. In addition, V
s
is clearly
measurable in the field (by geophysical techniques, e.g., Hoar and Stokoe, 1978, Woods and Stokoe, 1985,
Stokoe et al., 1994); this removes the ambiguity in site characterization contained in previous codes, and
makes it much more feasible for geotechnical engineers and earth scientists to obtain the information needed
to characterize the site.
In addition to the five site classes A through E for which site coefficients are provided, a fifth Site Class F
of special soils is also defined in the codes for which site coefficients are not provided. This Site Class F
requires site-specific studies. Site Class F includes: (1) soils vulnerable to potential failure or collapse under
seismic loading such as liquefied soils, quick and highly sensitive clays, and collapsible weakly cemented
soils; (2) profiles including a total thickness of 3m (10ft) or more of peat and/or highly organic clay; (3)
profiles including a total thickness of 8m (25ft) or more of very high plasticity clays of plasticity index, PI >
75; and (4) profiles including a total thickness of 36m (120ft) or more of soft/medium stiff clays. Cases (3)
and (4) were included in Site Class F so as to exclude soil profiles approaching the Mexico City case, where
amplifications much greater than those plotted in Fig. 25 are possible.

7.0 OTHER EFFECTS AND ONGOING STUDIES

Throughout this paper, a simplified one-dimensional approach with consideration of soil nonlinear effects
has been presented which eventually evolved into the new site amplification provisions in U.S. seismic
codes. Furthermore, based on a combination of actual evidence with theoretical and practical considerations,
a parameter considering only soil and rock properties to a depth of 30m (V
s
) is used to characterize the site
in these seismic codes. The consensus that these simplifications were appropriate, as well as other details
summarized in Table 1 and Figs. 24-25, was developed through a number of studies by many people which
culminated in a workshop in Los Angeles, California in November 1992 (Martin, 1994). Some of the results
of these studies have been summarized herein; a more extensive description of the whole process and of the
new site provisions is provided by Dobry et al. (2000).
While next section reviews the complex problem of the seismic response of soil at very large strains
including pore water pressure buildup and liquefaction (included in Site Class F described in the previous
section), it is useful to review briefly here other effects and ongoing studies. They relate to aspects which
were necessarily ignored in the simplifications implicit in these new U.S. seismic site provisions, and which
may be important for some sites, earthquakes and structural periods. In addition, a number of decisions
incorporated in the new U.S. code provisions summarized in Table 1 and Fig. 25, were arrived at by
extrapolating the recorded evidence to more intense levels of rock shaking using analysis and soil parameters
from laboratory soil tests (e.g., Figs. 15-16). Records from strong earthquakes in Northridge, California
(1994), Kobe, Japan (1995), as well as the more recent 1999 Turkey and Taiwan earthquakes are making it
possible to re-evaluate these decisions. The proceedings of three recent technical meetings contain a number
of papers which both discuss these other effects and re-evaluate the site coefficients of Fig. 25, mainly in
light of the acceleration records at a number of rock and soil sites obtained during the Northdrige and Kobe
earthquakes. These proceedings are: (i) Proceedings of the International Workshop on Site Response held in
Yokosuka, Japan in Jan. 1996 (Iai, 1996); (ii) Proceedings of the North America - Japan Workshop on the
Geotechnical Aspects of the Kobe, Loma Prieta and Northridge Earthquakes held in Osaka, Japan in January,
1996 (Bardet, Idriss, Adachi, Hamada and Ishihara, 1997); and (iii) Proceedings of the NEHRP Conference
and Workshop on Research on the Northridge Earthquake held in Los Angeles in August 1997 (Mahin,
1998).
In the list that follows, some of these ongoing studies and effects not considered in the current U.S.
seismic building codes are very briefly reviewed, with emphasis in providing key references to the interested
reader:
(i) Soil Nonlinearity: There is much discussion on the degree of nonlinearity included in the site
coefficients F
a
and F
v
in Fig. 25 for Site Classes C, D and E, as the level of rock shaking
increases to high values associated with A
a
= A
v
= S
l
0.4g or greater (S
s
1.0g or greater).
Some publications argue that the effect of nonlinearity is less than that used in Fig. 25 (that is,
the curves should decrease less or not decrease at all as the level of rock shaking increases). The
evidence of nonlinearity is clearer at short periods (F
a
) than at long periods (F
v
). Records from
the Northridge earthquake included a number of Class C and D sites near the source subjected to
intense rock shaking, while the Kobe earthquake provided records of some soft Site Class E sites
subjected to intense shaking (Boore et al., 1994, Borcherdt, 1996a,b, Ejiri et al., 1996,
Midorikawa et al., 1996, Dickenson and Seed, 1996, Dobry et al., 1997, 1999, Abrahamson and
Silva, 1997, Somerville, 1998, Joyner, 1998, Cultrera et al., 1999).
(ii) Soil and Rock Below 30m: Equation 1, expressing maximum amplification, is not a function of
the thickness of the soil stratum, h, in the model of Fig. 9; on the other hand, the period at which
this maximum amplification occurs is very much a function of h (Fig. 10). A number of site
response analytical studies confirm that the value of RRS
max
is more or less independendent of
soil thickness (Dobry et al., 1994). In models based on wave propagation energy, the site
amplification is also a function of the wave velocity (or the impedance) of the shallow soil rather
than of the thickness of the soil deposit (Joyner et al., 1981, Aki, 1988). These considerations
are at the root of the current seismic provisions in the U.S. that characterize a site mainly based
on the average shear wave velocity down to 30m. Boore et al. (1997) have argued that - at least
from a theoretical viewpoint - the depth over which V
s
should be averaged is one-quarter of the
wavelength associated with the period of interest, which can be greater than 30m. On the other
hand, as pointed out by the same authors, often the stiffness of the deeper layers is well
correlated with the wave velocity in the top 30m, and thus the current code provisions may
capture well the deeper amplification effects. However, high impedance contrasts and
associated seismic wave reflections at depths much greater than 30m, sometimes between stiff
soils and even harder rock below, may produce significant site amplifications at the ground
surface that have little to do with the properties of the top 30m; sometimes the overall
amplification is a combination of both shallow and deep effects. Probably the most critical case
in which the use of V
s
of the top 30m may not capture the correct amplification and may be
unconservative at long periods occurs in some deep, low damping deposits of stiff soil on much
harder rock; an example seems to be the destructive site amplification in the 1967 Caracas,
Venezuela earthquake (Papageorgiou and Kim, 1991, Dobry, 1997). Deep soil and rock effects
may have also contributed to site amplification in the 1989 Loma Prieta and 1995 Kobe
earthquakes (Chang and Bray, 1995, 1998, Motosaka and Nagano, 1997).
(iii) Basin Edge and Other 2D and 3D Effects: Analysis of records in both the Northridge and Kobe
earthquakes has shown the importance of the shapes of the boundaries of sedimentary valleys as
well as of deeper geologic structures in determining site response, especially at long periods of
the order of several seconds. These effects were significant in the Los Angeles Basin but not in
the San Fernando Valley in the 1994 Northridge earthquake, showing that specific relations
between source and valley are needed to produce the corresponding delayed surface waves
generated at the edge of the basin and traveling in the sediment. On the other hand, similar
surface waves were recorded in both Los Angeles Basin and San Fernando Valley during the
1971 San Fernando earthquake, having a different source. Both direct waves and surface waves
related to the Osaka Basin seem to have produced 2D and 3D effects that influenced the site
response in the 1995 Kobe earthquake (Drake and Mal, 1972, Hanks, 1976, Dobry et al., 1978,
Graves, 1995, Kawase, 1996, Pitarka and Irikura, 1996, Gao et al., 1996, Bardet et al., 1997,
Somerville, 1998.). Some of the corresponding site amplification effects are concentrated near
the edge of the basin while others extend over long distances into the valley. In some cases,
basin-edge effects which are important for weak earthquake motions tend to dissapear for strong
shaking due to the increased material damping in the sediment which strongly attenuates the
surface waves; this seems to have happened in the Marina District Basin in San Francisco in the
1989 Loma Prieta earthquake (Zhang and Papageorgiou, 1996).
8.0 LARGE STRAIN SITE RESPONSE AND LIQUEFACTION EFFECTS

8.1 Cyclic Soil Behavior at Large Shear Strains
In order to discuss the state of our knowledge on the complex nonlinear site response associated with
large shear strains, it is useful to review first our basic understanding of cyclic soil behavior over a wide
strain range. The stress-strain response of soil under cyclic loading with small to medium cyclic strain
amplitudes (generally between
c
10
-4
% and
c
10
-1
% or 1%), can be usually represented by a hysteresis
loop such as that of Fig. 13. This is true for dry soil, for water-saturated soil loaded in a drained condition,
and to a certain extent also for saturated soil in undrained condition. A main reason why the hysteresis loop
in Fig. 13 changes shape as
c
increases, is because the loop is bounded by upper and lower shear stress limits
specified by the shear strength of the soil. Figure 14 contains information on secant modulus, G and
damping ratio,
s
, up to
c
= 1%, associated with shapes of loops such as that sketched in Fig. 13. To fix
ideas, let us remember that in the very destructive 1985 Mexico City and 1989 Loma Prieta earthquakes, the
soft clay deposits experienced peak strains of the order of 0.1%. Therefore, the cyclic behavior summarized
by Fig. 13 and by charts like Fig. 14 represents well much of the nonlinear site response of Site Classes C
through E covered by the new U.S. seismic codes. As discussed before in this paper, the general trend is for
an increasing level of rock shaking to decrease the amplification at short periods, with a similar but lesser
amplification decrease at long periods. However, the soft soils of Site Class E may develop large cyclic
strains, say greater than 1%, if subjected to very strong rock shaking (e.g., beyond 0.5g); in this case, the
effect of soil nonlinearity on site amplification may become more complex.
Cyclic loading associated with larger strain amplitudes (
c
> 1%), produces a different type of stress-strain
behavior in many saturated soils - especially sands, silts and some gravels. In this case, the cyclic stress
strain response is strongly affected by the excess pore water pressures developed by the shaking and
corresponding changes in the effective stress of the soil. The upper and lower limits specified by the shear
strength of the soil under a drained condition - consistent with the loop of Fig. 13 - are no longer relevant to
the hysteresis loop of the soil in an undrained condition, because these limits are affected by the change in
the effective confining stress. The new ultimate limits may be specified by either the steady-state strength of
the soil or by the onset of cavitation in the pore water. Even more important for site response is the
appearance of a hard-spring type hysteresis loop involving a progressive increase in the shear strain
amplitude. Typical examples of change in effective confining stress and cyclic stress-strain relations as
measured in the laboratory, including this hard-spring behavior, are shown in Fig. 26.

x
y
/

m
o
'

x
y
/

m
o
'


Figure 26 : Sand behavior under undrained cyclic shear (Ishihara, 1985)

The difference between the "round" hysteresis loop shown in Fig. 13 and the "hard-spring" type hysteresis
loop shown in Fig. 26, can result in very different site responses. The liquefiable soil sites included in Site
Class F have a potential to undergo these large strain levels, because they are initially soft, and also because
the shaking induces in them excess pore pressures that decrease even more the initial values of G
max
and G.

8.2 Evidence from Field Measurements and Analysis
Distinctive evidence of large strain site response was obtained, among others, from a borehole array in
Kushiro, Japan, during the 1993 Kushiro-Oki earthquake (Iai et al., 1995). As shown in Fig. 27, the
subsurface profile of the Kushiro site down to a depth of about 50m consists of alternating layers of dense
and very dense sand, each a few meters in thickness, with Standard Penetration Test N-values being
alternately about 20 blows/foot and greater than 50. Silt layers occasionally appear and, when these layers
are encountered, the N-values decrease to less than 10. A dense sand layer exists below this with N-values
of 50 or greater down to a depth of 77m, which is the level of installation of the borehole accelerometer.
The Kushiro-Oki earthquake had a magnitude, M = 7.8, a focal depth of 107km and an epicenter located
15km south of the site. The recorded acceleration time histories are shown in Fig. 28. In this figure, the
major portion of the earthquake record is plotted from the original record, which had a total duration of 180
seconds; the origin of time is ten seconds before triggering. The accelerometers used were of force-balance
type with digital IC memory, having a flat frequency gain over the 0.01 to 35 Hz range. Loci of the
earthquake motions in a horizontal plane indicate that the velocity and displacement of the earthquake
motion were predominant in the NS direction, which is consistent with the fault movement of the earthquake.
By comparing the ground surface motion in
the NS direction shown at the top of Fig. 28, with
the NS motion recorded at a depth of 77m, a
distinctive change can be observed in the
response of ground after about 30 seconds. Most
of the high frequency accelerations are now
filtered out, with the cyclic motion having a
predominant period of about 1.5 seconds and
exhibiting a high-frequency spike at the peak of
each acceleration cycle. These features are less
remarkable in the EW component, probably due
to the smaller shaking in both velocity and
displacement in this direction. In contrast to this
distinctive response of the ground during the
main shock of the Kushiro-Oki earthquake, no
peculiarity was observed in the site response at a
smaller level of excitation at the same site (Iai et
al., 1995).
In order to study the mechanism of this
peculiar site response, a one-dimensional,
nonlinear, effective stress site response analysis
was performed. The model used was a strain
space multiple mechanism model, which
reasonably represents the large strain behavior of
soil discussed above. The recorded earthquake
motion at a depth of 77m was used as input in the
analysis. Soil samples from the deposit, obtained
after in-situ freezing of the soil were tested in the
laboratory to calibrate the analytical model. The results of the analysis were consistent with the instrumental
observations, as illustrated by Fig. 29. It was concluded that the recorded spiky acceleration time history at
the ground surface is due to the large strain cyclic behavior of the saturated sand, in which undrained
contractive and dilative behavior of the soil alternately makes the stiffness of the sand soften and stiffen
during cyclic loading, as represented by a hard-spring type hysteresis loop such as shown in Fig. 26.




Figure 27 : Boring log at Kushiro, Japan, site

Figure 29: Recorded and computed site response

8.3 Site Response Implication of Large
Strain Response.
The hard-spring type nonlinearity of soil
deposit associated with cycling at a large strain
implies a very complicated nonlinear site response.
By varying the level of input motion at a depth of
77m, with the peak acceleration ranging from 0.5
to 2.5 m/s
2
(this range includes the peak
acceleration of 2.04 m/s
2
recorded at 77m during
the main shock), time histories at the ground surface were calculated as shown in Fig. 30, with the
corresponding Ratios of Fourier Spectra of the ground surface over the input motion shown in Fig. 31. The
nonlinearity in the site response is more clearly seen in the time histories of Fig. 30 than in the Ratios of
Fourier Spectra of Fig. 31. Then, the parameters of the soil model were varied in a parametric study of the
site, ranging from Case-1 in the original model to Case-4 corresponding to a weaker cyclic resistance of the
soil.



Figure 28 : Recorded acceleration time histories
during 1993 Kushiro-Oki earthquake (Iai et al., 1995)

Figure 30 : Computed response at the ground
surface for various levels of
shaking (Iai et al., 1995)

Figure 31 : Frequency transfer functions of computed
response of the ground at various levels of
shaking for ratios of the ground surface
motions over the input motion at a depth of
77 meters (Iai et al., 1995)
The ground surface peak accelerations calculated in this parametric study are plotted versus input peak
acceleration in Fig. 32, and the corresponding amplification ratios of peak acceleration are plotted in Fig. 33
(Iai at al., 1996). The results in Fig. 33 indicate that the amplification factor for the peak acceleration
(comparable to short-period site factor F
a
in the U.S. codes), may be a complex function of input earthquake
motion and soil properties in liquefiable soils at large soil strains. These results are clearly different from the
simplified general trend we saw in previous sections of this paper for small to medium strain site response in
Sites C through E, of decreasing amplification as the level of input shaking increases. While of different
origin, the increased amplification at long periods recorded by a downhole array at the Wildlife Site in
California during the 1987 Superstition Hills earthquake, as the soil liquefied, confirms again the complexity
of large strain response, especially at liquefiable soil sites (Zoropapel and Vucetic, 1994).





A mild ground surface slope, even if it is only of a few degrees, may further complicate the large strain
site response of a liquefiable site. Early evidence was obtained, among others, from centrifuge tests of a
model sand deposit simulating liquefaction and lateral spreading (Dobry and Taboada, 1993). The centrifuge
model test was performed on a saturated sand deposit having a relative density of about 45%, and using a
laminar box as shown in Fig. 34. The test was performed with and without a ground slope. The model test
for perfectly level ground resulted in the recorded acceleration response shown in Fig. 35, where the
accelerations oscillate more or less symmetrically about the neutral axis. The response of the liquefying
sloping ground, however, is nonsymmetric as shown in Fig. 36. The spiky acceleration occurred every time
the sand deposit deformed downward along the slope, and this high-frequency spike was invariably directed
upslope. This has been explained by the presence of the static driving shear stress in the sloping ground due
to gravity. This driving shear stress mobilizes the shear strain in the soil deposit downward along the slope,
so that the soil deposit tends to accumulate strain in the downslope direction; simultaneously the stress-strain
response becomes dilative and the soil strength increases when the downward cyclic strain reaches a certain
value, thus explaining the upslope acceleration spikes. The effective stress model mentioned earlier was
successful in simulating this type of nonsymmetrical one-way spiky response of the sloping ground (Lacy et
al., 1993).

CASE-1
CASE-2
CASE-3
CASE-4


Figure 32 : Relation between peak acceleration at
the ground surface and input
acceleration calculated in the
parametric effective stress analyses (Iai
et al., 1996)
CASE-1
CASE-2
CASE-3
CASE-4

Figure 33: Shaking level dependency of amplification
factor of peak acceleration calculated in
the parametric effective stress analyses
(Iai et al., 1996)
Unit in meters
in prototype scale

Figure 34: Centrifuge model of soil deposit (Dobry and Taboada, 1993)








TIME (s)
Figure 35: Horizontal accelerations measured
in the centrifuge in level ground
(Dobry and Taboada, 1993)
TIME (s)


Figure 36: Horizontal acceleration measured in the
centrifuge in sloping ground (Dobry and
Taboada, 1993)
9.0 CODE SITE COEFFICIENTS AND MICROZONATION

It is useful to discuss very briefly the relation between generic site characterization and site coefficients
used in national seismic codes - such as described in this paper for the USA - and microzonation of smaller
areas, typically cities. While ground motions are only one of the geotechnical hazards considered when a
city or urban area is microzoned (others being liquefaction/ground failure and ladsliding, see ISSMGE,
1999), the two subjects are very much related. Microzoning based on a thorough study of the actual seismic
sources dominating the seismic hazard for the city, of the geotechnical and geophysical properties of the
existing soils and rocks including relevant geologic information/shape of the valley/deep structures, and of
the surface topography, may go a long way toward eliminating some of the problems and limitations of
national generic site coefficients and site classification systems. For example, the properties of the soil and
rock below 30m, as well as 2D and 3D effects related to the shape of the sedimentary valley, and appropriate
degrees of soil nonlinearity, can be naturally incorporated into the microzonation maps, without the need to
complicate the task of the user with unrealistic demands for subsurface information from large depths or
under other parts of the city. Also, microzonation can address directly as needed the influence on ground
motion amplification of liquefaction (previously discussed in this paper) and of surface topography (not
discussed).
One current problem with microzonation is the wide variety of ways available for collecting and
interpreting the necessary information, as well as the wide variety of field, laboratory and analytical
techniques being used. This can be partly solved by providing microzonation guidelines (e.g., ISSMGE,
1999). The best schemes for microzonation of ground motions are those which incorporate the installation of
a comprehensive seismic array covering the area of interest that is operated for a number of years. The
collection of a number of representative earthquake records on various geotechnical units in different parts of
the city, even if some extrapolation is still necessary to cover other earthquake sources and stronger shaking,
provides an invaluable tool for verifying theories and significance of effects, and for constraining the final
recommendations. Borehole arrays or pairs of soil/rock sites can be used to generate empirical Ratios of
Response Spectra (RRS) and Ratios of Fourier Spectra, which are directly applicable to both analysis
calibration and recommendations of site coefficients. This has been done in the Mexico City seismic code
(Fig. 37) and in the recently completed microzonation of the city of Medellin in Colombia (Municipio de
Medellin, 1999).
A possible overall approach is that taken by the new (1997-98) national seismic code provisions of
Colombia (Asociacion Colombiana de Ingeneria Sismica, 1998), that provides a generic system of national
site coefficients and site classification scheme similar to the situation in the USA described in this paper
and also explicitly encourages the microzonation of cities to improve these generic recommendations. So
far, several cities in Colombia have already been microzoned including the capital city of Bogota
(Ingeominas, 1997) and Medellin. National efforts in the U.S. and other countries to install massive numbers
of seismic instruments using new digital technology, including borehole arrays at selected locations, are an
important step toward the goal of combining national code provisions with the incorporation of local
conditions made possible by microzonation. In the U.S., the Geological Survey (USGS) is planning to install
about 3,000 free-field instruments for this purpose including borehole arrays - supplemented with other
3,000 instruments in buildings and other structures of interest, plus 1,000 instruments to be used for regional
earthquake monitoring. The deployment of these 7,000 instruments is contemplated as part of the new
Advanced National Seismic System (ANSS), see Frankel et al. (2000) and ANSS web site at
http://pasadena.wr.usgs.gov/eqhaz/ANSS.html

10.0 CONCLUSIONS

Both instrumental evidence during earthquakes and theory show that the shear wave velocity of the
shallow layers plays a key role in determining site amplification. The evidence of the large amplification of
spectra at both short and long periods is clearer for soft clays subjected to moderate rock motions ( 0.1g),
but significant amplifications are also observed in stiffer soils, with a key role in both cases played by the
soil/rock impedance contrast and the material damping ratio of the soil. Nonlinear stress-strain effects
generally tend to decrease the amplification as the level of rock shaking increases, but this trend may reverse
itself for very strong shaking and/or liquefaction effects associated with high levels of cyclic straining in the
soil.



Figure 37 : Microzonation of Mexico City into Zones I, II and III (Simn and Suarez, 1994)

The recent changes in the site coefficients and the way sites are characterized in national seismic
building codes in the USA are partially based on these findings. The average shear wave velocity in the top
30m of the site is used almost exclusively to characterize the site, two amplification site coefficients are
defined at short and long periods respectively, and both of these coefficients are made a function of the level
of seismic hazard in rock to account for soil nonlinearity. While these new code provisions necessarily
involve simplifications, they are consistent with the evidence, define any and every site without ambiguity,
and are practical to use. However, as many of the nonlinearity effects included in the current site
coefficients were obtained by analytical extrapolation of the instrumental evidence, discussion still continues
on some of the numbers to use, and the growing pool of records from recent earthquakes is being used for
comparisons and re-evaluation of the site coefficients.
Effects currently not considered by the site provisions in the U.S. codes include the influence of the soil
and rock below 30m, and basin edge and other 2D and 3D effects, which may not be adequately covered by
the code provisions for some combinations of sites, earthquakes and ranges of periods. Large strain site
response including liquefaction effects define another important area dominated by strong soil nonlinearity,
which is not completely covered by the code provisions. While some of these effects may be incorporated in
future code developments, another route is an increased use of microzonation of cities and urban areas.
Microzonation can be especially helpful in cases where local seismic arrays have provided enough results to
support and constrain the corresponding recommendations. Microzonation can be used to supplement and
refine the necessarily simplified generic provisions included in national seismic codes.
The specific seismic code site provisions discussed in this paper reflect the geological/geotechnical
conditions and design practice of buildings in the U.S. The authors have reservations about their general
applicability and recommend caution in any effort to implement them in other geographical areas around the
world.

11.0 ACKNOWLEDGEMENTS

The first author (RD) wants to acknowledge the Multidisciplinary Center for Earthquake Engineering
Research and the National Science Foundation, USA, for support in the writing of this keynote paper. The
second author (SI) would like to acknowledge the support by the Science and Technology Agency, Japan.

12.0 REFERENCES

Abrahamson, N.A., and Silva, W.J. (1997). "Empirical response spectral attenuation relations for shallow
crustal earthquakes". Seism. Res. Lett., 68, 94-127.
Aki, K. (1988). "Local site effects on strong ground motion". Proc. Specialty Conf on Earthquake
Engineering and Soil Dynamics II - Recent Advances in Ground-Motion Evaluation, (L. Von Thun, Ed.),
ASCE Geotechnical Special Publication No. 20, pp. 103-155, Park City, Utah, June 27-30.
Asociacion Colombiana de Ingenieria Sismica (1998). "NSR-98, normas colombianas de diseno y
construccion sismo resistente". (in Spanish), Ley 400 de 1997, Decreto 33 de 1998, Asociacion
Colombiana de Ingenieria Sismica, Vols. 1 and 2, Bogota, Colombia.
ATC (1978). "Tentative provisions for the development of seismic regulations for buildings". Report ATC 3-
06, San Francisco, California, Applied Technology Council.
Bachman, R., and Bonneville, D., (2000). "The seismic provisions of the 1997 Uniform Building Code".
Earthquake Spectra, 16, No. 1, pp. 85-100, February.
Bardet, J.P., Idriss, I.M., ORourke, T.D., Adachi, N., Hamada, M., and Ishihara, K., Editors (1997).
Proceedings of the North America-Japan Workshop on the Geotechnical Aspects of the Kobe, Loma
Prieta and Northridge Earthquakes, Dept. of Civil Engineering Report (University of Southern
California), Osaka, Japan, January 22-24, 1996.
Boore, D.M., Joyner, W.B., and Fumal, T.E. (1994). "Estimation of response spectra and peak accelerations
from western North American earthquakes: An interim report". Part 2, U.S. Geol. Surv. Open-File Rept.
94-127, 40 pp.
Boore, D.M., Joyner, W.B., and Fumal, T.E. (1997). "Equations for estimating horizontal response spectra
and peak acceleration from western North American earthquakes: A summary of recent work".
Seismological Research Letters, Vol. 68, No. 1, pp. 128-153.
Borcherdt, R.D. (1994a). "Simplified site classes and empirical amplification factors for site-dependent code
provisions". Proceedings of the 1992 NCEER/SEAOC/BSSC Workshop on Site Response During
Earthquakes and Seismic Code Provisions, (G.M. Martin, Editor), National Center for Earthquake
Engineering Research Special Publication NCEER-94-SP01 (Buffalo, NY), University of Southern
California, Los Angeles, November 18-20, 1992. (1993, Simplified site classes and empirical
amplification factors for site-dependent code provisions, in Proc., International Workshop on Strong-
Motion Data, Menlo Park, CA, Port and Harbour Research Institute, Iai, S., ed., 1993).
Borcherdt, R.D. (1994b). "Estimates of site-dependent response spectra for design (methodology and
justification)". Earthquake Spectra, 10, pp. 617-653.
Borcherdt, R.D. and Glassmoyer, G. (1994). "Influences of local geology on strong and weak ground
motions in the San Francisco Bay region, California and their implications for site-specific code
provisions". in The Loma Prieta earthquake of October 17, 1989 -- strong ground motion, R.D.
Borcherdt, ed., U.S. Geological Survey Professional Paper 1551-A, p. A77-A108.
Borcherdt, R.D. (1996a)."Strong ground motions generated by the Northridge and Hanshin-Awaji
earthquakes of January 17, 1994 and 1995; Implications for site-specific design factors". Eleventh World
Conference on Earthquake Engineering, Procs., CD-Rom.
Borcherdt, R.D. (1996b)."Preliminary amplification estimates inferred from strong ground-motion
recordings of the Northridge earthquake of January 17, 1994". Proc. International Workshop on Site
Response Subjected to Strong Earthquake Motions, S. Iai, ed., Jan. 16-17, Yokosuka, Japan, 2, 21-46.
Chang, C.Y., Power, M.S., Tang, Y.K. and Mok, C.M. (1989). "Evidence of nonlinear soil response during a
moderate earthquake". Proc. 12
th
Int. Conf. on Soil Mech. And Found. Engng., 3, International Society of
Soil Mechanics and Foundation Engineering.
Chang, C.Y. (1991). Personal communication to R. Dobry.
Chang, C.Y., Mok, C.M. and Tang, H.-T. (1996). "Inference of dynamic shear modulus from Lotung
downhole data". Journal of Geotechnical Engineering, ASCE, 122 (8), 657-665, August.
Chang, S.W. and Bray, J.D. (1995). "Seismic response of deep, stiff soil deposits in the Oakland, California
area during the Loma Prieta earthquake". Report. No. UCB/GT/95-06, Geotechnical Engineering, Dept.
of Civil and Environmental Engineering, U. of California, Berkeley, California.
Chang, S.W. and Bray, J.D. (1998). "Implications of recent strong motion data for seismic building code
design at deep, stiff sites". Proc. of the NEHRP Conference and Workshop on Research on the
Northridge, California Earthquake of January 17, 1994, California Universities for Research In
Earthquake Engineering (CUREe), Richmond, CA, Vol. II, pp. 90-99 (S.A. Mahin, editor), August 20-
22, 1997, Los Angeles.
Chin, B.-H. and Aki, K. (1996). "Local site effects study on ground motion during the 1994 Northridge
earthquake". Proceedings of the International Workshop on Site Response subjected to Strong
Earthquake Motions (S. Iai, Editor), Sponsored by the Science and Technology Agency, Japan,
Yokosuka, Japan, January 16-17, Vol. 1, pp. 34-39.
Crouse, C.B. and McGuire, J.W. (1996). "Site response studies for the purpose of revising NEHRP seismic
provision". Earthquake Spectra, Vol. 12, No. 3, August, pp. 407-439.
Cultrera, G., Boore, D.M., Joyner, W.B. and Dietel, C.M. (1999). "Nonlinear soil response in the vicinity of
the Van Norman Complex following the 1994 Northridge, California, earthquake". Bulletin of the
Seismological Society of America, 89:1214-1231.
Dickenson, S.E. and Seed, R.B. (1996). "Nonlinear dynamic response of soft and deep cohesive soil
deposits". Proc. International Workshop on Site Response Subjected to Strong Earthquake Motions,
edited by S. Iai, Jan. 16-17, Yokosuka, Japan, Vol. 2, pp. 67-81.
Dobry, R., Idriss, I.M. and Ng. E.N. (1978). "Duration characteristics of horizontal components of strong-
motion earthquake records". Bulletin of the Seismological Society of America, 68 (5): 1487-1520.
Dobry, R. (1991). "Soil properties and earthquake response". Proc., X European Conference of Soil
Mechanics and Foundation Engineering, Vol. IV, pp. 1171-1187, Florence, Italy, May 26-30.
Dobry, R. (Panel Leader), Bhatia, S.K., Cramer, C.H., Henke, R., Higgins, C, Jamiolkowski, M., Ko, H.-K.,
Ladd, R.S., Prakash, S., Saada, A. and Sirles, P.C., (1991). "Low- and high-strain cyclic material
properties". Proc. NSF/EPRI Workshop on Dynamic Soil Properties and Soil Characterization, Palo
Alto, CA, Nov. 9-10, 1989, Vol. 1, Chapter 3, pp. 3-1 to 3-100, Report EPRI NP-7337, June 1991.
Dobry, R. and Taboada, V.M. (1993). "Possible lessons from VELACS model No.2 results". Proceedings
Intl. Conf. on the Verification of Numerical Procedures for the Analysis of Soil Liquefaction Problems,
Vol. 2, pp.1341-1352, K. Arulanandan and R.F. Scott (Eds.), Balkema.
Dobry, R., Martin, G.M., Parra, E. and Bhattacharyya A., (1994). "Development of site-dependent ratios of
elastic response spectra (RRS) and site categories for building seismic codes". Proceedings of the 1992
NCEER/SEAOC/BSSC Workshop on Site Response During Earthquakes and Seismic Code Provisions
(G.M. Martin, Editor), National Center for Earthquake Engineering Research Special Publication
NCEER-94-SP01 (Buffalo, NY), University of Southern California, Los Angeles, November 18-20,
1992.
Dobry, R. (1997). "Lessons contributed by the 1967 Caracas earthquake to the incorporation of the soil
amplification effect into seismic codes". Proc. International Symposium on the July 29, 1967 Caracas
Earthquake: 30 Years Later, Caracas, Venezuela, July29-August 1, 1997.
Dobry, R., Ramos, R., and Power, M.S. (1997). "Site factors and site categories in seismic codes: a
perspective". Proceedings of the NCEER Workshop on the National Representation of Seismic Ground
Motion for New and Existing Highway Facilities, Report NCEER-97-0010, edited by Maurice S. Power
and Ronald L Mayes, May 29-30, San Francisco, pp. 137-170.
Dobry, R., R., Ramos, R., and Power, M.S. (1999). "Site factors and site categories in seismic codes". Report
MCEER-99-0010, Buffalo, New York: Multidisciplinary Center for Earthquake Engineering Research,
April.
Dobry, R., Borcherdt, R.D., Crouse, C.B., Idriss, I.M., Joyner, W.B., Martin, G.R., Power, M.S., Rinne, and
Seed, R.B. (2000). "New site coefficients and site classification system used in recent building seismic
code provisions". Earthquake Spectra, 16, No. 1, pp.41-67, February.
Drake, L.A. and Mal, A.K. (1972). "Love and Rayleigh waves in the San Fernando Valley". Bulletin of the
Seismological Society of America, 62:1678-1690.
Elgamal, A.W., Zeghal, M. and Parra, E. (1995a). "Identification and modeling of earthquake ground
response". Proc. 1
st
Int. Conf. on Earthquake Geotechnical Engineering, Special Keynotes and Theme
Lectures, Tokyo, 51-90.
Elgamal, A.-W., Zeghal, M., Tang, H.T. and Stepp, J.C. (1995b). "Lotung downhole array. I:evaluation of
site dynamic properties". Journal of Geotechnical Engineering, 121 (4), 350-362, April.
Ejiri, J., Sawada, S., Goto, Y. and Toki, K. (1996). "Peak ground motion characteristics". Soils and
Foundations, Special Issue on the Geotechnical Aspects of the January 17, 1995, Hyoken-Nambu
Earthquake, pp. 7-13, January.
Frankel, A., Benz, H. and Filson, J. (2000). "NEHRP advanced national seismic system will benefit
earthquake engineering". EERI Newsletter, Vol. 34, No. 6, pp. 2-4, June, Earthquake Engineering
Research Institute, Oakland, California.
Gao, S., Liu, H., Davis, P.M., and Knopoff, L., (1996). "Localized amplification of seismic waves and
correlation with damage due to the Northridge earthquake: evidence for focusing in Santa Mnica".
Bulletin of the Seismological Society of America, 86 (1B): S209-S-230.
Ghayamghamian, M.R. and Kawakami, H. (2000). "On-site nonlinear hysteresis curves and dynamic soil
properties". Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 126 (6), 543-555, June.
Graves, R.W. (1995). "Preliminary analysis of long-period basin response in the Los Angeles region from
the 1994 Northridge earthquake". Geophys. Research Letters, 22, 101-104.
Hanks, T.C. (1976). "Observations and estimation of long period strong ground motion in the Los Angeles
Basin". Earthquake Engineering and Structural Dynamics, Vol. 4, pp. 473-488.
Hardin, B.O. and Drnevich, V.P. (1972a). "Shear modulus and damping in soils: measurement and parameter
effect". Journal of Soil Mechanics and Foundations Division, ASCE, Vol.98, No. SM6, pp. 603-624
Hardin, B.O. and Drnevich, V.P. (1972b)."Shear modulus and damping in soils: design equation and curves".
Journal of Soil Mechanics and Foundations Division, ASCE, Vol.98, No. SM7, pp. 667-692.
Hoar, R.J. and Stokoe, K.H. II. (1978). "Generation and measurement of shear waves in situ". Dynamic
Geotechnical Testing, ASTM STP 654, American Society for Testing and Materials, pp. 3-29.
Holmes, W.T. (2000). "The 1997 NEHRP provisions for seismic regulations for new buildings and other
structures". Earthquake Spectra, 16, No. 1, pp. 101-114, February.
Housner, G.W. (1990). "Competing against time: report to governor George Deukmejian from the
Governors Board of Inquiry on the 1989 Loma Prieta Earthquake". May.
Iai, S., Morita, T., Kameoka, T., Matsunga, Y. and Abiko, K. (1995). "Response of a dense sand deposit
during 1993 Kushiro-Oki earthquake". Soils and Foundations, Vol.35, No.1, pp. 115-131.
Iai, S. Editor (1996). Proceedings of the International Workshop on Site Response subjected to Strong
Earthquake Motions, Sponsored by the Science and Technology Agency, Japan, Yokosuka, Japan,
January 16-17, Vols. 1 and 2.
Iai, S., Morita, T., Miyata, M. and Sakurai, H. (1996). "Earthquake response of saturated sandy ground".
Proceedings of the International Workshop on Site Response subjected to Strong Earthquake Motions (S.
Iai, Editor), Sponsored by the Science and Technology Agency, Japan, Yokosuka, Japan, January 16-17,
Vol. 1, pp. 63-68.
Idriss, I.M. (1990). "Response of soft soil sites during earthquakes". Proceedings of the Symposium to Honor
Professor H. B. Seed, Berkeley, May, pp. 273-289.
Idriss, I.M. (1991). "Earthquake ground motions at soft soil sites". Proceedings of the Second International
Conference on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, St. Louis,
Missouri, Vol. III, pp. 2265-2273.
Ingeominas (1997). "Microzonificacion sismica de Santa Fe de Bogota". (in Spanish), Ingeominas, Convenio
Interadministrativo 01-93, Bogota, Colombia.
Ishihara, K. (1985). "Stability of natural deposits during earthquakes". Proc. 11
th
ICSMFE, San Francisco,
Vol.1, pp. 321-376.
ISSMGE (1999). "Manual for zonation of seismic geotechnical hazards (revised version)". Technical
Committee for Earthquake Geotechnical Engineering, TC4 (Seco e Pinto and Ishihara, Eds.),
International Society for Soil Mechanics and Geotechnical Engineering, March.
Joyner, W.B. Warrick, R. and Fumal, T. (1981). "The effect of Quaternary alluvium on strong ground motion
in the Coyote Lake, California earthquake of 1979". Bull. Seism. Soc. Am., 71, 1333 - 1349.
Joyner, W.B. Fumal, T.E. and Glassmoyer, G. (1994). "Empirical spectral response ratios for strong motion
data from the 1989 Loma Prieta, California, earthquake". Proceedings of the 1992
NCEER/SEAOC/BSSC Workshop on Site Response During Earthquakes and Seismic Code Provisions
(G.M. Martin, Editor), National Center for Earthquake Engineering Research Special Publication
NCEER-94-SP01 (Buffalo, NY), University of Southern California, Los Angeles, November 18-20,
1992.
Joyner, W.B. (1998). "Implementation overview: strong ground motion and seismology". Proc. of the
NEHRP Conference and Workshop on Research on the Northridge, California Earthquake of January
17, 1994, California Universities for Research In Earthquake Engineering (CUREe), Richmond, CA,
Vol. II, pp. 34-43 (S.A. Mahin, editor), August 20-22, 1997, Los Angeles.
Kawase, H. (1996). "The cause of damage belt in Kobe: the basin-edge effect, constructive interference of
the direct S-wave with the basin-induced diffracted/Rayleigh waves". Seismological Research Letters,
67, No. 5, 25-34.
Lacy, S.J. Aubry, D., Benzenati, I., Modaressi, A., Roth, W. and Inel, S (1993). "Overview of VELACS
model No. 2". Proceedings Intl. Conf. on the Verification of Numerical Procedures for the Analysis of
Soil Liquefaction Problems, Vol. 2, pp. 1369-1381, K. Arulanandan and R.F. Scott (Eds.), Balkema.
Mahin, S.A. Editor (1998). Proceedings of the NEHRP Conference and Workshop on Research on the
Northridge, California Earthquake of January 17, 1994, California Universities for Research In
Earthquake Engineering (CUREe), Richmond, CA, Vols. I-IV, August 20-22, 1997, Los Angeles.
Martin, G.R. Editor (1994). Proceedings of the 1992 NCEER/SEAOC/BSSC Workshop on Site Response
During Earthquakes and Seismic Code Provisions, National Center for Earthquake Engineering
Research Special Publication NCEER-94-SP01 (Buffalo, NY), University of Southern California, Los
Angeles, November 18-20, 1992.
Midorikawa, S., Matsuoka, M. and Sakugawa, K. (1994). "Site effects on strong-motion records observed
during the 1987 Chiba-Ken-Toho-Oki, Japan earthquake". Proc. 9th Japan Earthq. Eng. Symp., 3, E085-
E090.
Midorikawa, S., Si, H. and Matsuoka, M. (1996). "Attenuation of peak acceleration and velocity observed
during recent large earthquakes in Japan - the 1994 Hokkaido-Toho-Oki and 1995 Hyogo-Ken Nanbu
earthquakes". Proceedings of the International Workshop on Site Response subjected to Strong
Earthquake Motions (S. Iai, Editor), Sponsored by the Science and Technology Agency, Japan,
Yokosuka, Japan, January 16-17, Vol. 2, pp. 201-210.
Mohraz, B. (1976). "Earthquake response spectra for different geologic conditions". Bull. Seism. Soc. Am.,
Vol. 66, pp. 915-935.
Motosaka, M. and Nagano, M. (1997). "Analysis of amplification characteristics of ground motions in the
heavily damaged belt zone during the 1995 Hyogo-ken Nanbu earthquake". Earthquake Engineering and
Structural Dynamics, 26 (3), 377-393.
Municipio de Medellin (1999). "Instrumentacion y microzonificazion del area urbana de Medellin". (in
Spanish), Municipio de Medellin, Medellin, Colombia.
NEHRP (1994). Recommended Provisions for Seismic Regulations for New Buildings, Building Seismic
Safety Council (BSSC), FEMA 222A/223A, Vol. 1 (Provisions) and Vol. 2 (Commentary).
NEHRP (1997). Recommended Provisions for Seismic Regulations for new Buildings and Other Structures,
Building Seismic Safety Council (BSSC), FEMA 302/303, Part 1 (Provisions) and Part 2 (Commentary).
Ordaz, M. and Arciniegas (1992). Personal communication to Ricardo Dobry.
Papageorgiou, A.S. and Kim, J. (1991). "Study of the propagation and amplification of seismic waves in
Caracas Valley with reference to the 29 July 1967 earthquake: SH waves". Bulletin of the Seismological
Society of America, 81 (6): 2214-2233.
Pitarka, A. and Irikura, K. (1996). "Basin structure effects on long-period strong motions in the San
Fernando Valley and the Los Angeles Basin from the 1994 Northridge earthquake and an aftershock".
Bulletin of the Seismological Society of America, 86 (1B): S126-S137.
Reinoso, E. and Ordaz, M. (1999). "Spectral ratios for Mexico City from free-field recordings". Earthquake
Spectra, Vol. 15, No. 2, pp. 273-295.
Roesset, J.M. (1977). "Soil amplification in earthquakes". Numerical Methods in Geotechnical Engineering,
edited by C. S. Desai and J.T. Christian, Chapter 19, pp. 639-682. New York: McGraw Hill.
Sato, K., Kokusho, T., Matsumoto, M. and Yamada, E. (1996). "Nonlinear seismic response and soil
property during strong motion". Soils and Foundations, Special Issue on the Geotechnical Aspects of the
January 17, 1995 Hyoken-Nambu Earthquake, pp. 41-52, January.
Schnabel, P.B., Lysmer, J. and Seed, H.B. (1972). "SHAKE: a computer program for earthquake response
analysis of horizontally layered sites". Report EERC 72-12, Earthquake Engineering Research Center,
University of California, Berkeley, CA.
Seed, H.B. and Idriss, I.M. (1970). "Soil moduli and damping factors for dynamic response analyses". Report
No. EERC 70-10, Univ. of California, Berkeley, Calif.
Seed, H.B., Murarka, R., Lysmer, J. and Idriss, I.M. (1976a). "Relationships between maximum acceleration,
maximum velocity, distance from source and local site conditions for moderately strong earthquakes".
Bulletin of the Seismological Society of America, 66 (4): 1323-1342.
Seed, H.B., Ugas, C. and Lysmer, J. (1976b). "Site dependent spectra for earthquake -resistant design".
Bulletin of the Seismological Society of America, 66 (1): 221-244.
Seed, H.B., Romo, M.P., Sun, J.I., Jaime, A. and Lysmer, J. (1988). "The Mexico earthquake of September
19, 1985 - relationships between soil conditions and earthquake ground motions". Earthquake Spectra,
Vol. 4, No. 4, pp. 687-729.
Simn, L.A. and Suarez, M.B. (1994). Reglamento de Construcciones para el Distrito Federal, (in Spanish),
Editorial Trilles, Mexico, D. F.
Somerville, P. (1998). "Research overview: strong ground motion". Proc. of the NEHRP Conference and
Workshop on Research on the Northridge, California Earthquake of January 17, 1994, California
Universities for Research In Earthquake Engineering (CUREe), Richmond, CA, Vol. II, pp. 15-27 (S.A.
Mahin, editor), August 20-22, 1997, Los Angeles.
Stokoe, K.H. II, Wright, S.G., Bray, J.A. and Roesset, J.M. "Characterization of geotechnical sites by SASW
method". Geophysical Characterization of Sites, Volume Prepared by ISSMFE Technical Committee
No. 10 (R.D. Woods, Ed.), pp. 15-25, International Science Publisher, NY.
UBC (1997). Uniform Building Code, International Council of Building Officials (ICBO), Whittier, CA.
Vucetic, M. and Dobry, R. (1991). "Effect of soil plasticity on cyclic response". Journal of Geotechnical
Engineering, ASCE, 117(1):89-107, January.
Woods, R.D. and Stokoe, K.H., II (1985). "Shallow seismic exploration in soil dynamics". Richart
Commemorative Lectures, ASCE, Detroit, Michigan (R.D. Woods, Ed.), pp. 120-156.
Zeghal, M., Elgamal, A.-W., Tang, H.T. and Stepp, J.C. (1995). "Lotung downhole array. II: evaluation of
soil nonlinear properties". Journal of Geotechnical Engineering, ASCE, 121 (4), 363-378, April.
Zhang, B. and Papageorgiou, A.S. (1996). "Simulation of the response of the Marina District Basin, San
Francisco, California, to the 1989 Loma Prieta earthquake". Bulletin of the Seismological Society of
America, 86 (5): 1382-1400.
Zoropapel, G.T. and Vucetic, M. (1994). "The effects of seismic pore water pressure on ground surface
motion". Earthquake Spectra, Vol. 10, No. 2, pp. 403-438.

Вам также может понравиться