Factor analysis of earthquake-induced geological disasters of the M7. 0 Lushan earthquake in China

Geodesy and Geodynamics

2013 , 4 ( 2) :22 - 29

http://www. jgg09. com Doi:10.3724/SP.J.1246.2013.02022

Factor analysis of earthquake-induced geological disasters of the M7. 0 Lushan earthquake in China Li Xue, Liu Xiaoli, Li Jinggang, Wang Qiuliang, Liao Wulin and Zhang Lifen Key Laboratory of EarthqWJlre Geodesy, lmtitute of Seimwlogy, China EarthqWJlre Administration, Wuhon 430071 , Chino

2

Abstract: The seismic intensities, lithologic characteristics and terrain features from a 3000 km -region near

the epicenter of the Lushan earthquake are used to analyze earthquake-induced geological disaster. The preliminary results indicate that secondary effects of the earthquake will affect specific areas , including those with glutenite and carbonate bedrock, a seismic intensity of IX, slopes between 40" and 50", elevations of less than 2500 m, slope change rates between 20" and 30", slope curvatures from - 1 to -0. 5 and 0. 5 to 1 , and relief between 50 and 100 m. Regions with susceptibility indices greater than 0. 71 are prone to landslides and collapses. The secondary features are mainly distributed on both sides of the ridges that extend from Baosheng to Shuangshi and from Baosheng to Longxing. Other features are scattered on both sides of the ridges that extend from Qishuping to Baosheng and from Masangping to Lingguan. The distribution of the earthquake-related features trends in the NE direction, and the area that was most affected by the Lushan earthquake covers approximately 52. 4 km 2 •

Key words: Lushan earthquake ; earthquake-induced geological disaster; factor analysis ; susceptibility index ; hazard distribution

113 aftershocks were greater than M3. 0; these includ-

1 Introduction

ed four aftershocks between MS. 0 and MS. 9, 21 aftershocks between M4. 0 and M4. 9 and 88 aftershocks

The M7. 0 Lushan earthquake occurred at 8 : 02 Bei-

between M3. 0 and M3. 9. The preliminary data showed

jing time on April 20, 2013. The epicenter ( 30. 3 "N,

that the earthquake had a rupture length of approxi-

103. O"E) was located between the towns of Longmen,

mately 35 to 40 km. The aftershocks defined a NNE

Pousheng and Taiping in Lushan county, Sichuan

trend along the Shuangshi-Dachuan fanit and had focal

province, approximately 100 km from Chengdu. The

depths between 15 and 25 km. The duration of the fo-

earthquake had a focal depth of 13 km and a maximum

cal rupture was approximately 30 seconds , and the

intensity of IX, and it affected an area of approximately

maximum slip on the fault plane was 1. 6 m['l. Hou-

18682 km2 • According to the China Earthquake Ad-

ses , roads and engineering facilities in the main shock

ministration website , the earthquake caused a total of

area were seriously damaged. The earthquake triggered

196 deaths and injured 11470 people; 21 people re-

many geological disasters , such as landslides , collap-

main missing. A total of 5657 aftershocks had been re-

ses and dammed lakes, which were significant challen-

corded through 20 : 00 on April 28th 2013, of which

ges to the disaster relief and reconstruction work. Chinese scholars have undertaken extensive research

Received: 2013-05-03; Accepted: 2013-05-11

Corresponding author:Li Xue, E-mail:leexuel211@ 126. com

on the causes , susceptibility and spatial distribution of geological

disasters['-' 1 •

This work is supported by the Director Foundation of the Institute of

earthquake-induced

The

Seismology, China Earthquake Administration( 201056076,201116002).

analysis of these secondary seismic effects using GIS ,

Li Xue, et al. Factor analysis of earthquake - induced geological disasters of

23

the Ml. 0 Lushan earthquake in China

No.2

RS, GPS and other spatial information technologies made considerable progress after the 2008 Wenchuan

2. 2

Geological structure

earthquake due to the rapid access to earthquake dam-

Both the Lushan earthquake and the Wenchuan earth-

age , the development of theories of seismic hazards and

quake occurred along the Longmenshan fault zone. The

the statistics of the distribution of earthquake-induced

study area is at the eastern margin of the Qinghai-Tibet

geological disasters[S-?l. This paper uses space-based

plateau and includes terrain with widely varying eleva-

technologies to perform a comprehensive quantitative a-

tions.

