Urban Expansion and Thermal Stress: A Remote Sensing Analysis of LULC and Urban Heat Islands in Ghaziabad, India

Abstract

The climate and environment of metropolitan areas have been negatively impacted by swift urbanization and industrialization. Surface Urban Heat Islands (SUHIs) are among the most critical environmental phenomena. This research focused on the spatiotemporal analysis of land use/land cover (LULC) changes in relation to surface urban heat islands and their interconnections from 1992 to 2022. Land Surface Temperature (LST), LULC, and LULC indices, such as the Normalized Difference Moisture Index (NDMI), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Built-up Index (NDBI), were generated using Landsat data. Urban hot spots (UHSs) were identified, and the Urban Thermal Field Variance Index (UTFVI) was then used to evaluate the spatiotemporal variation in thermal comfort. The results indicated LST values between a low of 14.24 and a maximum of 46.30. Urban areas and exposed surfaces, such as open or bare soil, exhibit the highest surface radiant temperatures. Conversely, regions characterized by vegetation and water bodies have the lowest. Additionally, this study explored the correlation between LULC, LULC indices, LST, and SUHIs. LST and NDBI show a positive relationship because of urbanization and industrialization (R² = 0.57 for the year 1992, R² = 0.38 for the year 2010, and R² = 0.35 for the year 2022), while LST shows an inverse relationship with NDVI and NDMI. Urban development should account for thermal sensitivity in densely populated regions. This study introduced an innovative spatiotemporal framework for monitoring long-term changes in urban surface environments. Furthermore, this research can assist planners in creating urban green spaces in cities of developing nations to minimize the adverse impacts of urban heat islands and improve thermal comfort.

1. Introduction

Urbanization refers to the migration of populations from rural to urban areas. Globally, urban populations have increased significantly over the past few decades, with estimates indicating that 55% of the world’s population resided in urban areas in 2018—a figure projected to rise to 68% by 2050 [1]. Since independence, India has witnessed considerable urban expansion, driven largely by a rapidly growing urban population. The proportion of urban residents in India increased from approximately 17.3% in 1951 to 31.16% in 2011 [2]. It is projected that by 2030, over 40% of India’s population will live in cities and contribute to 75% of the country’s GDP [3]. Urbanization, which involves both population migration to urban areas and land cover changes, has been strongly associated with environmental degradation, particularly in terms of waste generation, air pollution, and climate change [4,5]. The UHI effect is strongly linked to human-driven effects on energy consumption [6], air quality [7], and changes in precipitation patterns [8]. The UHI phenomenon, wherein the urban core experiences elevated temperatures compared to its surrounding suburban or rural areas [9,10], was initially observed by Luke Howard (1772–1864) [11]. Since then, numerous studies and literature reviews addressing the UHI phenomenon have been published shortly after their discovery, including literature reviews [12] and articles [13].

The recent surge in urban population is primarily attributed to extensive rural-to-urban migration, driven by enhanced economic opportunities and improved living conditions in cities and large metropolitan centers [14,15,16]. Urbanization transforms natural land into impervious surfaces, significantly changing its thermal characteristics and contributing to increased heat retention. These modifications intensify the thermal environment of urban areas and contribute to the formation of surface urban heat islands [17,18]. Understanding land use changes is essential for assessing the impacts of human activities on the landscape [19]. Changes in urban land use have led to several environmental issues, such as the emergence of localized microthermal and microclimatic zones [20]. Consequently, the decline in vegetation and the rapid growth of urban settlements are likely to elevate LST [21,22]. The integration of RS and GIS technologies has proven effective for modeling changes in LULC and LST [23,24,25].

UHIs are characterized by higher surface and air temperatures compared to their natural surroundings [26,27]. As a consequence of rapid urban expansion, significant LULC transformations have occurred in many cities. The most prominent consequence of LULC changes in Indian cities has been their impact on climate conditions, through rising temperatures, altered precipitation patterns, and declining air quality [20,28,29]. Several studies have quantitatively demonstrated a strong relationship between LULC changes and variations in land surface temperature [30,31,32].

The UHI effect can exacerbate extreme heat events, making them more intense and frequent [33,34]. Enhanced environmental pollution not only increases public health risks but also leads to higher energy consumption [35]. Urban heat islands (UHIs), air pollution, severe climatic conditions, and other human–nature conflicts are among the critical challenges that have arisen over the last century due to rapid urban expansion and LULC alterations [36]. Rapid urbanization and drastic land cover transformations have significantly altered surface energy balances, resulting in elevated land surface temperatures in urban centers [37]. Several studies have highlighted a strong association between LST, UHI intensity, and the spatial extent of built-up and vegetated areas [38,39,40]. The UHI effect leads to localized heat accumulation, resulting in higher urban temperatures relative to surrounding areas and posing serious health risks to urban populations [41,42]. The study of the UHI phenomenon is critical for developing effective thermal mitigation methods and improving the thermal resilience of urban environments, thereby contributing to the achievement of the United Nations Sustainable Development Goals [43]. Extreme temperatures observed on these high LST surfaces are referred to as urban hot spots [44]. Identifying and assessing UHSs is crucial for effective urban temperature regulation and maintaining ecological balance in the urban environment [45]. The main aim of this research is to identify urban hot spots within Ghaziabad, thereby assessing the prevailing levels of thermal comfort in the region.

Due to its proximity to the national capital, Ghaziabad has benefited from increased industrialization and associated economic opportunities. However, this rapid development has also introduced significant environmental and infrastructural challenges. The city’s thermal environment has experienced significant changes as a result of intensified industrial development and rapid urbanization. This research examines the dynamics of land use/land cover (LULC) and their interactions with land surface temperature (LST) and surface urban heat islands (SUHIs). Additionally, the study aims to identify urban hot spots and evaluate the correlations between different land use indices and LST. Unlike many previous remote sensing studies of urban heat islands, this research integrates SUHIs and UTFVI with UHS mapping to provide a more detailed assessment of thermal stress. By examining a three-decade period (1992–2022), it captures the long-term trajectory of UHI dynamics in Ghaziabad. The results of this investigation are expected to provide valuable insights into urban planning, environmental management, and the development of strategies aimed at mitigating the effects of climate change.

2. Materials and Methods

2.1. Study Area

Ghaziabad is located in the east of the national capital territory of Delhi and west of the state of Uttar Pradesh. It is in the National Capital Region (NCR), as shown in Figure 1. Due to its location adjacent to Delhi, it is known as the “Gateway of Uttar Pradesh.” Ghaziabad is surrounded by the district Meerut in the north, Gautam Budh Nagar and Bulandshahar in the south, Hapur on the east, and the national capital Delhi on the west side. The city is divided into the CIS Hindon Area (CHA) in the east and the Trans-Hindon Area (THA) in the west by the Hindon River, which flows through it. THA comprises 1/3 of the city’s area and population, while CHA comprises 2/3. Ghaziabad is a central industrial and corporate hub in Uttar Pradesh. In this area, similarly to other regions in northern India, there are mainly three distinct seasons: summer, winter, and the monsoon. The city experiences a dry and healthy climate, characterized by very high temperatures during the summer and significantly low temperatures in the winter. The monsoon season typically starts between late June and early July and continues until October. A significant amount of rainfall in the area occurs during the monsoon season. Ghaziabad, a significant district in Uttar Pradesh, is within the NCR and is part of the Meerut division. The 2011 census ranked it as the third-most populous district. Located in the center of the Ganga Yamuna doab, Ghaziabad has a roughly rectangular shape, measuring approximately 72 km in length and 37 km in width; the size of the sample area we examined is 177.9 km². Ghaziabad Metropolitan Area has a total population of 1,648,643, comprising 874,607 males and 774,036 females. Its geographical coordinates lie between 28°30′ and 28°59′ north latitude and 77°26′ and 78°10′ east longitude.

Different Landsat satellite images were used to assess LULC changes, extract LST, and examine UHI variations in the study area. All images featured a spatial resolution of 30 m for Bands 1–5 and 7 (

Table 1. Satellite datasets utilized for the study.

Sr. No	Satellite/Sensor	Acquisition Date	Cloud Cover (%)	Path/Row	Spatial Resolution of Spectral Bands (m)
1	Landsat 5 TM	16 March 1992	10.00	146/40	30
2	Landsat 5 TM	3 April 2010	0.01	146/40	30
3	Landsat 8	4 April 2022	0.00	146/40	30

). All Landsat images were sourced from the United States Geological Survey (USGS) platform (https://earthexplorer.usgs.gov/, accessed on 12 October 2024). These datasets are widely used for surface temperature estimations [46] and LULC change assessments [47]. The selected images had high spatial resolution and cloud cover of less than 2%. The shapefile for the study area was obtained from a verified source. Georeferencing was initially conducted using four Ground Control Points (GCPs), followed by careful digitization of the shapefile.

To investigate changes in LULC, LST, and NDVI within Ghaziabad, all data were transformed into the UTM/WGS84 coordinate system. Landsat images were geometrically corrected to the Zone 43 North UTM projection using the WGS84 geodetic datum. ArcGIS 10.8 software (Esri, Redlands, CA, USA) was utilized for spatial analysis, while ERDAS 2014 (Hexagon Geospatial, Sweden) was employed to assess LULC changes and process images.

2.2. Image Processing and Classification

The study area was delineated, and false color composite (FCC) images were generated by combining spectral bands to facilitate visual interpretation and improve classification accuracy. Visual interpretation provides valuable insights into land cover changes over time [48]. Image enhancement techniques have improved the interpretability of satellite imagery [49].

Remote sensing techniques employing Landsat images have been used to classify and map LULC changes by identifying distinct landscape elements [50]. More than 300 training samples were collected across all LULC classes to ensure accurate classification, with over 70 samples collected for each class. Five LULC categories were identified to assess the impact of land use changes on LST and SUHIs.

These categories—built-up area, agriculture, open land, vegetation, and water bodies—were classified using a supervised approach. The maximum likelihood algorithm was employed to classify the five land cover categories. The thermal band was excluded due to its non-reflective properties [49]. A detailed description of LULC classes is given in

Table 2. Description of various LULC classes.

LULC Class	Details of the LULC Class
Built-Up Area	Includes all residential, business, and industrial areas, settlements, villages, and infrastructure related to transportation.
Vegetation	Comprises natural forest, trees, shrubs, and all types of vegetation (excluding standing crops).
Agriculture	Encompasses cropland, including both large and small farms, as well as mechanized farms.
Water Bodies	Refers to anybody of open water, such as ponds, rivers, and lakes.
Open Land	Areas devoid of vegetation of vegetation are mainly made up of bare earth, exposed soil, and other natural rock surfaces.

.

2.3. Land Use/Land Cover (LULC) Extraction and Change Analysis

There are several ways to implement supervised classification; however, in this study, the commonly used maximum likelihood classifier (MLC) was employed together with ERDAS IMAGINE software. Ghaziabad was classified into five main LULC classes—built-up area, vegetation, agricultural, open land, and water bodies. Different spectral signatures of Landsat images were identified and validated using Google Earth images from the same period. Before assigning LULC classifications, band composition was applied to the Landsat images. The maximum likelihood classifier was applied using composite bands from the three study years as input parameters; a critical advantage of parametric algorithms, such as MLC, is their ease of application. Parametric techniques are especially effective for less complex terrains with well-defined LULC categories

Producer Accuracy (PA) measures the percentage of correctly classified pixels for a given class. In contrast, User Accuracy (UA) indicates the likelihood that a pixel classified as a particular category genuinely belongs to that category. The ratio of the accurately classified reference points to the total reference points was considered to calculate the overall accuracy.

2.4. Accuracy Assessment

Accuracy can be assessed by comparing reference data with information obtained through remote sensing techniques [51,52], emphasizing the importance of the number and representativeness of training samples in image classification. Additionally, it is vital to collect ground reference data as close as possible to the date of the historical imagery [53]. Ref. [54] notes that reference data were systematically generated for accurate evaluation. Ref. [55] reported using a stratified random sampling method to generate reference points for accuracy assessment. They also noted that a confusion matrix was employed to evaluate the accuracy of each LULC map [56]. A confusion matrix was generated in ERDAS IMAGINE 2014 to compute user accuracy, producer accuracy, and overall classification accuracy. The outcomes and confusion matrix for various LULC classifications are presented in

Table 3. User’s accuracy, producer’s accuracy, and overall accuracy assessment of land use/land cover classification: 1992–2022.

LULC CLASS	1992			2010			2022
	UA	PA	OA	UA	PA	OA	UA	PA	OA
Built-Up Area	98.65	97.95	95.40	100	93.57	93.27	99.92	98.34	98.16
Vegetation	93.07	96.58		96.56	96.98		99.12	97.19
Agriculture	87.66	83.85		67.808	85.34		88.30	98.28
Open Land	91.86	88.76		75.21	95.65		100	100
Water Bodies	94.78	99.09		83.89	88.39		100	99.10

. The overall accuracy of the LULC maps is above 90%.

2.5. Derivation of Spectral Indices

2.5.1. Normalized Difference Vegetation Index (NDVI)

The NDVI is an indicator of relative greenness used to assess vegetation. It is also effective for monitoring vegetation productivity and identifying changes in vegetation [57]. Surfaces without vegetation exhibited negative values [58]. NDVI analysis was conducted using ArcGIS 10.8 to identify green areas within Ghaziabad and to validate land use changes detected from satellite imagery. The required bands were added in ArcGIS, and the calculation was performed using a raster calculator. NDVI was calculated as suggested by [59] using Equation (1).

$$NDVI={\displaystyle \frac{NIR-R}{NIR+R}}$$

(1)

Whereas R corresponds to band 3 in Landsat 5 (TM), it represents band 4 in Landsat 8 OLI and TIRS. On the other hand, NIR signifies band 4 in Landsat 5 TM and band 5 in Landsat 8.

2.5.2. Normalized Difference Built-Up Index (NDBI)

Previous studies have shown that high NDBI values indicate urbanized areas, whereas negative values represent non-urban features such as vegetation and water bodies [60]. In urban settings and exposed soil, reflectance in the Short-Wave Infrared (SWIR) spectrum exceeds that of the Near Infrared (NIR). The infrared spectrum did not reflect water bodies. NDBI is determined using reflectance values from the SWIR and NIR bands in remote sensing data. NDBI values vary from −1 to +1, with negative values indicating water bodies and higher values signifying built-up areas. Vegetation typically exhibits low NDBI values.

NDBI was estimated using the method described by [60] using Equation (2).

$$NDBI={\displaystyle \frac{NIR-SWIR}{NIR+SWIR}}$$

(2)

Band 5 in Landsat 5 (TM) and band 6 in Landsat 8 (OLI/TIRS) correspond to the SWIR band.

2.5.3. Normalized Difference Moisture Index (NDMI)

NDMI was derived by computing the normalized difference between the NIR and SWIR bands. It indicates the moisture content of vegetation and assesses the water stress level. By analyzing the NDMI’s absolute value, it is possible to pinpoint areas of a field or farm experiencing water stress.

As suggested by [61], the NIR and SWIR bands were used to calculate the NDMI. The calculation was performed using Equation (3).

$$NDMI={\displaystyle \frac{NIR-SWIR}{NIR+SWIR}}$$

(3)

The information regarding bands is mentioned above.

2.6. Retrieval of Land Surface Temperature (LST)

LST was computed over three years using thermal bands from Landsat satellite imagery. ArcGIS 10.8 software was used to retrieve LST. The LST retrieval process involved three main steps: (i) determining the brightness temperature, (ii) estimating the amount of vegetation (Pv), and (iii) correcting emissivity using NDVI [62]. This is among the most widely used methods that differentiates between vegetated and non-vegetated areas. It is also used in the areas of mixed land cover, where emissivity varies significantly. The entire LST retrieval methodology for the different Landsat sensors is described below.

2.6.1. Retrieval of LST for Landsat 5 (TM)

Step 1: Top of Atmospheric Radiance

The equation suggested by Artis and Carnahan was used to calculate spectral radiance, as shown in Equation (4).

$$L\lambda =\left\{{\displaystyle \frac{L_{MAX}\lambda -L_{MIN}\lambda }{Qca{l}_{MAX}-Qca{l}_{MIN}}}\right\}\left(Qcal-Qca{l}_{MIN}\right)+L_{MIN}\lambda$$

(4)

where Lλ stands for spectral radiance; L_MAX λ for the maximum spectral radiance of the satellite sensor, scaled to correspond with the maximum quantized calibration value (Qcal_MAX); and L_MIN λ for the minimum spectral radiance of the satellite sensor, scaled to match the minimum quantized calibration value (Qcal_MIN).

Step 2: Convert Radiance into Brightness Temperature (BT) in Kelvin

The next step involved applying Equation (5) to calculate the brightness temperature (BT) from the top of the atmosphere (TOA) reflectance [63].

$$BT=\left\{{\displaystyle \frac{K2 }{\mathrm{Ln}\left(\frac{K1}{L\lambda }+1\right)}}\right\}$$

(5)

In the above equation, Lλ represents spectral radiance, and BT represents brightness temperature expressed in Kelvin. The constants K1 and K2 are radiometric calibration parameters provided in

Table 4. K values of thermal bands of Landsat satellites.

Sr. No	Satellite/Sensor	Thermal Band	K1	K2
1	Landsat 5 TM	6	607.76	1260.56
2	Landsat 8 OLI/TIRS	10	774.8853	1321.07

.

Step 3: Conversion of Kelvin into Degrees Celsius

The temperature values were converted from Kelvin to degrees Celsius using the following Equation (6).

T(°C) = BT − 273.15

(6)

In this equation, T(°C) refers to Land Surface Temperature, and BT is the measured temperature in Kelvin.

2.6.2. Extracting LST from Landsat 8 Data

(i) Top of Atmosphere (TOA) Radiance

The radiance rescaling factor transforms thermal infrared digital numbers into Top-of-Atmosphere (TOA) spectral radiance, as shown in Equation (7).

$$L\lambda =M_L\times Qcal+A_L$$

(7)

Lλ represents TOA spectral radiance, expressed in watts per square meter per steradian per micrometer (W·m⁻²·sr⁻¹·μm⁻¹). Qcal represents the pixel’s quantized and calibrated DN values in the standard product. M_L is the radiance multiplicative scaling factor for the particular band, while A_L is the corresponding additive factor.

(ii) Conversion of Top of Atmosphere (TOA)/Brightness Temperature (BT)

Thermal calibration constants from the Metadata file were used to convert spectral radiance data into top-of-atmosphere brightness temperature, as shown in Equation (8).

$$\mathrm{B}\mathrm{T}=\left\{{\displaystyle \frac{K2}{\mathrm{Ln}\left(\frac{K1}{L\lambda }+1\right)}}\right\}$$

(8)

BT is the brightness temperature at the top of the atmosphere, measured in Kelvin Lλ is the is the top-of-atmosphere spectral radiance, measured in watts per square meter per steradian per micrometer {W/ (m² × r × μm)}. K1 and K2 are constants specific to the bands being evaluated.

(iii) Normalized Differential Vegetation Index (NDVI)

NDVI has been calculated as previously described in Equation (1).

(iv) Proportion of Vegetation Index

The proportion of vegetation (Pv) was calculated using Equation (9).

$$Pv={\left\{{\displaystyle \frac{NDVI-NDV{I}_{min}}{NDV{I}_{max}+NDV{I}_{min}}}\right\}}^2$$

(9)

In this context, PV is the proportion of vegetation, NDVI is the DN value from the NDVI image, and NDVI_min and NDVI_max are the minimum and maximum DN values from the NDVI image, respectively.

(v) Land Surface Emissivity (LSE):

LSE is a parameter that describes the average emissivity of the earth’s surface, as determined by NDVI values, and is based on the radiative properties of natural land features. It is calculated by using Equation (10).

E = 0.004 Pv + 0.986

(10)

P_V represents the vegetation proportion, and E stands for land surface emissivity.

(vi) Land Surface Temperature (LST)

The radiative temperature, known as LST, was estimated by utilizing the wavelength of the emitted radiation, land surface emissivity, and brightness temperature at the top of the atmosphere, as shown in Equation (11).

$$LST=\left[{\displaystyle \frac{BT}{1+(\lambda \times \frac{BT}{\rho })\mathit{ln}\left(E\right)}}\right]-273.15$$

(11)

BT is the brightness temperature measured in Kelvin, $\lambda$ is the wavelength of the emitted radiation, E denotes land surface emissivity, and the $\rho$ value is 14,380.

2.7. Estimation of Surface Urban Heat Islands (SUHIs)

UHI represents a prominent environmental issue that impacts major cities where the air temperature is considerably higher than in surrounding suburban areas. The UHI effect also contributes to poor air quality and atmospheric haze. High-density urban infrastructures significantly contribute to the UHI effect since they not only restrict airflow but also release heat from solar energy that has been captured. The loss of wetlands and vegetation further reduces the ability of cities to release heat. Since the UHI impact poses a serious concern to the environment and human health [64,65], minimizing and mitigating it has become essential. SUHIs have been used in numerous prior studies to investigate the spatial extent and intensity of UHIs [66,67,68,69].

The surface urban heat island (SUHI) was calculated by using Equation (12).

$$SUHI=\left[{\displaystyle \frac{\left\{Ts-Tmean\right\}}{SD}}\right]$$

(12)

In this equation, Tmean is the mean LST, SD is the standard deviation of LST; Ts is the LST at a specific location.

2.8. Modeling of Urban Hot Spots (UHSs)

The UHS was modeled to identify areas with the highest land surface temperatures (LST). UHS typically occurs in isolated areas of highly densely populated land, barren land, or open land that manifests unusually hot temperatures and is unsuitable for human habitation. The UHSs were modeled using Equation (13), following the method of [45].

UHS = [LST > µ + 2 × δ]

(13)

In this equation, µ denotes the mean LST, while δ is the standard deviation of LST within the study area.

2.9. Estimation of UTFVI

The Urban Thermal Field Variance Index (UTFVI) is an extensively used metric for quantifying the surface urban heat island effect [70]. UTFVI values are typically higher and concentrated in areas experiencing elevated surface temperatures compared to their surrounding environments [71]. This index categorizes Ghaziabad into six thermal ecological zones: excellent, good, normal, bad, worse, and worst, based on levels of human thermal comfort. These categories represent a gradient of thermal conditions, ranging from optimal to severely uncomfortable [72]. Equation (14) was used to model the UTFVI [73].

$$UTFVI=\left[{\displaystyle \frac{\left\{Ts-Tmean\right\}}{Tmean}}\right]$$

(14)

The comprehensive methodology is shown in Figure 2.

3. Results

3.1. Land Use/Land Cover Dynamics: Unveiling the Changes

LULC dynamics play a crucial role in shaping the ecosystem. The state of the natural environment is closely tied to LULC, as changes in land use directly affect the natural environment and contribute to broader climate change phenomena, consequently worsening public health concerns. The ongoing process of LULC change is shaped by various factors, including socioeconomic conditions, political influences, cultural aspects, demographic trends, and environmental considerations [74]. To investigate the spatiotemporal development of urban heat in connection with urbanization, the study was conducted for the years 1992, 2010, and 2022. Analyzing UHIs and their relationship with urban expansion helps to identify climate and environmental vulnerabilities in urban areas. The analysis focused on Ghaziabad in Uttar Pradesh. While Ghaziabad has experienced rapid urbanization, the city’s growth rate remains relatively high, at 70.22%, as depicted in Figure 3 [75].

Figure 4 presents the supervised LULC classification. Three land use maps for 1992–2022 were created using supervised classification and post-classification techniques on Landsat data. In 1992, agricultural land covered 52.52 sq. km (28.38%); built-up area, 53.64 sq. km (29.01%); vegetation, 47.61 sq. km (26.74%); open land, 24.8 sq. km (13.9%), and water bodies, 3.39 sq. km (1.90%) (

Table 5. Changes in land use/land cover for different classes in Ghaziabad: 1992–2022.

LULC Class	(1992). Area (km2)	1992 (%)	2010 Area (km2)	2010 (%)	2022 Area (km2)	2022 (%)	% Change (1992–2010)	% Change (2010–2022)	% Change (1992–2022)
Agriculture	52.52	28.38	66	37.8	49.45	27.78	9.42	–10.02	–0.60
Built-up Area	51.64	29.01	79.17	44.48	106.3	59.76	15.47	15.28	30.75
Vegetation	47.61	26.74	8.29	4.66	11.65	6.55	–22.08	1.89	–20.19
Water Bodies	3.39	1.9	2.77	1.55	1.9	1.06	–0.35	–0.49	–0.84
Open Land	24.8	13.9	21.72	12.2	8.6	4.83	–1.70	–7.37	–9.07
Total Area	177.9		177.9		177.9

). The built-up areas in Ghaziabad were primarily found in the southwest, south, and central regions. The northeastern and riverbank regions of Ghaziabad were primarily covered by vegetation. Small pockets of open ground were found in Ghaziabad’s southern and northern regions.

The LULC trends recorded in Ghaziabad between 1992 and 2010 are shown in Table 5. There was a 9.42% increase in the built-up areas. In contrast, all classes, including vegetation, water bodies, and open land, showed declines of −22.08%, −0.35%, and −1.7%, respectively. This demonstrates that vegetation was the most adversely impacted land use category. The primary cause of these declines was the ongoing expansion of both agricultural and downtown areas, and an increase in population, housing demand, and agricultural development, which has turned open lands and natural vegetation into urban and agricultural zones.

Table 5 shows the patterns and areas of various LULC classes for 2022. Built-up areas are the only category that has consistently increased. This contributed 59.76 percent. Its area increased from 79.17 sq km in 2010 to 106.30 sq km in 2022, as presented in Figure 5 and Figure 6. Except for a slight increase in vegetation, all other land use categories showed a decline by 2022. Over the whole study period, four of the LULC classes—vegetation, water bodies, open land, and agriculture—have observed significant declines of −20.19%, −0.84%, −9.07%, and −0.6%, respectively, while built-up areas have increased (+30.75).

3.2. Transformation of Land Use/Land Cover (1992–2022)

Land use/land cover (LULC) in Ghaziabad experienced significant changes between 1992 and 2022, as shown in Table 5 and Figure 7. This indicates that the built-up area grew steadily for all three periods: 1992, 2010, and 2022. The built-up area recorded a consistent growth across the three study periods, increasing from 53.64 sq. km (29.01%) in 1992 to 79.17 sq. km (44.48%) in 2010, and further to 106.30 sq. km (59.76%) in 2022, with a total positive change of 30.75%. This represents the most significant increase among all LULC categories. However, during the same period, there was a decline in agricultural land (−0.60%), vegetation (−20.19%), water bodies (−0.84%), and open land (−9.7%). Among these, vegetation experienced the most significant decline.

Temporally, built-up areas increased from 29.01% in 1992 to 44.48% in 2010 and 59.76% in 2022, as shown in Figure 4. A downward tendency in open land was observed: from 13.90% in 1992 to 12.20% in 2010, then to just 4.83% in 2022. Additionally, vegetation cover declined from 26.74% in 1992 to 4.66% in 2010, followed by a slight increase to 6.55% in 2022. Water bodies also experienced a decline, from 1.90% in 1992 to 1.55% in 2010, and further to 1.06% in 2022. Anthropogenic influences were primarily responsible for this decline, including urbanization and infrastructure expansion. Moreover, agricultural land increased from 28.38% in 1992 to 37.80% in 2010, but declined again to 27.78% by 2022. The growth of urban settlements between 2010 and 2022 was a significant factor in the decline in agriculture.

Figure 4 shows the spatiotemporal dynamics of LULC in Ghaziabad in 1992, 2010, and 2022. Vegetation in the northeastern part in 1992 was transformed mainly into agricultural and built-up areas by 2022. Meanwhile, open land has also decreased continuously from 1992 to 2022. The built-up area has experienced a significant increase during the study period, as illustrated in Figure 5. Agriculture is mainly distributed in the surrounding areas. Its predominance was on the eastern and northeastern sides. Water bodies appear to have decreased from 1992 to 2022. Most built-up areas were converted from open vegetation and agricultural land.

Figure 6 provides a clear and accessible visualization of LULC changes between 1992 and 2010; vegetation was substantially converted into agricultural land and open areas, as shown in Figure 6. Meanwhile, the LULC class showed a slight recovery between 2010 and 2022. The built-up area was the only category that grew consistently during the study period. Owing to the transformation of the majority of vegetation into agricultural land, there was an increase in agricultural land from 1992 to 2010. The water bodies consistently shrank during the study period. The amount of open land steadily decreased throughout the study period. Open land is also a key factor for the high LST in built-up areas. Contributing to elevated LST in built-up areas, exacerbating the formation of UHIs in Ghaziabad. Figure 7 presents a more detailed representation of LULC change detection.

3.3. Spatiotemporal Analysis of Spectral Indices in Ghaziabad: 1992–2022

3.3.1. Spatiotemporal Analysis of Normalized Difference Vegetation Index (NDVI) in Ghaziabad: 1992–2022

The spatial distribution of NDVI values across Ghaziabad is illustrated in Figure 8. The NDVI values range from −0.13 to 0.63, −0.11 to 0.51, and 0.03 to 0.49 in 1992, 2010, and 2022, respectively, as shown in

Table 6. NDVI values in Ghaziabad: 1992–2022.

Date	NDVIMax	NDVIMin	NDVIMean
16 March 1992	0.63	−0.13	0.21
3 April 2010	0.51	−0.11	0.05
4 April 2022	0.49	0.03	0.17

. In 1992, higher NDVI values were primarily associated with areas of dense vegetation and active agricultural practices, particularly in the northeastern and eastern regions. In contrast, in 2010 and 2022, increased NDVI values were more distinctly observed along riverbanks and surrounding Hindon Airport. The NDVI values are closely linked to LULC, with the highest values observed over vegetation and agricultural areas and the lowest values over built-up areas, open land, and water bodies.

3.3.2. Spatiotemporal Variation in Normalized Difference Built-Up Index (NDBI) in Ghaziabad: 1992–2022

The distribution and variation of NDBI values across built-up areas are presented in

Table 7. NDBI values in Ghaziabad: 1992–2022.

Date	NDBIMax	NDBIMin	NDBIMean
16 March 1992	0.62	−0.33	0.07
3 April 2010	0.39	−0.29	0.17
4 April 2022	0.17	−0.37	−0.03

and Figure 8. The NDBI values range from − 0.33 to 0.62, − 0.29 to 0.39, and − 0.37 to 0.17 for 1992, 2010, and 2022, respectively. The mean NDBI values are presented. The mean NDBI values were 0.07 in 1992, 0.17 in 2010, and 0.03 in 2022. The built-up area in Ghaziabad increased over the study period. NDBI reached the highest individual value in 1992 (maximum of 0.62), though the mean peaked in 2010. The distribution of values can be seen over the southern, central, southwestern, and some northeastern parts of the research area because the cultivation of crops has been done, and the agricultural lands are vacant, which is why these areas show high values. For 2010, the NDBI values are lower than in 2022, but the distribution is mainly in the southern, central, and southeastern parts. The lowest NDBI values are obtained for 2022, concentrating primarily in Ghaziabad’s south, southeastern, and central areas. This pattern was impacted by industrialization and the expansion of transportation infrastructure during that period. The expansion of built-up areas has had a marked impact on the thermal environment of Ghaziabad, primarily due to the conversion of agricultural land into impervious surfaces.

3.3.3. Spatiotemporal Variation in Normalized Difference Moisture Index (NDMI) in Ghaziabad: 1992–2022

There is a wide range of NDMI values and variations during the study period. NDMI values fluctuated from −0.62 to 0.33 in 1992, −0.39 to 0.29 in 2010, and −0.17 to 0.37 in 2022, as shown in

Table 8. NDMI values in Ghaziabad: 1992–2022.

Date	NDMIMax	NDMIMin	NDMIMean
16 March 1992	0.33	−0.62	−0.07
3 April 2010	0.29	−0.39	−0.17
4 April 2022	0.37	−0.17	0.03

and Figure 8. The average NDMI values for 1992, 2010, and 2022 were −0.07, −0.17, and 0.03, respectively. The NDMI value in the study area for 1992 was higher than in 2010 but was lower than in 2022. Positive NDMI values were primarily observed near the riverine vegetation, while the entire study area showed negative values. The year 2010 had the lowest NDMI values, indicating minimal moisture content in the study area, primarily due to the prevalence of exposed open land. Only the Hindon River exhibited positive NDMI values, with most of the study area showing negative values for 2010. By 2022, the NDMI value had increased, indicating improved moisture conditions due to vegetation and agricultural land. Higher NDMI values in 2022 were observed around the river area, as well as in the upper southwestern part, and the eastern and northeastern regions.

3.4. The Spatiotemporal Variation in Land Surface Temperature in Ghaziabad: 1992–2022

The spatiotemporal distribution of LST in the Ghaziabad area is illustrated in Figure 9 below. According to

Table 9. LST in Ghaziabad.

Date	LSTMax	LSTMin	LSTMean	LSTSD
16 March 1992	28.35	14.24	20.81	2.14
3 April 2010	46.30	23.68	34.45	2.24
4 April 2022	41.61	25.64	35.01	1.77

, the LST values ranged from 14.24 to 28.35 on 16 March 1992, 23.68 to 46.30 on 3 April 2010, and 25.64 to 41.61 on 4 April 2022, as presented in Table 9. The mean LST values were 20.81 in 1992, 34.45 in 2010, and 35.01 in 2022. In 1992, the highest LST was observed in the southern, western, and southwestern parts of the study area. By 2010, the hot spots had shifted towards the southern, northern, northwestern, and southwestern regions. In 2022, the highest LST values were predominantly concentrated in Ghaziabad’s southern, southeastern, eastern, and northeastern parts.

3.5. The Spatiotemporal Variation in Surface Urban Heat Islands in Ghaziabad: 1992–2022

The temporal and spatial variations in the SUHI intensity across Ghaziabad are shown in Figure 9. The SUHI values ranged from −3.06 to 3.52 in 1992, −4.80 to 5.29 in 2010, and −5.28 to 3.77 in 2022. These ranges indicate a general intensification of SUHI effects over the 30-year study period. By 2022, the highest SUHI values were observed in the central and southern regions of Ghaziabad, with noticeable SUHI hot spots (high values) also emerging in the eastern and southeastern areas. These elevated SUHI values are mainly due to the concentration of built-up areas and limited vegetation cover in these regions. In contrast, areas dominated by water bodies, vegetation, and agricultural land consistently exhibited lower SUHI values.

Temporally, in 1992 and 2010, high SUHI values were largely associated with exposed open land. Whereas, by 2022, UHIs were primarily concentrated in densely built-up and populated areas. The high-intensity SUHI area progressively expanded, with the central, western, and southwestern regions of Ghaziabad experiencing the highest levels. The urban core of Ghaziabad exhibits progressive SUHI intensification over three decades: from relatively low intensity in 1992, through moderate levels by 2010, to very high intensity by 2022—with central areas experiencing the most severe thermal effects as a direct result of cumulative urban expansion and densification.

3.6. LST Distribution in Different LULC Classes

The spatiotemporal patterns of LST across various land use/land cover (LULC) categories from 1992 to 2022 are shown in Figure 10. The analysis indicates a consistent increase in average LST across all LULC classes during this period, primarily due to the disproportionate expansion of urban areas compared to green spaces. Notably, the vegetation category experienced an increase in mean LST, peaking in 2010, likely associated with the harvesting practices and a limited expansion in vegetation cover. The mean LST for built-up areas increased from 21.94 °C in 1992 to 35.64 °C in 2022. For agricultural areas, LST increased from 20.21 °C in 1992 to 34.18 °C in 2022. Although the absolute values are significantly lower, water bodies also showed an increasing trend in LST. Vegetation and open land displayed inconsistent but overall increasing LST patterns over the study period. Specifically, the mean LST for open land was 21.18 °C, which rose to 35.28 °C in 2010, followed by a slight decrease to 34.91 °C in 2022. For vegetation, the mean LST was 20.05 °C in 1992, increasing to 34.48 °C in 2010, then slightly declining to 33.67 °C in 2022. In summary, all LULC categories exhibited a net increase in LST from 1992 to 2022.

3.7. Urban Thermal Field Variance Index (UTFVI)

The UTFVI is a metric used to characterize the urban thermal environment and assess thermal comfort within urban contexts [76]. It is used to quantitatively evaluate the Urban heat island (UHI) phenomenon [77]. As noted by [73], UTFVI values are categorized into six thermal ecological levels. For the years 1992, 2010, and 2022, the UTFVI values for Ghaziabad were classified into six groups based on thermal comfort levels (

Table 10. UTFVI thresholds and temporal variation in Ghaziabad.

UTFVI	Urban Heat Island Phenomenon	Ecological Evaluation Index (EEI)	Proportion of Area in Percentage
			1992	2010	2022
<0	None	Excellent	81.3942	80.4438	83.4669
000–0.005	Weak	Good	0	14.7564	0
0.005–0.010	Middle	Normal	0	0	8.4681
0.015–0.015	Strong	Bad	13.0698	13.608	8.3187
0.015–0.020	Stronger	Worse	0	0	8.4015
>0.020	Strongest	Worst	83.5731	69.2289	69.381

) and are illustrated in Figure 11. The UTFVI map shows the spatiotemporal distribution of UTFVI extremes (excellent and worst) of thermal ecological assessment. Areas that are significantly warmer than their surroundings exhibit higher UTFVI concentrations. The UTFVI is widely employed to assess urban ecological quality, particularly emphasizing thermal comfort and UHI intensity. The figure indicates a shift in UHI intensity toward the southeastern region of Ghaziabad. In contrast, the southern region continues to display persistently high thermal stress due to dense built-up areas. Furthermore, the 2022 data show the emergence of smaller patches of middle-to-strongest UHI intensity across Ghaziabad, a pattern not evident in 1992.

3.8. Identification of Urban Hot Spots (UHSs)

The spatiotemporal variation in UHSs across the study area is illustrated in Figure 12. In 1992, small UHS patches were scattered across various parts of Ghaziabad, primarily associated with open land and emerging built-up zones. In 2010, Ghaziabad’s eastern and southern regions showed a higher concentration of UHSs. There are also some visible spots in the northwestern part of Ghaziabad. These hot spots were still driven by the widespread presence of open land. In 2022, the spatial distribution of UHSs shifted more toward the southern and eastern parts of Ghaziabad. At this stage, the primary driver of UHS formation was the high concentration of built-up land, replacing earlier open spaces.

4. Exploring the Interconnections of Land Use/Land Cover, LULC Indices, Land Surface Temperature (LST), and Urban Heat Islands (UHIs)

Land surface typologies and spatial configurations significantly affect the spatial distribution of UHIs. The pattern and distribution of LULC directly affect UHI intensity [78].

4.1. The Correlation Between LULC and LST

LST is strongly affected by the type and extent of LULC. Changes among various LULC types, especially the expansion of urban areas, exacerbate the UHI effect by increasing the number and distribution of thermal hot spots [30]. LST spatial patterns across Ghaziabad were created using Landsat thermal band data (digital numbers 0–255) and classified through the standard deviation method [79]. Table 9 presents the mean LST values for each LULC category in the years 1992, 2010, and 2022. The highest average LST values were observed in settlement areas, followed by open land, agricultural fields, vegetated areas, and water bodies.

4.2. The Correlation Between Spectral Indices and LST: 1992–2022

The linear relationships between LST and the spectral indices NDVI, NDBI, and NDMI used in the study are shown in Figure 13. A strong inverse correlation exists between NDVI and LST, such that decreases in NDVI values are associated with increases in LST. Numerous studies confirm that vegetation and LST are inversely correlated. For instance, NDVI and LST values in this study exhibited a negative relationship across all years [80,81]. Higher NDVI values are observed in Ghaziabad’s northern, southern, and southeastern parts and along the river. However, these areas correspond to regions with dense vegetation and extensive agricultural activity. Conversely, LST values were lower in areas dominated by vegetation and agricultural land. The highest LST values were observed mainly because of the decline in open land, agricultural land, and vegetation areas. Vegetation reduces thermal radiation and regulates surface temperature [82].

Pearson correlation analysis was employed to examine the relationship between LST and NDVI using Landsat data for 1992, 2010, and 2022. The analysis showed that areas with higher surface temperatures corresponded to lower NDVI values, confirming a negative correlation between vegetation cover and LST. The coefficients of determination (R²) for the correlation between LST and NDVI were 0.53 in 1992, 0.24 in 2010, and 0.31 in 2022. These results are graphically represented in Figure 13.

LST and NDBI have a positive association, which means that the growth of built-up areas causes an increase in LST. High NDBI values were predominantly concentrated in the central, eastern, and southeastern regions of Ghaziabad. These regions also corresponded with the highest LST values, reinforcing the positive association between built-up intensity and surface temperature. This increase in built-up area is primarily driven by industrial development and the conversion of open land, agricultural fields, and vegetated areas into impervious surfaces.

Pearson correlation analysis was conducted to examine the relationship between LST and NDBI, derived from Landsat data for 1992, 2010, and 2022, respectively. The results indicate higher NDBI values in regions with high temperatures, confirming a positive correlation between LST and NDBI. The correlation coefficients (R²) were 0.57 for 1992, 0.38 for 2010, and 0.35 for 2022. These trends are illustrated in Figure 13 where the scatter plots show a consistent increasing relationship between LST and NDBI.

The same correlation analysis assessed the relationship between NDMI and LST. A negative correlation was observed, indicating that higher surface moisture levels are associated with lower surface temperatures. This relationship is visually depicted in Figure 12.

5. Discussion

This study analyzed the spatiotemporal dynamics of LULC, their impact on the development of SUHIs, and their relationship with urbanization in Ghaziabad between 1992 and 2022, using multi-temporal Landsat datasets. Ghaziabad has undergone dramatic transformations in its LULC pattern, which have notably affected SUHI dynamics [39,83]. The analysis revealed marked spatiotemporal variations in all evaluated land use indices, including NDVI, NDBI, and NDMI. These changes are primarily attributed to urban expansion, increasing built-up areas that retain less moisture than natural surfaces [76,84]. This study explores how temporal shifts in LULC affect LST and SUHI patterns. The analysis was focused on detecting LULC changes in Ghaziabad and quantifying the SUHI phenomenon using LST derived from GIS and remote sensing. The research addressed the following objectives: (1) to assess the relationship between LULC indices and LST; (2) to investigate and track spatiotemporal variations in LULC; (3) to examine the dynamics of LST and SUHIs; and (4) to identify and analyze urban hot spots (UHSs).

Several studies have focused on the spatiotemporal dynamics of LULC [85,86,87,88,89] and their impact on surface urban heat islands [5,23,90,91,92,93,94]. Analysis of LULC changes revealed a significant transformation in land surface composition throughout the study period. This transformation was primarily driven by urban expansion, which reduced vegetation cover, open land, and water bodies [95]. Similar results were observed in a study on Sivas city and also some cities in India, such as Delhi, Noida, Meerut, Mumbai, etc., where vegetated areas declined between 1992 and 2010, followed by an increase from 2010 to 2022 [89,96,97,98]. NDVI values in 1992 showed substantial spatial variability, whereas in 2022, overall NDVI values were considerably lower. However, the notable NDVI increase observed in specific areas during 2022 may be attributed to plantation initiatives in the Delhi NCR region. NDBI values exhibited notable spatial variation across the study area in 2010, whereas changes were minimal in 1992, reflecting limited urban development in the earlier period. Therefore, to reduce the detrimental impacts of climate change, including UHIs, legislators and urban planners must prioritize integrating green spaces into city planning [99].

The maximum, minimum, and mean values of LST in Ghaziabad showed a significant fluctuation for all three datasets used in this study. Temporally, all land use classes have shown a noticeable variation. Built-up area was the category that experienced the highest temperature during both datasets, followed by open land, agriculture, vegetation, and water bodies, except for the data of 3 April 2010. Open land observed higher LST than built-up area in 2010 due to exposure of more open hard surfaces, concrete materials, and brick kilns. Similar patterns of vegetative decline and its thermal consequences in cities/urban areas around the world [78]. Studies suggest that reduced (diurnal/daily) temperature variation and rising surface temperatures, characteristic of UHIs, are key indicators of urbanization’s contribution to climate change [6].

In Ghaziabad, a significant shift in the SUHI pattern was observed between 1992 and 2010. These changes were primarily linked to barren surfaces and expanding settlements, particularly in the central, southern, and southwestern regions, where built-up land cover was extensive. Like SUHI patterns, urban hot spots (UHSs) also exhibited temporal and spatial shifts across Ghaziabad. These areas recorded the highest surface temperatures, largely due to heat-retaining materials such as concrete and asphalt [44]. The development of UHSs in Ghaziabad was primarily associated with impervious concrete surfaces and densely settled urban regions. Urban hot spots (UHSs) expanded from small, scattered patches in 1992 to a more widespread distribution across Ghaziabad by 2010. However, their highest concentration remained localized in specific built-up zones. However, in 2022, it is clear that inhabited regions have a notably higher proportion of UHSs.

This study also assessed how changes in LULC influenced LST patterns across the study area. Three geographical indices—the NDBI, NDVI, and NDMI—were used to evaluate their correlations with LST. Correlation analysis revealed a significant positive association between LST and NDBI, indicating that urban expansion contributes directly to increased surface temperatures. The continued expansion of urban features such as concrete and asphalt surfaces results in greater absorption of solar radiation, thereby elevating LST. In contrast, both NDVI and NDMI exhibited negative correlations with LST, as higher vegetation and moisture levels help to moderate surface temperatures [66,92].

The intensity of the UHI phenomenon has significantly increased in Indian cities, while urban thermal comfort has declined due to expanding impervious surfaces and the loss of vegetative cover [29,78]. Therefore, to reduce UHI impacts and enhance urban thermal comfort, it is crucial to integrate green infrastructure over barren surfaces and throughout residential and industrial areas. Urban planners and municipal authorities can utilize the findings of this study to inform the spatial planning and distribution of green spaces in Ghaziabad.

6. Conclusions

The current study used different Landsat datasets to evaluate the impact of urbanization on the LULC change of Ghaziabad, as well as the extent to which the observed changes in LULC can be connected to the spatiotemporal pattern of the LST and SUHIs in Ghaziabad during 1992–2022. Accuracy assessments of the findings demonstrated that over the study period, there was a decline in open land, water bodies, vegetation, and agriculture, and a dramatic increase in built-up areas. The primary catalyst for this expansion of built-up areas is the gradual transition from rural to urban residential communities. The city’s central, southern, and southwestern regions have seen the most growth. The findings also demonstrate the significant influence of all land use indices, including NDVI, NDBI, and NDMI on SUHIs. These also have an apparent association with Ghaziabad’s LST, with built-up area having a more significant influence from 1992 to 2022 than the others. The results obtained also reveal a different pattern of LST in Ghaziabad. Between 1992 and 2022, there was a consistent change in the SUHI and UHSs. The categories designated as “worse” and “worst” thermal ecological zones in the UTFVI showed a notable increase during the study period, signifying a deterioration in the thermal ecological conditions within Ghaziabad from 1992 to 2022. This study indicates that swift urban expansion contributes to the decline of the urban thermal environment in Ghaziabad. Though some regions of Ghaziabad have seen an increase in vegetation in recent years, the central, southern, and southwestern regions of Ghaziabad have experienced a rapid shift due to an unprecedented pace of urbanization.

Considering the study’s findings, Ghaziabad has a problem with UHIs and other related issues. As per the findings, locations with high LST and UHIs, such as industrial and densely populated areas, can be targeted for mitigation initiatives, including green roofs and lighter-colored surfaces. Determining UHIs is also essential for implementing appropriate mitigation measures, and the government should act immediately to address the urban heat island region by taking the steps needed. To achieve more accurate outcomes, future researchers are advised to consider other factors and locational impacts in their studies on LST and UHIs. This study has several limitations, such as using only three time series data; however, more time series data would be ideal for the analysis. If longer-period series data were used in the study, this would produce a better image of the UHI and UHS.