Next Article in Journal
The Impact of the Digital Economy on Industrial Eco-Efficiency in the Yangtze River Delta (YRD) Urban Agglomeration
Previous Article in Journal
Study on Trade Effects of Green Maritime Transport Efficiency: An Empirical Test for China Based on Trade Decision Model
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sustainability
Volume 15
Issue 16
10.3390/su151612329
sustainability-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Assessment of Land Use Land Cover Changes and Future Predictions Using CA-ANN Simulation for Gazipur City Corporation, Bangladesh

by
Md Shihab Uddin
1,
Badal Mahalder
1,* and
Debabrata Mahalder
2
1
Department of Water Resources Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh
2
Center for Natural Resource Studies, Dhaka 1213, Bangladesh
*
Author to whom correspondence should be addressed.
Sustainability 2023, 15(16), 12329; https://doi.org/10.3390/su151612329
Submission received: 6 June 2023 / Revised: 25 July 2023 / Accepted: 4 August 2023 / Published: 13 August 2023
Browse Figures
Review Reports Versions Notes

Abstract

:
Anthropogenic activities have a significant influence on land use and land cover (LULC) changes, especially in rapidly growing areas. Among several models, the combination of a cellular automata–artificial neural network (CA-ANN) model is being widely used for assessing future LULC changes using satellite images. This study aimed to investigate LULC changes in Gazipur City Corporation (GCC), Bangladesh, and the changes in LULC patterns over the last two decades (2002 to 2022). In this study, the maximum likelihood supervised classification technique was used for processing the available satellite images. The results show that the urban area and vegetation coverage increased by 150% and 22.78%, whereas the bare land and waterbody decreased by 7.02% and 78.9%, respectively, from 2002 to 2022 inside the GCC area. For future LULC predictions, the CA-ANN model was developed, the accuracy percentage of which was 86.49%, and the kappa value was 0.83. The future LULC prediction model results show that the urban area will increase by 47.61%, whereas the bare land and waterbody are supposed to decrease by 24.17% and 67.23%, respectively, by 2042. The findings of this study could be useful for future sustainable urban planning and management, as well as enabling decision making by authorities for improvements in environmental and ecological conditions in the study area.

1. Introduction

Land use and land cover (LULC) refers to the use of land, its resources, and the physical (bio) cover on the Earth’s surface (e.g., water bodies, plants, forests, agricultural land, and urban areas) [1,2,3]. It has been reported that LULC changes are the most significant anthropogenic disruption to the environment, causing a variety of microclimatic changes [4], which is important for the sustainable [5] and efficient management of natural resources and the environment. Proper planning for land use and other natural resources is dependent on the comprehensive knowledge of LULC for any area [6,7,8]. Furthermore, the qualitative and quantitative variations in land use, on a temporal scale, are also necessary for sustainable urban planning and management purposes [2,9]. Given the dynamic behavior of cities, these changes are unavoidable [10]; therefore, identifying trends in LULC change is necessary. Understanding changing patterns in LULC is also fundamental for formulating solutions to balance the demands for development, avoid conflicts, and the conservation of city planning and development [2].
Depending on the socioeconomic, political, and meteorological conditions of a city, changes in LULC are dependent on several factors. However, ‘population growth’ is the key for most urban settings [11]. The primary drivers of demographic, social, economic, and environmental development for emerging countries in recent decades are urbanization and industrialization [12]. The identification of LULC changes is also necessary, as LULC changes can alter the arrangement of spatial elements of different land surfaces [9,11,13,14]. However, to obtain correct LULC data from remotely sensed images, effective image classification techniques are fundamental [15]. Considering the requirements of increased accuracy and performance, machine learning classifiers have become powerful and widely used tools over the past few decades for LULC classifications [16]. In machine learning applications, numerous classification algorithms, such as supervised and unsupervised, parametric and nonparametric, hard and soft (i.e., fuzzy), and per-pixel- and subpixel-based classifiers, are available [17]. Recently, several models have also been developed for predicting future LULC dynamics for evaluating land use management policies.
Popular regression-based models used for the prediction of LULC changes include but are not limited to Markov chain analysis (MCA) or Markov models [18,19], cellular automation (CA) [20], cellular automata–Markov models (CA-Markov) [21], artificial neural networks (ANNs) [22], SLEUTH model [23], binary logistic regression, and fractal models. CA is a popular algorithm for pattern recognition, future prediction, and identifying the spatial distribution of LULC changes based on pixel analysis. However, the accuracy of the CA model in simulating urban expansion is low compared with other models, and it has limitations in handling the forcing of changing parameters in LULC studies [7]. Therefore, a combination of different models with the CA model is used by numerous researchers. Among the different combinations, ANNs are popular and widely used [9,13,24], as complex urban expansion patterns and nonlinear interactions among different components can be handled effectively using ANN models. In addition, compared to other regression-based models, the literature suggests that a combination of CA and ANN models performs better for LULC prediction [9,13,24,25]. Considering the capabilities of these two models (CA and ANN), they have been integrated (CA-ANN) to improve their performance in urban growth simulations, which was used in this study. Recently, in QGIS, the MOLUSCE (Modules of Land Use Change Evaluation) plugin was developed, where four regression-based models have been embedded for assessing LULC changes and future prediction. In machine-learning-based techniques, among several classification algorithms, maximum likelihood classification (MLC) is popular and widely used for image classification [26], using both supervised [27,28,29,30] and unsupervised classification [11,31,32,33]. In the literature, it is found that the MCA model has been extensively used for analyzing spatiotemporal satellite images to quantify environmental changes [21,34,35,36]. Several studies [2,10,25,37] used the amalgamation of the CA-ANN model for predicting LULC and other climatic parameters [38].
LULC and future land use prediction has been studied for different megacities or in fast-growing cities [7,9,13,14,39,40]. Gazipur city, adjacent to the capital city of Bangladesh, is an important industrial zone of the country. Previously, the area was mostly covered with dense forest, which is currently disappearing due to rapid industrialization [41]. In addition, because of rapid urbanization and economic activities, the land use patterns have also been altering. Abdullah et al. [41], Hassan et al. [42], and Arifeen et al. [3] attempted to study the LULC changes in Gazipur. Shapla et al. [33] studied the agricultural land cover changes of Gazipur District using unsupervised classification with different Landsat images. Abdullah et al. [41] investigated the causes of forest degradation, as well as quantifying the pattern of forest degradation in Gazipur using supervised classification based on the random forest algorithm. Arifeen et al. [3] studied the land use and land cover changes for Gazipur from 1990 to 2020. In Gazipur District, Gazipur City Corporation (GCC) is the main urban center, where the study of LULC changes and future land use change predictions is necessary for urban planning in order to develop a smart city. To the best of the authors’ knowledge, no study has been conducted to identify the LULC changes and predict the future LULC of the GCC area. Thus, in this study, an attempt was made to (i) assess the spatial and temporal changes in the LULC throughout the interlude of the past two decades (2002, 2012, and 2022) using multitemporal remote sensing satellite data for the GCC and (ii) to predict the future LULC changes using the CA-ANN model for the years 2032 and 2042.

