Analyzing the Relationship Between Vegetation and Temperature Changes in the Sylhet Region ^†

Abstract

As global temperatures continue to rise, understanding the relationship between climate and vegetation is crucial for agriculture and for mitigating and adapting to environmental changes. The complex interaction between vegetation and climate becomes even more significant as temperatures increase, making it essential to comprehend these changes in the environment. This study investigates vegetation dynamics in the Sylhet region of northeastern Bangladesh between 1988 and 2025, focusing on how long-term temperature changes have influenced vegetation cover. The research utilizes Landsat-derived Normalized Difference Vegetation Index (NDVI) datasets from 1988, 1995, 2005, 2015, and 2025, alongside monthly temperature records from NOAA’s NCEI. The temperature data was analyzed using Pearson’s correlation and descriptive statistical method to examine the relationship between vegetation and climate. The results show that dense vegetation declined sharply, with an estimated net loss of ~12.9 km² converting to sparse vegetation and ~1.5 km² converting to urban/barren land between 1988 and 2025. At the same time, sparse vegetation expanded, while urban area/barren land areas increased substantially. Water bodies also showed reductions. Over the same period, the mean annual temperature rose by +0.32 °C. These findings highlight the region’s ecological vulnerability to combined climatic and anthropogenic pressures, underscoring the urgent need for sustainable land management and adaptive strategies.

1. Introduction

An important environmental concern is climate change, which is manifested in changing weather patterns and rising global temperatures. As a sensitive indicator of ecological stress, vegetation also regulates the climate. A popular tool for evaluating the health and spatial distribution of vegetation in response to environmental factors like temperature and precipitation is the NDVI, which is obtained via satellite remote sensing [1,2]. Northeastern Bangladesh’s Sylhet region has a humid subtropical monsoon climate and is about 12,300 square kilometers in size, the study focuses on two rapidly urbanizing sub-regions Sylhet Sadar and South Surma. It is distinguished by a varied landscape of hills, valleys, and seasonal wetlands and receives a lot of rainfall each year [2]. In the past, dense vegetation was supported by these natural conditions [3]. Nevertheless, despite comparatively consistent rainfall patterns, rising temperatures have changed the composition and yield of the surrounding vegetation [4]. However, few studies have provided a long-term perspective that explicitly examines the relationship between vegetation and temperature in Sylhet using consistent multi-decadal data. This study addresses that gap by combining Landsat-derived NDVI from 1988, 1995, 2005, 2015, and 2025 with monthly temperature records from NOAA/NCEI, offering a 37-year temporal scope. Research has shown that there are intricate connections between NDVI and climate variables. NDVI had a stronger correlation with evapotranspiration than with rainfall in the northeast, including Sylhet [2]. Unlike earlier works that focused primarily on rainfall or evapotranspiration, this research emphasizes temperature as a direct climatic driver of vegetation change. Sylhet is of particular importance because it is simultaneously an agricultural hub, a rapidly urbanizing zone, and an ecologically sensitive landscape. Recent research has shown that urban expansion combined with climate variability is reshaping land use and land cover patterns in Sylhet, highlighting the urgency of sustained monitoring [5]. Sensitive wetland–forest ecosystems are undergoing measurable vegetation shifts, underscoring the need for long-term monitoring in regions like Sylhet [5]. Changes in vegetation cover therefore have immediate implications for food production, tea cultivation, urban planning, and water resource management. Another important factor influencing vegetation dynamics in a variety of biological zones is temperature [1]. This study attempts to measure how vegetation health and density have been impacted by rising temperatures over the years using historical temperature data and long-term satellite NDVI datasets. The NDVI and temperature changes for the Sylhet Sadar and South Surma from 1988 to 2025 are statistically analyzed in this work. This study advances in uniquely integrates long-term Landsat NDVI datasets with detailed temperature records to quantify vegetation–climate interactions in Sylhet.

2. Study Area

Northeastern Bangladesh’s Sylhet is a vibrant socioeconomic and biological area. The administrative zones of Sylhet Sadar and South Surma, which are undergoing fast environmental change and urbanization, are the subject of this study (Figure 1).

2.1. Topography

With a total area of 12,298.4 km² and a population density of 980 people/km², Sylhet is situated roughly at latitude 24°53′24″ N and longitude 91°51′36″ E [6]. Sylhet is located in a tectonically active region with low hills, alluvial plains, and haors (wetland depressions), roughly 197 km northeast of Dhaka. Originating in the Indian hills, the Surma and Kushiara rivers traverse the area before joining the Meghna River system, which is essential for both agricultural and regional hydrology [7].

2.2. Climate

The humid subtropical monsoon climatic zone includes Sylhet. About 3876 mm of rain falls there on average each year, and the relative humidity is high all year long [6]. The pre-monsoon season (March–May) delivers moderate precipitation, whereas the monsoon season (June–September) accounts for highest distribution of seasonal rainfall. The heaviest monthly rainfall usually occurs in July [8].

3. Materials and Methods

3.1. Data Collection

Monthly temperature data from 1988 to 2025 were collected from the National Centers for Environmental Information (NCEI) under the National Oceanic and Atmospheric Administration (NOAA). The data were extracted from NOAA’s publicly accessible climate database to align temporally and spatially with the selected satellite scenes [9].

3.2. Satellite Imagery Acquisition and Preprocessing

Landsat satellite imagery was utilized to assess vegetation dynamics across Sylhet Sadar and South Surma upazilas for the years 1988, 1995, 2005, 2015, and 2025. Due to the unavailability of cloud-free or usable data in the USGS Landsat archive for 1985, the analysis begins with the earliest available scene from 1988. Landsat 5 Thematic Mapper (TM) data were used for the years 1988 and 1995, Landsat 7 Enhanced Thematic Mapper Plus (ETM+) for 2005, and Landsat 8 Operational Land Imager (OLI) for 2015 and 2025 [10]. All images were downloaded from the USGS EarthExplorer platform and selected to represent dry-season conditions with minimal cloud cover, ensuring consistency for temporal NDVI comparison. Each image was radiometrically and geometrically corrected and projected to UTM Zone 46N using the WGS 84 datum. Subsetting was performed using administrative boundary shapefiles in ArcMap 10.8. NDVI was computed using red and near-infrared bands following standard methodology. All image preprocessing, including layer stacking, subsetting, and NDVI calculation, was conducted using ArcMap 10.8 software.

3.2.1. Calculation of NDVI

Utilizing reflectance values in the red and near-infrared (NIR) regions of the electromagnetic spectrum, the NDVI is a popular spectral vegetation measure (

Table 1. Calculation of NDVI Using Satellite Imagery.

Satellite	Year	NDVI Formula
Landsat 5	1988, 1995	(Band 4 − Band 3)/(Band 4 + Band 3)
Landsat 7	2005	(Band 4 − Band 3)/(Band 4 + Band 3)
Landsat 8	2015, 2025	(Band 5 − Band 4)/(Band 5 + Band 4)

). It works well for evaluating the condition, density, and spatial distribution of vegetation. In general, non-vegetated surfaces, including lake bodies or populated areas, are indicated by negative NDVI values; sparse or no vegetation is represented by values near 0; and dense, healthy vegetation is indicated by values near +1. Each Landsat image from 1988, 1995, 2005, 2015, and 2025 had its NDVI calculated using the relevant spectral bands according to sensor specifications. The Raster Calculator tool in ArcMap 10.8 was used to carry out the computation. The following NDVI formula is used for every Landsat sensor:

3.2.2. Classification of NDVI

Following NDVI computation, each raster image was classified into four distinct land cover categories based on pixel intensity values. The NDVI classification was defined as follows: Water Body, Urban Area / Barren Land, Sparse Vegetation, Dense Vegetation These thresholds were derived based on literature guidelines and visual inspection of NDVI distribution across the study years. Classified NDVI maps were generated using the Reclassify tool in ArcMap 10.8. Classification thresholds were determined using the Natural Breaks (Jenks) optimization method in ArcMap 10.8, which minimizes intra-class variance and maximizes inter-class differences. The resulting NDVI ranges were:

This classification (

Table 2. Classification of NDVI.

NDVI Class	Class Range
Water Body	<0.0
Urban Area/Barren Land	0.0 ≤ NDVI < 0.2
Sparse Vegetation	0.2 ≤ NDVI < 0.5
Dense Vegetation	NDVI ≥ 0.5

) approach is widely used in vegetation classification [2].

3.2.3. Analysis of Temperature

Descriptive statistics including Total, Mean, Average Temperature from 1988 to 2025 were calculated. This approach is widely used in climate studies for initial characterization of data series [11].

3.3. Correlation Analysis of NDVI and Temperature

Pearson’s correlation analysis was used to assess the linear relationship between NDVI and temperature. Coefficient values range from −1 to 1, where positive values indicate a direct relationship and negative values indicate an inverse relationship. A value near 1 suggests strong positive correlation and −1 suggests a strong negative correlation.

4. Results and Discussion

4.1. Spatial and Statistical Distribution of Vegetation

NDVI analysis over Sylhet City Corporation and South Surma across the years 1988, 1995, 2005, 2015, and 2025 shows a general pattern of vegetation increase over time, followed by signs of recent decline. The rising trend is likely linked to natural vegetation recovery and urban greening efforts in the early decades. The slight decline in the most recent year may be associated with intensified urban expansion and land conversion. These patterns are in line with observed greening trends in monsoon-influenced tropical regions.

Classified NDVI maps of Sylhet and South Surma for the years 1988, 1995, 2005, 2015, and 2025. The maps represent four land cover categories derived from NDVI thresholds: Dense Vegetation, Sparse Vegetation, Urban Area/Barren Land, and Water Body (Figure 2). Over time, a transition from Urban Area/Barren Land to Sparse Vegetation is evident in several peripheral areas, alongside an increase in urban/barren zones, particularly after 2005.

Substantial land cover transitions and temperature variations in the region occurred between 1988 and 2025 (

Table 3. NDVI and Temperature Change From 1988 to 2025.

Type of Change	Year (1988–1995)	Year (1995–2005)	Year (2005–2015)	Year (2015–2025)	Year (1988–2025)
Water Body—Water Body (km2)	8.012419	5.373532	1.329697	1.932187	5.373532
Water Body—Urban Area/Barren Land (km2)	9.735015	6.525312	1.319181	0.148101	5.996399
Water Body—Sparse Vegetation (km2)	4.273481	3.668904	0.429682	0.039027	5.206485
Water Body—Dense Vegetation (km2)	0.001913	0.649105	0.026132	0.042527	5.444283
Urban Area/Barren Land—Water Body (km2)	2.708974	0.001543	0.774978	4.063739	1.77701
Urban Area/Barren Land—Urban Area/Barren Land (km2)	248.174892	2.548474	63.329284	51.862553	127.334783
Urban Area/Barren Land—Sparse Vegetation (km2)	63.225581	232.57679	207.770414	9.94961	130.691764
Urban Area/Barren Land—Dense Vegetation (km2)	0.007295	54.066216	19.153803	1.551276	54.249509
Sparse Vegetation—Water Body (km2)	0.117156	0.347635	0.057183	1.39561	0.447443
Sparse Vegetation—Urban Area/Barren Land (km2)	31.009367	0.254644	2.754987	109.227988	33.449779
Sparse Vegetation—Sparse Vegetation (km2)	128.646017	48.517458	162.323755	231.586776	112.283359
Sparse Vegetation—Dense Vegetation (km2)	0.030635	161.902016	51.857713	28.567213	13.625689
Dense Vegetation—Water Body (km2)	0.010151	3.18865	0.0009	0.22457	0.02151
Dense Vegetation—Urban Area/Barren Land (km2)	0.618517	0.002786	0.024017	6.982332	1.455767
Dense Vegetation—Sparse Vegetation (km2)	17.678491	0.48087	0.286607	19.499991	12.871345
Dense Vegetation—Dense Vegetation (km2)	17.678491	0.008414	3.233718	47.565456	4.408378
Average Temperature (°C)	0.08	0.2	–0.11	0.14	0.32

). Notably, there is a marked increase in urban/barren land expansion, often replacing sparse and dense vegetation, particularly during 1995–2005 and 2015–2025. Dense vegetation showed a significant shift toward sparse vegetation, highlighting ongoing deforestation. Simultaneously, sparse vegetation exhibited a net increase in permanence but also a sharp conversion to urban land in 2015–2025. These land cover changes correspond with a gradual increase in average temperature (+0.32 °C overall), suggesting possible land-atmosphere feedback mechanisms.

4.2. Analysis of Temperature Trends

Long-term temperature trends were described using descriptive statistics, such as the total, mean, and average yearly temperatures from 1988 to 2025. The year-by-year temperature distribution is displayed (Figure 3). Emphasizing the variety and rising median values with time. Short-term anomalies may be indicated by the higher number of outliers in previous years.

Illustrates a positive linear trend, confirming a gradual rise in temperature over the study period. This reflects broader regional warming patterns (Figure 4). A linear trend analysis of annual mean temperatures (1988–2025) indicates a gradual warming (Figure 4). Complementing this, the boxplot (Figure 3) illustrates interannual variability and outliers, providing insight into both long-term trends and short-term anomalies relevant to ecological responses.

5. Relation Between Vegetation and Temperature Change

To assess how temperature changes relate to land cover transitions, we classified NDVI data into key land types across the years 1988, 1995, 2005, 2015, and 2025. We calculated the area changes between different land cover categories over time and compiled temperature data for the same years. Pearson’s correlation coefficient was then used to quantify the statistical association between each land cover transition and corresponding temperature changes.

The Pearson correlation results indicate clear statistical relationships between temperature variations and land cover changes in the region. Strong negative correlations were observed between transitions from water bodies to urban areas and vegetated land, which may reflect concurrent drying trends consistent with hydrological stress under warming conditions. Similarly, positive correlations were found between transitions from dense to sparse vegetation and from sparse vegetation to urban/barren land, suggesting that higher temperatures are statistically associated with vegetation reduction and land conversion. These associations should not be interpreted as direct causation but rather as evidence of co-occurring patterns that may be influenced by both climatic and anthropogenic factors. In contrast, land cover classes that remained unchanged (e.g., dense vegetation to dense vegetation or urban to urban) showed weak or no correlation with temperature, highlighting their relative stability (Figure 5). Overall, the findings demonstrate that rising temperatures are closely linked with reductions in natural vegetation and water bodies and are statistically associated with increases in urban and barren land.

6. Conclusions

Between 1988 and 2025, the region experienced substantial changes in climate and land cover. Indicating both ecological deterioration and increasing human influence, NDVI shows a continuing drop in dense vegetation and a considerable increase in sparse vegetation and urban/barren areas. Despite slight inter-decadal fluctuation, these changes are accompanied by an overall increase in average temperature, which throughout the course of the research period increased by around 0.32 °C. Strong negative associations between temperature rises and the survival of water bodies and dense vegetation are found in the Pearson correlation analysis, indicating that rising temperatures may be a major factor in the loss of these environmentally important places. A possible connection between warming and land conversion processes was suggested by the positive correlations found between temperature and transitions involving vegetation degradation and urban expansion. On the other hand, poor correlations in stable land cover classes indicate that the climate has little effect in those areas. These results highlight how susceptible Sylhet’s ecosystems are to changes in the climate, especially when it comes to plant loss and hydrological stress. There will be more environmental deterioration in the area if current patterns continue, with significant declines in vegetative density and water retention capacity. In order to maintain the region’s natural integrity, it is imperative that sustainable land management policies and climate-adaptive planning be implemented.