GROWTH AND SHRINKAGE PRE AND POST TSUNAMI IN FUKUSHIMA PREFECTURE, JAPAN

Abstract

Depopulation is a severe problem in many urban areas globally. Massive population migrations can occur due to relocation after natural disasters and significantly change the demographic composition of regions and cities. The 2011 Great Tsunami in Japan resulted in a combined total of deaths and missing persons of more than 24,500. Post-tsunami recovery efforts resulted in widespread population relocation of high-risk communities into lower-risk areas. Using the Fukushima Prefecture in Japan as the study area, a region characterized by several depopulating cities both pre and post-tsunami, this research examines how the population relocation efforts have either exacerbated or assisted in lessening the effects of urban shrinkage and decline after the earthquake and tsunami of 2011. The results show that 30 municipalities have seen population and economic growth since 2011, and 12 municipalities are underdoing trends toward decline within Fukushima. Negatively affected cities tend to have larger populations than positively affected cities. Most of the small towns and villages closer to the inundation area are fall into the category of negatively affected areas. Moreover, the population increases in many post-disaster cities are primarily due to significant increases in elderly populations with minimal young persons that will inevitably decline in the next decade. By determining the effects of their relocation efforts, the government can better develop targeted strategies that good for the prosperity and development of the Fukushima Prefecture.

2. INTRODUCTION

Excessive urban depopulation, sometimes referred to as shrinkage, is a severe problem in many urban areas globally, impacting development and having large-scale socio-economic effects (Gu et al., 2019; Newman et al., 2019). Relatedly, massive population migrations can occur due to relocation after natural disasters such as earthquakes and tsunami and can significantly change the demographic composition of regions and cities. Large scale demographic shifts due to natural disasters in regions already experiencing depopulation can either help protect cities from the problems associated with depopulation (or reverse the condition) or further amplify the condition. The 2011 Great Tsunami in Japan resulted in a combined total of confirmed deaths and missing persons of more than 24,578 (15,893 deaths, 2,533 missings, and 6,152 injured) (Matanle & Rausch, 2011). Post-tsunami recovery efforts resulted in widespread population relocation of high-risk communities into lower-risk areas (Gauntt et al., 2012). The massive population migrations resulting from the disasters significantly changed the demographic composition of the Fukushima region. Using the Fukushima Prefecture in Japan, a region characterized by several depopulating cities both pre and post tsunami, as the study area, this research examines how the population relocation efforts have either exacerbated or assisted in lessening the effects of urban shrinkage after the devastating earthquake and tsunami of 2011. The purpose of this study is to examine and compare the characteristics of growing (populating) and shrinking (depopulating) cities in terms of depopulation after the devastating Japanese disaster of 2011 and to offer future recovery and development plans and design suggestions. It is important to note that the Fukushima Prefecture is shaped by not only tsunami damage and post-tsunami relocation efforts, but also by post-nuclear-meltdown evacuation and relocation. To examine this, t-tests are used to evaluate significant changes in total population, youth population rate, labor force rate, elder population rate, and unemployment rate. Based on these findings, we then develop a ranking system to identify the municipalities in Fukushima are which have continued, began, or reversed trends toward urban decline, comparing pre and post tsunami spatial conditions. A population migration pattern and decline analysis is conducted to indicate the key characteristics of the residents affected and the migrants moving in and out the region. Such studies are essential for future urban design to understand the dynamics of local residents before and after the effects of such disaster events. This study provides a blueprint for designing for urban areas in Fukushima in regards to users, housing needs, economic conditions, development potential, and other attributes contributing to design programming and urban form needs.

3. LITERATURE REVIEW

3.1. Depopulation in Japan

Generally, a shrinking city can be referred to as a metropolitan area, including a city, part of a city, or a town, that has experienced population loss, economic downturn, employment decline, and social problems as symptoms of a structural crisis (Martinez-Fernandez et al., 2012). Shrinkage has become a significant urban design issue globally, impacting development, and having large-scale negative effects on communities. Various factors contributing to shrinkage include deindustrialization, suburbanization, war, natural or human-induced disasters, an aging or low-fertility rate population, and globalization (Hollander et al., 2009). Population aging and population decline draw increasing attention internationally. As of 2050, the estimated global population aged 60 years or over is expected to increase to reach 21.1%, an increase of 11.9% over 1990. The word’s elderly population living in developing countries will increase from 67% to nearly 80% by 2050 (Wang & Fukuda, 2019). Japan is one of the most aging countries in the world. For the next fifty years, Japanese children and working-age populations are projected to decline by nearly 50%, based on current projections (Matanle, 2014). The country is also becoming an aging society demographically due to people living longer lives combined with a lower birth rate (Matanle & Rausch, 2011). The people aged 65+ years composed about 23% of the total population of Japan in 2009, the highest elderly population globally (Bachev & Ito, 2017). Based on current projections, one-third of the total Japanese population will be 65+ years, and one-fifth will be 75+ years by 2030 (Muramatsu & Akiyama, 2011). In contrast, the birth rate of Japan is the lowest worldwide and reached a postwar level of 1.26 in 2005 (Matanle & Rausch, 2011). These circumstances affect all areas of Japan’s healthcare, educational resources, infrastructure, and environment. The high economic costs required to maintain a hyper-aged society stand at around a quarter of the national budget. Japan’s population changes are primarily due to the cuntry’s low fertility rate (Matanle & Rausch, 2011). Changes in the proportion of the population currently married as well as decreases in the amount of children among married couples are the two main factors contributing to this fertility decline (Atoh, 2001). As a result, the proportion of the aged sector of the population (over 65 years old) projects to be 40% by 2060 (Martinez-Fernandez et al., 2016).

Population dynamics associated with globalization can attract population and skills from unprosperous areas to the places with high levels of innovation and intellectual engagement. Under new globalization processes, many cities, towns, and villages are continuously losing capital and human resources, leading to significant social and economic challenges (Martinez-Fernandez et al., 2012). Spatially, depopulation effects many rural areas of Japan, mostly due to internal population migration. In general, younger adults are more likely to move from rural to urban areas for various opportunities in terms of education, employment, and social life; therefore, a large amount of older people have been left in neighbor-less communities (Matanle, 2014). As young populations migrate to prosperous areas, vacant lots, underused infrastructure, and closed commercial venues become common scenes, aggravating the isolation of elderly residents (Buhnik, 2010). Japan’s Ministry of Internal Affairs says that around 15,000 of Japan’s 65,000 or so communities have more than half of their population over the age of 65 (Semuels, 2017).

3.2. Disaster events and depopulation in Japan

Over the past decade, multiple large-scale natural disasters have significantly exacerbated urban shrinkage worldwide. Globally, natural disasters were responsible for, on average, 60,000 deaths per year. Earthquakes have the highest lethality among droughts, floods, and other events (Bates, 2019). Because of its unique geographic location, Japan has an intimate relationship with earthquakes and tsunamis. Disaster-induced problems, such as mortality, health, fertility, and migration, greatly contribute to Japan’s demographic size and composition (Frankenberg et al., 2014). During the evacuation process immediately following a disaster, scores of patients and elderly residents can die due to interruptions in essential life survival services such as medical care, food, and water (Hasegawa et al., 2016). People stricken by poverty and of low socio-economic status are at greater risk than others during disasters, in both the response and recovery phases. In general, people of low socio-economic status tend to live in fragile housing in disaster vulnerable areas, when compared with more affluent persons (Masterson et al, 2019; Kim and Newman, 2019). They also have lower likelihoods of receiving warnings of disasters, being able to evacuate promptly, or accessing post-disaster aid (Newman et al., 2018; Meyer et al., 2018; Reja et al., 2017). In addition to direct death, the long-term impacts stemming from a disaster can harm communities for years, as fertility trends can be altered (Bates, 2019). Birth rates may be reduced in the months following a disaster because of miscarriages, declines in coital frequency, and lower relationship quality; s imultaneously, many families that lost their children may intentionally try to conceive (Nobles et al. , 2015). This replacement of fertility changes local area population compositions.

Population migration is a natural way to deal with disaster shocks when property or sources of livelihood are destroyed. Population displacement – either voluntary or involuntary – also occurs during a large-scale disaster event. The level of voluntary displacement is linked with the risk of exposure to shock, poverty level, and the diversification of assets and income. Depopulated areas, small communities, and communities with higher vacancy rates are more likely to experience large population losses after disasters. In contrast, wealthier communities tend to retain their population (Cross, 2014). After a disaster, socioeconomic, political, social, and personal emotional factors all affect the return decision process (Morrice, 2013). Some displaced populations may return home while others may permanently move to other less affected places. Therefore, disasters do not only impact affected populations but also the communities from which the migrants have or are moving (Frankenberg et al., 2014).

3.3. A paradigm shift in Japanese urban planning

To address the shrinkage issue, there was a paradigm shift away from the modern expansion model to a shrinking model in urban design and development in Japan (Stefan et al., 2017). The contemporary shrinking model, known as a smart decline strategy, aims to ensure the quality life of remaining residents, emphasizes reducing the overall financial burden by spatial consolidation, and reassures future development possibilities in a much longer term (Heins, 2012). The challenge with the smart decline model is about how to efficently plan land uses that shape a concentrated urban structure while allowing for urban decline, but managing it (Stefan et al., 2017). However, acknowledging that some areas cannot be developed further can be difficult, especially after a large-scale disaster. The destroyed land, housing, and infrastructure in a disaster impacted area can sometimes promote economic development and increase population immigration into other areas (Albrechy, 2017). Relocation and recovery plans can bring related professionals, industrial workers, and their families from unaffected areas to affected areas. However, the influx of population changes after a disaster is a complicated issue and does not always follow a clear pattern. Demographic and social-economic changes occur over the long term as the impact of the disaster fades away. Usually, the reconstruction efforts tend to be expansion-oriented as the needs of reconstruction increases after a large-scale disaster. However, in an era of a sparse population like Japan, reconstruction plans need to be considered in the face of population shrinkage (Asano et al., 2018).

