SEVERITY OF MORTALITY

IN THE 2011 GREAT EAST JAPAN EARTHQUAKE:

AGE DEPENDENCY IN COMPARISON WITH

PAST MAJOR EARTHQUAKES

Yutaka OHTA1 and Maki KOYAMA2

1 Honorable Member of JAEE, Invited Researcher, Tono Research Institute of Earthquake Science,

Gifu, Japan, ohtay@mti.biglobe.ne.jp 2 Member of JAEE, Associate Professor, River Basin Research Center, Gifu University, Gifu, Japan,

maki_k@gifu-u.ac.jp

ABSTRACT: This paper aims at determining who is most vulnerable in disasters such as

an attack by an inland or oceanic earthquake. We found two major patterns (types U and

J, represented by English capital letters) when we introduced two-dimensional

coordinates having the X axis as the age group and the Y axis as age-specific mortality.

We also found two additional patterns of age independence: an initial case with the

lowest mortality near zero, and a case with extremely high mortality.

Key Words: The 2011 East Japan Earthquake, Age-Dependent Pattern, Severity of

Mortality, Comparison of Domestic and Overseas Earthquakes

1. INTRODUCTION

Authors are continuing their studies of the characteristics of mortalities from the 2011 Great East

Japan Earthquake compared with previous earthquakes and tsunamis. The purpose of this series of

study is the evaluation of mortalities. The first study1) is an evaluation of the severity of mortalities by

comparing the 2011 Great East Japan earthquake with other inland or subduction-zone earthquakes

around the world. The evaluation scale includes the severity of death and the severity of building

damage. However, these studies cannot consider individual attributes such as age, gender or social

roles, or characteristics of the earthquakes/tsunamis such as year or hour. Of course, these parameters

are very important for intrusive analysis. Based on this, the purpose of our second study is to model

the severity of mortalities by age groups.

It is well-known that the mortality rates for infants and aged people by natural disasters are high,

and that of the young generation is low. When the horizontal axis is age and the vertical axis is

mortality rate, its shape is called a bathtub curve. Hence, infants and the aged regarded as vulnerable

people. However, there is not enough evidence that these mechanisms are also same. On the other

hand, as an essential matter, what are the direct factors that affect the mortality rate? Of course, the

strength of the ground motion/tsunami is one of the factors. However, human behavior or the

surrounding environment are also large factors. The coupling effect of these factors is probably

important. Moreover, the mortality rate is affected by the status of the implementation of hardware and

- 37 -

software countermeasures. Data from previous earthquakes and tsunamis around the world are

collected and analyzed considering these various problems. As described above, this series study aims

to elucidate the problem of mortality in the 2011 Great East Japan earthquake. A comparative analysis

with previous earthquakes and tsunamis is fundamental to solve this problem. This paper focuses on

age dependency in mortality as an immediate issue.

2. PROBABILITY DENSITY FUNCTION OF BETA DISTRIBUTION FOR CASUALTY

MODELING

The probability density function of a beta distribution is suitable for modeling the mortality rate. This

is because this function is a continuous distribution from 0 to 1, and it can form various shapes

controlled by only two parameters ),( qp from this list2): simple increase, simple decrease, single

peaked pattern, or flat pattern. Therefore, this function is used in our analysis of mortality rates caused

by an earthquake or tsunami. We focus on the distribution profile pattern. Generally, the horizontal

axis shows the standardized age from 0 to 1, and the vertical axis shows the mortality rate. This

mortality rate is known as age-specific mortality. Here, age-specific mortality is represented by

formula (1):

21 ,,)( ConstqptBetaConstMortalityspecificAgeD (1)

where t is a variable of standardized age from 0 to 1. Both p and q are defined as positive

numbers. 1Const and 2Const are constant numbers to determine the absolute value of the

mortality rate.

In this study, we focused mainly on the differences in the distribution profiles; however, these

numbers have no crucial impact on the distribution profile. Therefore these constant values are only

used for arranging the shape. The distribution profile of the beta function is changed by a combination

of parameters p and q . Here, generally, forms of age-specific mortality by natural disasters do not

show a simple decrease. In many cases, characteristics of age-specific mortality are classified into four

types. First, mortalities of infants and aged people are high, and those of working-age people are low.

In this case, the shape of the distribution is similar to a capital letter U. Second, mortalities of aged

people are high, those of infants are slightly high, and those of working aged people are low. In this

case, the shape of distribution is similar to a capital letter J. Third, age-specific mortality shows a flat

shape and the mortality value is low. Fourth, age-specific mortality shows a flat shape, and the

mortality value is high. In particular, U and J types are often seen in many cases of natural disasters.

These four types of characteristics are shown in Fig. 1 (a). These four types are expressed when

parameters p and q of the beta function are under the following conditions: 0.1,0 qpp .

Fig. 1 (a) shows the probability density function of the beta distribution2), with the authors’ postscript.

This figure represents various transformations by values of the parameters p and q .

We focus on the relationship between parameters ),( qp and the shape of the functions. When

both p and q are less than 1, the shape of the function is U type. The U type appears only in this

case. In particular, the closer both p and q are to 0, the deeper the U depth. Type J appears in a

wide range when p is 1 or higher and q is less than 1. However, the beta function can express

various type-J shapes. For example, when p is higher than 5, only the aged people’s mortality rate is

high.

On the other hand, when both p and q are 1.0, this is the only case where the mortality rate is

the same in each age group. In this case, the shape of the figure is flat; therefore we named this the flat

type. There are two types of flat: a high mortality rate near 100% and a low mortality rate near 0%.

Based on these, the characteristics of the age-specific mortality rate are roughly divided into four

- 38 -

types: U, J, Flow, and Fextreme [Fig. 1 (b)].

Green part shows U-type distribution, and red part shows J-type distribution.

Yellow part shows flat-type distribution.

This means mortality rate is not dependent on age group.

Color hatching and comments were added by the authors.

Fig. 1 (a) Beta density function2)

Fig. 1 (b) Relationship between parameters ),( qp and type of function (U, J, Flow, or Fextreme)

Flat type is a singularity )0.1( qp

Of course, these types are not divided clearly, there are some intermediary types. In the next

section, we classify and coordinate the characteristics of the age-specific mortality rate of past

earthquakes, focusing on parameters p and q of the beta density function.

3. TARGET EARTHQUAKES

論文：採用版

J Type

U Type

Flat Type (Age Independent）

Beta密度関数 B{ｘ;p,q}でみる応答Types

J Type

0.0 1.0 2.0

P Value

1.0

0.0

q

Value

U Type

特異点 ｐ＝ｑ＝1.0Flat (Low, Extreme)

Singularity p=q=1.0

Flat(Low, Extreme)

- 39 -

Many researchers and governments collected and published the total number of deaths by previous

earthquakes or tsunamis. Generally, human biological and social attitude for living varies depending

on age, gender, economic situation, education, social level, etc. On the other hand, there are not many

statistical data about death that consider individual attributes such as those mentioned above. The

age-specific mortality rate was relatively obtainable in these data related to individual attributes.

Therefore, we focused on age-specific mortality. Because of casualties, the research field is highly

interdisciplinary, and casualty-related studies are published in a wide research field. Hence, data

collecting from a wide area including the medical field is needed. PubMed3) is the most popular

literature database in the wide range of medical field.

The available data for the age-specific mortality rate by earthquake are listed in Table 1. These are

the target earthquakes for our analysis. There are five Japanese earthquakes and five earthquakes that

occurred outside of Japan. Henceforth, any earthquake that occurred outside of Japan is called a

foreign earthquake. The typical active-fault earthquakes in Table 1 are the 1995 Great Hanshin-Awaji

earthquake in Japan, the 1970 Gediz earthquake in Western Turkey, the 1976 Çaldiran earthquake in

Eastern Turkey, the 1988 Armenia Earthquake in the former Soviet Union, and the 1999 Chi-Chi

earthquake in Taiwan.

Table 1 Target earthquakes

Earthquake Occ. Time M No. Deaths Remarks

1933 Showa Sanriku 02h31m, Mar. 3 8.3 1,522 Strong Shaking, Tsunami

1946 Nankai 04h19m, Dec. 21 8.1 1,331 Strong Shaking, Tsunami

1993 Hokkaido-Nansei-Oki 22h17m, July 12 7.8 230 Strong Shaking, Tsunami,

Fires

1995 Hanshin-Awaji 05h46m, Jan. 17 7.3 6,434 Strong Shaking

2011 Great East Japan 14h46m, Mar. 11 9.0 19,000 Giant Tsunami

1970 Gediz, Turkey 23h02m, Mar. 28 7.1 1,086 Shaking, Fires

1976 Çaldiran 14h22m, Nov. 24 7.1 3,840 Strong Shaking

1988 Armenia, USSR 11h41m, Dec. 7 6.9 25,000 Strong Shaking

1999 Chi-Chi, Taiwan 01h47m, Sept. 21 7.3 2,329 Strong Shaking

2004 Indian Ocean 07h58m, Dec. 26 9.3 230,000 Giant Tsunami 12

Countries

Marine earthquakes with tsunamis in Japan are the 1933 Showa Sanriku earthquake, the 1946

Showa-Nankai earthquake, the 1993 Hokkaido-Nansei-Oki earthquake, and the 2011 Great East Japan

earthquake. A particularly notable foreign marine earthquake with a tsunami is the 2004 Sumatra

earthquake in Indonesia. This earthquake took a heavy toll on more than 10 countries located on the

shores around the Indian Ocean. The severest damage in the 2004 Sumatra earthquake occurred in the

northern part of the Sumatra islands, located northwest of Indonesia. The following sections describe

our analysis focused on the age-specific mortality rate. A reference to the relevant data is mentioned

for each earthquake.

4. PATTERN CLASSIFICATION

As mentioned in Chapter 2, major types of age-specific mortality rates in a beta distribution function

are types U, J, Flow, and Fextreme. However, age-specific mortality patterns are not clearly divided into

these four types; rather, they are selected gradationally. Hence, some are called intermediate types, and

some cannot be determined. Regardless, we discuss the characteristics of the age-specific mortality

rate for each type.

- 40 -

4.1 TYPE U

For a Japanese earthquake with a tsunami, Yamashita4) reports on age-specific mortality rates. He is

one of the first researchers to address the theme of age dependency of casualties in Japan. He clearly

indicated that the mortality rate of people under 10 years old was enormously high during the 1933

Showa Sanriku earthquake. The characteristics of this age-specific mortality rate are classified as type

U. This earthquake occurred at 2:31 a.m. Japanese local time. People living near the Sanriku coast

woke up when they felt strong motion; however, they were not aware of the possibility of tsunami

attack and went back to sleep. Therefore many people failed to escape and died during the tsunami

attack. These areas were also affected by the 1896 Meiji Sanriku earthquake and tsunami. However,

people’s experience with the 1896 earthquake and tsunami was not utilized because of generational

shifts and a significantly high mortality rate. Shuto5) reported details of human behavior in affected

areas of the 1933 Showa Sanriku earthquake. In addition, Yamashita4) also reported about the 1993

Hokkaido-Nansei-Oki earthquake. The epicenter of this earthquake was located near the coast; hence,

the first tsunami wave attacked 5 min after the earthquake occurred, and people had very limited time

to evacuate or protect themselves. Moreover, the earthquake occurred at night. This is one of the

additional reasons for the high mortality rate among infants and aged people. In the 1933 Showa

Sanriku earthquake, age dependency was close to type U; and in the 1993 Hokkaido-Nansei-Oki

earthquake, age dependency was intermediate between types U and J. This is shown in Fig. 2.

Showa Sanriku Eq Hokkaido Nansei Eq

採 用

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0-10 11-20 21-30 31-40 41－５０ 51-61 60 +

Mo

rtal

ity

(%

)

Ten-year Age Intervals

1933 Shouwa Sanriku Eq 1993 Hokkaido Nansei oki Eq

Both age-specific population and mortality are shown in the original figure by Yamashita. Figure

shows age-specific mortality rate based on this population and death toll. Dotted line in the figure is a

polynomial approximation by the authors.

Fig. 2 Age-specific mortality rate

Left: 1933 Showa Sanriku earthquake

Right: 1993 Hokkkaido-Nansei-Oki earthquake

On the other hand, Miyano6) has conducted leading studies about human behavior that included

casualties of an earthquake or tsunami. One of these studies is of the 1993 Hokkaido-Nansei-Oki

earthquake. The field investigation team started its research immediately after the earthquake and

conducted a multilateral field survey. Based on this study, he reported that the age dependency of this

earthquake is intermediate, between the U and J types of this study, before Yamashita did4). Miyano

- 41 -

also introduced data on the age dependency of the 1933 Showa Sanriku earthquake, which is a type

belonging to U based on Sunamura’s survey. Miyano and Mochizuki7) conducted a sampling survey of

human behavior for the 1946 Nankai earthquake. They showed that infants’ and aged people’s

mortality rates were higher than those of other age groups. The distribution is similar to type U. In

addition, they also indicated that the female mortality rate was higher than that of the male. They

mentioned the reason is activities to protect the children, based on the high mortality rate of mothers

with children. The 1946 Nankai earthquake occurred immediately after World War II. Japanese land

was heavily attacked and many citizens lost their lives. The Japanese people and government were

fully shattered, and lacked every type of resource. Therefore, the Japanese people and government did

not have any power to consider against natural disaster countermeasures, from either a hardware or

software point of view.

Next, we mention a foreign earthquake. There are many studies about the 2004 Sumatra

earthquake; Doorcy et al8) conducted a particularly valuable study to understand age-dependent

mortality. They conducted a field survey of mortality considering age dependency, supported by the

Faculty of Medicine, Johns Hopkins University, USA. Their target area was the northern part of the

Sumatra islands, the Indian Ocean side facing the epicenter, and the eastern side of the peninsula.

Their four target sites suffered different tsunami effects. They took a common index of

Tsunami-displaced households as a statistical parameter in their study. Therefore, a multipoint

comparison is available. This study is valuable from this point of view. Fig. 3 shows a map of the

location of the epicenter and the target sites of their study8).

The two target sites (West Coast and Banda Ache) are on the Indian Ocean side of the peninsula.

These sites face the epicenter. The other target sites are Meulaboh, which is located 200 km mi south

of the peninsula, and the East Coast, which suffered no direct effect of the tsunami.

Fig. 3 Maps of 2004 Sumatra earthquake

Left: Location of epicenter and major affected countries

Right: Map of northern part of Sumatra island and location of target sites

Fig. 4 shows the age-specific morality rate of each site in Doorcy et al8). The authors summarized

the results for the West Coast and Banda Ache because the characteristics of the age-specific mortality

rates are very similar. This is shown by the red line in Fig. 4. The results for Meulaboh, located 200

km south of the west side of the peninsula, is shown by the blue line; and the result of the East Coast,

2004 Indian Ocean Earthquake (1)

Epicenter

West C.+Banda Aceh

Meulaboh

East Coast

- 42 -

which was not attacked by the tsunami, is shown by the green line. When the researchers selected their

investigation sites, they paid attention to tsunami strength but did not mention tsunami height.

Therefore, information about the tsunami height is cited from measurement results by Matsuyama9).

According to Matsuyama, a tsunami height of 10 m to 30 m was measured in the northeast area of the

peninsula at Banda Ache. In contrast, a tsunami height of less than 5 m was measured on the east side

of the peninsula. There is enough information about tsunami height at Meulaboh, located at the

southern part of the peninsula; it was estimated at approximately 5 m.

Fig. 4 Mortality rate of 2004 Sumatra earthquake

Left: Typical pattern of U in northern part of Sumatra islands

Right: Comparison of average mortality rates of investigation sites

Through a comparison of the right and left graph of Fig. 4, we see that the average mortality rate is

strongly affected by the height of the tsunami. Thus, the depth and height of the U distribution are

affected by the tsunami external force. Parameters p and q of the beta density function are both

under 1.0. In particular, the smaller the value, the deeper the U.

The description above shows the relationship between the mortality rate and tsunami strength. On

the other hand, there are many studies about the behavior of people living in the affected areas.

These studies, especially Takahashi et al.10), indicate a lack of countermeasures for both hardware

(coastal reinforcement, facilities for emergency contact networks, etc.), and software (disaster

education, evacuation behavior, etc.). These are examples of U-type distribution, in which infants’ and

aged people’s mortality rates are especially high. In essence, the following hypothesis is possible: If

the external force is massive and anti-disaster countermeasures are too inadequate, people are

defenseless and exposed the external force. If this is correct, the age-dependent mortality rate directly

shows age-dependent behavioral ability. This means that if the behavioral performance of infants and

aged people is low, then the mortality rate is high. Fig. 5 shows the relationship between age group

and behavioral performance. In Fig. 5, the graph on the left shows age-specific behavioral

performance by Kuroki11), and the graph on the right shows its inversion. These graphs indicate that

the inversion of human behavioral performance represents age-specific vulnerability, and its shape

shows the U pattern well. Nishikiori et al.2) also reported on the age-specific mortality rate of the south

coast of Sri Lanka, and it is similar to the U type.

2004年Indian Ocean Earthquake (2)

0

5

10

15

20

25

30

35

40

45

０－９

１０－１９

２０－２９

３０－３９

４０－４９

５０－５９

６０－６９

70

+

Mo

rtal

ity

( %

)

Ten-year Age Intervals

Northern Smatra

WC+BA

Meulaboh

East Coast

West Coast+Banda Ache

BB

- 43 -

Figure on right is inverse of figure on left. Figure on right shows age-specific behavioral

vulnerability.

Fig. 5 Human general age dependency of behavioral performance11)

(Dotted lines were added by authors)

4.2 Type J

A typical case of type J is the 1995 Great Hanshin-Awaji earthquake13). It is well known because it is

representative of an active fault earthquake that occurred in recent years in Japan. The age-dependent

mortality rate by this earthquake was in the area of JMA seismic intensity 7, which is shown in Fig. 6.

This earthquake occurred in the early morning (5:46 a.m.). Almost all residents were in their homes;

therefore there was the highest potential of child care or nursing care during the earthquake. In this

figure, the age-specific mortality rate is shown by gender, and is totaled. As shown by Fig. 6, the

differences in mortality rates by gender are far smaller than by age. Miyano and Sumiyoshi14) indicated

this. Therefore, we do not repeat the mortality differences by gender in this study.

The age-specific mortality rate of the 2011 Great East Japan earthquake is also a type-J

distribution. Fig. 7 (a) shows the age-specific mortality rate of three affected prefectures: Iwate,

Miyagi, and Fukushima. The horizontal axis indicates five-year age groups, and the vertical axis

indicates the mortality rate in areas washed away by the tsunami15). Here there are many missing

people because of the tsunami; however, we of course cannot know these people’s addresses.

Therefore, the missing people are not included in the mortality data of Fig. 7. Fig. 7 (a) indicates the

characteristic mortality rates of these three prefectures as follows: the trends of Miyagi and Fukushima

are similar, and Iwate has a lower mortality rate than Miyagi or Fukushima. One reason for these

differences is the strength of the tsunami. However, there are probably many other reasons, for

example, differences in coast types such as rias or plain. We need to conduct a deeper analysis.

Fig. 7 (b) is drawn for confirmation that the J pattern only appears at the prefecture or

municipality levels. The result is that the J pattern also appeared at the municipality level. There are

some other patterns in Fig. 7 (b); these are municipalities that recorded a very low death toll. The 2011

Great East Japan earthquake occurred at 2:45 p.m., and the average of allotted time until a tsunami

attacks is approximately 40 min in the three prefectures. Briefly, people had 40 min to evacuate to a

safe place. This makes a significant contribution to mitigate human casualties. In addition, the disaster

occurred on a weekday; therefore, most of the elementary school and junior high school students

stayed in their schools, where teachers could care for them (e.g. Sato 16)). This also contributes to

mitigating human casualties in the young generation. Moreover, the affected area of this earthquake

had historically been attacked by a tsunami many times. Consequently, there is a tradition known as

採 用

0

20

40

60

80

100

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

Be

ha

vio

ral

Pe

rfo

rm

an

ce

Five-year Age Intervals

Behavioral Performance

Male

Female

0

0.5

1

1.5

2

2.5

3

5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

Inve

rse

d P

erfo

rm

an

ce

Five-year Age Intervals

1/（Behavioral Performance）

0

5

10

15

20

25

30

0-4 5-14 15-44 45-64 65+

Mo

rta

lity

(D

ea

th p

er

Ye

ar :

10

5 p

ers

on

s)

Age Groupes

Daily Mortality in Dwelling ・Kuroki （2007）Geriatrics

・Naoi （1981）Architectural Planning

Reversed

- 44 -

Data is from the Editorial Committee for the Report on the Hanshin-Awaji Earthquake Disaster13).

Dotted line is the result of curve fitting for beta density function by the authors. Parameter p of

fitting curve is 6.25 for female, 10.21 for male, and 7.17 in total. However, q is near 1.0 for both

female and male. Mortality rate for aged people is approximately 10 times as that for infants.

Fig. 6. Age-dependent mortality rate of 1995 Hanshin-Awaji earthquake

2011 East Japan Earthquake

0

5

10

15

20

25

30

0-4

5-9

10

-14

15

-19

20

-24

25

-29

30

-34

35

-39

40

-44

45

-49

50

-54

55

-59

60

-64

65

-69

70

-74

75

-79

80

-84

85

+

Mo

rta

lity

(%

)

Five-year Age Intervals

Iwate Pref. Miyagi Pref. Fukushima Pref.

Generally, the mortality rate is calculated based on the death toll and population by national census

in an administrative district. In this case, the statistical parameters are not only in the affected area but

also the unaffected area in the same administrative district. According to the investigation results by

MLIT (Ministry of Land, Infrastructure and Transport), human life is strongly affected by tsunami

depth. In particular, the mortality rate is enormously high in the washed-away area. The mortality rate

in the washed-away area was calculated by Koyama et al15). Therefore, the mortality rate is not

calculated using the population of the unaffected area. The dotted lines are approximate curves for

reference by the authors. Mortality rates of aged people and infants are different among these three

prefectures. The mortality rate of aged people is eight times as high as that for infants in Fukushima,

twelve times in Miyagi, and fifteen times in Iwate.

Fig. 7 (a) Age-specific mortality rate in three affected prefectures: Iwate, Miyagi, and Fukushima

本採用

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80+

Mor

talit

y (%

)

Ten-year Age Intervals

Male Female Total

- 45 -

“Tsunami ten-den-ko,” which means that when an earthquake occurs, people should evacuate

immediately on an individual basis to save their own lives because there is no time to help evacuate

someone else in a tsunami situation. This tradition is also one reason for human loss mitigation.

The maximum value of mortality rate differs substantially among these three prefectures; therefore,

figures are drawn for each prefecture, and the scale of vertical axis is fitted for each maximum value.

Red bold line shows mortality rate of each prefecture; it is the same line as in Fig. 7 (a).

Fig. 7 (b) Age-specific mortality rate by each municipality in three affected prefectures

On the other hand, Armenian et al.19) reported on the age-dependent mortality rate of the 1988

Armenia earthquake. This is an initial study focusing on age dependency. However, they lumped the

mortality rate in the MSK seismic IX-X area together with the mortality rate in the VIII-IX area. It is

not certain that their result for the age-specific mortality rate is type J because the result included data

from different seismic intensity areas. One typical case of type J is the 1999 Chi-Chi earthquake. This

earthquake occurred by an inland reverse fault. Its age-specific mortality rate is similar to the recent

Japanese earthquake17). This earthquake is a typical inland reverse-fault-type earthquake. Therefore,

the mortality rate of the hanging-wall side (east side) is considerably larger than that of the foot-wall

side (west side) in the vicinity of the fault. The mortality ratio of the hanging-wall side and foot-wall

side is 2.4:118).

Fig. 8 The 1988 Armenia earthquake19)

Left: Map of investigated areas

Right: Age-specific mortality rate

改訂版（正）：3県市町村別の分布

Iwate Pref. Miyagi Pref. Fukushima Pref.

1988 Armenia Earthquake , Former USSR

- 46 -

4.3 TYPE FLAT

As we mentioned, there are two types of flat: a low-mortality-rate Flow and a high-mortality-rate Fextreme.

The mortality rates of both flat types are age independent. While the mortality rate of the Flow type is

nearly 0, that of Fextreme is nearly 100% as its ultimate state. These two flat types show a polar situation

of death. There are very few cases of type Fextreme; one example is a fishing village that was virtually

eradicated during the 2004 Sumatra earthquake. The population of the fishing village was

approximately 1,00020).

4.4 OTHER CASES

A mixed-type or special case of age-specific mortality rate occurred in two earthquakes in Turkey. The

epicenters of these earthquakes are shown in Fig. 9 (a)21). One is the 1970 Gediz earthquake, which is

a typical inland shallow earthquake. The affected area of this earthquake is located in western Turkey.

There are many wooden houses in this area because of a wealth of timber resources. The earthquake

occurred at 11:03 p.m., and large-scale fires also occurred in the epicenter area. It is rare in the world

for large-scale fires to occur during a midnight earthquake. Age-specific mortality rates were gathered

from three locations as near, middle, and far from the epicenter. The age-specific mortality rate of the

far locations is type Flow, and the others are type J. However, these J types have different amplitudes

and shapes. As seen above, the characteristics of the age-specific mortality rate are different based on

the strength of the ground motion. Whether it seems universal result but it is difficult to find

investigation data.

Fig. 9 (a) Two earthquakes in Turkey

The age-specific mortality rate of the 1978 Çaldiran earthquake is shown in the graph on the right

side of Fig. 9 (b). This earthquake occurred on the eastern end of Turkey, near the border with Iran and

near the cross-point of the northern and eastern Anatolian fault zones. The characteristic of age

dependency is unique from other characteristics. Its shape is similar to an L. The major types of

housing in these areas are adobe and stone buildings, and the aseismic capacity of these buildings is

enormously low. Therefore, the building collapse rate was also extremely high. The reason for the high

mortality rate among infants is that they stayed home at the time, and they were killed in a stampede

by their houses. In contrast, cattle breeding was active in this area. The earthquake occurred in

November, which was during harvesting season for feed grain. Most of the middle-aged and aged

people worked outside for harvesting operations. As a consequence, this is an anomalous case where,

the mortality rate of infants was high, and that of aged people was low. Additionally, if the earthquake

Earthquakes in Turkey (1)

-Inland and shallow-

1970 GedizWooden

1976 CardiranBrick

- 47 -

occurred at night, the type of age dependency would be U. Therefore, it may be said that the L pattern

is a special case, and we do not consider the L pattern as a typical type of age dependency.

Dotted lines are polynomial approximations added by the authors.

Fig. 9 (b) Mortality rate of two earthquakes in Turkey21)

5. CONCLUSION

5.1 SUMMARY OF RESULTS

Generally, vulnerable people in a disaster situation are known as CWAP (Children, Women, Aged

people, Poor or Patient)22). We focused on age-related factors such as children and aged people in

CWAP, and analyzed the age dependency of death during an earthquake or tsunami through a literature

survey. As a result, the characteristics of age-specific mortality rates are classifiable into four typical

patterns as type U, J, Flow, and Fextreme. These four types are ranked by the severity of death, as follows:

Flow < J < U < Fextreme

For type U, the mortality rate of both infants and aged people as vulnerable people is high. Typical

cases in this study are the northern part of Sumatra Island and the southern part of Sri Lanka during

the 2004 Sumatra earthquake. There are no typical U types in Japanese earthquakes; however, in the

case of old earthquakes such as the 1933 Showa Sanriku earthquake and the 1946 Nankai earthquake,

the age-specific mortality rates are near U or between the U and J patterns. In recent Japanese

earthquakes, type J is most like the 1995 Hanshin-Awaji earthquake and the 2011 Great East Japan

earthquake. Based on this analysis, we try to show a conceptual model of the age dependency of death.

The characteristic of age dependency is moved from Flow to J, U, and Fextreme in a phased manner.

Variables to change the type include the strength of the seismic intensity or the tsunami, age

dependence of behavioral ability, and the implementation status of hardware and software

countermeasures. Fig. 10 shows a conceptual model. This model shows that the mortality rate is

influenced by an external force as follows: When people are exposed to a small external force, their

mortality rate is low; however, the mortality rate increases as the external force becomes larger. Here,

death appears in aged people first; then the mortality rate of infants increases. If there is enough help

or care for infants, the age-specific mortality rate is type J. Inversely, if people have scarce knowledge

of anti-disaster protocols and countermeasures, the mortality rate of infants increases. As a result, the

Earthquakes in Turkey (3)

-Inland and shallow-

0

5

10

15

20

25

30

0-4 5-9 10-19 20-44 45-64 65+

Mor

talit

y (%

)

Age Groups

Akcaalan Gediz District

0

1

2

3

4

5

0-4 5-9 10-19 20-44 45-64 65+

Mor

talit

y (%

)

Age Groups

Spreading Fire J type

Age Independent

F type

U type (Deformed)

- 48 -

age-specific mortality rate changes to type U. In addition, when the external force increases, the

bottom of the U (the mortality rate of the young or middle aged) becomes elevated. Finally, the type of

age-specific mortality rate becomes Fextreme.

Fig. 10 Concept model for transition of age-dependent death patterns

5.2 FUTURE ISSUES

We constructed a concept model of the transition of age-dependent death patterns through an analysis

of age-specific mortality rate and relationships between death, external force, human behavioral

performance, and countermeasures. However, the result brings up new questions. One of them is the

meaning of the J type. The 2011 Great East Japan earthquake and the 1995 Hanshin-Awaji earthquake

are typical patterns of type J. If the age-specific mortality rate is took J type in the future earthquake,

we must consider some questions, as follows: How do we decrease the mortality rate of aged people?

It is difficult to care for and support both infants and aged people at the same time during a large

disaster. In particular, it is impossible in the case of a tsunami because time for everything is limited.

Among Japan's aging population, “elder-to-elder support” is required. Nevertheless, “elder-to-elder

support” is too severe for aged people in a disaster situation because of their human behavioral

performance. Without doubt, they are vulnerable in a disaster. The mortality rate of infants is

apparently low in the case of type J. At the same time, the situations for infants and aged people are

different even if both are vulnerable in a disaster. One of the reasons for the lower mortality rate of

infants may be that parents, neighbors, or teachers protect them. However, existing equations that

calculate the death toll cannot consider these human effects. Based on this, a more accurate equation

may be required. Therefore, we need more discussion about the human casualty model in a disaster.

On the other hand, Sawai23) indicates that the definition or representation scheme of the mortality

rate is different among the studies. There is also the problem of selecting a statistical parameter for

mortality rate calculation. As mentioned above, some studies adopt the population of municipalities,

some adopt the population of the investigated area, and other studies adopt the population of affected

area. Therefore, strictly speaking, each mortality rate means something different. This makes it

difficult to conduct a comparative study.

正採用：外力・人間（弱さ；防災力）・応答タイプ

Death Severity : Minor ・・ Medium ・・Large・・ Extreme

Better way for Protection

Transition of Age-dependent PatternsNo Awareness nor Protection

0 Flow U type Fextreme

J typeWith Awareness

and Protection

Increase of Seismic Shaking or Tsunami Height

- 49 -

ACKNOWLEDGMENT

We are deeply grateful to Dr. Fumiaki Kimata (present Senior Researcher of Tono Research Institute

of Earthquake Science), who was a reader of “Field Survey Team for Indonesia Ocean Earthquake and

Tsunami.” He provided much useful information and insightful comments about tsunami

circumstances and people’s countermeasures at the time. We also want to thank Dr. Michio Miyano,

who is a vice president of Osaka City University and has wide experience in casualty research. He

supports the gathering of related literature. We would like to thank Editage (www.editage.jp) for

English language editing. Funding for revision from the Cooperative Organizations of “Gifu Women

Empowerment Project,” supported by MEXT (Ministry of Education, Culture, Sports, Science and

Technology) is gratefully acknowledged

REFERENCES

1) Ohta, Y.: Severity evaluation of fatalities due to the 2011 East Japan Tsunamigenic Giant

Earthquake via introduction of a common evaluation scale and comparison with oceanic

and inland earthquakes, Report of Tono Research Institute of Earthquake Science, Seq.

No. 30, 2012, pp. 79–113 (in Japanese with Japanese title). 2) Suzuki, Y.: Statistics, New Mathematics Course 11, Asakura Publishing, 1989, p. 68 (in Japanese

with Japanese title).

3) PubMed: http://ja.wikipedia.org/wiki/PubMed.

4) Yamashita, F.: The highest number of deaths are recorded infants and aged

generation: The 1933 Showa Sanriku earthquake and tsunami, and the 1993

Hokkaido Nansei-Oki earthquake, Historical Earthquakes, 20, 2005, pp. 165–167 (in

Japanese with Japanese title).

5) Shuto, N.: Human behavior at the 1933 Showa Sanriku earthquake and tsunami:

Tsunami engineering technical report, 27, 2010, pp. 19–41 (in Japanese with

Japanese title).

6) Miyano M. et al.: Effect of age difference: Comprehensive studies on 1993 Hokkaido

nansei-oki earthquake, Urban Disaster Management Lab., Urban Research Center,

Tokyo Metropolitan University, 1994, p. 49. (in Japanese with Japanese title)

7) Miyano, M. and Mochizuki, T.: Follow-up survey of the 1946 Nankai Earthquake:

Casualty and human behavior at tsunami affected area, 35, 1988, pp. 75–85 (in

Japanese with Japanese title).

8) Doocy et al.: Tsunami mortality in Aceh Province, Indonesia, Bull. World Health Org.

85, 2007, pp. 273–278.

9) Matsuyama, M.: Measurement result by field survey team for Indonesia Ocean

Earthquake and Tsunami,

https://web.archive.org/web/20070221185721/http://www.eri.u-tokyo.ac.jp/namegaya/s

umatera/surveylog/survey_results.htm, accessed 6/Mar./2016 (in Japanese with

Japanese title).

10) Takahashi, M., Tanaka, S., and Kimata, F.: Damage and reconstruction of the 2004

Sumatra earthquake and tsunami, Kokon-Shoin, 2014, pp. 1–104 (in Japanese with

Japanese title).

11) Kuroki, T.: Health, aging, life: Culture magazine of human and life,

Chuokoron-Shinsha, 2007, p. 35 (in Japanese with Japanese title).

12) Nishikiori, N. et al.: Who died as a result of the tsunami? Risk factors of mortality

among internally displaced persons in Sri Lanka: a retrospective cohort analysis,

BMC Public Health 2006, 6:73.

- 50 -

13) Editorial Committee for the Report on the Hanshin-Awaji Earthquake Disaster:

Report on the Hanshin-Awaji Earthquake Disaster, omnibus, Maruzen Publishing,

2000, p. 63 (in Japanese).

14) Miyano, M. and Sumiyoshi, Y.: Study on Sex Difference in Casualties Due to the

Hyogo-ken Nanbu Earthquake, Report of Tono Research Institute of Earthquake

Science, Seq. No. 2. 1999, pp. 46–49 (in Japanese).

15) Koyama, M. et al.: Municipality‘s mortality rate according to inundation level and age

classes on the 2011 Great East Japan Earthquake, Journal of Japan Society of Civil

Engineers, Ser. A1 [Structural Engineering & Earthquake Engineering (SE/EE)], Vol.

69, No. 4, 2013, pp. I_161–I_170 (in Japanese).

16) Sato, T.: Safety Education and risk management, learn from the 2011 Great East

Japan Earthquake, keynote address, 12th Conference of the Japanese Association of

Safety Education in Joetsu University, Joetsu University of Education, 2011, pp.

10–13 (in Japanese with Japanese title).

17) Center for Disease Control: Mortality of the 1999 Earthquake in Nantou and

Taichung Counties, Epidemiology Bulletin, 16, No. 1, 2000, pp. 1–17.

18) Pai, Chih-Hung: A study on near-fault mortality from the 1999 Taiwan Chi-Chi

Earthquake, Paper No. 2542, 13th WCEE, 2004, Vancouver, Canada.

19) Armenian, H. K, et al.: Death and injuries due to the earthquake in Armenia, A cohort

approach, International. Jour. Epidemiology, 26, pp. 806-813, 1997.

20) Tabuchi, R.: Family sociological and demographical factors of earthquake and

tsunami disasters, The 3rd Investigation Report of 2004 Northern Sumatra

Earthquake, 2007, pp. 45–48 (in Japanese with Japanese title).

21) Ohashi, H. and Ohta, Y.: Study on earthquake disasters in Turkey in aiming at

development of seismic hazard reduction policy: (2) Empirical derivation of seismic

intensity distribution laws, Journal of Structural and Construction Engineering, 350,

1985, pp. 19–25, (in Japanese).

22) Yamamoto, Y.: International Disaster Study Group (Edit): Chapter 7, International

Society and Disaster, Disaster Medicine, Nanzando, 2002, p. 248, (in Japanese with

Japanese title).

23) Sawai, M.: ESCAP special report: Who is vulnerable during tsunamis? Experiences

from the Great East Japan Earthquake 2011 and the Indian Ocean Tsunami 2004,

2012, pp. 1–18.

(Original Japanese Paper Published: March, 2015)

(English Version Submitted: March 22, 2016)

(English Version Accepted: April 17, 2016)

- 51 -