of 14
1
Variations in Ground Motion Amplification in the Los Angeles Basin
due to the 2019 M7.1 Ridgecrest Earthquake: Implications for the Long
-
Period Response of Infrastructure
Monica D.
Kohler
, Ph.D.,
1
Filippos Filippitzis, Ph.D.
,
1
Robert Graves, Ph.D.
,
2
Anthony
Massari, Ph.D.
,
P.E.,
3
Thomas Heaton, Ph.D.
,
1
Robert Clayton, Ph.D.
,
4
Julian Bunn, Ph.D.
,
5
Richard Guy, Ph.D.
,
4
and K. Mani Chandy, Ph.D.
6
1
Division of Engineering and Applied S
cience, Department of Mechanical and Civil
Engineering, California Institute of Technology, Pasadena, CA 91125
;
email:
kohler@caltech.edu
2
U.S. Geological Survey, Pasadena, CA 91106
3
Department of Engineering Civil, Environmental and Geodetic Engineering,
The Ohio State
University, Columbus, OH,
43210
4
Division of Geological and Planetary Sciences, Seismological Laboratory, California Institute of
Technology, Pasadena, CA 91125
5
Division of Physics, Math and Astronomy, Center for Data
-
Driven Discovery,
California
Institute of Technology, Pasadena, CA 91125
6
Division of Engineering and Applied Science, Department of Computing and Mathematical
Sciences, California Institute of Technology, Pasadena, CA 91125
ABSTRACT
Coherent patterns and large variations in ground shaking amplification were observed in the Los
Angeles basin during the 2019 M7.1
Ridgecrest earthquake
. In particular, 3 s to 6 s re
sponses
showed variations due
to shallow basin geological structure that h
ave implications for the
response
to large earthquakes
of mid
-
rises, high
-
rises, long
-
span bridges, and fuel storage tanks
,
even if epicentral distances are several hundred kilometers.
The Ridgecrest strong
-
motion data
were recorded by
seismic stations from the spatially dense Community
Seismic Network, the
Southern California Seismic Network, and the California Strong Motion Instrumentation
Program. The mainshock observations are compared at the same locations with ground motion
simula
tions to examine the regions that experienced the largest shaking, and to investigate the
geological sources of large
-
amplitude shaking. The simulations were computed for the two most
commonly
-
used regional community seismic velocity models, CVM
-
S4.26.M01
(‘CVM
-
S’)
and
CVM
-
H 15.1.0
(‘CVM
-
H’)
. Both observations and simulations are used in dynamic analysis
with a finite
-
element model of an existing high
-
rise with ~6
-
second fundamental horizontal
periods, located in downtown Los Angeles. The geographical varia
tion in maximum story drift,
story
-
level
shear force, and
story
-
level moment
values
suggest that the excitation of a
2
hypothetical high
-
rise located in an area characterized by the largest
6
-
s PSA values
could be
significantly
larger than in a downtown Los
Angeles location
. G
round motion simulations using
the CVM
-
H velocity model
more closely predict
the
long
-
period site amplifications in greater
Los Angeles
, particularly in the south
-
central San Fernando Valley,
than simulations using
CVM
-
S.
INTRODUCTION
The M7.1 Ridgecrest earthquake added to a growing list of
large
-
magnitude
California
earthquakes that have produced
amplified
long
-
period ground shaking in greater Los Angeles,
even when epicentr
al distances are large (~200
+
km)
.
A pattern of geographical distribution of
long
-
period amplification
is present
for the
largest
(magnitude > 7)
southern California
earthquakes
that have occurred
since the
advent of modern, dense
,
seismic instrumentation
.
Specifically, t
he 1992
Mw7.3
Lande
rs
,
the 1999
Mw7.1
Hector Mine
,
and the 2010
Mw7.2
El
Mayor
-
Cucapah earthquakes showed large long
-
period motions in the Los Angeles basin
(Wald
and Graves, 1998; Grazier et al.,
2002; Hatayama and Kalkan, 2011), with associated distances
of ~170 km for Lan
ders, ~190 km for Hector Mine, and ~340 km for El Mayor
-
Cucapah (Fig. 1).
