“Clearer” or More “Blurred”? The Evolution of Urban–Rural Boundaries Since the Proposal of Urban–Rural Integrated Development: A Case Study of Zhengzhou

Abstract

As China’s urban–rural integrated development strategy advances, traditional urban–rural boundaries are undergoing rapid restructuring. However, it remains unclear whether these boundaries are becoming distinct through factor flow or blurring due to urban–rural functional coupling. To address this, this study examines the dynamic evolution of boundaries across three core dimensions—population, land, and function—to evaluate the efficacy of integration. We employ the Land Continuity Index, POI Diversity Index, and Gaussian Smoothing Index to characterize transitions in land use structure, spatial functional complexity, and population gradients, respectively. Additionally, a comprehensive Urban–Rural Fuzziness Index (URFI) is developed to quantify boundary blurring trends. Results indicate that Zhengzhou’s urban–rural boundaries exhibit a sustained weakening trend. Notably, changes in functional and population dimensions significantly outpace the land dimension, identifying functional urbanization and population mobility as the primary drivers of this blurring. Consequently, the URFI serves as a robust indicator of integration effectiveness. Overall, boundary blurring is not merely an external manifestation of urban expansion but a profound outcome of factor reorganization, spatial optimization, and the reshaping of urban–rural relationships. This study provides a novel quantitative tool for assessing policy effectiveness, offering both theoretical insights and practical implications for understanding urban–rural integrated development.

1. Introduction

Urban–rural integrated development constitutes a critical pathway toward achieving high-quality development and common prosperity. It is intrinsically linked to the modernization of national governance systems and capacities, holding particular practical significance within the Chinese context [1,2]. The urgency of this issue stems from a complex dynamic: while China has long grappled with a persistent “urban–rural dual structure”—characterized by substantial disparities in resource allocation [3,4]—rapid urbanization has simultaneously intensified the flow of factors between these regions, fundamentally reshaping the traditional pattern of “urban–rural opposition” [5]. Consequently, fostering urban–rural integration has become an imperative strategy for narrowing development gaps and stimulating the endogenous growth momentum of rural areas [6].

While urban–rural integrated development has emerged as a prominent research hotspot, current scholarship primarily focuses on institutional pathways, factor mobility mechanisms, spatial planning, and the equalization of public services [7]. Within this integration process, the urban–rural boundary transcends a mere physical demarcation line: it represents a complex interface where social, institutional, cultural, and identity factors intersect [8]. Consequently, the evolution of these boundaries serves not only as a result of adjusting urban–rural relations but also as a direct indicator of the depth and breadth of integration. On the one hand, infrastructure connectivity and policy integration are driving urban and rural spaces toward morphological unity, leading to a blurring of boundaries [9,10]. On the other hand, persistent disparities in lifestyles, social identities, and governance modes may simultaneously render these boundaries more explicit and solidified [11]. Thus, the relative “clarity” or “fuzziness” of urban–rural boundaries is intrinsic to the underlying dynamics of social integration. Nevertheless, systematic research and empirical analysis addressing this specific phenomenon remain insufficient.

Currently, research on urban–rural integrated development focuses on policy paths and mechanisms, factor flow, equalization of public services, regional differences, social identity, and cultural integration [12,13]. Specifically, studies on policies and mechanisms center on the realization paths of urban–rural integration and explore how reasonable policy design promotes the deep integration of urban–rural economy, society, and culture [14]. In this context, extensive studies scrutinize the policy frameworks, implementation strategies, and instruments employed by national and local governments to drive this process [15,16]. Concurrently, investigations into urban–rural factor flow mechanisms focus on the mobility of critical elements—such as labor, capital, and technology—and evaluate their catalytic role in fostering integration [17]. Numerous empirical studies employ data analysis to elucidate the driving mechanisms and impacts of urban–rural factor mobility [18]. Furthermore, the equalization of public services represents a core tenet of urban–rural integrated development [19,20]. Existing literature highlights that long-standing disparities in public service provision constitute a significant bottleneck restricting urban–rural integration [21]. Additionally, the role of regional differences in urban–rural integration is an important research direction [22]. Scholars reveal the impact of unbalanced regional development on the implementation effect of urban–rural integration by comparing the process in different regions [23]. Research shows that the progress of urban–rural integration in eastern China is relatively fast, while western and remote areas face more challenges [15]. Overall, the primary research focus regarding urban–rural integrated development transitions from the initial “how to evaluate urban–rural integrated development” to “how to judge the effects of urban–rural integrated development.” While earlier studies predominantly explore conceptual frameworks and policy paths and mechanisms at a theoretical level, recent scholarship—driven by sustained policy implementation and practical experience—increasingly emphasizes the empirical assessment of integration efficacy [24,25,26].

