In situ adaptation against sea level rise (SLR) in Bangladesh: does awareness matter?

This paper is intended to find out the influence of climate awareness on people's preference for adaptation strategies against the impact of climate change‐sea level rise (CC‐SLR) in coastal Bangladesh. Climate change is viewed as a gradual change in long‐term average conditions, greater variability within the range of “normal conditions” and changes in the types of extreme events which are possible or probable (Hare, 1991). Such change could be due to natural variability or as a result of human activity (intergovernmental panel on climate change (IPCC), 1996, p. 56). Climate change is manifested through changes in global mean temperature, drought, pattern of precipitation, storm surge, winter storm, amount of melting of snow and ice and rising global (mean) sea level (Neren et al., 2006; Leal Filho and Mannke, 2009). Despite the uncertainty about the extent of such changes in future, current scientific understanding has established a clear link among global warming, climate change and SLR (Houghton, 2004).

The third assessment report of the IPCC has projected a SLR of as high as 88 cm by the end of this century (IPCC: WG I, 2001). This has posed a formidable challenge for low‐lying deltaic coasts (Warner et al., 2009). In 22 coast‐lying countries 170 million people would be affected due to CC‐SLR (Middleton, 1999). The most severely affected countries will be Bangladesh, Kiribati, Marshall Islands, Antigua and Benin (Nicholls, 1995; Nicholls et al., 1995). Drawing from the study of Ahmed and Alam (1998) and (The) World Bank (2000, p. 11) has shown a projected (high‐end scenario) SLR of 30‐40 and 50 cm for the year 2030 and 2050, respectively, along Bangladesh's coast. For every year average rise is 1 cm. However, projected SLR of 12, 32 and 88 cm along the Bangladesh's coast for the year 2030, 2050 and 2100, respectively, are shown in the National Adaptation Program of Action for Bangladesh (GOB, 2005, 2006, p. 6). These figures are the basis of storyline used later of this paper.

Following the fourth assessment report of the IPCC (IPCC: WG II, 2007), CC‐SLR has been considered to be the most pressing problem against which the people of Bangladesh in general and the coastal people in particular have to adapt in this century (GOB, 2008). Adaptation involves some processes and actions in order to better cope with, manage or adjust to some changing conditions, stresses, hazards, risks or opportunities (Smith, 1997; Fankhauser et al., 1999; Smit and Pilifosova, 2001; Smit and Skinner, 2002, Brooks et al., 2005; Smit and Wandel, 2006). In a climate change context, IPCC viewed adaptation as an adjustment in natural or human systems in response to actual or expected climate stimuli or their effects, which moderate harm or exploit beneficial opportunities (McCarthy et al., 2001, p. 982). For adaptation against the impacts of SLR there are only three options: protection, accommodation and retreat (evacuation) (Klein et al., 2001). Experts already warn that for a land scarce country like Bangladesh, which may even lose one‐third of her landmass due to SLR, evacuation may not be a sustainable solution. Rather, protection and accommodation are considered to be suitable options as these do not require displacement (Ortiz, 1994). These two responses are basically in situ (in place) adaptation which often do not take place automatically.

In situ adaptation by and large depends on three important elements: timely recognition of the need to adapt, an incentive to adapt and ability to adapt (Fankhauser et al., 1999; Adger et al., 2005). Unless the people are aware about the dynamics of CC‐SLR they cannot comprehend the need for adaptation. Again their adaptation is not only contingent on their understanding of need to adapt but also depend on their abilities to adapt. When the question of ability to adapt arises, the conventional literature points a finger to the physical or other capital that a person posseses (Blaikie et al., 1994, Smithers and Smith, 1997, Pelling and High, 2005). Psycho‐social and behavioural aspects have also recently emerged as important dimensions in this regard (Patt and Gwata, 2002; Grothmann and Patt, 2005; Grothmann and Reusswig, 2006; Oppenheimer and Todorov, 2006). Some scholars even put higher emphasis on people's access to climate information and communication as means to enhance their climate knowledge, which in turn will help them to adapt (Phillips, 2003; Leal Filho, 2009; Leal Filho and Mannke, 2009; Nerlich et al., 2010). Climate awareness has therefore, emerged as an important sub‐dimension of adaptation research.

Despite availability of the literature on adaptation in general, literature concerning the influence of climate awareness on people's preference for adaptation strategy is very scanty, if not absent. Indeed, only a few qualitative attempts are observed; focused primarily in the policy arena of climate communication and relevant for developed countries (Dela Vega‐Leinert and Nichollls, 2008; Kont et al., 2008; Sterr, 2008; Tol et al., 2008; Leal Filho, 2009; Moser and Satterthwaite, 2009). In this respect, notable exceptions are Weber (1997), Grothmann and Patt (2005), Grothmann and Reusswig (2006), Semenza et al. (2008) and Blennow and Persson (2009) who have used quantitative approaches. Despite having very good explanatory power while dealing with climatic issues, the common limitation is neither of them focused exclusively on SLR.

Few works have been done so far on issues relating to adaptation in Bangladesh; most notable of which are “reduction of vulnerability to climate change” (RVCC, 2003; Schaerer and Ahmed, 2004; Thomalla et al., 2005). These are centred on qualitative inquiry. In fact, no study is observed to assess quantitatively the explanatory power of various factors, especially explanatory power of climate awareness to explain the variances in coastal people's preference for adaptation strategies against the impact of SLR on their livelihoods. The goal, here, is to assess quantitatively the influences of various dimensions of climate awareness on people's preference for adaptation strategies against the impact of future SLR. In fact, livelihood securities of coastal people in Bangladesh are in constant threat of CC‐SLR. The policy implication of the findings is these will help policy makers and planners to initiate programmes for encouraging in situ adaptation against the SLR.

Earlier works cited in the previous section have confirmed that various socio‐demographic, economic, cultural, information communication and contextual (spatial) factors influence local people's preferences for adaptation options against the impacts of climate change. The hypothesis of this research is that various dimensions of climate awareness are important determinants of coastal people's preferences for adaptation choices against the impact of future SLR.

Among the coast‐lying areas of Bangladesh, “Kalapara Upazila” (sub‐district) of Patuakhali District is worst affected by storm, surge, salinity intrusion, tidal flood and may continue to experience the same given the reality of SLR (Ortiz, 1994; Ali and Chowdhury, 1997; Huq et al., 1998; Ali, 1999; Ali Khan et al., 2000; (The) World Bank, 2000; Singh et al., 2001). Three coastal villages from three “Union Parishad (UP)” (lowest tier of local government unit) of “Kalapara Upazila” were selected. Study villages of “Dulasar”, “Mithaganj” and “Nilganj” UP are 5‐15 km from the Bay of Bengal and 10‐15 to 30‐40 cm (about one foot contour) above mean sea level. The whole area is criss‐crossed by numerous natural cannels and creeks having link with three river systems and the Bay of Bengal. People living here are, historically, prone to various natural calamities. The area may experience 10‐15 to 30‐40 cm inundation within the time frame of 2020‐2080 (Saroar and Routray, in press). Randomly selected 285 respondents from various occupation categories, age groups and gender composition (175 male and 110 female) were interviewed using semi‐structured questionnaires during January‐April 2009 Figure 1.

The word “awareness” has different connotations. In the disaster literature awareness means increasing levels of consciousness about risks and reduction of exposure to hazards (ISDR, 2004, p. 6). In the context of CC‐SLR, people having some understandings about the changing climatic events, plausible rise of sea level, and probable impacts of CC‐SLR are considered as aware about climate change. Awareness about climate change is very fundamental for a person to accurately appraise the level of future threat without which one cannot advance with rational choice of adaptation (Grothmann and Patt, 2005, p. 203). Thus, climate awareness is considered as an indispensable component while dealing with the issues of adaptation against the SLR.

However, the literature on climate awareness is still in its infancy stage. Concrete and agreed guidance as to how climate awareness can be assessed is yet to be developed, although a range of indicators have been identified that are assumed to be useful. Weber (1997) and Semenza et al. (2008), for instance, use “familiarity with climate change/weather extreme phenomena” as an indicator of awareness about climate change. Likewise, Grothmann and Patt (2005), Grothmann and Reusswig (2006) and Blennow and Persson (2009), use “perception/belief” about climate change‐related event as an indicator of climate awareness. While others, for example, Vedwan and Rhoades (2001) and Steel et al. (2005), use ideas very similar to “tacit/intuitive knowledge” as an indicator of climate awareness. This is because these indicators are in some ways related with determinants of many dimensions of climate awareness, such as familiarity with weather/climate extreme, belief/perception about climatic event and understanding about the impact of accelerated SLR. In this analysis, climate awareness is operationalized as a bundle of “familiarity with weather/climate extreme signal”, “perception/belief about climatic event” and “tacit/intuitive knowledge about the impact of SLR”. In the following section, specific indicators of the above three dimensions of climate awareness are operationalized.

Climate change/weather extreme phenomena are often felt by people living in any area for long time due to their interaction with local environments. In fact, their familiarities are often associated with particularly important outcomes. For example, a farmer gets used to planting a particular crop during the monsoon can easily notice if the rainy days are deviated from their traditional mental map of seasonal calendar. The way traditional mental map of seasonal calendar is used by apple grower to adjust with climatic variation in the “Kullu” valley in Himalayan India is well illustrated by Vedwan and Rhoades (2001). Of course, some of the signals of climatic variation and weather extreme phenomena are intrinsically noticeable; for example, drastic reduction of migratory birds during recent winters from a decade or so ago in tropical climates.

Keeping in mind the mental mapping technique (Vedwan and Rhoades, 2001), a set of ten questions was prepared to assess the “familiarity with weather/climate extreme signal” of the respondents (Table I). Inquiry was centred on climate‐related aspects that have some degree of “visual silence”. Visual silence is the intrinsic property of the phenomenon which allows the phenomenon to be felt/more observable (Vedwan and Rhoades, 2001). However, IPCC's report (IPCC: WG II, 2001, p. 15), Adger et al. (2003, pp. 182‐3), Dessler and Parson (2006) and Neren et al. (2006) have been appeared as very instrumental while framing the indicators/questions. Each response was originally rated in a four‐point scale: “no idea/can not remember if heard about”, “heard from others but do not agree”, “heard from others and agree” and “observed/felt by own”. However, for this paper the central two original responses, i.e. “heard from others but do not agree” and “heard from others and agree” are collapsed into one “heard from others” to maintain a three‐point uniform scale across three indices of climate awareness. For each of the respondents average “climate familiarity score” is computed to develop a “familiarity index”. The process of development of “index” is discussed later of this paper (Table II).

Perception is seen as enduring and transposable disposition of an individual which functions as the generative basis of structured, objectively unified practice (Bourdieu, 1977; Gregory, 1999). For building climate awareness, prevailing perception about climate change can be a determinant factor. However, constant flow of information, recurrent exposure to some unusual realities and in some cases value forming psycho‐social elements can change the prevailing wrong perception (O'Riordan and Jordan, 1999, p. 88). Hence, knowing people's perception about CC‐SLR is just the beginning of any intervention. Following Blennow and Persson (2009, p. 101), a set of five questions related with obvious outcomes of CC‐SLR (Wilbanks et al., 2007) were prepared for knowing their perception about CC‐SLR. Original five‐point response (analogous to Likert scale) is recoded for this paper into a three‐point scale by collapsing the terminal two responses in each side of the mid value (Table I). For each of the respondents, average “climate perception score” is computed to develop a “perception index” later of this paper (Table II).

Translating climate impacts into non‐specialist (lay people) public knowledge is essential to let people comprehend it and participate in/initiate appropriate adaptive responses (Delli Carpini and Keeter, 1996; Leal Filho, 2009). Since the respondents were not too familiar with the term “sea level rise”, the essence of SLR was communicated with them using practical means/example instead of using any abstract term. They were asked to identify ten possible impacts of SLR if farmlands next to their homesteads submerge – up to 30‐40 cm (almost knee height) below saline water or if height of storm‐surges increase a few meters than what they have experienced in the past. Following Steel et al. (2005) survey format was prepared enlisting major ten categories of impacts of SLR; mostly drawn from Smith (1997, p. 252), Houghton (2004), Choudhury et al. (2005), Wilbanks et al. (2007, pp. 216‐18) and Tol et al. (2008, pp. 438‐9). For each of the respondents average “tacit/intuitive (SLR impacts) knowledge score” is computed to develop an “impact knowledge index” later of this paper (Table II).

Reliability (internal consistency) of responses for questions concerning three dimensions of climate awareness were tested which are fairly above acceptance limit of 0.70 measured by Cronbach alpha (α) (George and Mallery, 2006) (Table II). Respondent's weighted mean score (index) for each of the three dimensions of climate awareness was separately computed using the formula ∑Wi/n (Wi = individual's weighted score (either of 0.33, 0.67 and 1.0) for each question, n = number of question). These three climate awareness indices along with other variables selected from factor analysis have been used later in the multinomial logistic regression (MLR) model to predict the variances in the respondents' preferences for choices of adaptation against three plausible scenarios of future SLR. From individual's weighted mean score (index), by using the formula ∑Wifi/∑fi (where Wi = individual's weighted score for each question, fi = frequency of that particular score) weighted average mean index for entire sample is prepared for each of the three dimensions of climate awareness for a comparison.

Respondents' familiarities with weather/climate extreme signals are higher than their perceptions about events associated with CC‐SLR and the impacts of SLR (Table II). On an average people are aware about all three dimensions of climate awareness. However, higher standard deviation (SD) in “familiarity index” implies that knowledge gap is higher among the respondents – meaning some people have had more familiarity with weather/climate extreme signals than many others.

The respondents were asked to identify the key sources of future SLR vulnerability that might force the local people to evacuate their present localities given that “no additional measures” apart from their own would be taken. Multiple response analysis shows that the dominant cause of displacement would be loss of income from current sources (35.8 per cent), followed by permanent disruption in access to various services (e.g. local health care, market place and schooling) (35.1 per cent), loss of free/low cost access to potable drinking water (30.4 per cent) and loss of food security (26.4 per cent) (Table III). It is a bit strange, however, to see that more people may evacuate due to perceived lack of availability of potable water than lack of production and availability of food stuffs.

Three story lines that tell likely scenarios of CC‐SLR for the year 2020‐2030, 2050‐2075 and 2080‐2100 were presented to the respondents and asked for their responses on preference for adaptation option (Table IV). They were asked what they or their descendents probably would do if farmlands next to their homesteads go – “almost ankle height (10‐15 cm)”, “half‐knee height (20‐25 cm)” and “knee height (30‐40 cm)” below saline water permanently by the year 2020‐2030, 2050‐2075 and 2080‐2100, respectively? Altogether eight adaptation pathways were identified. These are: raising homestead and continue with same occupation anyway; raising homestead and continue agriculture with saline and flood tolerant varieties; raising homestead and use of saline affected land for aquaculture; commuting for non‐farm occupation; evacuating and settle nearby safer localities; evacuating and take refuge nearby urban centres; evacuating the locality without predefined destination.

To make the preference for adaptation choices more distinct and meaningful similar adaptation pathways were consolidated and have been presented in three major distinct preference categories. These are in situ adaptation with same occupation, in situ adaptation with changed occupation, and evacuation/displacement (Table IV). In the immediate term (2020‐2030), no evacuation is expected. However, almost 30 and 60 per cent respondents believe that they or their offspring might prefer evacuation for the medium term (2050‐2075) and distant term (2080‐2100) scenarios of SLR.

Currently almost 17 million people inhabit in the coastal low lying areas and by 2050, 130 million more people will be added in Bangladesh many of whom will be in that fragile coastal tract (GOB, 2008). How this huge number of people will be relocated if two‐thirds of them eventually evacuate the coast‐lying areas due to SLR is of very critical concern. This massive relocation could only be avoided with careful in situ adaptation planning for which there is clear need for information about the drivers of in situ adaptation. In the following section the driving force of in situ adaptation especially the role of climate awareness is quantitatively assessed.

Initially, altogether 21 factors/variables which include various physical capitals/resources (identified by Blaikie et al. (1994), Smithers and Smith (1997), Kelly and Adger (2000) and Adger et al. (2003, 2005)), social capitals/resources (identified by Pelling and High (2005), Smit and Wandel (2006) and Fussel (2007)), information resources (identified by Phillips (2003) and Leal Filho (2009)) and psycho‐social/behavioural elements (identified by Patt and Gwata (2002), Grothmann and Patt (2005) and Grothmann and Reusswig (2006)) were entered in principal component analysis (PCA) for identifying major components. Among them 13 factors/variables were dummy coded following Hardy and Bryman (2002) and the remaining eight variables are measured in their respective SI units. However, as the factor/variable “frequent adaptation against storm/surge (dummy)” is a constant it is excluded and finally 20 factors/variables were loaded in PCA.

In this PCA as the value of determinant of correlation matrix was found >0 (i.e. 4.73E‐05), the Kaiser‐Meyer‐Olkin value for sampling adequacy was 0.57, and the Bartlett's test of sphericity was significant at p<0.001 the factor analysis is considered statistically valid (Field, 2005). Furthermore, a total of 20 variables for a sample size of 285 meet the requirement for factor analysis (i.e. 5:1 case/variable ration as recommended by Coakes and Steed (2001) as well. Finally, total eight components having Eigenvalue >1 were extracted using varimax rotation with Kaiser normalization to maximize intra‐component variances as suggested by Tabachnick and Fedell (1996). These eight components have explained almost 72 per cent of the variances which is higher than the threshold recommended by Hair et al. (2006). Component‐wise load factors (load factor <0.200 is not shown) of variables are presented in Table V.

The first component could be termed as “attachment with coastal environment”, which constitutes three variables that explains 13.89 per cent of the variances. The second component could be characterized as “wealth and social standing” which constitutes four variables that explains 13.51 per cent of the variances. The third component related with “social networking” constitutes two variables and explains 9.57 per cent of the variances. The fourth component has two variables characterized with “access to print media for flood information” and explains 7.89 per cent of the variances. Likewise, the fifth component is “coping and adaptation with recurrent hazard” which includes three variables and explains 7.70 per cent of the variances. The sixth component has two variables characterized with “spatial and demographic causes of exposure to climatic hazard” and explains 7.47 per cent of the variances. The seventh component is related to “exposure potential of dry spell due to types of occupation” which includes includes variables and explains 6.42 per cent of the variances. The last component is characterized as “gender difference in electronic media use for climate information” which constitutes two variables that explains 5.61 per cent of the variances. While PCA identified the components and factors having influence on respondents' adaptation choices, however, it does not quantified the extent of influence of each relevant variable/factor on respondents' adaptation preferences; a MLR model was used later to identify the extent of such influence.

After reclassification of adaptation pathways, the dependent variable “respondent's adaptation choice” has left three options. There are “in situ adaptation with same occupation”, “in situ adaptation with changed occupation” and “evacuation/displacement” (Table IV). It does not reflect any precise ordering hence MLR analysis is preferred over ordered logistic regression (Field, 2005). In the MLR model, in addition to three climate awareness indices such as: “familiarity index”, “perception index”, and “knowledge index”, all the 20 variables drawn from the PCA (Table V) were planned to use as independent/predictor variables. However, bivariate correlation analysis has identified strong colinearity between “total yearly income (Bangladeshi Taka (BDT))” and “total farmland (ha)” (r=0.91), “duration of living (yr)” and “age of respondent (yr) (r=0.81) and “membership status of any entity (dummy)” and “personal contact with official (dummy)” (r=0.73) (Table VI). Following Hair et al. (2006) the variable “total yearly income (BDT)”, “duration of living (yr)” and “membership status of any entity (dummy)” have been excluded from the model because of their strong colinearity (r>0.70). Thus, total PCA variable turned out to be 17.

Furthermore, in order to correctly fit with the MLR model, all the 12 dummies (independent variables) from these 17 variables were separately cross‐tabulated with the dependent variable “respondent's adaptation choice” to eliminate dummy variables (binary coded) having any empty cell or 0 frequency. Thus, after dropping “possession of television (dummy)” and “habit of reading newspaper (dummy)” the total explanatory variables became 18 (including three climate awareness indices). Finally, these 18 variables were entered into the MLR model. Here, the reference category is “evacuation”, which was compared with the other two choices, i.e. “in situ adaptation with same occupation” (Table VII), and “in situ adaptation with changed occupation” (Table VIII). As no incidence of evacuation was observed for immediate scenario (2020‐2030), the MLR model was used for only medium (2050‐2075) and distant term (2080‐2100) scenarios of CC‐SLR. After running the MLR model factors that significantly explain the variations in respondents' preferences for in situ adaptation (with same occupation or with changed occupation) over evacuation were identified. Particular stress was given towards examining the extent of influence of climate awareness on adaptation choice.

For the SLR scenario of 2050‐2075 (20‐25 cm inundation) occupational engagement, use of radio as source of weather/climate information, exposure to rainfall, perception about CC‐SLR appeared as the most significant predictors of intension for evacuation or in situ adaptation with same occupation (LR χ²=183.38, pseudo R²=0.54, p<0.001). Similarly, for the SLR scenario of 2080‐2100 (30‐40 cm inundation) in addition to the above cited four factors an additional four factors such as educational attainment, exposure to recurrent flood, dry spell, and climate familiarity, emerged as the most significant predictors of respondents' preferences (LR χ²=202.08, pseudo R²=0.60, p<0.001) (Table VII). However, for neither of the scenarios of SLR some common socio‐demographic, economic and cultural factors such as age, sex, social network and size of farmland holding have any significant influence on the respondents' preferences for adaptation or evacuation.

It is a bit surprising that contrary to general expectation even knowledge differences about SLR's impacts do not have significant influence on the respondents' preference for adaptive response. Similarly, contrary to expectation, no significant differences were observed in adaptation preferences for respondents living on the coast (within 5 km) and off the coast (within 10‐15 away from the coast (Table VII). The reason could be attributed to the fact that everyone cares much about cyclone and surges for which almost no differences in impacts are usually felt for even 50 km inward from the coast because the whole coastal tract is nearly flat.

People engaged in agriculture and allied occupations are more likely than others to prefer in situ adaptation over evacuation for both the scenarios of SLR. This occupational category prefers in situ adaptation primarily because this is the only skill that each individual possesses. They believe any evacuation may force them into a more challenging situation while struggling for their future security of livelihood. However, there is greater prospect for relocating people currently engaged in non‐agricultural professions. Failure to relocate this category of people may leave millions of souls eventually as climate refugees.

People having regular access to radio as sources of climate/weather information are more likely than those who do not have such access to prefer in situ adaptation over evacuation for both the scenarios of SLR. It is probably because weather information and warning disseminated via radio were credible enough to the listeners to take precaution to avoid or minimize loss in the event of climate related natural disasters. This finding supports the findings of other studies where the role of radio, especially community radio to disseminate the warning information before the onset of natural disasters are highly hailed (Steel et al., 2005; Kurita et al., 2006; Perry, 2007; Collins and Kapucu, 2008; Cretikos et al., 2008). This finding is substantive for policy makers because by increasing credible radio programmes on weather, climate and SLR it might be possible to build awareness for in situ adaptation among the people who have been living in the coastal regions for years.

People having experience of recurrent exposure to intense and prolong rainfall are more likely than those who do not have such experience to prefer in situ adaptation over evacuation for both the scenarios of SLR. The key reason could be the fact that this group of people believe that their indigenous coping and adaptation strategies, which they have developed over years, would be adequate to face the future challenge as well. It is hard to say whether they are overestimating their ability, which is often the case. In psychology this syndrome is termed as “illusion of control” (Wortman, 1976). However, any overestimation of current coping capacity may lead some people of this category to prefer evacuation eventually when they face actual situations of vulnerability in the future.

Respondents having better/accurate perception about CC‐SLR are more likely than others who do not have such perceptions to prefer in situ adaptation with same occupation over evacuation. This finding confirms with other studies which conclude that people having accurate/better perception about a problem are more rational while devising ways to counteract with the problem (Grothmann and Patt, 2005; Grothmann and Reusswig, 2006; Blennow and Persson, 2009). It implies that there is greater need for building accurate perception about the dynamics of CC‐SLR among the coastal inhabitants to encourage in situ adaptation. Appropriate information about the CC‐SLR may help them leave out any misconception that they might have derived from rumour concerning possible lose of settlement altogether due to accelerated SLR. This finding confirms with the theory of climate communication (Leal Filho, 2009; Nerlich et al., 2010) and is substantive for policy makers while designing programme components for raising climate awareness.

Even if most respondents do not care much about 20‐25 cm inundation by 2050‐2075, many of them are serious about the possible inundation of 30‐40 cm by 2080‐2100. For example, respondents with higher educational attainment than low achievers believe they or their descendents are more likely to prefer evacuation over in situ adaptation for the distant scenario (2080‐2100) of SLR (Table VII). It implies that substantial relocation of educated people is and needs to be handled with careful planning. This finding may be unexpected for a land‐scarce country like Bangladesh, but it is not unusual. This finding confirms very fundamentals of modern migration where it is theorized that educated people can more afford the risk of displacement than less/uneducated one. Here, future SLR might act as a push factor to force better educated groups to move out from the naturally vulnerable coastal tracts. Since, a growing body of migration literature confirms that educated people quit their place of origin for their developmental needs, it is very usual that they will evacuate the fragile coast for their survival needs as well, and resettle elsewhere.

Respondents having more experience of recurrent exposures to flood and seasonal dry spells are more likely than those who have less exposure of such kind to prefer evacuation over in situ adaptation with same occupation. Apparently, the finding is contrary to the general belief that people with more experience of coping are more likely to continue to better cope in future as well, and thus prefer adaptation over evacuation. However, in reality this may not be the case because of the limit of coping capacity, i.e. these people are already in their upper end of coping threshold. Any additional exposure may lead mass displacement due to their fear that the future recurrent hazard must exceed their coping ranges which confirms with the findings of Blaikie et al. (1994). Teversky and Kahneman (2008) and Crocker (1981) typified this fear as “availability heuristic” and is much like the way people with higher climate familiarity are more likely to prefer evacuation over in situ adaptation (Table VII).

The above finding does not confirm earlier findings, for example, Weber (1997) and Semenza et al. (2008). This contrasting finding is unusual but not altogether unexpected. Higher evacuation intention with increased familiarity of climatic events may be attributed to the possibility that such familiarity about CC‐SLR might have frightened the respondents and created panic about the possible impacts of SLR rather than making them aware about these. This finding has to be evaluated very carefully with further exploratory research because it signifies that if peoples are partially aware about the possible rises of sea level such awareness may even be counter productive‐ resulting in mass displacement. As a remedy, while designing awareness raising programme about CC‐SLR, programme components must be put in a way that these will not create panic among the coastal people. Rather these should build awareness through altering wrong notions and experiences, and encouraging them to adapt with the possible effects of CC‐SLR. Finally, it should be noted as well that some people having more familiarity with climate change/weather extreme still may prefer evacuation over in situ adaptation if those people are more cautious about their life and livelihood, and always bear risk aversion attitude.

For the SLR scenario of 2050‐2075 (20‐25 cm inundation) people engaged in agriculture and allied occupations are 7.68 times more likely than people of non‐agricultural occupations to prefer in situ adaptation with changed occupation over evacuation. However, for the SLR scenario of 2080‐2100 (30‐40 cm inundation) occupational engagement does not have any significant influence on peoples' preferences. Similarly for the SLR scenario of 2050‐2075 (20‐25 cm inundation) people who use radio as source of climate/weather information are 9.18 times more likely than who does not use radio to prefer in situ adaptation over evacuation. However, for the SLR scenario of 2080‐2100 (30‐40 cm inundation), preference for in situ adaptation with changed occupation over evacuation of the regular listener of radio is only 3.15 times higher than who does not listen regularly (Table VIII).

The above finding gives two clear messages. First, with the increased amount of SLR people will be more likely to prefer evacuation over in situ adaptation in general. Second, beyond certain level of SLR adaptation may not work – meaning adaptation has a limit. This finite nature of adaptive capacity confirms with the findings of Adger and Barnett (2009), Adger et al. (2009) and Barnett and Adger (2010) where they have given clear indication that after a certain levels people cannot adapt any more, thus the effort for innovative mitigation measures must continue.

However, for both the scenarios of SLR people who have the experiences of recurrent exposure to salinity are more likely than those who do not have such experiences to prefer in situ adaptation with changed occupation over evacuation. It is a stark contrast against the general belief that with increased salinity the tendency for evacuation will increase. But in reality people may prefer changed occupation over evacuation. It is primarily because over the years many people have adapted with salinity by investing in brackish water shrimp production. They believe, in the event of gradual SLR they would be able to harness the positive benefit out of it. Because, as of now, they have to spend quite a lot of money to bring saline water from the rivers/canals to their shrimp “gheers” (farms), however, for gradual SLR they may not need to spend additional money for it. This finding is also substantive; it indicates that the more the number of people in the coastal areas engaged in brackish water aquaculture the more their chances to follow in situ adaptation. Despite this general finding it is equally true that over times their preferences for evacuation may increase because for 20‐25 cm inundation (by 2050‐2075), preference for in situ adaptation over evacuation is 7.23 times higher but for the 30‐40 cm inundation (by 2080‐2100), preference for in situ adaptation over evacuation is only 4.88 times higher. Here, again the question of limit to adaptation comes (Adger and Barnett, 2009; Adger et al., 2009; Barnett and Adger, 2010).

Accurate estimation of the extent of possible impacts of CC‐SLR along Bangladesh coast is yet to be confirmed. Nonetheless, peoples in the coastal areas need to be prepared for anticipatory adaptation to avoid mass displacement. Because, Bangladesh may not be able to afford the cost of relocation of millions of families in the mainland which is already highly densely populated (GOB, 2008). For coastal communities to adapt against CC‐SLR there is need for full understanding about the various dynamics of climatic events and SLR. This study examined the local people's relative levels of familiarity with, perception about, and intuitive knowledge about possible impacts of CC‐SLR. Further, it examined how different dimensions of climate awareness such as familiarity, perception, knowledge about CC‐SLR and various socio‐demographic, economic, adaptive behavioural and climate information related factors influence the coastal people's preference for in situ adaptation over evacuation given the plausible scenarios of future SLR.

The finding has confirmed that while educational attainment, occupational engagement, use of radio for climate information, past exposure against intense rainfall, flood, salinity intrusion, climate familiarity and perception about CC‐SLR have significant influences on the peoples' likely preference for adaptation choice, many common socio‐demographic, economic, cultural factors such as age, sex, social network and farmland holdings do not have any significant influence. A most striking feature is that distance of settlement from the coast does not significantly influence the coastal inhabitants in this regard. Two dimensions of climate awareness such as familiarity with climatic events and perception about CC‐SLR have significant influences on peoples' choice of adaptation strategies. However, no difference is observed in terms of adaptation preference among respondents having higher intuitive knowledge and lower intuitive knowledge about the impacts of SLR.

The above findings have very strong policy implications. These can be summarized as follows:

• First, initiative must be taken to explore ways to retain educated and non‐agricultural professional group to avoid mass out‐migration. Significant out‐migration of better educated people may instigate others to evacuate as well. It is primarily because in the socio‐cultural context of rural Bangladesh the educated people, especially if they are elder as well, are highly regarded and others often follow them.

• Second, since the people who use radio as a source of climate/weather information prefer in situ adaptation over evacuation, policy makers may put more attention to the establishment of community radio and patronise the radio operating agencies for more information and awareness campaign related to adaptation with CC‐SLR. This initiative may prove very effective because coastal people, a significant portion of whom have been engaged in offshore and marine fishing for generations, still consider weather information and warning disseminated through radio as credible and make plans accordingly to avoid/minimize damages from inevitable natural disasters.

• Third, some people who have adapted against salinity intrusion in the past by heavily investing in brackish water aquaculture are very likely to continue this activity as long as it is feasible for them. Policy makers could explore the ways and means to support this group without undermining the livelihood struggle of other occupational groups especially rice growers.

• Fourth, campaigns targeting coastal people may be initiated to build positive perception and increase awareness about in situ adaptation against the impacts of CC‐SLR. It is primarily because people with better/accurate perception about CC‐SLR have shown their intention for in situ adaptation. Since transfer of (climate) knowledge is intergenerational, future generations will also benefit from current awareness campaigns (Gronhoj and Thogersen, 2009).

Further, it is equally important to keep in mind that people with more climate familiarity have shown their intention for evacuation which is a stark contrast with the general expectation. Policy makers and researchers alike must concentrate more on this. A concluding remark in this regard is that such familiarity probably has frightened them about the impact of CC‐SLR rather than making them aware about these natural processes. Carefully designed intervention needs to be initiated without further delay to let them acquire authentic knowledge about the adaptation against CC‐SLR from credible institutions. This may help them leave out any wrong conception and encourage them to think about in situ adaptation. However, the need for further exploratory research to see with deeper insight why this group of people are willing to evacuate contrary to general expectation of in situ adaptation is not ruled out. In this regard, the key research question for future research could be: whether the intention for evacuation of people having more familiarity with climate change/weather extreme is due to their past bad experiences from natural calamities.

The authors acknowledge the research grant provided by the Norwegian Ministry of Foreign Affairs as part of Scholarship for doing Doctor of Philosophy (PhD) of the first author at the Asian Institute of Technology in Thailand, supervised by the second author. The authors would like to thank anonymous reviewers and the editor for reviewing this paper and giving constructive comments. The views expressed here are the authors' alone.

Md Mustafa Saroar is currently finishing his PhD at the Asian Institute of Technology (AIT), in Bangkok, Thailand. He completed his Masters in Regional Development Planning at the Technical University of Dortmund, Germany. Saroar is an Associate Professor of Urban and Rural Planning Disciple, at Khulna University, Bangladesh. His research interest includes climate adaptation, wetland management, disaster management and urban environmental management. Md Mustafa Saroar is the corresponding author and can be contacted at: st105227@ait.ac.th; saroar.mustafa@yahoo.com

Jayant K. Routray is currently a Professor of Regional and Rural Development Planning, and Coordinator of the Disaster Preparedness, Mitigation and Management (Interdisciplinary Academic Program) at AIT, Bangkok. He has academic background in Geography and Regional Planning and been engaged in teaching, research and outreach activities over last 30 years with varied interests in Rural‐Regional Development Planning, Disaster Risk Reduction and Management, Climate Change Induced Adaptation and Sustainable Development.