Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Reports volume 15, Article number: 3271 (2025)
6
Metrics details
This study addresses the significant issue of rapid land use and land cover (LULC) changes in Lahore District, which is critical for supporting ecological management and sustainable land-use planning. Understanding these changes is crucial for mitigating adverse environmental impacts and promoting sustainable development. The main goal is to evaluate historical LULC changes from 1994 to 2024 and forecast future trends for 2034 and 2044 utilizing the CA-Markov hybrid model combined with GIS methodologies. Landsat images from various sensors (TM, OLI) were employed for supervised classification, attaining high accuracy (> 90%). Historical LULC changes from 1994 to 2024 were analyzed, revealing significant transformations in Lahore. The build-up area expanded by 359.8 km², indicating rapid urbanization, while vegetation cover decreased by 198.7 km² and barren lands by 158.5 km². Water bodies remained relatively stable during this period. Future LULC trends were projected for 2034 and 2044 using the CA-Markov hybrid model (CA-MHM), which achieved a high prediction accuracy with a kappa coefficient of 0.92. The research indicated significant urban growth at the expense of vegetation and barren land. Future forecasts suggest ongoing urbanization, underscoring the necessity for sustainable land management techniques. This research is a significant framework for urban planners, providing insights that combine development with ecological conservation. The results highlight the necessity of incorporating predictive models into urban policy to promote sustainable development and environmental preservation in quickly changing areas such as Lahore.
Land change refers to alterations in land’s natural or human-induced characteristics over time. These changes can include modifications in land use and land cover (LULC)1. Variations are critical indicators of anthropological activities and natural processes altering the Earth’s surface2. These changes encompass modifications in the physical and biological characteristics of the land, including changes in vegetation cover, urban expansion, and agricultural intensification3. These variations have significant consequences for ecosystems, biodiversity, and human well-being. For example, deforestation can lead to habitat loss and loss of agricultural land, leading to food security challenges4. Based on data from the United Nations (UN), in 2020, more than 54.4% of the population globally resided in urban areas, which is expected to rise to 60% by 2030, placing more pressure on urbanized resources5. The rate of this rise is more significant in developing nations, where populations are expanding and altering the urban environment rapidly6. The rapid growth without proper planning and infrastructure accelerates LULC changes, which are linked to the deterioration of ecological services and the welfare of people7. In developing countries, cities are usually defined by inadequate infrastructure planning, significant rates of citizenship, and a growing number of squatter communities8. These requirements include addressing distinct challenges and possibilities for adapting and mitigating urban environments and integrating them into urban growth strategies.
Pakistan is a country with diverse landscapes, ranging from mountains to plains. Pakistan has witnessed significant LULC transformations in recent decades, especially in metropolitan areas such as Lahore, Karachi, Islamabad, etc., primarily due to rapid urbanization, population growth, and decreasing agricultural expansions9. These changes have led to the transformations of natural ecosystems and profoundly impacted the environment and society. In Pakistan, land degradation is a significant concern, caused by unsustainable land use practices such as overgrazing and deforestation which lead to soil erosion. These impacts are particularly severe in areas where agriculture is the primary source of livelihood. The consequences of land degradation are far-reaching10. It has the potential to cause biodiversity loss, decreased agricultural output, and heightened susceptibility to natural disasters like droughts and floods. These impacts can have profound implications for the livelihoods of people dependent on the land for their food and income11. Researchers worldwide are interested in studying LULC patterns and changes due to the significant impact of land resources on environmental sustainability9. The transformation of LULC is particularly challenging in unplanned, rapidly changing areas such as urban settlements in developing countries. Understanding past LULC change patterns over the last few decades and predicting future changes is crucial for comprehending the effects of LULC alteration on the Earth’s surface12.
Remotely sensed information is often used to examine trends and alterations in LULC on a significant scale13. Several studies1,14,15,16have investigated LULC changes in various regions of Pakistan. Researchers used Landsat data and applied the most likely method to identify a 4.5% rise in residential areas and a decline in vegetation coverage in Okara from 2000 to 20209. A different study conducted in Khyber Pakhtunkhwa (KPK) revealed a reduction of 7.17% in vegetation cover, accompanied by an expansion of bare terrain and urban growth from 1990 to 201917. Additionally, a study in Islamabad, Pakistan, revealed a significant area of Natural vegetation and agricultural land had been converted to barren land between 1993 and 2018 based on Landsat images18.
Remote sensing (RS) and Geographical Information System (GIS) techniques are extensively used in weather prediction, climate change, and ecological research19. RS provides detailed Geo-spatial data, while GIS provides tools for managing environmental and ecosystem data more effectively20. Methods such as cross-correlated evaluation, imagery variance, post-classification comparison (PCC), Object pixel-based classification, and combining images have been employed to study LULC changes using RS data21. LULC change evaluation, a commonly used technique to measure changes in LULC, mainly depends on the use of multi-spectral remote sensing data. The use of data and multi-temporal remote sensing satellite images has presented multiple academic possibilities, especially in the study of LULC trends22. Satellite images, including those acquired from Landsat sensors, have played a crucial role in the analysis of changes to LULC23. These images also offer vital data on crops, food production, and environmental indicators, facilitating the tracking of ecological shifts over a time24.
Google Earth Engine (GEE) is a prominent example of a cloud-based platform for storing and processing large spatial datasets. It allows researchers to access high-performance computational resources and easily share algorithms with others. GEE facilitates LULC studies by efficiently handling large-scale spatial datasets. With access to extensive satellite imagery and geospatial data, researchers can analyze LULC changes over time28. Moreover, GEE supports modern Machine Learning techniques, improving the accuracy of LULC classification and change detection, thus significantly contributing to advancing research in this field29. GEE is regularly modified to accommodate recent techniques and Machine Learning (ML) techniques, as ML is crucial in spatial data analysis, especially in identifying intricate relationships and dependencies30. This capability is particularly beneficial for understanding spatial phenomena that exhibit complex behaviours.
The CA-Markov hybrid Model (CA-MHM), which combines CA with the Markov Chain, is frequently used for accurately evaluating LULC variations31. This model is particularly robust, making it suitable for modelling LULC changes in a complicated environment32. Many researchers15,33,34have successfully utilized the CA-MHM for LULC prediction. In this research, we employ the CA-MHM as it is a powerful and popular choice among researchers for detecting, predicting, monitoring, and simulating spatiotemporal changes in LULC35. The critical element of the CA-MHM is the transition rules, which are determined based on training data. Additionally, the performance is influenced by factors such as neighbourhoods class and cell size, which are essential for achieving optimal simulation or prediction results36. The CA-MHM efficiently combines remote sensing data with GIS, enabling the conversion of findings into geographically precise outputs that are essential for LULC studies37. The careful allocation and utilization of resources to balance ecological, social, and economic requirements is a prerequisite for sustainable land management37,38,39. Research has shown that predictive models, including Markov chain models and cellular automata, are indispensable instruments for comprehending and mitigating the consequences of agricultural intensification, deforestation, and urban expansion on ecosystems35. These models enable planners and policymakers to implement proactive measures that are consistent with sustainability objectives, including the preservation of biodiversity, the preservation of water quality, and the reduction of greenhouse gas emissions by forecasting LULC changes1. Valuable insights for the conservation of natural resources and the resilience of landscapes under changing environmental and socio-economic conditions are provided by the integration of LULC prediction models into land management strategies.
This study addresses a gap in LULC research in Pakistan, where few studies have used the hybrid CA-Markov model with GIS tools for long-term forecasting. By providing predictions for 2034 and 2044, it enables an evaluation of sustainable urban management strategies. Focused on Lahore, one of Pakistan’s fastest-growing cities, the study examines 50 years of LULC changes (1994–2044), offering insights into urban growth and environmental degradation while providing actionable recommendations for sustainable urban planning.
Key research questions include identifying the historical patterns and trends of LULC changes, assessing the impact of urbanization on the environmental and socio-economic landscape, and evaluating the accuracy of the CA-MHM projections. The study hypothesizes that urban areas in Lahore have significantly increased, resulting in the reduction of vegetation and barren land. Furthermore, it posits that the CA-MHM can accurately predict future LULC scenarios. This research fills a significant gap by providing a comprehensive, long-term analysis of LULC changes and validating the predictive capabilities of the CA-MHM in this context.
