Study on the driving mechanism of cultivated land change in the urban–rural fringe with Bayesian network modeling

Jianping Wang; Zhenhong Zhu; Meiqiu Chen; Yiguo Zhang

doi:10.1038/s41598-026-35760-7

. 2026 Jan 17;16:5599. doi: 10.1038/s41598-026-35760-7

Study on the driving mechanism of cultivated land change in the urban–rural fringe with Bayesian network modeling

Jianping Wang ^1,², Zhenhong Zhu ^1,³, Meiqiu Chen ^1,^2,^✉, Yiguo Zhang ³

PMCID: PMC12891723 PMID: 41548018

Abstract

Urbanisation accentuates human-land conflicts in the urban–rural fringe and poses significant threats to the preservation of cultivated land. Understanding the characteristics and mechanisms of cultivated land change is essential for balancing development and conservation in these regions. Based on the essential characteristics of the urban–rural fringe, a multidimensional feature index identification model suitable for long-term definition of the urban–rural fringe was developed. This model was used to identify the urban–rural fringe of Nanchang City from 2000 to 2024. A change trajectory analysis was utilized to describe the spatialtemporal pattern evolution of cultivated land, while a Bayesian network model was employed to uncover the underlying driving mechanisms. The results indicate the following: (1) The model demonstrated favourable feasibility and efficiency in the long-term sequential identification of urban–rural fringe areas. It delineated the extent of the urban–rural fringe in Nanchang City over the period from 2000 to 2024, and subsequent validation confirmed that the identification results are highly consistent with the fundamental characteristics of the urban–rural fringe; (2) In the urban–rural fringe, the total area of farmland transferred out exceeds that transferred in. Farmland transferred out is primarily converted into construction land. The transfer of farmland outwards is concentrated around the city centre and exhibits a relatively intense trend of gradual outward expansion. In contrast, the transfer of farmland inwards is more scattered and limited in spatial distribution, with farmland fragmentation becoming increasingly apparent; (3) The results of the sensitivity analysis indicate that the primary factor influencing changes in cultivated land use in the urban–rural fringe area of Nanchang City is construction occupation, followed by ecological occupation, the protective effect of policies and planning, and the degree of socioeconomic impact. These findings align with the actual development patterns observed in the urban–rural fringe area. The research results can not only directly provide policy references for the coordinated development of the urban–rural fringe and the protection of cultivated land in Nanchang City, but also offer useful references for the protection of cultivated land in similar urban–rural fringe areas.

Keywords: Urban–rural fringe, Cultivated land changes, Driving mechanism, Bayesian network model

Subject terms: Environmental sciences, Environmental social sciences, Geography, Geography

Introduction

Cultivated land, as one of Earth’s most precious natural resources, is fundamental to human survival and development. It serves various essential functions, including the preservation of cultural heritage, ecological conservation, and food production. As a cornerstone for achieving sustainable development goals, the sustainable management of cultivated land is vital to ensuring global food security, ecological balance, and socioeconomic stability¹. The tension between protecting cultivated land and fostering economic growth is particularly pronounced in rapidly urbanising regions. In these areas, substantial amounts of valuable cultivated land have been taken over by urban development, resulting in a continuous reduction and fragmentation of cultivated land area. This trend not only jeopardizes food security but also undermines the ecosystem services supplied by cultivated land^2,3. Against the backdrop of rapid urbanisation, a distinctive regional entity—the urban–rural fringe—has gradually emerged. It serves as a transitional zone between urban and rural areas, characterised by the rapid exchange of production factors and the rich integration of urban and rural features. This area is crucial for achieving integrated urban–rural development in the new era, yet it has also become a region where human-land conflicts are most concentrated⁴. Consequently, a series of problems have arisen, including land-use conflicts, the spread of urbanisation functions, the rapid loss of high-quality farmland, and ecosystem degradation. These issues not only pose a significant threat to food security but also restrict the further development of the urban–rural fringe and, indeed, the city as a whole. Therefore, the urban–rural fringe has increasingly become a key focus for coordinating the relationship between urban development and farmland protection.

Identifying the spatial extent of the urban–rural fringe is fundamental to its management and control. Due to its transitional, dynamic, and heterogeneous characteristics, the urban–rural fringe has long been regarded as a “Gray Box” within the urban–rural structure, rendering the precise definition of its spatial boundaries extremely challenging^5,6. Early research primarily relied on qualitative and empirical assessments. For example, Krueger et al.suggested that the urban–rural fringe lies within a radius of 50–150 km from the edge of urbanised areas⁷. Subsequently, scholars increasingly adopted quantitative research methods. Bryant attempted to delineate the urban–rural fringe using the ratio of the urban (non-agricultural) population to the agricultural population⁸, while Desai proposed a demarcation method based on a composite index derived from the aggregation index and the suburbanisation index⁹. With advances in remote sensing technology, researchers have accessed remote sensing image data—including night-time light data, population, and economic data—and combined these with catastrophe theory, threshold methods, and clustering models to define the urban–rural fringe more precisely and intuitively^10–12. However, some practical challenges persist. For instance, early night light data were constrained by detector limitations, with light values ranging only from 0 to 63, resulting in the loss of many saturated light areas and hindering the detailed representation of urban interiors¹³. Due to difficulties in obtaining humanistic spatial data, most current research relies on natural elements observable via remote sensing images to identify the urban–rural fringe, often employing relatively simplistic indicators that fail to capture its complexity. In addition, the temporal scope of most studies is generally short (1–10 years)^14,15, limiting their ability to fully characterise the urban–rural fringe at different stages of the urbanisation process.

The change in cultivated land use is a significant aspect of land use change. Research in this area facilitates the optimisation of industrial structures and promotes the coordinated development of urban and rural regions. Since the 1980s, scholars worldwide have conducted extensive studies on change in cultivated land, primarily focusing on depicting spatiotemporal differentiation characteristics¹⁶, identifying influencing factors¹⁷, and proposing strategies for spatial pattern optimisation¹⁸, all of which demonstrate strong practical relevance. Regarding research scales, existing studies encompass global¹⁹, national²⁰, provincial²¹, urban agglomeration, and municipal levels²². There are also investigations into specialised regions such as mountainous areas²³ and river basins²⁴; however, few have concentrated on the urban–rural fringe. To measure spatiotemporal differentiation characteristics, the academic community generally employs quantitative and spatially explicit methods, such as spatial autocorrelation²⁵ and gravity center analysis²⁶. Nevertheless, the change trajectory analysis method—which can clearly characterise morphological changes—is less used. In studies examining the influencing factors of cultivated land change, scholars have explored the impacts of natural endowments, policy effects, and socioeconomic development. However, certain deficiencies remain: ① In factor selection, economic and policy data are more difficult to obtain and spatially quantify than natural data, necessitating further refinement; ② In research methods, principal component analysis²⁷, geographical detector models²⁸, and FLUS models²⁹ are commonly used. While these methods effectively reveal statistical correlations between variables,they lack the capacity to analyse spatially explicit characteristics such as spatial distance, boundary effects, and driving mechanisms.Therefore, combining spatial optimisation methods or extended models is essential to enhance spatial expression capabilities³⁰. The Bayesian network model integrates graph theory and probability theory. Compared with traditional statistical models, it not only intuitively presents the complex relationships and dynamic evolution processes between cultivated land use change and its driving factors in a graphical form but also possesses strong reasoning and diagnostic capabilities. Moreover, it can analyse and explain implicit spatial laws by incorporating geographical elements such as ‘spatial distance’³¹.

In the context of rigorously implementing the national fundamental policy on cultivated land protection and actively promoting integrated urban–rural development, this study attempts to construct a multi-dimensional characteristic index identification model suitable for long-term dynamic monitoring to delineate the urban–rural fringe of Nanchang. Furthermore, it employs change trajectory analysis to capture the process of cultivated land use change in Nanchang’s urban–rural fringe and systematically analyses its driving mechanisms using a Bayesian network model, ultimately providing targeted recommendations for cultivated land protection in this area. The innovations of this study are as follows: first, it focuses on cultivated land change in the urban–rural fringe—a region where human-land conflicts are concentrated; second, based on the essential characteristics of the urban–rural fringe, it develops a systematic and operable multi-dimensional characteristic index identification model; third, it applies the Bayesian network model to dissect the driving mechanisms of cultivated land change. This research not only directly provides policy references for the coordinated development and cultivated land protection of Nanchang’s urban–rural fringe but also offers valuable insights for cultivated land protection in similar urban–rural fringe areas.

Research framework

Centering on the research objectives, this study first acquired and constructed a comprehensive raster database encompassing land use, geographical conditions, socioeconomic factors, and policy planning data using the multiple data platforms and ArcGIS 10.8.Secondly, owing to the intrinsic characteristics of the urban–rural fringe, a multidimensional characteristic index identification model was developed to delineate the spatial extent of Nanchang’s urban–rural fringe and analyse its evolutionary features.Subsequently, a change trajectory analysis method was employed to elucidate the quantitative structural changes and the spatiotemporal distribution of cultivated land within this defined fringe. To identify the complex relationships between changes in cultivated land use and their driving factors, a Bayesian network model was utilised. This model selected indicators across four dimensions—natural environment, geographical location, policy planning, and socioeconomic factors—to summarize the driving mechanisms behind cultivated land outflow. Finally, integrating these findings with the practical context of the study area, recommendations for the protection of cultivated land within the urban–rural fringe were proposed to harmonise societal development with land conservation. The overarching technical workflow is illustrated in Fig. 1.

Materials and methods

Overview of the study area

The study area is shown in Fig. 2. Nanchang City, the capital of Jiangxi Province, is strategically located between longitudes 115°27′E and 116°35′E and latitudes 28°10′N and 29°11′N, encompassing six districts and three counties. It serves as a crucial city in the development of central China, alongside Wuhan and Changsha. Since 2000, Nanchang has experienced significant urbanisation. The city’s population has increased from 4.4342 million to 6.6704 million by 2024, while its GDP has risen from 43.51 billion yuan to 780.037 billion yuan. Nevertheless, this rapid urbanisation has led to substantial changes in the spatial layout, threatening the stability of arable land. According to data from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (RESDC), the area of cultivated land in Nanchang City decreased by 52,144.68 hectares during 2000–2024. As a pioneering zone for urbanisation, the urban–rural fringe of Nanchang is characterised by significant expansion of construction land, resulting in a more severe reduction of cultivated land. Therefore, studying the changes in cultivated land use and their driving mechanisms in this region is both representative and necessary.

