Assessment of Land Use /Land Cover Changes and its Future Predictions Using CA-Markov with ANN Simulation for Berhampore, West Bengal, India

doi:10.21203/rs.3.rs-3407386/v1

Download PDF

Research Article

Assessment of Land Use /Land Cover Changes and its Future Predictions Using CA-Markov with ANN Simulation for Berhampore, West Bengal, India

https://doi.org/10.21203/rs.3.rs-3407386/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Land use and land cover (LULC) change is a multifaceted and dynamic process influenced by factors like population growth, economic development, and climate shifts. This study delves into the LULC changes spanning 1991 to 2021 in Berhampore, a district headquarters in Murshidabad, West Bengal, India. Notable findings include an expansion of built-up land from 5.25–9.30%, a reduction in agricultural land from 81.98–72.36%, and an increase in plantation or forest land from 8.45–13.23%. The change transition matrix highlights significant shifts, notably the transformation of agricultural land into built-up areas (15.92 km²) and conversion to plantation or forest land (25.96 km²) and water bodies (5.54 km²). A chord diagram visually represents the transition matrix's outcomes. Utilizing the Future Land Use Simulation (FLUS) model, the study forecasts Berhampore's LULC for 2031. Forecasts indicate an ongoing increase in built-up land and a decrease in agricultural land. Concurrently, the area of plantation or forest land is projected to expand. Employing remote sensing and GIS techniques, the research tracks LULC changes and engages in a focus group discussion with local stakeholders. Findings underscore the intricate interplay between LULC, urbanization, and environmental dynamics. The study underscores the urgency of sustainable city planning, resource management, and community involvement to manage these transformations while preserving community well-being and ecological equilibrium. As cities expand and populations grow, this research highlights the imperative to balance development with environmental preservation for the greater benefit of society and nature.

Artificial Neural Networks. Land Use Land cover

Chord Diagram

Remote Sensing

Geographic Information System

Markovchain

Geo-simulation

community engagement

Over epochs of geological history, the earth's surface transformations have been meticulously chronicled(Bhalli & Abdul, 2015). Yet, it was during the late Pleistocene that humanity's imprint on the environment commenced, amplifying with the advent of agricultural and urban societies(Naikoo et al., 2020; Showqi et al., 2014). While human endeavors have perennially reshaped the land for subsistence, the accelerated pace of exploitation has induced unparalleled alterations in local, regional, and global ecosystems(Amin & Fazal, 2012). Amidst these transformations, the escalating interplay of fast-paced population growth, industrial expansion, and economic prosperity emerges as prime drivers of global land use change(Alam et al., 2020; Bhalli & Abdul, 2015).

The visible natural layer that blankets the earth's surface is commonly known as land cover, while "land usage" signifies how this natural covering is practically employed (Naikoo et al., 2020). Typically, the way people use the land emerges from the interplay of a society's cultural background, its preferences, and the inherent capabilities of the land itself(Muthusamy et al., 2010; Rawat et al., 2013). Throughout the time, people have sculpted the land's countenance to mirror social, economic, and natural dynamics, engendering swift metamorphoses in land usage and land cover owing to an array of socio-economic activities and natural forces (Das & Sarkar, 2019; Fasona et al., 2022). This palpable transformation of the Earth's surface finds resonance with the global web of environmental dynamics, a synergy particularly discernible in regions undergoing rapid land use/cover transitions. These transitions tend to be more pronounced in developing regions, marked by their rapidity, delicate ecological balance, and limited natural resources (Patel & Tripathi, 2016). The dynamics of land use/cover changes pose paramount challenges for cultivating environmentally sustainable economic growth, emphasizing the imperative of judicious land exploitation (Rawat et al., 2013).

In light of this, optimizing land use becomes a pivotal conduit for enhancing economic resilience without exacerbating ecological strain (Muthusamy et al., 2010; Rawat et al., 2013). Crucially, comprehending the repercussions of human interventions upon the natural milieu lays the foundation for effective management strategies (Patel & Tripathi, 2016; Prakasam, 2010). This comprehension extends to the accurate mapping and prediction of evolving landforms, necessitating regular, real-time data updates (Bhalli & Abdul, 2015). The amalgamation of remote sensing (RS) and Geographic Information Systems (GIS) emerges as a potent methodology, facilitating the integration and analysis of land use/cover data (Bhalli & Abdul, 2015; Mhawish & Saba, 2016; Mishra et al., 2020; Prakasam, 2010).

Employing remotely sensed data expedites a cost-effective and precise exploration of land cover changes, circumventing temporal and spectral complexities(Muthusamy et al., 2010; Rawat et al., 2013). Beyond efficiency gains and precision, the digital demarcation of changes, fueled by multi-temporal and multispectral technologies, emerges as a robust tool for decoding landscape dynamics, permitting the delineation, cartography, and tracking of land use/cover evolution, regardless of causative factors(Rawat et al., 2013). Within this paradigm, landscape models furnish invaluable insights into the forces sculpting terrains, while geospatial analytical models decode dynamic transformations, substantiating land use and land cover tracking as a swift, reliable, and potent methodology, propelled by the ascendancy of advanced remote sensing and GIS technology (Das & Sarkar, 2019). The Markov model, a stalwart in the analysis of land use cover, metamorphoses, and future predictions, finds prominence as a tool of choice (Hamad et al., 2018). In the forthcoming discourse, this research paper endeavors to dissect the temporal evolution and transitions of land use and land cover in the study area, during the period from 1991 to 2021. Moreover, by employing the innovative fusion of Cellular Automata-Markov with Artificial Neural Networks (ANN), the paper aspires to forecast the forthcoming landscape mosaic of the study region in 2031, and predicated the patterns of land use and land cover shifts.

