2.1 Study area description
Most of the population of Bangladesh lives in flood plains with varying degrees of river flooding every year (Ferdous et al. 2019). Jamuna river floodplain, Bangladesh, was taken as a study area for this study. The Jamuna floodplain covers a vast area, including Gaibandha, Jamalpur and Sirajgonj districts. Jamalpur district is situated in the northern part of Bangladesh, taken as a case study. The district's general area is 2115.12 square kilometres, with 18.16 square kilometres of forest. Between 24°34 and 25°26 north latitudes, and 89°40 and 90°12 east longitudes, the district is located (BBS, 2011). The climate in Jamalpur is warm and temperate. There is significantly less rainfall in the winter than there is in the summer. This climate is classified as Cwa, according to Köppen and Geiger. The annual average temperature in Jamalpur is 26.0°C degrees Celsius. This area receives an average of 1963 millimetres of rainfall every year. This district's average yearly temperature ranges from a maximum of 33.3°C to a minimum of 12°C (BBS, 2011).
Comparatively, Jamalpur district is a warmer district than the others. The prominent rivers in this district include the Jamuna, Brahmaputra, Jhenai, Banar, Jirjira, and Chhatal (BBS, 2011). Nearly every monsoon carries river floods in this area. Five Upazilas, including Dewanganj, Islampur, Madargonj, Melandaha and Sharishabari, are comparatively low areas and are the worst sufferers of annual floods. That’s why these five Upazila have been selected for primary data collection. Jamalpur district have a population of 2292674 people with a population density of 1084 people per square kilometre, according to the 2011 census. The literacy rate is 38.4%, and the average household size is 4.06. The economy of Jamalpur district is primarily dependent on agriculture, with 62% of the population dependent on agriculture. Besides, most people live here by fishing. That’s why people of this area have to live and depend near the river (BBS, 2011).
2.2 Sampling strategy and data collection
Firstly, Jamalpur district has been selected from the Jamuna floodplain based on previous flood records. Five Upazilas, including Dewanganj, Islampur, Melandaha, Madarganj and Sharishabari, have been chosen as surveyed areas due to extensive flood damage in these areas. The study adopted quantitative techniques to collect primary data from the field. A pretested questionnaire has been prepared, and 400 households have been surveyed at the selected Upazilas of Jamalpur district in 2019. The household that directly affected by the flood considered target units for the household level survey. The Unions were considered clusters in the study, which uses cluster sampling techniques to identify the units of observations. The sample size was calculated using the following formulae based on a 20% indicator percentage (proportion of households affected by flood during the last occurrence in Jamalpur), a 95% confidence interval, 5% precision, and the highest response distribution with an assumed design effect of 1.5.
Here, P = the indicator percentage (0.2),
Z = the value of normal variants with 95% confidence interval (1.96),
d = the relative error margin,
Deff = the design effect (1.5).
According to the formulae, the minimum sample size is 369; that means the study must include at least 369 households. However, for equal distribution of households within the selected ten clusters, 400 households have been covered.
Out of five selected Upazilas, ten unions have been chosen purposively. From each union, 40 households were surveyed using simple random sampling. As the flood losses are severe in a rural area, that’s why rural areas have been surveyed in this study. In addition to the quantitative survey, qualitative surveys such as Focus Group Discussions (FGD), In-depth Interviews (IDI), and Key Informant Interviews (KII) were undertaken to develop a better understanding of the perspectives of various stakeholders. A total of 3 IDI, 2 FGD, and 5 KII have been conducted. The FGD was mainly comprised of flood-affected individuals. Flood victims, government and non-government authorities in charge of disaster management were among the IDI respondents. KII, on the other hand, was performed with the participation of government representatives.
For this research, both primary and secondary data were collected. Some secondary data, including MAP, Literacy rate, Employment status, the Sex ratio, has been collected from the DC office in Jamalpur. Furthermore, rainfall and discharge data were obtained from the BMD, and census data were obtained from the BBS. Several reports, journals, and published papers have been collected as secondary data for this study. A comprehensive literature study determined which factors related to exposure, sensitivity, and adaptive ability should be included in the questionnaire used to collect relevant data from the field. In the absence of a household head, female members such as wives and mothers were the respondents. To cover 400 households was one of the most challenging tasks, and it took almost eight days to collect data from households. One questionnaire took about 30–35 minutes to complete. After sorting out, the data has been analyzed using SPSS and EXCEL software. SPSS version 20 was used to analyze the primary data. The percentages for each indicator were calculated using descriptive statistics. Then the percentage value of the components has been entered into EXCEL for vulnerability analysis.
2.3. Indicators for vulnerability to floods
An index is a quantitative score measurement (Cutter et al. 2013) that can be obtained by combining variables according to certain rules (Sullivan et al. 2005). Nowadays, in disaster studies, there have been widely used indices. The use of indices in disaster studies simplifies the complex data into a single value (Cutter et al. 2013, Cutter et al. 2008). Indicators worked as a tool of decision and policy making in such studies. Indicator selection is most important in vulnerability assessment. Vulnerability is often measured both in quantitative and qualitative ways (Birkmann J, 2007). Absolute measurement of vulnerability using some indicators is not an easy task due to data limitation (Borden et al. 2007, Cutter et al. 2010). That's why some researchers have adopted proxy indicators to assess vulnerability in their studies (Qasim et al. 2017). The vulnerability of this study area was determined through the use of proxy indicators. The variables' results were calculated as percentages to avoid complications associated with using multiple units of measurement. Table 1 contains the identified vulnerability indicators used in this research.
Table 1
Vulnerability assessment indicators and their associated variables Source: (Adopted from Qasim et al. 2017)
Indicator
|
Variable
|
Expert weightage
|
Explanation
|
Justification & positive or negative impact on vulnerability
|
Exposure
|
Past flood experience
|
98
|
The percentage of households who have been impacted by floods in the past
|
Prior flooding experience increases flood vulnerability, +
|
Houses constructed near the river
|
90
|
The percentage of housing units constructed adjacent to flood-prone rivers.
|
those who live near river and seashore locations are more susceptible to flooding, +
|
Sensitivity/
Susceptibility
|
Poor building material
|
75
|
The percentage of housing units made of mud
|
Flood-prone houses are created from mud, +
|
Disabled people
|
70
|
The percentage of the population with physical or mental disabilities
|
Mobility and evacuation are hampered by physical and mental disabilities, +
|
Dependents
|
40
|
Percentage of dependent population ˃64 years plus percentage of Population ˂15 years
|
Larger numbers of dependents increase the community's vulnerability to floods, +
|
Illiteracy
|
60
|
Percentage of illiterate population
|
A greater illiteracy breeds more vulnerability, +
|
Human Loss
|
50
|
Percentage of population have lost due to flooding from HH
|
Loss of a human power from household increase vulnerability, +
|
Animal Loss
|
95
|
Percentage of cattle’s have lost due to flooding from HH
|
Loss of cattle’s from Household increase vulnerability, +
|
Adaptive Capacity
|
Information about extreme weather condition
|
90
|
Percentage of HH got the flood forecasting timely
|
Early forecasting reduce vulnerability, −
|
HH access to credit facilities
|
75
|
Percentage of HH who have life insurance
|
Credit facilities access decrease the vulnerability, −
|
Social networks
|
25
|
Percentage of population that have membership in any organization
|
More social capital means less vulnerability, −
|
Education
|
98
|
Percentage of population that have high school education
|
An educated community is less vulnerable, −
|
Working age group
|
90
|
Percentage of population from age group 15–64
|
Active people decrease vulnerability, −
|
Multiple income source
|
85
|
Percentage of population with multiple
income sources
|
People with diverse income streams are less vulnerable to floods, −
|
Employment
|
40
|
Percentage of population employed
|
Employed are less vulnerable to floods, −
|
Income
|
80
|
Percentage of households above poverty line
|
People above poverty line are less vulnerable to flood hazards, −
|
2.4. Vulnerability components and their accompanying variables
Three components were used in this study to determine the community vulnerability, including Exposure, Sensitivity/Susceptibility and Adaptive capacity. Exposure had determined in this study based on two variables including, the household's previous flood experience and location near a flood-prone river. The study location is one of the most vulnerable locations to flooding, and the majority of families have experienced flooding in the past. That's why past flood experience was chosen as an essential variable. The variable location represents the number of people currently living near a river prone to flooding (Qasim et al. 2017). Building materials, disability, dependent population and illiteracy were the related variables to assess sensitivity. The building material indicates the percentage of people who had mud houses. The bulk of the respondents from the area had houses made of mud and was vulnerable to floods. A large majority of these types of homes make them more susceptible to floods.
The presence of many disabled and dependents in a community makes it more susceptible to flood hazards (Qasim et al. 2017). So, we also included disability and dependent population for susceptibility measurement. Poverty and illiteracy are also played vital roles that make communities more vulnerable to flood hazards. Therefore, we also included these variables in measuring sensitivity to floods. To measure adaptive capacity, we selected six variables, including working-age group, social networks, education, income, employment and multiple livelihood sources (Qasim et al. 2017). This study's working-age group variable includes the percentage of the population from 15 to 64 years. The people in this age group are active and may decrease vulnerability to floods. They can actively participate in any kinds of physical activities which reduce vulnerability during floods. Social capital can also increase linkages and is considered to help people during disasters. The presence of social networks, therefore, makes communities less vulnerable to floods. Education is an essential variable because educated people must be less prone to disasters (Dufty, 2008). The family's income has an impact on flood vulnerability. With more significant money, people may build houses in safer regions and utilize flood-resistant materials to construct their homes. As a result, the higher a person's income, the less vulnerable they are to flooding. According to the HIES 2016, in Bangladesh, the minimum income of a household given for a rural area is 13442 BDT per month as a standard for poverty measurement. As a result, this definition was used in this study, and people earning less than BDT 13,500 per month were considered poor. Employment is also supposed to affect people's vulnerability to floods. The higher the percentage of individuals employed in a community, the more preventive measures are taken by themselves. Similarly, a community with numerous sources of income is less vulnerable to flooding. If one source of income is harmed, society may compensate with other sources of income.
2.5 Allocating weights to selected variables and calculating index
Adaptive capacity, sensitivity/susceptibility and exposure were the indicators chosen to assess flood vulnerability in the study area. Each of them consists of more than one related variable. The variable values are collected in percentage to avoid complications and simplify the calculations. Assigning proper weights to the variable is a significant and challenging task in indexing. Weights can be assigned based on their relative importance and locational importance (Mayunga JS, 2007). The weight allocation can be performed by either empirical or subjective methods (Cutter et al. 2010). Due to data limitation, the value weighting was a bit difficult. Therefore, this study uses several literature sources (Qasim et al. 2017, Shah et al. 2018) to choose the variables of vulnerability assessment and used expert judgment in weights allocation. The variable vulnerability index was calculated by dividing the weighted value for each variable by the percentage value collected from the field survey for the same variable. The variable vulnerability index (VVI) was calculated for all the selected variables by a similar process. Low values indicate reduced vulnerability, while high values indicate increased vulnerability for a variable. Then the component vulnerability was calculated by averaging the respective variable vulnerability indices. Following this, we derived the adaptive capacity vulnerability index (AVI), exposure vulnerability index (EVI) and sensibility/susceptibility vulnerability index (SVI). Then the composite vulnerability index (CVI) was calculated using the following formulae (Karmaoui et al. 2016).
Flood Vulnerability Index = (Exposure*Sensitivity)/Adaptive capacity.
Using this formula, CVI was calculated for the five selected Upazilas and compared.