Data sources and preliminary handling

This study has utilised several datasets spanning six periods (2000, 2005, 2010, 2015, 2020, and 2024) in Nanchang City, including land use data, Normalised Difference Vegetation Index (NDVI) data, Gross Domestic Product (GDP), Digital Elevation Model (DEM) data, nighttime light data, population figures, road vector data, urban development boundaries, and permanent basic farmland protection areas. The sources of the datasets are presented below:

The land use data encompass six time points: 2000, 2005, 2010, 2015, 2020, and 2024, with a spatial resolution of 30m × 30m. The dataset is provided by the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (RESDC) (http://www.resdc.cn). The first-level classification includes cultivated land, forest land, grassland, water area, residential land, and unused land. The interpretation accuracy exceeds 90%³², which satisfies the research needs. The 2024 data is based on the 2023 data,updated through field verification.
Both Normalized Difference Vegetation Index (NDVI) data and Digital Elevation Model (DEM) data were obtained via Google Earth Engine (GEE) (https://code.earthengine.google.com/), with both datasets having a resolution of 30 m.
The resolution of the nighttime light data is 500 m, and it was acquired from the National Earth System Science Data Center (http://www.geodata.cn/).
The population and GDP data at a 1 km resolution were sourced from the Resource and Environment Science and Data Center of the Chinese Academy of Sciences (www.resdc.cn).
The permanent basic farmland protection areas and urban development boundaries in Nanchang City (2023) have been determined using data from the third national land survey.
Road network data were obtained from OpenStreetMap (https://www.openstreetmap.org/).

Prior to formal analysis, the aforementioned datasets underwent preprocessing procedures, including mask extraction, followed by a unified projection transformation to the WGS_1984 _UTM_Zone_50N coordinate system. Furthermore, for data with spatial resolutions other than 30 m × 30 m, the Kriging interpolation method in ArcGIS 10.8 was employed for analysis, and all datasets were resampled to a uniform 30 m × 30 m spatial resolution.

Methodology

Multidimensional feature index identification model

Research unit determination.Taking administrative villages and communities as the analytical units to identify the urban–rural fringe, thereby enhancing the practical management value of the research¹⁵. The effective utilisation of land use/land cover (LULC) data and remote sensing imagery provides robust technical support and data reliability for the accurate identification and analysis of these units.Consequently, administrative villages and communities are selected as the research units in this study.
Selection of evaluation indicators.This study is based on the three core attributes of the urban–rural fringe widely recognized within the academic community^6,33 and develops a multidimensional identification and evaluation framework for the urban–rural fringe. Detailed indicators and their descriptions are provided in Table 1.In essence, the urban–rural fringe serves as a transitional zone where urban areas expand into rural territories, characterized by complex land use types and fragmented landscape patterns in terms of land utilization and landscape morphology³⁴. Although landscape heterogeneity encompasses multiple dimensions, such as form, configuration, and composition, in the urban–rural fringe area the degree of landscape fragmentation is often regarded as the primary spatial indicator for measuring the intensity of human disturbance and the degradation of habitat quality³⁵. Accordingly, to directly capture this characteristic and mitigate potential information redundancy and collinearity associated with multidimensional indicators, this study employs the landscape fragmentation index and the largest patch index to quantify spatial heterogeneity. These two indices quantitatively characterize the dispersion and dominance of the regional spatial structure from global and local perspectives, respectively.The urban–rural fringe, located at the interface between urban and rural areas¹⁵, exhibits distinct gradients in the spatial distribution of social, economic, and functional components. To capture its transitional nature, the nighttime light index and the proportion of construction land are utilized. Urban expansion leads to an increase in construction land and a decrease in vegetation coverage, indicating signifying changes in land use patterns. Accordingly, this study comprehensively employs the dynamic degree of land use and the normalized difference vegetation index to characterize its evolving features.

Table 1.

Multidimensional identification evaluation index system of urban–rural fringe.

Feature	Index	Explain	Property	Weight
Heterogeneity	Landscape fragmentation(x₁)	, N_i is the number of patches in landscape i, and A is the total area of the landscape	+	0.1883
Heterogeneity	Largest patch index(x₂)	, A_max is the area of the largest patch in the landscape, and A is the total area of the landscape	-	0.0248
Transition	Nighttime light index(x₃)	Calculate the results at the administrative village scale through the “Summary Statistics as Table” function in ArcGIS 10.8	+	0.1528
Transition	Proportion of construction land(x₄)	The proportion of construction land area to the total area	+	0.1603
Dynamic	Comprehensive dynamic degree of land use(x₅)	The meaning is as noted	+	0.4738
Dynamic	Dynamic degree of NDVI(x₆)	The NDVI change rate between different periods was determined using the Raster Calculator in ArcGIS 10.8. The analysis involved calculating the mean values at the administrative village/community scale for the years 2000 to 2024	-	0.0001

Open in a new tab

LC denotes the overall dynamic degree of land use within the research area. LU_i represents the initial area of the i-th land type at the commencement of the study. △LU_i-j indicates the absolute value of the net change in the i-th land type converted to the j-th land use type during the study period. T denotes the duration of the study.

(3)
Data standardization and weight assignment. The deviation standardization method was employed to normalize the data ,making it uniform and comparable by performing dimensionless processing on each index. To ensure objectivity in weight assignment, the entropy method was utilized to calculate the weights³⁶. The specific weight values are shown in Table 1.
(4)
Calculation of the characteristic index of the urban–rural fringe. The characteristic index of each evaluation unit is calculated using the a multi -factor comprehensive weighted model within the Map Algebra tool of ArcGIS 10.8, resulting in a spatial distribution raster representing the urban–rural fringe characteristic index³⁷.
(5)
Range delimitation based on the natural breaks method and limiting factor constraints. The natural breaks method is a classification approach based on the inherent statistical properties of data. Its core principle is to identify natural turning points within a data sequence to minimize variance within groups while maximizing variance between groups. In this study, the natural breaks tool embedded in ArcGIS 10.8 was used to categorize the calculated characteristic indices into three distinct types: urban areas, urban–rural fringes, and rural areas. To improve classification accuracy, the transitional nature of the urban–rural fringe was considered. Based on relevant literature³⁸, the nighttime light index and the proportion of construction land—both with thresholds set below 0.15—were used as limiting factors to refine the initial classification results. This adjustment aimed to eliminate isolated hotspots caused by landscape succession. Ultimately, the spatial boundaries of the urban–rural fringe in Nanchang City were delineated across six distinct periods.

Analysis method of change trajectory

The change trajectory analysis method examines the dynamic displacement and morphological evolution of geographical elements across spatial dimensions. This method is defined by the overlaying trajectory code values that represent different periods of land use types³⁹. For instance, cultivated land and non-cultivated land are denoted by 0 and 1, respectively. The change trajectory of a pixel is derived by overlaying the same pixel across the time series and is expressed mathematically as follows:

In the Eq. (1), T_ij denotes the code value of the i-th row and j-th column.The variable n signifies the total number of time nodes, which is 5 in this study. The variables (G₁)_ij, (G₂)_ij,…, (G_n)_ij denote land-class code values at the corresponding time nodes.

Bayesian network model

Proposed by Pearl, the Bayesian Network (BN) is a probabilistic graphical model that combines probability theory and graph theory to represent and reason about the interrelationships among uncertain phenomena in an intuitive yet rigorous manner³¹. When applied to analyze the driving mechanism of cultivated land changes in the urban–rural fringe of Nanchang City, this model not only enables the clear visualization of dependencies and causal relationships among variables through a directed acyclic graph (DAG) but also supports both forward and backward inference—facilitating rapid experimentation and tracing.The fundamental components of the Bayesian network model include constructing a directed acyclic graph, obtaining network parameters to establish a conditional probability table, and developing the model using Netica software, which supports sensitivity and diagnostic analysis.

Initially, key factors were selected as nodes to construct a directed acyclic graph. Changes in cultivated land reflect the landscape pattern resulting from the interaction between human activities and natural conditions in the urban–rural fringe. Drawing on previous studies of cultivated land use change mechanisms, as well as the specific characteristics of its cultivated land use, this study identified 11 factors across four dimensions (natural environment, geographical location, policy planning, and social economy) to investigate their impact on cultivated land change in the urban–rural fringe of Nanchang City^37,40. (1) Natural environment indicators, including topography and water source conditions, play a direct role in cultivated land change. Three indicators—elevation, slope, and distance from the water source—were selected to represent the natural environment’s influence and were designated as parent nodes. Generally, areas with high elevation, steep slopes, and greater distance from water sources are prone to transformations such as the reversion of cultivated land to forests, orchards, or grasslands. Cultivated land in these regions is predominantly converted into ecological land types, such as forests and grasslands. Consequently, ecological occupation was identified as a child node. (2) Geographic location indicators: The spatial variability of location drives the conversion of cultivated land to construction land. Accessibility to transportation and proximity to the city centre significantly influence the conversion of cultivated land to urban construction land. This study utilises four indicators—distances to railway stations, highway entrances and exits, main roads, and the city center—to represent the influence of geographic location as parent nodes, with construction occupation as the child node. (3) Policy and planning indicators: The urban–rural fringe has characteristics distinct from other regions and is notably influenced by policies, regulations, and spatial plans established by relevant land administrative authorities. Urban development boundaries and permanent basic farmland protection areas and are stringent policy indicators with a clear directive on changes in cultivated land. Hence, two indicators—urban development boundaries and permanent basic farmland protection areas—are chosen to represent the impact of policy and planning on changes of cultivated land as parent nodes, with the enforcement strength of policy and planning as the child node.(4) Socioeconomic indicators : Cultivated land change is also affected by socioeconomic indicators in the urban–rural fringe. Generally, frequent economic activities and concentrated populations tend to cause the phenomenon of the conversion of cultivated land to non-agricultural uses. Therefore, GDP and population are selected as socioeconomic indicators. The degree of socioeconomic impact is used as a child node, with GDP and population are used as parent nodes. The specific structure is illustrated in Fig. 3.