Conventionally, the urban–rural boundary is conceptualized as a physical demarcation between urban built-up areas and rural regions, primarily highlighting differences in spatial morphology [27]. However, advancing scholarship transcends this limited geographical definition, characterizing the boundary instead as a complex interface that synthesizes land-use patterns, population density, functional zoning, social behaviors, and institutional arrangements [28,29]. Early scholarship predominantly relies on physical indicators—such as land-use types, administrative divisions, and impervious surfaces extracted from remote sensing—to identify urban–rural boundaries [30]. By distinguishing between built-up environments and rural categories like cultivated land or woodland, researchers spatially delineate the urban from the rural [31]. These methodologies, which typically leverage high-resolution imagery and land-use classification datasets, are characterized by their operational simplicity and quantifiable nature [32,33]. Driven by advancements in spatial data acquisition and computational power, researchers increasingly incorporate comprehensive variables and sophisticated techniques into their analyses. One important method is the identification model based on population density. Specifically, this model evaluates the population distribution characteristics of a region through gridded population data. It delineates the boundary between highly agglomerated population areas (urban) and sparse areas (rural) [34,35,36]. In addition, researchers introduce nighttime light data as a proxy variable for urban activity intensity to track the expansion dynamics of urban fringes [37]. Alternatively, other studies approach boundary delineation from a functional perspective. By integrating multi-source data—including land-use, Point of Interest (POI) distributions, traffic accessibility, and industrial types—these approaches employ analytical techniques such as cluster analysis, discriminant models, and machine learning. Consequently, they effectively identify regions characterized by significant disparities in employment, public services, and commuting patterns [28,38,39]. In contrast to traditional approaches, boundary identification based on functional attributes more effectively reveals the heterogeneity between urban and rural areas regarding behavioral patterns and spatial utilization [40].

As urban–rural relationships evolve, the flow of factors—including population, resources, and functions—accelerates between urban areas and rural areas. This process drives profound transformations in the spatial, functional, and structural differences between urban and rural areas [40]. Amidst various observational approaches, the urban–rural boundary stands out as the most direct spatial manifestation of these changing dynamics. Consequently, academia increasingly regards this boundary as a critical metric for evaluating the progress of urban–rural integrated development [41]. Scholars widely recognize that the dynamic evolution of the urban–rural boundary effectively captures the expansion of urban areas into rural hinterlands, the urbanization of rural functions, and the growth of transitional zones where these spaces overlap [42]. Accordingly, an increasing number of studies investigate the boundary’s morphological transformations, positional shifts, and changing spatial characteristics to assess the speed and depth of urban–rural integrated development [43,44]. Existing research employs diverse technical approaches to identify and analyze the dynamics of urban–rural boundaries. Specifically, one strand of literature utilizes remote sensing imagery and land-use data to quantify indicators such as expansion direction, areal change, and edge morphology, thereby assessing the penetration and influence of urban forms on surrounding rural areas [45,46,47,48,49]. Alternatively, other studies incorporate nighttime light and population density data to construct integration intensity indices that characterize the integration status of boundary zones [50,51]. Further advancing this scope, researchers also superimpose multi-source data—including POI distributions, infrastructure coverage, and service accessibility—to model the functional evolution of these regions; this analysis determines whether urban functions expand into rural zones and whether rural areas possess preliminary urban functions [52,53]. Overall, urban–rural boundaries transcend their role as mere technical objects for spatial delineation. Instead, they function as a critical analytical lens for evaluating urban–rural spatial restructuring and the effectiveness of integration.

Drawing upon the existing literature, this study establishes a novel methodological approach to assess the progress of urban–rural integrated development by analyzing the evolution of urban–rural boundaries. Specifically, this study constructs a comprehensive “urban–rural fuzziness” index, incorporating multi-dimensional perspectives of population, land, and function, to dynamically evaluate the efficacy of integration across different periods. Increased boundary fuzziness signifies intensified population mobility, greater continuity in land use, and tighter functional linkages between urban and rural areas, thereby reflecting a higher level of integration effectiveness.

