strong motion data processing in taiwan and its engineering

STRONG MOTION DATA PROCESSING IN TAIWAN AND
ITS ENGINEERING APPLICATION
CHIN-HSIUNG LOH
Professor
Department of Civil Engineering, National Taiwan University
Taipei, Taiwan 106-17
e-mail: lohc0220@ccms.ntu.edu.tw
ABSTRACT
This paper presents the strong motion data processing and its engineering application from
TSMIP and TREIRS strong motion array in Taiwan. The TREIRS system provides almost real
time earthquake intensity distribution and the TSMIP data provides useful ground motion data
for both earth science and engineering study. Since the sampling rate of these 700 seismographs
of TSMIP array is 200/250 per second with resolution 16/24 bits, then the baseline correction is
not necessary applied to these data. Three different approaches to calculate ground velocity and
displacement from acceleration data are proposed. Engineering application of these free field
ground motion data covers: (i) Develop seismic design spectrum for Sa-value at short period and
long period, (ii) Generate spatial ground motion for Earthquake Emergency Responses, (iii)
Develop regional phase spectrum for simulation of ground motion, (iv) Study probabilistic
dynamic response analysis of non-stationary excitation.
INTRODUCTION
Taiwan is located at the active arc-continent collision region between the Luzon arc of the
Philippine Sea plate and the Eurasin plate. The Philippine Sea plate is colliding onto the
Eurasian continent at a rate of 7-8 cm/yr, resulting high seismicity in this region. Seismic
disaster in Taiwan over the last twenty years, particularly the 1999 Chi-Chi earthquake, had
caused a significant number of victims and direct physical damages. Figure 1 shows the
distribution of earthquake hypocenter and the active fault in Taiwan are shown. At present time
there are over 50 active faults were identified in Taiwan. Therefore, it is important to establish
seismic monitoring system in this area for earthquake hazard mitigation. Due to high seismic
activities in Taiwan the Central Weather Bureau (CWB) launched a Taiwan Strong Motion
Instrumentation Program (TSMIP/CWB) program in 1990, so as to increase the precision of
earthquake information determination. In the beginning of 1990, CWB began to install a seismic
network that includes 75 stations around Taiwan area, the Taiwan Rapid Earthquake Information
Figure 1: (a) 3-D distribution of earthquake hypocenter in Taiwan area; (b) Identified active faults in
Taiwan.
75 TREIRS Stations
650 Free-Field Stations
25
Latitude (N)
24
23
22
120
Figure 2:
121
Longitude (E)
122
Distribution of free-field strong motion instrumentation (over 700 instruments) and 75
real-time system under TSMIP/CWB program in Taiwan.
Release System, TREIRS (Lee and Shin, 1997). Digital telemetry and digital recording of
three-component high-quality force balanced accelerometers were used in this system for
seismic monitoring operations. This system can routinely release the location and magnitude of
a strong earthquake as well as the distribution of intensity about 10 seconds (or less) after the
occurrence of an inland earthquake. In addition, there are more than 700 free-field strong motion
observation stations island-wide distributed, as shown in Figure 2. The sampling rate of these
700 seismographs is 200/250 per second with resolution 16/24 bits. Site investigation at each
observation station was also launched since 2000. Using suspension PS-logger the bole hole
data was collected at each site of seismograph. The N-value, P-wave and S-wave velocities
along the depth are estimated. This paper presents the data processing technique and engineering
application of the strong motion array data collected from TSMIP and TREIRS array in Taiwan.
ROUTINE PROCESSING AND INTEGRATION OF RECORDS
There are over 5 different series of seismograph in TSMIP strong motion array, such as:
A800, A900, …etc. These seismographs provide accurate signal with frequency band up to 50
Hz. Ground motion data collected by CWB did not do any correction at all except the DC
correction. For engineering application three different approaches have been used to estimate
ground velocity and displacement from the acceleration record. A brief description of these
methods is made:
Method 1: Since the sampling rate of free-field seismograph is 200/250 per second with
resolution 16/24 bits, the recorded acceleration with frequency smaller than 50Hz represents
almost the true ground acceleration, and there is no need to conduct any baseline correction and
only DC correction is needed. To evaluate the velocity and displacement from the acceleration
record only a least square fit and/or low pass filter of the direct integrated velocity was used, as
shown in Figure 3. Two sets of ground acceleration data were selected to evaluate the method:
Chi-Chi earthquake data form station: TCU068 (near-fault ground motion data) and CHY088.
Figure 4a shows the result from direct integration of acceleration from station TCU068 and
Figure 4b shows the direct integration with least square fit on the velocity data. Comparison on
the calculated ground displacement using the result from Figure 4b provides more realistic
displacement because a permanent ground deformation was observed at station TCU068.
Similar procedure was also applied to data from station CHY088. Because data collected from
station CHY088 did not contain the characteristics of near-fault ground motion. To obtain
velocity or displacement either direct integration from acceleration or employ least square fitting
the result is almost the same, as shown in Figure 5. This method has been used in all the
engineering application of recorded ground acceleration.
Correct
Accelerogram a6(t)
Low-pass filter
Correct velocity
Least square fit
Integrate a6(t) for v4(t)
v5 (t ) = v4 (t ) − v02 − a3 (t )
v 7 (t ) = v 5 (t ) − v 6 (t )
to get v6(t)
Correct displacement
Low-pass filter to get
d2(t)
d 3 (t ) = d 1 (t ) − d 2 (t )
Integrate for displacement
d1(t)
Figure 3: Integration of strong motion acceleration to obtain velocity and displacement through
least square fit or/and low-pass filter modification.
TC U 068
TC U 068
1000
500
A ccel.(gal)
A c c e l.(g a l)
1000
0
-5 0 0
0
5
10
15
20
25
30
35
40
45
50
500
0
-500
-2 0 0
-4 0 0
0
5
10
15
20
25
30
35
40
45
50
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
Tim e (sec)
35
40
45
50
0
-200
-400
500
500
0
-5 0 0
-1 0 0 0
0
5
10
15
20
25
30
T im e (s e c )
35
40
45
(a)
50
D ispl.(cm )
D is p l.(c m )
0
200
0
V el.(cm /sec)
V e l.(c m /s e c )
200
0
-500
-1000
(b)
Figure 4: (a) Direct integration of acceleration to obtain the velocity and displacement, (b)
Direct integration of acceleration record to obtain the velocity and then linear regression the
velocity record, for record from TCU068.
C H Y 08 8
C H Y 08 8
200
A c c e l.(g a l)
A cc el.(ga l)
20 0
0
-2 0 0
-2 00
0
10
20
30
40
50
60
20
V e l.(c m /s e c )
V e l.(cm /s ec )
0
0
10
20
30
40
50
60
0
10
20
30
40
50
60
0
10
20
30
T im e (s e c )
40
50
60
0
-2 0
-2 0
0
10
20
30
40
50
60
20
D is p l.(c m )
10
D isp l.(cm )
0
20
0
10
0
-1 0
-1 0
0
10
20
30
Tim e (s ec )
40
50
60
(a)
(b)
Figure 5: (a) Direct integration of acceleration to obtain the velocity and displacement, (b) Direct
integration of acceleration record to obtain the velocity and then linear regression the velocity
record, for record from CHY088.
Inp ut A cceleration
a ’ (t )
E stim a te N o ise L evel
F rom P re-event R eco rd
F o u rier tran sform
F {a 1 (t)}
E rror correction on m easu rem ent sy stem :
L ow -pa ss filter: < 25 H z 100 %
H igh-pa ss filter: > 0.2 H z 100 %
A 1 (ω)
a 1 (t) = a’(t) - (noise level)
T ren d rem o val
(u sing lea st sq uare m eth od )
Inv erse Fo u rier T ran sform
O u tp ut: a(t) = F -1 {A 1 (ω)}
V 1 (ω )= A 1 (ω ) ? (1/i ω )
Inv erse Fo u rier T ran sform
O u tp ut: v(t) = F -1 {V 1 (ω)}
D 1 (ω )= V 1 (ω ) ? (1/i ω )
Inv erse Fo u rier T ran sform
O u tp ut: d(t) = F -1 {D 1 (ω)}
Figure 6: Estimation of ground velocity and displacement through filtering analysis in
frequency domain.
Method 2: Frequency domain approach was used in this method. After trend removal (DC
effect) on the acceleration data, data was transform to frequency domain. Both low-pass filter
(<50 Hz) and high pass filter (>0.1 Hz) are applied to the acceleration data. Inverse Fourier
transform is then used to obtain the correct acceleration data, a1 (t ) = F −1 {A1 (ω )} . The velocity
is obtained from the inverse Fourier transform of V1 (ω ) = A1 (ω ) ⋅ (1 ω ) and the displacement is
obtained from the inverse of Fourier transform of D1 (ω ) = V1 (ω ) ⋅ (1 ω ) , as shown in Figure 6.
Method 3: Empirical mode decomposition method was applied to the recorded acceleration.
Separate the decomposed acceleration into two parts: high frequency part and low frequency
part. Integrate the separated signals, aH(t) and aL(t), so as to obtain the fling effect. Figure 7
show the flow chart of the procedure of the method. The result of this analysis using data from
TCU068 (Chi-Chi earthquake) is shown in Figure 8.
Ground
acceleration
Empirical Mode Decomposition
or Wavelet analysis
Separate high frequency signal aH(t)
and low frequency signals aL(t)
Remove lowest frequency signal
From the original signal
Identify fling effect from
the decomposed waves
Figure 7: Application of EMD method to estimate fling effect of ground motion.
T C U 0 6 8 (s 1 1 -s 1 2 ): filte re d
T C U 0 68 (s 1 -s 10 ): filte red
50
A c c e l. (g a l)
A c c el. (g al)
1 00 0
5 00
0
-5 0
-5 00
0
5
10
15
20
25
30
35
40
45
50
V e l. (c m /s e c )
V e l. (c m /se c )
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
T im e (s e c )
35
40
45
50
100
1 00
0
-1 00
-2 00
0
-1 0 0
-2 0 0
0
5
10
15
20
25
30
35
40
45
50
50
1000
D is p l. (c m )
D isp l. (c m )
0
0
-5 0
0
-1 0 0 0
-2 0 0 0
0
5
10
15
20
25
30
T im e (s ec )
35
40
45
50
(b)
(a)
Figure 8: (a) Summation of decomposed acceleration (from No.1 to 10) and the velocity and
displacement from direct integration, (b) Summation of decomposed acceleration (from No.11
to 12) and the velocity and displacement from direct integration,
TC U 068 (s1-s12): filtered
A ccel. (gal)
1000
500
0
-500
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
Tim e (sec)
35
40
45
50
V el. (cm /sec)
200
0
-200
-400
D ispl. (cm )
1000
0
-1000
-2000
(c)
Figure 8: (c) Combine (a) and (b) to obtain the velocity and displacement.
ENGINEERING APPLICATION OF STRONG MOTION DATA
Ground motion data collected from the strong motion array in Taiwan have been applied to
earthquake engineering researches and hazard mitigation. In this section the application of
strong motion data will be introduced which covers the topics on:
(i) Development of seismic design spectrum for Sa-value at short period and long period,
(ii) Spatial ground motion estimation for earthquake emergency responses,
(iii) Simulation of ground motion using phase spectrum,
(iv) Probabilistic dynamic response analysis of non-stationary excitation,
Development of seismic design spectrum For the current development of seismic design
code in Taiwan, the elastic seismic demand is represented by the design spectral response
acceleration SaD corresponding to a uniform seismic hazard level of 10% probability of
exceedance within 50 years (return period of 475 years). The TSMIP strong motion array and
TREIRS real time strong motion array provide accurately observing ground motions from
seismic events. Based on the free field strong motion data collected by the Seismology Center of
the Central Weather Bureau during the past, the scenario earthquake induced intensity measure
(such as PGA or spectral intensity) around the island is calculated. The empirical attenuation
form can be expressed as:
IM = Yr = f (M,R) = b1 eb2 M [R + b4 exp (b5 M)] − b3
(1)
where IM is the intensity measure of seismic demand (it can be expressed as peak ground
acceleration, or spectral acceleration at any specified period); M and R are the earthquake
magnitude and the site-to-source distance, respectively; b1 through b5 are constants given
earthquake magnitude, focal depth, epicenter and/or surface fault rupture. The seismic demands
in terms of peak ground acceleration and response spectrum are calculated for user-specified
scenario earthquake. Table 1 shows the estimated model parameters from Taiwan earthquake
data.
Table 1: Model parameters of attenuation model shown in Eq.(1)
Case
b1
b2
B3
b4
b5
σ ln ( Err )
PGA
0.0036944
1.7537666
2.0564446
0.1221955
0.7831508
0.68-0.75
Sas
0.0097360
1.7348416
2.0857212
0.1136533
0.8003162
0.67-0.75
Sal
0.0027914
1.7730463
2.0419005
0.1154175
0.7713924
0.85
Figure 9 shows the PGA attenuation equation for M=5.0, 6.0 and 7.0, and the PGA data is also
shown in this figure. Data collected from TSMIP array since 1991 is also plot in this figure for
comparison. Normally the attenuation relationships of ground motion parameter predict the
ground intensity in rock site condition. Based on the uniform hazard analysis, the mapped
design 5% damped spectral response acceleration at short periods ( S SD ) and at 1 second ( S 1D )
are determined and prepared for each administration unit of village, town or city level.
PGA-Attenuation Curve, GeoMean[EW,NS)
Data : M = 5.0 --- 7.5; Depth = 0-35 km, Sall
4
3
2
ML=7.065, MW=7.54, Geo-Mean
1E+0
9
8
7
6
5
2
1E+09
8
7
6
5
4
4
3
2
Peak Groung Acceleration, (g)
3
Peak Ground Acceleration, [g]
2
1E-1
1E-19
8
7
6
5
4
3
Ca 2000, M=7
2
Ca 2000, M=6
Ca 2000, M=5
1E-29
8
7
6
5
4
Data, M=7.0-7.5
Data, M=5.5-6.0
2
Data, M=5.0-5.5
4
3
Comparison For Taiwan Campbell Form,
Norman Form and 921 Chi-Chi WQ. data
2
PGA, Taiwan Campbell Form
PGA, Norman Form, S/S
1E-2
Data, M=6.0-7.0
3
9
8
7
6
5
PGA, Norman Form, R/S
9
8
7
6
5
PGA, Norman Form, RHW
921 Chi-Chi EQ. Data
4
1E-39
921 Chi-Chi EQ. Data, HW
3
8
7
6
5
4
921 Chi-Chi EQ. Data, FW
2
PGA, Taiwan, For FW, Ratio = 0.475
3
2
1E-1
1E+0
1E+1
Distance, km
1E+2
1E-3
1E-1
2
3 4 5 6 789
1E+0
2
3 4 5 6 789
1E+1
2
3 4 5 6 789
1E+2
2
3 4 5
Surface Rupture Distance, (km)
Figure 9: (a) PGA attenuation for M=5, 6, and 7. The recorded PGA is also shown in the figure.
(b) Chi-Chi earthquake PGA data is plotted w.r.t. Taiwan PGA attenuation equation.
The TSMIP and
Ground Motion Estimation for Earthquake Emergency Responses
TREIRS systems provide accurately observing ground motions from seismic events. In
cooperate with these data the Earthquake Hazard Assessment Methodology (HAZ-Taiwan) was
developed. This system can estimate the ground motion immediately after the earthquake. This
system is aimed to support the central government and local governments to optimize its post
disaster management such as rescue, recovery and reconstruction. It aims at enabling the disaster
responders to take more effective measures. Under such a goal a simulation system of
earthquake disaster processes is being constructed. For a given earthquake information (i.e.
magnitude and hypocenter distance), the peak intensity of ground motion can be evaluated using
the attenuation model for hard site condition. For other site conditions, the estimation of ground
motion intensity should be modified by site amplification factor. This site amplification factor
can be developed in advance by using ground motion data collected from the strong motion
array data (from Taiwan Strong Motion Instrumentation Program) with different soil conditions.
For each specific site the revised intensity measure Ys can be evaluated using the following
equation (Chang et al. 2002):
Ys = (C0+ys C1)
(2)
where C0 and C1 are the coefficients of the linear regression form between recorded and
estimated intensity measure, as shown in Figure 10. The real-time free-field strong motion data
(from Taiwan rapid Information Release System, TREIRS) may also be used to upgrade the
ground motion estimation. Figure 11 shows the flowchart of the estimation of spatial ground
motion for a given earthquake magnitude and hypocenter. Not only the PGA attenuation
equation was used but also the site amplification factored the real-time TREIRS data are
implemented in this model. This system can also generate the shake map accurately immediately
after an earthquake (within 20 min.). Besides, this system can also be used to perform scenario
earthquake for hazard mitigation. To be more realistic, the scenario earthquake should be
selected based on probabilistic analysis in order to identify the important source areas that have
large contribution factor to each specific site for a prescribed probability.
Under the TSMIP a site investigation project was also launched. By using suspension
PS-logger the bole hole data was collected. This information provides researchers to identify the
site condition at each location of seismograph. One of the important application of these bole
hole data is to develop the relationship between S-wave velocity and SPT N-value in any
particular region, as shown in Figure 12. 45 borehole data were investigated in YunLin, ChiaYi
and TaiNan counties, as shown in Figure 13a. The reported data including soil profiles, SPT-N
values and wave velocities measured by Suspension PS Logger were used to develop an
empirical equation of shear wave velocities especially for alluvium deposits. It is found that
there is a linear relationship between shear wave velocity and depth. After corrected by standard
overburden pressure, the standard penetration test value was added into the linear empirical
equation. The empirical equation fits the measured data very well and the result is very useful to
site effect analysis and(or) other seismic issues, as shown in Figure 13b.
Because of the dense strong motion array (TSMIP), the site
Evaluation of Site Effect
response can be studied in terms of spectra ratio calculated by dividing of the site spectrum by
the reference spectrum estimated for a hypothetical “very hard rock” site. The developed
empirical source scaling and attenuation models can be used for the reference spectra calculation.
Due to the ample amount of free field ground motion recorded by TSMIP, this approach allows
us to evaluate the variability of spectral ratios due to uncertainties introduced by source and
propagation path effects and variability in the site response itself. In Taipei basin, there are about
50 seismograph been installed under TSMIP. It provides the opportunity for this research topic.
Figure 14 shows the comparison between theoretical spectral ratios obtained using 1-D model
and the empirical ratios for stations characterized by different thickness of Quaternary deposits.
Simulation Ground Motion Using Phase Spectrum
In order to generate spectrum
compatible ground motion both target amplitude response spectrum and phase spectrum need to
be presented as a prior. Generally, a primitive method to simulate design ground motions is to
use the phase spectrum from a certain observed ground motion. Actually, the importance of
phase spectrum is illustrated based on an ensemble of ground motion data. TSMIP data provides
this opportunity to develop models for regional phase spectrum. A theoretical derivation on
phase spectrum is introduced by Sato (1999) on the basis of group delay time that is defined as
the derivative of the phase spectrum with respect to circular frequency:
t gr (ω ) =
dΦ (ω )
dω
(3)
The mean value and standard deviation of group delay times within a certain frequency range
express the central arrival time and duration, respectively, of the earthquake motion with
frequency content at such a bandwidth. Therefore, it is much easier to model the group delay
time than to model the phase spectrum directly. The procedure to generate the mean value and
standard deviation of group delay time for each frequency band is shown in Figure 15. Based on
the recorded ground motion from a specific region (same site condition) attenuation equation (as
a function of earthquake magnitude and distance) of the mean and standard deviation of group
delay time for each specific frequency band can be generated. It can be found that the student
t-distribution with a degree of freedom φ=3 can be recognized as the representative distribution
of group delay times within the compact support of Meyer wavelet [Chai et al, 2002]. Based on
this result the phase spectrum can be generated and the simulation of ground motion can
incorporate with this phase spectrum.
Figure 10: Comparison on the estimated (from PGA or Sa attenuation equation) and recorded
PGA, Sa (T=0.3 sec) and Sa(T=1.0 sec). A regression line on the recorded ground intensity is
developed.
T S M IP d ata
E arth q u ak e P aram eters
(M agn itu d e, D ep th ,
E p icen ter etc.)
U p d ate estim ation
u sin g T aiw an
R ap id In form ation
R elease S ystem
IM = Y r = f (M ,R )
= b1 e b2 M [R + b 4 ex p (b5 M )] −b3
A tten u ation M od el
y a S = f (M ,R )
R =
S ite E ffect
M od ification
y sS = fS ( y a ,C 0 ,C 1 )
(R T D S )obs
(R T D S )y
S p atial D istrib u tion of
G rou n d M otion E stim ation
s
P D S = y sS × fS (R ,D 0 ,D 1 )
Figure 11: Flow chart indicates the procedures for estimation of ground motion intensity (PGA,
Sa -value) immediately after the catastrophic earthquake.
CHY001
0
O Y O P S -170
Logge r/R e corder
vs
N
vs
N
10
D e p th (m )
C able H ead
D iskette
w ith D ata
H ead R educer
CHY004
CHY002
vs
N
A rm ored 7 -C o nductor cable
W inch
20
30
U pper (R 2)
R eceiver
0.5 m
D epth refe rence locatio n
for R 1-R 2 analysis:
m id-point of R eceivers
40
0.5 m
Low er (R 1 )
R eceiver
0 10 20 30 40 50 0
1.0 7 m
200
400
0 10 20 30 40 50 0
200
400
0 10 20 30 40 50 0
200
400
1.5 7 m
Joint
D epth refe rence locatio n
for S -R 1 an alysis : m id
point of 3.14 m S -R 1 spacing
1.0 m flexible
Isolation C ylin der
CHY020
0
1.0 7 m
2.1 4 m
Joint
CHY025
CHY021
vs
N
vs
N
vs
N
3.7 m
10
S ource D river
1.0 5 m
W eigh t
T ip
D e p th (m )
C om bined S h and
P -w ave S ource (S )
20
30
O verall Len gth ~ 5 .8 m
N ot to S cale
40
0 10 20 30 40 50 0
200
400
0 10 20 30 40 50 0
200
400
0 10 20 30 40 50 0
200
400
Figure 12: Plot of estimated S-wave velocity and SPT-n value along the soil depth using
suspension PS-logger (shown on the left figure) at location of seismograph: Stations CHY001,
CHY002, CHY004, CHY020, CHY021, and CHY025.
300
v's (after depth corrected)(m/sec)
250
200
150
v's=139.1 + 2.0415 N1
Vs=139.1 + 2.0415N 1
100
50
0
10
20
30
N1(after corrected by overburden pressure)
40
Figure 13: Forty-five bole hole data were collected from YunLin, ChiaYi and TaiNan counties
(left ) to estimate the relationship between S-wave velocity and N1-value (right).
TAP27
TAP03
TAP07
2.0
2.0
A m p lific a tio n
A m p lific a tio n
A m p lific a tio n
2.0
1.0
1.0
0.0
0.0
0.0
2.0
4.0
6.0
8.0 10.0 12.0
1.0
0.0
0.0
2.0
Frequency, Hz
4.0
6.0
8.0 10.0 12.0
0.0
Frequency, Hz
2.0
4.0
6.0
8.0 10.0 12.0
Frequency, Hz
Figure 14: Theoretical and empirical spectral ratios (solid lines: mean and ±1 σ limits) of Taipei
basin calculated using 1-D models and TSMIP Data (Sokolov, Loh, Wen 2000).
Fourier
Transformation
x(t)
xj(t)
tjgr(ω)=∂Φ J(ω)/∂ω
Φ J(ω)
tj
µjtgr
gr(ω)
XJ(ω)
σjtgr
AJ(ω)
Meyer Wavelet Transformation
Mean Value µjtgr¡ G
 Central arrival time
Standard Deviation σjtgr
 Duration
Figure 15: Flowchart for generating the mean value and standard deviation of group delay time.
CONCLUSIONS
Two strong motion recording systems, TSMIP and TREIRS, which include over 750 free
field strong motion instrumentations all over the island, provide valuable data for engineering
application. The sampling rate of these 750 seismographs is 200/250 per second with resolution
16/24 bits. A very simple processing technique to evaluate the ground velocity and displacement
is proposed. Because of the high resolution data recorded from these array system there is almost
no need to do baseline correction on the acceleration data except the trend removal on constant
DC shift. To calculate the ground velocity direct integration of the acceleration data with the
implementation of low pass filter or a linear trend removal on the original velocity data. With
these high quality ground motion data several application were developed, which include:
1. Develop seismic design spectrum for Sa-value at short period and long period,
2. Generate spatial ground motion for Earthquake Emergency Responses,
3. Study the site effect, particularly on the basin effect,
4. Develop regional phase spectrum for simulation of ground motion.
REFERECES
Chen, M.H., K. L. Wen, and C. H. Loh, “A Study of Shallow Shear Wave Velocities for Alluvium
Deposits in Southwestern Taiwan,” J. of the Chinese Institute of Civil & Hydraulic Engineering, 2003.
Loh, C. H., C. H. Yeh, L. C. Chen, H. C. Hung, W. Y. Jean and W. I. Liao, “Damage Assessment
System HAZ-Taiwan for Seismic Loss Estimation in Taiwan,” NTU Journal, 2002
Sokolove, V. Yu., C. H. Loh, and K. L. Wen, “Empirical model for estimating Fourier amplitude spectra
of ground acceleration in Taiwan region,” J. of Earthquake Engineering and Structural Dynamics, 29,
339-357, 2000.
Sokolov, V., C.H. Loh and K.L. Wen, “Site-dependent design input ground motion estimation for the
Taipei area: A probabilistic approach,” Probabilistic Engineering Mechanics, 16, 2001, 177-191.
Wu, Y. M., W. H. K. Lee, C. C. Chen, T. C. Shin, T. L. Teng, and Y. B. Tsai, 2000, Performance of the
Taiwan Rapid Earthquake Information Release System (RTD) during the 1999 Chi-Chi (Taiwan)
Earthquake, Seismo. Res. Let., 71, pp. 338-343.
Chai, J.-F., Loh, C.-H. and Sato, T. (2002), Modeling of Phase Spectrum to Simulate Design Ground
Motions, Journal of the Chinese Institute of Engineers, Vol. 25, No. 4, pp. 447-459.
Sato, T., Murono, Y., and Nishimura, A., 1999, “Modeling of Phase Spectrum to Simulate Design
Earthquake Motion,” Optimizing Post-Earthquake Lifeline System Reliability, Edited By W.M.
Elliot and P. McDonough, Technical Council on Lifeline Earthquake Eng., ASCE, Monograph
No.16, pp. 804-813.
Sato, T. and Imabayashi, H,. 1999, “Real Time Conditional Simulation of Earthquake Ground Motion,”
Earthquake Engineering and Engineering Seismology, 1(1), 27-38.