For the more recent and more densely recorded 2010 El
-
Mayor
-
Cucapah earthquake,
Hatayama
and Kalkan (2011) found
multiple
zones of strong amplification in the L
os Angeles
basin, wit
h
the largest values
occurring in
a region in the western margin of the
central Los Angeles
basin,
from
analysis of
long
-
period (
4 s) Fourier spectra
relative to bedrock stations.
Recent studies of
peak ground motions and pseudo
-
spectral accelerations in
the Los
Angeles basin from the 2019 M7.1 Ridgecrest earthquake illustrate how spatially coherent
behavior occurs for periods
3 s
, and that the maximum site amplification factor is 10 for 6 s
(Kohler et al. 2020; Filippitzi
s et al. 2021). These findings
we
re computed from the ratio of
response spectral values combined for the two orthogonal horizontal directions, and are relative
to the average of three bedrock sites in the Santa Monica Mountains, the San Rafael Hills
,
and
the San Gabriel Mountains. Speci
fically, for long periods, the maximum peak ground motions
and site amplifications were found in the western part of the Los Angeles basin and in the south
-
central
San Fernando Valley sedimentary basin.
The strong
-
motion data used by Kohler et al.
(2020) a
nd Filippitzis et al. (2021) came from the
Southern Cal
ifornia Seismic Network (SCSN)
,
the California Strong Motion Ins
trumentation Program (CSMIP), and the
Community Seismic
Network
(CSN).
In this study, r
ather than
comparing
observed vs. simulated
g
round motions and their
spatial variations
,
we
investigate several engineering
param
eters
inter
-
story
drift, story
-
level
shears and story
-
level moments
for a common high
-
rise structure type
with long resonant
periods
.
One
goal
is
to examine the geograp
hical variations in predicted
maximum drift
s
, shear
force
s
, and moment
s
, and to compare these values
for
observed and simulated time histories
from the 2019 Ridgecrest earthquake response in urban Los Angeles
.
A second goal
is to
map
out region
s that experienced the largest of these values
in order
to m
ake a prediction for future
3
long
-
period
response
that
may occur
for large
-
magnitude earthquakes
.
We used the observed and
simulated Ridgecrest ground moti
ons in
linear
dynamic anal
ysis
with
a fini
te
-
element model
already developed and validated
for a
n existing
52
-
story high
-
rise
located in downtown Los
Angeles. In our
analysis
, we computed inter
-
story drift,
story
-
level
shear
force
s
, and story
-
level
moment
s
using this model
, where maxima in these v
alues are measures of building damage
potential. The geo
graphical variation in
these
values were compared with the response spectra
results described
in
Kohler et al. (2020) and Filippitzis et al. (2021
)
to investigate how well the
response spectra predict
ed building deformation, considering both observations and simulations.
Figure 1.
Topographic map of Southern California
.
Yellow stars show epicenters of
1992
M7.3
Landers,
1999
M7.1
Hector Mine, 2010
M7.2
El M
ayor
-
Cucapah and
2019
M7.1
Ridgecrest earthquakes. Red dashed box shows focus area of Ridgecrest study.
Black
dashed line shows shortest source
-
to
-
receiver path between 2019
M7.1
Ridgecrest epicenter
and Los Angeles.
LA=Los Angeles. WLA=West Los Angeles, SFV=San Fernando Valley,
LAB
=Los Angeles Basin, SGV=San Gabriel Valley.
GROUND MOTION SIMULATION
METHOD
The g
round motion simulations for the
2019
M7.1 Ridgecrest
earthquake
were
computed
with
the
Graves
(1996)
3D finite
-
difference method, which use
s the
Graves and Pitarka (2020)
stochastic finite
-
fault
rupture method.
The rupture model uses the Graves and
Pitarka
(2016)
kinematic rupture generator method
and is constrained by near
-
fault ground motion observations
.
4
The simulations use
a minimum shear
-
wave velocity
of 500 m/s and 100 m grid spacing,
resulting in velocity ground motions with energy content containing frequencies up
to 1 Hz.