The objectives of this study are: (i) To comprehensively assess and evaluate the dynamic trends of LULC variations in Lahore from 1994 to 2024. (ii) Predict and project future LULC trends for 2034 and 2044 using the CA-MHM, providing insights into potential urban expansion scenarios and vegetation decline. (iii) Enable the implementation of ecological land management practices by identifying key drivers and trajectories of LULC changes, thereby supporting informed decision-making and policy formulation.
The study focuses on Lahore, a metropolitan city in Punjab province in Pakistan. Punjab is Pakistan’s second-largest province in terms of area (205,344 km2) and population40. Change in crop production heavily impact the province due to climate change, as many rural residents depend on agriculture for their livelihoods41. Lahore is located between latitudes 31°15′–31°43′N and longitudes 74°10′–74°39′E in Pakistan’s central east, bordering the Indian states of Punjab and Himachal Pradesh42. The district covers a total area of 1772 km2and is bordered to the south by the Kasur district and to the north and west by the Sheikhupura district. With a population of approximately 14 million, Lahore is Pakistan’s second most populous district43. Due to its semi-arid environment, the district experiences warm, dry winters and hot, muggy summers, typically ranging from 36 to 46 ֠C44. For this research, the city of Lahore has been chosen as the region of study because of its rapid urbanization, large population expansion, and its role as a key industrial and cultural center in Pakistan. Because of these variables, Lahore is an important location for the study of the dynamics of LULC. In addition, the environmental vulnerability of Lahore, which is characterized by a decrease in vegetation and an increase in barren land, highlights the necessity of sustainable urban planning to reduce the negative effects on the natural world. Figure 1 provides a visual overview of our study area and data coverage, (a) depicts Pakistan within the global context, emphasizing population density distribution, (b) zooms into Punjab province, highlighting our specific study area with overlaid Landsat satellite path and row tiles, (c) offers a detailed close-up of Lahore, focusing on street-level mapping within our study boundaries. Together, these visuals illustrate the geographical context, satellite data coverage, and spatial scale of our research on LULC changes in Lahore.
Visual representation of study area Map (a) World map depicting Pakistan with population density distribution. (b) Overview of Punjab province with highlighted study area and Landsat satellite path and row tiles. (c) Close-up view of Lahore study area with street map detail.
This study involved a series of steps to analyse changes in LULC thoroughly. We acquired and processed images from Landsat-5, Landsat-8, and Landsat-9 data using the GEE data catalogues. The analysis was segmented into four distinct periods: 1994, 2004, 2014, and 2024, to consider the steady process of LULC change. Subsequently, we obtained images from Landsat-5 for the years 1994 and 2004 and from Landsat-8 and 9 for the years 2014 and 2024. The LULC analysis in this study utilized Landsat 5, 8, and 9, all with a spatial resolution of 30 m for reflecting bands. Although collected from various satellite missions, these datasets maintain identical spatial resolution, hence providing uniformity in LULC classification across multiple years. Detailed specifications of these datasets are provided in Table 1.
We processed Landsat images from 1994 to 2024 directly within the GEE platform to ensure high-quality data for our analysis. Using GEE’s pre-processed Landsat collections, which include standardized geometric corrections, we accessed and organized the spectral bands of each image into multi-temporal datasets suitable for analysis. This workflow provided consistent alignment of images across time, as GEE’s built-in corrections maintain spatial accuracy. Additionally, we projected the dataset to the Universal Transverse Mercator (UTM) coordinate system for uniformity in spatial representation across the study area, facilitating more precise multi-temporal analysis. This step ensured that all images were accurately aligned and represented in a standard geographic coordinate system. We filtered out images with less than 5% cloud coverage for all four mosaics to minimize cloud interference. Finally, we calculated the average values of every mosaic to combine the pictures, applying the average pixel technique. This methodology offered a more distinct representation of the annual characteristics in all regions, minimizing the influence of cloud influence and insufficient data on the evaluation of changes.
RSIs combine various spectral ands from remote sensing data into a single image, enhancing specific features of interest. Recognized for their ability to improve feature visibility and reduce noise45, this research selected four well-established RSIs from existing literature and utilized several indices detailed in Table 2. The selection of these indices was based on the diverse land use patterns in the research area to improve the accuracy of the classification method.
The study area focused on key LULC classes: water bodies, build-up areas, vegetation (including old and newly restored agricultural land), and barren land, as detailed in Table 3. A supervised classification method using a Random Forest and a Decision Tree Classifier (DTC) was employed, integrating various indices for enhanced accuracy. The DTC, a hierarchical model, recursively divides independent variables into homogeneous regions through decision rules, categorizing each pixel by binary decisions. This method, commonly used in satellite imagery, applies classification rules in three stages: gathering information, identifying factors using cognitive approaches, and generating criteria from observed data (Fig. 2).
Decision tree for LULC classes.
Assessing the precision of spatial data obtained from remote sensing images is essential to ensure accurate classified images50. In this study, the accuracy of images from 1994, 2004, 2014, and 2024 was evaluated through the error matrix. The accuracy assessment for the 1994, 2004, and 2014 LULC classification utilized a combination of ground truth data derived from historical maps, high-resolution satellite imagery, and records from field surveys. Due to the scarcity of direct field data from 1994, we employed reference points obtained from high-resolution images and corroborated them with previous data. Additionally, a semi-detailed survey conducted in 2024 provided precise information about soil trends, terrain, and environmental features. This data was used as a reliable reference for the entire research region. We employed a 70:30 ratio for training and validation points. A total of 500 ground control points were collected during a semi-detailed survey conducted in 2024, which covered various land classes (build-up areas, vegetation, barren land, and water bodies) to enhance the accuracy of the LULC classification. Consequently, the accuracy of each labeled image was assessed by calculating the values of the producer’s, user’s, overall, and Kappa coefficients, and Eqs. 1–4 outline the process for calculating them. Furthermore, the integration of GIS with remote sensing data and the Markov model was leveraged, highlighting the synergistic benefits of combining these technologies51. The total number of correctly identified pixels is divided by the number of reference pixels to obtain overall accuracy.
Ap represents the producer’s accuracy, Au stands for the user’s accuracy, and Ao denotes the overall accuracy. The kappa coefficient is symbolized by k.
This phase encompasses essential activities such as change detection, accuracy evaluation, and LULC majority filtering. To enhance classification accuracy, a majority filter is employed to remove isolated pixels, as detailed by52. This filter uses a tiny pixel’s neighborhood to identify the value that occurs most frequently. Implementing this filter before conducting accuracy assessments helps to remove noise and improve accuracy.
The final classified images were evaluated using metrics such as accuracy, recall, precision, and F1-Score (F1s), as described by14. These metrics are derived from True Positive (TP), True Negative (TN), False Positive (FP), and False Negative (FN) rates, offering a comprehensive assessment of the classifier’s accuracy. Precision, defined as (Eq. 5) the percentage of correctly identified examples (TP) to all instances identified by the algorithm, is an important metric for characterizing the final accuracy measurement.
Recall is the ratio of accurately recognized positive cases to the overall count of actual positive cases (TP + FN), as illustrated in Eq. (6). This metric emphasizes the model’s effectiveness in detecting positive cases accurately.
The F1-Score (F1s), which combines both precision and recall, quantifies the model’s overall performance, as illustrated in Eq. (7).
In addition to these accuracy assessment techniques, a confusion matrix is developed to compare actual class labels with projected ones. In this matrix, pixels in the proper rows and columns are incorrectly classified, whereas pixels along the diagonal are correctly classified.
The final step in LULC analysis is change detection. This process involves the comparison of images from different periods, usually through image subtraction, to highlight differences between the original and resulting images. In this study, change analysis was conducted to observe overall land changes, with a primary focus on assessing the impact of LULC.