Fig. 3 — Bayesian network structure of driving factors of cultivated land transfer in the urban–rural fringe of Nanchang City.

Secondly, the continuous variables were discretised using ArcGIS 10.8. Based on the research objectives and with reference to relevant literature⁴¹, the land use data were characterised by a spatial resolution of 30 m × 30 m, while the total area of the urban–rural fringe remained relatively small. The use of excessively large grid cells would result in the homogenisation of internal spatial variations; therefore, fishnet grids with dimensions of 500 m × 500 m, 300 m × 300 m, and 100 m × 100 m were generated for comparative evaluation. Preliminary experiments revealed that the 500 m grid contained a substantial number of mixed pixels, the 300 m grid was insufficient to accurately represent fragmented cultivated land patterns, whereas the 100 m grid effectively delineated boundaries between cultivated land and construction land while maintaining a computationally feasible workload. Consequently, the 100 m × 100 m grid configuration was selected for further analysis. Subsequently, data from 13,300 randomly selected sample points were input into the model for training, enabling the derivation of conditional probability tables for all nodes in the Bayesian network.Regarding the cultivated land conversion, instances where cultivated land transitioned to non-cultivated land were categorised as “Yes”, while cases where cultivated land remained unchanged were labelled as “No”, as outlined in Table 2. The corresponding conditional probability tables are depicted in Fig. 4.

Table 2.

Variable discrete classification table.

Variable type	Variable name	Grade code
Variable type	Variable name	1	2	3	4	5
Dependent variables	Y:Cultivated land transfer-out	YES	False
Location factor (A)	A1:Distance from the railway stations/m	< 3000	[3000,6000)	[6000,12,000)	[12000,20,000)	≥ 20,000
	A2:Distance from the highway entrances and exits/m	< 6000	[6000,12,000)	[12000,22,000)	[22000,32,000)	≥ 32,000
	A3:Distance from the main road/m	< 2000	[2000,5000)	[5000,8000)	[8000,12,000)	≥ 12,000
	A4:Distance from the city center/m	< 10,000	[10000,25,000)	[25000,55,000)	[55000,70,000)	≥ 70,000
	Y1:Construction occupation	YES	False
Natural factor (B)	B1:Elevation/m	< 50	[50,100)	[100,200)	[200,400)	≥ 400
	B2:Slope/°	< 2	[2,6)	[6,15)	[15,25)	≥ 25
	B3:Distance from the water sources/m	< 1000	[1000,2500)	[2500,4000)	[4000,5000)	≥ 5000
	Y2:Ecological footprint	Yes	False
Policy Planning (C)	C1:Permanent basic farmland protection area	Yes	False
	C2:Urban development boundary	Yes	False
	Y3:Policy planning protection force	Strong	Weak
Socioeconomic factor (D)	D1:GDP/(10⁴ yuan·km⁻²)	< 500	[500,1000)	[1000,2000)	[2000,3000)	≥ 3000
	D2:Population/(Person·km⁻²)	< 100	[100,300)	[300,500)	[500,1000)	≥ 1000
	Y4:Socioeconomic impact degree	Strong	Weak

Open in a new tab

Fig. 4 — Conditional probability table of the Bayesian network model for influencing factors of cultivated land transfer in the urban–rural fringe of Nanchang City.

Finally, sensitivity analysis and diagnostic analysis were employed to quantitatively assess the relationships among factors in the Bayesian network model⁴¹. Sensitivity analysis evaluates the impact of input variables on target variables by modifying the input variables and observing the resulting variation in the probability of the target variable. A higher percentage of entropy reduction or variance reduction indicates that the indicator exerts a stronger influence on the child node. Diagnostic analysis, which builds upon the results of sensitivity analysis, fixes the target variable to a specific state and examines the changes in factor probability. A higher change value indicates that the factor has a stronger substantial effect on the target variable.

Results analysis

Identification results of the urban–rural fringe in Nanchang City

The natural breaks method was applied to characteristic indices of the urban–rural fringe using ArcGIS 10.8. Areas with both nighttime light index and proportion of construction land below 0.15 were identified as overlay factors for correction. This process produced a spatial evolution map illustrating changes in the urban–rural fringe of Nanchang City over the study period (Fig. 5). To validate the accuracy of the identification, comparative analyses of landscape fragmentation, construction land proportion, and NDVI dynamics were conducted across urban, urban–rural fringe, and rural areas using data from six time points(2000–2024). As shown in Table 3, the urban–rural fringe exhibits more complex land use and vegetation characteristics than both urban and rural zones, confirming its inherent transitional nature. These findings verify the accuracy of the delineation method.

Fig. 5 — Spatial evolution process of the urban–rural fringe in Nanchang City from 2000 to 2024. Created by ArcGIS 10.8 https://www.esri.com/en-us/arcgis/products/arcgis-online/overview.

Table 3.

Comparison of different indicator values in different types of areas.

Types	Landscape fragmentation	The proportion of construction land	NDVI dynamic degree
Urban area	0.0037	0.7413	0.4518
Urban–rural fringe	0.0056	0.4840	0.7523
Rural area	0.0021	0.0413	0.2514

Open in a new tab

As shown in Fig. 5 and the supporting statistics, Nanchang’s urban–rural fringe expanded from 37 administrative villages and communities in 2000 to 144 in 2024, with its area increasing from 11,805.22 to 54,704.37 ha. Spatially, its distribution evolved from a “strip” pattern surrounding the urban core to a distinctive tilted “U”-shape encircling the main built-up area.

To comprehensively capture the dynamics of cultivated land within Nanchang’s urban–rural fringe, all administrative villages and communities identified as part of this fringe during the 2000–2024 period were included in the change trajectory and driving force analyses of this study. The specific study area is delineated in Fig. 6.

Fig. 6 — Research scope of cultivated land.Created by ArcGIS 10.8 https://www.esri.com/en-us/arcgis/products/arcgis-online/overview.

The spatio-temporal change characteristics of cultivated land in the urban–rural fringe of Nanchang City

Variation characteristics of the quantity and structure of cultivated land

Spatial analysis methods, including clipping, intersection, and change trajectory analysis, were employed using a GIS platform to determine the area and proportion of cultivated land inflow and outflow in the urban–rural fringe of Nanchang City across five periods spanning from 2000 to 2024 (Table 4). As shown in Table 4, three types of cultivated land changes that occurred during the study period: no change, transfer out and transfer in. Overall, the area of stable cultivated land decreased. The trends in cultivated land transfer in and transfer out exhibited a similar fluctuation pattern: an initial rise, then a decline, and a subsequent rise. However, the transfer out of cultivated land consistently exceeded the transfer in, resulting in a significant overall decline in cultivated land area in Nanchang City’s urban–rural fringe. Notably, between 2005 and 2010, the proportion difference between cultivated land transfer out and transfer in was the smallest at 32.95%, primarily due to a substantial inflow of cultivated land during this period. In contrast, during the other periods, the proportional difference between cultivated land transfer out and transfer in averaged around 80%, underscoring the significant challenge faced by Nanchang City in reconciling the acquisition and compensation of cultivated land.The transfer in area of cultivated land was minimal, totalling only 444.27 ha, from 2010 to 2015. Despite this, both transfer in and transfer out areas were not substantial during this period. Notably, Nanchang witnessed significant developments such as the completion of Subway Line 1 and 2 within this timeframe, although the land acquisition process occurred between 2005 and 2010. Furthermore, urban function optimisation and renewal initiatives were undertaken, resulting in the enhancement of existing urban construction land and resulting in relatively limited occupation of cultivated land during this period. From 2020 to 2024, the transfer of cultivated land decreased, but the total transfer area remained substantial at 6153.98 ha. This is related to the completion of several major infrastructure projects in Nanchang City in 2025, including the east extension of Metro Line 1, the north extension and the east extension of Metro Line 2. Moreover, a comprehensive analysis of the data reveals a proportional relationship between cultivated land outflow and inflow areas, indicating the proactive stance of the Nanchang municipal government in implementing policies to compensate for cultivated land acquisition. Nevertheless. Nevertheless, effectively balancing land occupation and compensation continues to pose a significant challenge.

Table 4.

Table of changes in the quantity and structure of cultivated land in the urban–rural fringe of Nanchang City from 2000 to 2024.

Transformation direction	Trajectory code	change type	2000–2005		2005–2010		2010–2015		2015–2020		2020–2024
Transformation direction	Trajectory code	change type	Area/ha	Proportion/%	Area/ha	Proportion/%	Area/ha	Proportion/%	Area/ha	Proportion/%	Area/ha	Proportion/%
Transfer out	12	Cultivated land → Forest land	244.27	2.14	212.09	4.23	169.45	3.52	220.43	2.66	151.67	2.27
	13	Cultivated land → Grassland	4.41	0.04	3.45	0.07	4.44	0.09	6	0.07	4.70	0.07
	14	Cultivated land → Construction land	9800.25	85.99	2812.02	56.05	4106.68	85.38	7182.88	86.73	5805.89	86.95
	15	Cultivated land → Water area	476.81	4.18	307.48	6.13	84.53	1.76	217.9	2.63	191.02	2.86
	16	Cultivated land → Unused land	0.09	0	0.27	0.01	0.71	0.01	0.48	0.01	0.70	0.01
	Total		10,525.82	92.36	3335.32	66.48	4365.82	90.76	7627.69	92.1	6153.98	92.16
Transfer in	21	Forest land → Cultivated land	225.36	1.98	360.9	7.19	178.77	3.72	207.58	2.51	143.67	2.15
	31	Grassland → Cultivated land	6.06	0.05	6.51	0.13	3.67	0.08	6.17	0.07	4.28	0.06
	41	Construction land → Cultivated land	90.47	0.79	746.85	14.89	177.1	3.68	309.49	3.74	270.32	4.05
	51	Water area → Cultivated land	548.78	4.82	567.39	11.31	84.44	1.76	129.9	1.57	104.62	1.57
	61	Unused land → Cultivated land	0.14	0	0.4	0.01	0.29	0.01	0.84	0.01	0.59	0.01
Total			870.8	7.64	1682.04	33.52	444.27	9.24	653.98	7.9	523.49	7.84
Unchanged	0	/	52,982.81	/	50,516.91	/	47,833.16	/	40,649.72	/	35,134.68	/

Open in a new tab

The conversion of cultivated land in the urban–rural fringe of Nanchang City primarily shifted towards construction land and water areas, with varying sources of cultivated land inflow observed across different periods. Between 2000 and 2005, the main sources of cultivated land were 548.78 ha of water areas and 225.36 ha of forest land. Subsequently, from 2005 to 2010, the cultivated land inflow area exceeded 150 ha, predominantly originating from construction land, water areas, and forest land, with respective areas of 746.85 ha, 567.39 ha, and 360.90 ha. In the periods spanning from 2010 to 2015 and from 2015 to 2020, cultivated land primarily resulted from the transformation of construction and forest land. Finally, between 2020 and 2024, cultivated land conversions originated from forest land, construction land and water areas.