nalysis of the seismic, geologic and terrain factors that

mountains, many of which are fragmented. River inci-

controlled the distribution of earthquake-induced geo-

sion is intense and has formed a series of narrow V-

logical disasters. The research provides a scientific

shaped valleys, and level I to III terraces have devel-

basis for post-earthquake rescue, hazard investigations

oped on both sides of the rivers. The towns and villages

and post-disaster reconstruction.

in the area are located on these terraces , and these ge-

The

study

area

includes

moderately

tall

ological and geomorphological characteristics make the

2 Geological structure and data sources

area prone to geological disasters. The distribution of faults and the bedrock lithologies

2. 1

are shown in figure 2. The lithologies of the study area

Location

can be divided into eight categories, including clay,

The epicenter of the M7. 0 Lushan earthquake is in

sandstone, conglomerate, mudstone, shale , phyllite ,

Lushan county, Yaan city, Sichuan province. Lushan

carbonates and igneous rocks.

county lies at the mountainous western margin of the Sichuan basin at the southern end of the Longmenshan

2. 3

Data sources

fault zone. The Lushan earthquake occurred less than

The data used in this study include Landsat-7 and AS-

100 km from the epicenter of the M8. 0 Wenchuan

TER GDEM images, both of which were downloaded

earthquake. The study area extends from 30° 4' to

from the website of the International Scientific Data

30°27'N and from 102° 36' to 103° 16' E, which is an

Service Platform ( http :II datamirror. csdb. en/index.

area of approximately 3000 km

2

(

Fig. 1 ) .

htm) . The remote sensing images were collected in 2003 at a spatial resolution of 15 m and were projected to the WGS_1984_UTM_Zone_48N coordinate system. The elevation data were collected in 2009 at a spatial resolution of 30 m and projected to the UTM/WGS84 coordinate system.

3 3.1

Analysis principles Factor selection and state grouping

Factor analysis is commonly used in GIS spatial analysis and digital terrain analysis. Many researchers have studied the spatial distribution of earthquake damage using terrain factors[S-!0]. Terrain factors can be divided into micro factors and macro factors. Micro factors Figure 1

Regional remote sensing image ( Landsat-7 ETM 7/4/2 band composite image)

include slope,

aspect,

slope change rate,

aspect

change rate , plan curvature , profile curvature and slope length. Macro factors include slope form, terrain

24

Geodesy and Geodynamics

Vol.4

roughness, relief amplitude, elevation coefficient of

the geological factor ( Fig. 2) . The values of the geo-

variation and depth of incision. Both the Lushan earth-

logical factor for the Lushan earthquake can be deter-

quake and the Wenchuan earthquake occurred along

mined using the statistical correlation between seconda-

the Longmenshan fault zone in areas of a similar ter-

ry seismic effects and lithology from the Wenchuan

rain. Therefore, the terrain factors identified in factor

earthquake. The values of the seismic factor are deter-

analysis research of the Wenchuan earthquake have

mined using the seismic intensity map released by the

been used to analyze the Lushan earthquake; these fac-

China Earthquake Administration [13 l .

tors include elevation , slope , slope change rate , curvature and relief amplitude[

The state groups for the seismic , geological and ter-

11 12 • J.

rain factors for the Lushan earthquake are shown in

The development and spatial distribution of earth-

table 1. The groups are based on the statistical results

quake-induced geological disasters are not only related

of the spatial relationships between the secondary

to terrain factors but also affected by seismic and geo-

seismic effects in the Wenchuan earthquake and the

logical factors. In this paper, the lithology is used as

seismic hazard factors [8 -! 3 J • Legend

•

Devonian sandstone, mudstone, carbonate rock

•

Siluria shale, limestone

D

Ordovician sandstone, limestone

•

Sinian limestone phyllite

D

Presinian gabbro

1ZJ Stratigraphic boundary

E E

K

Tianquan

~--~~--------------~------------------~

Figure 2

IZJ Pleistocene fault IZJ Late Pleistocene fault

* 8

Epicenter

County

Geological structure of the study area (fault information in downloaded from the China Earthquake Administration website "Seismic geological map of the Lushan earthquake" )

