Evaluating the Impact of Urban Encroachment and Land Cover Changes on World Cultural Heritage Site Taxila: A Spatio-Temporal Analysis from 1990 to 2024

Abstract

Rapid global urbanization during the late 20th and early 21st centuries has induced substantial land cover changes, posing significant threats to the United Nations Educational, Scientific and Cultural Organization’s (UNESCO) World Heritage Sites. In this study, we investigated the spatio-temporal change in urban development in response to land use transformations in the world cultural heritage site (CHS) of Taxila, Pakistan, to check the possible threats faced by the site. Land transfer matrices were used to assess the land cover change (LCC) between 1990 and 2024. Support vector machine and Getis–Ord Gi techniques were employed for LCC classification and spatial pattern interpretation, respectively, which were later evaluated by the high spatial resolution imagery of KH-9 (Keyhole-9), Google Earth Pro and Gaofen-2. The results indicate a significant increase in built-up area from 23.68 km² to 78.5 km², accompanied by a substantial rise in bare land from 8.56 km² to 26.5 km² between 1990 and 2024, which is quite irregular. LCC transformations were notable, with 13.1 km² of cropland and 44.8 km² vegetation being converted into 4.4 km² of built-up area and 14.5 km² into bare land during the 1990 to 2024 period. Getis–Ord Gi analysis observed a high Z-score value and showed low to high clustering patterns in the proximity of the Sarakhola and Bhir Mound sites from 1990 to 2024. Furthermore, high spatial resolution imagery indicates the loss of the core zone of the Sarakhola site from 0.0168 to 0.0032 km² from 2004 to 2024, which was the major threat to its outstanding universal venue (OUV) status. The findings of the current study indicate that the CHS under study is facing an alarming situation for conservation due to rapid urban development and encroachment. Therefore, local government should strictly implement the heritage law and revisit their policies to promote conservation efforts to maintain the authenticity and integrity of this world CHS.

1. Introduction

The world experiences significant land cover changes influenced by urbanization, infrastructure development, deforestation and resource exploitation, illustrating a dynamic relationship of anthropogenic activities [1]. The global population has been experiencing a significant surge in urbanization over the past two decades [2]. A similar pattern has been seen in the world cultural heritage sites in Taxila, Pakistan, which faces significant threats from rapid urbanization and unplanned infrastructure development [3]. Since 1990, the Taxila region’s population has more than tripled, driving profound land cover changes through unplanned urban development and infrastructure growth [4]. This unprecedented urban growth has led to extensive land cover changes that threaten both the physical integrity and the historical authenticity of this invaluable cultural heritage [5]. In 2010, the global urban population accounted for 51% of the worldwide population. This trend is projected to continue, with estimates suggesting that by the year 2050, a staggering 80% of the world’s population will reside in urban areas [6,7]. This rapid growth of urban populations exerts immense pressure on land resources, driving significant land cover transitions as cities expand the demand for housing, road connectivity and industry, and transforming natural landscapes into artificial surfaces [8,9]. This process pressurizes the natural landscape and strains existing resources [10], leading to the loss of archaeological relics that cannot be fully restored once destroyed [11,12].

Cultural heritage (CH) represents an invaluable collection of human history, transmitting knowledge, values and identities across generations. The preservation of CH, particularly at World Heritage Sites, ensures the continuity of cultural identity, historical authenticity, and integrity for future generations [13,14,15]. The preservation of the archaeological landscape emphasizes the outstanding universal value (OUV) of heritage sites, which highlights cultural identity and promotes a sense of belonging [16,17]. Furthermore, CH acts as a bridge between societies’ continued existence and evolution by transmitting knowledge, values and identities [18].