Characteristics of the spatial distribution change of cultivated land

Using the spatial visualisation tool in ArcGIS 10.8, we generated a map illustrating the changes in cultivated land within the urban–rural fringe of Nanchang City from 2000 to 2024 (Fig. 7). Figure 7 reveals a concentrated outward expansion of land transfers from the city centre, with the transferred-out areas primarily clustered around the central districts of Nanchang City, including Xihu District, Donghu District, Qingshanhu District, Xinjian District, and Qingyunpu District. Additionally, these transfers are evident around the central areas of county-level cities such as Jinxian County and Anyi County, indicating the continuous urban expansion of Nanchang City. Conversely, the transferred-in cultivated land exhibits a more scattered distribution compared to the transferred-out areas, suggesting significant land occupation within the urban–rural fringe and substantial pressure on land protection measures. Furthermore, the locations of transferred-in land are interspersed among unchanged and transferred-out areas, leading to a fragmented pattern of cultivated land distribution. The map illustrates a discernible pattern of outward migration of unchanged cultivated land, with a diminishing coverage area over time. Notably, the extent of unchanged cultivated land was prominent between 2000 and 2005; however, by 2020–2024, many previously stable cultivated areas from 2000 to 2005 had transitioned into converted-out cultivated land. This shift underscores the expansion of Nanchang City’s downtown area during the study period, resulting in substantial conversion of cultivated land into developed areas. Concurrently, the urban–rural fringe continued to expand outwards.

Fig. 7 — Spatial variation of cultivated land in urban–rural fringe of Nanchang City from 2000 to 2024.Created by ArcGIS 10.8 https://www.esri.com/en-us/arcgis/products/arcgis-online/overview.

Analysis of the driving mechanism of cultivated land change in the urban–rural fringe of Nanchang City

Results of sensitivity analysis

The analysis variable “Cultivated land transfer out Y” was selected to conduct a sensitivity analysis using the “Sensitivity to Findings” function in Netica software. This analysis aimed to quantify the impact of each explanatory variable on cultivated land transfer. Table 5 presents the sensitivity analysis results for cultivated land transfer out in the urban–rural fringe of Nanchang City. Mutual information indicates the degree of interdependence between variables, while the percentage of entropy reduction and variance are used to assess the marginal contribution intensity of explanatory variables to cultivated land transfer.

Table 5.

The results of the target variable sensitivity analysis.

Node	Mutual information	Entropy reduction percentage/%	Variance
Construction occupation (Y1)	0.66606	75.8	0.1764647
Ecological footprint (Y2)	0.01657	1.88	0.0053334
Policy planning protection force (Y3)	0.00095	0.108	0.0002647
Socioeconomic impact degree (Y4)	0.00063	0.0722	0.0001859
Distance from the city center (A4)	0.00797	0.907	0.0022467
Distance from the railway stations (A1)	0.00757	0.861	0.0022867
Distance from the highway entrances and exits (A2)	0.00463	0.527	0.0014578
Distance from the main roads (A3)	0.00125	0.143	0.0003881
Permanent basic farmland protection area (C1)	0.00063	0.0719	0.000179
Elevation (B1)	0.00019	0.0214	0.0000588
Slope (B2)	0.00013	0.0145	0.0000396
Population (D2)	0.00009	0.00993	0.0000253
Distance from the river water surface (B3)	0.00004	0.00459	0.0000122
GDP (D1)	0.00002	0.00226	0.0000058
Urban development boundary (C2)	0.00001	0.000912	0.0000023

Open in a new tab

Table 5 illustrates the hierarchy of influence of the four primary intermediate nodes (Y1 to Y4) on the conversion of cultivated land in the urban–rural fringe of Nanchang City. The order of influence is as follows: construction occupation (Y1) > ecological occupation (Y2) > policy planning and protection force (Y3) > socioeconomic influence construction (Y4). Specifically, the entropy reduction percentage of Y1 in the cultivated land conversion is 75.8%, with a variance of 17. This indicates that a large scale of urban construction occurred in the urban–rural fringe of Nanchang City during the study period, leading to the occupation of a significant portion of cultivated land, which aligns with the actual developmental trends in the urban–rural fringe. The entropy reduction percentage of Y2 is 1.88%. The urban–rural fringe facilitates metropolitan civilisation construction and addresses citizens’ ecological needs, leading to the utilization of a certain extent of cultivated land for ecological purposes. Y3 exhibits a relatively minor entropy reduction percentage of 0.108%. While policy planning has had some inhibitory effect on the cultivated land transfer, its overall influence on cultivated land conversion in the urban–rural fringe of Nanchang City is minimal. Y4 exerts the least influence on the conversion of cultivated land.

Regarding the multidimensional variables, the input variables A1 to A4,which pertain to the location factor, exhibit the most substantial influence on the conversion of cultivated land to other uses. Notably, A4 has the most pronounced impact on this conversion, leading to an entropy reduction of 0.907%. This is followed by A1, A2, and A3, which contribute entropy reductions of 0.143%,0.108%, and 0.0722%, respectively. This phenomenon is primarily observed at the urban–rural interface, where land conversion for construction purposes tends to occur closer to urban areas with convenient transport access. Regarding natural factors, elevation (B1) and slope (B2) significantly affect the conversion of cultivated land, while the distance from water sources (B3) has a comparatively minor influence, likely due to the water bodies abundance in the study area. Socioeconomic and policy planning factors have minimal effects on the cultivated land transfer. The permanent basic farmland protection area (C1) serves to safeguard cultivated land, whereas the urban development boundary (C2) has a minor influence on land transfer. Given the area’s status as a small-scale urban–rural fringe, variations in GDP (D1) and population (D2) are insignificant, resulting in limited impacts on land transfer within the study area.

Results of diagnostic analysis

To quantitatively characterise the causal relationship between cultivated land conversion and each driving factor, this paper conducts reverse inference using a Bayesian network. Setting the probability of “FALSE” for cultivated land conversion at 100%, the study then observes the probability of each factor, calculates the variance from the sensitivity outcome, and presents the findings of the diagnostic analysis (Fig. 8).

As shown in Fig. 8, under the scenario of “no transfer out of cultivated land”, the results for the four dimensional indicators-natural environment, geographical location, policy planning, and socioeconomic factors—all exhibit significant changes,These changes can be categorized into the following four points:

In terms of geographical location, the probability of the intermediate node “construction occupation” being “yes” decreases by 27.21%. The closer the area is to the city center, railway stations, highway exits and entrances, and main roads, the greater the decline in the probability of this factor. Conversely, the probability for more distant categories shows a compensatory increase, with the first two levels of proximity to railway stations, highway exits and entrances, and main roads being the most significant. This result further confirms that the expansion of construction land is the primary driving force behind the conversion of cultivated land, and that transportation accessibility further amplifies this impact.
In terms of the natural environment, the likelihood of the intermediate node of ecological occupation being in the “Yes” state decreased by 1.40%. The probability of the elevation factor “ < 50” increased by 0.13%, while the probabilities of the slope factor being “less than 2” and within the range [2, 6) increased by 0.06% and 0.03%, respectively. Conversely, the probability of the distance to the water sources “ < 100” decreased by 0.10%. These findings suggest that cultivated land on flat terrain is comparatively less susceptible to ecological occupation. However, being situated in an urban–rural fringe area, locations in close proximity to water bodies are more likely to satisfy the public’s ecological requirements and be converted into ecological land.
Regarding the policy planning, there has been a 0.60% increase in the likelihood of the policy planning protection force node being classified as “strong.” Notably, the two metrics within this domain exhibit divergent trends: the probability of designating the permanent basic farmland protection area as “yes” has risen by 0.60%, whereas the likelihood of marking the urban development boundary as “yes” has declined by 0.10%. This observation underscores the substantial positive impact of the permanent basic farmland system on the preservation of cultivated land stability.
In terms of the socioeconomic dimension, the likelihood of a “strong” state for the socioeconomic influence node decreased by 0.90%. Conversely, the probabilities of the top three levels in the population factor all increased. This suggests that in areas with lower population density, human activities occur less frequently, resulting in reduced damage and interference with cultivated land. Consequently, cultivated land in these regions tends to be more stable.

Discussion

This study focuses on changes in cultivated land within the urban–rural fringe. Drawing on the three core characteristics of the urban–rural fringe, a relatively systematic and highly operational indicator system was developed to construct a model for the long-term sequential identification of the urban–rural fringe boundary. Verification results confirm the model’s feasibility and efficiency. Furthermore, this study employs a Bayesian network model to analyse the driving mechanisms behind cultivated land changes within the urban–rural fringe, aiming to fully capture the causal relationships among variables and the spatial regularity of distance factors, thereby providing a reference for formulating cultivated land protection policies in these areas.