2. Materials and Methods

2.1. Study Area

The GCC is one of the twelve city corporations of Bangladesh, covering the largest area (321.90 sq. km) among all of the city corporations in the country. It was established in 2013, and the geographic location of the GCC area is shown in Figure 1. The main rivers of the GCC area are Turag and Bangshi [43]. In the city corporation area, the population is approximately 6.5 million. The total road length inside the GCC area is approximately 1552.83 km. The city is one of the major hubs for employment, commerce, and entertainment, which accounts for nearly one-third of the national GDP. In addition, 75% of the country’s total garments industry is located in the GCC [43]. Therefore, people from neighboring districts, particularly from rural areas, are settling in the GCC area, as the job market is expanding rapidly. Therefore, in-migration is the main cause of the recent population growth in the GCC [44]. Because of this rapid urbanization, the city area is facing massive energy issues and shortages of natural resources.

2.2. Data Collection

To conduct this study, digital elevation model (DEM) data were required, which were collected from the Shuttle Radar Topography Mission (SRTM) of the United States Geological Survey (USGS). Multiple spectral and temporal Landsat 7 and 8 data, between the years 2002 and 2022, covering the GCC area were collected from the USGS website (www.earthexplorer.usgs.gov, accessed on 23 April 2022). Distances from major roads were calculated using vector layers and data collected from the OpenStreetMap (OSM). The details of the data used in this study are listed in Table 1. These datasets had been radiometrically corrected and projected to UTM 46N in GIS for further analysis. The radiometric correction of the pixel values or faults in the values was performed, which enhances the quality and readability of remote sensing data. The radiometric correction and calibration were of primary significance, since this research compared various datasets collected from different sources and different temporal ranges. Furthermore, Google Earth images, which refer to base maps, were used to carry out the classification of urban areas, waterbodies, vegetation, etc., as well as to recognize the spectral signature on the image of several LULC classes.

2.3. Methodology

2.3.1. Multitemporal Data Processing and LULC Monitoring

Landsat imageries were covered by clouds, and specifically, Landsat 7 exhibited scan line errors in various areas. Such flight-line errors are common with the Landsat 7 (along-track) satellite. The ‘Landsat Toolbox’ feature within ArcMap 10.3.1 software was used to address these issues. After eliminating cloud cover and rectifying scan line errors, classification procedures were conducted. Finally, a supervised classification system was performed for this study.

2.3.2. Supervised Classification

Various LULC classes were selected as training datasets, utilizing False Color Composite (FCC) and appropriate band combinations. False-color images are used to highlight or reveal details that might remain hidden or partially visible. This study used the mid-infrared, near-infrared, and blue bands to assemble images and select a training dataset. The Maximum Likelihood Supervised Classification (MLSC) method was used to categorize Landsat images into four major LULC groups for the years 2002, 2012, and 2022: bare land, urban areas, vegetation, and waterbodies (Table 2).
Approximately 20 samples were trained to create LULC maps for each class, resulting in a total of 80 samples being trained for the four land use classes. To assess the accuracy of land cover maps, 160 ground truths points were obtained from Google Earth images. The accuracy of the image classification was evaluated using kappa statistics and the confusion matrix, widely considered as robust quantitative measurements for image classification accuracy assessment [39]. Note that Landsat 9 images are available from the beginning of 2022, and their verification was not found within the study period. As a result, Landsat 8 images were used in this study.

2.3.3. LULC Change Prediction

To predict LULC trends in the GCC area, the MOLUSCE tool, available in the 2.18.3 edition of QGIS software, was used in conjunction with the CA simulator [45]. The CA function in QGIS operates in a Markovian manner, as it relies on the current state of land use, rather than the preceding one [46]. This process takes inputs from prior and current land use maps, along with spatial factors, and generates the desired outputs, as shown in Figure 2. For the modeling of transitional potential in LULC, ANN algorithms were chosen due to their higher accuracy compared to alternative methods [47]. The ANN is composed of neurons similar to those in the human brain and works to unveil relationships in data, as shown in Figure 3.
LULC maps representing the years 2002, 2012, and 2022 were used to predict LULC maps for the years 2032 and 2042. Within the input module, independent variables (LULC maps) were paired with dependent variables such as DEM (elevation), slope, aspect, distance from major roadways, distance from waterbodies, and population data. To generate a simulated LULC map, two separate LULC maps from different years were used as inputs. The difference in the year of the input LULC map acted as the difference in the predicted map year with the final input LULC map year. For instance, when using the LULC maps of 2012 and 2022, the predicted LULC map represented the year 2032.

2.3.4. Preparation of Parameters for the Future LULC Simulation

Existing literature highlights the significance of various parameters for future LULC prediction, including elevation, slope, aspect, distance to transport network, distance to the waterbody, distance to vegetation, population, etc. [9]. In this study, considering available data and other land use patterns, six parameters were selected for future prediction. Among these parameters, slope and elevation data were generated using SRTM DEM. Topographic features such as elevation, slope, and aspect play a crucial role in the transformation of LULC [48]. Furthermore, distances to major roads were calculated using vector layers obtained from the open street map (OSM). The proximity to road networks is a pivotal aspect of urban development. People usually migrate to the city areas for work due to the concentrated industrial presence in the GCC area. Given the prevalence of migration to urban areas for employment, the distance from the roads can significantly influence land use change predictions. Distance from waterbody and population data were downloaded from the DIVA-GIS website (https://www.diva-gis.org/gdata, accessed on 30 June 2023). Population density significantly impacts LULC change, with higher population densities leading to more rapid LULC changes. Conversely, greater distances from water bodies tend to reduce the frequency of LULC conversion, while shorter distances tend to increase conversion rates of LULC [9]. The parameters utilized in this study are shown in Figure 4.

2.3.5. Transitional Prospective of ANN Modelling

In this study, the learning rate was set to 0.01, where 50 pixels were randomly selected for individual classes, the maximum number of iterations was set to 200, and the momentum was set to 0.01 in transient prospect simulation [39]. Simulations of one iteration implied that the simulation would be conducted once and, based on the difference between the beginning and final year, the simulation of the next time span will be generated.

2.3.6. CA-ANN Model Validation

The null model method was used for validation purposes, representing a fundamental validation technique [32]. In the MOLUSCE framework, simulation was guided by the t1 time feed map, while validation involved comparing the simulated t2 map with the observed t2 map. In this methodology, if the null model (t1 map) exhibited greater similarity to the t2 map than the simulated map did, the simulation was deemed unsatisfactory. Furthermore, kappa values were also assessed within the MOLUSCE, encompassing kappa statistics such as kappa histogram, local kappa, and total kappa ranging from 0 to 1.