2. Materials and Methods

2.1. Study Area

To rigorously examine the evolutionary trends of urban–rural boundary fuzziness, this study selects Zhengzhou City, Henan Province, as the empirical case (Figure 1). Situated in the geographical center of China, Zhengzhou possesses a vast hinterland and dense rural settlements, serving as a quintessential region where traditional agricultural zones and modern urban areas intersect and integrate [54]. In recent years, rapid population growth and spatial expansion have driven the continuous extension of the urban built-up area into the periphery, progressively blurring the once-distinct urban–rural spatial boundaries. Notably, within specific “junction zones”—such as areas along beltways, subway extensions, and the interface between the main urban district and surrounding counties—distinct phenomena have emerged: significant population migration to fringe villages, the conversion of rural land for construction, and the outward diffusion of urban services. These trends exemplify the specific processes of “urban function sinking” and rural spatial restructuring. Overall, given its spatial structural diversity and the evident characteristics of boundary restructuring during its evolution, Zhengzhou stands as an exemplary case for measuring urban–rural boundary fuzziness and evaluating integration effectiveness, offering significant representativeness and generalizable value.

2.2. Study Data

2.2.1. LandScan Data

Developed and continuously maintained by the Oak Ridge National Laboratory (ORNL), LandScan data stands as one of the most widely adopted high-resolution global population distribution datasets [55]. Operating on a global 1 km spatial resolution raster system, this dataset synthesizes multi-source geographic information—including nighttime lights, road networks, topography, and land cover types—with official national census data to model and estimate global geospatial population patterns. In contrast to traditional static demographic statistics based on administrative divisions, LandScan data offers superior spatial continuity and versatile applicability, thereby more accurately reflecting dynamic population distribution characteristics across different time frames and regions. Characterized by annual updates and global coverage, the dataset demonstrates excellent comparability and timeliness, making it extensively applicable in diverse fields such as population resource assessment, disaster risk analysis, urban planning, public safety, and environmental modeling [56]. For this study, population distribution data are obtained from the official LandScan Global dataset (https://landscan.ornl.gov/) for the years 2013, 2017, 2021, and 2025, with the 2025 data representing the average value from January to November (Figure 2).

2.2.2. Land-Use Data

Land-use data for this study are derived from the China Land Cover Dataset (CLCD), developed by the School of Remote Sensing and Information Engineering at Wuhan University [57] (Figure 3). As one of the preeminent high-quality remote sensing products available, the CLCD is characterized by its nationwide coverage, extensive temporal span, and superior spatial resolution. Consequently, it was widely adopted across various disciplines, including land resource management, ecological monitoring, urban expansion analysis, and land-use change research. Constructed based on 30 m Landsat imagery, the dataset employs an object-oriented classification approach combined with multi-source data fusion strategies to yield a continuous annual record of land cover across China from 1985 to 2025. The classification system encompasses key categories such as cropland, forest, grassland, water bodies, and construction land, thereby accurately capturing the spatiotemporal evolutionary characteristics of China’s land-use structure [58]. Validation results provided by the development team indicate an overall classification accuracy exceeding 85%, confirming the dataset’s substantial academic reliability and practical application value.

2.2.3. POI Data

POI data for this study are acquired from the Amap open platform. As a fundamental resource for characterizing the distribution of urban spatial functions, POI datasets encompass critical attributes such as geographic coordinates, category labels, names, and addresses. These data accurately depict the precise location and configuration of urban facilities and services, thereby supporting broad applications in fields such as urban research, population activity simulation, functional zoning, and traffic behavior modeling [59]. Distinguished by high spatial resolution, rich semantic content, frequent updates, and extensive coverage, Amap POI data are particularly well-suited for analyzing human settlements and spatial structures within the Chinese urban context. To effectively differentiate between urban and rural functions, this study selected specific POI categories, including commercial services, education and research, medical services, public management, transportation facilities, and cultural leisure. To ensure analytical rigor and consistency, all data undergo classification and cleaning based on unified standards [60]. The dataset for Zhengzhou City, covering the years 2013, 2017, 2021, and 2025, is retrieved via the Amap API (https://lbs.amap.com/). The raw dataset initially contained 12,574,123 records; however, following rigorous preprocessing procedures—such as deduplication, coordinate standardization, and classification field matching—a final dataset of 8,546,357 valid records is retained for analysis (Figure 4).