The constructed multidimensional feature index identification model

Based on the essential characteristics of the urban–rural fringe, a multidimensional feature index identification model was developed to delineate the urban–rural fringe of Nanchang City. Compared with previous studies⁴², the nighttime light index in the constructed indicators utilised the “NPP-VIIRS-like” nighttime light dataset from the National Earth System Science Data Center. This approach resolved the issue of incompatibility between data from different sensors (DMSP-OLS and NPP-VIIRS) and offered advantages such as long-term consistency and high data quality²¹. Furthermore, the spatial evolution results of the urban–rural fringe identified by the model corresponded closely with the actual development of Nanchang City. Overall, this study provides a novel and feasible method for the long-term identification of the urban–rural fringe.

Spatio-temporal characteristics and driving mechanisms of cultivated land changes in the urban–rural fringe of Nanchang City

From 2000 to 2024, the area of cultivated land in the urban–rural fringe of Nanchang exhibited a statistically significant decreasing trend, with the primary conversions being to construction land and water bodies—consistent with findings in the existing literature⁴³. This phenomenon is chiefly attributed to Nanchang’s rapid urbanisation and industrialisation during this period, as evidenced by large-scale upgrading and optimisation of infrastructure (including power grids, road networks, pipeline networks, and water supply systems), alongside the construction or expansion of multiple industrial parks and economic development zones. This process drove the rapid outward expansion of the main urban area, facilitating the spatial shift of the urban–rural fringe and consequently leading to extensive conversion of cultivated land into construction land.

Based on the results of the Bayesian network model, the urban–rural fringe is subject to the dual impacts of urban expansion and rural development. Among the factors affecting its cultivated land, construction occupation exerts the most significant influence, followed by ecological occupation, while socioeconomic factors have a relatively minor effect—this aligns with the findings of Shen et al. (2023)⁴¹. However, the conclusion that construction occupation is the primary driver of cultivated land changes in the urban–rural fringe contradicts the results of Meng et al. (2025)⁴⁴. Using a geographically weighted regression (GWR) model to explore the driving factors of cultivated land occupation by urban expansion in the Chengdu metropolitan area, Meng et al. (2025) identified population density, non-agricultural population, and grain output as the dominant factors⁴⁴. This discrepancy is primarily attributed to the conceptual distinction between metropolitan areas and urban–rural fringes. First proposed by the United States in 1910, the metropolitan area has gradually been defined by its core concept as a region consisting of a central city with a specific population size and adjacent surrounding areas with close socioeconomic ties, encompassing urban areas, urban–rural fringes, and rural areas. Thus, the drivers of cultivated land changes in metropolitan areas may overlap with those in urban–rural fringes but lack sufficient granularity, leading to divergent results. Additionally, the driving mechanisms of cultivated land changes in the urban–rural fringe differ considerably from those in other regions, particularly in typical mountainous areas. For instance, Chen et al. (2024) employed a system dynamics model and found that the key factors contributing to cultivated land abandonment in Luxi County, a typical mountainous region, can be summarised as agricultural income levels, agricultural labour force, and agricultural insurance⁴⁵. Given the dramatic changes in cultivated land within the urban–rural fringe and the distinctiveness of its influencing factors compared to other regions, it is implicitly indicated that targeted cultivated land protection measures tailored to the characteristics of the urban–rural fringe are necessary.

Suggestions for cultivated land protection in the urban–rural fringe of Nanchang City

Over the past twenty years, the urban–rural fringe in Nanchang has undergone consistent expansion, with its construction land area increasing from 7632.25 ha to 43,975.49 ha. Concurrently, the area of cultivated land has decreased by 48.27%. This trend has intensified the conflict between urban development and the preservation of cultivated land. To enhance the spatial organisation of cultivated land and harmonise the relationship between societal progress and cultivated land conservation, the following recommendations are proposed, based on the results of the Bayesian network analysis and the above conclusions presented earlier in the text.

The research findings indicate a shift in the development focus of Nanchang City’s urban–rural fringe from incremental expansion towards quality improvement and efficiency enhancement. This suggests that Nanchang should currently refrain from indiscriminate expansion. Instead, the city should prioritise urban renewal to improve livability, functionality, competitiveness, and inclusiveness. Furthermore, the trajectory table of cultivated land changes reveals that more than 50% of the transferred-out cultivated land has been converted into construction land, with the cultivated land lost due to construction primarily concentrated in areas with convenient transportation. Therefore, it is imperative that a consistent emphasis on protecting cultivated land is maintained throughout the entire transportation project construction process. This necessitates strengthening the integration of transportation construction and cultivated land protection planning, as well as leveraging the overarching coordination role of territorial spatial planning. Moreover, the occupation of cultivated land should be a pivotal consideration in scheme comparison and evaluation, with appropriate technical measures employed to minimise cultivated land occupation.

The diagnostic results indicate that the demarcation of permanent basic farmland has set boundaries for urban expansion and various types of construction, as well as a firm baseline for farmland preservation. In the urban–rural fringe, delineating permanent basic farmland requires a delicate balance between flexibility and strictness. On the one hand, high-quality arable land characterised by concentrated distribution, optimal infrastructure, fertile soil, reliable water sources, and gentle slopes should be given prioritized for inclusion within the permanent basic farmland zone to ensure rigorous protection. Additionally, studies indicate that urban development boundaries have limited effectiveness in preventing the conversion of cultivated land. Therefore, implementing a “skylight opening” approach can effectively safeguard prime farmland within the boundary, meeting both the need for strict preservation and the multifunctional use of cultivated land in the urban–rural fringe. In addition, given the likelihood of the urban–rural fringe being converted into urban areas in the future, strategic measures should be taken to avoid key transportation corridors, major infrastructure nodes, and core areas of principal development clusters in the short term. Moreover, it is advisable to designate a buffer zone of non-permanent basic farmland between permanent basic farmland and urban construction zones ,serving as a flexible area for potential urban expansion.

Limitations and future research

Although this study has achieved certain findings, it is not without limitations, including issues related to the selection of fishnet scales, the timeliness of partial data, inadequacies in the indicator system, and the relative generality of recommendations for cultivated land protection. To address these shortcomings, future research should be optimised in the following areas.

Firstly, the systematic selection of fishnet scales and the comparison of results should be improved. In constructing the fishnet size for analysing the driving mechanisms, this study primarily focused on the urban–rural fringe and its changes in cultivated land. Given the 30 m × 30 m spatial resolution of the cultivated land data and the relatively limited extent of the urban–rural fringe, this approach differs from that of Jin et al. (2025), who selected a 3 km × 3 km fishnet for their analysis of large-scale, highly homogeneous ecological quality issues⁴⁶. Additionally, a number of preliminary experiments were conducted in this study. Comparing the results of cultivated land change driving mechanisms across different scales will facilitate a more systematic and detailed understanding of these mechanisms in the urban–rural fringe, which also represents a promising direction for future research.

Secondly, the timeliness of policy planning factor data should be more rigorously assessed. To incorporate policy planning factors into this study and achieve spatial quantification, the permanent basic farmland protection zones and urban development boundaries considered are all derived from established spatial planning data. However, given the time-sensitive nature of policy planning, future research must evaluate the currency of policy planning data to more accurately measure the impact of policy planning on changes in cultivated land.

Thirdly, the range of influencing factors should be broadened. On the one hand, this study employs landscape fragmentation and the largest patch index within a multidimensional feature index identification model to characterise spatial heterogeneity, effectively revealing the characteristics of the urban–rural fringe. However, heterogeneity is a multi-dimensional concept, and a comprehensive characterisation should also include edge density, landscape shape index, and the Shannon index, among others^47,48. Future research that obtains higher-precision land use data and multi-dimensional landscape indicators will be better able to delineate the differences between the urban–rural fringe, urban areas, and rural areas—representing a clear and significant extension of this study. On the other hand, beyond the indicators selected here, certain factors that are difficult to obtain or accurately quantify spatially, such as government decision-making and engineering techniques, also influence changes in cultivated land within the urban–rural fringe. Subsequent research should seek to further quantify these potential influencing factors.

Fourth, formulate practical and viable strategies for the protection of cultivated land in the urban–rural fringe. By analysing the quantitative conversion of cultivated land and its underlying driving mechanisms, this study proposes corresponding recommendations for cultivated land protection, which hold considerable reference value. The findings regarding influencing factors also indicate that the occupation of cultivated land by construction exerts the most significant impact. Therefore, future research should consider increasing the quality and multi-functionality of cultivated land to guide the direction of urban development.

Through these systematic improvements and in-depth analyses, future studies will offer more scientific, rigorous, and practical recommendations for the protection of cultivated land in the urban–rural fringe of Nanchang and comparable regions, thereby promoting the coordinated development of urban–rural integration and cultivated land conservation.