Table 1 Factor state groups Factors

State Group

2

3

4

5

6

vm

vn

VI Shale

Phyllite

20002500

25003000

>3000

20-30

30-40

40-50

50-60

>60

10-20

20-30

30-40

>40

-2.0-1.5

-1.5-1.0

-1.0-0.5

-0.50

00.5

0.51.0

Category

Name

Seismic factor

Seismic intensity

(I)

IX

Geological factor

Lithology ( L)

Clay

Elevation (E)

<1000

10001500

15002000

Slope( S)

<10

10-20

Slope change rate(R)

<10

Curvature (C)

< -2.0

Relief amplitude (A)

<20

Terrain factor

Sandstone Glutenite Mudstone

20-50 50 - 100 100 - 180

> 180

7

8

9

100

1.52.0

>2.0

Carbonate Igneous rock rock

1.01.5

Li Xue, et al. Factor analysis of earthquake - induced geological disasters of

No.2

3. 2

25

the Ml. 0 Lushan earthquake in China

sually occur in areas of hard rock, while landslides de-

Analysis method

velop more often in soft rocks. In this paper, the den-

The Susceptibility Index ( SI) is used to analyze the

sity of earthquake-induced geological disasters in dif-

distribution of secondary seismic features in the study

ferent lithologies during the Wenchuan earthquake is

area. The formula for SI is shown in equation 1. /Fi

used as the geological factor for the Lushan earth-

represents the factors that are involved in the calcula-

quake.

tion' and wi is the weight of the factor.

The topography of the study area ranges from 523 m to 3448 m. The elevations are divided into six catego-

Sf=

L WJFi

(1)

ries at 500 m intervals ( Fig. 4) . Approximately 80% of the landslides triggered by the Wenchuan earthquake

The value of /Fi depends on the value of the factor state group. The weights of the factors ( wi) can be calculated using

the

Analytic

Hierarchy

Process

( AHP). AHP is a multi-indicator analysis method that can compare the factors in pairs [4 J • The AHP method can avoid problems that arise when an individual

were at elevations below 2500 m, and more than 50% occurred between 1000 m and 2000 m[ 4 • 6 • s]. The largest percentage of landslides occurred between 1000 m and 1500 m. In this paper, the elevation distribution of landslides from the Wenchuan earthquake is used as the elevation factor for the Lushan earthquake.

weight is unreasonable. The Consistency Rate ( CR) is used to measure whether the weights of the factors are appropriate.

4

Factor analysis of earthquakeinduced geological disasters

4. 1

Factor assignment

Based on the seismic intensity map released on April 25th by the China Earthquake Administration, the

Mingshan VII

10 Tianquan km "

ll vr

maximum intensity of the Lushan earthquake was IX , and the distribution of isoseismal lines trends to the NE. The seismic intensities in the study area are shown in figure 3 . The areas of intensities IX , VIII ,

VH

and

VI

Figure 3

Seismic intensity map of the Lushan eathqueke

( adapted from the China Earthquake Administration website)

are 208 km2 , 1167 km 2 , 1133 km 2 and 426 km 2 , reN

spectively. Based on statistical correlations between

A

seismic intensity and landslides[ 12 J , the landslide density is used as the value of the seismic factor. Mter vectorizing the 1 : 250000 scale geological

*

map, the stratigraphy and lithology of the study area were grouped and merged into the geological map ( Fig. 2 ) . Correlations between the distribution of gy[9, 10] demonstrate that earthquake-induced geological disasters are most common in areas underlain by hard

shale, are also at risk for these features. Collapses u-

. .

Epicenter

<1000 1000-1500

Mingshan

1t.a . Tianquan

rock, such as igneous rocks, carbonates and sandstone. Areas of soft rock, such as slate, phyllite and

*

Lusban

damage from the Wenchuan earthquake and litholo-

Figure 4

2500-3000 >3000

Elevation classification map in the study area

26

Geodesy and Geodynamics

Vol.4

Slope indicates the inclination of the ground surface ,

is the difference between the highest and lowest eleva-

while the slope change rate is the second derivative of

tions. These two indicators are both macro terrain fac-

the ground elevation. The slope in the study area ran-

tors. Most of the curvature values in the study area are

ges from 0 to 84. 2 ° , and the slope change rate ranges

between - 2 ° and 2 ° , and the relief amplitudes are

from 0 to 49. 7°. The classification maps of the slope

mainly between 0 and 180 m. The curvature and relief

and slope change rate are shown in figure 5. In the

amplitude classification maps are shown in figure 6.