Anelastic attenuation
is
defined by
=
50
where
is shear
-
wave velocity in km/s
, and
is related to
damping
d
, by
d
=1/2
.
The simulations were
carried out using
3D
seismic velocity models
CVM
-
H15.10.0
(Shaw et al., 2015)
(
hereafter referred to as ‘CVM
-
H’
)
and
CVM
-
S4.26.M01
(Lee et al., 2014)
(hereafter referred to as ‘CVM
-
S’)
.
Although both models
cover approximately the same region
and depth
range
, there are fundamental differences in the types of data that constrain parts of the
models, particularly in the Los Angeles sedimentary basin (‘LAB’ in Fig. 1).
Fig. 2 comp
ares the
results for 6
-
s pseudo
-
spectral accelerations
(PSA)
using 2% damping for
the M7.1 Ridgecrest
earthquake using the method described in Filippitzis et al. (2021). This figure illustrates the
geospatial variation in
the PSA values computed
for data (Fig. 2a) and synthetic gro
und mo
tions
computed using CVM
-
H (Fig. 2b) and CVM
-
S
(Fig. 2c).
(a)
(b)
(c)
Figure 2
.
Geospatial variations in 6
-
s PSA value
s with
2% damping using M7.1 Ridgecrest
acceleration
s for: a)
D
ata; b
) CVM
-
H
simulations;
c
) CVM
-
S
simulations. PSA amplitude
scales are identical. Analysis is from Filippitzis et al. (2021).
HIGH
-
RISE
DYNAMIC ANALYSIS
METHOD
The high
-
rise
finite
-
element
model is based on an existing 52
-
story (+ 5 basement levels) dual
steel moment and braced frame
system building located in Los Angeles. This building’s lateral
system consists of a braced frame core surrounded by a steel moment frame. The floor plans
contain various setbacks and notches along the building’s vertical profile. The structural system
con
sists of three major components: an interior concentrically braced core, outrigger beams
spanning the core to the building perimeter, and eight exterior outrigger columns. The beams
support gravity loads, act as ductile moment
-
resisting beams between the c
ore and exterior frame
columns, and enhance the overturning resistance of the building by engaging the perimeter
columns to the core columns (Taranath, 1997). The first
two
translational resonant
period
s of this
building in the
horizontal directions are ap
proximately
5.7 s and 1.7 s (Kohler et al., 2016).
In
5
dual lateral systems such as this, the
overall story
stiffness is
controlled by the interpl
ay between
the
mome
nt frames and the braced frames
,
whose properties vary
vertically.
At the bottom of the
buil
ding, t
he stiffness
contribution
from
the moment frame
s
relative to the braced frame
s
is
larger than at the top
, resulting in
vertical variations in relative story
-
level shear and story
-
level
moment
.
These in turn will contribute
complexity
to the maximum
inter
-
story
drift values
when
comparing location to location
.
T
he computations described below
are carried out with
a
linear
finite
-
element model of
the building
which wa
s developed based on detaile
d information obtained from a complete set of
structural drawings
. The major structural and connection elements obtained from the drawings
are modeled using object
-
based physical
-
member modeling, such as built
-
in steel sectio
ns and
braces (ETABS,
Computers and Structures, Inc.
). More details on the model construction and
validation with local earthquake data are described in Kohler et al. (2016).
Here we focus on
linear elastic dynamics for the recorded
acceleration time
serie
s from observed and sim
ulated
M7.1 Ridgecrest data for
20
seismic
station
s
in the Los Angeles basin
that had the largest
recorded 6
-
s PSA values (
Table 1
)
.
The linear
-
elastic dynamic response to the earthquake
acceleration inputs in the EW and NS directions are computed using the ETABS model.
A
broad
band Butterworth filter
i
s applied to the
observations and
simulation r
esults for
frequencies
between
0.1 and
25 H
z. The res
ponse is modeled in
a fixed
-
base coordinate system.
RESULTS
W
e compute
commonly
-
used
engineering parameters
in order to
examine the geographical
variations in
their values for the largest of the observed long
-
period M7.1 Ridgecrest datas
ets
,
and for their associated simulated accelerations.
In addition, we examine the vertical variations
in these values within the building
for complexity from
multi
-
modal
contributions.
W
e focused
on the
20
seismic stations that produced the largest 6
-
s
PS
A
values
for the high
-
rise dynamic
analysis
. The goal was to show
how well the
maxim
a
in predicted (
i.e.
computed) drifts,
story
-
level
shear and
story
-
level
moments for a high
-
rise would compare
in station locations
with the
largest
6
-
s PSA values
. These
20
stations consisted of SCSN, CSMIP and CSN stations, mostly
located in West Los Angeles and the south
-
central San Fernando Valley
(Table 1)
.
Fig. 3a shows the maximum building
-
EW component inter
-
story drift results of the
linear
analysis for the high
-
rise, using M7.1 Ridgecrest recorded data as time history input into the
ETABS model. The curves show the maximum values for each floor, independent of the time at
which the maximum occurred; thus they are not all from the sam
e instance of time. All 20 station
inputs resulted in parameters that are well within the linear regim
e; however, while most values
a
re small, one station (CE.24805 in Northridge, CA) resulted in significantly larger parameters
than the rest, and primarily
in the building
-
EW direction. Fig. 3b shows the analogous results for
seismic velocity model CVM
-
H, and Fig. 3c shows the results for CVM
-
S. The CVM
-
H
simulations also predict the largest inter
-
story drift values for a location in the San Fernando
Valley
(CE.
Q0057
in Van Nuys, CA); however, Figs. 3b and 3c indicate that the simulations,
6
particularly those that use CVM
-
S,
significantly
underpredict the long
-
period ground motions in
general in both the San Fernando Valley and West Los Angeles.
Table 1
.
Se
ismic stations
in greater Los Angeles
with maximum
6
-
second
PSA values
from
the M7.1 2109 Ridgecrest earthquake
.
PSA=pseudo
-
spectra
l acceleration for 6 seconds
with
2% damping.
PGA=Peak ground a
cceleration.
Network code
s:
CE=CSMIP,
CI=Southern California S
eismic Network, CJ=Community Seismic Network.
Fig
s
. 3
abc show several
vertical profile curves
(i.e.
different
seismic station
s
) that cross
over each other
as a function of building height
.
This may indicate the impact of lowest
-
frequency vs. higher
-
frequency modal
contributions to the relative
displacements within the
building.
The
building
-
NS component results for all three datasets are smaller
-
amplitude
,
and
show similar geographical varia
tion across seismic stations
,
as well as similar relative amplitudes
between observed data and simulations
.
As with the EW results, the NS drift curves also show
some cross
-
over, indicating low
-
frequency vs. higher
-
frequency modal contributions to the NS
relative displacements
that
are
dependent on
the frequency content of the
time history input
.
The
results
for
maximum
EW
story
-
level shear
force
and moment
values
can be seen in
Fig
s
. 4
and 5
, and as before are not all from the same time instance
.
The shear forces
are
computed by ETABS using constitutive relations determined by the structural elements
making
up the model, together with
the computed deflection
s, integrated at every story along each
direction of interest.
Here, we refer to the EW moments as those that are caused by forces
Station
name
Net
-
work
Station location
Station coordinates
PGA
(%g)
6
-
sec
PSA
(
%
g
)
24805
CE
Northridge, CA
34.2289
-
118.5289
4.31
5.61
14221
CE
Manhattan Beach, CA
33.8883
-
118.4098
1.49
4.97
24806
CE
Winnetka, CA
34.2218
-
118.5714
2.83
4.61
LAF
CI
Torrance, CA
33.8689
-
118.3314
1.22
4.3
0
24866
CE
Reseda, CA
34.1934
-
118.5484
3.25
4.14
PDR
CI
Playa Del Rey, CA
33.9627
-
118.4370
1.54
3.92
T001227
CJ
Southcentral Los Angeles, CA
33.9773
-
118.2843
2.07
3.88
T001213
CJ
Southcentral Los Angeles, CA
33.9713
-
118.2761
2.01
3.82
T000887
CJ
View Park
-
Windsor Hills, CA
34.0059
-
118.3357
2.25
3.77
Q0057
CI
Van Nuys, CA
34.1876
-
118.4771
3.45
3.77
T001211
CJ
Southcentral Los Angeles, CA
33.9776
-
118.2758
2.24
3.74
T001214
CJ
Southcentral Los Angeles, CA
33.9696
-
118.2670
2.53
3.74
LCG
CI
Ladera Heights, CA
34.0003
-
118.3779
1.81
3.73
24013
CE
Canoga Park, CA
34.1941
-
118.5888
2.43
3.59
BHP
CI
View Park
-
Windsor Hills, CA
33.9905
-
118.3617
1.51
3.58
T001228
CJ
Southcentral Los Angeles, CA
33.9862
-
118.2939
2.06
3.58
T001229
CJ
Southcentral Los Angeles, CA
33.9858
-
118.2884
1.94
3.57
T001215
CJ
Southcentral Los Angeles, CA
33.9680
-
118.2604
1.64
3.56
T001212
CJ
Southcentral Los
Angeles, CA
33.9796
-
118.2680
1.93
3.49
14820
CE
View Park
-
Windsor Hills, CA
33.9934
-
118.3425
1.99
3.47
7
applied in the EW direction.
Fig
s
. 4
a
and 5a show
the
story
-
level shear
an
d moment
results of the
linear analysis for the high
-
rise using the M7.1 Ridgecrest recorded data as input.
Fig
s
. 4b
and 5b
show
the analogous results for seismic velocity model CVM
-
H, and Fig
s
. 4c
and 5c show
the
results for CVM
-
S.
As with inter
-
story dri
ft analysis, the CVM
-
H simulations predict the largest
story
-
level shear
and moment
values for
station
CE.Q0057 located
in Van Nuys
, CA
; however,
the simulations also
significantly
underpredict the overall story
-
level shear
and moment
values
for all 20
locations.
As with drift, Fig
s
. 4
abc
also
show several
vertical profile
curves that cross
Fi
gure 3
.
Maximum
EW
i
nter
-
s
tory drift computations using
finite
-
element
model
of high
-
rise
with
M7.1 Ridgecrest
time histories from
:
a)
D
ata; b
) CVM
-
H
simulations
;
c
) CVM
-
S
simulations
.
H
orizontal ampli
tude scales
are identical
.
S
tation locations
are shown in
Table 1
.
8
over
others, vertically within the building, indicating
the
varying contributions of the
lowest
-
frequ
ency vs. higher
-
frequency modes
to the
shear force distribution. The NS
-
component res
ults
for all three datasets
show similar geo
graphical variation across seismic stations
,
as well as
similar relative amplitudes between observed data and simulations.
Figure 4
.
Maximum
EW
s
tory
-
level
shear computations
usi
ng
finite
-
element
model of high
-
rise
with
M7.1 Ridgecrest time histories from
: a)
D
ata; b
) CVM
-
H
simulations
;
c
) CVM
-
S
simulations
.
Horizontal amplitude scales
are identical
.
Station locations are shown in
Table 1.
9
Figure 5
.
Maximum EW story
-
level moment computations using
finite
-
element
model of
high
-
rise
with
M7.1 Ridgecrest time histories from: a)
D
ata; b) CVM
-
H simulations;
c)
CVM
-
S simulations
.
Horizontal amplitude scales
are identical
. Station locations are
in
Table 1.
The
geo
spatial variations in maximum values for
drift, and
story
-
level
shear
forces
and
momen
ts
are shown in Figs. 6
-
8
for the
20
stations
listed in Table 1. Fig
s
. 6
a
, 7a and 8a
show
the
variations in values fo
r the observed data.
Figs. 6b
,
7b,
and
8b
and
Figs.
6
c
,
7c,
and
8c
show the
variations for the simulation input
s
using CVM
-
H and CVM
-
S
respectively
. The smaller gray