Conclusion

Based on multi-source data and Bayesian network model, this research delineates the urban–rural fringe of Nanchang City over the past twenty years and elucidated the spatio-temporal dynamics and causal factors influencing cultivated land within this transitional zone. The principal findings include:

The multidimensional characteristic index identification model for the urban–rural fringe has yielded favourable results in long-term spatial identification. Since 2000, the area of the urban–rural fringe in Nanchang City has exhibited an expanding trend, characterised by a dual process of outer-ring expansion and inner-ring contraction. This identification model can also be applied to delineate similar urban–rural fringe areas, such as Changsha (Hunan Province) and Wuhan (Hubei Province).
An analysis of the quantitative structural changes in cultivated land within the urban–rural fringe of Nanchang City reveals that cultivated land utilisation between 2000 and 2024 was characterised by the coexistence of transfer-out, transfer-in, and stable statuses. Cultivated land transfer-out was predominantly directed towards construction land and water areas, while transfer-in primarily originated from forest land, construction land, and water areas. A larger scale of transfer-out corresponded to a subsequent increase in transfer-in, reflecting the effective local implementation of the cultivated land occupation-compensation balance policy.
The spatial characteristics of cultivated land change in the urban–rural fringe of Nanchang City indicate that the transferred-out cultivated land was concentrated around the city center, gradually expanding outward. In contrast, the transferred-in cultivated land was more dispersed, reflecting the progression of the inner-ring urban–rural fringe integrating into the city and the rural areas surrounding the city evolving into the outer-ring urban–rural fringe.
The driving factors for the change of cultivated land in the Nanchang City’s urban–rural fringe are ranked as follows: construction occupation holds the greatest significance, followed by ecological occupation, policy planning protection force, and socioeconomic influence. Notably, the construction occupation emerges as the predominant factor. The level of transportation convenience directly correlates with the likelihood of cultivated land occupation, aligning with the observed trends in the development of Nanchang City’s urban–rural fringe.
It is imperative to optimise the cultivated land layout through multiple channels and coordinate the relationship with urban development, including shifting from carpet-style expansion to urban renewal, comprehensively planning transportation layout and adopting land-saving measures, and strengthening the balance between rigidity and flexibility in the demarcation of permanent basic farmland in the urban–rural fringe.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 42461041), 2024 Annual Nanchang City Social Science Planning Project (Grant No. GL202405), Technology Innovation Center for Land Spatial Ecological Protection and Restoration in Great Lakes Basin, MNR, and the Research Center on Rural Land Resources Use and Protection of Jiangxi Agricultural University.

Author contributions

Jianping Wang: Conceptualization, Methodology, Data curation, Formal analysis, Visualization, Writing—original draft, Funding acquisition; Zhenghong Zhu: Writing—review & editing, Methodology, Data curation, Software; Meiqiu Chen: Project administration, Funding acquisition, Conceptualization, Supervision, Writing—review & editing; Yiguo Zhang: Methodology, Data curation.

Funding

Data availability

The datasets used and analysed during the current study available from the corresponding author(Cmq12@263.net) on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1.Li, Q., Jiang, W. T., Gao, W. & Lyu, J. How to build a framework for soil health of cultivated land assessment and monitoring: Insights from the perspective of government assistance to smallholders. Land Use Policy157, 107645. 10.1016/J.LANDUSEPOL.2025.107645 (2025). [Google Scholar]
2.Luo, X. Z., Ye, Y. C., Guo, X., Zhao, X. & Kuang, L. Multifunctional evolution response mechanisms to urbanization processes on peri-urban cultivated land, Nanchang City, China. Land14, 259. 10.3390/LAND14020259 (2025). [Google Scholar]
3.Qu, Y. B., Zhan, L. Y., Wei, C. C., Zhang, Q. & Wang, J. Interactive transition of cultivated land and construction land during china’s urbanization: A coordinated analytical framework of explicit and implicit forms. Land Use Policy138, 1047049. 10.1016/J.LANDUSEPOL.2024.107049 (2024). [Google Scholar]
4.Ning, J., Ma, H. Z., Sun, Y., Wang, N. & Wang, M. Research on the characteristic identification and multidimensional dynamic evolution of urban-rural fringe in Harbin, China. Land14, 359. 10.3390/LAND14020359 (2025). [Google Scholar]
5.Wang, K., Mell, I. & Carter, J. Characterising the urban-rural fringe area (URFA) in China: A review of global and local literature on urban-rural fringe areas. Town Plann. Rev.95(6), 617–642. 10.3828/TPR.2024.26 (2024). [Google Scholar]
6.Chen, Y. Q. & Wu, W. Analysis on the characteristics and evolution mechanism of human-land system in urban-rural fringe. Geogr. Sci.18(5), 27–33. 10.13249/j.Cnki.Sgs.1998.05.004 (1998) (in Chinese). [Google Scholar]
7.Krueger, R. R. & Bryfogle, R. C. Urban Problems: A Canadian Reader (Holt, Rinehart and Winston of Canada, 1971). [Google Scholar]
8.Bryant, C. R., Russwurm, L. H. & Mclellan, A. G. The city’s countryside, land and its management in the rural-urban fringe (Addison-Wesley, 1982). [Google Scholar]
9.LeSage, J. P. & Charles, J. S. Using home buyers’ revealed preferences to define the urban-rural fringe. J. Geogr. Syst.10(1), 1–21. 10.1007/s10109-007-0055-z (2008). [Google Scholar]
10.Dadashpoor, H. & Ahani, S. Explaining objective forces, driving forces, and causal mechanisms affecting the formation and expansion of the peri-urban areas: A critical realism approach. Land Use Policy102, 105232. 10.1016/j.landusepol.2020.105232 (2021). [Google Scholar]
11.Ji, D., Tian, J., Zhang, J. H., Zeng, J. & Namaiti, A. Identification and spatiotemporal evolution analysis of the urban-rural fringe in polycentric cities based on k-means clustering and multi-source data: A case study of Chengdu City. Land13(11), 1727. 10.3390/land13111727 (2024). [Google Scholar]
12.Hu, Y. N., Connor, D. S., Stuhlmacher, M., Peng, J. & Turner, B. L. T. More urbanization, more polarization: Evidence from two decades of urban expansion in China. npj Urban Sustain.4(1), 33. 10.1038/s42949-024-00170-z (2024). [Google Scholar]
13.Duan, H. R., Du, F. G., Zhang, Y. J., Jiang, X. J. & Chen, B. An urban-rural fringe extraction method based on combined urban-rural fringe index (CUFI). Geocarto Int.39(1), 2311211. 10.1080/10106049.2024.2311211 (2024). [Google Scholar]
14.Yin, G. M., Feng, Y., Jiang, Y. X. & Bao, Y. A Scene-object-economy framework for identifying and validating urban-rural fringe using multisource geospatial big data. Appl. Sci.14(22), 10191. 10.3390/app142210191 (2024). [Google Scholar]
15.Cao, H. et al. Multi-dimensional identification and feature analysis of Nanjing’s urban-rural fringe. J. Nat. Resour.38(3), 795–807. 10.31497/zrzyxb.20230315 (2023) (in Chinese). [Google Scholar]
16.Ye, S. J. et al. Spatial pattern of cultivated land fragmentation in mainland China: Characteristics, dominant factors, and countermeasures. Land Use Policy139, 107070. 10.1016/j.landusepol.2024.107070 (2024). [Google Scholar]
17.Zhang, X. & Han, H. Spatiotemporal dynamics and driving forces of cultivated land in China. Sci. Rep.15(1), 31149–31149. 10.1038/S41598-025-11755-8 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Wang, X., Song, X. Q., Wang, Y., Xu, H. X. & Ma, Z. H. Understanding the distribution patterns and underlying mechanisms of non-grain use of cultivated land in rural China. J. Rural. Stud.106, 103223. 10.1016/j.jrurstud.2024.103223 (2024). [Google Scholar]
19.Potapov, P. et al. Global maps of cropland extent and change show accelerated cropland expansion in the twenty-first century. Nat. Food.3(1), 19. 10.1038/s43016-021-00429-z (2022). [DOI] [PubMed] [Google Scholar]
20.Liu, S. C., Xiao, W., Ye, Y. M., He, T. T. & Luo, H. Rural residential land expansion and its impacts on cultivated land in China between 1990 and 2020. Land Use Policy132, 106816. 10.1016/j.landusepol.2023.106816 (2023). [Google Scholar]
21.Chen, L. L., Zhao, H. S., Song, G. & Liu, Y. Optimization of cultivated land pattern for achieving cultivated land system security: A case study in Heilongjiang Province, China. Land Use Policy108, 105589. 10.1016/j.landusepol.2021.105589 (2021). [Google Scholar]
22.Zhu, J., Li, X. T., Zeng, X. C., Zhong, K. Y. & Xu, Y. F. Cultivated land-use benefit evaluation and obstacle factor identification: Empirical evidence from Northern Hubei, China. Land11(9), 1386. 10.3390/land11091386 (2022). [Google Scholar]
23.Zhou, J., Li, C., Chu, X. T. & Luo, C. Y. Is cultivated land increased by land consolidation sustainably used in mountainous areas?. Land11(12), 2236. 10.3390/land11122236 (2022). [Google Scholar]
24.Zhu, Z. Y. et al. Soil quality evaluation of different land use modes in small watersheds in the hilly region of southern Jiangsu. Ecol. Indic.160, 111895. 10.1016/j.ecolind.2024.111895 (2024). [Google Scholar]
25.Gao, X. et al. Response of cultivated land quality to landscape pattern changes at the watershed scale. J. Environ. Manag.392, 126818. 10.1016/J.JENVMAN.2025.126818 (2025). [DOI] [PubMed] [Google Scholar]
26.Wu, M. Q. & Chen, X. G. Changes in arable land area in Qinzhou City and its driving force based on the principle component analysis. Geogr. Sci.34(01), 54–59. 10.13249/j.cnki.sgs.2014.01.017 (2014). [Google Scholar]
27.Qian, F. K., Jiao, S. Y., Yu, Y. J., Wang, X. G. & Shao, T. Y. Cultivated land quality assessment and obstacle factors diagnosis in Changtu County, Northeast China. Land Degrad. Dev.35(17), 5065–5077. 10.1002/ldr.5278 (2024). [Google Scholar]
28.Xiang, J. W., Han, P. & Chen, W. X. Coordinated development efficiency between cultivated land spatial morphology and agricultural economy in underdeveloped areas in China: Evidence from western Hubei province. J. Geogr. Sci.33(4), 801–822. 10.1007/s11442-023-2107-9 (2023). [Google Scholar]
29.Wu, T., Wu, S., Hu, S. G. & Zhang, Q. Simulating future cultivated land using a localized SSPs-RCPs framework: A case study in Yangtze River Economic Belt. Habitat Int.154, 103210. 10.1016/j.habitatint.2024.103210 (2024). [Google Scholar]
30.Yang, H. et al. Sustainable utilization of cultivated land resources based on “element coupling-function synergy” analytical framework: A case study of Guangdong, China. Land Use Policy146, 107316. 10.1016/j.landusepol.2024.107316 (2024). [Google Scholar]
31.Heckerman, D. Bayesian networks for data mining. Data. Min. Knowl. Disc.1, 79–119. 10.1023/A:1009730122752 (1997). [Google Scholar]
32.Xu, X. L., Liu, J. Y. & Zhuang, D. F. Remote sensing monitoring methods of land use/cover changes in national scale. J. Anhui Agric. Sci.40(4), 2365–2369. 10.13989/j.cnki.0517-6611.2012.04.025 (2012) (in Chinese). [Google Scholar]
33.Ahani, S. & Dadashpoor, H. A review of domains, approaches, methods and indicators in peri-urbanization literature. Habitat. Int.114, 102387. 10.1016/j.habitatint.2021.102387 (2021). [Google Scholar]
34.Peng, J., Hu, Y. N., Liu, Y. X., Ma, J. & Zhao, S. Q. A new approach for urban-rural fringe identification: Integrating impervious surface area and spatial continuous wavelet transform. Landsc. Urban. Plan.175, 72–79. 10.1016/j.landurbplan.2018.03.008 (2018). [Google Scholar]
35.Liang, C. & Zeng, J. Three-dimensional dynamic identification and spatial evolution mechanism of urban-rural fringe: A case study of Xiamen City. Geogr. Res.40(05), 1404–1420 (2021) (in Chinese). [Google Scholar]
36.Kumar, R. et al. Revealing the benefits of entropy weights method for multi-objective optimization in machining operations: A critical review. J. Mater. Res. Technol.10, 1471–1492. 10.1016/j.jmrt.2020.12.114 (2021). [Google Scholar]
37.Wang, J. Y., Wang, S. J. & Zhou, C. S. Quantifying embodied cultivated land-use change and its socioeconomic driving forces in China. Appl. Geogr.137, 102601. 10.1016/j.apgeog.2021.102601 (2021). [Google Scholar]
38.Marzialetti, F., Gamba, P., Sorriso, A. & Carranza, M. L. Monitoring urban expansion by coupling multi-temporal active remote sensing and landscape analysis: Changes in the Metropolitan Area of Cordoba (Argentina) from 2010 to 2021. Remote Sens.15(2), 336. 10.3390/rs15020336 (2023). [Google Scholar]
39.Li, Y. L. et al. Study on transformation trajectory and driving factors of cultivated land in the Yellow River Delta in recent 30 years. China Popul. Resour. Environ.29(9), 136–143. 10.12062/cpre.20190601 (2019) (in Chinese). [Google Scholar]
40.Li, H. H. & Song, W. Spatial transformation of changes in global cultivated land. Sci. Total. Environ.859(1), 160194. 10.1016/j.scitotenv.2022.160194 (2023). [DOI] [PubMed] [Google Scholar]
41.Shen, Y., Gong, J., Li, X. D., Wang, W. W. & Tao, R. Driving mechanism and policy implications of cultivated land use change in Wuhan Metropolitan Area. China Land Sci.37(08), 123–133. 10.11994/zgtdkx.20230804.163543 (2023) (in Chinese). [Google Scholar]
42.Yang, Y. L., Ma, M. G., Tan, C. & Li, W. P. Spatial recognition of the urban-rural fringe of Beijing using DMSP/OLS nighttime light data. Remote Sens.9(11), 1141. 10.3390/rs9111141 (2017). [Google Scholar]
43.Wang, L., Lv, T. G. & Tan, Y. Z. Understanding the characteristics and coupling evolution trends of the non agricultural transformation of “population-land-industry”: A case study in Nanchang City. Environ. Dev. Sustain.10.1007/S10668-025-06270-9 (2025). [Google Scholar]
44.Meng, B., Wang, X. X., Zhang, Z. F. & Huang, P. Spatio-temporal pattern and driving force evolution of cultivated land occupied by urban expansion in the Chengdu Metropolitan Area. Land.11(9), 1458. 10.3390/land11091458 (2022). [Google Scholar]
45.Cheng, Y. Y., Chen, W. B. & Lai, S. S. Research on precise identification and driving mechanisms of abandoned cultivated land in typical mountainous Areas. China Land Sci.38(12), 93–104. 10.11994/zgtdkx.20241209.092816 (2024) (in Chinese). [Google Scholar]
46.Jin, A. B., Li, H., Wang, X. R. & Wang, Z. Y. Spatiotemporal assessment of ecological quality and driving mechanisms in the Beijing metropolitan area. Sci. Rep.15(1), 13136–13136. 10.1038/S41598-025-97156-3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Xiong, Y. G. & Jin, A. B. Spatial gradient effects of landscape pattern on ecological quality along the grand canal. Land14(6), 1310–1310. 10.3390/LAND14061310 (2025). [Google Scholar]
48.Zhang, J. M., Wang, P. H. & Jin, A. B. Multidimensional characteristics of urban green space and its impact in mitigating urban heat Island effects: A case study of Guangzhou. Sci. Rep.15(1), 39959–39959. 10.1038/S41598-025-23773-7 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The datasets used and analysed during the current study available from the corresponding author(Cmq12@263.net) on reasonable request.