In this study, the CA-MHM was utilized to predict future LULC trends. By leveraging transition probabilities, this model forecasts upcoming LULC scenarios based on current conditions. The CA-MHM approach is a straightforward statistical method that uses a transition probability matrix, integrating neighbourhood influences through a spatially aware algorithm54. However, while the transition probability matrix for each class is precise, the spatial dispersal of these occurrences remains uncertain. This limitation arises from the inability to project the spatial allocation of LULC categories accurately55. To address this issue, the CA-MHM incorporates a CA filter, enhancing the model’s precision in detecting changes across various land-use classes by adding a spatial dimension56. We conducted the CA-Markov-Chaining modelling in the IDRISI Selva environment, which served as the primary IDE for implementing these methods on raster data. IDRISI Selva facilitated the efficient processing of spatial data and allowed us to apply CA transitional rules for land cover change simulation. This study did not include multi-criteria decision-making approaches, such as AHP, because it primarily concentrated on spatial-temporal LULC changes, which the model successfully addresses. Future studies could integrate multi-criteria decision-making technologies to enhance land use planning and decision-making processes.
The CA model posits that the state of a grid cell is influenced by both its dynamics and those of its neighbouring cells34. This model offers an adaptive assessment by establishing a dataset of land uses at a specific time and subsequently predicting the chances of these modifications for a future timeframe51. In contrast, change detection in LULC involves predicting the transformation of image pixels from one land-use category at an initial time (t1) to another later (t2). The applied models are expressed in Eq. 857.
Pij represents the transitional probability matrix, reflecting the probability of a transition from one specific land use type (n) to another, with values ranging from 0 to 1. St denotes the current land-use status at time t, while St + 1 represents the subsequent land-use state at time t + 1.
When combined with CA, The Markov model accurately forecasts and predicts variations in LULC over time and space. This comprehensive methodology excels in accurately representing and forecasting intricate LULC types32. The CA-MHM employed in this study utilizes two variables: the discrete variable in time and space and the local variable, typically assigned to interactions. Core components of the CA-Markov model include grid sizes, cell neighbourhoods, cell spaces, time phases, and transition rules58. Cells, adjacent or close in any dimensional space, can only be in one state at a time, defining the system’s attributes. Transition rules dictate that a cell’s state changes based on its neighbours’ states. The model assigns weight factors to determine neighbours, with closer cells having a higher weight factor, thus aiding in predicting the states of neighbouring cells. The LULC changes were calculated using an appropriate Eq. 959.
(:Si) represents land-use change. (:L{U}_{(i,t1)}) refers to land-use change during an earlier period, while (:L{U}_{(i,t2)}) indicates land-use change during a later period. (:LU{A}_{i}) is the area where no change has occurred.
Future LULC patterns for the studied future dates were predicted to depend on historical patterns. This study used LULC images from two time periods: initial land use, t1, and later land use, t2, following the methodology illustrated by Hyandye and Martz32. Each image undergoes a typical 5 × 5 pixel proximity filter to determine the nearby cells for every land-use category. To exert a substantial impact, every individual cell was encircled by a matrix consisting of a 5 × 5 arrangement of cells60. This study utilized adjacent pixels to produce geographically adjacent weights for projecting LULC categories for the years 2034 and 2044. Equation 10 demonstrates the application of the proximity filter in evaluating LULC change.
The methodology for examining, modeling, and forecasting LULC changes using the CA-MHM involves four main steps:
The CA-MHM was used to calculate transition probability matrices for the years 1994, 2004, 2014, and 2024.
These transitional matrixes were subsequently utilized to produce a series of conditional probabilistic datasets for various land types spanning from 1994 to 2024.
The transitional probability matrixes for the periods 1994–2004, 2004–2014, and 2014–2024, in conjunction with the conditional probabilistic data and LULC classification maps for 2014 and 2024, were combined using the CA-Markov geospatial operators.
This integration was used to simulate LULC maps for 2034 and 2044.
According to Eastman61, the accuracy of any predicted change algorithm is highly based on the validation procedure. Before projecting future LULC patterns for the years 2034 and 2044, it was necessary to validate the outputs of the model. A validation module achieved this by comparing the predicted maps with the classified maps and assessing the degree of agreement between them. The validation process involved comparing the predicted LULC results for 2024 with their corresponding observed dataset. Kappa coefficients were then computed to assess the accuracy of the model. The Kappa coefficient (Eq. 11) is a statistical metric that differentiates between errors in quantity and errors in location in two qualitative maps, providing a comprehensive evaluation of the model’s performance62.
P0 represents the proportion of cells that are correctly classified. PCdenotes the hypothetical probability of chance agreement between the actual LULC map for 2024 and the projected LULC maps for the same year. The Kappa values were divided according to56 as shown in Table 4. Additionally, Fig. 3 illustrates the methodology used for LULC evaluation and prediction using the CA-Markov hybrid model.
Methodological framework.
The findings show that the accuracy assessment and kappa values for various land-use classes in the Lahore area were acceptable for the years 1994, 2004, 2014, and 2024 (Figs. 4 and 5). This study achieved classification accuracy that met the required criterion of at least 80% accuracy for the sensor data. Figures 4 and 5 demonstrate the results of our accuracy assessment. The high F-1 scores, which exceed 0.90 for most categories, indicate the robustness of our classification, with a slight exception for barren land, which scored 0.88. The consistency in performance is observed across various classes, with Build-up areas, water bodies, and vegetation showing the highest accuracy. Conversely, barren land exhibited lower accuracy, with an average recall score of 0.89. Importantly, all land use classes surpassed the 0.90 threshold in precision, further validating the robustness of our results. The confusion matrices illustrate the high agreement among actual data and predicted classes, with overall accuracy scores of 90.1%, 91.1%, 92.3%, and 92.6% for 1994, 2004, 2014, and 2024, respectively. The higher accuracy of the data has been greatly influenced by the better quality of the Landsat 8/9 images. The comprehensive evaluation of correctness not only highlights the strength and reliability of our LULC categorization but also facilitates informed decision-making and dependable studies using map information.
Evaluation measures such as F1-score, precision, and Recall are used to measure the correctness of each LULC category for the corresponding years.
Confusion matrices for the years 1994, 2004, 2014, and 2024 described the detailed assessment of classification accuracy for each land use land cover class.
The findings of the maximum likelihood method applied to satellite images via supervised classification identified four LULC classes in the study area: Build-up, Vegetation, Barren, and water bodies. Figure 7 presents the quantitative data for the four LULC categories across different periods. Additionally, Fig. 6 displays the arrangement of LULC classes using classified maps created from satellite imagery of the Lahore region for the years 1994, 2004, 2014, and 2024. The various LULC classes on these maps are shown in easily identifiable colours.
LULC classification maps (a) 1994 (b) 2004 (c) 2014 and (d) 2024.
LULC classes area from 1994–2024.
The analysis of LULC variations over the period from 1994 to 2024 in the Lahore area reveals significant shifts in various categories, as shown in Table 5. The area of water bodies showed fluctuations, starting at 21.77 km² (1.16%) in 1994, slightly increasing to 23.19 km² (1.23%) in 2004, reaching a peak of 30.24 km² (1.61%) in 2014, before decreasing to 19.20 km² (1.02%) in 2024. There was a consistent decline in vegetation cover from 1011.07 km² (53.9%) in 1994 and reaching 812.29 km² (43.3%) in 2024. Barren land also decreased over the study period, starting at 235.07 km² (12.5%) in 1994 and further to 76.58 km² (4.08%) in 2024. In contrast, Build-up areas saw a substantial increase from 608.66 km² (32.43%) in 1994 to 778.11 km² (41.4%) in 2004, rising to 815.13 km² (43.4%) in 2014, and reaching 968.50 km² (51.6%) in 2024. These trends highlight the ongoing urbanization and reduction in both vegetation and barren land in the Lahore area over the past three decades. The shifts in LULC categories indicate significant environmental and developmental changes, with urban expansion being the most pronounced.
To acquire an understanding of the fluctuating alterations to LULC types in the Lahore area, the yearly variations in such types were calculated for the three different years under research (1994–2004, 2004–2014, and 2014–2024), and an overall change from 1994 to 2024, as shown in Fig. 8. According to the data in Table 6, Between 1994 and 2004, the area of water bodies increased by 1.42 km², followed by a more substantial rise of 7.05 km² from 2004 to 2014. However, there was a significant decline of 11.05 km² in the period from 2014 to 2024, resulting in an overall decrease of 2.57 km² over the entire 30-year period. Vegetation cover decreased annually by 74.7 km², 36.6 km², and 87.4 km² for the three periods, respectively, with the greatest reduction occurring between 2014 and 2024. Overall, there was a significant loss of 198.7 km² of vegetation cover from 1994 to 2024. The barren land category also saw a notable reduction over the years. There was a substantial decrease of 96.2 km² between 1994 and 2004, followed by a smaller decline of 7.43 km² from 2004 to 2014. The period from 2014 to 2024 saw a further reduction of 54.88 km². In total, barren land decreased by 158.5 km² over the three decades. Build-up areas expanded significantly, with annual increases of 169.4 km² from 1994 to 2004, 37 km² from 2004 to 2014, and 153.3 km² from 2014 to 2024. Over the entire study period, the Build-up area expanded by a remarkable 359.8 km².
Furthermore, Fig. 8shows the temporal variation of LULC Classes change in the Lahore region in km2 throughout the study period. The results show that the build-up area increased over time but also showed a noticeable decrease in barren land. The rise in build-up also shows rising population growth in Lahore and an increase in buildings, development, and growing domestic things, all of which are the main drivers of LULC changes. These trends highlight the region’s rapid urbanization and significant environmental changes over the past three decades.
LULC changes for annual years in lahore.
Figure 9 presents the changes in LULC classes in the Lahore region across three time periods: 1994–2004, 2004–2014, and 2014–2024. The data highlight significant trends in the transition of LULC types over these periods. Furthermore, Fig. 10illustrates the spatial variations of LULC classes over the study region. The analysis of LULC changes from 1994 to 2004 shows that the most significant change occurred in the vegetation-to-vegetation category, covering approximately 733.39 km2. This was followed by a build-up to build-up with 414.13 km2, indicating considerable urban stability. Urban expansion is also evident, with 247.09 km2changing from vegetation to build-up. Other notable changes included 105.70 km2from barren to build-up and 158.40 km2from build-up to vegetation. From 2004 to 2014, vegetation-to-vegetation changes remained the largest, with an area of 670.92 km2. Build-up to build-up increased to 550.53 km2, showing sustained urban development. The transitions from vegetation to build-up and build-up to vegetation also saw significant areas of change, 211.14 km2and 174.54 km2, respectively. Smaller changes were observed in categories like barren to build-up, covering 78.20 km2, and build-up to water, with 19.47 km2.
Between 2014 and 2024, the vegetation-to-vegetation category remained the largest change, covering 666.18 km2. The buildup-to-buildup category showed a significant increase to 729.85 km2. Changes from vegetation to build-up amounted to 197.99 km2, while build-up to vegetation recorded 98.65 km2. Other notable changes included 95.39 km2from barren to build-up and 20.34 km2from Water to Buildup. Overall, from 1994 to 2024, the largest LULC change occurred in the vegetation-to-vegetation category, accumulating 573.20 km2. The build-up to build-up followed by 449.00 km2, showing consistent urban development. The transition from Vegetation to Buildup covered 412.99 km2, indicating a significant conversion of vegetative areas to urban areas. Other important changes included 51.70 km2from barren to vegetation and 142.44 km2 from build-up to vegetation.
The data indicate a consistent pattern of urban expansion in Lahore, primarily at the expense of vegetation and barren land. The vegetation-to-vegetation changes remained the highest in all periods, reflecting areas where vegetative cover is maintained. These transitions highlight the dynamic nature of land use changes over time, with notable urban growth and the conversion of native landscapes into metropolitan cities.
Changes in LULC classes between 1994–2024.
LULC changes (a) 1994–2004 (b) 2004–2014 (c) 2014–2024 and (d) 1994–2024.
Tables 6, 7 and 8 illustrate the transition probability matrices derived from LULC categories in the Lahore region for 1994–2004, 2004–2014, and 2014–2024, respectively. The tables show the likelihood of each LULC category changing into another over the specified periods.
From 1994 to 2004 (Table 7), water bodies had a 16.31% chance of remaining water, with the highest probability (51.42%) of converting to build-up areas. Vegetation had a high persistence rate of 72.54%, while barren land had a 36.27% probability of remaining the same and a 44.97% chance of becoming build-up areas. Build-up areas had a 68.04% probability of remaining unchanged.
From 2004 to 2014 (Table 8), water bodies showed a higher persistence of 23.01%, with a 37.28% likelihood of transitioning to build-up areas. Vegetation had a persistence of 71.65%, and barren land had a decreased persistence of 21.42%, with a higher probability (56.30%) of becoming build-up areas. Build-up areas remained relatively stable with a 70.75% chance of persistence.
In the period from 2014 to 2024 (Table 9), water bodies had a slightly increased persistence rate of 24% and a 59.82% probability of converting to Build-up areas. Vegetation maintained a high persistence rate of 75.62%, while barren land had a low persistence of 10.68%, with a significant likelihood (84.35%) of transitioning to build-up areas. Build-up areas had the highest persistence rate of 86.01%.
The data in these tables demonstrate the shifts in LULC over time, highlighting the increasing urbanization and changes in land use in the Lahore region. For example, the probability of build-up areas remaining as Build-up areas were 69.75% from 1994 to 2004, and cultivated lands had a 20.36% probability of changing to urban areas. Barren land and water bodies had greater persistent probabilities of 76.33% and 76.49%, respectively, in 2004.
The spatial distribution maps for LULC in 2024, both actual and predicted, reveal several key insights (Fig. 11). The actual map for 2024 indicates that Build-up areas dominate, accounting for the largest percentage of land cover, followed closely by vegetation. Water bodies and barren land constitute a smaller portion of the total area. The predicted map for 2024, generated using the CA-MHM, closely aligns with the actual distribution, though some discrepancies are noted. The model predicts a slightly higher percentage of vegetation, and a marginally lower percentage of Build-up areas compared to the actual data. Barren land is predicted to cover less area than observed, while the extent of water bodies remains consistent between the actual and predicted maps. These results underscore the efficiency of CA-MHM in capturing the overall LULC trends, though minor variations highlight areas for further refinement in predictive modelling.
The CA-MHM was validated in this research by comparing the actual LULC map for 2024 with the projected LULC map for that year. The model’s overall accuracy was evaluated using various essential metrics. The CA-Markov model attained an overall classification accuracy of 93.6%, signifying its capability to accurately forecast the geographical distribution of different land cover types in Lahore until 2024. This elevated percentage verifies that the model accurately reflects real LULC changes, indicating its dependability for predicting future developments. The Kappa coefficient was calculated to assess the concordance between the predicted and actual LULC maps, so further validating the model. The Kappa coefficient was 0.92, signifying nearly complete concordance between the two datasets. Per Kappa criteria, a result exceeding 0.81 indicates almost perfect agreement, hence substantiating the model’s predictive robustness. The validation results underscore the efficacy of the CA-MHM in precisely forecasting LULC changes over time. The amalgamation of a high overall accuracy percentage and a robust Kappa coefficient indicates that the model is a dependable instrument for modeling future land cover transitions, instilling confidence in its forecasts for 2034 and 2044.
Table 10 compares the actual and predicted LULC changes for 2024. The actual data indicates that water covers 1.02% of the area, which is predicted to decrease to slightly 1.01%. Vegetation, currently at 43.29%, is expected to rise to 46.47%. Barren land, which covers 4.08%, is predicted to decrease to 3.20%. Build-up areas make up 51.61% of the land but are expected to reduce to 49.32%. These predictions suggest an increase in vegetative cover at the expense of barren and build-up areas.
LULC change (a) Actual 2024 (b) Predicted 2024.
The CA-MHM was utilized to forecast the future LULC trends for the years 2034 and 2044. As shown in the thematic maps in Fig. 12, significant changes are expected in the land use and cover of Lahore. The projections indicate a substantial increase in urban areas, which will continue to expand at the expense of vegetated and barren lands. By 2034, urban areas are anticipated to become more dominant, and this trend is expected to continue into 2044. Vegetation is predicted to decrease steadily over both periods, reflecting the ongoing urbanization. Water bodies are forecasted to see a slight increase, while barren lands are expected to shrink considerably. These future scenarios underscore the ongoing trend of urban sprawl and the reduction of natural landscapes, highlighting the importance of effective urban planning and sustainable land management to address these changes. Figure 13 shows the change maps from 2024 to 2034 and 2034–2044.
Spatial Distribution Maps of LULC prediction (a) 2034 (b) 2044.
Future prediction LULC changes (a) 2024–2034 (b) 2034–2044.
The future predictions for LULC changes in Lahore show significant shifts between 2024, 2034, and 2044 (Table 11). The area covered by water bodies is expected to decline slightly by 0.45 km² from 2024 to 2034, and further reduce by 1.88 km² by 2044. Vegetation is projected to decline substantially, decreasing by 23.28 km² between 2024 and 2034, and another 56.30 km² by 2044. Barren land is anticipated to shrink dramatically, with a reduction of 23.70 km² in the first decade and an additional decrease of 7.51 km² in the subsequent decade. Conversely, Build-up areas are predicted to expand significantly, increasing by 47.07 km² from 2024 to 2034, and by 65.68 km² from 2034 to 2044. These trends highlight a continuing pattern of urbanization at the expense of natural and undeveloped land.
This study comprehensively analyzes LULC changes in Lahore from 1994 to 2024 and provides projections for 2034 and 2044 using the CA-MHM. The most important socioeconomic factors that are contributing to changes in LULC in Lahore are urbanization, population increase, and greater agricultural intensification. A direct result of the expansion of built-up areas is the decline in the amount of vegetation and land that is the absence of vegetation. To create specific measures that promote sustainable land management and eliminate the detrimental environmental impacts of urban expansion, it is essential to have a solid understanding of these factors. This study acknowledges the significance of rural transformations and agricultural shifts, even though urban development is the most prominent pattern that has been noticed in Lahore’s LULC alterations. Not only does the transformation of agricultural land into build-up areas diminish the possibility of food production, but it also changes the ecosystems of the surrounding area, which influences the levels of biodiversity and natural resources. As a result of the fact that rural regions supply crucial services such as food production and carbon sequestration, these changes are essential for the preservation of sustainable land management. To ensure the region’s ecological sustainability and food security over the long term, it is vital to take a balanced approach to regulating both the growth of urban areas and the transformations of rural areas.
Accurately and regularly updating LULC maps is significant for accurately calculating climatological factors. A study was conducted to examine how rainfall patterns in the large towns of Shanghai have changed over time based on LULC data. The study analyzed LULC data from 1979, 1990, 2000, and 2010, as well as daily rainfall trends from 1979 to 2010. The findings indicated that rain trends vary in different areas of Shanghai, including the fringes, suburbs, and the urban core. Specifically, the suburban areas experienced a notable increase in precipitation, while residential areas showed a decreasing trend in rainfall64. Another research conducted on the Yangtze River Delta in China observed that stations in highly urbanized areas from 1990 to 2015 showed the most pronounced negative annual trend in wind speed, highlighting the significant impact of urbanization in reducing wind speed65. A study examining the effect of LULC alternations on LST in the San Luis Potosí Basin, Mexico, from 2007 to 2020, found an annual rise of 11 °C in LST. The results indicated that areas classified as scarce vegetation or desert land had higher temperatures, whereas regions with dense foliage and water bodies showed lower temperatures66. The study conducted by67 examined the correlation between PM2.5 pollution and LULC alternations, focusing on Dhaka, Indonesia, which is known for high particulate matter pollution. They used Rapid Eye Landsat imagery to classify LULC in 2012 and 2018. Results showed a decrease in water bodies and vegetation cover by 2% and 5%, respectively, while barren lands increased by 4% and 3.4%, respectively. The study found a positive correlation between the decline in cultivated land and water bodies and the rise in PM2.5levels, highlighting the importance of LULC modifications in improving air quality. Researchers68 utilized various supervised classification methods, such as random forest and machine learning algorithms, to analyze Landsat imagery spanning from 2017 to 2020. Their study, conducted in the Tangail district of Bangladesh, aimed to investigate changes in LULC, particularly the rapid expansion of urban areas.
A study conducted in Iraq explores LULC dynamics under two business-as-usual scenarios, one assuming the continuation of past trends under UN sanctions until 2023, and the other assuming post-sanctions trends until 2023. The Cellular Automata-Markov chain model is used for simulation, with land use classes classified using Random Forests. The results indicate a trend towards stable and homogeneous areas by 2023, particularly in the scenario after the end of UN sanctions, which is beneficial for the park69. In a study focusing on northern China, the WRF model was employed to analyze climate variations influenced by environmental and socio-economic conditions. Simulations for 2001 and 2010 considered the dynamic interactions between LULC and climate. The study computed four biophysical parameters based on LULC changes: albedo, vegetation fraction, leaf area index (LAI), and emissivity. Results showed that LULC changes led to a significant reduction in summer temperatures and an increase in winter temperatures. Improved performance of the WRF model was noted with the use of updated LULC data33. Research conducted by34focuses on the Jiangle region in China, specifically a hilly area. It employs the CA-Markov model to predict land use patterns in 2025 and 2036. The CA-Markov model combines cellular automata and Markov chain analysis to forecast future land use based on historical data. Validation of the predictions using actual land use data from 2014 yielded a Kappa index of 0.8128, indicating a good fit. The researcher used CA-Markov to study LULC changes in Cholistan and Thal deserts, Punjab, Pakistan, for 1990, 2006, and 2022. Random Forest classified Landsat imagery with over 87% accuracy. Forecasting with CA-Markov showed LULC changes in 2022, extended to 2038. Urban sprawl analysis using Random Forest indicated growth in high and low-density residential areas by 203815. A study utilized spatial analysis techniques to examine LULC changes in Ravansar from 1992 to 2015. The CA-Markov model was then used to project the spatial pattern changes of LULC up to 2030. The findings revealed a significant increase in build-up and agricultural land (both aquatic and non-aquatic) from 1992 to 2015, leading to a decrease in gardens, range, and bare lands in the region70.
The results of this research concern specifically Lahore, a swiftly urbanizing metropolis with distinct socio-economic and environmental attributes. The model accurately represents the LULC variations in Lahore; however, the urban growth patterns, environmental pressures, and land use determinants may vary in other areas of Punjab or Pakistan. Consequently, the findings must be understood about the distinct urban dynamics of Lahore. Additional research is necessary to see if the trends identified in Lahore may be extrapolated to other places, especially rural areas or cities with varying developmental frameworks. However, because our study was constrained by the lack of comprehensive LULC change drivers, these projections should be interpreted with caution. Future research should focus on identifying the variables that drive LULC changes to improve the accuracy of these predictions. Our findings emphasize the need for the Lahore government to address developing urban areas into productive farmland to promote sustainable development. Agricultural practices should aim to maintain productivity while minimizing harm to ecosystems, aligning with FAO’s warning about increased land degradation due to rapid population growth.
This study conducted a thorough measurement of the uncertainty related to the CA-MHM model to improve the scientific reliability of our forecasting analyses. We assessed principal sources of uncertainty, including the quality and resolution of input data, which comprised Landsat imagery. By choosing exclusively high-quality images with under 5% cloud cover and applying geometric modifications, we achieved consistency across various time datasets. Furthermore, the transition probabilities obtained from historical land use alterations were examined for possible discrepancies induced by unforeseen socio-economic or environmental influences. Cross-validation methods and diverse calibration intervals were utilized to alleviate the impact of these anomalies. The model’s robustness was additionally validated by sensitivity analysis, which indicates negligible fluctuations in outcomes with alterations to input parameters. Our findings demonstrate a robust concordance between predicted and real land cover, evidenced by a Kappa coefficient of 0.92 and an overall accuracy of 93.6%. Although the model exhibits effectiveness, more studies could diminish uncertainty by incorporating supplementary socio-economic and environmental factors.
The results from this analysis provide crucial insights into the primary drivers of LULC changes in Lahore, particularly the fast urbanization and reduction in vegetation. These factors underscore the necessity for proactive ecological land management solutions that combine urban development with environmental preservation. With the expansion of metropolitan areas, it is imperative to adopt land management strategies that reduce the environmental consequences of these developments, such as habitat destruction, deforestation, and heightened strain on water supplies. This study’s findings can assist policymakers in formulating evidence-based strategies that enhance sustainable land management. Urban growth boundaries can be instituted to curtail sprawl and save agricultural land. The study’s LULC projections for 2034 and 2044 can assist urban planners in forecasting future development trends and strategizing to preserve green spaces and ecosystems. The incorporation of green infrastructure into urban planning is a principal recommendation from this study. Green infrastructure, including parks, green belts, and urban forests, contributes to ecological equilibrium, mitigates heat islands, and promotes biodiversity. Furthermore, prioritization of water management strategies is essential to minimize the observed decline in water bodies through the incorporation of rainwater harvesting and wetland restoration in urban initiatives.
It is recognized that LULC prediction models are crucial tools to promote sustainable land management. Policymakers and urban planners can make informed decisions that impact a balance between urban expansion and environmental conservation due to these models, which estimate future patterns in land use. A study highlights the use of geospatial models in the management of land incursions in sensitive areas25. Another study further highlights how predictive models can assist in evaluating the influence that land conversion has on the quality of habitat26. By predicting changes in catchment areas, the researcher demonstrates how LULC prediction models can contribute to the development of sustainable land–water management27. To provide insights that are essential for the development of plans for sustainable urban expansion, this study builds on these ideas by employing the CA-Markov model to anticipate changes in the LULC in Lahore. It is possible to ensure that principles of sustainability drive urban development by incorporating predictive models into land use plans. This will allow for the preservation of natural resources while also accommodating the growth of the city.
This study underscores the importance of using predictive LULC models, such as the CA-Markov model, as a tool for sustainable land management. By identifying the principal factors influencing LULC changes and incorporating these insights into policy development, decision-makers can guarantee that Lahore’s urban expansion is conducted in an environmentally sustainable manner.
The CA-Markov model, though useful for predicting LULC changes based on historical trends, has notable limitations. It assumes that land transitions occur linearly and predictably, which often fails to capture the complexities of real-world changes. Influences such as rapid urbanization, economic fluctuations, shifts in agricultural practices, and climate change can lead to irregular LULC patterns that this model does not effectively represent. To address these limitations, future studies should explore more sophisticated approaches that incorporate social, economic, and environmental dimensions. Agent-Based Models (ABMs), for example, simulate the behaviours and interactions of individual agents (households, businesses, or governmental bodies) with their environments, enabling more comprehensive, dynamic modelling. Additionally, machine learning techniques like random forests and neural networks can model complex, non-linear relationships and uncover hidden patterns, integrating variables like population growth, economic development, climate impacts, and land use regulations. These advanced models could lead to more accurate, multidimensional LULC forecasts, supporting sustainable urban planning and policymaking.
This study evaluated the effectiveness of the CA-MHM in predicting and modelling future LULC trends. By analyzing LULC changes from 1994 to 2024 and projecting trends for 2034 and 2044, it was found that the Markov model, integrated with remote sensing data, accurately predicts future LULC trends with high accuracy rates of 93% The study highlighted extensive urban expansion over the past three decades, with build-up areas increasing significantly about 359.8 km2at the expense of vegetation (−198.7km2) and barren lands (−158.5km2). Water bodies exhibited minor fluctuations but remained relatively stable. Transition probability matrices underscored the high likelihood of vegetated and barren lands transitioning into urban areas, reflecting continuous urbanization pressure. Future projections indicate these trends will persist, emphasizing the need for effective urban planning and sustainable development strategies in Lahore. The study identified a notable increase in urban sprawl due to population growth and industrial activities, while newly reclaimed agricultural lands remain insufficient to ensure sustainable food security. These findings underscore the urgent need for comprehensive land management strategies. By integrating RS, GIS, and Markov modelling, this study demonstrates an effective approach for mapping and monitoring LULC changes, though future research should refine the model by incorporating additional variables. Overall, this study provides valuable insights into the dynamic LULC changes in Lahore, highlighting significant urban expansion and its implications for natural landscapes, underscoring the urgency of adopting sustainable development practices.
Policymakers should prioritize the incorporation of greenery into development plans to control Lahore’s rapid growth while providing a sustainable environment. This involves the conservation of gardens, city forests, and green areas, which are vital for sustaining ecological equilibrium and reducing the urban heat island phenomenon. Furthermore, safeguarding agricultural land from uncontrolled urban expansion is essential. Enacting zoning regulations and instituting urban growth limits will aid in the preservation of arable land, hence enhancing food security and mitigating land degradation. Moreover, it is essential to execute water conservation measures. The reported reduction in water bodies necessitates the incorporation of measures like rainwater collecting, wetland restoration, and sustainable drainage systems into urban development strategies to safeguard water resources. The application of GIS and remote sensing techniques must be enhanced for the ongoing monitoring of land use changes. These instruments deliver precise, real-time information, allowing policymakers to make informed judgments and react promptly to land degradation or unlawful land use alterations. By implementing these guidelines, urban planners may foster equitable growth while preserving Lahore’s natural and agricultural environments.
Data will be available on the reasonable request from the 1st author.
Hu, Y. et al. Land Use / Land Cover Change Detection and NDVI Estimation in Pakistan ’ s Land Use / Land Cover Change Detection and NDVI Estimation in Pakistan ’ s Southern Punjab Province. (2023). https://doi.org/10.3390/su15043572
Haseeb, M., Tahir, Z., Mahmood, S. A., Batool, S. & Farooq, M. U. Spatial soil loss prediction impacted by long-term land use/land cover change: a case study of Swat District. Environ. Monit. Assess. 196, (2024).
Abdul Athick, A. S. M. & Shankar, K. Data on land use and land cover changes in Adama Wereda, Ethiopia, on ETM+, TM and OLI- TIRS landsat sensor using PCC and CDM techniques. Data Br. 24, 103880 (2019).
Article Google Scholar
Zhu, L., Song, R., Sun, S., Li, Y. & Hu, K. Land use/land cover change and its impact on ecosystem carbon storage in coastal areas of China from 1980 to 2050. Ecol. Indic. 142, 109178 (2022).
Article CAS Google Scholar
Wang, S. W., Munkhnasan, L. & Lee, W. K. Land use and land cover change detection and prediction in Bhutan’s high altitude city of Thimphu, using cellular automata and Markov chain. Environ. Challenges. 2, 100017 (2021).
Article CAS MATH Google Scholar
Kedia, S., Bhakare, S. P., Dwivedi, A. K., Islam, S. & Kaginalkar, A. Estimates of change in surface meteorology and urban heat island over northwest India: impact of urbanization. Urban Clim. 36, 100782 (2021).
Article Google Scholar
Khanal, N., Uddin, K. & Matin, M. A. Automatic detection of Spatiotemporal Urban expansion patterns by fusing OSM and Landsat Data in Kathmandu. (2019). https://doi.org/10.3390/rs11192296
Mallupattu, P. K. & Sreenivasula Reddy, J. R. Analysis of land use/land cover changes using remote sensing data and GIS at an urban area, Tirupati, India. Sci. World J. 2013, 1–7 (2013).
Article Google Scholar
Hussain, S., Mubeen, M. & Karuppannan, S. Land use and land cover (LULC) change analysis using TM, ETM + and OLI Landsat images in district of Okara, Punjab, Pakistan. Phys. Chem. Earth Parts A/B/C. 126, 103117 (2022).
Article Google Scholar
Sahana, M., Ahmed, R. & Sajjad, H. Analyzing land surface temperature distribution in response to land use/land cover change using split window algorithm and spectral radiance model in Sundarban Biosphere Reserve, India. Model. Earth Syst. Environ. 2, 81 (2016).
Article MATH Google Scholar
Yazdanpanah, M. et al. The impact of Livelihood assets on the Food Security of Farmers in Southern Iran during the COVID-19 pandemic. Int. J. Environ. Res. Public. Health 18, (2021).
Ojima, D. S., Galvin, K. A. & Turner, B. L. The global impact of land-use change: to understand global change, natural scientists must consider the social context influencing human impact on environment. Bioscience 44, 300–304 (1994).
Article MATH Google Scholar
Rane, N., Achari, A., Choudhary, S. & Giduturi, M. Effectiveness and Capability of Remote Sensing (RS) and Geographic Information Systems (GIS): A Powerful Tool for Land use and Land Cover (LULC) Change and Accuracy Assessment. 8, 286–295 (2023).
Waleed, M. & Sajjad, M. Leveraging cloud-based computing and spatial modeling approaches for land surface temperature disparities in response to land cover change: evidence from Pakistan. Remote Sens. Appl. Soc. Environ. 25, 100665 (2022).
MATH Google Scholar
Asif, M. et al. Modelling of land use and land cover changes and prediction using CA-Markov and Random Forest. Geocarto Int. 38, (2023).
Zhao, Q. et al. Evaluation of Land Use Land Cover changes in response to Land Surface temperature with Satellite indices and Remote Sensing Data. Rangel. Ecol. Manag. 96, 183–196 (2024).
Article MATH Google Scholar
Khan, R. et al. Monitoring land use land cover changes and its impacts on land surface temperature over Mardan and Charsadda Districts, Khyber Pakhtunkhwa (KP), Pakistan. Environ. Monit. Assess. 194, 409 (2022).
Article CAS PubMed Google Scholar
Sadiq Khan, M., Ullah, S., Sun, T., Rehman, A. U. R. & Chen, L. Land-Use/Land-Cover Changes and Its Contribution to Urban Heat Island: A Case Study of Islamabad, Pakistan. Sustainability 12, (2020).
Sarfo, A. K. & Karuppannan, S. Application of Geospatial Technologies in the COVID-19 fight of Ghana. Trans. Indian Natl. Acad. Eng. 5, 193–204 (2020).
Article PubMed PubMed Central MATH Google Scholar
Mukherjee, S. et al. Aquatic eco-systems under influence of Climate Change and anthropogenic activities: potential threats and its mitigation strategies. in 307–331 (2022). https://doi.org/10.1002/9781119870562.ch14
Kharazmi, R. et al. Monitoring and assessment of seasonal land cover changes using remote sensing: a 30-year (1987–2016) case study of Hamoun Wetland, Iran. Environ. Monit. Assess. 190, 356 (2018).
Birhanu, A., Masih, I., van der Zaag, P., Nyssen, J. & Cai, X. Impacts of land use and land cover changes on hydrology of the Gumara catchment, Ethiopia. Phys. Chem. Earth Parts A/B/C. 112, 165–174 (2019).
Article ADS Google Scholar
Choate, M., Rengarajan, R., Storey, J. & Lubke, M. Geometric calibration updates to Landsat 7 ETM + instrument for Landsat Collection 2 products. Remote Sens. 13, (2021).
Zoungrana, B. J. B., Conrad, C., Thiel, M., Amekudzi, L. K. & Da, E. D. MODIS NDVI trends and fractional land cover change for improved assessments of vegetation degradation in Burkina Faso, West Africa. J. Arid Environ. 153, 66–75 (2018).
Article ADS Google Scholar
Mozumder, C. & Tripathi, N. Geospatial scenario based modelling of urban and agricultural intrusions in Ramsar Wetland Deepor Beel in Northeast India using a multi-layer perceptron neural network. Int. J. Appl. Earth Obs Geoinf. 32, 92–104 (2014).
Google Scholar
Upadhaya, S. & Dwivedi, P. Conversion of forestlands to blueberries: assessing implications for habitat quality in Alabaha river watershed in Southeastern Georgia, United States. Land. use Policy. 89, 104229 (2019).
Article Google Scholar
Ouma, Y. O. et al. Land-Use Change Prediction in Dam Catchment Using Logistic Regression-CA, ANN-CA and Random Forest Regression and Implications for Sustainable Land–Water Nexus. Sustainability 16, (2024).
Kumar, L. & Mutanga, O. Google Earth Engine applications since inception: usage, trends, and potential. Remote Sens. 10, 1–15 (2018).
Article Google Scholar
Avci, C., Budak, M., Yagmur, N. & Balcik, F. B. Comparison between random forest and support vector machine algorithms for LULC classification. Int. J. Eng. Geosci. 8, 1–10 (2023).
Article MATH Google Scholar
Naghdyzadegan Jahromi, M. et al. Developing machine learning models for wheat yield prediction using ground-based data, satellite-based actual evapotranspiration and vegetation indices. Eur. J. Agron. 146, 126820 (2023).
Article Google Scholar
Sibanda, S. & Ahmed, F. Modelling historic and future land use/land cover changes and their impact on wetland area in Shashe sub-catchment, Zimbabwe. Model. Earth Syst. Environ. 7, 57–70 (2021).
Article MATH Google Scholar
Hyandye, C. & Martz, L. W. A Markovian and cellular automata land-use change predictive model of the Usangu Catchment. Int. J. Remote Sens. 38, 64–81 (2017).
Article Google Scholar
Ganjirad, M. & Bagheri, H. Ecological Informatics Google Earth Engine-based mapping of land use and land cover for weather forecast models using landsat 8 imagery. Ecol. Inf. 80, 102498 (2024).
Article MATH Google Scholar
Liping, C., Yujun, S. & Saeed, S. Monitoring and predicting land use and land cover changes using remote sensing and GIS techniques—A case study of a hilly area, Jiangle, China. PLoS One. 13, 1–23 (2018).
Article Google Scholar
Dey, N. N., Rakib, A., Kafy, A., Raikwar, V. & A.- Al & Geospatial modelling of changes in land use/land cover dynamics using multi-layer Perceptron Markov chain model in Rajshahi City, Bangladesh. Environ. Challenges. 4, 100148 (2021).
Article Google Scholar
Majumder, M. Introduction to Model Development for Prediction, Simulation and Optimization. (2023).
Li, Y. et al. Dynamics of Land Use/Land Cover Considering Ecosystem Services for a dense-Population Watershed based on a hybrid dual-subject Agent and Cellular Automaton modeling Approach. Engineering https://doi.org/10.1016/j.eng.2023.10.015 (2024).
Article Google Scholar
Bashir, O. et al. Simulating Spatiotemporal Changes in Land Use and Land Cover of the North-Western Himalayan Region Using Markov Chain Analysis. Land 11, (2022).
Belay, T. & Mengistu, D. A. Impacts of land use/land cover and climate changes on soil erosion in Muga watershed, Upper Blue Nile basin (Abay), Ethiopia. Ecol. Process. 10, 68 (2021).
Article Google Scholar
Amin, M. et al. Monitoring agricultural drought using geospatial techniques: a case study of thal region of Punjab, Pakistan. J. Water Clim. Chang. 11, 203–216 (2020).
Article MATH Google Scholar
Abbas, F. Analysis of a historical (1981–2010) temperature record of the Punjab Province of Pakistan. Earth Interact. 17, 1–23 (2013).
Article ADS MATH Google Scholar
Minallah, M. Retrieval of Land Surface temperature of Lahore through Landsat-8 TIRS Data. Int. J. Econ. Environ. Geol. 10, 70–77 (2019).
Google Scholar
Khokhar, M. F., Mehdi, H., Abbas, Z. & Javed, Z. Temporal assessment of NO2 pollution levels in urban centers of Pakistan by employing ground-based and satellite observations. Aerosol Air Qual. Res. 16, 1854–1867 (2016).
Article CAS Google Scholar
Basheer, M. A. & Waseem, M. A. Spatiotemporal Analysis of Urban Growth and Land Surface temperature: a case study of Lahore, Pakistan. Int. Arch. Photogramm Remote Sens. Spat. Inf. Sci. – ISPRS Arch. 48, 25–29 (2022).
Article MATH Google Scholar
Xue, J. & Su, B. Significant remote sensing vegetation indices: A review of developments and applications. J. Sensors (2017). (2017).
Kumari, M. & Sarma, K. Changing trends of land surface temperature in relation to land use/cover around thermal power plant in Singrauli district, Madhya Pradesh, India. Spat. Inf. Res. 25, 769–777 (2017).
