A study on the use of cyclone shelters in Bangladesh

Abstract

Bangladesh's geography renders it vulnerable to cyclones. The most severe cyclone occurred in 1970, resulting in severe damages and over 300 000 deaths. The government has implemented cyclone disaster mitigation primarily by providing cyclone shelters. Recently, BRAC University proposed a new type of cyclone shelter, the house‐shaped “mini‐shelter.” This study aims to contribute to the government's cyclone mitigation schemes, especially regarding the appropriate utilization and construction of cyclone shelters that can be used throughout cyclone‐prone areas. Investigations in two cyclone‐affected areas reveals that cyclone shelters, which are intended for a few hours of occupation during inundation, function as temporary living quarters during post‐cyclone reconstruction. Moreover, most cyclone shelters are elementary schools, and their storage and staff rooms may be locked during cyclones owing to poor management, which complicates their use as evacuation centers or temporary post‐cyclone living quarters. The “mini‐shelters” can be used as both evacuation centers and livestock storage during cyclones. Although locals doubt the buildings' safety because the wooden second floor is too low for protecting against storm surges, “mini‐shelters” can be useful in remote coastal areas with limited space for cyclone shelters. Hence, a combination of these cyclone shelter types affords a holistic solution to cyclone disaster mitigation.

Introduction Research background

Bangladesh is a cyclone-prone country owing to its social and geographical conditions. Approximately 50% of the land is 7 m or less above sea level and faces the Bay of Bengal. During the cyclone disaster in 1970, more than 300 000 people died.[Note 1] In recent years, coastal areas suffered serious damages caused by two devastating cyclones, Sidr in 2007 and Aila in 2009. The Bangladesh government and international cooperation agencies have attempted to mitigate cyclone disasters, primarily through implementing early warning systems and constructing cyclone shelters since achieving independence in 1971. Although the death toll has decreased significantly with the increased construction of cyclone shelters, the following are still ongoing issues in cyclone countermeasures: the inability to use cyclone shelters as evacuation centers in emergencies owing to inadequate maintenance,¹ and factors that hinder peoples' willingness to evacuate owing to gender issues, such as “women cannot go to places where men gather” as “women should protect their homes” and “cannot run away with one's belongings.”² Furthermore, cyclone shelters, which are typically public facilities, lack necessities for evacuation use.³

Hence, domestic and international nongovernmental organizations (NGOs) have proposed the construction of cyclone shelters with the participation of local residents and use of a new management method. In particular, the house-shaped shelter proposed by BRAC University has attracted attention as an innovative project that proposes “homes that do not require evacuation to cyclone shelters,” and 43 house-shaped shelters have been constructed in 2011 funding from the United Nations Development Programme (UNDP).⁴

Literature review

Numerous surveys and studies have been conducted regarding cyclone damage and countermeasures in Bangladesh. The Bangladesh government and the Bangladesh University of Engineering and Technology have reviewed the number of cyclone shelters built in Bangladesh by 1993, as well as their maintenance status and have published a standard design for cyclone shelters.⁵ The Public Works Research Institute, an independent administrative institution, systematically summarized the cyclone damage in Bangladesh in its publication, “Factor Analysis of Water-Related Disasters in Bangladesh” and presented the history and issues concerning cyclone shelters.⁶ In addition, Paul considered evacuation behaviors during cyclones and discovered that appropriate maintenance was not being performed after the construction of cyclone shelters at the coastal areas in Bangladesh.¹ Ikeda investigated cyclone-related gender issues and clarified that some people could not evacuate cyclone shelters owing to social and religious reasons.² Hence, Saito et al. classified the management and operation of cyclone shelters into three categories: residents, NGOs, and administrative bodies.⁷ Furthermore, they participated in the development of cyclone shelter guidelines by residents and indicated that the management and operation by residents affects decision-making during a disaster. Kusakabe's research on the evacuation behavior on Hatiya Island after a cyclone disaster revealed that cyclone shelters were constructed for temporary evacuation in case of inundation by storm surges; therefore, they were not considered for use in evacuation refugees' daily lives.³ Furthermore, cyclone shelters have been miniaturized, and the Japanese Red Cross Society identified small shelters that were used as warehouses in detached houses, which housed 10-20 people from the 1991 cyclone disaster.⁸ However, studies on the use of cyclone shelters during temporary evacuation are few, especially those considering refugees' daily lives during reconstruction. Although the project objectives and construction process records of BRAC University's house-shaped shelters project led by Fuad have been reported,⁹ the shelters have not been damaged by the cyclone since their construction in 2011, and no surveys have been conducted.