2.3. Methods

2.3.1. Land Continuity Index

In the domain of urban–rural spatial analysis, the urban–rural interface is characteristically defined as a transitional zone of land-use types. Unlike distinct administrative borders, the transition between urban and rural areas lacks sharp delineation, presenting instead a relatively ambiguous or “fuzzy” spatial morphology. To systematically quantify the fuzziness and structural characteristics of this transition zone, we introduce the Land Continuity Index. This methodological approach measures regional spatial continuity and the smoothness of land-use gradients by quantifying the similarity of land-use types between adjacent spatial units. The index effectively captures the degree of similarity between a focal region and its neighborhood, thereby facilitating the identification of areas exhibiting significant land-use heterogeneity—areas that likely constitute the urban–rural transition [61,62]. Furthermore, the calculation of the Land Continuity Index relies on the influence of multiple adjacent regions within the area. By establishing a specific neighborhood range for each focal region $i$, we compute the similarity relative to its surrounding areas to derive a comprehensive and robust Land Continuity Index.

The calculation formula is presented below:

$$C_i={\displaystyle \frac{1}{n}}\sum _{j\in N\left(i\right)}{S}_{i,j}$$

$$S_{i,j}={\displaystyle \frac{|L_i\cap L_j|}{|L_i\cup L_j|}}$$

where $C_i$ is the Land Continuity Index of region $i$, and it measures the overall similarity of land-use types between this region and its adjacent regions. $N\left(i\right)$ represents the neighborhood set of region $i$. Additionally, $S_{i,j}$ is the similarity index between region $i$ and neighbor $j$. We derive this value by calculating the degree of overlap or similarity of land-use types between the two regions. $n$ is the number of neighborhood regions. In the calculation of the Spatial Continuity Index, $L_i$ and $L_j$ respectively represent the set of land-use types in spatial unit $i$ and its neighboring unit $j$. Specifically, this study first uses raster spatial units as the basic analytical units to statistically analyze land-use data within each spatial unit. After deduplication, all land-use categories appearing in the unit form the set $L_i$. For example, if a spatial unit contains three types—construction land, cultivated land, and forest land—its corresponding land-use type set can be expressed as follows:

$$L_i=\{construction land, cultivated land, and forest land\}$$

Building upon this foundation, the set operation $L_i\cap L_j$ represents the number of land-use type categories that spatial unit $i$ and its neighboring unit $j$ both possess, while $L_i\cup L_j$ represents the total number of land-use type categories after merging the two. Therefore, the similarity index $S_{i,j}$ is used to measure the degree of consistency in land-use structure between adjacent spatial units. When the land-use type composition of two spatial units is highly similar, the proportion of the intersection to the union is high, and the $S_{i,j}$ value approaches 1; conversely, it indicates a significant difference in land-use structure and weaker spatial continuity. The neighborhood range $N\left(i\right)$ of spatial unit $i$ is defined based on spatial adjacency relationships. This study adopts a fixed neighborhood method based on spatial topological relationships, defining units that are directly adjacent to spatial unit $i$ along edges or at vertices as its neighboring units, thereby constructing a first-order neighborhood set (Queen adjacency method). This neighborhood definition can adequately account for spatial continuity while avoiding the excessive smoothing effect caused by an overly large neighborhood range, thus more effectively characterizing the transitional features of land-use types in urban–rural fringe areas. The number of neighboring units is denoted as $n$ and is used to normalize the similarity indices of the various adjacent units.

2.3.2. Shannon Index

To systematically quantify the structural disparities in urban and rural functions, this study employs the Shannon Diversity Index to analyze POI data, thereby evaluating the level of functional diversity across different regions [63]. As spatial proxies for urban functions, the distribution and typology of POIs provide critical insights into a region’s service capacity, spatial organization, and degree of functional complexity. Empirically, urban areas are characterized by high-density, heterogeneous POI clusters, whereas rural regions typically exhibit lower densities with monofunctional attributes. By constructing a POI-based diversity index, we can objectively capture the complexity of these functional structures and the transitional nature of the boundary. Specifically, this method quantifies the richness of functional types, serving as a robust indicator for determining urban–rural fuzziness [64]. Unlike approaches limited solely to POI density, the Shannon Index integrates both the “quantity” and “diversity” dimensions, enabling a more holistic characterization of the functional mix and spatial organization within the study area. Specifically, the study area is first divided into regularized spatial units $i$, which are consistent with the spatial units used for the Spatial Continuity Index. Within each spatial unit $i$, the POI data are then statistically analyzed. For a given spatial unit $i$, assume it contains a total of $K_i$ types of POI functional categories. If the number of POI points belonging to the $k$-th category in unit $i$ is $N_{i,k}$, then the proportion of that category can be expressed as follows:

