Pure and Applied Geophysics

, Volume 168, Issue 3, pp 731–738

Seismic Hazard and Risk Assessments for Beijing–Tianjin–Tangshan, China, Area

Authors

  • Furen Xie
    • Institute of Crustal DynamicsChina Earthquake Administration
    • Kentucky Geological Survey
  • Jingwei Liu
    • Institute of Crustal DynamicsChina Earthquake Administration
    • Institute of GeologyChina Earthquake Administration
Article

DOI: 10.1007/s00024-010-0115-z

Cite this article as:
Xie, F., Wang, Z. & Liu, J. Pure Appl. Geophys. (2011) 168: 731. doi:10.1007/s00024-010-0115-z

Abstract

Seismic hazard and risk in the Beijing–Tianjin–Tangshan, China, area were estimated from 500-year intensity observations. First, we digitized the intensity observations (maps) using ArcGIS with a cell size of 0.1 × 0.1°. Second, we performed a statistical analysis on the digitized intensity data, determined an average b value (0.39), and derived the intensity–frequency relationship (hazard curve) for each cell. Finally, based on a Poisson model for earthquake occurrence, we calculated seismic risk in terms of a probability of I ≥ 7, 8, or 9 in 50 years. We also calculated the corresponding 10 percent probability of exceedance of these intensities in 50 years. The advantages of assessing seismic hazard and risk from intensity records are that (1) fewer assumptions (i.e., earthquake source and ground motion attenuation) are made, and (2) site-effect is included. Our study shows that the area has high seismic hazard and risk. Our study also suggests that current design peak ground acceleration or intensity for the area may not be adequate.

Keywords

Seismic hazardseismic riskseismic hazard analysishazard curve

1 Introduction

The study area is located in the northeast part of the north China plain and includes several major cities, for example Beijing, Tianjin, and Tangshan (Fig. 1). The area has a long history of earthquakes and has experienced many strong and large earthquakes. The largest historical event is the Sanhe-Pinggu earthquake (M = 8.0) of September, 1679. The most devastating earthquake occurring in the study area was the Tangshan earthquake (M = 7.8) of July 26, 1976, which leveled the whole of Tangshan City, killed more than 240,000 people, and caused huge economic loss. Thus, the area is facing significant seismic hazard and risk. Furthermore, the area is the political, economical, and cultural center of China. Therefore, specific measures, including better seismic design of buildings and infrastructures, are needed to mitigate seismic hazards and to reduce seismic risk in order to prevent a major disaster, like the 1976 Tangshan earthquake, in the area. Development of a sound mitigation measure requires better estimates of seismic hazard and risk.
https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig1_HTML.gif
Fig. 1

The study area and 0.1° × 0.1°cells

Although the terms “seismic hazard” and “seismic risk” have been used interchangeably, they are two fundamentally different concepts (Wang, 2006, 2007, 2009a). More importantly, seismic risk is more useful for engineering design and other policy consideration. Seismic hazards generally describe “earthquake-related natural phenomena such as ground shaking, fault rupture, or soil liquefaction” (Reiter, 1990, p. 3) or “a property of an earthquake that can cause damage and loss” (McGuire, 2004, p. 7). Seismic risk generally describes “the probability of occurrence of these consequences (i.e., adverse consequences to society such as the destruction of buildings or the loss of life that could resulted from seismic hazards)” (Reiter, 1990, p. 3) or “the probability that some humans will incur loss or that their built environment will be damaged” (McGuire, 2004, p. 8). Therefore, in general or qualitative terms, seismic hazard describes the natural phenomenon or property of an earthquake whereas seismic risk describes the probability of loss or damage when humans and their built environment (i.e., vulnerabilities) are exposed to a seismic hazard (Wang, 2009a). The relationship between seismic hazard and risk can be expressed qualitatively as:
$$ {\text{Seismic}}\,{\text{Risk}} = {\text{Seismic}}\,{\text{Hazard}} \times {\text{Vulnerability}}. $$
(1)
In quantitative terms, seismic hazard is determined on the basis of three measurements: physical measurement (i.e., fault rupture, strong ground motion, liquefaction, etc.), spatial measurement (where the event will take place), and temporal measurement (when or how often the event), with associated uncertainties. Seismic hazard is assessed on the basis of instrumental, historical, and geological observations. Seismic risk is determined by four variables: probability, level of severity (i.e., a physical or monetary measurement), and spatial and temporal measurements (Wang, 2009a). Seismic risk quantification is complicated and somewhat subjective, because it depends on the desired measurement (i.e., magnitude, ground motion, fatalities, or economic loss), how the hazard and vulnerability interact in time and space, and physical measures. In order to estimate seismic risk, a model has to be assumed or introduced to describe how an earthquake occurs in time. The most commonly used model for seismic risk estimation is the Poisson model. If earthquake occurrence in time follows a Poisson distribution (Cornell, 1968; Milne and Davenport, 1969; Wang, 2006, 2007, 2009a), then seismic risk, expressed in terms of a probability p of an earthquake exceeding a specified magnitude (M), can be estimated by use of the equation:
$$ p = 1 - e^{ - t/\tau } , $$
(2)
where τ is the average recurrence interval of an earthquake of M or greater, and t is the exposure time for a given vulnerability. Equation 2 can also be used to estimate seismic risk in terms of the probability of the intensity exceeding a specified level (I) (Cornell, 1968; Milne and Davenport, 1969; Bozkurt et al., 2007). Equation 2 is also commonly used to estimate flood and wind risks (Gupta, 1989; Sachs, 1978).

In this paper, we estimated seismic hazard and risk for the Beijing–Tianjin–Tangshan area from historical intensity observations. First, we digitized the historical intensity records for 0.1° × 0.1° grid cells. Then, we performed analyses on the digitized intensity records and determined the intensity–frequency relationship (hazard curve) for each cell. Finally, we calculated seismic risk for each cell and the area.

2 Intensity Data

China has long and rich historical records on earthquakes (CEA, 1999a, b). According to Huanget al., (1994), the earthquake catalog is complete for MS ≥ 4.75 since 1484 in north China. From the earthquake catalogs (CEA, 1999a, b), we obtained 73 earthquakes and their intensity observations (maps) since 1500. The intensity scale used in the study is the Chinese intensity scale with 12 grades, I through XII. Some of the intensity maps only showed felt range with no intensity values. We used the intensity attenuation relationship of Wang and Wu (1993) to calculate intensity I
$$ I = 2.429 + 1.488M - 1.391\ln (R + 11), $$
(3)
where M is magnitude and R is radius from the average axis in kilometers. Based on the coverage, geological characteristic, and population density, the study area was divided into 0.1° × 0.1° cells (Fig. 1). The intensity maps were digitized using ArcGIS under the WGS1984 coordinate system. As shown in Fig. 1, the intensity (I ≥ 4) in each cell varies from 7 (lowest) in the north to 52 (highest) in the south-west.

3 Hazard Analysis

The purpose of seismic hazard analysis is to determine ground motion or intensity with its associated occurrence interval or frequency, and the associated uncertainties, at a site of interest. Similar to the Gutenberg–Richter relationship, earthquake intensity and its occurrence frequency for an individual cell follows:
$$ \log (f) = a - b \cdot I, $$
(4)
where f is the frequency with which the intensity exceeds I, and a and b are constants determined by least-squares fitting (Milne and Davenport, 1969; Bozkurt et al., 2007). Figures 2a, b, c show the data points, the intensity–frequency curve, the a and b values obtained, and the standard deviations for the Beijing, Tianjin, and Tangshan cells (Fig. 1). The least-squares residual (∏) for the frequency–intensity curve is:
$$ \Uppi = {\frac{{\left[ {\log \left( {f_{\text{obs}} } \right) - \log \left( {f_{\text{fit}} } \right)} \right]^{2} }}{{N_{I} }}}, $$
(5)
where NI is number of intensity data, ∏ is least-squares residual, and fobs and ffit are observed and predicted frequencies, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig2_HTML.gif
Fig. 2

The frequency–intensity curves for the Beijing, Tianjin, and Tangshan cells

The range of a and b values for all cells are from −1.31 to 2.07 and from 0.11 to 0.95, respectively. In order to limit the effect on the b value of the limited number of intensities and the individual high intensity, we applied an average b value of 0.39 to all cells and derived the a value. In other words, the ratio of strong to weak shaking is taken as constant, but the frequency of shaking is permitted to vary (Bozkurt et al., 2007). Figures 2d, e, f show the data points, intensity–frequency curve, and value of a obtained, with standard deviation, with b = 0.39, for the Beijing, Tianjin, and Tangshan cells. As shown in Fig. 2, the intensity–frequency curves for constant and varied b values are quite similar for the Beijing and Tianjing cells, but significantly different for the Tangshan cell. The difference for the Tangshan cell is caused by the high intensity (XI) of the 1976 Tangshan earthquake. Figure 3 shows the least-squares residual (∏), ranging from 0.0002 to 0.1746, with a constant b value of 0.39.
https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig3_HTML.gif
Fig. 3

