Natural Hazards 29: 387–403, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
387
Building Damage and Casualties after an Earthquake Relationship between Building Damage Pattern and Casualty Determined Using Housing Damage Photographs in the 1995 Hanshin-Awaji Earthquake Disaster LU HENGJIAN1,2, MASAYUKI KOHIYAMA1,3, KEI HORIE1,4, NORIO MAKI1,4, HARUO HAYASHI1,4,5 and SATOSHI TANAKA5
1 Earthquake Disaster Mitigation Research Center (EDM), Institute of Physical and Chemical Research (RIKEN); 2 Presently, Earthquake Engineering Institute of Seismological Bureau of Shanghai, China, No. 87 Lan Xi Road, Shanghai, 200062, China; 3 Presently, Institute of Industrial Science (IIS), University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo, 153-8505, Japan; 4 Presently,
Earthquake Disaster Mitigation Research Center (EDM), National Research Institute for Earth Science and Disaster Prevention (NIED), 2465-1 Mikiyama, Miki, Hyogo, 673-0433, Japan; 5 Disaster Prevention Research Institute (DPRI), Kyoto University, Gokasho, Uji, Kyoto, 611-0011, Japan (Received: 6 November 2000; accepted: 9 August 2001) Abstract. The relationship between building damage patterns and human casualties in Nishinomiya City – one of the most heavily damaged cities in the 1995 Hanshin-Awaji Earthquake Disaster – was investigated using photographs of damaged buildings. First, the photographs of buildings in which casualties occurred were identified, and the building damage patterns were judged based on the photographs considering the existence of survival space. Then the relationship between the building damage pattern and casualty occurrence, and the characteristics of casualty distribution, were investigated. The main findings were as follows: Most casualties occurred in relatively old two-story wooden buildings in which the ground floor completely collapsed without survival space; casualties occurred at all building damage levels including “no damage”, and it can be seen that building damage is the major, but not the sole cause, of casualties in an earthquake; in Nishinomiya City, the regional distributions of casualties due to the collapse of buildings that left no survival space is similar to that of casualties due to other types of building damage. Key words: 1995 Hyogo-ken Nanbu (Kobe) Earthquake, disaster, human casualties, building damage, survival space, Nishinomiya City, photograph.
1. Introduction Building collapse due to earthquakes is one of the major causes of human casualties. Until now, many studies on clarifying the relationship between building damage and human casualty have been conducted using building damage categories, such as “major damage”, “moderate damage” and “minor damage”. However, categorized data of damaged buildings are accumulated for emergency risk judgment, or the accumulation of statistical data of building damage after
388
LU HENGJIAN ET AL.
Figure 1. Major damaged area and location of Nishinomiya City (Building Research Institute of Japan, 1996).
an earthquake, and these categories do not reflect the different types of building damage patterns. For example, buildings categorized under “major damage” exhibit many kinds of damage patterns, from the pattern in which there is space for survival to the pattern of complete collapse. Since it is clear that the patterns of building damage greatly influence the number of human casualties, it is necessary to clarify the relationship between human casualty and building damage pattern. In this article, the relationship is investigated using photographs of buildings damaged in the 1995 Hanshin-Awaji Earthquake Disaster.
2. Damage in Nishinomiya City and Construction of Database The target of this study is Nishinomiya City, one of the cities most heavily damaged in the 1995 Hanshin-Awaji Earthquake Disaster. The damage in Nishinomiya City was typical of the disaster in terms of topography, socioeconomic situation and ground motion intensity which was 7 on the JMA scale. The major damaged area (Building Research Institute of Japan, 1996) and the location of Nishinomiya City are shown in Figure 1. Nishinomiya City had a population of 411,882 and approximately 96,000 buildings at the time of the earthquake (Administrative Bureau of the Ministry of Home Affairs, 1994). In addition, Nishinomiya City has an area of about 100 km2 . As shown in Figure 2, the northern area is mountainous, and there are few buildings in this area. The southern seaside area also has relatively few buildings. The buildings
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
389
Figure 2. Distribution of buildings.