$$p_{i,k}={\displaystyle \frac{N_{i,k}}{\sum _{k=1}^{K_i}{N}_{i,k}}}$$

Based on this, the calculation formula of the Shannon Diversity Index is as follows:

$$H_i=-\sum _{k=1}^{k_i}{p}_{i,k}·\mathrm{l}\mathrm{n}\left(p_{i,k}\right)$$

where $H_i$ represents the level of functional diversity within a spatial unit $i$, and $k_i$ is the total number of POI functional categories within spatial unit $i$. Additionally, $p_{i,k}$ represents the proportion of the $k$-th type of POI within spatial unit $i$, and this value is the quantity of this category of POI divided by the total number of POIs in the region. If the Shannon Index value is higher, it indicates that the functional types of the region are richer. Consequently, the spatial structure is more complex, and the region possesses stronger urban attributes. Conversely, a lower value indicates that the regional function is single, and the area tends towards rural areas or non-built-up areas.

2.3.3. Gaussian Smoothing Index

To identify the spatial transition characteristics in population distribution and subsequently characterize the urban–rural fuzziness, this study introduces the Gaussian Smoothing Index to process LandScan population data. By performing a spatial Gaussian convolution on the original population raster, this method effectively attenuates noise and extracts underlying density trends. Crucially, this process reveals the spatial gradient shifts from urban core areas to rural transition zones [65], thereby serving as a robust spatial proxy for the degree of fuzziness along urban–rural boundaries. The calculation of the Gaussian Smoothing Index relies on a two-dimensional Gaussian kernel function, and it performs weighted average processing on the values of each raster pixel and the pixels in its neighborhood [66].

The specific calculation formula is as follows:

$$GS{I}_{\alpha ,\beta }=\sum _{m=-k}^k\sum _{n=-k}^k{P}_{\alpha +m,\beta +n}·G(m,n,\sigma )$$

where $\alpha$ represents the index of the spatial analysis unit within the study area along the grid row direction, and $\beta$ represents the index along the column direction. Together, they uniquely determine the location of a population raster cell $\left(\alpha ,\beta \right)$. $GS{I}_{\alpha ,\beta }$ denotes the Gaussian smoothing result of the population raster at the location $\left(\alpha ,\beta \right)$, where $\left(\alpha ,\beta \right)$ is the center location of the population raster to be smoothed; $(m,n)$ represents the discrete spatial offset relative to the center cell $\left(\alpha ,\beta \right)$; and $P_{\alpha +m,\beta +n}$ is the population raster value at the absolute spatial coordinates $\left(\alpha +m,\beta +n\right)$. $G(m,n,\sigma )$ is the two-dimensional Gaussian function, defined as follows:

$$G(m,n,\sigma )={\displaystyle \frac{1}{2\pi {\sigma }^2}}\mathrm{e}\mathrm{x}\mathrm{p}(-{\displaystyle \frac{m^2+n^2}{2{\sigma }^2}})$$

Specifically, it should be noted that the mean of this Gaussian function is always located at the origin (0, 0) of the relative coordinate system, rather than at $(m,n)$ or the absolute coordinates of the population raster $\left(\alpha +m,\beta +n\right)$. The variables $m$ and $n$ are used solely to describe the spatial distance of neighboring cells relative to the center cell, and through distance decay weighting, they reflect the strength of their influence on the central location $\left(\mathsf{\alpha },\mathsf{\beta }\right)$. The core idea of the Gaussian Smoothing Index involves converting the population value of each raster into the overall population distribution trend of its neighborhood. This process effectively attenuates local outliers while preserving macroscopic patterns of population gradients. Consequently, significant fluctuations in GSI values signify rapid spatial transitions in population density; such variations typically characterize the urban–rural interface and serve as distinct markers of boundary features.

2.3.4. Urban–Rural Fuzziness Index (URFI)

To comprehensively characterize the spatial fuzziness of the urban–rural boundary from three dimensions—population distribution, land-use structure, and urban functional characteristics—this study constructs a URFI based on the aforementioned Spatial Continuity Index, functional diversity index, and Gaussian Smoothing Index. This index aims to reflect the degree of continuous spatial transition of urban–rural attributes, rather than a simple binary classification result. For each spatial analysis unit $i$, a feature vector is constructed:

$${\mathbf{x}}_i=[C_i^{*},| H_i^{*},| GS{I}_i^{*}]$$

where $C_i^{*}$, $H_i^{*}$, and $GS{I}_i^{*}$ are the normalized results of the Spatial Continuity Index, Shannon Diversity Index, and Gaussian Smoothing Index, respectively, calculated for the spatial unit $i$ in Section 2.3.1, Section 2.3.2 and Section 2.3.3. Since the three types of indicators differ in their units of measurement and value ranges, this study applies the min–max normalization method to all features before model input:

$$X_i^{*}={\displaystyle \frac{X_i-\mathrm{m}\mathrm{i}\mathrm{n}\left(X\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left(X\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(X\right)}}$$

where $X_i$ represents any original indicator value, and $X_i^{*}\in [0, 1]$ is the normalized result. This process ensures that different features participate in model learning within the same numerical interval, avoiding weight bias caused by scale differences.

