Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)

EERI, November 12-15, 2000/ Palm Springs. California

LOCAL SITE EFFECT OF KOBE BASED ON MICROTREMOR

MEASUREMENT

Yutaka Nakamura, Tsutomu Sato, and Masayuki Nishinaga

ABSTRACT

The Hyogo-Ken-Nanbu Earthquake caused heavy damage in Kobe city and its
surrounding area. To clear the relationship between damage and local site effect, a
month after the earthquake, microtremor measurement was performed for about a year.
According to the measurement results in the damaged zone, the amplification factor (A)
ranges between 2 and 3 which is not so high. However the predominant frequency (F)
ranges between 1.5 and 2 Hz which corresponds to that of strong motion. Distribution
of vulnerability index Kg value for ground confirmed the damage belt.

Introduction
The 1995 Hyogo-Ken-Nanbu Earthquake caused heavy damage to Kobe city and its
surrounding area. Especially, a part of area where extensive damage concentrated,
spreads from east to west in the region. With its appearance the area called as Damage
Belt. Many studies have been done for investigating the reason why did the damage
occurred in such a belt shape.
In this paper, to clear the relationship between damage and local site effect,
microtremor measurement was made in damaged and surrounding area. Dynamic
characteristics of surface ground (predominant frequency F, amplification factor A) are
estimated. Furthermore, vulnerability 46

index Kg value (Nakamura,1996) is also

calculated, and compared with real 42
Sea of Japan
observed damage information. As a
result, 1: Predominant frequencies (F) 38
Pacific Ocean

have a trend to gradually small value Kobe

T okyo

from mountain side to sea side. 2: 34

34.74
RO1-0 1

Amplification factor (A) and RO2-0 1

SU2-0 1

SU1-0 1

30
Kobe City
vulnerability index Kg value have a 128 132 136
34.72
140 144 148
NA1 -01
NA2 -01

trend to the high value in damage belt

SMG-01 SU2-2 1
RO2-1 4
34.7 SMA-01
RO1-1 5
NAG-5 0 SMB-1 6 SU1-0 9
MO-0 1
NA1 -15 NA2 -16

area. 3: As the measurement points 34.68

NAG-2 3
KO-0 1

SMB-0 1
SMG-25
SMA-13
Rokko
Island

come nearer to a sea side, Kg value NAG-1 0

NAG-1 1
KO-0 6
Port
Island

become higher these points 34.66

NAG-5 2 NAG-3 7
Osaka Bay

NAG-3 6
: Measurement Points
corresponded to liquefaction area.
NAG-5 3
●

34.64
135.12 135.14 135.16 135.18 135.2 135.22 135.24 135.26 135.28

Therefore, it is suggested that the

distribution of damage is caused by local Fig.1 Measurement points in Kobe city
site effect.

Measurement
The investigated area was in Kobe city and showed in Fig.1. The area is surrounded by
Mt. Rokko in north side, Osaka bay in south side, and after the earthquake this site was

System and Data Research Co., Ltd., 3-25-3 Fujimidai, Kunitachi-shi, Tokyo, 186-0003, Japan
 Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)
EERI, November 12-15, 2000/ Palm Springs. California

damaged like a belt spread from east to west. Kobe city which had huge damage is a
narrow area about 5km from northern mountain side to southern sea side, has grown on
the typical irregular ground. This was the reason, that the damage occurred in a belt of
1-2km width to east-west direction which attracted many researchers to concern this
phenomenon.
In this paper, our main purpose is to prove that the cause of damage belt was because
of the influence of local site effect. A month after the earthquake microtremor
measurement was made from February 1995 to February 1996 for this purpose.
Measurement points showed in Fig.1. Each profile has 15-20 points, total 25 profiles
spreaded north-south through mountain side, damage area, and sea side. The distance
between each points was 100 to 300m, and each profiles length was 1-2km. Total
measurement points were about 400. In this paper analyzed 15 profiles will be discussed,
to be concentrate on a central part in Kobe city.

Analysis
An instrument named Portable Intelligent Collector (PIC) was used for microtremor
measurements. A sensor is set on the asphalt or the soil, are measured at the 2 horizontal
components (NS and EW direction) and a vertical component same time. Sampling
interval is 1/100 sec and the length of each record is 40.96sec. Measurements was
repeated three times at each observation points. After measurements, Fourier spectrum
for each components are calculated. One frequency spectrum of one component was
estimated by averaging the three Fourier spectra. Then, from a spectral ratio of
horizontal to vertical components QTS spectrums (Quasi-Transfer Spectrum) are
calculated, Nakamura(1989). Predominant frequency F and amplification factor A
which represents dynamic characteristics of the ground are found from this analysis and
Vulnerability index Kg are calculated as explained below. Details of the methodology
can be found in Nakamura(1989, 1996).

Vulnerability Index Kg values for Ground

For the vulnerability index Kg of surface ground, shear strain γ is considered
(Nakamura, 1996). According to the Ishihara (1982) ground soil becomes plastic state at
about γ=1000 x10-6 and for γ>10000 x10-6 landslide or collapse of foundation occurs.
Fig. 2. Shows the simplified shear deformations of the surface ground.
Ag ×d

H Vf

Basement Ground vb

Figure 2. Shear deformation of surface ground.

Average shear strain γ can be estimated as γ=Ad/H, where A is amplification factor of

surface layer, H is thickness of surface layer, and d is seismic displacement of basement
 Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)
EERI, November 12-15, 2000/ Palm Springs. California

layer. Details about formulation can be find in Nakamura(1996). Without going into a
details, we are going to write shear strain as follows (Nakamura, 1996);
A2 a
γ = ⋅ 2
F π ⋅ vb
2
In this equation, A /F is called as Vulnerability index, Kg value for surface ground. a is
the acceleration in the basement. vb is the S wave velocity of the basement.

Results and Discussion

Relationship between the damage and Predominant frequency (F), Amplification factor
(A), and Vulnerability index (Kg) value calculated from microtremor measurement is
discussed.
Distribution of Predominant frequency F
Fig.3 shows distribution of predominant frequency (F) estimated from QTS in each
measurement points. The value of F divided into 4 ranges (F<1.5Hz, 1.5<F<2.5Hz,
2.5<F<3.5Hz, 3.5<Hz) to easily catch the whole tendency. This area is Mt. Rokko in
northern part, and sea in southern part. Distribution of F in the Fig.3 presents large F
value on northern part overall, and small F value on southern part. However, the area of
seismic intensity of 7 (on the Japanese scale) does not show the same phenomenon as
F.

RO1-01
SU2-01
34.73 RO2-01 SU1-01

Kobe City
NA2-01

NA1-01
34.71 SMG-01
Intensity 7
SMA-01 SU2-21
NAG-50 RO2-14
MO-01 RO1-15 SU1-09

KO-01 NA1-15
34.69 NA2-16
NAG-23 Rokko
SMA-13 Island
SMG-25
KO-06
Port
34.67 NAG-10
Island
NAG-11

F<1.5
NAG-37
1.5<F<2.5
34.65
NAG-53 NAG-52 NAG-36 Osaka Bay 2.5<F<3.5
3.5<F

34.63
135.12 135.14 135.16 135.18 135.2 135.22 135.24 135.26 135.28

Fig.3 Distribution of Predominant Frequency F

More detailed graph of F in the damage belt is shown in Fig.4, F values are little less
then 2Hz in general. This frequency is predominant frequency of strong motion records
in 1995 Hyogo-Ken-Nanbu Earthquake recorded in this area, this was guessed to be a
reason of resonance of strong motion in this area.

Distribution of Amplification factor A

Fig.5 shows distribution of amplification factor (A) in each points. This distribution
again divided into 4 ranges as we did in F, intervals are A<2.0, 2.0<A<2.5, 2.5<A<3.0,
 Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)
EERI, November 12-15, 2000/ Palm Springs. California

3.0<A. For whole area, while damaged area of seismic intensity of 7 have larger A
value in measurement points, slightly damaged area have smaller A. Overall distribution
has the characteristic that A has large value in the highly damaged area.
Fig.4, shows more detailed graphs of A. A does not show clear and characteristic trend
as F. However, in lots of measurement points on high damage  area, A increase and tends
to show value of  3. On 10.00
F(Hz) F(Hz)
10.00

8.00 8.00
A(times)
the other hand, Damage Zone
6.00
A(times)
Damage Zone

Ｆ and A
6.00

Ｆ and A
distribution of A in 4.00 4.00

2.00 2.00

MO and KO lines, in 0.00 0.00

0 500 1000 1500 2000 2500 3000 3500 4000
the slightly damaged (Sea side)
Measurement Points (NO.53-NO.10)
(Mt. side) (Sea side)
Distance （ｍ）

Measurement Points (NO.11-NO.52)

(Mt. side)
0 1000
Distance （ｍ）
2000 3000

area (the place was cut 10.00 10.00

F(Hz) F(Hz)
the damage belt.) A in A(times)
8.00

6.00
A(times)
8.00

Damage Zone 6.00

F and A
F and A

Damage Zone
almost all the points 4.00 4.00

in this area show 2.00

0.00
2.00

0.00

smaller value same as (Sea side)

0 1000

(Mt. side)
2000

(Sea side)
3000

Distance （ｍ）
4000 5000 6000

(Mt side)
0 1000 2000
Distance (m)
3000

Measurement Points (NO.23-NO.36) Measurement Points (NO.37-NO.50)

the points in mountain
side . 10.00 10.00

F(HZ) F(Hz)
Briefly, we can say 8.00 A(times) A(times)
8.00
F and A

6.00 6.00

F and A
that amplification 4.00 Damage Zone
4.00

factor (A) has large 2.00 2.00

values in the area 0.00

0.00
0 1000 2000 3000
0 1000 Distance (m) 2000 (Mt. side) Distance (m)
(Sea side) (Sea side) (Mt. side)
Measurement Points (KO) Measurement Points (SMG)
where heavy damage 10.00 10.00

occurred. If we F(Hz)
A(times)
8.00
F(Hz)
A(times)
8.00

consider the behavior 6.00 6.00

F and A
F and A

Damage Zone
4.00 4.00

of F and A together we 2.00 2.00

can say that these 0.00

0 1000 2000 3000
0.00
0 1000 2000
Distance (m) Distance (m)
(Sea side) (Mt. side)
characteristics caused (Sea
Measurement Points (SMA)
(Mt. side)
Measurement Points (MO)

severe vibration of Fig.4 Distribution F and A in each profile

strong motion coming
from basement.

RO1-01

SU2-01
34.73 RO2-01 SU1-01

NA2-01
Kobe City
NA1-01
34.71
SMG-01 SU2-21
SMA-01 RO2-14
NAG-50 MO-01 RO1-15
NA1-15 SU1-09
NA2-16
KO-01
34.69 NAG-23
Rokko
SMA-13 Island
SMG-25

KO-06
Port
34.67 NAG-10
NAG-11 Island
3.0<A
2.5<A<3.0
NAG-52
NAG-37 2.0<A<2.5
34.65 NAG-53 NAG-36
Osaka Bay A<2.0

34.63
135.12 135.14 135.16 135.18 135.2 135.22 135.24 135.26 135.28

Fig.5 Distribution of Amplification Factor A

 Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)
EERI, November 12-15, 2000/ Palm Springs. California

Distribution of Vulnerability index Kg value, comparison between Kg value and

detail damage data
Fig.6 is a distribution of vulnerability index Kg value calculated from the relationship
of F and A. This figure also given for 4 ranges like F and A. Kg has larger value in the
damage area spreaded like a belt, and small value in the other areas. Kg value is related
to damage distribution. In measurement points of large Kg value near coastline (NA2
and SU1 lines), although the area is out side of seismic intensity of 7, liquefaction was
observed extensively.
RO1-01
SU2-01
34.73 RO2-01 SU1-01

NA2-01
Kobe City
NA1-01
34.71
SMG-01 SU2-21

SMA-01 RO2-14
NAG-50 MO-01 RO1-15
SU1-09
NA1-15 NA2-16
34.69 NAG-23 KO-01
Rokko
SMA-13 Island
SMG-25
KO-06
34.67 NAG-10 Port
NAG-11 Island
4.0<K
NAG-52 NAG-37 3.0<K<4.0
34.65 NAG-53 NAG-36
2.0<K<3.0
Osaka Bay K<2.0

34.63
135.12 135.14 135.16 135.18 135.2 135.22 135.24 135.26 135.28

Fig.6 Distribution of Vulnerability index Kg

10.0
Sea side Mountain side Sea side Mountain side Sea side Mountain side
10 10.0 10 10.0 10
9 Kg Value 9 Ｋg Value 9
8.0 Ｋ g Value 8 8.0 8 8.0 8
Damage Damage 7 Damage 7
Ｋg Value

Ｋg Value

7
Ｋg Value

6.0 6 6.0 6 6.0 6

5 5
5

Damage
4.0 4
Damage

4.0
4.0 4
4 3
Damage

3
3 2.0 2 2.0 2
2.0 2 1 1
1 0.0 0 0.0 0
NO.52

NO.22

NO.21

NO.20

NO.19

NO.18

NO.17

NO.16

NO.15

NO.14

NO.13

NO.12

NO.11

0.0 0
NO.37

NO.38

NO.39

NO.40

NO.41

NO.42

NO.43

NO.44

NO.45

NO.46

NO.49

NO.48

NO.47

NO.50
NO.53

NO.54

NO.55

NO.56

NO.57

NO.58

NO.1

NO.51

NO.2

NO.3

NO.4

NO.5

NO.6

NO.7

NO.8

NO.9

NO.10

Measurement Points Measurement Points

Measurement Points
Sea side Mountain side Sea side Mountain side Sea side Mountain side
10.0 10 10.0 10 10.0 10
9 9 9
8.0 Ｋg Value 8 Ｋg Value 8.0 Ｋg Value 8
Damage 8.0 8 Damage
7 Damage 7
Ｋg Value

Ｋg Value

7
Kg Value

6.0 6 6.0 6 6.0 6

5 5 5
Damage

4
Damage

4.0
Damage

4.0 4 4.0 4
3 3 3
2.0 2 2.0 2 2.0 2
1 1 1
0.0 0 0.0 0 0.0 0
NO.35

NO.36

NO.34

NO.33

NO.32

NO.31

NO.30

NO.29

NO.28

NO.27

NO.26

NO.25

NO.24

NO.23

MO-15

MO-14

MO-13

MO-12

MO-11

MO-10

MO-09

MO-08

MO-07

MO-06

MO-05

MO-04

MO-03

MO-02

MO-01
KO-06

KO-05

KO-04

KO-03

KO-02

KO-01

Measurement Points Measurement Points Measurement Points

Sea side Mountain side Sea side Mountain side Sea side Mountain side
10.0 10 10.0 10 10.0 10
9 Ｋg Value 9 Ｋg Value 9
8.0 Ｋg Value 8 8.0 8 8.0 8
Damage Damage 7 Damage 7
Ｋg Value

7
Ｋg Value
Ｋg Value

6.0 6 6.0 6 6.0 6

5 5 5
Damage
Damage

4.0 4
Damage

4.0 4 4.0 4
3 3 3
2.0 2 2.0 2 2.0 2
1 1 1
0.0 0 0.0 0 0.0 0
NA1-15

NA1-14

NA1-13

NA1-12

NA1-11

NA1-09

NA1-10

NA1-08

NA1-07

NA1-06

NA1-05

NA1-04

NA1-03

NA1-02

NA1-01
SMG24

SMG22

SMG20

SMG19

SMG17

SMG15

SMG13

SMG11

SMG9

SMG7

SMG3

SMG4

SMG1
SMA13

SMA12

SMA11

SMA10

SMA9

SMA8

SMA7

SMA6

SMA5

SMA4

SMA3

SMA2

SMA1

Measurement Points Measurement Points Measurement Points

Sea side Mountain side Sea side Mountain side Ssea side Mountain side
10.0 10 10.0 10 10.0 10
Ｋg Value 9 9
8.0 8 8.0 Ｋg Value 8 9
Damage Damage 8.0 Ｋg Value 8
7
Ｋg Value

7
Ｋg Value

Damage 7
Ｋg Value

6.0 6 6.0 6
5 5 6.0 6
Damage

5
Damage

4.0 4 4.0 4
Damage

3 3 4.0 4
2.0 2 2.0 2 3
1 1 2.0 2
0.0 0 0.0 0 1
0.0 0
NA2-15

NA2-16

NA2-14

NA2-13

NA2-12

NA2-11

NA2-10

NA2-08

NA2-09

NA2-07

NA2-06

NA2-05

NA2-04

MA2-03

NA2-02

NA2-01

RO1-15

RO1-14

RO1-11

RO1-12

RO1-13

RO1-10

RO1-9

RO1-8

RO1-7

RO1-6

RO1-5

RO1-4

RO1-3

RO1-2

RO1-1

RO2-14

RO2-13

RO2-12

RO2-11

RO2-10

RO2-9

RO2-8

RO2-7

RO2-6

RO2-5

RO2-4

RO2-3

RO2-2

RO2-1

Measurement Points Measurement Points Mesurement Points

Fig.7 Comparison between Kg and damage ratio

 Proceedings of the Sixth International Conference on Seismic Zonation (6ISCZ)
EERI, November 12-15, 2000/ Palm Springs. California

Kg value reflects local site effect and corresponded with the damage belt and
liquefaction as well. It is possible to conclude that the damage of 1995 Hyogo-Ken-
Nanbu Earthquake was related to dynamic characteristics of the ground.
Fig.7 compares Kg value and damage ratio for small-scale structures. Damage ratio
divided into 5 ranges; 1: no damage, 2: 0-12.5%, 3: 12.5-25%, 4: 25-50%, 5: 50-100%.
In each figure, distribution form of Kg value and damage ratio has similar tendency.
Especially, in NAG, SMA, SMG area, distribution form of Kg, very well corresponds
with damage ratio of structure from mountain side to sea side. The damage area
between mountain parts and sea parts shows larger Kg value and bigger damage ratio.
On the other hand, distribution of Kg in in almost all the points on MO and KO lines (in
lightly damaged area), show smaller value. With these results Kg value which is defined
as vulnerability index for ground, found to be related also with damage ratio of
structures.
To compare Kg value and damage ratio overall, in Kobe city, it has been found that
some damage occurred for Kg>2. For acceleration value of 200-300gal and S-wave
velocity of 300m/s in the basement of this city, ground strain estimated as γ = 1 ~ 2 × 10 −3
when earthquake occurred in the area. This value corresponds to arise ground damage.
In Fig.7, Kg value does not agree with damage ratio in some parts. Points with large
Kg value in spite of small damage ratio are almost located in liquefaction area. To find
out the reason for this, building characteristics in this area will be checked in detail in
further part of the study.

Conclusion
To explain dynamic characteristics of the ground in the damage of 1995-Hyogo-Ken-
Nanbu Earthquake, microtremor measurement was made in Kobe city. Amplification
factor A and vulnerability index Kg value are showed larger value in damage belt.
Vulnerability index Kg value in spite of for the ground agreed with damage ratio for
structure very well. Therefore, it can be conclude that damaged belt occurred after the
earthquake had a unique character of local site effect in the area.

Acknowledgment
Microtremor measurement in Kobe city was preformed with Railway Technical
Research Institute (RTRI). Thank for their cooperation and help during measurements.
And we would like to thank you Dr. E. Dilek Gurler for her help and comments.

References
Nakamura, Y,. A Method for Dynamic Characteristics Estimation of Subsurface using
Microtremor on the Ground Surface, Quarterly Report of RTRI, Railway Technical
Research(RTRI), 1989, Vol.30, No.1.
Nakamura, Y,. Seismic Vulnerability Indices For Ground and Structures Using Microtremor,
World Congress on Railway Research in Florence, Italy, November, 1997.
Building Research Institute Ministry of Construction, The report of damage on the 1995
Hyogo-Ken-Nanbu Earthquake, 1996