The study Area

The present research work focuses on the Berhampore CD Block (Fig. 1), strategically positioned within the Ganges-Bhagirathi Basin, a distinctive elongated river basin characterized by its unique geographical attributes. The Bhagirathi River intricately divides this expanse, forming a natural demarcation between the Rarh zone to the west and the Bagri zone to the east. With an average elevation of 18 meters above sea level, the region boasts fertile soil, rendering it exceptionally suitable for agricultural practices. Encompassing the geographical coordinates from 23°50' N to 24°10' N latitudes and 88°10' E to 88°25' E longitudes, Berhampore exhibits a climatic imprint defined by an average annual precipitation of 1465.5 mm. The temperature range, spanning from 8.8°C to 42.4°C, encapsulates the diverse climatic nuances experienced within this locale. According to the 2011 Census of India, the Berhampore CD Block sustains a total population of 642,110 individuals, distributed between its rural and urban landscapes. Among these, 337,623 inhabitants reside in rural expanses, while 304,487 dwell within urban precincts. The gender distribution reveals a slight male majority at 51.22 percent, with females constituting 48.78 percent of the populace. The significance of Berhampore as a pivotal administrative center draws an influx of inhabitants, consequently instigating rapid transformations in land use and land cover (LULC) patterns across the study area. Over the past two decades, the district's central position has amplified its appeal, leading to a remarkable surge in population. Notably, the population escalated from 413,910 in 1991 to an impressive 642,110 in 2011, signifying an impressive growth rate of nearly 55 percent over this two-decade span.

Database

The task of delineating Land Use and Land Cover (LULC) involved the meticulous analysis of multispectral images obtained through remote sensing techniques. The chosen tool for LULC mapping in this study is Landsat Imagery, primarily due to its ready availability and consistent acquisition schedule (Bose & Chowdhury, 2020). The USGS Earth Explorer website proved invaluable, offering cost-free access to a repository of satellite images. Specifically, cloud-free multispectral images from various months across each year were employed(Maity et al., 2020). Comprehensive details pertaining to these images are meticulously outlined in Table 1. To augment the dataset, road data was sourced from an open street map, while crucial service areas were marked using Google Earth. The creation of a slope map and the assessment of settlement proximity, derived from classified satellite images, were facilitated by ASTER Global Dem Version 3. The entire dataset was processed using ArcGIS 10.5 software, efficiently integrating road networks, service centers, settlements, and Digital Elevation Model (DEM) data. Incorporating demographic insights, population data spanning the years 1991, 2001, and 2011 was harnessed from the District Census Handbook of Murshidabad. A comprehensive summary of the utilized data and their respective origins is presented in Table 2. Additionally, the exploration of the primary drivers of LULC dynamics was facilitated through a series of focus group discussions (FGD) involving three separate groups of the study area. Each focus group consisted of 4 to 7 participants, representing diverse perspectives including local elders, community leaders, school teachers, members of non-governmental organizations (NGOs), and agricultural extension workers. Each discussion, each spanning approximately 45 minutes, offered a platform to engage in-depth conversations and insights. Furthermore, a total of 15 key informants from the local population, who had been residing in the study area for over 40 years, were identified. These key informants were willing participants in interviews aimed at acquiring additional insights into the factors contributing to LULC changes. The key informant interviews provided valuable qualitative data that complemented the quantitative analysis and added depth to the understanding of the intricacies of LULC dynamics (Abebe et al., 2022). The integration of qualitative methods like FGDs and KIIs alongside quantitative techniques enhanced the comprehensiveness of the research methodology, allowing for a more holistic exploration of the various drivers (Bufebo & Elias, 2021) and dimensions of land-use and land-cover changes in the study area. The procedural framework is visually encapsulated in Fig. 2, elucidating the steps undertaken to execute this research. The core objective of this study's methodology rests in the classification of satellite images, aimed at extracting thematic insights for both land use and land cover categorizations.

Table 1

Details of satellite images used in the study
Year	Acquisition Date	Satellite	Sensor	Path/Row	Resolution (m)	Projection
1991	24-01-1991	Landsat 5	TM	139/43	30	UTM-WGS84
2001	19-01-2001	Landsat 5	TM	139/43	30	UTM-WGS84
2011	14-12-2010	Landsat 5	TM	139/43	30	UTM-WGS84
2021	25-12-2020	Landsat 8	OLI/TIRS	139/43	30	UTM-WGS84

Source: USGS Earth Explorer

Table 2

Data and sources used in the study
Data	Source	Year
Satellite Images	https://earthexplorer.usgs.gov/	1991,2001,2011,2021
ASTER Global Dem Version 3	https://earthdata.nasa.gov/	2013
Road	https://www.openstreetmap.org	2021
Service centers	Google Earth Pro	2015
Settlement Proximity	Classified Satellite Image	2021
Census Data	Census of India Districts Hand book	1991,2001,2011

Source: Compiled by authors

Methodology

Image pre-processing

Employing ERDAS IMAGINE 2014 software, thematic data were gleaned from the Landsat images. The initial step involved transforming Landsat images from Digital Numbers (DN) to surface reflectance through the DOS1(Dark Object Subtraction) method, carried out using QGIS software (Das & Sarkar, 2019). These surface reflectance data products play a pivotal role in environmental analysis, especially in cases involving supervised classifications. Notably, the conversion to surface reflectance enhances accuracy, a notable advantage when multiple images are employed for assessing land cover changes (Chavez, 1996). To establish distinctive signatures for each land use class, visual perception methods were utilized. These signatures were then employed for precise identification purposes. Subsequently, utilizing ERDAS IMAGINE, data preprocessing was executed for activities like mosaicking and image sub-setting, tailored to the area of interest (AOI). Supervised classification was conducted through the Maximum Likelihood Classification (MLC) technique, a pixel-oriented approach embedded within ERDAS IMAGINE 2014. This technique, widely recognized for its effectiveness in identifying land use and land cover classes, underpins the detection process. Initially, each land use/cover class was defined, and a series of training pixels were collected, which were then combined to form a class. The classified satellite images underwent necessary corrections (Recode) to generate the Land Use and Land Cover (LULC) map. The approach of supervised classification using MLC has been embraced by recent studies like Rahman et al. (2012), Kaliraj et al. (2017), Garai and Narayana (2018), Chowdhury et al. (2020), Kangabam et al. (2019), Choudhury et al. (2019), Gharaibeh et al. (2020), Mishra et al. (2020), Alam et al. (2021), Roy and Kasemi (2021), Saber et al. (2021), all converging on its efficacy in detecting LULC changes. In line with this methodology, this study identified and described four primary land use and land cover categories, namely, water bodies, agricultural land, plantation or forest land, and built-up area, as elaborated in Table 3.