Wenchuan earthquake, 86. 8% of earthquake-induced

Most of earthquake-induced geological disasters in the

geological disasters were located in areas with slopes

W enchuan earthquake occurred in areas where the cur-

between 20° and

The maximum density of

vature was between - 1 and - 0. 5 and between 0. 5

these secondary seismic effects occurred in areas with

and 1 and in areas where the relief amplitude was be-

slopes between 40° and 50°. The density of features in

tween 50 and 100 m[s, 9 l. The distributions of these

the Wenchuan earthquake was used for the slope factor

secondary seismic effects by curvature and relief ampli-

of the Lushan earthquake. Most of the earthquake-in-

tude from the Wenchuan earthquake were used as the

duced geological disasters in the Wenchuan earthquake

values of the curvature and relief amplitude factors for

occurred in areas where the slope change rate was be-

the Lushan earthquake.

soo[S-!0].

tween 20° and 30° [sJ. The density of features in the

To eliminate the influence of the dimensions of the

Wenchuan earthquake was used for the slope change

different factors , a normalization process was applied

rate factor of the Lushan earthquake.

after the values of the different states were applied to

Curvature is a quantitative measure of the degree that the ground surface curves, while relief amplitude

each factor. The normalized values of the earthquake hazard factors are shown in table 2.

N

A

•. J Baoxing

Baoxing

* Lushan

s 10 Tianquan -==km

o

(a) Slope

(b) Slope change rate

Figle 5

Classification maps of slope and slope change rate

/

N

A

Baoxing

*

* .

Lushan Mingsh~

10km Tianquan

· <-2 = ·2-1.5 -1.5--1 ·1-0.5 0.5-1 · 1-1.5 . 1.5-2

•o >2

(a) Curvatire

Fig. 6

..

Lushan ) Mingshan. 10 ; Tianquan " Jan

I

/

(b) Relief amplitude

Classification maps of curvature and relief amplitude

N

A

Li Xue, et al. Factor analysis of earthquake - induced geological disasters of

No.2

27

the M7. 0 Lushan earthquake in Chioa Table2

Seismic intensity (I)

State group

Lithology (L)

State gronp

Elevation (E)

State group

Slope (S)

State group

Slope change rate (R)

State group

Curvature (C)

State gronp

Relief amplitude (A)

State group

Values of earthquake hazard factors

Normalized landslide density

Normalized hazard distribution

Normalized landslide area ratio

Normalized hazard density

Nonnalized im=al distriOOiion ratio

Nonnalized im=al distriOOiion ratio

Nonnalized im=al distrihution ratio

II

12

13

I4

0.87

0.09

0.03

0.01

Ll

L2

L3

L4

LS

1.6

L7

L8

0.02

0.06

0.19

0.09

0.09

0.08

0.19

0.28

El

E2

E3

E4

E5

E6

0.11

0.28

0.19

0.17

0.14

0.11

Sl

S2

S3

S4

S5

S6

S7

0.04

0.13

0.16

0.21

0.27

0.10

0.09

Rl

R2

R3

R4

R5

0.07

0.27

0.4

0.2

0.06

Cl

C2

C3

C4

C5

C6

C7

C8

C9

CIO

0.01

0.03

0.27

0.15

0.02

0.05

0.18

0.24

0.04

0.01

AI

A2

A3

A4

A5

0.02

0.18

0.65

0.12

0.03

Table 4

4. 2

Judgment matrix of factor weights

Calculation of weights I

L

E

s

R

c

A

w,

I

2

1/2

114

4

6

8

0.1676

L

1/2

I

1/3

115

2

3

5

0. 0964

E

2

3

I

1/3

4

5

6

0. 2110

s

4

5

3

6

7

8

0.3973

R

1/4

1/2

1/4

1/6

I

3

5

0. 0671

c

1/6

1/3

1/5

117

1/3

3

0.0384

A

1/8

115 1/6 1/8 115 1/3

I

0.0222

Factors

In this paper, the Analytic Hierarchy Process { AHP) method is used to calculate the weights of the factors. The importance of each pair of factors is first determined based on the scale values shown in table 3. A judgment matrix of factor weights can then be obtained {Tab. 4). Finally, the weight of each factor is calculated by the matrix. The calculated weights of seismic intensity , lithology , elevation , slope , slope change

rate, curvature and relief amplitude are 0. 1676, 0. 0964, 0. 2110, 0. 3973, 0. 0671, 0. 0384 and 0. 0222 , respectively. The consistency ratio of the

4. 3

Results and verification

judgment matrix is 0. 0588, because the ratio is less

Mter the state grouping and assignment, rasterization

than 0. 1, the weights are reasonable.

and weighting are performed. The susceptibility index of secondary features from the Lushan earthquake is calculated using equation 1. The values of SI are nor-

Scale value

Meaning

malized to the interval [ 0, 1] and are classified. The

I

The factors have the same importance

distribution of earthquake-induced geological disasters

3

The fii'St factor is slighdy more important

from the Lushan earthquake is shown in figure 7. The

5

The fii'St factor is more important

mean SI is 0. 37, and the standard deviation of the SI

7

The fii'St factor is much more important

values is 0. 17. Based on the histogram of SI and

9

The first factor is very much more important

knowledge of the regional seismicity , the study area is

lntennediate values between the values described above

divided into five levels of secondary seismic features,

2,4,6,8

IIi( i = 1,2, ··· ,9) Opposites of the values described above

including very low, low, medium, high and very high. The statistics of each level are shown in table 5. Regions with SI values greater than 0. 71 are extremely

28

Vol. 4

Geodesy and Geodynamics

prone to landslides and collapses. The earthquake-in-

and the total area of high density is approximately

duced geological disasters are mainly distributed on

247. 9 km

both sides of the ridges that extend from Baosheng to

than 0. 62 are medium-low areas. These regions mainly

Shuangshi and from Baosheng to Longxing. In addi-

distributed in areas of relatively flat terrain and are

tion , several features are scattered on both sides of the

unlikely to be affected by earthquake-induced geolog-

ridges that extend from Qishuping to Baosheng and

ical disasters.

from Masangping to Lingguan. The region with a very

2

•

Regions in which the Sl values are less

In figure 7 ,

CD - ®

show post -earthquake remote

high density of features trends to the NE , and the total

sensing images provided by the Lushan County Geo-

area with a very high density of features is approximate-

graphic Information Publishing Platform [14 J • The col-

2

Regions in which the Sl values range

lapses and landslides are marked with red dots in the

from 0. 62 to 0. 71 are also prone to secondary seismic

images. The post -earthquake remote sensing images

effects. Areas of high densities occur on both sides of

show the consistency of the analysis results with the

the ridges from Qishuping to Chengxiang, Qishuping to

earthquake-related features.

Fenghe and Masangping to Nanmuping. The region of a

demonstrates the effectiveness and feasibility of the

high density of disasters also has a general NE trend,

proposed method.

ly 52. 4 km

•

The

Tianquan 10 illlllllllli:::==::::Skm·

Figure 7

Analysis and validation of the distribution of secondary seismic features

good

consistency

Li Xue, et al. Factor analysis of earthquake - induced geological disasters of

No.2

29

the M7. 0 Lushan earthquake in China

Table 5

553/100342/100342/

Statistics of the distribution of [2]

secondary seismic features Area (km')

SI

Danger assessment of earthquake-induced geological disasters in

Area ratio (%)

<0.37

1185.5

40.4

Low

0. 37 -0.54

978.5

33.4

Medium

0. 54-0.62

467.0

15.9

Very low

High Very high

mdex. html.

Lin Fengmin, Zhang Lihai, Lin Haiqing and Zhang Yecheng.

0. 62-0.71

247.9

8.5

>0. 71

52.4

1.8

China. JoumalofGeomechanics, 2006,12(2):127-131. (in

Chinese) [ 3]

Yang Xin, Tang Guoan, Liu XuejlDl, Li Fayuan and Zhu Shijie. Digital terrain analysis: Theory, method and application. Acta Geog"Phica Smica, 2009, 64(9) ,1058-1070. (in Chine•e)

[ 4]

Xu Chong, Dai Fuchu, Yao Xing, Cheng Jian, Tu Xinbing, Sun Yu and Wang Zhiyi. GIS-based landslide susceptibility assessment using analytical hierarchy process in Wenchuan earthquake region. Chinese Journal of Rock Mechanics and Engineering,

2009 , 28 ( 2) , 3978 - 2985. ( in Chine•e)

5

[5]

Conclusions

Wang Xilin , Xie Baoyuan and Guan Yuanbin. Ecological security assessment of heavy debris flow regions: A case study in Wen-

chnan county. Chinese Jonmal o f Ecology, 2008 , 27 ( 11 ) ,

1 ) Earthquake-induced geological disasters will form in regions with the following characteristics : glutenite and

1990 -1996. ( m Chine.e) [6]

Han Yongs1nm, Lang Chuan, Cui Peng, Han Jon, Xue Jiao and

carbonate bedrock , a seismic intensity of IX , slopes

Zhang Yongxiang. Susceptibility of mountain hazards triggered by

between 40° and 50°, elevations less than 2500 m,

Wenchuan earthquake to topographic factors. Journal of Sichuan

slope change rates between

zoo

Univemty(Engineering Science Edition), 2010, 42(Snpp.l),

and 30°, slope curva-

tures from - 1 to - 0. 5 and 0. 5 to 1 , and relief be-

15 - 21. (in Chinese) [7]

tween 50 and 100 m.

duced landslides: An overview. Geological Review, 2010, 56

2) Regions in which the susceptibility index is greater than 0. 71 are prone to landslides and collapses.

(6), 860-874. (m Chine.e) [8]

Wang Qiuliang, Xue Hongjiao, Yang Qiang, Wang Dun, Li linggang and Zhang Lifen. Analysis on terrain factors of earthquake

The earthquake-induced geological disasters are mainly

disaster in Beichuan area. Journal of Geodesy and Geodynamics,

distributed on both sides of the ridges that extend from Baosheng to Shuangshi and from Baosheng to Longx-

Xu Chong, Dai Foehn and Xu Xiwei. Wenchuan earthquake-in-

2008, 28(6) ,69-72. (m Chin""') [ 9]

Xu Chong, Dai Foehn, Yao Xin,

Zhao Zhou and Xiao Jianzhang.

ing. In addition, several features are scattered on both

GIS platform and certainty factor analysis method based Wen-

sides of the ridges that extend from Qishuping to

chuan earthquake-induced landslide susceptibility evaluation.

Baosheng and from Masangping to Lingguan. The dis-

Journal of Engineering Geology, 2010, 18 ( I) , 15 - 26.

tribution of disasters has a NE trend , and the region

[10]

tribution rules of geohazards induced by Wenchuan earthquake on

that was most affected by the Lushan earthquake covers

12th May 2008. Chinese Journal of Rock Mechanics and Engi-

an area of approximately 52. 4 km 2 • The analysis resnlts are consistent with the actual distribution of fea-

Huang Runqiu and Li Welle. Research on development and dis-

neering, 2008, 27(12) ,2585-2592. (in Chine"') [11]

Xu Chong, XuXiwei, WuXiyan, DaiFuchu, YaoXinandYao Qi. Detailed catalog of landslides triggered by the 2008 Wen-

tures.

chuan earthquake and statistical analyses of their spatial distribu-

3) Many factors affect the distribution of earthquake-

tion. Jonmal of Engffieering Geology, 2013, 21 (I) , 25 - 44.

induced geological disasters. Further research is necessary to increase knowledge of the influence of these fac-

(m Chinese) [ 12 ]

tors. Meanwhile, regional differences in the factors

Huang Runqiu and Li Weile. Analysis on the number and density

of landslides triggered by the 2008 Wenchuan earthquake , China. Journal of Geological Hazards and Environment Preserva-

shonld be considered to improve the accuracy of the a-

tion, 2009, 20(3) ,I -7. (m Chin""')

nalysis of earthquake-related damage. [ 13]

Seismic intensity map of the Ml. 0 earthquake in Sichuan province Lushan April20. 2013. http://www. cea. gov. en/publish/

References

dizhenj/464/478/ 20130425153642550719811/index. html. [ 14]

[ 1 ]

April 20, 2013 MI. 0 earthquake in Sichuan Province Lushan special covemge. http://www. cea. gov. cn/publish/izhenj/468/

Lushan County earthquake Ceogmphic information release plat-

form. http :1/www, scgis. net/ LSXEarthquake/.