and population are concentrated in the relatively narrow area between the Hanshin Railway and the Sanyo Shinkansen Line. It was reported that 1108 casualties, 19,500 major damaged buildings and 16,300 moderately damaged buildings were caused by the Hanshin-Awaji earthquake disaster in Nishinomiya City by 28 March 1996 (The 21st Century Hyogo Project Association, 1997). The geographic information system database of the earthquake disaster in Nishinomiya City, the Nishinomiya Built Environment Database, has been constructed (Maki et al., 1998; Lu et al., 1999), and it contains the following datasets: (1) a digital map including the information on buildings and structures in all areas of Nishinomiya City before the earthquake; (2) casualty data based on the results of autopsies by the Hyogo-ken Police Department; (3) building damage data investigated by the Architectural Institute of Japan and the City Planning Institute of Japan (AIJ & CPIJ) (Building Research Institute of Japan, 1996), and those investigated by Kobe University (Field Investigation Team on Great Hanshin Earthquake, Department of Civil Engineering, Faculty of Engineering, Kobe University, 1995); (4) building property data based on real-estate tax roll involving building attributes, such as structure, age and number of stories, and including the building damage certification issued by Nishinomiya City; (5) more than 11,700 building photographs taken of the entire disaster area after the earthquake; and (6) data on the distributions of simulated seismic intensity and ground conditions. One of the characteristics of this database is that it contains photographic data of damaged buildings and the locations and direction of “take” are recorded for each photograph. In addition, all data have been linked to each building unit. The database
390
LU HENGJIAN ET AL.
Figure 3. Components of the Nishinomiya Built Environment Database.
construction and components are shown in Figure 3. Using this database, building damage situations in Nishinomiya City in the 1995 Hanshin-Awaji Earthquake Disaster can be reproduced and the building damage patterns can be classified based on photographic data.
3. Standard of Classifying Building Damage Patterns Okada and Takai (1999) studied data in relation to the classification of building damage patterns and proposed the two charts shown in Figures 4 and 5. The chart in Figure 4 is for classifying wooden building damage patterns caused by ground motion. In this chart, these patterns were classified into 19 categories, which are described in Table I, based on the degree of damage. The chart in Figure 5 is for classifying nonwooden building damage patterns caused by ground motion. In this chart, the damage patterns were classified into six grades from D0 to D5. Horie et al. (2000) proposed the chart shown in Figure 6. Using this chart, the damage patterns of wooden buildings caused by ground liquefaction were classified into three levels based on the degree of subsidence and inclination of the damaged buildings. These three charts cover the wide variety of building damage patterns observed after an earthquake and provide a standard for classifying the patterns of building damage. To distinguish the building damage patterns in Nishinomiya City in the 1995 Hanshin-Awaji Earthquake Disaster using photographs, the above-mentioned charts were used.
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
391
Figure 4. Chart of wooden building damage patterns caused by ground motion (Okada and Takai, 2000).
392
LU HENGJIAN ET AL.
Table I. Damage patterns of wooden buildings (Okada and Takai, 1999) Damage Grade
Damage pattern
Type of damage
D0 (none) D1 (slight damage) D2 (moderate damage) D3 (heavy damage)
Nd0
No damage.
Md1
Walls have cracks and covering materials are peeled off.
Md2
Roofing tiles and wall mortar are widely peeled off.
Gd3
Portions of columns, beams or walls on the ground floor present structural damages, but the internal space remains unaffected. Portions of columns, beams or walls on the second floor present structural damages, but the internal space remains unaffected. Portions of columns, beams or walls on every floor present structural damages, but internal space remains unaffected. Most of the roofing tiles are collapsed, especially inwards. Portions of columns, beams or walls of a single-story building present structural damages, but the internal space remains unaffected. Loss of internal space due to damages in columns and/or beams of the ground floor. Loss of internal space due to damages in columns and/or beams of the upper floor. Loss of internal space due to damages in columns and/or beams of both floors. Loss of internal space due to damages in columns or beams of a single-story building. The roof of the first floor has fallen down or is nearly touching the ground. The roof of the first floor has fallen down or is nearly touching the ground, and the second floor also presents damages. The upper floor is damaged or has collapsed. The upper floor is damaged or has collapsed, and the ground floor also has considerable damages. Dwelling space of a single-story building is considerably lost, and the roof has fallen down or is nearly touching the ground. The roof of the second floor has fallen down or is nearly touching the ground. Completely collapsed and turned into rubble.