Table 1

LULC Class Description
LULC Classes	Descriptions
Agriculture	Land that is mainly used to grow crops and vegetables. Only Cropland
Built-up	man-made & non-porous structure like Settlement, Road etc.
Plantation or Forest	The light vegetation consists of areas where crops are grown and some small trees, groves and some orchards.
Water Bodies	There are rivers, ponds, water-logged areas etc.

Source: Compiled by authors

Accuracy assessment

Assessing the accuracy of data holds paramount significance in gauging the reliability of information, especially within the realm of image classification and the detection of land use and land cover transitions(Andualem et al., 2018; Cheruto et al., 2016; Das & Sarkar, 2019). Given the inherent limitations in satellite imagery resolution, precision is inherently constrained. To ensure rigorous evaluation of the classified images, ERDAS IMAGINE 2014 conducted a comprehensive assessment. For each land use map, a collection of 100 random reference points was generated, serving as the basis for determining both the overall accuracy and the kappa coefficient. These reference points encompass processed satellite imagery, local insights, and Google Earth maps. The evaluation of land use and land cover classification accuracy was undertaken using two widely adopted methodologies: the overall accuracy and the Kappa accuracy. The overall accuracy pertains to the ratio of accurately identified pixels to the total count of reference points employed for validation(John et al., 2019). Meanwhile, the Kappa coefficient emerges as a widely trusted tool for assessing accuracy (Bose & Chowdhury, 2020). The calculation of expected frequency (ef) involves the sum of diagonal frequency (a), the total data (N), and is expressed through the formula: (Row Total × Column Total) / N. Notably, ERDAS IMAGINE software streamlines this process by automatically computing the expected frequency.

Post-Classification Change detection

Following the creation of Land Use and Land Cover (LULC) maps, the classified images were subjected to the 3 × 3 majority filter in order to remove isolated pixels (Duraisamy et al., 2018); the next phase involved employing ArcGIS 10.5 to execute spatial vector analysis. This encompassed converting the raster LULC maps into vector formats, thereby facilitating the extraction of changes. In this transformation process, a series of geoprocessing operations were executed, including dissolve, union, among others. To delve deeper into the dynamic shifts in land use and land cover, a pivotal tool emerged—the transition matrix. This matrix was instrumental in unveiling both the extent and direction of LULC alterations. Widely embraced, the change detection matrix played a central role in comprehending land encroachments across various land categories. Notably recognized as the "from-to" change detection, this approach enabled a comprehensive grasp of the transformation dynamics(Addae & Oppelt, 2019; Aneesha Satya et al., 2020; Das & Angadi, 2022).

Geo-simulation Modelling

This study employs the Future Land Use Simulation (FLUS) model to anticipate forthcoming scenarios. This model integrates an Artificial Neural Network (ANN) with a Cellular Automata (CA) model, incorporating a self-adaptive inertia competition mechanism(Liang et al., 2020; Liang et al., 2018; Liu et al., 2017). The process of generating a future potential Land Use and Land Cover (LULC) map follows several crucial stages. Initially, the Markov chain technique is employed to construct a prospective land change matrix and quantify the projected LULC. The Markov chain model, a well-established methodology, finds common use in analyzing LULC change(Bose & Chowdhury, 2020; Wang et al., 2021). This model provides a realistic approach to simulating the intricate nature of land dynamics. By randomly selecting variables and simulating probabilities and transition regions from two-time series, the Markov model predicts forthcoming LULC changes based on spatial, stochastic, and transition-based dynamics(Adedeji et al., 2015; Bose & Chowdhury, 2020; Gharaibeh et al., 2020; Wang et al., 2020). The Markov mechanism, characterized by predicting a system's future state solely from its immediate past state, underpins the study's explanation of LULC shifts over time. To enhance the spatial aspect of the Markov model, cellular automata (CA) is integrated, allowing for the probability of spatial changes in specific regions over time(Das & Sarkar, 2019; Hua, 2017; Liping et al., 2018). The study leverages land use/cover maps from two time periods (t1 and t2) to evaluate the historical distributional difference of land use/cover groups. The Markov chain, accounting for historical transition potential, quantifies the groups that may transition from t2 to the predicted time period t3 (Halmy et al., 2015). Transition probability matrices (Table 4) represent the likelihood of land use/cover classes transitioning to others(John et al., 2019), with the transformation probability matrix obtained using interpolation if t3 is not an even multiple between t1 and t2. The next step involves employing the FLUS model's Multiple Cellular Automata Allocation method, simulating future spatial patterns based on projected land use needs. The CA model components include cell, cell space, neighbor, law, and time. The CA model's filter calculates neighbor interactions, with the weight factor increasing as the distance between cells decreases. The weight factor interacts with transformation probabilities to predict neighboring grid cell states, ensuring non-random land-use change (Liping et al., 2018). The FLUS model, a multi-type land-use scenario modelling tool, amalgamates human and natural influences. This model incorporates both a top-down System Dynamics (SD) model and a bottom-up Cellular Automata (CA) model. The SD model predicts land-use scenario demands driven by socio-economic and environmental factors. To capture the complex interplay among land-use categories, the CA model incorporates a self-adaptive inertia and competition mechanism.

Table 4

Conversion Probability Matrix
Types of LULC	Agriculture	Water Bodies	Plantation or Forest	Built-up
Agriculture	0.874	0.022	0.067	0.037
Water Bodies	0.233	0.721	0.032	0.013
Plantation or Forest	0.276	0.012	0.605	0.107
Built-up	0.288	0.007	0.095	0.610

Source: Computed by authors

Table 5

Selection of Drivers for simulation modelling
Authors (Year)	Road Proximity	Service (Commercial) centre Proximity	Slope	Built-up Proximity
Lukas, P et al. (2023)	•		•	•
Kamaraj M.et al. (2022)	•		•	•
Mitra, S. et al. (2022)	•		•	•
Abbas, Z et al. (2021)	•		•	•
Dey, N. N. et al. (2021)	•	•	•	•
Gharaibeh, A. et al. (2020)	•	•	•
Novin, M. S. et al. (2019)	•		•	•
Al-Darwish Y. A. et al. (2018)	•	•		•
Li X. et al. (2012)	•	•		•
Liu X. et al. (2010)	•	•		•
Source: Compiled by authors

The use of Artificial Neural Networks (ANNs) estimates the probability of occurrence using driving factors like slope, proximity to roads, existing settlements, and service centres (Fig. 3) chosen factors are determined through a review of existing literature (Table 5). The FLUS model's SD component enhances its long-term simulation capabilities by connecting top-down demand with bottom-up CA modelling. Ultimately, the FLUS model provides future Land Use and Land Cover Change (LUCC) scenarios, uniting human and natural dynamics. It combines a top-down SD model with a bottom-up CA model, demonstrating its utility across various domains. The study draws on the work of Liu et al. (2017) and provides access to the FLUS software and its applications (https://www.geosimulation.cn/FLUS.html).

Geo- simulation model validation

The FLUS model was executed to project future land use and land cover scenarios, following the classification and accuracy assessment of all imagery data. Subsequently, utilizing the classified images from 2001 and 2011, along with driving variables, a prediction for 2021 was generated. This projection was subjected to a chi-square test, comparing it against the classified 2021 image. Upon successful model validation, recalibration was performed to forecast potential land use and land cover scenarios for 2031. It's noteworthy that the chi-square test has been employed by Das et al. 2019 and Bose et al. 2020 as well, underscoring its relevance in validating predictive land-use models. Moving forward, this study places a significant emphasis on model validation. To validate the accuracy of the projected land use and land cover for 2021, Pearson's Chi-square statistic was employed to compare the actual 2021 map with the projected 2021 map. This comparison involves assessing the simulated outcomes against the observed outcomes for 2021. The Chi-square (X²⁾ test is computed using the formula X² = ∑(O-E)2/E, where O is the predicted value and E is the actual value of the class. Additionally, the model validation includes the utilization of the kappa index, which measures the agreement between the actual and predicted raster classifications for LULC. Kappa-overall and Kappa-location were employed to assess the overall and location-specific performance of the simulation, respectively. Kappa statistics range from 0% (no agreement) to 100% (perfect agreement)(Aneesha Satya et al., 2020).

LULC Classification

The dataset on land-use change, presented across the years 1991, 2001, 2011, and 2021, provides a comprehensive portrayal (Fig. 4) of the dynamic evolution within various Land Use and Land Cover (LULC) categories. The dataset encapsulates (Table 6) the extent of each LULC class in terms of both area coverage and percentage representation, alongside their corresponding relative changes, captures the dynamic flux of the study area's landscape over time. A closer examination of the data underscores several significant trends. The realm of agriculture, constituting 81.98% (264.88 km²⁾ of the area in 1991, has witnessed a discernible reduction in coverage, culminating at 72.36% (233.79 km²) by 2021. The domain of water bodies has exhibited subtle fluctuations. Beginning at 4.32% (13.96 km²) in 1991, it has marginally expanded to 5.11% (16.51 km²) in 2021.Remarkably, the expanse of plantation or forest areas has exhibited a remarkable expansion. Commencing at 8.45% (27.29 km²) in 1991, it has burgeoned to 13.23% (42.75 km²) in 2021. Concurrently, the built-up has undergone noticeable transformation. Starting at 5.25% (16.96 km²) in 1991, it has burgeoned to 9.30% (30.04 km²) in 2021. The available census data made it evident that the population of cities is growing more quickly (Fig. 5). Another fascinating finding was that the data showed people switching (Fig. 6) from agricultural work to other occupations, as evidenced by the steadily rising numbers of household workers and other workers.

Figure 4: Land Use/Cover map of Berhampore (1991, 2001, 2011 and 2021)

Table 6

Quantification and Magnitude of change for each LULC Class
LULC Classes	1991		2001		2011		2021		Relative Change (1991–2021)
LULC Classes	km	%	km	%	km	%	km	%	km	%
Agriculture	264.88	81.98	259.44	80.30	241.60	74.78	233.79	72.36	-31.09	-9.62
Water Bodies	13.96	4.32	14.08	4.36	14.78	4.57	16.51	5.11	2.55	0.79
Plantation or Forest	27.29	8.45	28.91	8.95	39.00	12.07	42.75	13.23	15.46	4.79
Built-Up	16.96	5.25	20.66	6.39	27.71	8.58	30.04	9.30	13.08	4.05

Source: Computed by authors

Accuracy assessment of classified images

The classified image's accuracy assessment reveals varying Kappa Accuracy (K) values and overall accuracy percentages over different years. In 1991, a Kappa Accuracy of 0.89 corresponded to an overall accuracy of 98%. Similarly, in 2001, a Kappa Accuracy of 0.9 aligned with a 98% overall accuracy. In 2011, a Kappa Accuracy of 0.89 yielded a 96% overall accuracy, and in 2021, a Kappa Accuracy of 0.86 resulted in a 92% overall accuracy. These accuracy evaluations show how well the classification algorithm classified the image pixels' land cover accurately. Greater overall accuracy values and Kappa values indicate that the classified results and the ground truth data are more in agreement. The accuracy assessment results are shown in a Table 7.

Table 7

Accuracy assessment of classified images
Year	Overall Accuracy (%)	Kappa Accuracy (K)
1991	98	0.89
2001	98	0.9
2011	96	0.89
2021	92	0.86
* K value near 1 indicates that the referenced image and the classified image are more accurate

Land use/cover transition

The findings from the analysis of the change transition matrix (Table 8) revealed significant land-use transformations within the study period. Notably, around 15.92 km²of land transitioned from agricultural use to built-up areas, 25.96 km² shifted to plantation/forest, and 5.54 km² became water bodies. Conversely, approximately 3.68 km² of water bodies transformed into agricultural land. The analysis further indicated that a substantial portion of plantation or forest land (8.78 km²) experienced conversion to agricultural land, while 4.31 km² transitioned to built-up areas. This observation (Fig. 7) underscores a trend where a considerable portion of converted land, particularly in the context of plantation or forest areas, shifted towards both agricultural and built-up categories.

Table 8

Transition Matrix of LULC (1991–2021)
Land-Use/Land Cover		Year 2021 (km²)
Land-Use/Land Cover		Agriculture	Water Bodies	Plantation or Forest	Built-up	Total
Year 1991 (km²)	Agriculture	217.46	5.54	25.96	15.92	264.88
	Water Bodies	3.68	9.17	0.92	0.19	13.96
	Plantation or Forest	8.78	1.64	12.56	4.31	27.29
	Built-up	3.87	0.16	3.31	9.62	16.96
	Total	233.79	16.51	42.75	30.04	323.09

Source: Computed by authors

FLUS simulation modelling result

The outcome derived from the simulated map of 2031(Fig. 8) provides significant insights into the evolving land use patterns. It is evident that the extent of agricultural land covers an area of 226.21 km². In contrast, the categories of Water Bodies, Plantation or Forest, and Built-Up encompass areas of 14.87 km², 47.14 km², and 34.84 km² respectively. This pattern of change closely mirrors (Fig. 9) the trends observed in the earlier phases of land use transition, with only a marginal decline in the water body category. Notably, the majority of the converted agricultural land has transitioned into plantation cover (Table 9), accounting for 18.76 km², and built-up areas, spanning 9.92 km², with relatively minor variations. An intriguing observation arises from the conversion of water bodies into agricultural land, encompassing an area of 4.71 km². This phenomenon reflects the ongoing dynamics of repurposing croplands into built-up and plantation regions, as well as the transformation of wetlands into agricultural fields, as visually depicted in Fig. 10.

Table 9

Transition Matrix of LULC (2021–2031)
LULC Classes		Year 2031(km²)
LULC Classes		Agriculture	Water Bodies	Plantation or Forest	Built-up	Total
Year 2021 (km²)	Agriculture	201.4	3.71	18.76	9.92	233.79
	Water Bodies	4.71	10.66	0.7	0.44	16.51
	Plantation or Forest	15.92	0.41	22.01	4.41	42.75
	Built-up	4.18	0.09	5.67	20.1	30.04
	Total	226.21	14.87	47.14	34.87	323.09

Source: Computed by authors

Validation of FLUS model

The most crucial aspect of simulation modelling is validation. It was observed that the chi square (X²) value (0.29) indicates that there was no appreciable difference between the predicted and actual values (Table 10). The obtained X² value of 0.29 is below the significance level of X² = 0.05(df = 3), which is 7.815. Therefore, it can be said that the Markov model for predicting LULC changes in the study area is trustworthy. Additionally, Kappa index (Table 11) validated module exhibited robust performance, as evident from Kappa values: Kappa (histogram) at 0.978, Kappa (overall) at 0.869, and Kappa (location) at 0.887. The percentage of correctness was calculated at 93.57%, indicating a strong alignment between the projected and actual LULC situations in 2021 (Fig. 11) for Berhampore. This further reinforces the model's reliability and underscores the consistency achieved between projected and observed outcomes (Fig. 12).

Table 10

Validation of simulation result by chi-square test
LULC Classes	Simulated (O)	Actual(E)	(O-E)	(O-E⁾²	(O-E)²/E
Agriculture	233.51	233.79	-0.28	0.08	0.00
Water Bodies	14.72	16.51	-1.79	3.20	0.19
Plantation or Forest	43.18	42.75	0.43	0.18	0.00
Built-up area	31.68	30.04	1.64	2.69	0.09

Source: Computed by authors

Table 11

Kappa index for model validation (simulated 2021 with reference 2021)
Kappa Index	Value
% of Correctness	93.571
Kappa(overall)	0.869
Kappa(histogram)	0.979
Kappa(location)	0.888

Source: Computed by authors

Perception of LULC change in local community

Based on the findings of the focus group discussion, it is apparent that the expansion of the built-up area is occurring at an accelerated pace. The conversion of agricultural land and plantation or forest land for the purpose of constructing new buildings, shops, offices, and markets is currently taking place. In built-up areas, old houses are predominantly being demolished to make way for the construction of new apartments/multi-stored buildings. Moreover, the enlargement of structures often involves the clearance of backyard gardens. Berhampore, functioning as a block headquarters, offers a diverse range of goods and services, including markets, schools, colleges, hospitals, courts, government offices, theatres, and shopping centres, attracting people from the nearby regions. Consequently, the mounting population pressure is leading to the transformation of agricultural land, as well as plantation or forest land, into built-up areas. A significant portion of agricultural land situated near roads is undergoing conversion to accommodate shops, markets, offices, and residential houses. This shift is prompted by the unprofitability of agriculture, along with the strenuous and labour-intensive nature of farming activities.

Figure11: Land use/Cover Transition from 2021 to 2031

Consequently, individuals are turning towards fields that offer higher wages than agriculture. As a result, a considerable proportion of agricultural land has been transformed into plantation or forest areas, as well as built-up structures. The transition from agriculture to plantation farming is primarily motivated by factors such as reduced maintenance costs, lower labour intensity, minimal plant care requirements, and the potential for significant earnings at specific intervals. Water bodies surrounding settled areas are gradually being filled in to make room for new constructions. The encroachment upon many water bodies is driven by the increasing population pressure, as individuals aim to create residential apartments. Furthermore, there is a growing trend of reclaiming low-lying and marshy areas adjacent to agricultural fields for agricultural purposes, primarily due to the frequent absence of water in these water bodies. While a few new water bodies have been artificially created, some reclaimed ponds or water bodies have been filled with mud, debris, and water hyacinths. The Wetlands Protection Committee of Murshidabad districts is actively working to counteract the unauthorized filling of water bodies and the reclamation of deteriorating ponds. The local people now grapple with issues like extensive traffic congestion, insufficient parking provisions, an underdeveloped sewerage system, and waterlogging concerns. The absence of proper waste disposal facilities has led to public dumping, resulting in unpleasant odours and unsightly piles of garbage throughout the area, thereby contributing to an unfavourable living environment. The study area is also grappling with water scarcity, further exacerbating the situation. The surge in mosquito and fly populations due to these conditions has resulted in the proliferation of diseases like dengue and malaria.

Social and policy making relevance of the present study

After analysing the fact, it is believed that the present study has following social and policing making relevance:

Urban Planning and Development: Land use changes in urban areas can significantly affect the quality of life for residents. Understanding and predicting these changes can inform urban planning decisions, such as infrastructure development, housing, transportation, and green spaces, which directly impact the daily lives of people living in Berhampore.

Environmental Conservation: Changes in land use and land cover can have environmental consequences, including deforestation, habitat loss, and changes in local climate patterns. The paper's findings may be relevant to efforts aimed at conserving natural resources and biodiversity in the region.

Agricultural Practices: In regions like Berhampore, where agriculture is a significant part of the economy, land use changes can impact agricultural practices and food security. Farmers and policymakers can benefit from insights into how land use changes may affect crop patterns and agricultural sustainability.

Disaster Management: Land use changes can influence vulnerability to natural disasters like floods and landslides. The paper's predictions may be useful for disaster management and preparedness efforts, helping to reduce the social and economic impacts of such events.

Socio-economic Impacts: Land use changes can have socio-economic implications, including changes in property values, access to services, and livelihoods. Understanding these changes can help policymakers and communities adapt to evolving circumstances.

Policy and Governance: The findings of the paper may inform land use policies and regulations in Berhampore and similar regions. Better land use policies can contribute to sustainable development and improved living conditions.

Public Awareness and Engagement: Research on land use change can raise awareness among the general public about the importance of responsible land use and its consequences. Informed citizens are more likely to engage in discussions and actions that promote sustainable land management.

Education and Research: The paper can serve as a valuable resource for students, researchers, and academics studying land use change, modeling techniques, and environmental science in the context of Berhampore and the broader West Bengal region.

In conclusion, the 1991–2021 land-use dataset shows the dynamic evolution of various Land Use and Land Cover (LULC) categories. This dataset illustrates the study area's dynamic flux over time by showing each LULC class's area coverage, percentage representation, and relative changes. Closer inspection of the dataset revealed several significant trends. Agriculture covered 81.98% or 264.88 km² in 1991 but only 72.36% or 233.79 km² by 2021. Water bodies expanded slightly from 4.32% or 13.96 km² in 1991 to 5.11% or 16.51 km² in 2021. The plantation or forest areas have grown significantly. It grew from 8.45% or 27.29 km² in 1991 to 13.23% or 42.75 in 2021. Similarly, the built-up area increased from 5.25% or 16.96 km² in 1991 to 9.30% or 30.04 km² in 2021. The change transition matrix revealed major changes during the study. Notably, 15.92 km² of land went from agriculture to built-up areas, 25.96 km² to plantations or forests, and 5.54 km² to water bodies. Conversely, 3.68 km² of water bodies became agricultural land. The analysis showed that 8.78 km² of plantation or forest land was converted to agricultural land, while 4.31 km² was changed to built-up areas. This shows that a lot of converted land, especially plantation or forest land, became agricultural and built-up. The 2031 simulated map revealed changing land use patterns. The dataset shows a 226.21 km² area of agricultural land. However, the categories of Water Bodies, Plantation or Forest, and Built-Up cover 14.87 km², 47.14 km², and 34.84 km², respectively. This pattern of change closely mirrors earlier land use transition trends, with only a marginal decline in water bodies. Most of the converted agricultural land has become plantation cover (18.76 km²) and built-up areas (9.92 km²), with minor variations. An intriguing observation is the conversion of 4.71 km² of water bodies into agricultural land. This reflects the ongoing process of repurposing croplands for development and plantation. The focus group discussion also showed that the built-up area is growing rapidly. Agricultural and plantation or forest land is being converted for new buildings, shops, offices, and markets. Old houses in built-up areas are mostly demolished to build apartments. Structure expansion often clears backyard gardens. Berhampore, a block headquarters, attracts people from nearby regions with its markets, schools, colleges, hospitals, courts, government offices, theatres, and shopping centres. Thus, population pressure is turning agricultural, plantation, and forest land into built-up areas. A lot of agricultural land near roads is being turned into shops, markets, offices, and homes. This shift is caused by agriculture's unprofitability and laborious nature. Therefore, people are moving to higher-paying fields than agriculture. Thus, much agricultural land has been converted into plantations, forests, and buildings. The switch from agriculture to plantation farming is driven by lower maintenance costs, labour intensity, plant care requirements, and the potential for high earnings at certain times. Waterways around settled areas are being filled in for new constructions. Population pressure to build residential apartments is encroaching on many water bodies. Due to the lack of water availability in low-lying and marshy areas near agricultural fields, they are being reclaimed for agricultural use. Some reclaimed ponds and water bodies are filled with mud, debris, and water hyacinths, while others water bodies are artificially created. Wetlands Protection Committee of Murshidabad districts is fighting illegal filling and reclamation of deteriorating ponds. The locals face traffic congestion, parking shortages, a poor sewerage system, and waterlogging. Public dumping due to a lack of waste disposal facilities has caused foul odours and unsightly garbage piles, making the area unlivable. The study area also has water shortages, worsening the situation. The rise in mosquito and fly populations due to these conditions has spread dengue and malaria. The results obtained from the study highlight the complex and dynamic relationship between land use, urbanization, and environmental dynamics. They emphasize the urgency of sustainable city planning, resource management, and community engagement to navigate these transformations while safeguarding both the well-being of the population and the ecological balance of the area.

Acknowledgement

The second author of this study is thankful to the University Grant Commission (UGC), New Delhi, India for providing the Senior Research Fellowship (SRF) during this research work. The authors are highly indebted to Aliah University, Kolkata for providing research environment. The authors also grateful to the USGS for making the Landsat data freely accessible. The authors are highly thankful to the anonymous reviewer(s) for their scholarly comments and suggestions which help to improvement the manuscript.

We know of no conflicts of interest associated with this publication, and there has been no significant financial support for this work that could have influenced its outcome. As Corresponding Author, I confirm that the manuscript has been read and approved for submission by all the authors.