3. Results

3.1. Land Use Distribution

The MLSC algorithm was used to evaluate various LULC classes for the years 2002, 2012, and 2022, as previously mentioned. The results from the analysis are presented in Table 3 and Figure 5. For the accuracy assessment, 40 samples from each class were chosen. The LULC data for each year were meticulously cross-verified with Google Earth to ensure precision in the accuracy measurement. To evaluate the performance of a given classifier, confusion matrices, user accuracy, and producer accuracy were used. From Table 4, it is evident that the overall classification accuracy achieved by MLSC was 91.875%, 92.5%, and 92.5% for the years 2002, 2012, and 2022, respectively. This level of overall accuracy indicates that more than 90 percent of errors were successfully avoided with the classification procedure.
From image analyses, it has been found that bare land constituted the highest area among all other classes inside the GCC area, though the percent of coverage varied from year to year. In 2002, bare land covered nearly 45% of the total land use, while it was increased to 51.40% in 2012. However, it was decreased to 41.50% in the year 2022. Among other classes, vegetation cover displayed a consistent pattern in each year, exhibiting minimal changing patterns. Waterbody accounted for about 22% of the total GCC area in 2002, whereas they accounted for about 4.6% in 2022 (Table 3).
Urban areas were found only covering 9.90% in 2002, whereas it was found expanded to 25.1% in the year 2022. The results strongly indicated a significant surge in urbanization from the year 2002 to 2022, which also supports the overall in-migration pattern to the area. For further clarification, Figure 6 illustrates the changes in different LULC from 2002 to 2022. In the year 2002, vegetation cover was 75.64 km2 (Table 3), which was almost 23.5% of the total area. Vegetation cover increased to 26.60% in 2012 and further to 28.90% in 2022. A smaller increase in vegetation cover could have been influenced by the classification methods as park areas, playgrounds, trees, grasslands, croplands, and fallow land were defined as vegetation cover. Moreover, the increase in vegetation cover could have also been influenced by various private and governmental initiatives within the GCC area.

3.2. Multitemporal LULC Change Assessment

A key aspect of forecasting future LULC changes is the analysis of land use changes between different LULC classes. The relative changes in LULC during different temporal scales, as found from this study, are presented in Figure 7. With the exception of one land use class, bare land, both increasing and decreasing trends have been observed for all classes. The vegetation and the urban areas showed an increasing pattern, whereas the waterbodies consistently showed a decreasing pattern within the GCC area. As stated earlier, bare land was the sole class displaying irregularities. It was observed that bare land initially increased from 143.55 km2 to 165.36 km2 during the year 2002 to 2012 and then decreased to 133.479 km2 from 2012 to 2022. A possible explanation for this observation is that during the first decade, some waterbodies might have transformed into bare land, leading to their classification as such.
Results from this study also showed that overall bare land decreased to 7.02% from the year 2002 to 2022, whereas vegetation increased by 22.78% during the same period. Figure 8 represents the interchanges among various LULC classes, the gains and losses for LULC classes, and net changes from 2002 to 2022. It can be observed that from 2002 to 2022, the waterbody decreased by 78.9%. The reduction was at its maximum during 2002 to 2012, which was almost 59.60%, and during the next decade it was reduced by 47.90%. As reported earlier, urban area within the GCC area was found to be increasing; it increased by 32.3% during the years 2002 to 2012, and further 90.70% during the subsequent decade (2012 to 2022). So, the overall change in urban area was found to be about 150% from the year 2002 to 2022.
From Figure 8, it can also be found that the majority of bare land was transformed into urban areas, likely involving reclamation from waterbodies. As reported in Figure 8, it can also be observed that major gains or losses during the last two decades were primarily within the bare land and waterbodies. Decreasing the bare land and waterbodies contributed to the increase in other land use classes. However, bare land contributed most to the overall increase in the urban areas, followed by vegetation and waterbodies. From the results, another interesting trend in land use class changes was observed for the period 2012–2020, wherein the net changes for bare land and waterbodies were negative.

3.3. Performances of CA-ANN Simulation for Future Prediction

Using the LULC datasets of 2002, 2012, and 2022, the CA-ANN model was developed to forecast future land use for the years 2032 and 2042. The CA-ANN model was initially used to predict the LULC for the year 2022 in order to ensure the accuracy of prediction results using LULC data from the year 2002 and 2012. For model validation, the MOLUSCE plugin of QGIS was used to calculate the kappa coefficient. Following the validation processes for the year 2022, the values obtained for kappa local, kappa histogram, and kappa overall were 0.85, 0.87, and 0.83, respectively. Overall, 86.49% validation accuracy was found for the CA-ANN model.

3.4. LULC Prediction Using CA-ANN

The CA-ANN simulation was carried out to predict future land use changes utilizing LULC inputs from the years 2012 to 2022 as spatial and temporal variables. Results from this study also indicated that the vegetation area is supposed to increase by 5.11% from the year 2022 to 2032, followed by a decrease of 0.9% from 2032 to 2042. However, model results indicate that the overall vegetation cover will remain nearly similar, at approximately 30.33% and 30.06% in 2032 and 2042, respectively. Figure 9 shows the predicted LULC of the GCC area. It can also be observed that the bare land area is supposed to be decreased by 17.11% and 8.51% in 2032 and 2042, respectively.
The model results also indicate a continuous decline in waterbodies by 41.17% in 2032 and 44.29% in 2042. This suggests that the available waterbody area could constitute about 2.72% of the total GCC area in 2032, which might be further reduced to 1.52% in the year 2042. As expected, from the year 2022 to 2032, urbanization is supposed to be increased by 30.04%, followed by a more moderate increase of about 13.52% in 2042. Thus, the model findings suggest that total urban area will comprise approximately 32.58% and 36.98% of the GCC area in 2032 and 2042, respectively. This implies that the areas previously occupied by bare land, vegetation, and waterbodies might undergo transformation into urban areas (Table 5). Note that in this CA-ANN model, among various input variables, the existing roadway networks have been used. Currently the GCC does not have any masterplan for future roadway expansion; therefore, the model might have underpredicted the extent of future urban areas.