Article MATH Google Scholar
Jaswal, S. & Thakur, P. Correlation between LST, NDVI and NDBI with reference to Urban Sprawling – A Case Study of Shimla city. Int. J. Multidiscip Res. 5, 1–14 (2023).
MATH Google Scholar
Haseeb, M. et al. Enhancing Carbon Sequestration through Afforestation: evaluating the impact of Land Use and Cover changes on Carbon Storage dynamics. Earth Syst. Environ. https://doi.org/10.1007/s41748-024-00414-z (2024).
Article MATH Google Scholar
Zhao, H. & Chen, X. Use of normalized difference bareness index in quickly mapping bare areas from TM/ETM+. in Proceedings. 2005 IEEE International Geoscience and Remote Sensing Symposium, 2005. IGARSS ’05. vol. 3 1666–1668 (2005).
Al-aarajy, K. H. A., Zaeen, A. A. & Abood, K. I. Supervised Classification Accuracy Assessment Using Remote Sensing and Geographic Information System. 13, 396–403 (2024).
Selmy, S. A. H. et al. Detecting, Analyzing, and Predicting Land Use / Land Cover (LULC) Changes in Arid Regions Using Landsat Images, CA-Markov Hybrid Model, and GIS Techniques. (2023).
Salvi, M., Acharya, U. R., Molinari, F. & Meiburger, K. M. The impact of pre- and post-image processing techniques on deep learning frameworks: a comprehensive review for digital pathology image analysis. Comput. Biol. Med. 128, 104129 (2021).
Article PubMed MATH Google Scholar
Abdul Rahaman, S., Aruchamy, S., Balasubramani, K. & Jegankumar, R. Land use/land cover changes in semi-arid mountain landscape in Southern India: a geoinformatics based Markov chain approach. Int. Arch. Photogramm Remote Sens. Spat. Inf. Sci. – ISPRS Arch. 42, 231–237 (2017).
Article Google Scholar
Firozjaei, M. K., Sedighi, A., Argany, M., Jelokhani-Niaraki, M. & Arsanjani, J. J. A geographical direction-based approach for capturing the local variation of urban expansion in the application of CA-Markov model. Cities 93, 120–135 (2019).
Article Google Scholar
Nouri, J., Gharagozlou, A., Arjmandi, R., Faryadi, S. & Adl, M. Predicting Urban Land Use Changes using a CA–Markov Model. Arab. J. Sci. Eng. 39, 5565–5573 (2014).
Article Google Scholar
Aliani, H., Malmir, M., Sourodi, M. & Kafaky, S. B. Change detection and prediction of urban land use changes by CA–Markov model (case study: Talesh County). Environ. Earth Sci. 78, 546 (2019).
Article ADS MATH Google Scholar
Ma, C., Zhang, G. Y., Zhang, X. C., Zhao, Y. J. & Li, H. Y. Application of Markov model in wetland change dynamics in Tianjin Coastal Area, China. Procedia Environ. Sci. 13, 252–262 (2012).
Article CAS MATH Google Scholar
Keshtkar, H. & Voigt, W. A spatiotemporal analysis of landscape change using an integrated Markov chain and cellular automata models. Model. Earth Syst. Environ. 2, 10 (2015).
Article MATH Google Scholar
Mannan, A. et al. Application of land-use/land cover changes in monitoring and projecting forest biomass carbon loss in Pakistan. Glob Ecol. Conserv. 17, e00535 (2019).
MATH Google Scholar
Laari, S. K. S. P. B., Szabó, S. & S. M. P. K. S. & Modelling of land use land cover change using earth observation data-sets of Tons River Basin, Madhya Pradesh, India. Geocarto Int. 33, 1202–1222 (2018).
Article ADS Google Scholar
Eastman, J. R. Guide to GIS and Image Processing. (2009).
Pan, S. et al. Runoff Responses to Climate and Land Use/Cover Changes under Future Scenarios. Water vol. 9 at (2017). https://doi.org/10.3390/w9070475
McHugh, M. L. Interrater reliability: the kappa statistic. Biochem. Med. 22, 276–282 (2012).
Article MATH Google Scholar
Jiang, Q. et al. Spatiotemporal analysis of land use and land cover (lulc) changes and precipitation trends in Shanghai. Appl. Sci. 10, 1–21 (2020).
Article Google Scholar
Zhang, G. et al. Rapid urbanization induced daily maximum wind speed decline in metropolitan areas: a case study in the Yangtze River Delta (China). Urban Clim. 43, 101147 (2022).
Article Google Scholar
Ovalle, A. G. C., Tristán, A. C., Amador-Nieto, J. A., Putri, R. F. & Zahra, R. A. Analysing the land use/land cover influence on land surface temperature in San Luis Potosí Basin, México using remote sensing techniques. IOP Conf. Ser. Earth Environ. Sci. 686, 12029 (2021).
Article Google Scholar
Zarin, T. & Esraz-Ul-Zannat, M. Assessing the potential impacts of LULC change on urban air quality in Dhaka city. Ecol. Indic. 154, 110746 (2023).
Article CAS MATH Google Scholar
Talukder, A., Mim, S. M., Ahmed, S., Syed, M. & Rahman, R. M. Machine learning and remote sensing technique for urbanization change detection in Tangail District. in Intelligent Sustainable Systems (eds Nagar, A. K., Jat, D. S. & Marín-Raventós, G.) (2022). & Mishra, D. K.) 241–249 (Springer Nature Singapore, Singapore.
Chapter Google Scholar
Hamad, R., Balzter, H. & Kolo, K. Predicting Land Use/Land Cover Changes Using a CA-Markov Model under Two Different Scenarios. Sustainability 10, (2018).
Karimi, H., Jafarnezhad, J., Khaledi, J. & Ahmadi, P. Monitoring and prediction of land use/land cover changes using CA-Markov model: a case study of Ravansar County in Iran. Arab. J. Geosci. 11, 592 (2018).
Article Google Scholar
Download references
The authors extended their appreciation to the Researchers Supporting Project number (RSPD2025R951), King Saud University, Riyadh, Saudi Arabia.
The authors acknowledge the Researchers Supporting Project number (RSPD2025R951), King Saud University, Riyadh, Saudi Arabia.
Institute of Space Science, University of Punjab, Lahore, 54780, Punjab, Pakistan
Zainab Tahir, Muhammad Haseeb & Syed Amer Mahmood
Centre For Integrated Mountain Research, University of the Punjab, Lahore, 54780, Punjab, Pakistan
Saira Batool
Department of Software Engineering, College of Computer and Information Sciences, King Saud University, Riyadh, 11543, Saudi Arabia
M. Abdullah-Al-Wadud
Department of Water Resources and Environmental Engineering, Nangarhar University, Jalalabad, Nangarhar, 2600, Afghanistan
Sajid Ullah
School of Resources and Environmental Engineering, East China University of Science and Technology, Shanghai, 200237, People’s Republic of China
Sajid Ullah
Department of Wildlife Fisheries and Aquaculture, College of Forest Resources, Mississippi State University, Mississippi State, MS, 39762-9690, USA
Aqil Tariq
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
Zainab Tahir and Muhammad Haseeb contributed significantly to this research, leading the conceptualization, methodology, software implementation, validation, formal analysis, and investigation. They were responsible for the initial drafting of the manuscript and subsequent revisions, providing critical insight throughout the editing process. Aqil Tariq contributed to the manuscript’s refinement through detailed review and editing. M. Abdullah-Al-Wadud and Sajid Ullah provided essential funding support, enabling the research to be conducted with the necessary resources. Saira Batool and Syed Amer Mahmood supervised the project, offering guidance to authors and ensuring the research met high academic standards. All authors have reviewed and approved the final manuscript, contributing to its completion and validation.
Correspondence to Muhammad Haseeb or Sajid Ullah.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Tahir, Z., Haseeb, M., Mahmood, S.A. et al. Predicting land use and land cover changes for sustainable land management using CA-Markov modelling and GIS techniques. Sci Rep 15, 3271 (2025). https://doi.org/10.1038/s41598-025-87796-w
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-025-87796-w
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
© 2025 Springer Nature Limited
Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research, free to your inbox weekly.