Ud3
Ed3
Rd3 Sd3
D4 (very heavy damage)
Gd4 Ud4 Ed4 Sd4
D5 (destruction)
Gd5− Gd5+
Ud5− Ud5+ Sd5
Cd5− Cd5+
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
393
Figure 5. Chart of nonwooden building damage patterns caused by ground motion (Okada and Takai, 2000).
4. Relationship between Human Casualty and Building Damage Pattern In Nishinomiya City, investigators of AIJ & CPIJ took more than 11,700 photographs of damaged buildings immediately after the earthquake, covering the entire disaster area. The sites in the 11,426 photographs were confirmed on the digital map of the Nishinomiya Built Environment Database, and these photographs, shown in Figure 7, were linked to the buildings. Among the linked 11,426 photographs, there are 408 photographs of buildings in which casualties occurred.
394
LU HENGJIAN ET AL.
Figure 6. Chart of wooden building damage patterns caused by liquefaction (Horie et al., 2000).
Figure 7. Area of photographs taken.
The number of buildings in which casualties occurred was 328 and the number of casualties was 471. Using the Nishinomiya Built Environment Database, first, the photographs of the buildings in which casualties occurred were selected, and then these buildings were examined using the above-mentioned charts to classify their damage patterns. Based on the results of classification, the number of damaged buildings with the same pattern and the number of casualties occurring in buildings with the same damage pattern were collected. Regarding the different building damage patterns, the number of damaged buildings and casualties were compared to clarify the re-
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
395
Table II. Number of buildings and casualties in each damage pattern Damage level
Damage pattern
Buildings
Casualties
Major damage without survival space
Cd5+ Cd5− Gd5+ Gd5− Ud5+ Ud5− Sd5 D5 Ud5+ Ud5− Sd5 Ed4 Gd4 Ud4 Sd4 D4 Rd3 Sd3 D3 Md2 Md1 D1 Nd0 D0 (Demolished)
62 37 118 16 0 0 1 3 1 1 0 6 6 1 5 0 2 1 2 4 6 0 7 13 36 328
92 47 159 20 0 0 1 27 1 1 0 7 7 1 6 0 2 1 3 4 7 0 8 15 62 471
Major damage with survival space
Moderate damage
Minor damage
No damage Unknown Total
lationship between building damage pattern and human casualty. In addition, we comparatively analyzed the building damage patterns and the number of casualties occurring in the damaged buildings of different damage levels, such as “no damage”, “minor damage”, “moderate damage” and “major damage”. In the buildings that collapsed with or without survival space, however, the level of “major damage” was further classified into two levels, namely, “major damage with survival space” and “major damage without survival space”. From the photographs, it can be seen that there were no casualties in the buildings damaged by ground liquefaction. Table II shows the results of the judging damage patterns of buildings in which casualties occurred. “Ud5+”, “Ud5–” and “Sd5” are further classified into “major damage with survival space” and “major damage without survival space” based on
396
LU HENGJIAN ET AL.
Building-ID: 48694 Casualties: 1 Damage Pattern: Gd5+
Building-ID: 47280 Casualties: 6 Damage Pattern: D5
Photographs 1–2. Patterns of total damage without survival space (1).
Building-ID: 78398 Casualties: 1 Damage Pattern: Gd5−
Building-ID: 69813 Casualties: 1 Damage Pattern: Cd5−
Photographs 3–4. Patterns of total damage without survival space (2).
Building-ID: 40217 Casualties: 1 Damage Pattern: Ed4 Photographs 5–6. Patterns of major damage with survival space.
Building-ID: 3386 Casualties: 1 Damage Pattern: Cd4
397
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
Building-ID: 39562 Casualties: 1 Damage Pattern: D3
Building-ID: 70792 Casualties: 1 Damage Pattern: Rd3
Photographs 7–8. Patterns of moderate damage.
Building-ID: 8489 Casualties: 1 Damage Pattern: Md2
Building-ID: 40286 Casualties: 1 Damage Pattern: Md1
Photographs 9–10. Patterns of minor damage.