3.5. LULC Transition Matrices from 2032 to 2042

The transition matrix holds significance in studying the temporal variations across various LULC categories. In this study, the future land use change matrices of LULC classes during the years 2032 to 2042 are presented in Table 6. The rows indicate the categories during the initial year, while the columns depict the same sequence of LULC categories in the final year. Diagonal entries indicate the extent of stability within a specific land class, while off-diagonal entries represent the scale of transition from one land class to another. Entries near 1 on the diagonal scale signify the stability of a particular category. Urban areas, bare land, and vegetation display the most consistent trends, with transition values of 0.969, 0.897, and 0.890. However, waterbody, with a score of 0.762, are expected to undergo rapid fragmentation. The waterbody class is projected to contribute 0.229 to the bare land category, while vegetation is anticipated to contribute 0.110 to the urban area category. The matrices indicate that vegetation area has the lowest probability of changing compared with other land use classes.

4. Discussion

Numerous studies have focused on LULC classification and the prediction of future LULC patterns based on various GIS tools and machine learning algorithms. Though the classification and prediction outcomes can vary depending on several key factors/parameters, the selection of appropriate machine learning algorithms remains a pivotal component in accurate future LULC predictions. In this study, the MLSC method was used to classify the land use within the GCC area. This method has been extensively used in the literature because of its robust performance and straightforward classification techniques [16]. In addition, the CA-ANN machine learning model was developed to predict future LULC trends. The CA-ANN model not only predicts future LULC patterns but also highlights the impact of each conditioning parameter on the prospective changes in land use [49]. Given the socio-economic and demographic importance of the study area, future LULC prediction is important for policy making and the formulation of sustainable urban development plans.
Findings from this study demonstrated the model’s efficacy in correctly identifying the LULC maps, achieving an accuracy level surpassing 90%. Furthermore, the validation accuracy for the predicted LULCs exceeded 80%. Literature suggests a minimum desirable LULC classification accuracy of around 70% in remote sensing data analysis [50,51]. Validation accuracy values of 80% or higher are considered indicative of a robust prediction model for future land use projections [39,52]. The findings of this study, therefore, demonstrated the model’s performance within the context of its application.
In general, the results from our study aligned with the overall trend of LULC observed in the Gazipur area, consistent with previous studies [3,33,41,42]. In this study, it has been observed that the vegetation cover was increasing, where forests, shrubs, grasslands, and agricultural lands have been included in this classification. It has been identified that because of the close proximity to the capital city of Bangladesh and the rapid growth of the population in Gazipur, food security is a major issue. Therefore, several private and governmental initiatives have been taken to increase the crop production. As a result, more bare land and low-lying waterbodies have transitioned into agricultural fields, leading to an increase in vegetation cover within the GCC area. The upward trend in vegetation covers mirrors the findings of other studies as well [3,33].
Due to rapid urbanization, deforestation and land grabbing are expected. From the historical image analysis, it has been observed that the waterbodies are reducing at an alarming rate within the GCC area. Moreover, the high valuation of land and its scarcity have spurred industrial encroachments on natural water bodies (e.g., rivers and lakes), resulting in the significant alteration of the natural waterbodies inside the GCC area. The model results also indicated a substantial decline in waterbody in the future, with projected waterbody coverage estimated at approximately 2.72% in 2032 and diminishing further to 1.52% by 2042 within the GCC area. The analysis also unveils the unauthorized conversion of waterbodies into bare land due to land grabbing (Figure 6 and Figure 9). These findings are also supported by the observations of similar studies conducted in proximity to the GCC area [41].
As expected, our results reveal that for the years 2022 to 2032, urbanization is supposed to be increased by 30.04%, followed by a slower growth rate of approximately 13.52% in the subsequent decade. These trends in increasing urbanization rate aligned with observations from other cities in Bangladesh, as found from the literature. Kafy et al. [53] studied the future urbanization increasing trend of Rajshahi city and reported an approximate 15.06% increase by the year 2029. Similarly, projections for Dhaka city indicated potential 18.35% growth in urbanization by 2055. Since Dhaka city is already developed, the future expansion is expected to be more subdued [54], whereas the GCC area is undergoing rapid urbanization, contributing to the higher growth rate projected for the year 2032 in our study. However, it is also expected that once major urban settings have been developed, the growth rate will be reduced. The future urbanization rate in the year 2042 also supports this hypothesis, and the urbanization rate was found to be similar ranges as those reported in the literature [53,54].
The results from this study also demonstrate yet another important parameter, i.e., the transition matrix, which is important in analyzing LULC dynamics. The significance has also been substantiated by preceding studies of similar natures. The transition matrix provides a valuable tool for drawing insights into LULC dynamics by analyzing the net change, rate of change, and the types of changes that transpire between two distinct time periods [55]. In literature, it has been found that during future land use prediction, for a land use class, if the transition matrix value is close to unity, it is more likely that the land use class will be changed to others [10,39,56]. In this study, similar findings have been observed (Table 6), which were then used for generating future land use change maps.
Among various machine learning algorithms, the CA-ANN model has been found to be a viable model for future LULC predictions, despite its classification as a ‘black box’ due to occasional inaccuracies [9]. The CA-ANN model assembles training samples by collecting input from different layers without considering their relative relevance while picking the most significant variables. The system lacks weighting requirements for the input parameters [57]. Moreover, the accuracy of LULC simulation utilizing the CA-ANN model can vary based on the type and quantity of input parameters. However, notwithstanding its shortcomings, the model has been widely utilized worldwide to predict future LULC patterns [14,58,59,60,61,62]. Although the CA-ANN model produces credible findings, as found in this research, it is recommended to combine CA with other machine learning techniques such as random forest (RF) or decision tree (DT) algorithms to achieve higher accuracy. Acknowledging the overall limitations of this study, a similar study design could be formulated by calculating more parameters such as Land Surface Temperature (LST), Urban Heat Index (UHI), NDVI, etc., for future LULC investigations in the GCC area.

5. Conclusions

The objective of this study was to assess the LULC changes and forecast the future land use patterns of the GCC area, Bangladesh. Utilizing the MLSC algorithm, in this study, we assessed LULC changes within the GCC area from the year 2002 to 2022. Furthermore, a CA-ANN model was developed for predicting future land use patterns of the year 2032 and 2042. From the study results, it has been found that the urban area was expanded by approximately 150% between 2002 to 2022, with a substantial increase of nearly 100% occurring during the last decade. The waterbodies witnessed a sharp decline of 78.9% from 2002 to 2022, subsequently being replaced by bare land, vegetation, and urban areas. In the GCC area, bare land decreased by 7.02%, while vegetation increased by 22.78% in 2022 from the year 2002. The future LULC prediction using the CA-ANN model of the GCC area revealed a 30.04% increase in urban areas from 2022 to 2032, followed by an additional 13.52% increase by the year 2042. Overall urban areas are supposed to be increased by about 47.61% in 2042 compared with 2022. The model result also predicted that bare land would decline continuously in both 2032 and 2042. Among others, the CA-ANN model predicted a decreasing trend for waterbodies, which will be about 2.72% and 1.52% of the total area in 2032 and 2042, respectively. The decreasing trends of bare land and waterbodies from the year 2022 to 2042 are supposed to be about 24.17% and 67.23%, respectively. Meanwhile, vegetation cover is projected to exhibit an increasing trend in 2032, followed by a subsequent decline in 2042.
Considering the model results, the GCC area is poised to emerge as one of the fastest growing urban areas of Bangladesh over the next decade. The acceleration of present and future industrial development is anticipated to play a pivotal role in the rapid urbanizations. Therefore, this study could be helpful to the GCC administrative authorities, urban planners, policymakers, and various stakeholders involved in future city development and planning. However, in this study, no future development plan or microclimatic changes or scenarios have been incorporated. This could be the subject of future investigation for the GCC area and other rapidly growing urban areas of Bangladesh for the sustainable environmental management of natural resources.

