3.
2 Selection of Study Area
Khagrachari Sadar Upazila, located in the southeastern hilly region of Bangladesh, was
selected as the study area for this research. It is part of the Chattogram Hill Tracts (CHT),
characterized by a diverse terrain of forests, agricultural lands, water bodies, and
emerging built-up areas. The region has witnessed considerable land cover
transformations over the past two decades due to urban expansion, shifting cultivation,
and environmental pressures.
Rangamati is a hilly district in southeastern Bangladesh, known for Kaptai Lake, diverse
ethnic communities, and scenic landscapes. It covers around 6,116 sq km and is home to
Chakma, Marma, Tripura, and other indigenous groups. The district has seen rapid
tourism development, especially near Kaptai Lake, with rising numbers of hotels,
resorts, and transport facilities. This has triggered significant land use change,
environmental degradation, and pressure on water resources. Rangamati was selected
to understand how tourism and land transformation affect local communities, water
quality, and forest ecosystems. Fieldwork included household surveys and water sample
collection from both surface and groundwater sources.
Sajek Valley, located in Baghaichhari Upazila under Rangamati District, is a popular
highland tourism destination near the India border. Positioned at about 1,800 feet above
sea level, it has become a fast-growing resort hub in recent years. Indigenous
communities like the Lusai, Pankhoa, and Chakma live in the area. Rapid development
in a fragile environment has led to deforestation, water scarcity, and cultural shifts. Sajek
was chosen due to its ecological vulnerability and tourism-driven changes. Surveys and
water tests were conducted here to understand local perceptions, resource challenges,
and the socio-environmental impacts of tourism growth.
3.3 Land Use and Land Cover Change Detection
Understanding how land use and land cover (LULC) have changed over time is essential
for environmental planning, resource management, and policy-making. In this study,
LULC change detection was carried out for Khagrachhari Sadar Upazila using multi-
temporal satellite imagery from 2005 and 2025. The change detection process followed
several critical stages to ensure reliable results. These steps are described below.
3.3.1 Data acquisition
The study utilized cloud-free satellite imagery sourced from the United States Geological
Survey (USGS) Earth Explorer platform. Two distinct temporal datasets were selected for
the years 2005 and 2025.
• 2005: Landsat 5 Thematic Mapper (TM)
• 2025: Landsat 8 Operational Land Imager (OLI)
Both images had a 30-meter spatial resolution and were captured during the dry season
(December to March). This seasonal selection helped reduce the impact of vegetation
variation and cloud cover, ensuring better consistency across both datasets.
3.3.2 Data preparation
To prepare the data for analysis, each image was clipped to the boundary of
Khagrachhari Sadar Upazila using a shapefile collected from national GIS sources like
BBS and LGED. This step ensured that the analysis was spatially limited to the study area
and excluded irrelevant surrounding regions. The shapefile delineation also provided
administrative accuracy for further GIS processing.
3.3.3 Image preprocessing
Several important preprocessing tasks were conducted to ensure image quality and
comparability.
• Clipping: The satellite images were clipped using the defined study area shapefile.
• Cloud Masking: Cloud-contaminated pixels were removed using the Quality
Assessment (QA) bands and the pixel_qa/fmask tool in Google Earth Engine (GEE).
• Atmospheric Correction: This step was not applied because the study focused on
relative classification based on clearly identifiable surface features. Since both
images were collected under similar seasonal and environmental conditions,
atmospheric correction was deemed unnecessary.
These preprocessing steps minimized distortions caused by external factors and
prepared the imagery for accurate classification.
3.3.4 Image classification
For classifying land cover types, the study used a supervised classification approach
through the Maximum Likelihood Classifier (MLC) in ArcGIS. This method was selected
for its ability to incorporate class variance and covariance, leading to more accurate
results. Five land cover classes were defined for this study:
1. Waterbody
2. Agricultural Land
3. Built-up Area
4. Vegetation
5. Dense Vegetation
Training samples for each class were collected through visual interpretation of satellite
images, Google Earth references, and field data. A ground truth KML file developed
during a field survey that divided the area into 25 blocks, enhanced classification
precision by ensuring spatial representation and validation.
3.3.5 Accuracy assessment
Accuracy assessment is a crucial component in any land use/land cover (LULC)
classification process, as it quantifies the degree to which the classified image matches
the actual ground conditions. In this study, an initial accuracy evaluation was performed
using a confusion matrix, which compares the classified pixel values with known
reference data. This comparison provides a structured means to evaluate how well each
land cover class was identified.
Based on the confusion matrix results, several key accuracy metrics were derived:
• Overall Accuracy: The overall accuracy of the classification is calculated as the
total number of correctly classified pixels divided by the total number of reference
pixels. In this study, the overall accuracy is 62%, indicating that 62 out of every 100
pixels were correctly classified across all land cover categories.
• User's Accuracy (U_Accuracy): This measures the reliability of the classification
from the perspective of the map user – in other words, it shows the probability that
a pixel labeled as a specific class actually represents that class on the ground. The
User's Accuracy varies by class:
◦ C_1: 43.75%
◦ C_99: 100%
◦ C_100: 37.5%
◦ C_184: 83.33%
◦ C_257: 67.65%
These values indicate that some classes, like C_184, have high map reliability, while
others like C_100 are more prone to misclassification.
• Producer's Accuracy (P_Accuracy): This metric evaluates how well the
classification process correctly labeled reference data. It reflects the perspective of
the data producer. The Producer's Accuracy in this study is:
◦ C_1: 100%
◦ C_99: 33.33%
◦ C_100: 64.29%
◦ C_184: 55.56%
◦ C_257: 82.14%
These values highlight that while some classes, like C_99, are well classified, others,
such as C_1, were frequently misclassified as other land cover types.
• Kappa Coefficient: The Kappa statistic is a more robust measure than overall
accuracy, as it accounts for the agreement occurring by chance. The calculated
Kappa value is 0.5203, which indicates moderate agreement between the classified
image and the reference data. A value above 0.5 is generally considered acceptable
in remote sensing classifications, but further improvements are encouraged to
enhance the model's reliability.
3.4 Identification of causes and effects of LULC change
3.4.1 Questionnaire preparation
To identify the causes and effects of LULC changes in Khagrachhari, a structured
questionnaire was prepared using a closed-ended format. The questionnaire was
designed exclusively in English to maintain uniformity and simplify data processing. It
consisted of multiple-choice and Likert-scale questions to ensure consistency and
facilitate quantitative analysis.
The questionnaire focused on the following key areas:
• Observed Land Use Changes: Questions targeted the types of LULC changes
perceived by the respondents over the past two decades, such as forest decline,
urban expansion, agricultural intensification, and infrastructure development.
• Causes of Change: Respondents were asked to select the most relevant driving
factors, including population growth, shifting cultivation, tourism activities,
government projects, and market expansion.
• Environmental Impacts: Questions assessed the perceived impacts of land use
changes on soil erosion, water availability, forest degradation, and wildlife habitat
loss.
• Socioeconomic Effects: Respondents were asked about changes in livelihood
patterns, access to land, economic opportunities, and resource conflicts as a result
of land transformation.
• Community Responses: The questionnaire also included items on awareness,
local adaptation strategies, and community involvement in land management.
The use of only closed-ended questions ensured ease of comparison across respondents
and reduced subjectivity during data analysis. No pre-testing of the questionnaire was
conducted prior to field implementation.
3.4.2 Sample size determination and surveying
A total of 201 respondents were surveyed across different unions of Khagrachhari Sadar
Upazila using stratified random sampling. The sampling ensured representation from
various stakeholder groups including:
• Indigenous communities (Chakma, Marma, Tripura)
• Farmers
• Local government officials
• Business owners
• Education and youth
This sample size was deemed statistically sufficient to reflect the views of the broader
population, providing a 95% confidence level with a marginal error below 5%.
The survey was conducted through face-to-face interviews to ensure better engagement,
especially in rural and remote areas where literacy levels vary. Interviewers were trained
to follow ethical research practices, ensuring informed consent and confidentiality for all
participants.
3.4.3 Data processing and analysis
Following data collection, the survey responses were systematically processed and
analyzed. The workflow included:
• Data Entry: All responses were digitized using Microsoft Excel and SPSS software.
• Coding of Open-ended Questions: Descriptive responses were categorized into
thematic groups for quantitative analysis.
• Descriptive Statistics: Frequencies, percentages, and cross-tabulations were used
to summarize the main causes and impacts as reported by respondents.
• Visualization: Pie charts, bar graphs, and frequency tables were generated to
illustrate patterns in community perceptions.
• Comparative Analysis: Survey findings were compared with remote sensing
results to validate observed land cover changes and to explain socio-
environmental dynamics from both demand and satellite perspectives.
The integration of social survey data with satellite image analysis enriched the study by
linking land cover transformations with real-life local drivers and consequences. This
approach helped uncover not only “what” has changed in Khagrachhari, but also
“why” and “with what consequences” – offering a more comprehensive
understanding for policymakers and planners.
Software and Tools Used
• Google Earth Engine (GEE): For cloud masking and data acquisition
• ArcGIS 10.x: For classification, accuracy assessment, and change detection
• Google Earth Pro: For ground truth verification
• Microsoft Excel: For change matrix tabulation and statistical analysis
3.5 Socio-economic and environmental impact of
tourism on local community
Tourism has emerged as a transformative force in the Chittagong Hill Tracts (CHT),
especially in Sajek Valley, Khagrachhari, and Rangamati. While it contributes to
economic development, the unregulated nature of tourism expansion has led to
significant social, environmental, and cultural impacts on indigenous communities. This
section outlines the methodology adopted to assess the socio-economic and
environmental impacts of tourism on local people. It includes the development of
questionnaires, sample size determination, surveying process, and methods for
analyzing the collected data.
3.5.1 Questionnaire preparation
To assess tourism’s impact on the local community, a structured and closed-ended
questionnaire was developed in English. The questionnaire included both multiple-
choice and Likert-scale questions to simplify the analysis and ensure consistency among
respondents. The questionnaire was designed to cover a broad range of social,
economic, and environmental indicators influenced by tourism.
The major themes included:
• Income and Occupation Changes: Changes in livelihood before and after the
tourism boom, involvement in tourism-related jobs (e.g., resort work, transport,
handicraft selling).
• Cultural Impact: Perceptions of cultural commodification, language shifts, and
displacement of traditional customs.
• Environmental Concerns: Questions on changes in air, water, and soil quality due
to increased tourism activity.
• Social Conflict and Safety: Tourism-related crime, overcrowding, and tension
between tourists and locals.
• Infrastructure Development: Whether the development of roads, markets, and
other services due to tourism has benefited or burdened the community.
The questionnaire was reviewed by faculty advisors and field assistants for logical flow
and ease of understanding, especially in the rural regions where education levels vary.
3.5.2 Sample size determination and surveying
A total of 137 respondents were selected across three locations: Rangamati Sadar,
Baghaichhari (Sajek), and Khagrachhari Sadar. The respondents were chosen using
stratified random sampling to ensure equal representation from various occupational
and ethnic groups, including:
• Resort and hotel workers
• Local transport service providers
• Farmers and small business owners
• Indigenous community leaders
• Government officials and school teachers
Survey were conducted through face-to-face interviews by trained field enumerators to
ensure accurate data collection. The interviews were conducted in local dialects or
Bengali as per the comfort of the respondents. Informed consent and confidentiality
were ensured for all participants.
The survey period lasted two weeks, and responses were recorded on paper-based
forms before being transferred into a digital spreadsheet.
3.5.3 Data processing and analysis
All the data collected from the tourism impact survey were first coded in Microsoft Excel.
The responses were cleaned and formatted before being imported into SPSS for deeper
analysis. Various analytical tools were used, including:
• Descriptive Statistics: Frequencies and percentages to summarize responses.
• Cross-tabulation: To explore patterns such as tourism’s effect on income by
occupation type or gender.
• Visual Presentation: Bar charts, pie charts, and tables were used to illustrate key
findings.
Key insights from the analysis include:
• Income Shift: 71% of respondents noted an increase in income due to tourism,
while others reported increased living costs.
• Cultural Shifts: Many community elders (over 70%) expressed concern about the
loss of traditional language and practices due to tourist influence.
• Environmental Effects: 73 respondents confirmed deforestation for tourism
infrastructure, while 38 reported reduced access to clean water during peak
seasons.
3.6 Community Perceptions on Water Availability,
Quality and Challenges in Hill Tracts
Water is a critical resource in the Chittagong Hill Tracts, but in recent years, local
communities have increasingly faced challenges related to water availability, pollution,
and seasonal scarcity. This section outlines the methodology used to explore
community perceptions of water issues and to assess water quality using field samples
and laboratory tests.
3.6.1 Questionnaire preparation
A separate structured questionnaire was developed to assess community perception
regarding water availability, quality, and challenges. It included closed-ended questions
focusing on:
• Sources of Water: Spring, stream, tube well, rainwater, pond, etc.
• Seasonal Scarcity: Availability of water during dry and wet seasons.
• Perceived Pollution: Visible changes in water color, odor, and taste.
• Health Impacts: Experience with waterborne diseases such as diarrhea or skin
infections.
• Tourism Impacts: Whether respondents feel tourism has worsened water
conditions.
• Adaptation Strategies: Household-level responses such as rainwater harvesting or
bottled water use.
This questionnaire was developed in consultation with local field staff and translated
verbally during interviews to match local comprehension.
3.6.2 Sample size determination and surveying
A total of 106 respondents participated in this survey. Respondents were drawn from
both urban and rural areas in Rangamati, Sajek, and Khagrachhari, with a mix of
household heads, farmers, hotel workers, and women (who often manage water
collection for the household).
Sampling was purposive and stratified, ensuring diverse representation of both
upstream and downstream communities, especially those most affected by tourism and
environmental degradation.
Data was collected through household visits, and each interview lasted approximately
20-30 minutes. Interviewers were trained to approach the topic sensitively, as water
scarcity is a matter of deep community concern.
3.6.3 Water quality assessment
To verify the community’s perception of water issues, 10 water samples were collected
from key sources, including:
• Spring water in Sajek
• Tube wells in Khagrachhari and Rangamati
• Surface streams near resorts and villages
Water samples were tested in a certified environmental laboratory. Parameters tested
included:
• pH
• Turbidity
• Total Dissolved Solids (TDS)
The results revealed serious issues:
• Fecal contamination in Sajek and Rangamati exceeded WHO standards in 60% of
tested samples.
• Turbidity was highest in surface water used by villagers, indicating sediment and
pollution.
• Some tube wells showed acidic pH, suggesting contamination from septic systems.
3.6.4 Data processing and analysis
The survey responses and water test results were digitized and analyzed in Microsoft
Excel and SPSS. Descriptive statistics such as frequencies and cross-tabulations were
used to summarize community concerns.
Key findings included:
• 78% of Sajek households reported water scarcity during the tourist season.
• 65% in Khagrachhari blamed tourism and agriculture for groundwater depletion.
• 60% in Rangamati reported a decline in water quality, especially near Kaptai Lake.
• Over 45% of households resorted to purchasing bottled water, showing both
awareness and vulnerability.
Community concerns were validated by laboratory data, strengthening the study’s
conclusions regarding the degradation of water quality due to land use change,
unregulated tourism, and lack of proper water infrastructure.
