key: cord-283169-v1ly657m
authors: Zaman, Shamrita; Sammonds, Peter; Ahmed, Bayes; Rahman, Taifur
title: Disaster risk reduction in conflict contexts: Lessons learned from the lived experiences of Rohingya refugees in Cox's bazar, Bangladesh
date: 2020-06-01
journal: Int J Disaster Risk Reduct
DOI: 10.1016/j.ijdrr.2020.101694
sha: 
doc_id: 283169
cord_uid: v1ly657m

Bangladesh is currently hosting more than one million stateless Rohingya refugees, who fled from the Rakhine State to avoid serious crimes against humanity persecuted by the Myanmar Army. The newly arrived Rohingyas were accommodated in overcrowded refugee camps in Cox's Bazar District (CBD). The camps are highly vulnerable to landslides, tropical cyclones, flash-flooding, and communicable disease outbreak. Although a number of improvement measures are ongoing, however, no study to date has addressed Rohingyas' self-adopted strategies to mitigate disaster risks. Consequently, this paper aims to explore how refugees cope with risks associated with environmental hazards in the Kutupalong Rohingya Camp in CBD. A mixed-methods research strategy incorporating both quantitative household questionnaire survey and qualitative focus group discussions (FGDs) techniques were applied. In total 250 Rohingya refugees were selected for the questionnaire survey using a stratified random sampling method from camps 17 and 19, and two FGDs (male and female-only) were carried out in camp 13 involving 21 Rohingya participants. Results derived from the study show that responding to early warning systems, storing dried food and medicine, utilising available resources, relocating to safer shelters, and keeping hopes high were some of the coping strategies practised by the respondents. Literacy level imposed a significant impact over respondents' perception to accept various measures. For instance, the probability of storing dried food in preparation for disasters was 4 times higher among literate Rohingya compared to their illiterate counterparts. Similarly, for literate respondents, the probability was 20 times higher to store medicine than for illiterate. Guaranteed distribution of shelter strengthening kits among all refugee households, inclusion of disaster risk awareness and preparedness trainings, ensuring safe and dignified return in Myanmar, and global and regional cooperation to address the refugee crisis are some of the propositions recommended in this study for improving Rohingyas' future adaptation strategies in a humanitarian context.

Disasters triggered by natural hazards cause loss of life, property damage, forced displacement, hunger, and disease outbreaks. The world's poorer or less-resilient nations, and especially marginalised groups such as minorities, displaced people and refugees are also highly impacted by natural hazards [1] . Considerable emphasis has been given by the scientific community to identifying traditional coping strategies practised by the urban or rural poor in a disaster context [2] [3] [4] [5] [6] . However, there is a gap in conducting research solely on refugees in such a meticulous way. The present study attempts to focus on exactly this point by considering the Rohingya refugees crisis in Cox's Bazar District (CBD), Bangladesh as an example.

By definition, refugees are persons "someone who is unable or unwilling to return to their country of origin owing to a well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group, or political opinion" [7] . From this definition, the arrival of refugees is considered as a temporary phenomenon for the host country. The durable solution regarding refugees is that they will be able to return to their homeland voluntarily as soon as the situation, which forced them to flee from their territory, ends [8] . But the real situation is quite different. A study by the United Nations (UN) High Commissioner for Refugees (UNHCR) reported that around 70.8 million individuals, including 25.9 million refugees (registered only), were forcibly displaced worldwide at the end of 2018. Currently, 33% of the global total refugees are being hosted by the least developed countries where they live very distressing lives. Bangladesh is ranked 8 th , accommodating nearly 1 million Rohingya refugees from Myanmar [9] .

Fragile camp settings with limited access to basic provision of infrastructures and aid, difficult camp locations (e.g., too arid or too hilly), overpopulation and rickety shelters are some of the common phenomena with which the refugees are forced to deal with recurrently. Such fragile physical settings act as driver to increased vulnerability to natural hazards [10] . For instance, communications between the Leitchuor refugee camp in Ethiopia and the surrounding community were interrupted for a month because of the disruption of the camp's main access road due to flooding in July 2014. Again, 250 shelters on steep hillslopes in refugee camps in Rwanda were destroyed due to a landslide in 2014. A survey conducted by UNHCR in 16 countries suggested that out of 3.2 million refugees, 200,000 were affected and 100,000 were displaced by disasters associated with natural hazards during 2013 and 2014 owing to living in vulnerable camp settings [10] .

Consequently, the question is how the refugee communities can build coping strategies to respond to potential disaster risks. Vulnerability of the poor in their own territory and vulnerability of refugees in a host country are not similar phenomena. The vulnerable in their own country may enjoy freedom of movement, better access to relief aid, alternative livelihoods, access to evacuation shelters, internet, telecommunications, vaccines and so on, to mitigate disaster risks, which typically are denied for refugees.

Rohingyas are the second largest ethnic group of the Rakhine State in Myanmar [11] . They have distinguished culture, heritage, language and religion [8] . Their citizenship was denied by the Myanmar government in 1982 [11] [12] . Being targeted by the state-sponsored persecution, Rohingyas fled from Myanmar and started arriving in Bangladesh chronologically, first in 1978, then in 1991-1992, 2012, 2016 , and lastly in late August 2017. Nevertheless, it was August 2017 when Bangladesh experienced by far the largest and fastest refugee influx [13] . The International Court of Justice (ICJ) has instructed Myanmar to take actions to stop genocide of the Rohingyas on 23 January 2020, and the final verdict on the punishment of the crime of genocide is still pending [14] . Nearly 708,985 Rohingya refugees (officially known as Forcibly Displaced Myanmar Nationals) including more than 460,000 children (27.3% boys and 26.2% girls) and 217,000 adult women have fled into CBD since August 2017, and they are now settled in two main camps -Kutupalong and Nayapara [13] . CBD is currently hosting 859,161 Rohingyas ( Fig. 1 ) who are officially registered by the UNHCR [15] . [15] . In addition, the crisis has adversely impacted more than 444,000 host community members. The 2020 Joint Response Plan (JRP) is seeking US$ 877 million to respond to the critical humanitarian needs in CBD [13] . The Rohingya refugees are not only unique in terms of their great numbers, but also in the sense of their exposure to adverse weather condition in the refugee camps [16] . Every year, the Rohingyas are facing catastrophic impacts of natural hazards in CBD. For instance, heavy downpour for five consecutive days with landslides on July 2019 affected more than 3,000 Rohingya families and took away lives of two people [17] . Again, a total of 55,057 Rohingyas were affected by the rainfall-triggered landslides, flash floods and waterlogging during May-December in 2018. Moreover, the devastation caused by the Cyclone Fani in May 2019 recognised the need for better preparedness against extreme storm events in the Rohingya camps [18] .

In this context, the paper aims is to assess refugees' adopted strategies to cope with the shifting risks (i.e., the type of risks associated with natural hazards that has changed over time with the change of location) and also, the level of organisational involvement in implementing disaster resilience interventions in the camps. The central argument of the paper is that a new dimension can be added to the studies of coping strategies for the physical and built environment field by exploring disaster risk reduction (DRR) strategies developed and practised by a refugee community. Accordingly, the research question is 'despite facing many challenges, how do Rohingya refugees sustain their existence against the growing vulnerability raised by extreme weather events'? The alternative hypothesis of this study states that Rohingyas have developed their own coping mechanisms, in an attempt, to protect themselves against shifting risks posed by natural hazards. The null hypothesis is that no such coping mechanisms have been developed by the Rohingyas.

The study area, Kutupalong Balukhali Expansion Site or here in short Kutupalong Rohingya camp (located in between 21°10'20'' and 21°13' north latitudes and in between 92°8' and 92°10'15'' east longitudes), is bounded by Bandarban district on the northeast, Teknaf Upazila (sub-district) on the south, the Rakhine State (formerly known as Arakan) of Myanmar on the east, and the Bay of Bengal on the southwest. The Kutupalong camp is the largest refugee settlement in the world which is sheltering over 40,000 Rohingya per square kilometre (e.g. population density of London, Tokyo, and New York cities are respectively 4542, 6263, and 10194 per km 2 ). The study area is mostly composed of small hillocks and valleys. The surface geology is primarily composed of sandstone, clay and siltstone. The average temperature of CBD is 25.6 °C and average annual rainfall is 4,200 mm. The total area of the main Kutupalong camp is approx. 14.57 km 2 , and around 30% and 17% areas are highly susceptible to flooding and landslides, respectively. Camps 13, 17, and 19 were selected for this study (Fig. 2 ) because of their high exposure to different natural hazards [19] [20] [21] . For example, from April to November 2019, over 1,400 landslide incidents impacted 3,184 households, 499 wind/storm incidents impacted 10,127 households, and 68 flooding incidents impacted 5,706 households across all camps. Overall, the incidents were responsible for 76 individual refugee injuries and 10 fatalities, and 3,972 households were displaced [21] . Some basic information about the selected camps are described in Table 1 . Families with persons of special needs (%) 29 30 29 Source: UNHCR, 2020 [22] .

The study deployed a mixed-methods research strategy combining both quantitative household-based questionnaire survey and qualitative focus group discussions (FGDs) techniques. FGDs were conducted followed by the household questionnaire survey to verify the results obtained from the fieldwork. It is well-accepted that integration of quantitative and qualitative data analysis is necessary to capture the overall vulnerability scenario of a community. While questionnaire surveying enables to extract information regarding socio-economic and demographic issues of a community, FGDs enable the participants to share their perception, knowledge and future plans [23] .

First, household questionnaire surveying was conducted with the newly arrived adult Rohingyas (over 18 years old) who entered into Bangladesh following the August 2017 violence. The survey was administered face-to-face individually with 250 Rohingyas in camps 17 and 19 in order to acquire reliable primary data. A 'stratified random sampling' method (also known as a random walk process) was adopted in this research, because it offers greater precision than a simple random or systematic sample. Only participants living on extremely vulnerable hill slopes or floodplains were selected by ensuring gender balance. The questionnaire consisted of both open and close-ended questions yielding quantitative data, highlighting Rohingyas' exposure to hazards, adopted ways to cope with shifting risks, organisational response to hazards, and the scopes and challenges of implementing good practice. A draft questionnaire was piloted and tested in April 2019. The first draft of the questionnaire had some issues related to redundancy, length, selecting appropriate options for different questions, adjusting cultural understanding and contextualisation. The overall testing performance was satisfactory. A revised questionnaire (see Appendix-I) was prepared for the final round of fieldwork which was carried out in May 2019. All the field surveyors were given necessary training and background instructions prior to conducting the survey, covering issues such as the selection of survey participants, gender balance, due diligence, taking field notes, transcription and translations, risk assessment, research ethics, principles of reliability and validity, permissions, security and safety, unconscious bias, data protection, and professional code of conduct. Participants were informed about anonymity and project objectives, and their consent was sough beforehand. Fieldwork permissions were officially granted by the Office of the Refugee Relief and Repatriation Commission (RRRC) in Cox's Bazar. Five field surveyors (three males and two females) conducted the whole survey who were selected based on their fluency in Rohingya and Bengali languages, and successful completion of the training.

Second, two FGDs were conducted in camp 13 on 16 February 2020 to address the limitations in quantitative methods and validate the results obtained from the questionnaire survey. The FGD participants were divided into men-only (aged between 35 and 50) and women-only (aged between 25 and 35) groups comprising ten and eleven members respectively. The reason behind forming gender-specific group was to understand their gendered experiences. Particular time schedule and suitable location were fixed beforehand. The FGDs were conducted in the Heath Management BD (HMBD) Foundation's premise inside camp 13 by maintaining all relevant fieldwork ethics and regulations. Two trained field enumerators (one male and one female) conducted the FGDs with the presence of HMBD doctors and nurses.

As the study involved human participation, institutional ethical approval (UCL project ID: 15843/001, and data protection ID: Z6364106/2019/05/20) and risk assessment were in place before the commencement of the face-to-face questionnaire surveying and FGDs in the camps.

Responses from complete questionnaires were compiled preparing a database in SPSS software [24] and used for the statistical analyses. Test of two proportions, chi-square test of homogeneity (2 X C) and binomial logistic regression are three powerful statistical tests that were used in this study. The test of two proportions and chi-square test of homogeneity (2 X C) were used to determine whether the difference between the binomial proportions of independent variables on a dichotomous dependent variable is statistically significant or not. The basics of these two tests are mostly similar except the nature of dealing with the number of independent categories. Test of two proportions is applicable where the experiment demands to explore statistical significance between two independent categories [25] [26] , whereas the chi-square test of homogeneity (2 X C) is applicable if independent variable contains three or more categories [25] [27] . Binomial logistic regression is another test that was used to predict the probability that an observation falls into one of two categories of a dichotomous dependent variable based on one or more independent variables [28] [29] [30] . Binomial logistic regression proved to be the best fit for examining the type of interaction that was sought between the dependent and independent variables in this study. During the FGDs, detailed notes were taken in Bengali language, and later it was translated into English for further analysis.

Of the 250 Rohingyas surveyed, 41% were male and 59% were female.

Respondents' age ranged from 20 to above 50 years old. Nearly 48% of the respondents were aged between 20 and 30, while about 39% were aged between 31 and 50, and the rest of 13% were above 50 years old. Approximately 80% of the respondents were from Maungdaw township of Rakhine State; others were from Akyab, Buthidaung and Rathedaung townships. Only 24% were literate which means that the literacy rate is very low in this community.

Rohingyas' primary source of income in Myanmar was agriculture and fishing. Other occupations were business, teaching, tailoring, day labouring, and so on (Fig. 3a) .

Male participants showed diversification in their livelihoods; while only a minority of women had varied livelihoods (e.g., farming, business) other than being a housewife. The average monthly income for those who did business or fishing was above 200,000 Kyat (Kyat is the currency of Myanmar; 1 USD = 1526 Kyat during this research). Other professionals, like physician, tailor, teacher and farmer had a monthly income of above 100,000 Kyat on an average. In contrast, above 70% of men and 90% of women were found unemployed in the selected refugee camps (Fig. 3b) in CBD. Less than 20% of men were involved with 'cash for work' programmes run by the UN agencies or non-governmental organisations (NGOs). Both men and women described working as volunteers or running small-scale informal businesses. 'Block Head' (Majhi in Bengali) is a type of unpaid service with which male refugees were reported to be involved. The unemployed refugees are entirely dependent on aid, more specifically on food rations. The refugees, who have entered Bangladesh since August 2017, have not yet received any legal approval to work [31] . Such barrier to income generation has amplified their vulnerability to environmental disasters including communicable disease outbreak. 

Results suggest that more than 80% of the respondents came from Maungdaw township in Rakhine State. Around 62% of the respondents reported their exposure to cyclone, 45% to flooding (riverine and coastal) and only 2% stated their exposure to landslide hazards in Rakhine. During the FGDs, Rohingyas also recognised their low exposure to landslides in Rakhine as most of them lived in low-lying flat land. In contrast, Rohingyas are now highly exposed to rainfall-triggered landslide in the camps in CBD. The Rohingyas who participated in the survey confirmed three natural hazards to which they were vulnerable (Fig. 4) , namely extensive rainfall, windstorm and landslides. During the FGDs, Rohingyas living on steep slopes have expressed no fear to flash flooding, but showed concern about the rainfall-triggered landslides. They gave credit to the spiritual power/almighty (Allah) for not having encountered any catastrophic cyclone in the last three years.

Rohingyas revealed through the survey that receiving an early warning message and acting accordingly, maintaining storage spaces above the ground to protect non-food items, storing of medicines and dry foods were very common type of measures in preparation for a disaster. The study investigated whether the demographic features such as age or education made an impact on adapting coping strategies or not.

Receiving and understanding early warnings among different age groups of Rohingya population produced mixed results. Respondents aged under 30 years old were more able to receive and understand early warning messages compared to other age groups (Fig. 5) . A sensitivity analysis was conducted to split the respondents within several age clusters.

Pattern of receiving and understanding early warning messages among various age groups in Rohingya population in CBD.

Using chi-square test of homogeneity (2 X C), the recorded proportional difference of receiving and understanding early warning messages within different age groups was found not to be statistically significant at the 0.05 level (p = 0.98). It indicates that irrespective of age groups, respondents had the similar ability in receiving and interpreting early warning messages. This hypothesis was supported by the fact that while in Rakhine, most Rohingya faced at least one type of natural hazard-induced disaster that taught them how to react after receiving early warnings in the event of a disaster [32] . While in Rakhine, about 58% of respondents of different age groups practised deliberate relocation to the nearest cyclone shelter or other ad hoc shelter (e.g., monastery, school) based on early warnings received. Apart from the relocation, they used to practise some other tactics in Rakhine based on the early warnings received, which they discussed in the FGDs. For instance, some of them used to take shelter on the higher ground before flooding and would come back to the plains after draining out of water. Again, some of them used to move all their furniture, kitchen utensils and necessary documents to the upper storey of twostoried building before flooding.

The study also aimed at determining the impact of Rohingya literacy on the implementation of coping strategies before disasters. A binomial logistic regression model was used to predict the probability to store dried food before a disaster strikes based on Rohingya literacy status. It was important to know the overall model evaluation, statistical significance of the individual predictors and validation of predicted probabilities to assess the soundness of the model [30] which are given in Table 2 . Rohingya literacy status was a significant predictor for storing dried foods as a coping mechanism before disaster (p <0.05). The logistic regression model was statistically significant (χ 2 = 19). The model explained 10.8% (Nagelkerke R 2 ) of the variance in the storage of dried food and correctly classified 72.8% of cases. Sensitivity (the proportion of correctly classified events) was 43.5%, specificity (the proportion of correctly classified non-events) was 84%. The positive predicted value (the percentage of correctly predicted cases with the observed characteristic [29] , i.e., participants responded as 'Yes' in case of storing dried food compared to the total number of cases predicted as 'Yes' in case of storing dried food in advance of the disaster) was 50.8% and the negative predicted value (the percentage of correctly predicted cases without the observed characteristic [29] , i.e., participants responded as 'No' in case of storing dried food compared to the total number of cases predicted as 'No' in case of storing dried food in advance of the disaster) was 79.6%. Another important component is the odds ratio which deals with the change in the odds with the increase of one unit of the independent variable [29] . Here, for the Rohingya literacy status, an increase in one unit (i.e., being literate) increases the odds by 4. It means that the odds of storing dried food before disasters ('Yes' category) is 4 times higher for literate as against illiterate Rohingya (Table 2) .

Similarly, another regression model was run to predict the probability of storing medicines in advance of disasters based on Rohingya literacy status. Results showed that the variable 'Rohingya literacy status' was statistically significant (p<0.05). The logistic regression model was statistically significant (χ 2 = 79). The model explained 39% (Nagelkerke R 2 ) of the variance in the storage of medicines before disasters and correctly classified 84% of cases. The sensitivity was 64%, specificity was 92%, positive predicted value was 75%, and negative predicted value was 87%. Literates have 19.5 times higher odds to exhibit storing medicines before disasters than illiterates.

The study also analysed how respondents perceived relocation to a safer shelter during the disaster period in the selected Rohingya camps. The majority of respondents were reluctant to evacuate to safer shelter; roughly 38% of them were willing to move in the case of any event. Further analysis indicated that literate Rohingya had the tendency to avoid relocation to safer shelters compared with their illiterate counterparts. Around 68% of literate Rohingya reported unwillingness to move to safer shelters/places while the figure was around 60% for the illiterate Rohingya.

Using the test of two proportions, existing proportional difference between illiterate and literate Rohingyas in case of avoiding relocation was found to be not statistically significant at the 0.05 level (p = 0.29). This supported the hypothesis that irrespective of the literacy status, all the respondents were uncomfortable to leave their current shelter and relocate to a safer place during any disaster. Surveys revealed the reasons behind Rohingyas' tendency to avoid relocation to safer shelter which include possible unsuitability of the relocation site, fear of losing the individual's own shelter, surrounding kinship bond, and weak relationship with the camp-in-charge and target population.

The Rohingya respondents were asked about their perception of present shelter locations. About 94% of the respondents reported that they either live on unstable hilltops/slopes or at the edge of hills. Only 6% marked their shelter were located in safer zones. Among them, approximately 90% and 58% respondents pointed out that their shelters are exposed to landslides and windstorms, respectively (Fig. 6 ). . 6 . Perception of Rohingya on shelters exposed to natural hazards.

Construction of permanent structures is strictly prohibited in the camps as instructed the Government of Bangladesh (GoB) and RRRC. Refugees' shelter condition was found unsatisfactory both in terms of living standards and structural integrity to withstand natural hazards. A number of humanitarian actors have initiated several steps for shelter improvements to protect the refugees. The distribution of emergency shelter kits (ESKs) and upgraded shelter kits (USKs) by humanitarian actors were ongoing in the camps. Each USK consists of 4 bamboo 'Borak', 60 Bamboo 'Mulli' ('Borak' and 'Mulli' are local categories of large and small-sized bamboos respectively, which are currently being used in the Rohingya camp for bamboo housing construction purpose), tarpaulin and ropes [13] . In this study, nearly 40% respondents stated that they received USKs.

In addition, shelter tie down kits (TDKs) were distributed on an emergency basis to provide additional strength to the shelters to withstand against the strong winds and cyclones. Each TDK consists of 60m of 6mm rope, steel pegs, 10 sandbags, printed infographic materials on how to use TDKs to secure shelters, and 2 pieces waterproof plastic bags [13] . Around 70% respondents of this study received shelter TDKs. During the FGDs, it was also found that while most of the refugees used TDKs, some of the recipients sold them to get some cash. Those who got TDKs used it in various ways to strengthen their shelters. For instance, respondents used the tie down ropes to resist uplifting forces (Fig. 7a ) from strong wind, placed biodegradable sandbags at the edge of the cluster of shacks to prevent them from blown away (Fig. 7b) , and used extra bamboos and plastic bags to reinforce the shelters to tackle monsoon rains and winds (Fig. 7c) . A binomial logistic regression model was run to predict the probability of respondents to use TDKs based on their capacity to receiving and interpreting early warning messages about upcoming disasters. Results showed that the variable 'receiving and interpreting early warning messages in the camp' was statistically significant (p<0.05). The logistic regression model was statistically significant (χ 2 = 125). The model explained 39% (Nagelkerke R 2 ) of the variance in the storage of medicines before the disasters and correctly classified 84% of cases. The model explained 55.7% (Nagelkerke R 2 ) of the variance in the use of TDKs and correctly classified 88% of cases. Sensitivity was 93.7%, specificity was 75%, positive predicted value was 89.6%, and negative predicted value was 83.8%. Rohingyas who were able to interpret early warning messages exhibited 44 times higher probability of using shelter TDKs than those who were unable to interpret the alerts.

The stability of shelters in the hilly area is related with the stability of the hills itself.

Respondents were asked about organisational strategies to stabilise the hills against rainfall-triggered landslides. They conferred some strategies that have been implemented in the camp such as placement of bamboos (Fig. 8a) and sandbags (Fig. 8b ) on steep hillslopes, construction of drainage system (Fig. 8c) , retaining walls, large-scale mechanised work to level the steep hilltops, and plantation of vetiver grass across the hill slopes to avert soil erosion and the risk of landslides.

Around 85% of the participants stated that government officials take substantial measures to clear debris from roads or drainage system during the post-landslide period. Also, 84% respondents did not get any DRR training. The participants in the FGDs reported that before the cyclone season, NGOs provided some training in the form of briefings. However, they wanted drill exercises in their blocks. Overall, participants expressed scepticism about the success of ongoing small-scale DRR trainings in the camp. 

Space constraint in the refugee camp remains one of the greatest push factors increasing disaster vulnerability. Congested settlements, scarcity of open spaces, poor access to roads and other basic provisions, and restriction of movement are increasing disaster vulnerability. The availability of land has been expanded to a total of 6,500 acres in the Ukhiya and Teknaf Upazilas [13] . Still, the GoB is struggling to resolve the overcrowded situation in the camp in order to meet basic international humanitarian standards for refugees. Relocation to safer places can only be a possible solution in such scenarios.

Bhasan Char Island (see Appendix-II) has been identified as a potential site to relocate around 100,000 refugees by the GoB. It is a 526-hectare silted-up island floating in the Bay of Bengal and located in Hatiya Upazila of Noakhali District (22°22'05.68''N and 91°22'38.96''E, elevation -2m) in Bangladesh [13] . Nearly 88% of the respondents had heard about relocation option to Bhasan Char Island. Approximately 6% of them were willing to move from their present location, and 4%

were not sure about their decision. Those who wanted to relocate themselves to the island identified 4 factors that influenced their decision: access to safe drinking water, diversified income opportunities, permanent housing and more secured life in Bhasan Char than in Myanmar (Fig. 9 ). They believe that even though the island's situation is unfavourable, life on the island is much better than tolerating violence in Rakhine, Myanmar. About 90% of respondents did not want to relocate in the island. They figured out 4 major factors (multiple answers were captured) in favour of their decisions: instability of the island, inappropriate for human settlement, adverse weather condition, and fear to lose surrounding neighbourhoods (Fig. 10) . Fig. 10 . Perception of the Rohingya in the context of avoiding relocation to Bhasan Char Island.

A broader percentage (around 61%) of the respondents were inclined to repatriate to Myanmar with proper citizenship. Other top responses were access to proper housing, freedom of movement outside the camp, and access to diversified income opportunities (Fig. 11) . Differences of opinion regarding repatriation to Myanmar between male and female groups emerged from the FGDs. Men emphasized on right to property, citizenship, justice, and freedom of movement. The women participants were very concerned regarding their safety and security in Rakhine. Similar findings were reported by the Department of Peace and Conflict Studies at the University of Dhaka in their recent publication on the 2017 Rohingya exodus [33] .

Attempts to adapt to changing risks posed by natural hazards has become an integral part of Rohingyas' life since moving from the plain land in Rakhine to the mountainous regions in CBD. The findings clearly demonstrate a positive correlation between disaster coping strategies and literacy level of Rohingya refugees. Regression analysis shows that the educated community is far ahead compared to the illiterate community in dealing with disasters by adopting various strategies. Similar results were found by Ronan et al. (2010) who reported that youth involved in hazards education programme had an increased level of disaster risk perception [34] . Mishra and Suar (2007) established that people having prior disaster education and experience were more prepared to tackle flooding and heatwave events in Orissa, India [35] . Education is essential to make the Rohingya community resilient to disasters. However, above 30% of children (aged 3-14 years) and 83% of adolescents and youths (aged 15-24 years) still do not have access to any educational or skills development activities [13] . No formal education is permitted in the camps and the learning centres neither follow the Bangladesh nor the Myanmar curriculums. In January 2019, the GoB has approved the UN partners to provide informal education through a newly designed programme known as the Learning Competency Framework and Approach (LCFA). The LCFA covers English and Burmese language, mathematics, life skills and science across levels 1 to 5. But more than 90% Rohingya children do not have learning competencies at LCFA grades 3 or above [13] . Some of the obstacles are socio-cultural norms of restricting girls' mobility after puberty (purdah), allocation of learning centres in highly disaster-prone zones, difficulties in achieving education facilities, and absence of educated Rohingya learning facilitators [13] . The humanitarian actors are addressing these issues by emphasizing on sustainable learning approaches, including the establishment of new learning centres, mainstreaming weather and disaster management related issues in learning materials, and alternative learning modalities (i.e., home-based learning, mobile learning, radio-based teaching) [13] . Marlowe and Bogen (2015) found that young people from refugee backgrounds in New Zealand can be leaders in DRR within their communities [36] , which necessitates proving wider DRR education and trainings among the Rohingya children and adolescents.

The culture of using shelter strengthening kits and the involvement of humanitarian actors in distributing those kits suggest that institutional cooperation enables Rohingya to involve in DRR initiatives. Being able to understand the early warning message means to get a clear idea about the severity of the imminent disaster. Rohingya who got access and capacity to interpret early warning messages were most likely to use TDKs to increase the robustness of their shelter to withstand against natural hazard-induced disasters. Receiving early warning messages about impending disasters in time requires improved access to information systems, which is being addressed by the Communication with Communities Working Group [13] .

It should be noted that the Inter Sector Coordination Group (ISCG), in close association with the GoB, manages all the 117 active project partners and 160 currently ongoing projects in the Rohingya refugee camps in CBD. The Natural Hazard Risk Analysis Taskforce (NatHaz TF) operating under the ISCG produces and validates hazard maps, and provides guidelines on DRR strategies. The study reveals that about 94% respondents did not receive any DRR trainings in the selected camps. There exists a growing necessity to integrate DRR and emergency preparedness through a community-based approach. The JRP 2020 reports that only 3% shelters meet desired performance standard, 20% households are benefitting from treated bamboo, and 0% households have updated multi-hazard operational plan [13] . The 'Preparation and Contingency Plan' proposed by the JRP is based on a scenario where most camps and displaced Rohingyas are likely to be threatened by the landfall of a Category-1 cyclone in CBD. Conceptually it is a continuing process where community learns how to respond to early warning, where to seek assistance and how to keep themselves protected during disaster period. The simulation exercise plan also highlights prevailing limitations regarding evacuations. One limitation is that construction of permanent or semi-permanent structures for the purpose of evacuation, like cyclone shelters is strictly prohibited at any part of the camp [21] . Refugees are obliged to use transitional shelters or other types of houses for the evacuation purpose which are not robust as like a permanent cyclone shelter.

Considering all the constraints stated above, the GoB decided to transfer about 100,000 Rohingyas to the island of Bhasan Char although there are many arguments for and against this proposition. The project with a construction cost of US$ 268.84 million has been implemented on Bhasan Char which includes facilities like concrete built houses, flood prevention systems, cyclone shelters, potable water and solar power systems, livelihood opportunities (fishing, agriculture, and raising livestock), hospital, police and fire station, and office premises for the partners [13] . However, about 90% of the respondents in this survey did not want to be relocated to Bhasan Char. The island is distinctively vulnerable in terms of its formation with silt, uneven ground with mangroves and flat land, frequent change of shoreline over the past 20 years, and extreme threats of cyclones, flooding and storm surges [13] . Some of these issues were raised by the respondents as a major cause to avoid relocation to Bhasan Char.

The respondents were also asked about how they aspire in the long-term for better future. The answer to this question expresses their ultimate desire for repatriation to Myanmar. As per the signatory of the Memorandum of Understanding (MoU) between UNHCR and Bangladesh on 13 April 2018, no Rohingya are allowed to return until conditions in Myanmar are conducive for them. Rohingya return to Myanmar has to be voluntary, safe, dignified and sustainable in line with international standards. Failure of the recent repatriation plan on 22 August 2019 suggests that still Rohingya are in fear of facing torture and violence in Myanmar [37] .

The overall disaster context in the Rohingya camps in CBD is dynamic and scenarios change quite frequently. The GoB and humanitarian agencies have undertaken a number of emergency preparedness initiatives (as updated on March 2020) in each camp -such as formation of a disaster management committee, identifying temporal communal shelters (existing learning centres) and emergency back-up distribution sites, reinforcing critical infrastructure, prepositioning adequate emergency shelter stock (tarpaulins, ropes and flood mats), providing emergency rapid food assistance, mapping infrastructure and services in high risk zones, protection messaging and awareness-raising, ensuring minimal disruption to critical services like health, water, sanitation, and hygiene (WASH), and protection, and continued training for temporary learning centre facilitators, teachers and students [38] .

The novel Coronavirus (COVID-19) pandemic is another major concern for the Rohingya refugees in CBD, as they do not have freedom of movement, are living in exceedingly overcrowded camps, and the existing health centres are not equipped with necessary testing and treatment facilities (e.g. intensive care unit, oxygen supply and ventilators, and personal protective equipment for health workers). The COVID-19 outbreak, in addition to natural disasters, would result in a severe humanitarian crisis. As instructed by the UN COVID-19 Global Humanitarian Response Plan [39] , the GoB, ISCG and partners are advancing with the construction of an isolation and treatment centres, reducing activities to essential services and assistances only, promoting hygiene activities, training healthcare workers, and ensuring social distancing inside the camps [40] . As of 8 April 2020, the entire CBD including the camps are now locked down, and no Rohingya is even allowed to move between two camps until further notice.

Unprecedented deforestation and indiscriminate hill cutting activities took place in CBD to build makeshift shelters and supply cooking materials (wood) for the Rohingya refugees. Because of it, nearly 5,800 hectares of forest land cover disappeared [41] that has also increased landslide disaster risk significantly. To address this matter, the partners are now providing liquid petroleum gas (LPG) cylinders and refilling them monthly. Almost 92% Rohingya households are now covered under this scheme for the purpose of cooking and household-level lighting [13] . The GoB is also considering to initiate the Myanmar national curriculum framework in 2020 to ensure more formal education for the Rohingya children and adolescents and facilitate their sustainable repatriation in Myanmar [13] .

Regarding the Rohingya relocation to Bhasan Char, the GoB is not insisting to implement the plan as it will be voluntary. The UN desires to assess the overall situation in Bhasan Char in terms of its safety and sustainability, exposure to natural hazards, freshwater availability and the standard of protection facilities. As of today, there is no further progress on the UN assessment and the decision of relocating the Rohingyas is halted [13] . In the meantime, on 23 March 2020, the GoB has instructed low-income Bangladeshi citizens, who are unable to maintain their livelihood in urban or rural areas, to take refuge or use the facilities in Bhasan Char amid the growing threats of Coronavirus (COVID-19) outbreak.

To withstand during the cyclone and monsoon season (May to October), immediate steps should include strengthening and reinforcing refugees' makeshift shelters and critical infrastructures. In the short-term, wider level and continuing education should be in place to promote good practices, skill development trainings and education on DRR.

This is a protection crisis, and the Rohingya refugees specially women and girls face extreme gender-based violence, many of them are subject to human trafficking, and the Rohingyas' top priorities are shelter materials, food, clean drinking water, fuel and electricity [13] . Therefore, in the long-run, it is necessary to safeguard them, continue life-saving assistance, promote peaceful co-existence with host communities, and achieve sustainable solutions in Myanmar to create an environment for their voluntary, dignified, and safe repatriation. Fig. 12 illustrates Rohingyas' physical, social, economic, environment, cultural and institutional dimensions of vulnerability to environmental hazards. The outcomes validates that disasters are not exclusively natural [42] . It is the (flawed) decision making process and blending of complex socio-economic and socio-cultural aspects [43] that are accountable for converting hazards into catastrophic disasters. The decision makers, camp in charge (CiC) and emergency managers should have indepth knowledge on the root-causes of disaster vulnerability for achieving a sustainable solution them. 

The study aims to find out how Rohingya are coping with extreme weather events, such as floods, heavy rains, landslides, and windstorms in the Kutupalong camp in Cox's Bazar District. The results show some preventive and impact minimising strategies developed and adopted by the Rohingyas to withstand disasters. These include practices in individual household (e.g., receiving multi-hazard early warning messages and acting accordingly, maintaining storage spaces above the ground to protect non-food items, storing medicine and dry foods before any impending disaster) to collective efforts (e.g., strengthening shelters by means of shelter improvement kits distributed by the humanitarian actors). Furthermore, slope stabilisation works such as placement of bamboos or sandbags on the hillslopes, construction of retaining walls and drainage, implementation of mechanised work to level steep hilltops, and plantation of vetiver grass are some examples undertaken to make the camps resilient to natural hazards. It is understood that Rohingyas have adopted several risk mitigation strategies to sustain their existence with the paradigm of shifting risks posed by natural hazards. Consequently, the study gets enough evidence to accept the alternative hypothesis of this study.

The study has three major contributions in the field of DRR. First, it provides a solid baseline on how a refugee community in a humanitarian context might differ compared to other grassroot communities in terms of coping with disasters. Second, it contributes in achieving some of the priorities outlined by the Sendai Framework for Disaster Risk Reduction (understanding disaster risk) [44] , and the UN Sustainable Developments Goals (ensuring education for all, leaving no one behind, promoting peaceful and inclusive societies, and making cities and human settlements resilient) [45] . Lastly, the findings and recommendations would support the Government of Bangladesh, United Nations and humanitarian stakeholders to identify some of the gaps in mitigating disaster risks from extreme natural hazards in the Rohingya camps in Bangladesh.

The authors have no competing interests to declare. More peaceful and secure than our habitats in Rakhine II. I receive aids from humanitarian agencies III. I receive food without doing any work IV. We are treated here with dignity; not as an illegal immigrant or danger V. All of the above VI. Others, please specify 22.2 If no, why? What were the benefits in Rakhine?

I. Property II. Access to formal schooling III. Access to health facilities IV. Employment V. It was our own land VI. All of the above VII. Others, please specify 23. What else could be provided to make things better for you?

Repatriation with proper citizenship of Myanmar II.

Freedom of movement in Bangladesh III.

If proper housing is provided, I will stay in Bangladesh IV.

Access to schooling in Bangladesh V.

Diversified livelihood opportunity VI.

More upgraded DRR training VII. All of the above VIII. Others, please specify

Climate change and disaster management

Response to dynamic flood hazard factors in peninsular Malaysia

Flooding, vulnerability and coping strategies: local responses to a global threat

Local disaster risk reduction in Latin American urban areas

The double bind of poverty and community disaster risk reduction: a case study from the Caribbean

Assessing risks from climate variability and change for disaster-prone zones in Bangladesh

Convention and protocol relating to the status of refugees. The United Nations High Commissioner for Refugees (UNHCR)

The Plight of the Stateless Rohingyas: Responses of the State, Society & the International Community

Global Trends: Forced Displacement

Displacement and Disaster Risk Reduction. The United Nations High Commissioner for Refugees (UNHCR)

Memories of Burmese Rohingya refugees: contested identity and belonging

The State, Vulnerability, and Transborder Movements: The Rohingya People in Myanmar and Bangladesh

Inter Sector Coordination Group (ISCG) -Bangladesh. United Nations Office for the Coordination of Humanitarian Affairs (OCHA)

Application of the Convention on the Prevention and Punishment of the Crime of Genocide (The Gambia v. Myanmar). International Court of Justice (ICJ), the Hague

Joint Government of Bangladesh -UNHCR Population map as of 15

Humanitarian disaster for Rohingya refugees: impending natural hazards and worsening public health crises. The Lancet Global Health

Thousands Displaced as Monsoon Rains Pound Rohingya Camps. International Organization for Migration (IOM)

Aid workers race to prepare Bangladesh's Rohingya refugee camps for monsoon wind, rain. International Organization for Migration (IOM)

Rohingya refugee crisis and forest cover change in Teknaf

Cyclone risk assessment of the Cox's Bazar district and Rohingya refugee camps in southeast Bangladesh

The International Organization for Migration -Bangladesh Mission

Bangladesh: UNHCR Camp Settlement and Protection Profiling -Round 6

Measuring community vulnerability to environmental hazards: a method for combining quantitative and qualitative data

IBM SPSS Statistics 25 Brief Guide

Pearson-Fisher chi-square statistic revisited

Test of two proportions using SPSS Statistics. Statistical tutorials and software guides

Chi-square test of homogeneity (2 x C) using SPSS Statistics. Statistical tutorials and software guides

Statistics review 14 : logistic regression

Binomial logistic regression using SPSS Statistics. Statistical tutorials and software guides

An introduction to logistic regression analysis and reporting

Rohingya refugees' perspectives on their displacement in Bangladesh: uncertain futures

Multi Hazard Risk Assessment in the Rakhine State of Myanmar

The Rohingya Exodus 2017: Issues and Implications for Stability, Security and Peace in South Asia. Institute for Risk and Disaster Reduction (IRDR)

Correlates of hazards education for youth: a replication study

Do lessons people learn determine disaster cognition and preparedness

Young people from refugee backgrounds as a resource for disaster risk reduction

UNHCR Statement on Voluntary Repatriation to Myanmar. The United Nations High Commissioner for Refugees (UNHCR)

Inter Sector Coordination Group (ISCG) -Bangladesh. United Nations Office for the Coordination of Humanitarian Affairs (OCHA)

Global Humanitarian Response Plan COVID-19. The United Nations Office for the Coordination of Humanitarian Affairs (OCHA)

COVID-19: Preparedness and response for the Rohingya refugee camps and host communities in Cox's Bazar District. Inter Sector Coordination Group (ISCG) -Bangladesh. United Nations Office for the Coordination of Humanitarian Affairs (OCHA)

Application of geospatial technologies in developing a dynamic landslide early warning system in a humanitarian context: the Rohingya refugee crisis in Cox's Bazar

Disaster by Choice: How our actions turn natural hazards into catastrophes

Framing vulnerability, risk and societal responses: the MOVE framework

Sendai Framework for Disaster Risk Reduction 2015-2030. The United Nations Office for Disaster Risk Reduction (UNISDR)

Sustainable Development Goals. The United Nations (UN)

Questionnaire for household survey in the Rohingya camps in Cox's Bazar, Bangladesh Survey date: Survey time: Survey location

Female III. Other, please specify 2. Specify your: Education (class) or illiterate? Where you live in Burma? (Village, District/state) Previous occupation (in Rakhine) Your avg. monthly income in Rakhine

When did you arrive in Bangladesh? SECTION B -ROHINGYA REFUGEES' PERCEPTION ON NATURAL HAZARDS AND DISASTER RISK REDUCTION (DRR)

If yes, what were the natural disasters to which you were exposed in the Rakhine state of Burma?

Chronological development of settlements (2016-2020) in the Bhasan Char Island in Bangladesh

We are grateful to the Ministry of Disaster Management and Relief (MoDMR), Ministry of Foreign Affairs (MOFA), Office of the Refugee Relief and Repatriation Commission (RRRC), University of Dhaka (DU), and the Cox's Bazar Development Authority (CoxDA) for allowing us to enter the Rohingya camps in Cox's Bazar and supporting us to conduct necessary fieldwork/surveys. We are also indebted to the respective Rohingya refugees and members of the host communities for assisting us on numerous occasions.

This research is funded by the British Academy as part of the project, "Rohingya Journeys of Violence and Resilience in Bangladesh and its Neighbours: Historical and Contemporary Perspectives" (Award Reference: SDP2\100094), supported under the UK Government's Global Challenges Research Fund (GCRF). The first author was a Commonwealth Scholar funded by the UK Government.