Abbas, Z., Yang, G., Zhong, Y., & Zhao, Y. (2021). Spatiotemporal change analysis and future scenario of LULC using the CA-ANN approach: A case study of the greater bay area, china. Land, 10(6), 584.
Abebe, G., Getachew, D., & Ewunetu, A. (2022). Analysing land use/land cover changes and its dynamics using remote sensing and GIS in Gubalafito district, Northeastern Ethiopia. SN Applied Sciences, 4(1), 30.
Addae, B., & Oppelt, N. (2019). Land-use/land-cover change analysis and urban growth modelling in the Greater Accra Metropolitan Area (GAMA), Ghana. Urban Science, 3(1), 26.
Adedeji, O. H., Tope-Ajayi, O. O., & Abegunde, O. L. (2015). Assessing and predicting changes in the status of Gambari forest reserve, Nigeria using remote sensing and GIS techniques. Journal of Geographic information system, 7(03), 301.
Al-Darwish, Y., Ayad, H., Taha, D., & Saadallah, D. (2018). Predicting the future urban growth and it's impacts on the surrounding environment using urban simulation models: Case study of Ibb city–Yemen. Alexandria engineering journal, 57(4), 2887-2895.
Alam, A., Bhat, M. S., & Maheen, M. (2020). Using Landsat satellite data for assessing the land use and land cover change in Kashmir valley. GeoJournal, 85, 1529-1543.
Alam, N., Saha, S., Gupta, S., & Chakraborty, S. (2021). Prediction modelling of riverine landscape dynamics in the context of sustainable management of floodplain: a Geospatial approach. Annals of GIS, 27(3), 299-314.
Amin, A., & Fazal, S. (2012). Land transformation analysis using remote sensing and GIS techniques (a case study).
Andualem, T. G., Belay, G., & Guadie, A. (2018). Land use change detection using remote sensing technology. Journal of Earth Science & Climatic Change, 9(10), 6.
Aneesha Satya, B., Shashi, M., & Deva, P. (2020). Future land use land cover scenario simulation using open source GIS for the city of Warangal, Telangana, India. Applied Geomatics, 12, 281-290.
Bhalli, M., & Abdul, G. (2015). Use of geospatial techniques in monitoring urban expansion and land use change analysis: a case of Lahore, Pakistan. Journal of Basic & Applied Sciences, 11, 265-273.
Bose, A., & Chowdhury, I. R. (2020). Monitoring and modeling of spatio-temporal urban expansion and land-use/land-cover change using markov chain model: a case study in Siliguri Metropolitan area, West Bengal, India. Modeling Earth Systems and Environment, 6, 2235-2249.
Bufebo, B., & Elias, E. (2021). Land use/land cover change and its driving forces in Shenkolla watershed, south Central Ethiopia. The Scientific World Journal, 2021.
Chavez, P. S. (1996). Image-based atmospheric corrections-revisited and improved. Photogrammetric engineering and remote sensing, 62(9), 1025-1035.
Cheruto, M. C., Kauti, M. K., Kisangau, D. P., & Kariuki, P. C. (2016). Assessment of land use and land cover change using GIS and remote sensing techniques: a case study of Makueni County, Kenya.
Choudhury, D., Das, K., & Das, A. (2019). Assessment of land use land cover changes and its impact on variations of land surface temperature in Asansol-Durgapur Development Region. The Egyptian Journal of Remote Sensing and Space Science, 22(2), 203-218.
Chowdhury, M., Hasan, M. E., & Abdullah-Al-Mamun, M. (2020). Land use/land cover change assessment of Halda watershed using remote sensing and GIS. The Egyptian Journal of Remote Sensing and Space Science, 23(1), 63-75.
Das, S., & Angadi, D. P. (2022). Land use land cover change detection and monitoring of urban growth using remote sensing and GIS techniques: a micro-level study. GeoJournal, 87(3), 2101-2123.
Das, S., & Sarkar, R. (2019). Predicting the land use and land cover change using Markov model: A catchment level analysis of the Bhagirathi-Hugli River. Spatial Information Research, 27, 439-452.
Dey, N. N., Al Rakib, A., Kafy, A.-A., & Raikwar, V. (2021). Geospatial modelling of changes in land use/land cover dynamics using Multi-layer Perceptron Markov chain model in Rajshahi City, Bangladesh. Environmental Challenges, 4, 100148.
Duraisamy, V., Bendapudi, R., & Jadhav, A. (2018). Identifying hotspots in land use land cover change and the drivers in a semi-arid region of India. Environmental monitoring and assessment, 190(9), 535.
Fasona, M. J., Akintuyi, A. O., Adeonipekun, P. A., Akoso, T. M., Udofia, S. K., Agboola, O. O., Ogunsanwo, G. E., Ariori, A. N., Omojola, A. S., & Soneye, A. S. (2022). Recent trends in land-use and cover change and deforestation in south–west Nigeria. GeoJournal, 87(3), 1411-1437.
Garai, D., & Narayana, A. (2018). Land use/land cover changes in the mining area of Godavari coal fields of southern India. The Egyptian Journal of Remote Sensing and Space Science, 21(3), 375-381.
Gharaibeh, A., Shaamala, A., Obeidat, R., & Al-Kofahi, S. (2020). Improving land-use change modeling by integrating ANN with Cellular Automata-Markov Chain model. Heliyon, 6(9).
Halmy, M. W. A., Gessler, P. E., Hicke, J. A., & Salem, B. B. (2015). Land use/land cover change detection and prediction in the north-western coastal desert of Egypt using Markov-CA. Applied Geography, 63, 101-112.
Hamad, R., Balzter, H., & Kolo, K. (2018). Predicting land use/land cover changes using a CA-Markov model under two different scenarios. Sustainability, 10(10), 3421.
Hua, A. (2017). Application of Ca-Markov model and land use/land cover changes in Malacca River Watershed, Malaysia. Applied Ecology and Environmental Research, 15(4), 605-622.
John, J., Chithra, N., & Thampi, S. G. (2019). Prediction of land use/cover change in the Bharathapuzha river basin, India using geospatial techniques. Environmental monitoring and assessment, 191, 1-15.
Kaliraj, S., Chandrasekar, N., Ramachandran, K., Srinivas, Y., & Saravanan, S. (2017). Coastal landuse and land cover change and transformations of Kanyakumari coast, India using remote sensing and GIS. The Egyptian Journal of Remote Sensing and Space Science, 20(2), 169-185.
Kamaraj, M., & Rangarajan, S. (2022). Predicting the future land use and land cover changes for Bhavani basin, Tamil Nadu, India, using QGIS MOLUSCE plugin. Environmental Science and Pollution Research, 29(57), 86337-86348.
Kangabam, R. D., Selvaraj, M., & Govindaraju, M. (2019). Assessment of land use land cover changes in Loktak Lake in Indo-Burma Biodiversity Hotspot using geospatial techniques. The Egyptian Journal of Remote Sensing and Space Science, 22(2), 137-143.
Li, X., Zhang, Y., Liu, X., & Chen, Y. (2012). Assimilating process context information of cellular automata into change detection for monitoring land use changes. International Journal of Geographical Information Science, 26(9), 1667-1687.
Liang, X., Liu, X., Chen, G., Leng, J., Wen, Y., & Chen, G. (2020). Coupling fuzzy clustering and cellular automata based on local maxima of development potential to model urban emergence and expansion in economic development zones. International Journal of Geographical Information Science, 34(10), 1930-1952.
Liang, X., Liu, X., Li, D., Zhao, H., & Chen, G. (2018). Urban growth simulation by incorporating planning policies into a CA-based future land-use simulation model. International Journal of Geographical Information Science, 32(11), 2294-2316.
Liping, C., Yujun, S., & Saeed, S. (2018). 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(7), e0200493.
Liu, X., Li, X., Shi, X., Zhang, X., & Chen, Y. (2010). Simulating land-use dynamics under planning policies by integrating artificial immune systems with cellular automata. International Journal of Geographical Information Science, 24(5), 783-802.
Liu, X., Liang, X., Li, X., Xu, X., Ou, J., Chen, Y., Li, S., Wang, S., & Pei, F. (2017). A future land use simulation model (FLUS) for simulating multiple land use scenarios by coupling human and natural effects. Landscape and Urban Planning, 168, 94-116.
Lukas, P., Melesse, A. M., & Kenea, T. T. (2023). Prediction of future land use/land cover changes using a coupled CA-ANN model in the upper omo–gibe river basin, Ethiopia. Remote Sensing, 15(4), 1148.
Maity, B., Mallick, S. K., & Rudra, S. (2020). Spatiotemporal dynamics of urban landscape in Asansol municipal corporation, West Bengal, India: a geospatial analysis. GeoJournal, 1-19.
Mhawish, Y. M., & Saba, M. (2016). Impact of population growth on land use changes in Wadi Ziqlab of Jordan between 1952 and 2008. International Journal of Applied Sociology, 6(1), 7-14.
Mishra, P. K., Rai, A., & Rai, S. C. (2020). Land use and land cover change detection using geospatial techniques in the Sikkim Himalaya, India. The Egyptian Journal of Remote Sensing and Space Science, 23(2), 133-143.
Mitra, S., Roy, S., & Hore, S. (2023). Assessment and forecasting of the urban dynamics through lulc based mixed model: evidence from Agartala, India. GeoJournal, 88(2), 2399-2422.
Muthusamy, S., Arunkumar, R. X., Raj, N. T., Lakshumanan, C., & Jayaprakash, M. (2010). Land Use and Land Cover Changes Detection Using Multitemporal Satellite Data, Cuddalore Coastal Zone, Se-Coast of India. International Journal of Geomatics and Geosciences, 1(3), 610.
Naikoo, M. W., Rihan, M., & Ishtiaque, M. (2020). 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. Journal of Urban Management, 9(3), 347-359.
Novin, M. S., & Ebrahimipour, A. (2019). Spatio-temporal modelling of land use changes by means of ca–Markov model. Modeling Earth Systems and Environment, 5, 1253-1263.
Patel, L. K., & Tripathi, S. (2016). A Geospatial Approach to Analyze the Impact of Population Growth on Bundelkhand Landscape, Central India. International Journal, 5(6), 1755-1767.
Prakasam, C. (2010). Land use and land cover change detection through remote sensing approach: A case study of Kodaikanal taluk, Tamil nadu. International Journal of Geomatics and Geosciences, 1(2), 150.
Rahman, A., Kumar, S., Fazal, S., & Siddiqui, M. A. (2012). Assessment of land use/land cover change in the North-West District of Delhi using remote sensing and GIS techniques. Journal of the Indian Society of Remote Sensing, 40, 689-697.
Rawat, J. S., Biswas, V., & Kumar, M. (2013). Changes in land use/cover using geospatial techniques: A case study of Ramnagar town area, district Nainital, Uttarakhand, India. The Egyptian Journal of Remote Sensing and Space Science, 16(1), 111-117.
Roy, B., & Kasemi, N. (2021). Monitoring urban growth dynamics using remote sensing and GIS techniques of Raiganj Urban Agglomeration, India. The Egyptian Journal of Remote Sensing and Space Science, 24(2), 221-230.
Saber, A., El-Sayed, I., Rabah, M., & Selim, M. (2021). Evaluating change detection techniques using remote sensing data: Case study New Administrative Capital Egypt. The Egyptian Journal of Remote Sensing and Space Science, 24(3), 635-648.
Showqi, I., Rashid, I., & Romshoo, S. A. (2014). Land use land cover dynamics as a function of changing demography and hydrology. GeoJournal, 79, 297-307.
Wang, S. W., Gebru, B. M., Lamchin, M., Kayastha, R. B., & Lee, W.-K. (2020). Land use and land cover change detection and prediction in the Kathmandu district of Nepal using remote sensing and GIS. Sustainability, 12(9), 3925.
Wang, S. W., Munkhnasan, L., & Lee, W.-K. (2021). Land use and land cover change detection and prediction in Bhutan's high altitude city of Thimphu, using cellular automata and Markov chain. Environmental Challenges, 2, 100017.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Assessment of Land Use /Land Cover Changes and its Future Predictions Using CA-Markov with ANN Simulation for Berhampore, West Bengal, India

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

The study Area

Database

Source: USGS Earth Explorer

Source: Compiled by authors

Methodology

Image pre-processing

Source: Compiled by authors

Accuracy assessment

Post-Classification Change detection

Geo-simulation Modelling

Source: Computed by authors

Results and discussion

Source: Computed by authors

Accuracy assessment of classified images

Land use/cover transition

Source: Computed by authors

FLUS simulation modelling result

Source: Computed by authors

Validation of FLUS model

Source: Computed by authors

Source: Computed by authors

Perception of LULC change in local community

Social and policy making relevance of the present study

Conclusions

Declarations

References

Additional Declarations

Status:

Version 1