1. Introduction
Climate change, influenced significantly by human activities, is a complex phenomenon under extensive scientific scrutiny [
1]. Anthropogenic activities, primarily fossil fuel combustion and deforestation, have amplified greenhouse gas emissions—namely nitrous oxide (N
2O), methane (CH
4), and carbon dioxide (CO
2)—which are primary contributors to global warming [
2,
3]. These climate change impacts pose severe threats to environmental sustainability, biome services, and overall human well-being [
4].
Among the critical outcomes of climate change are changes in land use and land cover (LULC) and increases in land surface temperature (LST), which exacerbate biodiversity loss and induce urban heat islands (UHI) [
5]. Consequently, these have become focal points in remote sensing research [
6].
The past few decades have seen remarkable demographic, socioeconomic, and environmental transformations, mainly attributable to global urbanization [
7]. This dynamic phenomenon has sweeping implications, leading to significant LULC alterations, as revealed by [
8]. The transition from natural and pastoral areas to built-up environments significantly contributes to biodiversity loss and ecosystem service depletion.
Urbanization further leads to the creation of UHIs, impacting the energy balance and microclimate of cities [
9]. Studies, including the work of [
10], emphasize how urban morphology, such as building density and placement, intensifies the UHI effect. These changes underscore the need for a comprehensive understanding of urbanization’s repercussions for effective urban planning and sustainable development strategies [
11].
Unchecked LULC changes and rapid urbanization significantly influence Earth’s thermodynamic, hydrological, and radioactive phenomena, potentially exacerbating climate change impacts and intensifying heat waves [
12,
13]. In the absence of sufficient monitoring, the growth of built-up areas often occurs at the expense of green cover, worsening the UHI effect through a substantial rise in LST [
14,
15]. This unrestrained urban growth, resulting in significant LST increases, bears wide-ranging impacts on the UHI effect, ecosystems, and local and regional climates [
16].
The UHI phenomenon, significantly influenced by urbanization-induced LST increases, has considerable implications for local temperature, regional weather patterns, and biodiversity [
17]. UHI effects are prevalent in built-up regions due to the presence of heat-absorbing and radiating impermeable surfaces, bringing numerous adverse impacts on urban populations [
18]. To understand and quantify UHI impacts, researchers often employ LST estimates from remote sensors with high spatiotemporal resolutions [
19].
Achieving sustainable environmental improvement in cities necessitates an in-depth understanding of changing LULC trends [
20]. Given the transient nature of LULC, understanding these changes requires comprehensive knowledge at appropriate scales, bolstered by reliable time-series data [
21]. Establishing a clear relationship between LULC alterations and changing urban climates is crucial for understanding the environmental effects induced by these changes [
22].
Researchers worldwide have employed a variety of mathematical indices to understand changing LULC patterns [
23]. The normalized difference vegetation index (NDVI) and normalized differential water index (NDWI) are such measures that utilize the red and NIR bands and NIR and MIR bands of satellite imagery, respectively. They provide valuable information about vegetation status and water scarcity, aiding in LULC change evaluations and water availability studies [
24,
25,
26].
Despite the wealth of research in this area, knowledge gaps persist, highlighting the need for further investigations. This study intends to bridge this gap by examining the LULC and LST changes in the Kamrup Metropolitan District region of Northeast India from 2000 to 2022 and projecting future scenarios for the year 2032. In doing so, the aim is to gain insights into the LULC dynamics and temperature trends in the study area, and how these changes relate to the increasing phenomenon of UHIs. This study also seeks to evaluate the effectiveness of the supervised machine learning algorithm in classifying LULC over the years and to assess the reliability of the CA-ANN model’s predictions for LULC and LST in 2032. The findings of this research are anticipated to contribute valuable knowledge to urban planning, environmental management, and climate change mitigation strategies.
2. Study Area
Situated in the northeastern region of India, the state of Assam encompasses the Kamrup Metropolitan District, defined by the geographical coordinates of 26.05° N latitude and 91.60° E longitude (
Figure 1). Spanning a substantial expanse in the southwest of Assam, the district is characterized by its diverse topography, comprising plains, mountains, and river valleys. The district shares its boundaries with other Assam districts, such as Kamrup Rural, Nalbari, and Darrang.
Recent years have witnessed accelerated urbanization within Kamrup Metropolitan, evidenced by an 18.34% increase in its urban population since the 2001 Census, according to the Census of India (2011). This rapid urban expansion has driven an increase in urban communities, growth in built-up areas, and a surge in population density, now averaging 2010 individuals per square kilometer. Such urbanization can be attributed to rural–urban migration, economic opportunities, and infrastructure development, all of which underline the district’s significant urban transformation.
A significant part of the Kamrup Metropolitan district corresponds to the jurisdiction of the Guwahati Metropolitan Development Authority, with the city of Guwahati acting as the district’s administrative center. As a major urban hub and the capital of Assam, Guwahati’s geographical coordinates span from 91°33′ E to 91°52′ E longitude and 26°2′ N to 26°16′ N latitude. The study area encompasses the Guwahati Municipal Corporation (GMC) zone, which covers 176.2 km
2. Here, escalating urbanization has prompted an expansion of built-up areas, often at the cost of plant cover and water bodies, such as wetlands [
27].
Despite their importance in moderating surface temperatures, these ecological attributes have considerably declined due to urbanization. Guwahati’s distinct physical features, including mountains, rivers, lakes, and wetlands, contribute to the regional variances in its prevailing mild, humid subtropical climate. The region experiences significant annual rainfall averaging 1082 mm, predominantly during the southwest monsoon period from May to September.
Given the study’s context, the city’s demographics are of prime importance. Guwahati has experienced the most rapid urbanization compared to other northeastern Indian municipalities. The city is nestled between the southern bank of the Brahmaputra River and the lower reaches of the Shillong Plateau. Since becoming the capital of Assam in 1972, Guwahati’s population growth has surged. The GMC’s population has exponentially grown from 43,615 in 1951 to 962,334 in 2011, as per the Census of India 2011. This sharp population increase has led to substantial unplanned urbanization, impacting land use and land cover (LULC) patterns, local climate, and urban health [
28,
29].
4. Results
4.1. LULC Change Analysis
We employed a supervised machine learning algorithm for the extraction of four distinct land use and land cover (LULC) classes spanning the years 2000, 2014, and 2022 [
41]. To ascertain the accuracy of the LULC maps’ classification, we randomly selected 100 data points for each year (as depicted in
Table 2). The LULC maps’ classification proved to be highly accurate, with an overall accuracy exceeding 85% across all three years [
42]. Additionally, the Kappa coefficient, which assesses the concordance between the classified maps and reference data, yielded values surpassing 0.88 throughout the study period (refer to
Table 2). This attests to the reliability and consistency of the LULC classification for the time frame under analysis [
43].
In 2000, cultivated land was the most extensive land cover class, spanning 208.11 square kilometers. However, its area significantly diminished to 134 square kilometers in 2014 and further dwindled to 121.01 square kilometers in 2022. Conversely, the area of uncultivated land increased over this period, growing from 31.58 square kilometers in 2000 to 95 square kilometers in 2014, albeit slightly declining to 94.89 square kilometers in 2022 (
Table 3).
Vegetation, encompassing forests, grasslands, and other natural vegetation types, initially covered an expansive 668.99 square kilometers in 2000. This coverage, however, declined to 600 square kilometers in 2014 and further contracted to 459.92 square kilometers by 2022. Additionally, built-up areas, which accounted for 48.63 square kilometers in 2000, witnessed significant growth to 130 square kilometers in 2014 and 166.43 square kilometers in 2022.
The extent of water bodies, comprising lakes, rivers, and reservoirs, was 33.36 square kilometers in 2000, and it remained relatively stable, with a slight decrease to 32 square kilometers in 2014 and 31.67 square kilometers in 2022.
Two key trends emerged from 2000 to 2022. Firstly, the built-up area experienced significant expansion, while the vegetation cover and water bodies underwent contraction. The built-up area surged by a noteworthy percentage over this period, with an average annual growth rate of 9.81 square kilometers per year. Conversely, vegetation cover and water bodies decreased by negative percentages (−35.25% and −5.06%, respectively) from 2000 to 2022, with an average annual decline rate of −17.42 and −0.14 square kilometers per year, respectively. These dynamic trends underscore the evolving nature of the land surface, characterized by the proliferation of built-up areas at the cost of natural land covers and water bodies (as illustrated in
Figure 2). The implications of these changes warrant further investigation to assess their potential impact on the ecosystem and local communities [
44,
45,
46].
4.2. Land Surface Temperature (LST) Analysis
The spatiotemporal distribution of land surface temperature (LST) was evaluated utilizing a combination of equations and Landsat thermal bands throughout the study duration.
Figure 3 provides a visual interpretation of the annual LST distribution from 2000 to 2022, demonstrating an escalating trend. The maximum temperature recorded in 2000 was 23.65 °C, significantly increasing to 67.45 °C in 2022, with an annual average shift of 1.990 °C. Correspondingly, the minimum temperature registered in 2000 was 11.85 °C, exhibiting a remarkable rise to 59.26 °C in 2022, with an annual average shift of 2.166 °C. These observations underscore the increasing trend in temperatures during the study period, implicating the potential influence of various factors on LST dynamics [
47].
In 2022, the mean LST was noted as 23.652 °C, representing the average land surface temperature. The year’s peak temperature was recorded at 31.322 °C, whereas the lowest temperature was at 18.569 °C. When compared to 2014, a minor decrease in the mean LST was observed. The mean LST in 2014 was 22.606 °C, signifying a slightly higher average temperature than in 2000. The highest and lowest temperatures in 2014 were 29.897 °C and 16.373 °C, respectively. In comparison to the year 2000, there was a substantial increment in the mean LST. The mean LST in 2000 was 16.249 °C, significantly lower than the subsequent years. The maximum and minimum temperatures in 2000 were recorded as 23.657 °C and 11.856 °C, respectively. Overall, these statistics depict a persistent rise in LST from 2000 to 2022, punctuated by a significant increase in mean temperatures. The maximum temperatures also portray an escalating trend, suggesting possible alterations in temperature patterns [
47].
4.3. Validation
To verify the accuracy of the prediction results, we initially utilized the cellular automata–artificial neural network (CA-ANN) model to predict land use/land cover (LULC) and land surface temperature (LST) for the 2020s. We subsequently compared the forecasted and estimated maps using the QGIS-MOLUSCE Plugin and Terrset (v 2020) Software programs, employing diverse kappa settings (
Table 4,
Table 5,
Table 6 and
Table 7) [
48,
49]. The assessment results were generally favorable, considering all kappa parameters, accuracy percentages, and overall kappa values surpassing 0.75 and 0.80, respectively. Moreover, the mean error value across all parameters was approximately 20%, further corroborating the reliability of our findings. Additionally, a cross-tabulation was conducted to enhance our understanding of the validation results, which is elaborated upon in
Section 4.4.
4.4. Cross-Tabulation
Using SIMULATED_IMAGE_2022_ (columns) vs. REFERENCE_IMAGE_2022_ (rows), we performed a cross-tabulation.
4.5. Forecasted Land Surface Temperature (LST)
A predictive model was developed based on the prior land surface temperature (LST) dataset to forecast potential LST conditions for the year 2032 (
Figure 4). As per the model’s predictions, the creation of higher temperature zones is more probable in the northern and northwestern regions, primarily in areas experiencing extensive urban expansion [
47]. The model projects maximum and minimum LST values exceeding 26 °C and below 19 °C, respectively (
Table 8). This simulation aligns with the observed increase in LST from 2000 to 2022, largely attributed to the expansion of built-up areas, thereby considerably affecting LST values.
The intensified urban heat island (UHI) effect and the surge in LST are directly linked to the broadening urban footprints and depletion of vegetative cover [
47]. The contributing factors for escalating temperatures and UHI impact are multifold, and include urbanization, alterations to the greenhouse gas effect, global warming, and modifications in surface features [
50,
51,
52]. The projected LST scenarios underline the grave concerns associated with this rising trend, such as the heightened UHI effect.
4.6. Interplay between Land Use/Land Cover (LULC) Indices
The presented correlation coefficients (R
2) (refer to
Figure 5) that detail the relationship between the normalized difference built-up index (NDBI) and land surface temperature (LST) for the years 2000, 2014, and 2022 exhibit compelling patterns. There is a subtle uptick in the correlation coefficient from 2000 to 2022, denoting an intensifying relationship. The linear regression models for each year suggest a consistent positive relationship between LST and NDBI, albeit with variable slopes and intercepts.
In 2000, the positive slope of 10.675x signals a strong influence of LST on NDBI, with an intercept of 24.674. The R2 value of 0.3632 implies that approximately 36.32% of the NDBI variability can be accounted for by LST during this year. Similarly, in 2014 and 2022, the models project positive slopes of 12.218x and 6.4392x, respectively, indicating an ongoing positive relationship between LST and NDBI, albeit with divergent magnitudes. The intercepts for 2014 and 2022 stand at 25.615 and 16.246, respectively. The respective R2 values of 0.3638 and 0.3844 imply that roughly 36.38% and 38.44% of the NDBI variability can be ascribed to LST during these years. This suggests that the influence of built-up areas on temperature patterns has become more significant, with NDBI serving as a more potent predictor of LST variations.
Conversely, an analysis of the relationship between LST and the normalized difference vegetation index (NDVI) for the Kamrup Metro Region across 2000, 2014, and 2022 reveals a different dynamic. The models demonstrate a negative relationship between LST and NDVI, with fluctuating slopes and intercepts. In 2000, the equation projects a steeper negative slope of −5.9821x, suggesting a considerable impact of LST on NDVI, with an intercept of 26.19. Nevertheless, the R2 value of 0.1192 indicates that only around 11.92% of the NDVI variability can be accounted for by LST during this year.
In 2014 and 2022, the models present negative slopes of −6.6917x and −2.5775x, respectively, signifying a weakening relationship between LST and NDVI. The intercepts for 2014 and 2022 stand at 25.37 and 17.184, respectively. The respective R
2 values of 0.2286 and 0.2346 suggest that around 22.86% and 23.46% of the NDVI variability can be ascribed to LST during these years. These findings indicate that although vegetation coverage (measured by NDVI) might influence temperature patterns, other factors, such as urbanization, surface characteristics, and localized climatic conditions, likely play a more decisive role in governing LST fluctuations [
53,
54].
4.7. Urban Heat Island Profile
The urban heat island (UHI) profile, as depicted in
Figure 6, is based on a west–east transect drawn across the city of Guwahati, an area characterized by high concentrations of built-up regions. The temperature distribution along this line demonstrates that urban centers typically exhibit higher temperatures compared to their surrounding suburban counterparts. However, a detailed examination reveals a series of temperature peaks and valleys, highlighting distinct thermal variations across the city.
These fluctuations are directly linked to the city’s unique geographic features. The valleys, which correspond to areas with relatively cooler temperatures, are often associated with farmlands, bodies of water, and parks enriched with green spaces. Nearby vegetation-covered areas also play a pivotal role in creating these cooler zones.
In 2000, the highest UHI values appear at a distance of 0.7 on the transect, interspersed with several valleys or dips in the graph at distances of 0.02, 0.06, and 0.18. These dips record UHI values of less than 0.3, implying the presence of significant vegetation cover.
As we progress from 2000 to 2014, there is a noticeable increase in temperatures. The peak at a distance of 0.14 escalates to a UHI value of 0.76, while the valleys at distances of 0.02, 0.06, and 0.18 simultaneously rise to values of 0.31, 0.34, and 0.38, respectively. This uptrend continues into 2022, with the UHI value at a distance of 0.14 surging to 0.78.
The 2014 and 2022 profiles portray an increasing formation of heat islands. This trend is marked by a growth in the number of temperature peaks (high temperatures) and a decrease in the number of valleys (low temperatures), indicating an escalating UHI effect as urban areas expand and vegetation cover reduces. This development is corroborated by the land use/land cover (LULC) analysis conducted from 2000 to 2022.
5. Discussion
The projected future scenario of the Kamrup Metro Region can be extrapolated from the various factors explored in this study. The land use/land cover (LULC) changes revealed a notable transformation in the land surface composition over time, characterized by an expansion of built-up areas at the expense of natural land cover and water bodies [
55,
56,
57,
58]. This trend not only demonstrates an ongoing process of urbanization but also suggests its likely continuation in the future.
The land surface temperature (LST) analysis further underscored a persistent and significant uptick in land surface temperatures, especially a remarkable rise in mean temperatures from 2000 to 2022. This escalation in temperature is intrinsically tied to the proliferation of built-up areas, thus, spotlighting the urban heat island (UHI) effect [
59]. Predicted LST scenarios for 2032 suggest the emergence of hotter zones, predominantly in the northern and northwestern regions, which are undergoing significant urban development. These predictions signal an impending amplification of the UHI effect, raising serious concerns about its implications for the region’s socio-economic and environmental sustainability.
Moreover, the correlation analysis between the LULC indices exposed a critical insight. While the built-up areas are gaining more influence on temperature patterns, vegetation cover seems to have a diminishing effect on temperature modulation. This trend suggests a systemic imbalance between urbanization and environmental preservation, reinforcing the consequential rise in temperatures and UHI effect [
60].
Collectively, these results indicate that the Kamrup Metro Region is on a trajectory towards increased urbanization, heightened temperatures, and amplified UHI effect. The potential impacts of this pathway are vast, encompassing changes in local climate, increased energy consumption, public health risks, and threats to biodiversity, among others [
61,
62,
63].
These findings underscore the urgency of implementing sustainable land management and urban planning strategies [
64]. Efforts must focus on preserving and restoring natural land cover to counterbalance the adverse effects of urban sprawl and climate change [
65]. Green infrastructure, such as urban green spaces, green roofs, and street trees, could play a crucial role in mitigating the UHI effect. Moreover, sustainable urban planning can promote compact development, reducing the need for land conversion and, thus, preserving natural habitats. Policymakers and urban planners should also prioritize climate resilience in their strategies, considering the vulnerability of urban areas to climate change.
This study serves as a clarion call for a balance between urban growth and environmental preservation, illuminating the critical need for sustainable solutions to manage urban heat islands and mitigate climate change impacts. It underscores the imperative to bridge the divide between economic development and environmental conservation, as the future of the Kamrup Metro Region hangs in the balance.
6. Conclusions
The interplay between land use and land cover (LULC) and land surface temperature (LST) in the Kamrup Metropolitan District region of Northeast India has been deeply examined in this study. The research unfolded through the timescale from 2000 to 2022, offering a well-founded look into the future with projections until 2032. While the conclusions derived are multi-layered and extensive, they are also intricately tied to the limits of the study, opening doors for future research directions. The methodology anchored on a supervised machine learning approach demonstrated that robust and accurate results are achievable. This approach presented reliable mappings of LULC transformations, highlighting an evident shift from cultivated land to built-up areas, painting a picture of growing urbanization. Similarly, the upward trend of LST underscored a warming landscape influenced strongly by urban development and decreasing green cover. The robustness of the CA-ANN model used for future LULC and LST predictions was validated, despite the inherent uncertainties in predictive modeling. However, it is worth acknowledging that the model’s accuracy depends heavily on the quality of input data and how representative it is of future conditions, signaling a potential area for enhancement in future studies. Future projections pointed towards heightened LST, particularly in areas projected for urban growth. This correlation between urban expansion and increasing LST was supported by the relationships between LST, NDBI, and NDVI. The study found that the impact of built-up areas on temperature variations overpowered that of vegetation cover, underscoring the significant role of urbanization in escalating temperatures. The analysis of the urban heat island (UHI) effect extended the discussion further by revealing that urban centers exhibited higher temperatures than their suburban counterparts. The fluctuating peaks and valleys in the temperature profile aligned with the spread of built-up areas and shrinking vegetation, providing evidence of the manifestation and intensification of the UHI effect. While the study provides critical insights into the ongoing urbanization process, increasing temperatures, and growing UHI effects, it also points out several limitations and gaps. The reliance on remote sensing data, although powerful, also carries uncertainties related to data quality and resolution. Furthermore, while the machine learning model provides accurate predictions, it also introduces an inherent degree of uncertainty due to its stochastic nature. These limitations not only indicate the need for caution while interpreting the results but also point towards future research directions. In addition, the study leaves some questions unanswered. While it highlights the importance of vegetation in mitigating UHI effect, the detailed role of different types of vegetation and green infrastructure is left unexplored. Additionally, how urban planning strategies can effectively utilize these natural resources to combat the UHI effect warrants further research. Overall, the study provides a comprehensive understanding of the changing LULC and LST dynamics in the Kamrup Metro Region. It emphasizes the significance of sustainable land management and urban planning strategies in mitigating the adverse effects of urban expansion. At the same time, it throws light on areas needing further exploration, ultimately guiding the region towards a sustainable and climate-resilient future.