[CR1] 1.Li, Q., Jiang, W. T., Gao, W. & Lyu, J. How to build a framework for soil health of cultivated land assessment and monitoring: Insights from the perspective of government assistance to smallholders. Land Use Policy157, 107645. 10.1016/J.LANDUSEPOL.2025.107645 (2025). [Google Scholar]

[CR2] 2.Luo, X. Z., Ye, Y. C., Guo, X., Zhao, X. & Kuang, L. Multifunctional evolution response mechanisms to urbanization processes on peri-urban cultivated land, Nanchang City, China. Land14, 259. 10.3390/LAND14020259 (2025). [Google Scholar]

[CR3] 3.Qu, Y. B., Zhan, L. Y., Wei, C. C., Zhang, Q. & Wang, J. Interactive transition of cultivated land and construction land during china’s urbanization: A coordinated analytical framework of explicit and implicit forms. Land Use Policy138, 1047049. 10.1016/J.LANDUSEPOL.2024.107049 (2024). [Google Scholar]

[CR4] 4.Ning, J., Ma, H. Z., Sun, Y., Wang, N. & Wang, M. Research on the characteristic identification and multidimensional dynamic evolution of urban-rural fringe in Harbin, China. Land14, 359. 10.3390/LAND14020359 (2025). [Google Scholar]

[CR5] 5.Wang, K., Mell, I. & Carter, J. Characterising the urban-rural fringe area (URFA) in China: A review of global and local literature on urban-rural fringe areas. Town Plann. Rev.95(6), 617–642. 10.3828/TPR.2024.26 (2024). [Google Scholar]

[CR6] 6.Chen, Y. Q. & Wu, W. Analysis on the characteristics and evolution mechanism of human-land system in urban-rural fringe. Geogr. Sci.18(5), 27–33. 10.13249/j.Cnki.Sgs.1998.05.004 (1998) (in Chinese). [Google Scholar]

[CR7] 7.Krueger, R. R. & Bryfogle, R. C. Urban Problems: A Canadian Reader (Holt, Rinehart and Winston of Canada, 1971). [Google Scholar]

[CR8] 8.Bryant, C. R., Russwurm, L. H. & Mclellan, A. G. The city’s countryside, land and its management in the rural-urban fringe (Addison-Wesley, 1982). [Google Scholar]

[CR9] 9.LeSage, J. P. & Charles, J. S. Using home buyers’ revealed preferences to define the urban-rural fringe. J. Geogr. Syst.10(1), 1–21. 10.1007/s10109-007-0055-z (2008). [Google Scholar]

[CR10] 10.Dadashpoor, H. & Ahani, S. Explaining objective forces, driving forces, and causal mechanisms affecting the formation and expansion of the peri-urban areas: A critical realism approach. Land Use Policy102, 105232. 10.1016/j.landusepol.2020.105232 (2021). [Google Scholar]

[CR11] 11.Ji, D., Tian, J., Zhang, J. H., Zeng, J. & Namaiti, A. Identification and spatiotemporal evolution analysis of the urban-rural fringe in polycentric cities based on k-means clustering and multi-source data: A case study of Chengdu City. Land13(11), 1727. 10.3390/land13111727 (2024). [Google Scholar]

[CR12] 12.Hu, Y. N., Connor, D. S., Stuhlmacher, M., Peng, J. & Turner, B. L. T. More urbanization, more polarization: Evidence from two decades of urban expansion in China. npj Urban Sustain.4(1), 33. 10.1038/s42949-024-00170-z (2024). [Google Scholar]

[CR13] 13.Duan, H. R., Du, F. G., Zhang, Y. J., Jiang, X. J. & Chen, B. An urban-rural fringe extraction method based on combined urban-rural fringe index (CUFI). Geocarto Int.39(1), 2311211. 10.1080/10106049.2024.2311211 (2024). [Google Scholar]

[CR14] 14.Yin, G. M., Feng, Y., Jiang, Y. X. & Bao, Y. A Scene-object-economy framework for identifying and validating urban-rural fringe using multisource geospatial big data. Appl. Sci.14(22), 10191. 10.3390/app142210191 (2024). [Google Scholar]

[CR15] 15.Cao, H. et al. Multi-dimensional identification and feature analysis of Nanjing’s urban-rural fringe. J. Nat. Resour.38(3), 795–807. 10.31497/zrzyxb.20230315 (2023) (in Chinese). [Google Scholar]

[CR16] 16.Ye, S. J. et al. Spatial pattern of cultivated land fragmentation in mainland China: Characteristics, dominant factors, and countermeasures. Land Use Policy139, 107070. 10.1016/j.landusepol.2024.107070 (2024). [Google Scholar]

[CR17] 17.Zhang, X. & Han, H. Spatiotemporal dynamics and driving forces of cultivated land in China. Sci. Rep.15(1), 31149–31149. 10.1038/S41598-025-11755-8 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Wang, X., Song, X. Q., Wang, Y., Xu, H. X. & Ma, Z. H. Understanding the distribution patterns and underlying mechanisms of non-grain use of cultivated land in rural China. J. Rural. Stud.106, 103223. 10.1016/j.jrurstud.2024.103223 (2024). [Google Scholar]

[CR19] 19.Potapov, P. et al. Global maps of cropland extent and change show accelerated cropland expansion in the twenty-first century. Nat. Food.3(1), 19. 10.1038/s43016-021-00429-z (2022). [DOI] [PubMed] [Google Scholar]

[CR20] 20.Liu, S. C., Xiao, W., Ye, Y. M., He, T. T. & Luo, H. Rural residential land expansion and its impacts on cultivated land in China between 1990 and 2020. Land Use Policy132, 106816. 10.1016/j.landusepol.2023.106816 (2023). [Google Scholar]

[CR21] 21.Chen, L. L., Zhao, H. S., Song, G. & Liu, Y. Optimization of cultivated land pattern for achieving cultivated land system security: A case study in Heilongjiang Province, China. Land Use Policy108, 105589. 10.1016/j.landusepol.2021.105589 (2021). [Google Scholar]

[CR22] 22.Zhu, J., Li, X. T., Zeng, X. C., Zhong, K. Y. & Xu, Y. F. Cultivated land-use benefit evaluation and obstacle factor identification: Empirical evidence from Northern Hubei, China. Land11(9), 1386. 10.3390/land11091386 (2022). [Google Scholar]

[CR23] 23.Zhou, J., Li, C., Chu, X. T. & Luo, C. Y. Is cultivated land increased by land consolidation sustainably used in mountainous areas?. Land11(12), 2236. 10.3390/land11122236 (2022). [Google Scholar]

[CR24] 24.Zhu, Z. Y. et al. Soil quality evaluation of different land use modes in small watersheds in the hilly region of southern Jiangsu. Ecol. Indic.160, 111895. 10.1016/j.ecolind.2024.111895 (2024). [Google Scholar]

[CR25] 25.Gao, X. et al. Response of cultivated land quality to landscape pattern changes at the watershed scale. J. Environ. Manag.392, 126818. 10.1016/J.JENVMAN.2025.126818 (2025). [DOI] [PubMed] [Google Scholar]

[CR26] 26.Wu, M. Q. & Chen, X. G. Changes in arable land area in Qinzhou City and its driving force based on the principle component analysis. Geogr. Sci.34(01), 54–59. 10.13249/j.cnki.sgs.2014.01.017 (2014). [Google Scholar]

[CR27] 27.Qian, F. K., Jiao, S. Y., Yu, Y. J., Wang, X. G. & Shao, T. Y. Cultivated land quality assessment and obstacle factors diagnosis in Changtu County, Northeast China. Land Degrad. Dev.35(17), 5065–5077. 10.1002/ldr.5278 (2024). [Google Scholar]

[CR28] 28.Xiang, J. W., Han, P. & Chen, W. X. Coordinated development efficiency between cultivated land spatial morphology and agricultural economy in underdeveloped areas in China: Evidence from western Hubei province. J. Geogr. Sci.33(4), 801–822. 10.1007/s11442-023-2107-9 (2023). [Google Scholar]

[CR29] 29.Wu, T., Wu, S., Hu, S. G. & Zhang, Q. Simulating future cultivated land using a localized SSPs-RCPs framework: A case study in Yangtze River Economic Belt. Habitat Int.154, 103210. 10.1016/j.habitatint.2024.103210 (2024). [Google Scholar]

[CR30] 30.Yang, H. et al. Sustainable utilization of cultivated land resources based on “element coupling-function synergy” analytical framework: A case study of Guangdong, China. Land Use Policy146, 107316. 10.1016/j.landusepol.2024.107316 (2024). [Google Scholar]

[CR31] 31.Heckerman, D. Bayesian networks for data mining. Data. Min. Knowl. Disc.1, 79–119. 10.1023/A:1009730122752 (1997). [Google Scholar]

[CR32] 32.Xu, X. L., Liu, J. Y. & Zhuang, D. F. Remote sensing monitoring methods of land use/cover changes in national scale. J. Anhui Agric. Sci.40(4), 2365–2369. 10.13989/j.cnki.0517-6611.2012.04.025 (2012) (in Chinese). [Google Scholar]

[CR33] 33.Ahani, S. & Dadashpoor, H. A review of domains, approaches, methods and indicators in peri-urbanization literature. Habitat. Int.114, 102387. 10.1016/j.habitatint.2021.102387 (2021). [Google Scholar]

[CR34] 34.Peng, J., Hu, Y. N., Liu, Y. X., Ma, J. & Zhao, S. Q. A new approach for urban-rural fringe identification: Integrating impervious surface area and spatial continuous wavelet transform. Landsc. Urban. Plan.175, 72–79. 10.1016/j.landurbplan.2018.03.008 (2018). [Google Scholar]

[CR35] 35.Liang, C. & Zeng, J. Three-dimensional dynamic identification and spatial evolution mechanism of urban-rural fringe: A case study of Xiamen City. Geogr. Res.40(05), 1404–1420 (2021) (in Chinese). [Google Scholar]

[CR36] 36.Kumar, R. et al. Revealing the benefits of entropy weights method for multi-objective optimization in machining operations: A critical review. J. Mater. Res. Technol.10, 1471–1492. 10.1016/j.jmrt.2020.12.114 (2021). [Google Scholar]

[CR37] 37.Wang, J. Y., Wang, S. J. & Zhou, C. S. Quantifying embodied cultivated land-use change and its socioeconomic driving forces in China. Appl. Geogr.137, 102601. 10.1016/j.apgeog.2021.102601 (2021). [Google Scholar]

[CR38] 38.Marzialetti, F., Gamba, P., Sorriso, A. & Carranza, M. L. Monitoring urban expansion by coupling multi-temporal active remote sensing and landscape analysis: Changes in the Metropolitan Area of Cordoba (Argentina) from 2010 to 2021. Remote Sens.15(2), 336. 10.3390/rs15020336 (2023). [Google Scholar]

[CR39] 39.Li, Y. L. et al. Study on transformation trajectory and driving factors of cultivated land in the Yellow River Delta in recent 30 years. China Popul. Resour. Environ.29(9), 136–143. 10.12062/cpre.20190601 (2019) (in Chinese). [Google Scholar]

[CR40] 40.Li, H. H. & Song, W. Spatial transformation of changes in global cultivated land. Sci. Total. Environ.859(1), 160194. 10.1016/j.scitotenv.2022.160194 (2023). [DOI] [PubMed] [Google Scholar]

[CR41] 41.Shen, Y., Gong, J., Li, X. D., Wang, W. W. & Tao, R. Driving mechanism and policy implications of cultivated land use change in Wuhan Metropolitan Area. China Land Sci.37(08), 123–133. 10.11994/zgtdkx.20230804.163543 (2023) (in Chinese). [Google Scholar]

[CR42] 42.Yang, Y. L., Ma, M. G., Tan, C. & Li, W. P. Spatial recognition of the urban-rural fringe of Beijing using DMSP/OLS nighttime light data. Remote Sens.9(11), 1141. 10.3390/rs9111141 (2017). [Google Scholar]

[CR43] 43.Wang, L., Lv, T. G. & Tan, Y. Z. Understanding the characteristics and coupling evolution trends of the non agricultural transformation of “population-land-industry”: A case study in Nanchang City. Environ. Dev. Sustain.10.1007/S10668-025-06270-9 (2025). [Google Scholar]

[CR44] 44.Meng, B., Wang, X. X., Zhang, Z. F. & Huang, P. Spatio-temporal pattern and driving force evolution of cultivated land occupied by urban expansion in the Chengdu Metropolitan Area. Land.11(9), 1458. 10.3390/land11091458 (2022). [Google Scholar]

[CR45] 45.Cheng, Y. Y., Chen, W. B. & Lai, S. S. Research on precise identification and driving mechanisms of abandoned cultivated land in typical mountainous Areas. China Land Sci.38(12), 93–104. 10.11994/zgtdkx.20241209.092816 (2024) (in Chinese). [Google Scholar]

[CR46] 46.Jin, A. B., Li, H., Wang, X. R. & Wang, Z. Y. Spatiotemporal assessment of ecological quality and driving mechanisms in the Beijing metropolitan area. Sci. Rep.15(1), 13136–13136. 10.1038/S41598-025-97156-3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Xiong, Y. G. & Jin, A. B. Spatial gradient effects of landscape pattern on ecological quality along the grand canal. Land14(6), 1310–1310. 10.3390/LAND14061310 (2025). [Google Scholar]

[CR48] 48.Zhang, J. M., Wang, P. H. & Jin, A. B. Multidimensional characteristics of urban green space and its impact in mitigating urban heat Island effects: A case study of Guangzhou. Sci. Rep.15(1), 39959–39959. 10.1038/S41598-025-23773-7 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Study on the driving mechanism of cultivated land change in the urban–rural fringe with Bayesian network modeling

Jianping Wang

Zhenhong Zhu

Meiqiu Chen

Yiguo Zhang

Abstract

Introduction

Research framework

Fig. 1.

Materials and methods

Overview of the study area

Fig. 2.

Data sources and preliminary handling

Methodology

Multidimensional feature index identification model

Table 1.

Analysis method of change trajectory

Bayesian network model

Fig. 3.

Table 2.

Fig. 4.

Results analysis

Identification results of the urban–rural fringe in Nanchang City

Fig. 5.

Table 3.

Fig. 6.

The spatio-temporal change characteristics of cultivated land in the urban–rural fringe of Nanchang City

Variation characteristics of the quantity and structure of cultivated land

Table 4.

Characteristics of the spatial distribution change of cultivated land

Fig. 7.

Analysis of the driving mechanism of cultivated land change in the urban–rural fringe of Nanchang City

Results of sensitivity analysis

Table 5.

Results of diagnostic analysis

Fig. 8.

Discussion

The constructed multidimensional feature index identification model

Spatio-temporal characteristics and driving mechanisms of cultivated land changes in the urban–rural fringe of Nanchang City

Suggestions for cultivated land protection in the urban–rural fringe of Nanchang City

Limitations and future research

Conclusion

Acknowledgements

Author contributions

Funding

Data availability

Declarations

Competing interests

Footnotes

References

Associated Data

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases