august 15-17, 2016 site effects on strong motion …...5th iaspei / iaee international symposium:...

5th IASPEI / IAEE International Symposium: Effects of Surface Geology on Seismic Motion August 15-17, 2016

SITE EFFECTS ON STRONG MOTION RECORDS OF THE 2011 TOHOKU, JAPAN EARTHQUAKE

Saburoh Midorikawa and Yoshihiro Nogi Tokyo Institute of Technology

1

500 km

The 2011 Tohoku earthquake of Mw9.0 produced many strong motion records. We have collected about 3000 records on ground from different sources;

K-NET 673 KiK-net 511 JMA 396 Pref.-net 952 MLIT 153 NEXCO-East 126 PARI 37 BRI 25 Tohoku Tech 17 Tohoku Univ. 16 JR-East 12

Total 2918 2

PGV cm/s PGA cm/s2

The dataset includes records with large amplitudes. The records provide a unique opportunity for better understanding of ground motion at strong shaking level.

Distribution of Peak Horizontal Acceleration and Velocity

3

Objectives

Site effects on ground motion have been studied using observed ground motion records. In many cases, weak motion records not strong ones were used because of the data availability.

To examine site effects at strong shaking level, nonlinear soil response analyses have been also conducted. The validity of the nonlinear analyses, however, have not fully examined by observed strong motion records.

In this study, we select about 500 strong motion records observed in the high seismic intensity area of the Tohoku earthquake. Then we discuss preliminarily site factors at strong shaking level from the average spectra at different site classes.

4

Target Area JMA Seismic Intensity

We selected the high intensity area where the JMA seismic intensity of 6 lower (MMI 9) or greater is observed as the target of the study. The number of the sites in the area is 467. At most of the sites, the recorded PGA is higher than 300 cm/s2.

5

Among the 467 sites, the soil profile data is available only at 92 sites. At the other 375 sites, the NEHRP site class is determined by the H/V spectral ratio of microtremors. The criteria for site classification from the H/V spectral ratio have been proposed (Nogi and Midorikawa, 2016).

Period (s)

H/V

Spectr

al R

atio

Band Width:0.3Hz n=9

0.05 0.5 2

0.1

110

B Ave ± σ

Period (s)

H/V

Spectr

al R

atio


0.05 0.5 2

0.1

110

C Ave ± σ

Period (s)

H/V

Spectr

al R

atio


0.05 0.5 2

0.1

110

D Ave ± σ

Period (s)

H/V

Spectr

al R

atio


0.05 0.5 2

0.1

110

E Ave ± σ

category

Ave

(0.3～

0.8

s)

/ Ave

(0.1～

0.2

s)

B C D E

0.1

11

0

Ave(0.3～0.8s) / Ave(0.1～0.2s)

y = 1.6

y = 0.44

category

va

ria

nce

(in

lo

g)

B C D E

0.0

02

0.0

10

.05

0.5

0.1～2s

y=0.007

category

Ave

(0.6～

1.2

s)

/ Ave

(0.1～

0.3

s)

B C D E

0.1

11

02

0

Ave(0.6～1.2s) / Ave(0.1～0.3s)

y=2.7

Site Classification

6

Finally, using the soil profile data and microtremor data, the sites are classified into 30 class B sites, 153 class C sites, 207 class D sites and 77 class E sites.

The PHAs and PHVs at site class B which is considered to be a reference site are about 0.45 g and 25 cm/s in average, respectively.

Site Class (# sites)

Pea

k g

rou

nd

acc

eler

atio

n (

cm/s

/s)

Fault distance (km)

PGAの距離減衰特性

Si and Midorikawa (1999) Mw=9.0 Mw=8.3

100

101

102

103

100

101

102

103

7

Strong site effects are found on the records. These three sites are located within several kilometers, and the difference in the spectra is mainly due to site effects.

h=0.05

Site Class (# sites)

8

Kunimi Town Office Building was damaged and demolished.

9

Acceleration

-100

0

100

Acc.

(cm

/s/s

)

NS (peak:- 17.8 cm/s/s)

-100

0

100A

cc.

(cm

/s/s

)EW (peak: 17.9 cm/s/s)

-100

0

100

Acc.

(cm

/s/s

)

0 10 20 30 40 50 60 70 80 90 100Time (sec)

UD (peak: 6.0 cm/s/s)

Velocity

-5

0

5

Ve

l. (

cm

/s) NS (peak: 2.23 cm/s)

-5

0

5

Ve

l. (

cm

/s) EW (peak: 2.41 cm/s)

-5

0

5

Ve

l. (

cm

/s)

0 10 20 30 40 50 60 70 80 90 100Time (sec)

UD (peak: 0.43 cm/s)

Fourier Spectrum (Time:0-60s, Parzen:0.2Hz)

0.05

0.1

0.5

1

5

10

50

Fo

uri

er

Am

plit

ud

e (

cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

NSEWUD

Pseudo Vel. Response Spectrum (h=5%)

0.05

0.1

0.5

1

5

10

20

Pse

ud

o V

elo

city R

esp

on

se

(cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

0.001(cm)

0.010.1

1

10

0.1(

cm/s/s)

1

10100

1000 NS

EWUD

2011/03/10 03:16, Site: 8A4 Fore Shock, JMA Intensity: 3.0

Acceleration

-500

0

500

Acc.

(cm

/s/s

)

NS (peak:- 405.8 cm/s/s)

-500

0

500

Acc.

(cm

/s/s

)

EW (peak:- 438.7 cm/s/s)

-500

0

500

Acc.

(cm

/s/s

)

30 40 50 60 70 80 90 100 110 120Time (sec)


Velocity

-100

0

100

Ve

l. (

cm


-100

0

100

Ve

l. (

cm

/s) EW (peak:- 82.13 cm/s)

-100

0

100

Ve

l. (

cm

/s)

30 40 50 60 70 80 90 100 110 120Time (sec)

UD (peak:- 15.79 cm/s)


1

5

10

50

100

500

1000

Fo

uri

er

Am

plit

ud

e (

cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

NSEWUD


5

10

50

100

500

Pse

ud

o V

elo

city R

esp

on

se

(cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

0.1(cm) 1

10

100

1(cm

/s/s)

10

10010

00

1000

0

NSEWUD

2011/03/11 14:48, Site: 8A4 Main Shock, JMA Intensity: 6.0

Acceleration

-100

0

100

Acc.

(cm

/s/s

)

NS (peak: 41.1 cm/s/s)

-100

0

100

Acc.

(cm

/s/s

)

EW (peak:- 58.4 cm/s/s)

-100

0

100

Acc.

(cm

/s/s

)

0 10 20 30 40 50 60 70 80 90 100Time (sec)


Velocity

-20

0

20

Ve

l. (

cm


-20

0

20

Ve

l. (

cm

/s) EW (peak:- 14.28 cm/s)

-20

0

20

Ve

l. (

cm

/s)

0 10 20 30 40 50 60 70 80 90 100Time (sec)

UD (peak:- 2.02 cm/s)


0.2

0.5

1

5

10

50

100

200

Fo

uri

er

Am

plit

ud

e (

cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

NSEWUD


0.5

1

5

10

50

100

Pse

ud

o V

elo

city R

esp

on

se

(cm

/s)

0.05 0.1 0.5 1 5 10 20Period (sec)

0.01(cm)

0.1

1

10

100

1(cm

/s/s)

10

100

1000 NS

EWUD

2011/04/11 17:16, Site: 8A4 After Shock, JMA Intensity: 4.4

0 50 100 Time (sec)

0 50 100 Time (sec)

0 50 100 Time (sec)

Main Shock PGA=0.44g

Foreshock PGA=0.02g

Aftershock PGA=0.06g

Nonlinear behavior of site response is also found on the records. This figure shows comparison of response spectra of the main shock, foreshock and aftershock records at JMA Wakuya. The site is located on the alluvial lowland, and classified to the site class E. 10

Average Response Spectra To generalize the site effects on the main shock records, the average spectrum for each site class is calculated. The velocity response spectra tend to be constant with period at class B. The spectral amplitudes become larger at softer site classes. At site class E, the amplitudes tend to be larger at around 1 second.

11

Response S

pectr

al R

atio

Period (s)

Class E/ Class B Class D/ Class B Class C/ Class B

0.1 1 100.1

1

10

The average and standard deviation of PHAs at site class B are 0.45 g and 0.19 g, respectively. This suggests that the shaking level inputted to the bedrock in the area may not have large difference. Therefore, the ratio of the average spectrum at class C, D or E with respect to that at class B is considered to be an approximate of site amplification at strong motion level. The amplifications of site classes C and D to B are almost two, and rather similar to each other. The amplification of site class E to B, however, is larger at longer periods and smaller at short periods, showing stronger site effects.

Spectral Ratio of Average Spectra

12

Site Amplification at Weak Motion Level We have evaluated the amplification factor at weak motion level with respect to Vs30, using small or moderate amplitude records observed at nearby soil-rock station pairs (Yamaguchi and Midorikawa, 2012).

Schematic Figure for Evaluation of Amplification Records at Nearby Station Pairs 13

Response S

pectr

al R

atio

Period (s)

AVS30(150m/s)/ AVS30(800m/s) AVS30(250m/s)/ AVS30(800m/s) AVS30(500m/s)/ AVS30(800m/s) Class E/ Class B Class D/ Class B Class C/ Class B

0.1 1 100.1

1

10

The amplifications evaluated for weak motion are shown by red lines. In general, the amplification from strong motion is smaller at short periods and larger at longer periods than that from weak motion. The difference in both amplifications is smaller at site class C (stiff site) and larger at site class E (soft site).

Class E / Class B (Weak Motion) Class D / Class B (Weak Motion) Class C / Class B (Weak Motion) Class E / Class B (Strong Motion) Class D / Class B (Strong Motion) Class C / Class B (Strong Motion)

14

Class E

Class C

Class D

after Seyhan and Stewart (2014)

Shorter Period Amp.

Longer Period Amp.

Shorter Period Amp.

Longer Period Amp.

Class B

We compared our site factors from Japanese strong motion data with those from NGA2 GMPE and semi-empirically developed by Seyhan and Stewart (2014) . Our results are consistent with those from the U.S. data and simulation, but show weaker nonlinear effects for longer period site factor at site class E.

15

0.1

110

0.1 1 10

Sp

ect

ral

Ra

tio

Period(s)

h=0.05 n=3

MeanC/B ratio

C

C

B E

25km

C

Reference lock (Site Class B)

PGAr =0.272(g)

1)

C / B

0.1

110

100

1000

0.1 1 10Acc

ere

lati

on

Re

spo

nse

(cm

/s/

s)

Period(s)

h=0.05

Site Class BSite Class C

2)

3) 4)

To examine the amplitude dependence of the site factors more quantitatively, we are evaluating the amplification factors using records observed at nearby station pairs (Hori and Midorikawa, 2016). We select a site with site class B as a reference rock. For soil sites within 25 km and the rock site, the response spectra are calculated. From the spectral ratio, the amplifications for short-period Fa and for longer period Fv, are calculated. The amplification versus the peak ground acceleration at reference rock, PGAr, is plotted.

16

Fa Fv

PGAr (g)

Fa(0

.1-0

.5s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class C

PGAr (g)

Fv(0

.4-2

.0s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class C

PGAr (g)

Fa(0

.1-0

.5s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class D

PGAr (g)

Fv(0

.4-2

.0s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class D

PGAr (g)

Fa(0

.1-0

.5s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class E

PGAr (g)

Fv(0

.4-2

.0s)

0.01 0.05 0.5 1

0.1

0.5

52

0

Site class E

This Study average

Hori and Midorikawa

Poster Session P104C

17

This figure shows the amplifications for short-period Fa and for longer period Fv from the different station pair records. For site class C, the amplifications are constant with the amplitude. For site class D, the amplitude dependence of the amplifications is slightly observed. For site class E, the amplitude dependence is clearly observed at Fa, but weaker at Fv.

Conclusions To discuss site effects at strong shaking level, 467 strong motion records observed in the high seismic intensity area of the 2011 Tohoku earthquake are analyzed. The site amplification is preliminarily evaluated from the ratio of the average spectrum at class C, D or E with respect to that at class B. The amplifications of site classes C and D to B are almost two, and rather similar to each other. The amplification of site class E to B, however, is larger at longer periods and smaller at short periods, showing stronger site effect. The amplification is compared with the site amplification derived from weak motion records. The difference in both amplifications is smaller at stiff site and larger at soft site, which is consistent with recent results from the U.S. data.

18

Examples of Vs Profile at Site Class B

Vs (m/s)

De

pth

(m

)

0 1000 2000

30

20

10

0

California

Tohoku

20

august 15-17, 2016 site effects on strong motion …...5th iaspei / iaee international symposium:...

Documents