The least-squares residual (∏) for constanta b value of 0.39

We can estimate the frequency (f) or return period (1/f) for each cell by use of Eq. 4. We can also estimate the intensity for a given frequency or return period for each cell. Table 1 lists the return periods of different intensities for Beijing, Tianjin, and Tangshan. Figure 4 shows the intensity distribution for a return period of 100 years in the study area.
Table 1

Return period of different intensity for Beijing, Tianjin, and Tangshan

City

Coordinate

Cell number

Number of observations

Return period (years)

Longitude

Latitude

I = 7

I = 8

I = 9

Beijing

116.364°E

39.934°N

669

42

109

266

351

Tianjin

117.182°E

39.143°N

358

41

105

258

340

Tangshan

118.190°E

39.623°N

528

31

52

128

169

https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig4_HTML.gif
Fig. 4

Intensity distribution for a return period of 100 years

4 Risk Estimate

We can use the intensity–frequency relationships (hazard curves) and Eq. 2 to calculate seismic risk in terms of the probability of an intensity exceeding a specified level (I) in a certain period (i.e., 10, 30, or 50 years). For example, we estimated the 50 year exceedance probability for I = 8 (Fig. 5). Table 2 lists the 50 year exceedance probabilities for I = 7, 8, and 9 for major cities in the study area. Table 2 shows that the exceedance probabilities for I = 9 in 50 years are greater than 10 percent for all the major cities, except Chengde. According to the People’s Republic of China National Standard (PRCNS, 2001), there is a relationship between intensity and peak ground acceleration (PGA) (Table 3). In other words, we can also estimate seismic risk in terms of the probability of PGA exceeding a specified level in a certain period. Figure 6 shows the PGA map for 10 percent probability of exceedance in 50 years in the Beijing–Tianjin–Tangshan area.
https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig5_HTML.gif
Fig. 5

Exceedance probability for I = 8 in 50 year in the study area

Table 2

Seismic risk for major cities in the study area

City

Coordinate

Cell number

50 year probability of I ≥ 7/%

50 year probability of I ≥ 8/%

50 year probability of I ≥ 9/%

Longitude

Latitude

Botou

116.565°E

38.065°N

16

32

15

11

Cangzhou

116.859°E

38.313°N

99

31

14

11

Huanghua

117.348°E

38.368°N

137

33

15

12

Renqiu

116.087°E

38.709°N

208

35

16

12

Baoding

115.497°E

38.860°N

255

29

13

10

Tianjin

117.182°E

39.143°N

358

38

18

14

Zhuozhou

115.964°E

39.494°N

465

29

13

10

Langfang

116.687°E

39.587°N

512

46

22

17

Tangshan

118.190°E

39.623°N

528

58

33

26

Beijing

116.364°E

39.934°N

669

37

17

13

Chengde

117.916°E

40.967°N

1,084

16

7

5

Table 3

Relationship between intensity and peak ground acceleration (PRCNS, 2001)

Peak ground acceleration (g)

<0.05

0.05

0.10–0.15

0.15–0.20

0.20–0.30

0.30–0.40

≥0.40

Earthquake intensity

<VI

VI

VII

VII

VIII

VIII

≥IX

https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig6_HTML.gif
Fig. 6

PGA map with a 10 percent probability of exceedance in 50 years

5 Discussion

Many methodologies have been used to estimate seismic hazard and its associated uncertainties in time and space, and the estimates have been applied to seismic risk assessment. Among these methodologies, probabilistic seismic hazard analysis (PSHA) and deterministic seismic hazard analysis (DSHA) are the most commonly used worldwide (Cornell, 1968; Reiter, 1990; Krinitzsky, 2002; McGuire, 2004). Although many advantages have been acclaimed, PSHA is not based on valid physics and mathematics (Wang and Zhou, 2007; Wang, 2009b). Thus, the resulting hazard estimate from PSHA does not have a clear physical and statistical meaning and has caused so many problems (Wang, 2005, 2006, 2007, 2009b). For example, PSHA could result in consideration of a PGA of 10 g for engineering design of nuclear repository facilities at Yucca Mountain in Nevada (Steppet al., 2001). Even though DSHA has been labeled as an unreliable approach, it has actually been more widely used for seismic hazard assessment because it has clear physical and statistical bases. For example, Dinget al., (2004), Panet al., (2006), and Wang and Zhou (2007) estimated ground motion hazards from simulations of the 1697 Sanhe-Pinggu and 1976 Tangshan earthquakes. The biggest drawback of DSHA is that the temporal characteristics (i.e., the recurrence interval or frequency of ground motion) are often time neglected. This is one of the areas that must be addressed in DSHA because the frequency is also an important aspect of risk assessment and policy consideration (Wang, 2006, 2007, 2009a).