Author Contributions

Conceptualization, M.S.U., B.M. and D.M.; validation, M.S.U. and B.M.; resources, M.S.U.; data curation, M.S.U. and B.M.; software, M.S.U.; supervision, B.M. and D.M.; visualization, M.S.U.; writing—original draft preparation, M.S.U., B.M. and D.M.; writing—review and editing, B.M. and D.M. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

All the data used and produced are reported in this manuscript. However, if required, data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgments

The authors are grateful for three anonymous reviewers’ comments and suggestions for enhancing the quality of the manuscript. The authors also acknowledge the financial support from Bangladesh University and Technology (BUET) for covering the APC charges through the Basic Research Fund received by the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Mishra, V.N.; Rai, P.K. A remote sensing aided multi-layer perceptron-Markov chain analysis for land use and land cover change prediction in Patna district (Bihar), India. Arab. J. Geosci. 2016, 9, 249. [Google Scholar] [CrossRef]
  2. Saputra, M.H.; Lee, H.S. Prediction of Land Use and Land Cover Changes for North Sumatra, Indonesia, Using an Artificial-Neural-Network-Based Cellular Automaton. Sustainability 2019, 11, 3024. [Google Scholar] [CrossRef] [Green Version]
  3. Arifeen, H.M.; Phoungthong, K.; Mostafaeipour, A.; Yuangyai, N.; Yuangyai, C.; Techato, K.; Jutidamrongphan, W. Determine the Land-Use Land-Cover Changes, Urban Expansion and Their Driving Factors for Sustainable Development in Gazipur Bangladesh. Atmosphere 2021, 12, 1353. [Google Scholar] [CrossRef]
  4. Roberts, D.; Batista, G.; Pereira, J.; Waller, E.; Nelson, B. Change Identification Using Multitemporal Spectral Mixture Analysis: Applications in Eastern Amazonia; Ann Arbor Press: Ann Arbor, MI, USA, 1999; pp. 137–158. [Google Scholar]
  5. Belenok, V.; Hebryn-Baidy, L.; Bielousova, N.; Gladilin, V.; Kryachok, S.; Tereshchenko, A.; Alpert, S.; Bodnar, S. Machine learning based combinatorial analysis for land use and land cover assessment in Kyiv City (Ukraine). J. Appl. Remote Sens. 2023, 17, 014506. [Google Scholar] [CrossRef]
  6. Sheeja, R.V.; Joseph, S.; Jaya, D.S.; Baiju, R.S. Land use and land cover changes over a century (1914–2007) in the Neyyar River Basin, Kerala: A remote sensing and GIS approach. Int. J. Digit. Earth 2011, 4, 258–270. [Google Scholar] [CrossRef]
  7. Singh, B.; Venkatramanan, V.; Deshmukh, B. Monitoring of land use land cover dynamics and prediction of urban growth using Land Change Modeler in Delhi and its environs, India. Environ. Sci. Pollut. Res. 2022, 29, 71534–71554. [Google Scholar] [CrossRef]
  8. Rabby, Y.W.; Li, Y.; Abedin, J.; Sabrina, S. Impact of Land Use/Land Cover Change on Landslide Susceptibility in Rangamati Municipality of Rangamati District, Bangladesh. ISPRS Int. J. Geo-Inf. 2022, 11, 89. [Google Scholar] [CrossRef]
  9. Shahfahad; Naikoo, M.W.; Das, T.; Talukdar, S.; Asgher, S.; Asif; Rahman, A. Prediction of land use changes at a metropolitan city using integrated cellular automata: Past and future. Geol. Ecol. Landsc. 2022, 1–19. [Google Scholar] [CrossRef]
  10. Baig, M.F.; Mustafa, M.R.U.; Baig, I.; Takaijudin, H.B.; Zeshan, M.T. Assessment of Land Use Land Cover Changes and Future Predictions Using CA-ANN Simulation for Selangor, Malaysia. Water 2022, 14, 402. [Google Scholar] [CrossRef]
  11. Kafi, K.M.; Shafri, H.Z.M.; Shariff, A.B.M. An analysis of LULC change detection using remotely sensed data; A Case study of Bauchi City. IOP Conf. Ser. Earth Environ. Sci. 2014, 20, 012056. [Google Scholar] [CrossRef]
  12. Dewan, A.M.; Yamaguchi, Y. Land use and land cover change in Greater Dhaka, Bangladesh: Using remote sensing to promote sustainable urbanization. Appl. Geogr. 2009, 29, 390–401. [Google Scholar] [CrossRef]
  13. Rahman, F.; Rahman, M.T.U. Use of cellular automata-based artificial neural networks for detection and prediction of land use changes in North-Western Dhaka City. Environ. Sci. Pollut. Res. 2023, 30, 1428–1450. [Google Scholar] [CrossRef] [PubMed]
  14. Lukas, P.; Melesse, A.M.; Kenea, T.T. Prediction of Future Land Use/Land Cover Changes Using a Coupled CA-ANN Model in the Upper Omo–Gibe River Basin, Ethiopia. Remote Sens. 2023, 15, 1148. [Google Scholar] [CrossRef]
  15. Hazarika, N.; Das, A.K.; Borah, S.B. Assessing land-use changes driven by river dynamics in chronically flood affected Upper Brahmaputra plains, India, using RS-GIS techniques. Egypt. J. Remote Sens. Space Sci. 2015, 18, 107–118. [Google Scholar] [CrossRef] [Green Version]
  16. Zhang, P.; Gong, M.; Su, L.; Liu, J.; Li, Z. Change detection based on deep feature representation and mapping transformation for multi-spatial-resolution remote sensing images. ISPRS J. Photogramm. Remote Sens. 2016, 116, 24–41. [Google Scholar] [CrossRef]
  17. Shetty, S. Analysis of Machine Learning Classifiers for LULC Classification on Google Earth Engine. 2019. Available online: http://essay.utwente.nl/83543/ (accessed on 23 September 2022).
  18. Sivakumar, V. Urban Mapping and Growth Prediction using Remote Sensing and GIS Techniques, Pune, India. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. 2014, XL-8, 967–970. [Google Scholar] [CrossRef] [Green Version]
  19. Redowan, M.; Phinn, S.; Roelfsema, C.; Aziz, A.A. Modeling forest cover dynamics in Bangladesh using multilayer perceptron neural network with Markov chain. J. Appl. Remote Sens. 2022, 16, 034502. [Google Scholar] [CrossRef]