3.4. The 2011 Japanese tsunami

On March 11, 2011, a devastating tsunami following a magnitude 9.0 earthquake produced catastrophic damage to many coastal cities in northeast Japan. This tsunami caused widespread damage to buildings, roads, communications, and regional electrical power (Gauntt et al., 2012), and resulted in a combined total of more than 24,578 confirmed deaths and missing persons (nearly 15,893 deaths, 2,533 missing, and 6,152 injured) (See Table 1). Miyagi, Iwate, and Fukushima prefectures had the most significant number of victims as a powerful tsunami wiped out entire communities (Bachev & Ito, 2017). Another primary consequence of the tsunami was a meltdown of three reactors at the Fukushima Daiichi nuclear power plant, resulting in aftermath that was given a rating of 7 (the highest score) on the International Atomic Energy Agency (IAEA) nuclear event scale. It is still impossible to gauge the widespread health and environmental effects due to the radioactive leakage from the Fukushima nuclear accident (Gauntt et al., 2012).

Table 1.

Number of confirmed deaths, missing, and injured person associated with the March 2011 earthquake (March 10, 2017)

Prefectures	Deaths	Missing	Injured	Prefectures	Deaths	Missing	Injured
Hokkaido	1	-	3	Gunma	1	-	42
Aomori	3	1	112	Saitama	-	-	45
Iwate	4,673	1,123	213	Chiba	21	2	258
Miyagi	9,540	1,230	4,145	Kanagawa	4	-	138
Akita	-	-	11	Nigata	-	-	3
Yamagata	2	-	29	Yamanashi	-	-	2
Fukushima	1,613	197	183	Nagano	-	-	1
Tokyo	7	-	117	Shizuoka	-	-	3
Ibaraki	24	1	712	Mie	-	-	1
Tochigi	4	-	133	Kochi	-	-	1
Total					15,893	2,553	6,152

^*Source: National Police Agency

Japan’s triple disaster of 2011 - earthquake, tsunami, and Fukushima Daiichi nuclear accident – largely impacted population movements, employment situations, and the gross domestic products (GDP) of many cities. Compared to 2010, the total net migration of disaster-stricken Tohoku prefectures of 2011 increased by 30,799 people. Because of the nuclear climate, the net migration of Fukushima was much larger than pre-disaster, which occupied around 80% of the total net immigration from the three disaster-stricken Tohoku prefectures. This resulted in a significant change in the employment structure, post-disaster. The labor force openings in the construction industry and public service sectors largely increased in many disaster-stricken areas in the Tohoku Prefecture, due largely to the reconstruction after the disaster. However, job hunters seeking local industrial jobs were less fortunate in the tsunami-hit region (Takabe & Inui, 2013). The Japanese economy was significantly affected by the disaster; the country’s total GDP reduced between 0.2 to 0.5 percentage points, however, its overall economic growth still increased by nearly 1% (Nanto et al., 2011).

3.5. Relocation in Japan

There were more than 76,000 residents within a 20 km area surrounding the Fukushima Daiichi Nuclear Power Plant (FDNPP) prior to the nuclear incident (Nanto et al., 2011). Immediately after the accident, the Fukushima government ordered 1,864 residents in only 2km radius around the FDNPP to evacuate at 19:03 on Mar.11, 2011. By the evening of the same day, as the evacuation area expanded to 3 km, the affected residents increased to 5,800, including residents under instructions to take shelter within 10km of FDNPP. After the 1st explosion at nuclear reactor No.1, the government forced over 50,000 residents who lived within 20km from the FDNPP to evacuate on March 12, 2011 (Gemenne et al., 2012). By March 13, 170,000 people were obligated to evacuate due to the possibility of a meltdown. On March 15, more than 97% of residents living in the 20 km radius had evacuated (Hasegawa et al., 2016).

The government further instructed the residents living between 20km-30km radius from the Nuclear Power plant to stay inside their homes. On April 22, even residents within a 30km zone were recommended to relocate. The government designated the “restricted area” within a 20km radius around the FDNPP (See Image 1), prohibiting entry into the area, excluding those engaged in emergency response (Fukushima Beacon for Global Citizens Network, 2017). In May, the government designated all areas having an air radiation dose of 20 mSv/year as “deliberate evacuation areas.” Residents from these areas were required to evacuate by the government. Until September 2011, more than 100,000 residents in Fukushima were affected under several evacuation orders, a large degree having to relocate away from their homes (Gemenne et al., 2012).

Figure 1: Restricted area post-tsunami.

*Source: Fukushima On the Globe

By March 2012, the areas to which evacuation orders had been issued were rearranged by the government into three areas according to the annual cumulative nuclear dosage:

1) The evacuation order cancellation preparation zone was designated areas where the evacuation orders were ready to be lifted in and the annual integral dose of radiation was below 20mSv. People were allowed to enter and pass through the areas temporarily, but the overnight stay was prohibited.

2) The restricted residence zone designated areas where people were not recommended to enters the annual radiation dosage was 20 mSv or more, but were allowed to enter during the daytime.

3) The difficult to return zone defined areas where people were unable to stay as the annual integral dose of radiation was expected to be 20 mSv or more within five years and the current integral dose of radiation per year was 50 mSv or more. (Fukushima On the Globe, n.d.).

The government gradually lifted the evacuation orders from some areas, including Kawamata Town, Namie Town, Litate Village, and Tomioka Town. The evacuation designated zones occupy about 2.7% of the entire area of Fukushima Prefecture (FukushimaVoice, 2013) (See Image 2).

Figure 2: Evacuation zones during the 2011 Japanese tsunami.

Source: Fukushima On the Globe

3.6. Population redistribution post-tsunami in Japan

After the tsunami, municipal buildings, schools, and gymnasiums became shelters for evacuees who were eventually moved to transitional shelters, funded by the central government. The sites of transitional shelters were selected by the municipality. After, evacuees could either find public housing to rent or choose to build their own houses in permitted areas for permanent living (Gemenne et al., 2012). Among the evacuees, many who feared the radioactive leakage evacuated voluntarily, especially those in zones experiencing a severe lack of resources.

In Fukushima Prefecture, the number of displaced residents increased from 86,308 in March 2011 to 99,205 in June 2011. Moreover, voluntary migrants who moved out of Fukushima increased by 62,831 after the first year of the disaster, especially among families with children and younger populations (Hasegawa et al., 2016). Due to an exodus of younger persons, issues associated with aging populations in the affected areas were accelerated. In fact, in 2011,10% (210,000) of Fukushima’s population were relocated to other places within the Prefecture after the disaster, including 77,000 in the restricted area, 10,000 in the deliberate area, 26,000 in the evacuation prepared area, and 117,000 in other areas within 30km (FukushimaVoice, 2013).

4. METHODS

4.1. Objectives and Data

This research examines the urban socio-economic recovery of cities and villages in Fukushima prefecture, a major victim of the 2011 Tōhoku earthquake and tsunami, comparing conditions before and after the disaster. The data was obtained from the Japan Population Census Data, which is conducted every 5 years since 1920. The unit of analysis is the household, formed by a group of individuals that share livelihood in the same residence. A family living in one house is a typical example. In addition, a member who is not a relative can also be counted as a member of a Household Unit when they share the same residence and livelihood. However, a person who lives independently can also be considered a Household Unit, even if he or she shares the same residence with others. Besides, to take economic characteristics into consideration, Gross Domestic Product (GDP) data, covering the annual GDP from 2003 to 2015 of each city and village provided by SNA (National Accounts of Japan), is utilized to evaluate the economic growth after the disaster.

4.2. Variables

The census data released by the National Statistics Center, Japan, are all in different scales and formats, and do not align to typical U.S census or parcel based analytic approaches. In this study, a 500m side mesh area tabulation is investigated to capture the most accurate population change in Fukushima Prefecture. The mesh area tabulation, known as “Regional Mesh Grid Areas”, is based on the Japan Standard Rectangular Grid Square System. Three types of mesh areas are covered in this data set: a) 500m Side Mesh Areas; b) 1 km Side Mesh Areas; and c) 10 km Side Mesh Areas, each is aggregated into territories. These mesh areas’ boundaries are not modified over time, so they are suitable to compare the census data results in different years.

For comparison of pre and post-conditions of the disaster of the 2011 Tōhoku earthquake and tsunami, census data from 2005, 2010, and 2015 are used to show the reduction or increase of socio-economic development speed before and after the disaster. This study is divided into 2 steps, focusing on the individual socio-economic characteristics, and a comprehensive ranking of the recovery of socio-economic aspects of cities and villages in Fukushima prefecture by using a t-test on before and after socio-economic characteristics and the changing rates of each mesh in each city or village.

During the analysis, urban decline issues, population loss, economic downturns, employment declines, and social problems are assumed to be symptoms of a structural crisis (Martinez-Fernandez et al., 2012). Population, age structure, GDP, labor force, and unemployment rate are the primary socio-economic attributes to be tested in this study. In this study, several socio-economic characteristics are interpreted for comparison. Besides the total population, as the aging problem is a vital part of population study, median age, elder population (population of age above 65), and minor population (population of age under 17) are taken into consideration. The unemployment rate and labor force are also utilized as essential indicators of socio-economic recovery. In addition, the GDP of each city and village is another indicator of the economic growth. To examine the patterns of recovery within the Fukushima prefecture, mesh data are dissolved and merged into city scaled data. The socio-economic characteristics prior to the Tsunami disaster are calculated with 2005 and 2010 census data. In order to normalize the size of cities and villages, changing rates are calculated to present their socio-economic changes before and after the disaster.

4.3. Ranking

To test the mathematic significance of the change of socio-economic characteristics before and after the tsunami, this study utilizes a Welch’s t-test. Each mesh in each city is considered as an observation. Selected socio-economic characteristics, including population, median age, elderly population, minor population, labor force, and unemployment rates, are then tested using this method. Using “x1” as the change rate before the tsunami and “x2” as the change rate after the tsunami, the null hypothesis assumes that there are no changes before and after the tsunami (“x1” – “x2” =0). If the p value of the test is smaller than 0.05, indicating rejecting the null hypothesis, we suggest that the socio-economic characteristics changing rate after the disaster are mathematically significantly different than it was before the tsunami. In this case, the result of “x1” – “x2” will show whether it is growing faster (“x1” – “x2” <0), or declining faster (“x1” – “x2” >0).

The ranking of the cities and villages in Fukushima prefecture is based on the following attributes: t- test results of population, elderly population, minor population, labor force, unemployment. Shapiro tests are also conducted to determine if the samples are normally distributed prior to the t-tests. If t-test results of population, minor population, and labor force are significant at 0.05 level under “x1” – “x2” <0 condition, indicating it increases after the disaster, ranking counts will increase by one. If it is significant under “x1” – “x2” >0 condition, the ranking counts will reduce by one. Otherwise, the ranking counts stay the same because there is no significant difference between pre and post-situation. To ensure the consistency of the result, we didn’t count GDP and median age into the ranking system because they are not the mash level data. However, we integrated GDP and median age data into later analyses to imply which cities or villages have comparatively recovered better and which ones are comparatively worse.

5. RESULTS

5.1. Ranking results

In the ranking results, there are 12 municipalities’ socio-economic attributes that appear to be worse than they were before the tsunami. There are 4 cities, 7 towns, and 1 village among them (see table 2). The one village that was negatively affected is Kawauchi Village, which has a minus 3 ranking counts. There are 6 partially negatively affected areas t hat have minus 2 ranking counts. Meanwhile, there are 5 minor negatively affected areas of minus 1 ranking counts. On the other hand, there are 30 municipalities with positive ranking counts, include 4 cities, 15 towns, and 11 villages (see table 3). The one positively affected area, which has a positive 3 ranking count, is Inawashiro Town. There are 15 partially positively affected areas with 2 positive ranking counts and 14 minor positively affected areas with 1 positive ranking counts. Figure 3 shows a clear pattern that negatively affected areas are more clustered around coastal flooded areas, positively affected areas are away from disaster areas, and neutral affected areas are scattered between them. Within the 20 km radius, towns and villages were hit the hardest compared to cities. Population compositions of Tamura city and Iwaki city were basically unaffected by the disaster, but people living in Kawauchi village, Naraha town, and Hirono town had difficulty returning to heir hometown after the disaster. There are few municipalities that do not agree with this pattern. Kaneyama town and Nishiazu town, the two of the most inland towns, are identified as partially negatively affected areas. Soma city and Shinchi town, located near the inundation areas are rated as partially positively affected areas.

Table 2.

Ranking results of negatively affected areas

City	Total Population	Youth population ( Age <14)	Labor Force	Elderly population (Age >65)	Unemployment	Ranking
Kawauchi mura	−1	−1	0	−1	0	−3 (Major negatively affected)
Hirono machi	−1	−1	0	−1	1	−2 (Partially negatively affected)
Kagamlishi machi	−1	0	−1	−1	1	−2 (Partially negatively affected)
Kaneyama machi	−1	−1	−1	0	1	−2 (Partially negatively affected)
Kawamata machi	−1	−1	0	−1	1	−2 (Partially negatively affected)
Naraha machi	−1	−1	0	0	0	−2 (Partially negatively affected)
Nishiaizu machi	−1	−1	0	−1	1	−2 (Partially negatively affected)
Date shi	−1	−1	1	−1	1	−1 (Minor negatively affected)
Fukushima shi	−1	−1	1	−1	1	−1 (Minor negatively affected)
Hanawa machi	−1	−1	1	−1	1	−1 (Minor negatively affected)
Koriyama shi	−1	0	0	−1	1	−1 (Minor negatively affected)
Minamisoma shi	−1	−1	1	−1	1	−1 (Minor negatively affected)

Table 3.

Ranking results of positively affected areas

City	Total Population	Youth population ( Age <14)	Labor Force	Elderly population (Age >65)	Unemployment	Ranking
Inawashiro machi	0	1	1	0	1	3 (Major positively affected)
Asakawa machi	0	1	1	−1	1	2 (Partially positively affected)
Hirata mura	0	1	1	−1	1	2 (Partially positively affected)
Izumizaki mura	0	0	1	0	1	2 (Partially positively affected)
Minamiaizu machi	0	0	1	0	1	2 (Partially positively affected)
Mishima machi	0	0	1	0	1	2 (Partially positively affected)
Ono machi	0	0	1	0	1	2 (Partially positively affected)
Samegawa mura	0	1	1	−1	1	2 (Partially positively affected)
Shimogo machi	0	0	1	0	1	2 (Partially positively affected)
Shinchi machi	1	0	0	0	1	2 (Partially positively affected)
Shirakawa shi	0	0	1	0	1	2 (Partially positively affected)
Soma shi	1	0	1	−1	1	2 (Partially positively affected)
Tadami machi	0	0	1	0	1	2 (Partially positively affected)
Yabuki machi	0	0	1	0	1	2 (Partially positively affected)
Yamatsuri machi	0	0	1	0	1	2 (Partially positively affected)
Yugawa mura	0	0	1	0	1	2 (Partially positively affected)
Aizumisato machi	0	0	1	−1	1	1 (Minor positively affected)
Aizuwakamatsu shi	0	0	1	−1	1	1 (Minor positively affected)
Bandai machi	0	0	1	0	0	1 (Minor positively affected)
Hinoemata mura	0	1	0	0	0	1 (Minor positively affected)
Kitashiobara mura	1	−1	0	0	1	1 (Minor positively affected)
Kunimi machi	−1	0	1	0	1	1 (Minor positively affected)
Miharu machi	0	0	1	−1	1	1 (Minor positively affected)
Motomiya shi	0	0	1	−1	1	1 (Minor positively affected)
Nakajima mura	0	0	1	0	0	1 (Minor positively affected)
Nishigo mura	0	0	0	0	1	1 (Minor positively affected)
Otama mura	0	0	1	−1	1	1 (Minor positively affected)
Tamakawa mura	0	0	1	−1	1	1 (Minor positively affected)
Tanagura machi	0	0	1	−1	1	1 (Minor positively affected)
Tenei mura	−1	0	1	0	1	1 (Minor positively affected)

Figure 3.

Distribution of positively and negatively affected areas

5.2. Socio-economic attributes

The specific socio-economic conditions are shown in Table 4. Population wise, compared with the previous rate before the tsunami, 10% of the positively affected areas’ population increased, while 100% of the negatively affected areas’ population decreased. In the youth population ratio analysis, 80% of the positively affected areas did not have significant change from the previous condition, but 40% of them were positively skewed. On the contrary, 83% of the negatively affected areas lose their youth population dramatically. When it comes to the elder population ratio, 63% of the positively affected areas have the same ratio changing rate with the former condition, and 83% of the negatively affected areas have in creased ratio changing rates comparing to the former condition. Meanwhile, 87% of the positively affected areas have increased labor force ratio changing rate, while only 33% of the negatively affected areas have increased rates. When it comes to the labor force ratio, the percentage of the positively affected areas with significantly increased labor force rates is 37% higher than of the negatively affected areas. Albeit there is a gap between positively affected areas and negatively affected areas in terms of labor force rate, 50% of negatively affected areas are positively skewed. Interestingly, almost all the areas, whether positively or negatively affected, have a decreased unemployment population ratio, decreased average age, and increased GDP. In a nutshell, after the tsunami, most positively affected areas have relatively stable total population, youth population ratio, and elder population ratio, increased labor force ratio and GDP, and decreased average age. For negatively affected areas, most of them have decreased total population, youth population ratio, and unemployment population ratio, and increased elder population ratio and GDP.

Table 4.

Distribution of different conditions among positively affected and Negatively affected areas

	Total population		Youth population ratio		Elderly population ratio		Labor force population ratio		Unemployment population ratio		Average age		GDP
	PAA (30)	NAA (12)	PAA (30)	NAA (12)	PAA (30)	NAA (12)	PAA (30)	NAA (12)	PAA (30)	NAA (12)	PAA (30)	NAA (12)	PAA (30)	NAA (12)
Decreased	7%	100%	3%	83%	0%	0%	0%	17%	90%	83%	97%	75%	17%	8%
Didn’t change	83%	0%	80%	17%	63%	17%	13%	50%	10%	17%	3%	0%	0%	0%
Increased	10%	0%	17%	0%	37%	83%	87%	33%	0%	0%	0%	25%	83%	92%
Positive skew	27%	25%	40%	8%	43%	0%	13%	50%	3%	0%	-	-	-	-
Normal	47%	75%	23%	84%	44%	83%	80%	50%	94%	83%	-	-	-	-
Negative skew	27%	0%	37%	8%	13%	17%	7%	0%	3%	17%	-	-	-	-

^*Positively Affected Areas (PAA), Negatively Affected Areas (NAA)

^*The underlined values indicate the most significant percentage under each category.

Additionally, investigation of the socio-economic conditions of the affected areas reveals much when based on the status of the municipality (see table 5). In Japan, three levels of the municipality - city, town, and village - are decided by the prefectural government based on the total population. Generally, a municipality with a population of above 5,000 is defined as a city; otherwise, it is regarded as a town or a village. Among city level samples, negatively affected cities’ labor force and GDP changing trend appear to be more positive than positively affected areas, including outliers. Among town level samples, negatively affected areas elderly population, average age, and labor force changing trend appear more positive than positively affected areas, including outliers. Among village level samples, negatively affected areas elder population, labor force, and GDP changing trend are better than positively affected areas, including outliers.

Table 5.

Socio-economic attributes changing rates comparison

Mean	Pop0510	Pop1015	Trend	You0510	You1015	Trend	Old0510	Old1015	Trend	Age0510	Age1015	Trend
NAA (city level)	−2%	−6%	−4%	−7%	−20%	−13%	12%	18%	7%	7%	6%	−1%
PAA	−2%	−2%	0%	−7%	−9%	−2%	8%	12%	5%	6%	4%	−3%
Difference	0%	4%	4%	0%	11%	11%	−4%	−6%	−2%	−1%	−2%	−2%
NAA (town level)	−7%	−20%	−14%	−10%	−33%	−22%	6%	1%	−5%	8%	0%	−9%
PAA	−8%	−8%	0%	−11%	−8%	4%	9%	13%	3%	9%	3%	−6%
Difference	−1%	12%	14%	−1%	25%	26%	3%	11%	8%	1%	4%	2%
NAA (village level)	−10%	−29%	−19%	−13%	−60%	−47%	5%	7%	2%	21%	11%	−10%
PAA	−4%	−6%	−2%	−10%	−7%	3%	8%	15%	7%	17%	5%	−12%
Difference	5%	23%	17%	3%	53%	50%	3%	8%	5%	−4%	−6%	−2%

Mean	Lab0510	Lab1015	Trend	Unem0510	Unem1015	Trend	GDP0510	GDP1015	Trend
NAA (city level)	−5%	4%	9%	17%	−34%	−51%	−4%	24%	28%
PAA	−1%	−1%	1%	47%	−39%	−86%	0%	21%	20%
Difference	4%	−5%	−8%	30%	−5%	−35%	5%	−3%	−7%
NAA (town level)	−1%	14%	15%	24%	−50%	−74%	−14%	25%	39%
PAA	−4%	2%	5%	41%	−53%	−94%	−13%	29%	42%
Difference	−3%	−12%	−9%	17%	−3%	−20%	1%	4%	3%
NAA (village level)	−5%	26%	31%	36%	−35%	−71%	−13%	26%	39%
PAA	−3%	5%	9%	38%	N/A	N/A	−6%	21%	27%
Difference	2%	−21%	−22%	2%	N/A	N/A	7%	−5%	−12%

6. DISCUSSION AND CONCLUSIONS

The results show that 30 municipalities (4 cities, 15 towns, and 11 villages) have seen population and economic growth since 2011, and 12 municipalities (4 cities, 7 towns, and 1 village) are underdoing trends toward decline within the Fukushima Prefecture. To assis t in interpretation, we characterize each city according to their demographic composition and GDP, finding that most cities have increased their GDP, decreased unemployment rates, and decreased the average age of residents. Expanding municipalities tend to have increased in total population, less aged populations, and more labor force than shrinking cities. However, many now populating cities which were depopulating prior to the tsunami, are doing so primarily due to significant increases in elderly populations with minimal young persons, indicating a short-term increase in population that will inevitably decline in the next decade or so in many cities.

The results also imply that negatively affected cities tend to have larger populations than positively affected cities. Large cities like Fukushima, Date, and Koriyama are more vulnerable than relatively smaller cities like Soma, and Tamura post-disaster. This abnormal condition could be caused primarily by the location of these cities. Most of the negatively affected cities are closer to the inundated area and are more likely to be influenced by the tsunami, while the smaller cities with more space for new residents appear to be the beneficiaries of the relocation of the residents from the disaster area. Although the government gave particular financial support for the 2011 disaster survivors, it was difficult for many who came from small towns and villages to pay additional living expenses to live in large cities. Meanwhile, negatively affected towns tend to have smaller populations than positively affected towns, and negatively affected villages tend to have a smaller population than victory villages. For example, large towns like Inawashiro, Tadami, and Minamiaizu are doing much better than small towns. When it comes to smaller municipalities, they oftentimes lack the means, such as hospitals, rescue groups, related professionals, and mature autonomous structures to recover their populations sustainably (Matanle, 2014). The elderly population distribution in smaller towns and villages could be another factor that exaggerates this situation, considering the excess mortality of elderly persons during the disaster (Morita et al. 2017).

From results it appears that most of the municipalities closer to the inundation area are fall into the category of negatively affected cities, according to the ranking results, including Kawauchi Village, Naraha Town, Hirono Town, and Kawamata Town. Nuclear radiation concerns and governmental relocation plans could be reasons for the loss of population and economic decline. Kawauchi Village, Naraha Town, and Hirono Town were formerly in the restricted zone in the evacuation plan (Fukushima On The Globe, n.d.). There are also 2 partially negatively affected cities that are far away from the inundation area, Nishiazu Town and Kaneyama Town. Both have only negative changes in the total population and youth population. The reason for this could be the national aging issue and the fertility problem in Japan (Matanle & Rausch, 2011). Among the municipalities, there are 2 positively affected cities close to the inundation area, Shinchi Town, and Soma City. Both have significantly increased in population and significantly dropped in the unemployment rate. The recovery project recruitment process could largely contribute to this phenomenon. Since the Japan Census Data include migrates as survey units, rescue teams, research groups, construction workers, and their families for the recovery project in the disaster area could be anchored in these closest and safe locations, increasing the local population and labor force. The detailed distribution of such a migrate population requires further research.

Post-disaster populating and depopulating cities require planners to adopt specific recovery strategies tailored to their characteristics carefully. The impact of reconstruction efforts on city structure sometimes surmounts even the disaster itself. The reconstruction efforts tend to be expansion-oriented as the needs of reconstruction increases after a large-scale disaster. In an era of the sparse population increases like Japan, reconstruction plans need to be considered in the face of population shrinkage (Asano et al., 2018). Below, we propose several suggestions to assist planners in developing proper recovery strategies for both populating and depopulating cities after the 2011 triple disaster.

For positively affected areas, most of which are small cities, large towns, and large villages, population changes need to be carefully monitored and predicted to avoid being fooled by illusions of population growth. Indeed, the destroyed land, housing, and infrastructure during a disaster impacted areas can sometimes increase the value of comparable land, housing, and infrastructure in surrounding areas, thereby potentially promoting the economic development of other areas (Albrechy, 2017). Also, the relocation plans and recovery plans have brought some of the population, such as related professionals, industrial workers, and their families, from other parts of Japan and even overseas to Fukushima prefecture. As the impact of the disaster fades away, such situation will gradually get back on original track. Although some evacuator who was living the temporary housing have started new lives in new cities, many will eventually return to their original communities due to emotional bonding and lack of confidence in living in a strange environment (Murakami et al., 2014). Moreover, the population increases in many post-disaster cities are primarily due to significant increases in elderly populations with minimal young persons. This short-term increase in population that will inevitably decline in the next decade. In the past, most housing developments for survivors have not been included as part of city urban designs or planning efforts, leading an inappropriate site selection for these houses. Unfortunately, the target group of housing development for survivors was rarely open to current residents, resulting in some of lots remaining unsold, especially for those with poor locational conditions. Additionally, since land acquisition is quicker and cheaper in the outer area of the city, it is easy to run counter to the compact development guided by infill-style development. Also, as a result of the demographic aging of the relocation, some new developments may not be used (or reused) on a permanent basis. Green infrastructure, such as community gardens, bioswales, retention ponds, constructed wetlands, or pocket parks, could be a way to achieve social and ecological services simultaneously. Therefore, for cities with a temporality increased population, sustainable urban designs and plans that can satisfy the rational demand of survivors and existing residents without overestimation and overplanning should become the goals for designers (Asano et al., 2018). Governments and designers must fully understand the long-term implications to avoid excessive investment, which would impose a greater financial burden on future generations.

In the case of the tsunami disaster in Fukushima prefecture, population loss, especially in the small towns and villages close to the inundation area, is severe. The reconstruction strategies for these negatively affected towns and villages based on agriculture or fishery, such as Kawauchi village, Naraha town, and Hirono town, should be different from other positively affected regions. We need to keep in mind that it is difficult to restore or revive the population in these places, because many residents have relocated to higher lands, and there was a severe depopulation issue before the disaster. The reconstruction process of these areas should aim to return disaster-prone areas to their original ecosystems. Looking to the past, rural and ecological approaches are more likely to be overwhelmed by civil engineering, such as building concrete seawalls and raising lower-lying land. Emphasizing only economic services while neglecting ecosystem services can lead to increased social vulnerability. For example, sea walls can help prevent inundation by tsunamis but also can destroy fishery industries by cutting off the connection to the sea. However, some ecological approaches, such the multiple lines of defense approach and shore-parallel structural risk reduction strategies, have been approved to affect inundation reduction positively. Integrating these green infrastructure elements may help open up some of the coastal areas for fishing activities of remaining residents. Also, the vacant low-lying lands that cannot be used as residential areas could accommodate farmland, parks, shops, and factories to meet the needs of economic services, social services, and ecological services (Murakami et al., 2014; Morris-Suzuki, 2015).

Monitoring and predicting demographic composition and social-economic changes under close scrutiny is essential to avoid excessive investments when developing urban design plans for disaster effected shrinking cities, as it could impose a greater financial burden on future generations. In terms of urban planning and political decision making, switching from top-down mechanisms to a bottom-up model that involves the participation of residents is extremely important to build health cities under urban shrinkage. No matter if the areas are positively or negatively affected after the tsunami, envisioning a shrinkage-based model of urban design instead of an expansion model during disaster recovery processes is critical, given the background of the population decline situation in Japan.

7. LIMITATIONS AND FUTURE WORKS

Since the Japan Census Data are surveyed and collected every 5 years, we can only interpret the socio-economic data in 2005, 2010, and 2015. The tsunami occurred in 2011, so the census data may not precisely catch the changes before and after the disastrous event. The demographic data right before the disaster happened (2010 to 2011) is hard to incorporate into the analysis. Because of the 5-year interval of the census data, we have to include 2010, the year before the 2011 disaster, to investigate the demographic changes after the disaster. Besides, since the variables of socio-economic attributes from the census data are limited, there are some potential overlaps in out investigated variables, like youth population, elder population, and labor force ratio, which to some degree weaken the argument on ranking criteria. Regardless of these limitations, Japan Census Data are the most comprehensive and irreplaceable data to shed light on the demographic composition of the country. In addition, some geographical data are absent in this study, which constrains the investigation of the reasons for the uneven distribution of population change. When it comes to the relocation process, the amounts of immigrants from each municipality are unknown, which makes it difficult to rationalize the sharper population loss in big cities. Additional geographical analysis should be included in the future to help determine the difficulty of the relocation of the victim population from the inundation area.