In this paper, we used about 500 years of intensity observations (records) to estimate seismic hazard and risk for the Beijing–Tianjin–Tangshan area. The advantages of using historical intensity observations are:

  1. 1.

    they are as free as possible of modeling assumptions; and

     
  2. 2.

    inclusion of site-effect.

     

There are also some limitations of this method, however. One of the limitations is that the period (i.e., 500 years) may not be enough to reflect the recurrence intervals of large earthquakes in the area. Wang (1984) and Liuet al., (1997) estimated that the recurrence interval of the Tangshan earthquake (M = 7.8) is about 1,500–7,500 years. Xianget al., (1988), Qiuet al., (1997) found that the recurrence interval of the Sanhe-Pinggu earthquake (M = 8.0) is about 7,000 years. This limitation may be compensated by the fact that the observed intensities were from all earthquake sources, not from single one. Another limitation is that a large individual intensity, such as those of the Sanhe-Pinggu and Tangshan earthquakes, affect the results. This can be seen clearly in Figs. 5 and 6 in which the higher intensities are concentrated in the Sanhe-Pinggu and Tangshan areas. This limitation may be corrected by using an average b value. As shown in Fig. 2, the b value (0.22) for the Tangshan cell is much lower than the average b value (0.39). This lower b value (0.22) for the Tangshan cell is caused by the high observed intensity (IX) of the 1976 Tangshan earthquake.

The hazard and risk estimates from this study are not only a good alternative but also an independent test of other methods. Figure 7 shows the seismic ground motion parameter zonation map for the Beijing–Tianjin–Tangshan area (PRCNS, 2001). The corresponding exceedance probability for the seismic ground motion parameter zonation map of China is 10 percent in 50 years (PRCNS, 2001). From Figs. 6 and 7, we can see that seismic risk derived from this study is higher than that being used for building seismic design in the study area. This suggests that the zonation map of China (PRCNS, 2001) might underestimate the seismic design ground motion parameter for the studied area. The damage observations from the May 12, 2008, Wenchuan earthquake also showed that the seismic ground motion parameter zonation map of China (PRCNS, 2001) is not sufficient in the epicentral area (Xieet al., 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs00024-010-0115-z/MediaObjects/24_2010_115_Fig7_HTML.gif
Fig. 7

Design peak ground acceleration for the Beijing–Tianjin–Tangshan area (PRCNS, 2001)

6 Conclusion

Seismic hazard and risk in the Beijing–Tianjin–Tangshan area were estimated from historical intensity observations since 1500. The advantages of using the intensity observations are:

  1. 1.

    fewer assumptions are made;

     
  2. 2.

    site-effect is included; and

     
  3. 3.

    intensity is directly related to damage.

     

If the past seismicity continues into the future, our study shows that the Beijing–Tianjin–Tangshan area has high seismic hazard and risk. Intensity 7 or greater could be expected in the Beijing–Tianjin–Tangshan area in the next 100 years. The probability of experiencing intensity 8 or greater in the area is larger than 10 percent. Our study illustrates that there are large uncertainties involved in seismic hazard and risk assessments. Hence, a specific confidence level should be considered for seismic design code formulation and seismic design of critical facilities. Our study also suggests that current design peak ground acceleration (PRCNS, 2001) for the area may not be adequate.

Acknowledgments

We thank Yanju Peng and Jinshi Hao for their help in ArcGIS digitization, and Yan Zhao, Xiaoliang Zhang, and Jiyang Ye for their assistance in data analyses. We thank Meg Smath of the Kentucky Geological Survey for editorial help. We also thank two anonymous reviewers for their valuable comments and suggestions that improved this manuscript greatly.

Copyright information

© Birkhäuser / Springer Basel AG 2010