  20. Sinha, S.; Sharma, L.K.; Nathawat, M.S. Improved Land-use/Land-cover classification of semi-arid deciduous forest landscape using thermal remote sensing. Egypt. J. Remote Sens. Space Sci. 2015, 18, 217–233. [Google Scholar] [CrossRef] [Green Version]
  21. Subedi, P.; Subedi, K.; Thapa, B. Application of a Hybrid Cellular Automaton C Markov (CA-Markov) Model in Land-Use Change Prediction: A Case Study of Saddle Creek Drainage Basin, Florida. Appl. Ecol. Environ. Sci. 2013, 1, 126–132. [Google Scholar] [CrossRef] [Green Version]
  22. Mozumder, C.; Tripathi, N.K. 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. 2014, 32, 92–104. [Google Scholar] [CrossRef]
  23. Osman, T.; Divigalpitiya, P.; Arima, T. Using the SLEUTH urban growth model to simulate the impacts of future policy scenarios on land use in the Giza Governorate, Greater Cairo Metropolitan region. Int. J. Urban Sci. 2016, 20, 407–426. [Google Scholar] [CrossRef]
  24. Gharaibeh, A.; Shaamala, A.; Obeidat, R.; Al-Kofahi, S. Improving land-use change modeling by integrating ANN with Cellular Automata-Markov Chain model. Heliyon 2020, 6, e05092. [Google Scholar] [CrossRef]
  25. Rahman, M.T.U.; Tabassum, F.; Rasheduzzaman, M.; Saba, H.; Sarkar, L.; Ferdous, J.; Uddin, S.Z.; Islam, A.Z.M.Z. Temporal dynamics of land use/land cover change and its prediction using CA-ANN model for southwestern coastal Bangladesh. Environ. Monit. Assess. 2017, 189, 565. [Google Scholar] [CrossRef] [PubMed]
  26. Allam, M.; Bakr, N.; Elbably, W. Multi-temporal assessment of land use/land cover change in arid region based on landsat satellite imagery: Case study in Fayoum Region, Egypt. Remote Sens. Appl. Soc. Environ. 2019, 14, 8–19. [Google Scholar] [CrossRef]
  27. 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 ABC 2022, 126, 103117. [Google Scholar] [CrossRef]
  28. Nagne, A.D.; Vibhute, A.D.; Dhumal, R.K.; Kale, K.V.; Mehrotra, S.C. Urban LULC Change Detection and Mapping Spatial Variations of Aurangabad City Using IRS LISS-III Temporal Datasets and Supervised Classification Approach. In Data Analytics and Learning; Nagabhushan, P., Guru, D.S., Shekar, B.H., Kumar, Y.H.S., Eds.; Springer: Singapore, 2019; pp. 369–386. [Google Scholar]
  29. Uddin, M.S.; Mahalder, B. Land use and land pattern changes estimation due to cyclone Amphan for Tala upazila, Satkhira using google earth engine. In Proceedings of the 5th Annual Paper Meet and 2nd Civil Engineering Congress, Dhaka, Bangladesh, 29–30 July 2022. [Google Scholar]
  30. Vivekananda, G.; Swathi, R.; Sujith, A. Multi-temporal image analysis for LULC classification and change detection. Eur. J. Remote Sens. 2021, 54, 189–199. [Google Scholar] [CrossRef]
  31. Lang, R.; Shao, G.; Pijanowski, B.C.; Farnsworth, R.L. Optimizing unsupervised classifications of remotely sensed imagery with a data-assisted labeling approach. Comput. Geosci. 2008, 34, 1877–1885. [Google Scholar] [CrossRef]
  32. Saponaro, M.; Tarantino, E. LULC Classification Performance of Supervised and Unsupervised Algorithms on UAV-Orthomosaics. In Computational Science and Its Applications—ICCSA 2022 Workshops; Gervasi, O., Murgante, B., Misra, S., Rocha, A.M.A.C., Garau, C., Eds.; Springer International Publishing: Cham, Switzerland, 2022; pp. 311–326. [Google Scholar]
  33. Shapla, T.; Park, J.; Hongo, C.; Kuze, H. Agricultural Land Cover Change in Gazipur, Bangladesh, in Relation to Local Economy Studied Using Landsat Images. Adv. Remote Sens. 2015, 4, 214–223. [Google Scholar] [CrossRef] [Green Version]
  34. Halmy, M.W.A.; Gessler, P.E.; Hicke, J.A.; Salem, B.B. Land use/land cover change detection and prediction in the north-western coastal desert of Egypt using Markov-CA. Appl. Geogr. 2015, 63, 101–112. [Google Scholar] [CrossRef]
  35. Roy, S.; Farzana, K.; Papia, M.; Hasan, M. Monitoring and Prediction of Land Use/Land Cover Change using the Integration of Markov Chain Model and Cellular Automation in the Southeastern Tertiary Hilly Area of Bangladesh. Int. J. Sci. Basic Appl. Res. IJSBAR 2015, 24, 125–148. [Google Scholar]
  36. Vázquez-Quintero, G.; Solís-Moreno, R.; Pompa-García, M.; Villarreal-Guerrero, F.; Pinedo-Alvarez, C.; Pinedo-Alvarez, A. Detection and Projection of Forest Changes by Using the Markov Chain Model and Cellular Automata. Sustainability 2016, 8, 236. [Google Scholar] [CrossRef] [Green Version]
  37. Zeshan, M.T.; Mustafa, M.R.U.; Baig, M.F. Monitoring Land Use Changes and Their Future Prospects Using GIS and ANN-CA for Perak River Basin, Malaysia. Water 2021, 13, 2286. [Google Scholar] [CrossRef]
  38. Gupta, R.; Singhal, S. Prediction of Global Solar Radiation in India using Artificial Neural Network. J. Sustain. Dev. Energy Water Environ. Syst. 2016, 4, 94–106. [Google Scholar] [CrossRef] [Green Version]
  39. Hossain, M.T.; Zarin, T.; Sahriar, M.R.; Haque, M.N. Machine learning based modeling for future prospects of land use land cover change in Gopalganj District, Bangladesh. Phys. Chem. Earth Parts ABC 2022, 126, 103022. [Google Scholar] [CrossRef]
  40. Kamaraj, M.; Rangarajan, S. Predicting the future land use and land cover changes for Bhavani basin, Tamil Nadu, India, using QGIS MOLUSCE plugin. Environ. Sci. Pollut. Res. 2022, 29, 86337–86348. [Google Scholar] [CrossRef] [PubMed]
  41. Abdullah, H.M.; Islam, I.; Miah, M.G.; Ahmed, Z. Quantifying the spatiotemporal patterns of forest degradation in a fragmented, rapidly urbanizing landscape: A case study of Gazipur, Bangladesh. Remote Sens. Appl. Soc. Environ. 2019, 13, 457–465. [Google Scholar] [CrossRef]
  42. Hassan, M.A.; Mahjabin, R.; Islam, M.R.; Imtiaz, S. Land Cover Classification and Change Detection Analyzing Multi-Temporal Landsat Data: A Case Study of Gazipur Sadar, Bangladesh between 1973 and 2017. Geogr. Environ. Sustain. 2019, 12, 104–118. [Google Scholar] [CrossRef] [Green Version]
  43. GCC. At a Glance. Bangladesh National Portal. 2018. Available online: http://gcc.gov.bd/site/page/0f4394e7-0406-422e-9e43-f2ec6455ecb9/ (accessed on 23 September 2022).
  44. Ahmed, N. There’s Still Time to Save Gazipur. The Daily Star, 28 September 2021. Available online: https://www.thedailystar.net/views/opinion/news/theres-still-time-save-gazipur-2185336(accessed on 26 September 2022).
  45. NEXTGIS. MOLUSCE—Quick and Convenient Analysis of Land Cover Changes. NEXTGIS, 4 November 2013. Available online: https://nextgis.com/blog/molusce/(accessed on 26 September 2022).
  46. Lu, Y.; Wu, P.; Ma, X.; Li, X. Detection and prediction of land use/land cover change using spatiotemporal data fusion and the Cellular Automata–Markov model. Environ. Monit. Assess. 2019, 191, 68. [Google Scholar] [CrossRef]
  47. Gašparović, M.; Jogun, T. The effect of fusing Sentinel-2 bands on land-cover classification. Int. J. Remote Sens. 2018, 39, 822–841. [Google Scholar] [CrossRef]
  48. Birhane, E.; Ashfare, H.; Fenta, A.A.; Hishe, H.; Gebremedhin, M.A.; Wahed, H.G.; Solomon, N. Land use land cover changes along topographic gradients in Hugumburda national forest priority area, Northern Ethiopia. Remote Sens. Appl. Soc. Environ. 2019, 13, 61–68. [Google Scholar] [CrossRef]
  49. Al Kafy, A.A.; Faisal, A.A.; Rahman, S.; Islam, M.; Al Rakib, A.; Islam, A.; Khan, H.H.; Sikdar, S.; Sarker, H.S.; Mawa, J.; et al. Prediction of seasonal urban thermal field variance index using machine learning algorithms in Cumilla, Bangladesh. Sustain. Cities Soc. 2021, 64, 102542. [Google Scholar] [CrossRef]
  50. Shahfahad; Mourya, M.; Kumari, B.; Tayyab, M.; Paarcha, A.; Asif; Rahman, A. Indices based assessment of built-up density and urban expansion of fast growing Surat city using multi-temporal Landsat data sets. GeoJournal 2021, 86, 1607–1623. [Google Scholar] [CrossRef]
  51. Naikoo, M.W.; Rihan, M.; Ishtiaque, M.; Shahfahad. Analyses of land use land cover (LULC) change and built-up expansion in the suburb of a metropolitan city: Spatio-temporal analysis of Delhi NCR using landsat datasets. J. Urban Manag. 2020, 9, 347–359. [Google Scholar] [CrossRef]
  52. Congalton, R.G. A review of assessing the accuracy of classifications of remotely sensed data. Remote Sens. Environ. 1991, 37, 35–46. [Google Scholar] [CrossRef]
  53. Kafy, A.-A.; Naim, N.H.; Khan MH, H.; Islam, M.A.; Al Rakib, A.; Al-Faisal, A.; Sarker, M.H.S. Prediction of Urban Expansion and Identifying Its Impacts on the Degradation of Agricultural Land. In Re-Envisioning Remote Sensing Applications, 1st ed.; Singh, R., Ed.; CRC Press: Boca Raton, FL, USA, 2021; pp. 85–106. [Google Scholar] [CrossRef]
  54. Kafy, A.-A.; Naim, N.H.; Subramanyam, G.; Faisal, A.-A.; Ahmed, N.U.; Al Rakib, A.; Kona, M.A.; Sattar, G.S. Cellular Automata approach in dynamic modelling of land cover changes using RapidEye images in Dhaka, Bangladesh. Environ. Chall. 2021, 4, 100084. [Google Scholar] [CrossRef]
  55. Chang, Y.; Hou, K.; Li, X.; Zhang, Y.; Chen, P. Review of Land Use and Land Cover Change research progress. IOP Conf. Ser. Earth Environ. Sci. 2018, 113, 012087. [Google Scholar] [CrossRef]
  56. Tuljapurkar, S.; Steinsaltz, D. Stochastic Models for Structured Populations. In Handbook of Statistics; Elsevier: Amsterdam, The Netherlands, 2019; pp. 133–155. [Google Scholar] [CrossRef]
  57. Shatnawi, N.; Qdais, H.A. Mapping urban land surface temperature using remote sensing techniques and artificial neural network modelling. Int. J. Remote Sens. 2019, 40, 3968–3983. [Google Scholar] [CrossRef]
  58. Berberoğlu, S.; Akın, A.; Clarke, K.C. Cellular automata modeling approaches to forecast urban growth for adana, Turkey: A comparative approach. Landsc. Urban Plan. 2016, 153, 11–27. [Google Scholar] [CrossRef]
  59. Shafizadeh-Moghadam, H.; Asghari, A.; Tayyebi, A.; Taleai, M. Coupling machine learning, tree-based and statistical models with cellular automata to simulate urban growth. Comput. Environ. Urban Syst. 2017, 64, 297–308. [Google Scholar] [CrossRef]
  60. Muhammad, R.; Zhang, W.; Abbas, Z.; Guo, F.; Gwiazdzinski, L. Spatiotemporal Change Analysis and Prediction of Future Land Use and Land Cover Changes Using QGIS MOLUSCE Plugin and Remote Sensing Big Data: A Case Study of Linyi, China. Land 2022, 11, 419. [Google Scholar] [CrossRef]
  61. Islam, M.Y.; Nasher, N.M.R.; Karim, K.H.R.; Rashid, K.J. Quantifying forest land-use changes using remote-sensing and CA-ANN model of Madhupur Sal Forests, Bangladesh. Heliyon 2023, 9, e15617. [Google Scholar] [CrossRef] [PubMed]
  62. Amgoth, A.; Rani, H.P.; Jayakumar, K.V. Exploring LULC changes in Pakhal Lake area, Telangana, India using QGIS MOLUSCE plugin. Spat. Inf. Res. 2023, 31, 429–438. [Google Scholar] [CrossRef]
Sustainability 15 12329 g001
Figure 1. Study area.
Figure 1. Study area.
Sustainability 15 12329 g001
Sustainability 15 12329 g002
Figure 2. Flowchart of methodology.
Figure 2. Flowchart of methodology.
Sustainability 15 12329 g002
Sustainability 15 12329 g003
Figure 3. CA-ANN model for future LULC prediction.
Figure 3. CA-ANN model for future LULC prediction.
Sustainability 15 12329 g003
Sustainability 15 12329 g004
Figure 4. Parameters used for LULC prediction: (a) existing DEM of study area, (b) slope, (c) aspect, (d) distance from road (m), (e) distance from waterbody, and (f) population density.
Figure 4. Parameters used for LULC prediction: (a) existing DEM of study area, (b) slope, (c) aspect, (d) distance from road (m), (e) distance from waterbody, and (f) population density.
Sustainability 15 12329 g004
Sustainability 15 12329 g005
Figure 5. Changes in LULC classes in percentage from 2002 to 2022.
Figure 5. Changes in LULC classes in percentage from 2002 to 2022.
Sustainability 15 12329 g005
Sustainability 15 12329 g006
Figure 6. Historical LULC map for the years (a) 2002, (b) 2012, and (c) 2022.
Figure 6. Historical LULC map for the years (a) 2002, (b) 2012, and (c) 2022.
Sustainability 15 12329 g006
Sustainability 15 12329 g007
Figure 7. Comparison of changes between 2002 and 2012 and 2012 and 2022.
Figure 7. Comparison of changes between 2002 and 2012 and 2012 and 2022.
Sustainability 15 12329 g007
Sustainability 15 12329 g008aSustainability 15 12329 g008b
Figure 8. Increases and decreases, net changes, and cumulative contributing percentage of urban area between 2002 and 2022: (a) gain and losses, (b) net changes, and (c) total contribution of net changes in urban area.
Figure 8. Increases and decreases, net changes, and cumulative contributing percentage of urban area between 2002 and 2022: (a) gain and losses, (b) net changes, and (c) total contribution of net changes in urban area.
Sustainability 15 12329 g008aSustainability 15 12329 g008b
Sustainability 15 12329 g009
Figure 9. Predicted LULC from CA-ANN simulation for the year (a) 2032 and (b) 2042.
Figure 9. Predicted LULC from CA-ANN simulation for the year (a) 2032 and (b) 2042.
Sustainability 15 12329 g009
Table 1. Data source and availability date.
Table 1. Data source and availability date.
Data LayerSourceAvailable PeriodResolution (m)
DEMUSGS201430 × 30
Landsat 7
Enhanced Thematic Mapper (ETM+)		USGSMarch, 200230 × 30
Landsat 7
Enhanced Thematic Mapper (ETM+)		USGSMarch, 201230 × 30
Landsat 8
Operational Land Imager (OLI)		USGSFebruary, 202230 × 30
Table 2. Description of LULC types.
Table 2. Description of LULC types.
LULC TypeDescription
Bare LandVacant land, bare soils, sand and landfill sites
Urban AreasResidential, commercial, industrial and transportation network
VegetationPark, playground, trees, grassland, cropland and fallow land
WaterbodyRiver, wetlands, lakes, ponds and reservoirs
Table 3. Statistics of different LULC classes from 2002 to 2022.
Table 3. Statistics of different LULC classes from 2002 to 2022.
LULC ClassArea (km2)Percent Area (%)
200220122022200220122022
Vegetation75.6485.6792.8823.5026.6028.90
Bare Land143.56165.37133.4844.6051.4041.50
Waterbody70.7328.5814.8922.008.904.60
Urban Area31.9742.2980.659.9013.1025.10
Total321.90321.90321.90100100100
Table 4. Accuracy assessment of land cover classification for 2002, 2012, and 2022.
Table 4. Accuracy assessment of land cover classification for 2002, 2012, and 2022.
YearClassified ClassValidation Points for Different LULC Classes
WaterbodyUrbanVegetationBare LandTotalUser Accuracy
2002Waterbody380204095
Urban Area136034090
Vegetation033704094.9
Bare Land040364090
Total39433939160
Producer Accuracy97.483.794.992.3Overall Accuracy 91.875Kappa Coefficient 89.16
2012Waterbody370304092.5
Urban Area036044090
Vegetation033704092.5
Bare Land020384095
Total37414042160
Producer Accuracy10087.892.590.5Overall Accuracy 92.5Kappa Coefficient 90
2022Waterbody370304092.5
Urban Area137024092.5
Vegetation043604090
Bare Land020384095
Total38433940160
Producer Accuracy97.58697.395Overall Accuracy 92.5Kappa Coefficient 90
Table 5. Projected LULC classes in 2032 and 2042.
Table 5. Projected LULC classes in 2032 and 2042.
LULC ClassEstimated Area (km2)Estimated Area (%)Change from 2022 (%)
203220422032204220322042
Vegetation97.62996.753330.3330.065.114.17
Bare Land110.637101.21834.3731.44−17.11−24.17
Waterbody8.764.882.721.52−41.17−67.23
Urban Area104.874119.048732.5836.9830.0447.61
Total321.9321.9100100
Table 6. LULC transition matrices from 2032 to 2042.
Table 6. LULC transition matrices from 2032 to 2042.
LULC ClassesLand Use 2042
WaterbodyVegetationBare LandUrban
Land Use 2032Waterbody0.7620.0020.2290.007
Vegetation00.89000.110
Bare Land00.0570.8970.046
Urban00.0000.0310.969
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Uddin, M.S.; Mahalder, B.; Mahalder, D. Assessment of Land Use Land Cover Changes and Future Predictions Using CA-ANN Simulation for Gazipur City Corporation, Bangladesh. Sustainability 2023, 15, 12329. https://doi.org/10.3390/su151612329

AMA Style

Uddin MS, Mahalder B, Mahalder D. Assessment of Land Use Land Cover Changes and Future Predictions Using CA-ANN Simulation for Gazipur City Corporation, Bangladesh. Sustainability. 2023; 15(16):12329. https://doi.org/10.3390/su151612329

Chicago/Turabian Style

Uddin, Md Shihab, Badal Mahalder, and Debabrata Mahalder. 2023. "Assessment of Land Use Land Cover Changes and Future Predictions Using CA-ANN Simulation for Gazipur City Corporation, Bangladesh" Sustainability 15, no. 16: 12329. https://doi.org/10.3390/su151612329

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop