Incorporating site response into earthquake hazard microzonation maps


Silva, W.J.; Wong, I.; Silvia, L.; Gregor, N.; Wright, D.

Seismological Research Letters 71(1): 247

2000


using
the
Federal
Emergency
Management
Agency
and
National
Institute
of
Building
Safety
sponsored
code,
HAZUS,
developed
to
provide
a
national
stan-
dard
for
earthquake
loss
estimation.
We
have
incorporated
a
recently
developed
soils
map
for
the
state
of
California,
a
preliminary
map
of
liquefaction
suscepti-
bility
and
hazard
curves
for
each
of
5858
census
tract
centroids.
Seven
soil
con-
ditions
are
tied
to
the
1997
Uniform
Building
Code
soils
classification
scheme,
that
is
based
on
shear-wave
velocity
in
the
upper
30
meters.
Liquefaction
sus-
ceptibility
is
based
on
previous
Division
of
Mines
and
Geology
earthquake
planning
scenarios.
Building
inventory
and
fragility
relations
are
supplied
in
HAZUS.
Preliminary
results
suggest
about
$5
billion
(1990
dollars,
1990
inventory)
in
total
loss
annually
statewide.
Of
that,
about
3
billion
is
direct
structural
and
non-structural
damage
to
structures,
$1.5
billion
to
low-rise
wood
frame
structures.
The
remaining
$2
billion
is
contents
losses
and
direct
economic
losses.
The
$5
billion
represents
an
effective
annual
tax
of
about
$200
per
year
for
each
and
every
Californian.
.
Los
Angeles
County
is
subject
to
the
greatest
expected
total
annual
loss
of
over
$1.5
billion,
fully
30%
of
the
state-
wide
total.
This
results
principally
from
the
great
exposure
in
metropolitan
Los
Angeles
County.
Alameda
County
is
a
distant
second,
with
$424
million
annual
loses.
Loss
estimates
have
a
high
uncertainty
in
the
earthquake
hazard
model,
structure
fragility
and
structure
inventory.
We
estimate
an
uncertainty
of
at
least
a
factor
of
4
at
2
standard
deviations.
SISPRO:
A
Scientific
Program
to
Quantify
Seismic
Effects
on
a
Subsurface
Nuclear
Waste
Installation
RODRIGUES,
D.,
CEAJDIF/Laboratoire
de
Detection
et
de
Geophysique,
BP
12,
91680
BRUYERES-LE-CHATEL,
FRANCE,
rodrigues@dase.bruyeres.cea.fr
In
1991,
the
french
Government
decided
to
fund
a
15
years
research
program
in
order
to
establish
a
long
term
solution
for
nucl6r
waste
disposal.
More
recently,
in
the
framework
of
this
program,
it
has
been
asked
to
study
viability
of
an
interim
storage
in
a
subsurface
installation,
at
a
depth
of
typically
few
10
meters.
For
this
concept,
seismic
aspects
have
to
be
carefully
taken
into
account,
to
show
for
instance
how
the
seismic
exposure
is
modified
when
the
installation
is
builded
in
the
subsurface
domain.
More
generally,
3
axis
of
research
are
inspected
to
circumvent
and
predict
seismic
effects
:
analysis
of
in
situ
seismic
records
in
depth,
centrifugal
experiments
and
numerical
modeling.
Subsurface
seismic
displacements
will
be
recorded
in
a
well
known
area,
in
the
south
of
France,
which
has
been
intensively
prospected.
Boreholes,
with
mechanical
characterization
of
the
geological
layers,
and
seismic
campaigns
have
been
con-
ducted
to
establish
a
3D
geological
map.
Seismic
records
will
be
used
to
quan-
tify
the
rule
of
different
parameters,
such
as
magnitude,
directivity
and
depth
of
local
earthquakes,
in
the
attenuation
with
depth
of
seismic
waves
and
to
validate
numerical
modeling.
A
centrifugal
machine,
capable
to
apply
a
100
G
accelera-
tion
to
a
1
ton
object,
will
be
used
to
make
strong
motion
experiments
on
an
small
size
model,
representative,
when
centrifugated,
of
large
scale
size
effects.
3D
finite
differences
and
strong
motion
codes
are
used
in
the
numerical
domain.
Their
synthetic
results
will
be
validated
with
records.
They
are
the
basis
for
the
methodology
which
will
be
builded
in
order
to
be
applicable
to
any
site
of
interest.
This
is
a
several
years
program
designed
to
assess
methods
to
tackle
seismic
hazard
for
nuclear
waste
disposal,
seen
from
seismologists
point
of
view.
First
results
from
records
analysis
and
numerical
simulations
will
be
shown.
New
Scenario
Earthquake
Hazard
Maps
for
the
San
Francisco
Bay
Area
SCHNEIDER,
J.
F.,
Impact
Forecasting,
Chicago,
IL
60606,
jschneider@aon.com
;
SILVA,
W.
J.,
GREGOR,
N.,
LI,
S.,
Pacific
Engineering
&
Analysis,
El
Cerrito,
CA
94530,
pacific@crl.com
;
and
WRIGHT,
D.,
URS
Greiner,
Oakland,
CA
94607,
douglas_wright@urscorp.com
We
have
developed
detailed
seismic
hazard
maps
for
two
major
earthquake
sce-
narios
in
the
San
Francisco
Bay
area.
These
maps
incorporate
the
effects
of
source
rupture
directivity,
crustal
wave
propagation
and
near-surface
soil
response,
as
well
as
the
associated
uncertainty.
The
maps
will
be
useful
to
a
broad
audience
of
earthquake
risk
analysts
to
provide
improved
estimates
of
ground
motion
hazard
for
input
to
probable
maximum
loss
analysis,
structural
designs
and
building
codes,
and
urban
disaster
planning.
The
maps
present
estimates
of
ground
motion
hazard
for:
a
474-km
long
M
7.9
on
the
four
northern
segments
of
the
San
Andreas
(repeat
of
1906);
and
an
87-km
long
M
7.1
on
the
northern
and
southern
segments
of
the
Hayward
fault.
To
generate
rock
motions,
the
approach
combines
the
results
of
a
numer-
ical
ground
motion
simulation
model
with
attenuation
relations.
In
the
numer-
ical
simulation
model,
the
earthquake
rupture
and
crustal
wave
propagation
are
represented
using
a
stochastic
finite-fault
simulation
method
with
random
vibration
theory
(RVT)
to
generate
suites
of
response
spectra
for
each
scenario.
Four
alternative
attenuation
models
were
used
that
are
appropriate
and
com-
monly
used
to
estimate
ground
motions
in
California.
To
accommodate
the
effects
of
near-surface
amplification
in
soil
(or
soft
rock),
we
developed
fre-
quency-
and
amplitude-dependent
amplification
factors.
These
factors
take
into
account
depth
to
bedrock
(for
alluvial
sites),
soil
type
(based
on
near-surface
geology),
and
nonlinear
amplification
effects
at
high
strains.
The
amplification
factors
were
applied
to
rock
motions
across
a
dense
geographic
grid,
then
aver-
aged
across
models
at
several
discrete
spectral
periods.
The
results
were
con-
toured
to
yield
hazard
maps
for
5%
damped
response
spectra
at
periods
of
0
sec
(i.e.,
PGA),
0.3
sec,
and
1
sec,
and
at
the
median
and
1-sigma
levels.
Incorporating
Site
Response
into
Earthquake
Hazard
Microzonation
Maps
SILVA,
Walter,
Pacific
Engineering
&
Analysis,
El
Cerrito,
CA
94530;
WONG
,
Ivan,
URS
Greiner
Woodward
Clyde,
Oakland,
CA
94607;
LI,
Sylvia
and
GREGOR,
Nick,
PE&A;
WRIGHT,
D.,
URSGWC
As
part
of
a
USGS-supported
program
to
develop
microzonation
maps
for
the
Portland,
OR,
and
Salt
Lake
City,
UT,
metropolitan
areas;
the
Albuquerque-
Santa
Fe,
NM,
corridor;
and
the
Wasatch
Front,
UT,
we
have
incorporated
the
site
response
of
unconsolidated
sediments
and
underlying
rock
to
compute
surf-
icial
ground
motions.
All
four
areas
are
located
in
alluvial
basins
and
so
site
response
effects
are
expected
to
be
significant
in
future
large
earthquakes.
Based
on
the
available
geologic
data,
we
define
site
response
categories
baed
on
surfi-
cial
lithology
and,
when
data
are
available,
the
total
thicknesses
of
unconsoli-
dated
units
above
a
specified
reference
rock
datum.
Each
category
was
further
subdivided
to
account
for
ranges
of
thicknesses,
e.g.,
10-50
ft.,
50-100
ft.,
etc.
Each
site
category
is
characterized
by
an
average
shear-wave
velocity
profile;
based
on
this
profile,
30
randomized
profiles
are
computed
to
account
for
the
horizontal
and
vertical
variability
in
velocities
and
category
thicknesses
using
a
correlation
model
developed
by
G.
Toro.
Recently
developed
shear
modulus
reduction
and
damping
curves
(e.g.,
Silva
et
al.,
1997)
are
assigned
to
the
vari-
ous
site
categories
to
account
for
strain-dependent
nonlinear
soil
response
and
uncertainties
in
nonlinear
properties.
Based
on
the site
category
profiles
and
degradation
curves,
amplification
factors
are
calculated,
using
the
stochastic
numerical
ground
modeling
approach
coupled
with
an
equivalent-linear
meth-
odology.
Amplification
factors
are
for
5%-damped
acceleration
response
spectra
for
each
site
category
and
are
developed
for
expected
rock
outcrop
peak
acceler-
ations.
The
point-source
stochastic
methodology
is
used
to
generate
rock
spec-
tra
for
a
controlling
earthquake,
which
are
then
propagated
up
through
the
velocity
profiles.
The
event
is
placed
at
several
distances
to
produce
expected
outcropping
rock
peak
accelerations
f
0.05
to
1.25
g.
In
the
Portland
metro-
politan
area,
unconsolidated
units
consisted
of
alluvial
and
flood
deposits
up
to
500
ft.
thick
atop
Columbia
River
basalt.
Amplification
factors
ranged
from
0.5
to
1.4
g
for
peak
horizontal
acceleration,
and
0.6
to
2.2
g
for
1.0
second
spectral
acceleration
relative
to
Columbia
River
basalt.
At
high
ground
motions
(>
0.5
g),
nonlinear
site
response
is
significant.
A
Comparison
of
Methodologies
to
Achieve
a
Site-specific
PSHA
SILVA,
W,
Pacific
Engineering
and
Analysis,
311
Pamona
Ave,
El
Cerrito,
CA
94530,
pacific@crl.com
;
LEE,
R.,
Bechtel
Savannah
River
Company,
P.O.
Box
616,
Aiken,
SC
29809,
richard02.lee@srs.gov
;
McGUIRE,
R.,
Risk
Engineering
Inc.,
4155
Darley
Ave,
Ste
A,
Boulder,
CO
80303,
mcguire@riskeng.com
;
CORNELL,
A.,
Stanford
Univ,
Dept
of
Civil
Eng,
Terman
Engineering
Center,
Stanford,
CA
94305,
cornell@ce.stanford.edu
.
The
objective
in
developing
site
specific
soil
motions
for
engineering
design
is
to
produce
seismic
demands
that
reflect
a
desired
hazard
level
or
degree
f
con-
servatism
that
is
uniform
across
structural
frequency.
An
essential
aspect
of
this
process
is
the
accommodation
of
appropriate
degrees
of
uncertainty
and
vari-
ability
in
source,
path,
and
site
processes.
The
usual
approach
t
developing
site
specific
soil
motions
involves
defining
regionally
generic
rock
(or
very
firm
con-
ditions)
outcrop
motions
and
then
performing
site
response
analyses
to
accom-
modate
the
effects
of
local
soils.
In
this
approach
the
hazard
level
is
usually
set
at
the
base
of
the
soil
column
(in
defining
the
control
motions)
and
the
actual
hazard
level
of
the
resulting
soil
motions
is
generally
poorly
known.
To
provide
conservatism,
that
is
to
insure
that
the
resulting
soil
motions
are
not
likely
to
reflect
a
lower
hazard
level
than
the
control
motions
at
some
frequencies,
para-
metric
site
response
analyses
are
performed
to
incorporate
both
uncertainty
and
variability
in
dynamic
material
properties
as
well
as
site
response
model
defi-
ciencies.
The
resulting
suite
of
soil
motions
is
then
either
smoothly
enveloped
or
mean
values
estimated.
Since
the
effects
f
site
variability
have
been
counted
twice,
once
in
developing
the
control
(rock
outcrop)
motions
and
again
in
the
parametric
site
response
analyses,
the
resulting
soil
motions
can
reflect
signifi-
cantly
higher
hazard
levels
than
desired
as
well
as
hazard
levels
that
vary
with
Seismological
Research
Letters
Volume
71,
Number
1
January/February
2000
247