Based on the information above, this study was conducted to clarify the actual use of cyclone shelters as evacuation centers during and after cyclone events and the possible use of house-shaped shelters.

Research objective and methodology

This study aims to propose the appropriate use of conventional cyclone shelters and the construction of cyclone shelters that ensure the safety of all residents in cyclone-hazard areas. The objectives of this study are as follows: (i) to clarify how cyclone shelters are used during temporary evacuation and in evacuation refugees' lives, and (ii) to elucidate the maintenance status of house-shaped shelters and residents' awareness of disaster prevention to consider the potential of house-shaped shelters.

Hence, literature review and field survey were conducted. To comprehend the cyclone damage in Bangladesh and their evolving countermeasures as well as cyclone shelter construction projects, literature, previous studies, and project reports issued by international organizations and NGOs were referenced. To address point (1), a questionnaire survey was administered to 70 households living in the vicinity of a cyclone shelter from August 9 to August 20, 2015 in Hatiya Island, where the number of cyclones is high and local NGOs are actively engaged in disaster-prevention education activities. To achieve objective (2), a questionnaire survey and a measurement survey of house-shaped shelters were conducted from August 30 to September 18, 2015 for 41 households living in house-shaped shelters in Padmapukur, a construction site. Figure 1 shows a segmented map of cyclone hazard areas published by the Bangladesh Disaster Management Bureau.¹⁰ All surveyed areas belonged to the high-risk area.

Figure 1. Cyclone risk zone (Ref. Disaster Management Bureau of Bangladesh)10

This research is part of a joint project on cyclone shelters between the Graduate School of Global Environmental Studies, Kyoto University and the Department of Architecture, BRAC University.

Cyclone Damages and Countermeasures in Bangladesh Outline of Bangladesh

The People's Republic of Bangladesh, with an area of 147 570 km² and a population of approximately 159.4 million, is located on the world's largest delta, the Bengal Delta, which comprise more than 310 rivers. Its climate is tropical monsoon, and cyclones often hit this area from April to May and October to November annually. Approximately 80% of the population lives in rural areas, with a per capita Gross Domestic Product (GDP) of $1110, ranking 159th among 190 countries.[Note 2]

Approximately 70% of the land is flood plains of major rivers and one-third of it experiences floods at least once annually, although embankment developments change the irrigation pattern in the Bengal Delta. In these areas, residents secure their living spaces on premises with several meters in the flood plain and are connected to the embankment by a narrow causeway.¹¹ Therefore, when causeways between the fields are flooded in the rainy season, the residence sites become isolated islands. Many people in these coastal lowlands have lost their lives during cyclone storm surges.⁶

Cyclone damage in Bangladesh

The cyclone that hit Bangladesh in 1970, before its independence, was one of the largest natural disasters worldwide in the 20th century, with more than 300 000 deaths. Since then, cyclones have hit the country almost every other year, causing serious damages to people and livestock as well as significant economic loss. Table 1 shows the major cyclones and their damages from 1970 to the present. The number of deaths has decreased owing to cyclone measures; however, the number of victims remains high. Livestock damage results in not only livestock loss, but also causes the spread of infectious diseases. Foot and mouth disease, which is an endemic in South Asia, is highly contagious and easily transmitted by floods.

Table 1 Death toll and affected people by major cyclones since 1970 (Ref. World Bank (2010) and Government of Bangladesh (2010))

Date	Name	Death toll	Affected people
November 12, 1970	Bhola	300 000	3 648 000
April 29, 1991	Gorky	138 000	14 438 849
November 15, 2007	Sidr	4234	9 000 000
May 25, 2009	Aila	190	3 935 341

Bangladesh's coastal areas and islands have a population of approximately 15 million people, of whom approximately 6.4 million live in high-risk areas where storm surges reach a depth of more than 1 m and may result in the loss of many lives.

Cyclone countermeasures in Bangladesh Hardware countermeasures

The Bangladesh government has taken measures against cyclone damages, including the construction of seawalls, planting mangroves, building a hill called killa, and constructing cyclone shelters. The construction of the seawalls started as part of flood control measures when modern infrastructure was introduced in the British territory before independence from India. A master plan was established during the period of East Pakistan, which was 7500 km long in 1971 when Bangladesh became independent.⁶ However, many issues regarding its use were raised, including the following: the levee was destroyed by the residents after its construction because consent from beneficiaries were not obtained prior to its construction; its maintenance was not performed appropriately; and the effects of the waterway on the ecosystem, agriculture, and transportation were not considered.¹² The establishment of tide-prevention mangrove forests began in 1996 to protect the embankments. However, logging for export timber and fuel has progressed, and currently, the land where the mangroves are cut down is being converted to shrimp aquaculture land, defeating these forests' initial purpose. Killa is a circular mounded hill proposed by the Bangladesh Red Crescent Society and developed to evacuate livestock during cyclone storm surges. A total of 146 killas existed in 1993.⁵ Both circular and elevated grounds piled up for flood-tide evacuation are called killa by local residents.

The construction of cyclone shelters has been planned since the late 1960s in East Pakistan. Because they saved many lives in the aftermath of the 1991 cyclone, shelter construction became the cornerstone of cyclone damage countermeasures. As of 2007, 3976 cyclone shelters exist.¹ The Ministry of Disaster Management announced that 2000 more cyclone shelters would be required owing to population growth; subsequently, 807 were confirmed to be constructed by the Ministry of Primary & Mass Education, local governments, and NGOs.¹⁰ The architectural features of these cyclone shelters include (i) a multistory building, (ii) an open space on the ground floor in case of flood tides, and (iii) a capacity of 500-2500 people (Figure 2A).¹³ However, many cyclone shelters became unavailable owing to inadequate maintenance outside the cyclone season. Since the master plan formulation by the Bangladesh government in 1991, these shelters have been used as schools during non-cyclone times (Figure 2B).

Figure 2. (A) Cyclone shelter; (B) Room used as classroom

Early warning system

In 1972, the International Red Cross/Red Crescent Society and the Bangladesh Red Crescent Society provided support to cyclone victims. Owing to the cyclone disaster of 1970, an early warning system, the Cyclone Preparedness Programme (CPP), was organized to provide cyclone warning and evacuation guidance to residents in remote areas without electricity. First, information from the storm warning center is transmitted to CPP headquarters and then to the field office; subsequently, local CPP volunteers use sirens and megaphones to directly inform villagers of the impending cyclone. As of 2011, more than 50 000 local CPP volunteers have participated in the program.¹⁴

Challenges of cyclone countermeasures Factors that hinder willingness to evacuate

Even though cyclone shelters are constructed, many residents choose not to evacuate. Their decision is governed by economic reasons, such as unwilling to leave livestock and household goods or the lack of daily income after evacuation; the location of cyclone shelters, such evacuation sites that are far away; and poor road conditions. In addition, many women have been unable to evacuate their homes during a cyclone disaster because the Bangladesh Islamic precepts state that women should not be in the presence of men other than their relatives and that women's role is to protect livestock, household goods, and children. In the 1991 cyclone disaster, more deaths occurred among women than men across all age groups, with women reported to be 4.5 times more likely to die than men, particularly at age 20-49.⁷

Maintenance of cyclone shelters

As of 2007, 3976 cyclone shelters exist in Bangladesh; however, 1576 are physically unusable owing to inadequate maintenance and land subsidence caused by riverbank erosion.¹ Even at cyclone shelters used during cyclone disasters, many cases occurred in which principals and teachers who had the shelter's key were not in the village when the cyclone struck, and residents had to break in to enter the shelter. In addition, during this investigation, many cyclone shelters had collapsed concrete walls, pillars, and beams, exposing the interior's reinforcing steel. Although they were included in the total count above, these cyclone shelters may no longer be useable. One of the reasons for the inadequate maintenance mentioned above is the lack of appropriate operating guidelines for cyclone shelters.

Research Sites Hatiya Island, Noakhali District, Chittagong Division

Hatiya Island, located in the Noakhali District, Chittagong Division, is an island with a population of 350 000 and is located at the estuary of the Brahmaputra, Ganges, and Meghna rivers in Bangladesh. Following a cyclone disaster that killed more than 10 000 people on the island in 1991, the construction of cyclone shelters has progressed rapidly. As of 2015, 151 cyclone shelters exist on the island. In addition, activities to raise awareness on disaster prevention are being performed, focusing on the opening of radio stations by the NGO Dwip Unnayan Shongsta (DUS, Bengali for “island development organization”), which was established after the cyclone in 1970, and local volunteers. In November 2015, with the support of the Japan International Cooperation Agency and the cooperation of the authorized Basic Human Needs (BHN) Telecom Support Council, a Japanese nonprofit organization, the first community radio station on Hatiya Island was launched as part of disaster prevention activities. On July 30, 2015, Cyclone Komen destroyed five embankments in the northern part of the island, and residents evacuated to cyclone shelters.

A field survey was conducted between August 9 to 20, 2015, immediately after Cyclone Komen, in cooperation with the NGO DUS and the BHN Telecom Support Council. Five cyclone shelters that functioned as temporary shelters after Cyclone Komen hit were selected for the study, as local DUS volunteers were available for guidance. The 70 households surveyed were randomly selected to include 14 households living within a 20-minute walk from each cyclone shelter. Oral interviews were conducted based on a semistructured questionnaire.

Padmapukur Union, Satkhira District, Khulna Division

Padmapukur Union, which belongs to the Satkhira District, Khulna Division, is located in a brackish water area where seawater and fresh water are mixed. This is a highly dangerous cyclone area with a population of 26 000. However, compared with Hatiya Island, cyclones do not affect the area as severely, and the mangrove forests located in the south render the area less susceptible to wind damages, therefore, cyclone countermeasures have been delayed. In the aftermath of Cyclone Aila in 2009, the lack of the CPP prevented residents from receiving cyclone warnings, resulting in only two cyclone shelters being open for evacuation: Consequently, 10 people died, 4432 households (22 163 people) were affected, and 12 220 houses were destroyed.[Note 3] Since then, cyclone countermeasures have progressed rapidly, and eight cyclone shelters and 43 house-shaped shelters have been constructed. As of 2015, however, no embankments have been repaired after Cyclone Aila, and boats have been used as the primary mode of transportation because roads are flooded when the river water level increased owing due to rainy seasons or high tides.

The field survey in Padmapukur was performed with the cooperation of BRAC University. The survey targeted 41 households living in 43 house-shaped shelters and 25 households living within a 20-minute walk from the cyclone shelter nearest to the house-shaped shelter that functioned as a temporary shelter during Cyclone Aila. The oral interviews were conducted based on a semistructured questionnaire.

Evacuation to Cyclone Shelters

The evacuation to cyclone shelters on Hatiya Island (number of households surveyed: 70) in 2015 during Cyclone Komen and that in Padmapukur (number of households surveyed: 66) in 2009 during Cyclone Aila were compared and analyzed based on the questionnaire survey results.

Preparedness

In Hatiya, 80% of the respondents were prepared for a possible evacuation before a cyclone hit, whereas only 13% in Padmapukur was prepared (Figure 3A). The measures implemented on Hatiya Island included the preparation of bags for evacuation (Figure 4A), which included items such as clothing, kitchenware, dry food (Figure 4B), saline (oral rehydration therapy to combat dehydration), bedding, kata (Bengali for a thick large cloth), books, and certificates of graduation from elementary and junior high schools. Every year before the cyclone season, women would prepare a bag for their household: essential items would be places in it for evacuation; subsequently, it would be sewed up, draped with a waterproof vinyl, and hung from a beam. Evacuation life in cyclone shelters is considered a normal part of life in Bangladesh; therefore, two types of kitchenware exist, one for daily life and the other for evacuation life.

Figure 3. Percentage of people who (A) prepared for evacuation before the cyclone; (B) evacuated to a cyclone shelter during the cyclone; (C) remained in a cyclone shelter after the cyclone had passed

Figure 4. (A) Emergency supplies stored in a bag hanging from a beam; (B) Dry rice in a bag

In Hatiya Island, households whose monthly income is less than 3000 BDT (Bangladesh Taka, or approximately 4600 Japanese yen[Note 4]) do not have the financial capacity to perform cyclone countermeasures; therefore, the percentage of households that perform cyclone countermeasures is approximately 20% lower than that of households whose income is 3000 BDT or more.

Evacuation behavior

On Hatiya Island, 87% of respondents evacuated to cyclone shelters, whereas only 47% of Padmapukur evacuated (Figure 3B). Regarding pre-evacuation activities, 80% of Hatiya Island respondents were prepared to “take livestock to a safe place” and “move household goods into the attic,” whereas only 14% of Padmapukur were prepared to perform those activities. In Hatiya, 77% of households started evacuating when the inundation height was below the waist, whereas in Padmapukur, 75% had not started evacuating even though the height was above the chest. One of the reasons for this occurrence is that CPPs were not introduced in Padmapukur prior to the cyclone disaster; hence, notification was delayed. Furthermore, even on Hatiya Island, where the evacuation rate was high, the rate of evacuation of cyclone shelters within a 10-minute walk was 100%, whereas the rate of evacuation of residents who had to walk more than 20 minutes was as low as 64%. The distance to the cyclone shelter is hence considered as a significant factor in evacuation decision-making.

Evacuation refugee life

In Hatiya, where 87% of the inhabitants evacuated to cyclone shelters and returned home after the floods receded, most of the households stayed more than 1-2 days, although some stayed more than 10 days. Meanwhile, 47% of Padmapukur residents evacuated to cyclone shelters during the cyclone; however, the number of households that stayed in evacuation shelters increased to 64% during the recovery period (Figure 3C), and the maximum length of stay was one and a half years. The increase occurred because refugees who temporarily evacuated to places other than the cyclone shelter such as nearby houses, roofs, or trees, but could not return to their homes because their houses were completely destroyed, lived temporarily in the cyclone shelters while their houses were being rebuilt. Hence, it is clear that cyclone shelters serve as evacuation centers for residents who evacuated before the cyclone struck in accordance with cyclone warnings; however, they also serve as shelters for evacuation refugee life after the disaster.

Use of Cyclone Shelters: A Case Study on Hatiya Island

In Hatiya Island, a questionnaire was administered to five cyclone shelters (Figure 5A) that functioned as evacuation shelters during Cyclone Komen in July 2015 and to 70 households of its surrounding inhabitants. Figure 5B shows an enlarged map around CS (1) in Figure 5A. The following is based on the information obtained from the questionnaire and interview surveys.

Figure 5. Map of Hatiya Island (CS: cyclone shelter; Source: BHN); (B) Site map with a cyclone shelter (CS1 of 5a) and its surroundings (drawn by author)

Management of cyclone shelters

Among the five cyclone shelters, four were built with the support of the Saudi government and one was constructed with the support of a local government; all are used normally as elementary schools. Each school's key is managed by a principal, a janitor, or cyclone shelter members. When a cyclone strikes, the key-holder unlocks the building when a warning is issued. Maintenance comprised only daily cleaning by the janitor; whereas toilet, windows, doors, and stair handrails that have become obsolete have not been repaired since their construction. The cyclone shelters, constructed with the support of the Saudi government, comprised three floors (Figure 6A) and can accommodate 1000 people. Its emergency toilets were on the roof, and daily toilets were outside the building (Figure 6B). The cyclone shelter, built with the support of the local government, comprised two floors, which can accommodate 100 people; an outdoor toilet for daily use; and no emergency toilet.

Figure 6. (A) Cyclone shelter constructed by Saudi Arabia Aid; (B) Outdoor toilet

Use of cyclone shelters during cyclone disaster Temporary evacuation

During Cyclone Komen, 61 among 70 households evacuated to cyclone shelters. Furthermore, 45 among 61 households took their livestock to the cyclone shelters. According to interviews with the residents, cows and goats were placed on the ground floor, whereas chickens and ducks on the second floor. As the cyclone shelter was used as a school, the warehouse and staff room remained locked and unusable even during the evacuation. Because most residents were Muslims, the rooms were categorized by gender; however, only two of the five shelters were categorized as such. The room allocation was determined by the shelter committee, and the women's room was larger than the men's to allow children to evacuate with their mothers (Figure 7). The hallway was a unisex space (shaded portion of Figure 7); however, men working in the hallway were regarded as invading women's privacy and hence were not allowed to stand and talk therein. Among the remaining three shelters, two were intentionally housed as a mixed-gender shelter owing to requests from families who wished to evacuate together. The final shelter had only one evacuation room; therefore, both men and women had to evacuate together. Men and women typically use separate toilets in daily life, and it is not desirable for both to use the same public toilet; however, because there was only one emergency toilet at the time of evacuation, they had to share it.

Figure 7. Separation of evacuation room by gender during the cyclone (drawn by author)

Evacuation refugee life

The number of days spent in the cyclone shelter depends on the scale of the damage. After Cyclone Komen, 46 among 61 households returned home within 2 days, whereas those whose houses were completely destroyed lived in shelters for more than 10 days.

In the case of evacuees living at the shelters for 1 week, most families had only dry food to eat; however, they brought portable cookers (Figure 8A) and cooked in an empty space in the shelter. Some families returned to their flooded homes to cook on their beds and then brought their meals to the cyclone shelter.

Figure 8. (A) Portable cooker; (B) Temporary house constructed along a roadside

Use of cyclone shelters during housing reconstruction

Temporary housing during reconstruction included relatives' houses, cyclone shelters, temporary housing built by residents (Figure 8B), and tents provided by NGOs. Life at the cyclone shelters during the recovery period differed between the men and women. During the day, men went to the market, rebuilt their homes, and returned to the shelter at night. Meanwhile, women often spend the day in cyclone shelters. Families with no prospects for rebuilding their homes used the open space on the ground floor of the cyclone shelter to store their furniture.

Issues of cyclone shelters

Owing to insufficient maintenance, the roof and doors of the evacuation stairs and emergency toilets deteriorated, rendering them unusable during evacuations. In some cyclone shelters, paint and mortar on pillars and walls came off, and the embedded steel was exposed and became rusty (Figure 9A). At three of the four cyclone shelters constructed with Saudi government support, roofs and doors of emergency toilets on the roof were damaged or missing (Figure 9B). Interviews with residents revealed issues from a gender perspective, such as the fact that only one emergency toilet is available for hundreds of evacuees, and that it cannot be used by either gender during evacuation, as a person of another gender may be using it. Hence, many residents evacuated in the open space on the rooftop or on the ground floor with their livestock, rendering the environment in the shelter unsanitary. In addition, the government's guidelines for cyclone shelters assumed that all rooms would be used as evacuation shelters when they calculated the number of occupants. However, in all five shelters, warehouses and staff rooms where important documents were stored were locked during evacuation. Some residents requested for the room to be unlocked by the principle for increased space and separation.

Figure 9. (A) Exposed steel in a pillar; (B) Emergency toilet without roof or door

Potential of House-shaped Shelters: A Case Study in Padmapukur Union Construction project of house-shaped shelters

After Cyclone Aila in 2009, BRAC University received funding from the UNDP for the construction of 43 residential shelters and one school/cyclone shelter[Note 5] for low-income families living on land provided by the Bangladesh government, called Khas Land in Padmupukur (Figure 10), Shamnagar District, Satkhila District, and Khulna Division, which was the most affected area in Bangladesh.

Figure 10. Site plan showing Padmapukur (drawn by author)

The construction project was designed to accomplish the “construction of houses that do not require evacuation to cyclone shelters” and “independence of the affected people through participation in the project”, it was designed by a professor at BRAC University based on a model developed by participants of a design workshop held in the capital city Dhaka, and Padmapukur. The prototype of the house-shaped shelter was a piloti with reinforced concrete (RC) columns on the ground floor, which was built on the premise of extension by residents, and a wooden house on the second floor (Figure 11A). The second floor, which includes a veranda, was approximately 4 m by 4 m wide and functions as a shelter during the event of a cyclone. The capacity was calculated to be 3.3 people per 1 m², and it was assumed to be able to accommodate 54 people (based on an interview with the person in charge). After the prototype construction in 2011, an NGO led by BRAC University replaced the wooden walls on the second floor with brick walls; however, these organizations were not involved in the extension of the ground floor. According to the questionnaire survey, all residents independently extended their shelters (Figure 11B). The primary building materials used for the extension of the ground floor were wood, galvanized iron, bricks, and mortar, in which wood was the most typically used material for the wall removed from the renovation of the second floor.

Figure 11. (A) Prototype of house-shaped shelter (Ref. UNDP); (B) House-shaped shelter modified by a resident

Use of house-shaped shelter

The ground floor was extended and modified in all houses and was used as a bedroom, kitchen, and store room. Furthermore, the second floor, which functions as a cyclone shelter, was used as a bedroom in 27 buildings, as a bedroom/food storage space in 10 houses, and as a storeroom for one house, while the remaining four households did not use the second floor.

The ground floor was extended and modified by local carpenters and masons, and the cost was borne by the residents. The extension and modification can be categorized into two forms: (i) building extension separated from the house-shaped shelter, and (ii) continuous extension of a room from the house-shaped shelter. In both patterns, the toilet, livestock shed, and kitchen were built as separate buildings (Figure 12). According to the questionnaire survey, problems regarding the house-shaped shelters included leaky tiled roofs during the rainy season, erosion of soil foundations by rain and floods, and deterioration of wooden parts on the second floor owing to rain leaks. Hence, the residents performed measures such as reinforcing the gaps between the joints of bricks and tiles with mortar, roofing using galvanized iron, attaching vinyl sheets to the ceiling, and covering the foundation with bricks.

Figure 12. (A) Extension of a building separated from a house-shaped shelter; (B) Continuous extension of a room from a house-shaped structure. B: bedroom; F: Food stock; K: kitchen; Ls: Livestock shed; S: Storage; T: toilet; W: Well; Wt: Water tank. (drawn by author)

Additionally, BRAC University held a workshop for residents in 2015 because “residents should understand the measures against rain leakage on the second floor and the appropriate extension and modification methods of the ground floor.” In the workshop, the appropriate use of building materials and construction methods were presented in which methods to use metal fittings for extensions and reconstructions that were attached to RC beams were explained.

Disaster prevention consciousness of residents

A questionnaire survey regarding residents' disaster prevention consciousness and the availability of house-shaped shelters during cyclone disasters were analyzed. In response to the question, “in the future, if a cyclone warning (No. 8)[Note 6] of the same scale as Cyclone Aila is issued, will you evacuate to the cyclone shelter,” the residents answered, “remain in a house-shaped shelter” (61%) and “take refuge in a cyclone shelter” (39%). Regarding their reasons for remaining in house-shaped shelters, some respondents answered, “house-shaped shelters were built as cyclone shelters,” whereas others answered, “I am concerned about the safety of the wooden house-shaped shelter, however, I still choose it over the RC cyclone shelter because the latter is located more than one hour away on foot.” When a No. 5 cyclone warning was issued in 2014,[Note 7] 68% of residents remained in residential shelters, whereas 29% evacuated to schools and shelters built under the project. In addition, 78% of the residents responded that they “evacuated livestock upstairs in case of a cyclone,” indicating that house-shaped shelters may be used as livestock shelters.

Issues of house-shaped shelters

House-shaped shelters were intended for residents in the neighborhood; however, the housing security was strengthened by the extension and modification of the ground floor, which resulted in the difficult access to the house-shaped shelters on the second floor. In this regard, even relatives could not access without the permission of their parents, husbands, or brothers because the second floor was occupied by their parents or brothers, or managed as a warehouse. In addition, the response of the interviewees that “the structure of the second floor is made of wood; therefore, it is unsafe to use it as a cyclone shelter” indicates that it is a highly likely that local people in the vicinity will not be able to use house-shaped shelters when a cyclone occurs. The height from the ground to the second floor of the house-shaped shelter was 3.3 m including the foundation, and the height to the roof was approximately 5 m. However, the height of cyclone affected storm surges can be 5-9 m, therefore, the safety of the shelter is of concern. Furthermore, although the residents themselves were taking measures against the problems of roof leaks and the aging of wooden parts on the second floor, many have complained that “BRAC University should fix the house-shaped shelters because they built them.” Therefore, it cannot be concluded that the “independence of the residents” is achieved through the project.

The coastal cyclone high-risk area in Khas Land is an agricultural and fishing village.; The residents are of the low-income bracket, and they typically live in a detached house with an pond added in front of the house, which is used to store fresh water to bathe, wash clothes, and cultivate fish; furthermore, they manage livestock by constructing a barn in the front yard of the house. The construction cost of a house-shaped shelter was approximately 120 000 BDT (approximately 185 000 Japanese yen) and approximately 15 times that of a typical home in a rural village in Bangladesh (7500 BDT, or approximately 11 500 Japanese yen).[Note 8] Considering that most of the people living in cyclone high-risk areas are of the low-income bracket, dissemination without financial assistance would be difficult.

Conclusion

Cyclone shelters, which have been promoted as a countermeasure against cyclone disasters, are not only used for immediate shelter for some hours after flood tide inundation, but also for several days to months after the disaster. However, they lack essential elements because the toilets and stairs for evacuation are broken and cannot be used. Previously, the rooms for evacuation were separated by gender for religious reasons. However, some residents at cyclone shelters have requested to be segregated by family (mixed sex); therefore, it is necessary to review the residents' traditions and awareness, which may have changed in recent years, and consider new forms of evacuation spaces that reflect those changes. In addition, the rooms typically used as staff rooms and warehouses in schools remained locked even during evacuation, preventing shelters from achieving their intended occupancy levels and ratio. Therefore, cyclone shelter designers should plan the allocation of rooms for evacuation based on the use of staff rooms and warehouses at the planning stage, and explain to school staff and local residents regarding their use during evacuation.

The house-shaped shelters built by BRAC University can be used as shelters and warehouses for livestock as well as residents, and residents who have not evacuated because they do not want to leave their livestock and household goods will be able to evacuate. Meanwhile, the safety of cyclone shelters poses challenges, and it is difficult to disseminate them without financial assistance, considering the expensive construction costs. In addition, residents indicated challenges in obtaining financial support and state that BRAC University should fix and modify the shelters that they built. Hence, maintenance by residents must be firmly established, eg, by regularly holding workshops pertaining to extension and reconstruction.

The number of cyclone shelters constructed as a measure against cyclones since the 1960s, before Bangladesh's independence, was approximately 4000 in 2007. Although these shelters have saved many lives, they could not accommodate all the residents of the cyclone high-risk areas located on islands in coastal deltas. Meanwhile, house-shaped shelters may be difficult to build in large scales owing to the construction cost; however, they can be used as small cyclone shelters for small-scale communities such as islands in cyclone hazard areas where conventional cyclone shelters cannot be constructed. House-shaped shelters, however, are an experimental project that started after Cyclone Aila in 2009 and have not been subjected to cyclone damage; therefore, no research has been conducted on their use as evacuation facilities. The results of the survey provided herein were limited to discussions on the actual conditions of cyclone use in normal times. The availability of cyclone shelters in times of cyclone attacks could not be investigated. Hence, a long-term investigation should be conducted to clarify changes in safety owing to the aging of house-shaped shelters and the maintenance of residents, as well as the actual conditions of their use as shelters during a cyclone disaster. Furthermore, a comprehensive cyclone shelter layout plan should be developed including the introduction of small-scale cyclone shelters from the viewpoints of evacuation distance, construction cost, and capacity for communities such as those in Khas Land, where it is difficult to construct conventional cyclone shelters owing to their island location in coastal deltas; and to secure evacuation routes during cyclone storm surge floods.

Acknowledgments

This work was supported by the Japan Public-Private Partnership Student Study Abroad Program: TOBITATE! Young Ambassador Program. The authors express their gratitude to the funding source.

Disclosure

The authors have no conflict of interest.

Notes:

^{Note 1)}The number of fatalities in 1970 has been reported by a number of organizations; however, the figures are based on the 2013 edition of the White Paper on Disaster Management published by Japanese Cabinet Office.¹⁵

^{Note 2)}The figures for Bangladesh's GDP and GDP growth rate were obtained from the Statistical Yearbook published by the Bangladesh Bureau of Statistics in 2014.¹⁶

^{Note 3)}Although there were many cyclone damage reports in 2009, the description of the damage report published in 2009 by the local government of Shamnagar District, Bangladesh was adopted in this study.

^{Note 4)}1 BDT = 1.54 Japanese Yen (2016).

^{Note 5)}The reconstruction of an elementary school as a cyclone shelter in Khas Land was part of the project after Cyclone Aila destroyed it. The school has a reinforced concrete structure and brick walls, four classrooms, separate toilets for men and women, and a rainwater storage tank on the second floor. It can accommodate 500 people with livestock.

^{Note 6)}The Ministry of Disaster Management in Bangladesh classifies cyclone warnings (warning signal) into 11 categories, in which No. 1 has the lowest risk and No. 11 the highest.

^{Note 7)}In 2014, a No. 5 cyclone warning was issued around Padmapukur; however, the cyclone did not affect the area and house-shaped shelters were not damaged (according to an interview with local residents).

^{Note 8)}A questionnaire survey revealed that the German Red Cross provided 7,500 BDT per house for the reconstruction of houses after the 2009 Cyclone Sidr, and that the residents used this money for rebuilds from the foundation.