Building-ID: 40473 Casualties: 1 Damage Pattern: Nd0 Photographs 11–12. Patterns of no damage.
Building-ID: 44176 Casualties: 1 Damage Pattern: D0
398
LU HENGJIAN ET AL.
the photographs. It can be seen that most of the damaged buildings in which casualties occurred were wooden buildings with damage pattern of “Gd5+”. In addition, numerous casualties also occurred in wooden buildings with damage patterns of “Gd5–”, “Cd5+” and “Cd5–”. These building damage patterns are characterized by a completely collapsed structure without survival space. Among the buildings in which casualties occurred, the number of buildings with damage patterns “Gd5+”, “Gd5–”, “Cd5+” and “Cd5–” comprise 71% of the total number of buildings, and 98% of the damage level of “major damage without survival space”. Each typical damage pattern of the building in which casualties occurred is shown in photographs 1–12. The results of judging the photographs indicated that the most casualties occurred in relatively old two-story wooden buildings in which the ground floor completely collapsed without survival space. There were also quite a number of cases in which both the ground and second floor collapsed. Based on the results of judging the damaged buildings, the number of damaged buildings and casualties were collected regarding the different building damage levels, such as “no damage”, “minor damage”, “moderate damage”, “major damage with survival space” and “major damage without survival space”. The percentage of damaged buildings and the mortality were determined in terms of each damage level. As shown in Figure 8, the results of the comparison show that more than 84% of the total number of casualties in Nishinomiya City occurred in the damaged buildings categorized as “major damage without survival space”. In addition, among the buildings in which casualties occurred, as shown in Figure 9, the number of damaged buildings with the damage level of “major damage without survival space” comprised 81% of the total number of buildings. The figure also shows the tendency that casualties occurred mostly in buildings that collapsed completely without survival space. Although casualties were very few in buildings with other damage levels, they occurred at all levels of building damage from “no damage” to “major damage with survival space”; moreover, the number of casualties did not increase with increasing extent of building damage among damage levels that left survival spaces.
5. Distributions of Casualties The total number of deaths due to the 1995 Hyogo-ken Nanbu Earthquake in Nishinomiya City was 1,108, including related deaths (e.g., deaths due to illness contracted as a result of the earthquake) (The 21st Century Hyogo Project Association, 1997). In this article, the casualty data are based on the results of autopsies and exclude the number of related deaths. Based on the Nishinomiya Built Environment Database, it was confirmed that 973 of the total number of deaths occurred in 711 houses. Based on the distributions of casualties in Nishinomiya City, there were few casualties in the northern mountainous and southern seaside areas. The
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
399
Figure 8. Percentage of casualties by building damage levels.
Figure 9. Percentage of buildings with casualties of different damage levels.
casualties were concentrated in the urban areas between the Hanshin Railway and the Sanyo Shinkansen Line, as shown in Figure 10. To investigate the relationship between the distribution of casualties and the building damage pattern, casualties were classified into two groups: those due to the collapse of buildings that left no survival space, and those due to other types of building damage based on the above-mentioned results of judging building damage patterns. From Figure 11, it is clear that the casualties due to the collapse of buildings that left no survival space are widely distributed in nearly all the urban areas of Nishinomiya City. On the other hand, the casualties due to other types of
400
LU HENGJIAN ET AL.
Figure 10. Distribution of casualties.
building damage are also widely distributed in nearly all the urban areas of the city. The regional distributions of casualties are similar to that of casualties due to the collapse of building that left no survival space. There are no obvious differences between the two categories of building damage patterns. In addition, the relationship between the distribution of casualties and the distribution of seismic intensity in Nishinomiya City has been investigated. As shown in Figure 12, it is clear that all urban areas between the Hanshin Railway and the Sanyo Shinkansen Line were subjected to high seismic intensity, and the northern mountainous area and the southern seaside area were subjected to lower seismic intensity. In the entire urban area, the simulated peak ground velocity was over 60 cm/sec. In particularly, in the area from the JR Tokaido Line to Sanyo Shinkansen Line, the simulated peak ground velocity was over 80 cm/sec and the highest simulated peak ground velocity was over 100 cm/sec. Comparing the regional distributions of casualties and the seismic intensity, it can be seen that the regional distributions of casualties are in agreement with the regional distributions of high seismic intensity. Because the urban areas and population of Nishinomiya City were concentrated in a relatively narrow area, and high seismic intensity was distributed throughout most of the urban area, the casualties caused due to the collapse of buildings that left no survival space and to other types of building damage both occurred throughout the entire urban area. There are no clear differences in the regional distributions
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
Figure 11. Distributions of casualties in buildings with and without survival space.