Addressing these issues is vital to fulfilling global conservation commitments, particularly in the context of the United Nations’ Sustainable Development Goals (SDGs) [19]. This commitment was further reinforced in 2015 when the United Nations adopted the Sustainable Development Goals (SDGs), including Target 11.4, which focuses on protecting and conserving cultural and natural heritage as a foundation of sustainable development. However, translating these global objectives into local action remains a challenge, as evidenced by UNESCO’s reports on the world cultural heritage sites of Taxila in Pakistan [20,21]. The United Nations Educational, Scientific and Cultural Organization (UNESCO), through the World Heritage Convention (WHC), has long emphasized the global importance of cultural heritage (CH) and has established a framework for its preservation [22]. A review of UNESCO’s reports from 1995 to 2013 (1995, 1999, 2000, 2013) reveals ongoing challenges in preserving Taxila’s archaeological landscape. While the 1995 report highlighted uncontrolled vegetation as a threat [23], later reports suggest this issue persisted [24]. New concerns also emerged, including urban development and encroachment and urban utilities, affecting the heritage site’s integrity and authenticity. Overall, these reports reveal a continuous effort to manage Taxila effectively, and UNESCO potentially assists the local government in preserving Taxila’s cultural heritage [23,24,25,26,27].

The natural threats and anthropogenic activities indicated by the studies have affected World Heritage Sites. Natural threats such as soil erosion, landslides, sea level rise and human activities like urban development, tourism, and warfare significantly contribute to the destruction of cultural heritage sites [28]. Urban development is the primary drive for change in the area’s land cover, landscape and road networks that threaten World Heritage Sites, which are reported in the endangered list of world heritage by the World Heritage Committee [29,30]. Most of the world’s cultural heritage listed sites are located in urban areas, making them susceptible to the challenges posed by urban development [31].

Remote sensing introduces modern techniques to archaeological investigations, providing high-resolution multispectral data for mapping and monitoring features and landscapes [32]. This technology offers detailed high spatial resolution and allows for the precise detection, monitoring and mapping of archaeological features and spectral information of buried archaeological features [33]. Li, Guangdong et al. highlighted urban expansion’s potential threat to cultural heritage sites [34]. Urbanization patterns and their effects on cultural heritage sites were examined by Gwendolyn et al. through the analysis of aerial photographs, satellite imagery and archaeological archives. Similarly, in 2018, a study conducted by Zhang et al. examined the extent of urbanization using nighttime data (NTD) around 152 Buddhist monasteries across the Tibetan plateau [35]. In 2020, Xiao et al. calculated the Urban Intensity Index (UII) globally, utilizing nighttime data, human settlement patterns and world population statistics [36]. Similarly, a 2023 study by Nath et al. revealed that Kaziranga Park in India lost grasslands, water and cropland over 30 years with rising built-up areas, threatening its UNESCO World Heritage status [37]. Guo et al. (2023) found that 70% of World Heritage Sites faced worsening land cover change between 2015 and 2020, which threatened conservation efforts for these cultural treasures [38].

Despite the global success of such approaches, these methodologies have not been systematically applied to Taxila, leaving critical gaps in understanding and managing its conservation needs. Pakistan’s rapid economic development is driving urban development, placing significant pressure on land resources and potentially threatening the World Heritage Site of Taxila [25,39]. Since 1990, Taxila’s population has increased more than three times, leading to significant land cover changes driven by unplanned urban development and infrastructure growth. These pressures have altered the site’s surroundings, eroded its cultural landscape, and introduced challenges that hinder its effective preservation. A review of UNESCO’s reports on Taxila (1995–2013) highlights persistent threats, including urban encroachment, and urban utilities, which compromise the site’s integrity. Despite these well-documented challenges, no comprehensive studies have explored the spatio-temporal dynamics of land cover changes or assessed their specific impact on the Taxila World Heritage Site. This gap in research limits the ability to develop targeted conservation strategies to mitigate the effects of urban expansion. Therefore, there is a need to monitor, investigate, and highlight the threats caused by urban development to Taxila’s world cultural heritage. So, considering the new emerging issues, such as urban development and the conservation literature, this study aims to (1) monitor spatio-temporal land cover change scenarios from 1990 to 2024; (2) determine the land cover transition and its trends; and (3) determine the spatio-temporal change in built-up area around the core zone of the world cultural heritage site of Taxila using hot spot and cold spot analysis. By focusing on Taxila within the broader context of global urbanization and SDG-11.4, this research provides actionable insights to preserve its archaeological landscape while contributing to the global dialogue on heritage conservation.

2. Study Area and Data

2.1. Study Area

The study area focuses on Taxila (Figure 1), designated as a UNESCO world cultural heritage site in the Pothohar region, approximately 25 km from Rawalpindi, Punjab, Pakistan. Taxila’s strategic location in the subcontinent serves as a pivotal nexus connecting the western areas of China to the eastern part of Pakistan, Afghanistan, and further southwest to Iran. Its landscape consists of a wide valley surrounded by notable hills and ridges, including the Murree Hills, Hazara Hills, and Margalla Hills. The city is situated at an elevation of approximately 549 m (1801 feet) above sea level. It experiences a subtropical climate characterized by hot summers with heavy monsoon rains, and the temperature ranges from 15 °C to 40 °C during the warmer months. In contrast, cooler temperatures in the region range between 5 °C and 15 °C during the winter. These geographical and climatic conditions significantly influence the area’s ecological dynamics and human activities, making them important considerations for conservation planning.

Taxila’s historical significance is equally remarkable. Its rich heritage dates back to prehistoric times, as evidenced by the Khanpur Cave, which reveals traces of early human existence. During the late Neolithic era, early urban settlements (Sarakhola) marked a significant development in the subcontinent [40]. Additionally, the historic entry of Alexander the Great into the Bhir Mound in Taxila established a crucial connection between Taxila and the ancient Silk Road, extending into the western regions of China within the subcontinent. These pivotal events, alongside the early formation of urban settlements in Taxila, serve as criteria for UNESCO to designate the cultural heritage sites for outstanding universal status [41]. Notably, Chinese Buddhist pilgrims travel this route from west China to Taxila for pilgrimage purposes. Moreover, it served as the primary gateway to enter the subcontinent from the west, highlighting its historical significance [42,43].

In recent years, the population of Taxila has increased from 222,840 (1990) to 739,244 (2023) [4], nearly tripling in the last three decades. This upsurge has put pressure on existing resources and caused land cover changes around the world CHS of Taxila. Therefore, comprehensive monitoring and change detection were necessary at this site to initiate appropriate conservation measures for the preservation of this cultural heritage asset.

2.2. Remote Sensing Imagery

The analysis of land use and cover is critical to this study. Landsat satellite series datasets were employed to assess the dynamics of land cover change at Taxila spanning from 1990 to 2024, with intervals of ten years. These datasets were acquired from the United States Geological Survey (USGS) Earth Explorer website [44], as outlined in the provided

Table 1. Multispectral satellite dataset description.

Imagery	Row/Path	Satellite	Sensor	Acquisition Time	Cloud Cover (%)	Total Scenes
1990	150,037	Landsat 5	TM	Jan–March	<3	3
2000	150,037	Landsat 7	ETM	Jan–March	<2	4
2010	150,037	Landsat 5	TM	Jan–March	<3	4
2024	150,037	Landsat 8	OLI	Jan–March	<2	4

. Late January and March were selected as target months due to their typically favorable atmospheric conditions and reduced haze and cloud coverage in the study area. The satellite images were employed to analyze the impact of land cover changes on Taxila over three decades.

It was taken care that downloaded imageries met the minimum criteria requirements such as cloud coverage, target months free from scan line error, clear study area and good quality. The sensor specifications were a thematic mapper (TM), enhanced thematic mapper and operational land imagery (OLI). The spatial resolution of all imageries was 30 m. Additionally, historical remote sensing images (declassified KH-9) were obtained from the USGS Earth Explorer. Furthermore, high spatial resolution imagery from Maxar satellites (Google Earth Pro) and Gaofen-2 from Natural Resources Satellite Sensing were utilized. The Gaofen-2 images underwent a pan sharpening process using the Gram–Schmidt technique to achieve a high spatial resolution of 1 m the details of the satellite imageries are presented in

Table 2. Details of the high spatial resolution satellite imagery.

Date	Satellite	Spatial Resolution
10 September 1997	KH-9 Hexagon	0.6–1.2 m
15 July 2004	Maxar satellite	0.3–15 m
28 December 2015	Gaofen-2	0.78–4 m
24 January 2024	Gaofen-2	0.78–4 m

.

2.3. Satellite Imageries and Classification

The USGS currently offers surface reflectance data, which undergoes pre-processing steps, including radiometric calibration and atmospheric correction. This dataset is instrumental in assessing land cover changes with enhanced accuracy and reliability [45,46]. For land cover change classification, the training data were collected with different spectral band combinations to the image with the help of historical Google Earth Pro software 7.3 [47], a high multispectral spatial resolution Gaofen-2 image and background knowledge of the study area. Subsequently, the image underwent classification into six distinct land cover classes, namely built-up area, tree cover, cropland, vegetation, water bodies, and bare land, by applying supervised classification methodologies. A sufficient number of training samples were gathered for each class to ensure robust representation across all land type classes. Additionally, satellite image enhancement was attained using different band combination techniques to get maximum accuracy for the classification [48,49]. The acquisition of training samples involved polygons based on spectral signature values corresponding to each land type class, supplemented by background knowledge of the study area of the feature type. Subsequently, a supervised classification approach was employed, utilizing vector machine algorithms supported by ENVI 5.6 software.

2.4. Post-Classification, Accuracy Assessment, and Change Detection

Following the post-classification phase, accuracy assessment and spatial change detection procedures were conducted to validate the classification outcomes. Such verification processes were essential to ensure the reliability and accuracy of the classification results [50]. So, the accuracy assessment was collected by using the ground truthing technique [51] of the images from 1990, 2000, 2010 and 2024, respectively, with the help of Google Earth Pro software 7.3 archived imagery [52] and high-resolution imageries from Gaofen-2. The accuracy samples were systematically collected utilizing ArcMap 10.8 software. Subsequently, the Kappa coefficient and overall accuracy measures were applied as a metric to evaluate the accuracy of the classification, providing a scientifically rigorous assessment of the classification’s reliability and precision.

Kappa coefficient is a statistic that measures the agreement between observed and expected classifications. The Kappa coefficient was calculated [53] using the following Equation (1):

$$\mathrm{k}\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{a}={\displaystyle \frac{\mathrm{P}\mathrm{o}-\mathrm{P}\mathrm{e}}{1-\mathrm{P}\mathrm{e}}}$$

(1)

where P_o is the relative observed agreement among rasters (overall accuracy) [53], and P_e is the hypothetical probability of chance agreement. P_e can be calculated using the following Equation (2):

$$\mathrm{P}\mathrm{e}={\sum }_i({\displaystyle \frac{row sumi\times column sum i}{{(Total number of Classification)}^2}})$$

(2)

The overall accuracy was measured using the following Equation (3):

$$\mathrm{O}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{l}\mathrm{l} \mathrm{a}\mathrm{c}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}={\displaystyle \frac{\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t} \mathrm{c}\mathrm{l}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r} \mathrm{o}\mathrm{f} \mathrm{c}\mathrm{l}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}}$$

(3)

The land transition matrix technique was utilized after the land cover classification (details are presented in

Table 3. Description of the classifications.

LC Type	Description
Built-up area	Buildings, roads, concrete and asphalt
Tree cover	Trees and shrub
Cropland	Agriculture land, arable land
Vegetation	Green belt, sparse vegetation
Water bodies	Water bodies, reservoirs, ponds and lakes
Bare land	No permeant built up, natural rock

) to assess the land cover changes in the study area [54]. The transition matrix is an n × n matrix where each entry T[i,j] represents the proportion of land cover type that transitioned from land cover class ‘i’ at the time ‘t₁’ to class ‘j’ at time ‘t₂’. The transition matrix can be calculated by the following Equation (4) [55]:

$$\mathrm{T}=\left[\begin{array}{c}\mathrm{T}\left[1,1\right] \mathrm{T}\left[1,2\right] \dots T\left[1,n\right]\\ \mathrm{T}\left[2,1\right] \mathrm{T}\left[2,2\right] \dots T\left[2,n\right]\\ \begin{array}{c}\begin{array}{ccc}⋮& ⋮& \begin{array}{cc}\ddots & ⋮\end{array}\end{array}\\ \mathrm{T}\left[\mathrm{n},1\right] \mathrm{T}\left[\mathrm{n},2\right] \dots T\left[n,n\right]\end{array}\end{array}\right]$$

(4)

where each row ‘i’ represents the initial land cover class at ‘t₁’. Each column ‘j’ represents the resulting land cover class at ‘t₂’. Diagonal elements T[i’i] capture the persistence of each land cover type, indicating areas where no change occurred.

2.5. Hot Spot and Cold Spot Analysis (Getis–Ord Gi)]

The Getis–Ord Gi* statistic was employed to identify spatial clusters of urban development around the Taxila world cultural heritage site (CHS). This method calculates Z scores for each feature, revealing significant clustering or dispersion relative to its neighbors [56]. While effective, limitations such as sensitivity to distance thresholds, boundary distortions and challenges with 30 m resolution data required careful mitigation. A sensitivity analysis tested distance thresholds of 100 m, 200 m and 500 m, with 200 m selected for its stability and consistency with patterns observed in high-resolution imagery (KH-9, Gaofen-2 and Google Earth Pro). Buffering reduced boundary effects, and validation through satellite imagery and field surveys ensured accuracy. Despite the limitations of Landsat data, high-resolution datasets and refinements enhanced robustness. The analysis revealed significant clustering, particularly around Sarakhola and Bhir Mound, highlighting progressive urban encroachment. The Gi* statistic was computed following Equation (5) [57].

$$Getis-Ord{G}_i^{*}={\displaystyle \frac{{\sum }_{j=1}^n{W}_{ij x_j-\overline{\mathrm{X}}{\sum }_{j=1}^n{w}_{i.j}}}{\sqrt[S]{\frac{[n\sum _{j=1}^n{w}_{i.j}^2-{\left({\sum }_{j=1}^n{w}_{i.j}\right)}^2]}{n-1}}}}$$

(5)

The Z score and p-value presented in

Table 4. Thresholds of the Z scores and p-values from hot spot and cold spot analyses.

Z Score	p-Value	Type of Spatial Pattern	Color
<−2.58	0.01	High Cluster
−2.58 to −1.96	0.05	-
−1.96 to −1.65	0.10	Low Cluster
−1.65 to 1.65	-	Random
1.65–1.96	0.10	Low Cluster
1.96–2.58	0.05	-
>2.58	0.01	High Cluster

, essential for determining the significance of the computed index, are among the many important metrics that this analysis computes for each characteristic. The Z score standardizes a feature’s value to the mean and standard deviation, giving information about the feature’s relative position within the dataset. This indicates probable hot spots (high values) or cold spots (low values) in the urban region by letting us know if the feature’s value is noticeably greater or lower than the average [58]. A detailed methodology flowchart is attached as Figure 2.

3. Results

3.1. Land Cover Changes

The distribution of land cover change, with observations at 10-year intervals (except for the final interval), was divided into six distinct land types. In the Pothohar region, Taxila’s land classification map prominently highlights major land types: built-up area, bare land and cropland, along with vegetation, are shown in Figure 3, and Table 3 elaborates on the dynamics of land cover changes from 1990 to 2024 for the World Heritage Site of Taxila. On the other hand, the classification accuracy for the region of interest was assessed using the Kappa coefficient and overall accuracy, based on 250 sample points corresponding to built-up, bare land, cropland, tree cover, vegetation, and water bodies. The results indicate that the Kappa coefficient and overall accuracy were approximately 0.85 and 88% for 1990, respectively. In 2000, these values were around 0.84 and 87%. By 2010, the Kappa coefficient had decreased to 0.83, with an overall accuracy of 87%. However, by 2024, the Kappa coefficient increased to 0.87, with an overall accuracy of 90%, as shown in

Table 5. Overall accuracy and Kappa coefficient of the six land cover types.

Year	Overall Accuracy (%)	Kappa Coefficient	Built-Up Area (%)	Tree Cover (%)	Cropland (%)	Vegetation (%)	Bare Land (%)	Water Bodies (%)
1990	88	0.85	85	80	92	83	88	95
2000	87	0.84	90	85	90	87	89	94
2010	87	0.83	88	83	91	85	86	93
2024	90	0.87	92	88	93	89	90	96

. A further confusion matrix of all land type classes is attached as Tables S1–S4 in the Supplementary Data.

By analyzing Figure 3 and

Table 6. Trends in LCC of the World Heritage Site of Taxila between 1990 and 2024 (km²).

	1991	%	2000	%	2010	%	2024	%
Built-up area	23.68	4.2	27.4	4.84	40.6	7.2	78.5	13.8
Tree cover	133.92	23.74	142.45	25.97	147.5	25.1	134.8	23.58
Cropland	116.02	20.43	110.69	19.5	85.91	15.13	74.83	13.18
Vegetation	278.03	48.95	259.83	45.2	267.85	47.16	248.43	43.78
Bare land	8.56	1.5	5.35	0.94	23.78	4.19	26.5	4.66
Water bodies	6.72	1.18	11.12	1.95	7.26	1.27	5.41	0.95
Total	567.5	100	567.5	100	567.5	100	567.5	100

, clear trends in land cover change in the study area can be observed over the past three decades (1990–2024). The built-up area increased sharply from 4.2% in 1990 to 7.2% in 2010 and further to 13.8% in 2024. This trend, evident in the red areas of Figure 3, highlights ongoing urbanization and infrastructure development, which likely corresponds to population growth and increased human activity in the region. A notable decrease in cropland is observed, from 20.43% in 1990 to 13.18% in 2024. Similarly, vegetation declined consistently, from 48.95% in 1990 to 43.78% in 2024. These losses can be attributed to the expansion of built-up areas and bare land, as shown in Figure 3. Tree cover initially increased from 23.74% in 1990 to 25.97% in 2000, possibly due to conservation efforts. However, a subsequent decline to 23.58% by 2024 indicates a reversal, potentially due to increased urban sprawl.

Bare land, though initially low at 1.5% in 1990, increased significantly to 4.19% in 2010 and 4.66% by 2024. This increase, represented in grey areas in Figure 3, highlights the abandonment of previously utilized land.

Figure 4a shows a comparative bar chart illustrating how different land cover types have changed over time. By comparing the heights of these bars across years, one can observe trends such as increases in built-up area or decreases in cropland. Figure 4b shows a three-dimensional perspective of the same land cover data shown in Figure 3; by viewing the data in a 3D format, it becomes easier to visualize relative differences in the magnitude of land cover type in the study area. Furthermore, in the Figure 4c line graph, the focus shifts from absolute areas to proportions or percentages of each land cover category over time, making trends clearer.

There is a steady decline in bare land and a corresponding increase in built-up area. In Figure 4d, the trend line illustrates population growth in Taxila at various time intervals based on census data from the Pakistan Bureau of Statistics. In 1951, the population was approximately 50,000, rising to 322,840 by 1998 and reaching around 739,244 by the 2023 census.

3.2. LCC Transition over the Period of 1990–2024

We have employed a land transfer matrix to further analyze the transition of LCC from 1990 to 2024, which provided the results for alterations in each land cover type to the other over the time span of the last three decades. As can be seen from

Table 7. Land transfer matrix from 1990 to 2024 (area in sq.km).

Classes		2024
		Bare Land	Built-Up Area	Cropland	Tree Cover	Vegetation	Water	Total
1990	Bare land	5.9	0.3	0.0	0.0	1.6	0.0	8.5
	Built-up area	0.1	17.6	0.2	1.9	3.5	0.0	23.2
	Cropland	4.4	13.1	41.7	6.6	49.8	0.2	115.8
	Tree cover	1.0	2.3	8.7	84.2	37.3	0.4	133.8
	Vegetation	14.5	44.8	23.1	39.5	157.3	0.3	279.5
	Water	0.8	0.0	0.0	0.5	0.9	4.5	6.7
	Total	26.6	78.1	73.8	133.3	250.3	5.4	567.5

and Figure 5, a significant portion of the cropland and vegetation was converted into a built-up area of 13.1 km² and 44.8 km², respectively, from 1990 to 2024. This is similar to the 4.4 km² of cropland and 14.5 km² of vegetation land transfer into bare land and the minor change of bare land (1.6 km²) into vegetation land from 1990 to 2024.

3.3. Impact of Urban Development and Encroachment on CHS Taxila

To check the impact of urban development and encroachment on Taxila’s world CHS, this study has applied the Getis–Ord Gi technique. The results in Figure 6 indicate that the spatial pattern of the urban cluster has a significant value from 1990 to 2024. All heritage sites observed a cold-spot pattern with low Z values in 1990, indicating a high to low cluster formation. By 2000, the Z value for the Sarakhol site increased to 1.96, demonstrating a low cluster pattern as a hot spot, while the remaining sites continued to show cold spot characteristics (high cluster). In 2010 and 2024, the Sarakhol site observed maximum hot-spot values, indicating high clusters. Bhir Mound showed a cold spot (low cluster) in 2010 and shifted to a hot spot (low cluster) in 2024, while other sites showed no significant changes.

Figure 7 illustrates the impact of urban development and encroachment on Taxila’s world cultural heritage sites. The graph reveals that Sarakhol and Bhir Mound were the most affected sites by urban development. Sarakhol exhibited a significant increase in area, expanding from 0.2423 km² in 1990 to 0.3397 km² in 2000 and further growing to 0.554 km² and 0.627 km² in 2010 and 2024, respectively. Similarly, Bhir Mound showed considerable growth, with its area increasing from 0.0665 km² in 1990 to 0.074 km² in 2000, 0.114 km² in 2010, and 0.2733 km² in 2024. In contrast, other sites depicted in Figure 7b demonstrated less significant changes than Sarakhol and Bhir Mound. High spatial resolution images in Figure 7c,d indicate increasing urban encroachment around these two sites. In 2004, the Sarakhola core zone was 0.0168 km², shrinking to 0.0077 km² and 0.0032 in 2015 and 2024, respectively. Concurrently, a similar trend is observed at the Bhir Mound site, where the distance between the core zone of the CHS and the adjacent urban residential areas has been progressively decreasing.

4. Discussion

Urban development has pressured the region surrounding Taxila, a world-renowned CHS. Effective management and monitoring are essential to preserve this archaeological treasure. The widespread availability and affordability of remote sensing technology facilitate reliable data acquisition, which is crucial for developing innovative monitoring systems. These systems support preservation efforts and mitigation strategies aimed at safeguarding the region’s cultural heritage and landscape [59,60]. This study quantitatively analyzed urban development in the Taxila world CHS between 1990 and 2024. A marked increase in built-up area, the predominant land use class, was observed, expanding from 23.68 km² in 1990 to 78.5 km² by 2024 (Table 4). This rapid growth, particularly evident post-2010, is attributed to the development of new urban housing communities in response to the growing population in Pakistan [61,62]. Indeed, the increase in population required new commercial and residential areas, as well as public utilities, which was the main problem for conservation efforts toward CHSs [54]. Trend line analysis and high-resolution Gaofen-2 imagery (Figure 4 and Figure 7) verify the urban development phenomena. Furthermore, our results indicate the increase in built-up area was mainly due to the loss of crop and vegetation land. Concurrently, a chord diagram (Figure 5) illustrates a significant transformation of cropland (13.1 km²) and vegetation (44.8 km²) into built-up area (4.4 km²) and bare land (14.5 km²) land cover classes. Notably, this pattern aligns with trends in Pakistan’s urban planning policies. For instance, the Ravi Urban Development Authority (RUDA) in Lahore has acquired approximately 400 km² of farmland for urban projects, demonstrating how farmland, rather than forested areas, is often targeted for development [63,64]. These findings underscore the escalating pressure on the site’s conservation efforts and highlight the urgent need for effective land use planning and management strategies to mitigate the negative impacts of urban development on the World Heritage Site.

Getis–Ord Gi analysis revealed predominantly positive Z statistic values, indicating a random cluster that is likely due to the low resolution of the dataset. However, proximity to the cultural heritage site revealed spatial variations in the cluster patterns (Figure 6). Getis–Ord analysis revealed high Z-score values for Sarakhol and Bhir Mound between 1990 and 2024, which indicated a shift from low to high clustering. Conversely, the remaining four sites exhibited low Z-score values, suggesting cold-spot conditions [58]. These findings were thoroughly validated by the results from 500 m built-up area and high spatial resolution KH-9, Google Earth Pro and Gaofen-2 imagery from 1971, 2004, 2015 and 2024, respectively. Sarakhola’s built-up area expanded from 0.2423 km² in 1990 to 0.627 km² in 2024, reflecting a 159% increase. In contrast, Bhir Mound exhibited an even more substantial rise from 0.0665 to 0.2733 km², corresponding to a 311% expansion. Figure 7c shows reductions in the Sarakhol core zone from 0.0168 km² in 2004 to 0.0077 km² and 0.0032 km² by 2015 and 2024, respectively. These results underscore the critical need for Pakistan to enhance monitoring and strictly implement the Antiquities Act 1975 to sustain its OUV status from UNESCO [14,27].

Analyzing the land cover change and spatial urban distribution patterns reveals a pronounced clustering around the archaeological landscape over time. While the environmental consequences of urban expansion can vary, it is evident that unplanned urban growth consistently exacerbates negative impacts. Urban expansion poses a significant threat to the integrity of Taxila and mirrors challenges faced by numerous developing and developed countries [35]. Furthermore, urban development and encroachments around archaeological sites, such as Saraikhol, Bhir Mound, and other sites, highlight a lack of buffer zone protection [10]. This inadequate protection is one among other ongoing issues that significantly irrevocably damage these invaluable CHSs.

In the study area of Taxila, the expansion of urban areas from 1990 to 2024 has resulted in a significant loss of cropland and vegetation, as illustrated in Figure 5. This transformation is primarily characterized by converting agricultural and vegetated areas into built-up area. The challenges in Taxila reflect global issues faced by cultural heritage sites. The Giza Pyramids in Egypt face urban encroachments from Cairo’s rapid expansion [65]. Similarly, Hue monuments in Vietnam struggle with unplanned urbanization causing damage to archaeological zones [66], and in Jordan, uncontrolled urban growth threatens the site’s integrity [67]. These examples highlight the need for effective strategies to safeguard Taxila’s world cultural heritage. Additionally, the lack of an adequate buffer zone around the core heritage zone has been a critical factor in this urban encroachment. Consequently, residential areas, commercial developments, quarrying activities, road networks, and public utilities have increasingly encroached upon the core zone [25]. This uncontrolled urban sprawl threatens the integrity of the world cultural heritage site and disrupts the historical landscape. The absence of stringent land use regulations and effective urban planning strategies has exacerbated this situation. Additionally, the region’s rapid population growth and economic development in the region have intensified the pressure on these culturally significant areas, potentially compromising their OUV, as recognized by UNESCO.

5. Conclusions

While this study provides valuable insights into the land cover changes and urban encroachment affecting Taxila’s world cultural heritage site (CHS), it is not without limitations. The reliance on Landsat imagery (30 m resolution) may overlook small-scale or recent encroachments, which introduce data uncertainty in detecting finer landscape transformations. Moreover, methodological constraints, such as the choice of classification algorithms and the resolution of data used in hot spot and cold spot analyses, could influence the observed clustering patterns. Despite these constraints, our findings reveal a significant expansion in built-up area from 23.68 km² to 78.5 km² between 1990 and 2024, primarily at the expense of cropland and vegetation (reduced by 13.1 km² and 44.8 km², respectively). Sarakhol and Bhir Mound experienced particularly intense urbanization, with Sarakhol’s core zone diminishing by 81% between 2004 and 2024.

Moving forward, integrating socioeconomic data such as demographic trends, land ownership, and economic indicators would yield a more holistic understanding of the drivers behind rapid urbanization near heritage sites. Additionally, employing higher-resolution or multisensor imagery such as Sentinel-2, commercial satellites and LiDAR may improve the detection of subtle encroachment patterns. The conservation of CHSs is critical for preserving their historical and cultural significance, as these sites offer tangible links to our past. To safeguard Taxila’s integrity, buffer zones should be delineated by integrating data from field surveys, high-resolution satellite imagery, spatial data, topography and the historical and cultural significance of the area. By incorporating these data delineation, the buffer would be determined. For instance, a 100 m buffer around the most significant historical and cultural feature could ensure the utmost protection, with 200 m surrounding areas that are still of historical importance but may allow limited developments under strict conservation regulation. Additionally, these buffer zones can be integrated into local urban planning policies through zoning laws, creating a legal framework to limit encroachment and safeguard the integrity of the World Heritage Site. Finally, revising outdated regulations such as the Antiquities Act 1975, enhancing monitoring systems and implementing stricter penalties for violations would strengthen heritage protection and maintain the outstanding universal value of these archaeological treasures for future generations.