Figure 12. Distribution of simulated peak ground velocity.
401
402
LU HENGJIAN ET AL.
of casualties due to the collapse of buildings that left no survival space and to other types of building damage. 6. Conclusions The results of this study revealed the relationship between building damage pattern and casualty caused by the Hanshin-Awaji Earthquake Disaster. The main findings were as follows: (1) More than 84% of casualties occurred in buildings that collapsed without survival space. Most casualties occurred in relatively old two-story wooden buildings in which the ground floor collapsed completely without survival space. There were also quite a number of cases in which both the ground and second floors collapsed completely without survival space. (2) Although casualties were very few at other building damage levels, casualties occurred at all building damage levels including “no damage”. From this result, it can be seen that building damage is the major, but not the sole, cause of casualties in an earthquake. (3) In Nishinomiya City, the buildings and the population were concentrated in a relatively narrow area, and high seismic intensity was distributed throughout most of the urban areas. The regional distribution of casualties due to the collapse of buildings that left no survival space is similar to that of casualties due to other types of building damage. There are no clear differences between the two categories of building damage patterns. Acknowledgements This research was conducted as a project of the Earthquake Disaster Mitigation Research Center (EDM), the Institute of Physical and Chemical Research (RIKEN); EDM was placed under the National Research Institute for Earth Science and Disaster Prevention (NIED) on 1 April 2001. The photographs used here were provided by the Architectural Institute of Japan and the City Planning Institute of Japan. The authors wish to thank Dr. Akiyoshi Nishimura of Shiga University of Medical Science for the casualty data. References Administrative Bureau of the Ministry of Home Affairs: 1994, The Population Handbook of Resident Fundamental Register, p. 78 (in Japanese). Building Research Institute of Japan: 1996, Final Report of Damage Survey of the 1995 HyogokenNanbu Earthquake (in Japanese). Field Investigation Team on Great Hanshin Earthquake, Department of Civil Engineering, Faculty of Engineering, Kobe University: 1995, 2nd Report on Great Hanshin Earthquake (in Japanese). Horie, K., Maki, N., Kohiyama, M., Lu H., Tanaka, S., Hashitera, S., and Hayashi: 2000, Development of building damage chart for post disaster management, Proceedings of the 12th World Conference on Earthquake Engineering, CD-ROM.
BUILDING DAMAGE AND CASUALTIES AFTER AN EARTHQUAKE
403
Lu H., Maki, M., Tanaka, S., Hashitera, S., and Hayashi, H.: 1999, Construction of a built environmental inventory database from the Great Hanshin-Awaji Earthquake, Proceedings of the 6th Japan/United States Workshop on Urban Earthquake Hazard Reduction, J-7-7, pp. 581–584. Maki, N., Lu, H., Tanaka, S., Hashitera, S., Nishimura, A., and Hayashi, H.: 1998, Database construction of the building collapse after the Great Hanshin Awaji Earthquake Disaster, Papers of the Annual Conference of the Institute of Social Safety Science, No. 8, pp. 78–83 (in Japanese). Okada, S. and Takai, N.: 1999, Classifications of structural types and damage patterns of buildings for earthquake field investigation, J. Structural and Construction Engineering, No. 524, pp. 65–72 (in Japanese). Okada, S. and Takai, N.: 2000, Classifications of structural types and damage patterns of buildings for earthquake field investigation, Proceedings of the 12th World Conference on Earthquake Engineering, CD-ROM. The 21st Century Hyogo Project Association: 1997, The Reconstruction of the Great Hanshin-Awaji Earthquake Disaster, No. 1, p. 59 (in Japanese).