Different from traditional linear weighting methods, this study does not preset fixed weights for each indicator. Instead, it employs the Feature Tokenizer Transformer (FT-Transformer) model to nonlinearly fuse the aforementioned features. The core function of this model is not a simple weighted summation, but rather to dynamically characterize the relative contributions and interactive relationships of different features in determining urban–rural fuzziness through a self-attention mechanism [67,68].

Specifically, each component in ${\mathrm{x}}_i$ is regarded as a feature token. After linear mapping, it is input into the Transformer network. The multi-head self-attention mechanism then calculates the dependencies between different features and generates a comprehensive feature representation. This process can be formalized as follows:

$${\widehat{y}}_i=f_{\theta }\left({\mathbf{x}}_i\right)=W_0\cdot \mathrm{Transformer}\left({\mathbf{Z}}_i\right)+b_0$$

where ${\mathrm{Z}}_i$ is the representation after feature embedding, $\theta$ is the model parameter, and ${\widehat{y}}_i$ is the model’s comprehensive score for the urban–rural fuzziness of the spatial unit $i$.

It should be noted that this model does not contain explicit, fixed linear weight parameters that are required to sum to 1. The importance of different features is adaptively learned by the model during training. Their “weights” are implicitly contained within the attention coefficients and dynamically change according to the spatial unit and feature combinations.

To give the model’s output a clear physical meaning, this study applies a normalization constraint on the prediction results at the output layer, ensuring that the URFI satisfies the following condition:

$$URF{I}_i={\widehat{y}}_i\in \left[\mathrm{0,1}\right]$$

where $URF{I}_i$ represents the urban–rural fuzziness level of the spatial unit $i$. Its numerical meaning can be interpreted as follows.

When $URF{I}_i\approx 0$, it indicates that the population, land-use, and functional structures all show clear, singular characteristics of either urban or rural areas, signifying a distinct urban–rural boundary.

Conversely, when $URF{I}_i\approx 1$, it means multiple types of features are highly mixed in space, with gentle population gradients, continuous land-use, and composite functional structures, showing significant transitional characteristics between urban and rural areas.

Intermediate values represent varying degrees of continuous urban–rural transitional states.

3. Results

3.1. Analysis of Urban–Rural Boundaries Based on Land-Use Data

Based on CLCD land-use classification data, Figure 5 illustrates the results of the Land Continuity Index for the years 2013, 2017, 2021, and 2025. Throughout this period, the index consistently exhibits a “low-center, high-periphery” spatial distribution pattern. This indicates that the urban core features complex and highly mixed land-use types, whereas the urban periphery displays greater homogeneity and stronger continuity. This distinction arises because the outer fringe areas largely retain contiguous agricultural landscapes or consist of low-density construction land. However, from a temporal perspective, this spatial configuration underwent sustained evolution from 2013 to 2025. Notably, the Land Continuity Index within the urban core exhibited a progressive decline, indicating an intensification of land-use mixing and increasingly complex interfaces between different land types. This trend reflects the deepening refinement and functional compounding of the internal urban spatial structure. Specifically, the intertwining of commercial, residential, and public service functions amplifies land-use heterogeneity, thereby blurring the previously “clear edges” of the land-use structure. Secondly, while urban peripheral areas retain a relatively high Land Continuity Index, they exhibit a distinct trend of contraction over time. In regions originally characterized as typical rural areas or urban–rural transition zones, the expansion of construction land, the extension of infrastructure, and the conversion of rural land to urban functions drive a diversification of land-use types. Consequently, this diversification reduces the continuity index, resulting in an expanding “blurred zone” within the urban–rural spatial structure.

Overall, from 2013 to 2025, land-use boundaries within the urban core gradually become blurred, while the urban–rural transition zones in the periphery continuously expand. These areas display diversified land-use characteristics, manifesting as enhanced fuzziness in typical urban–rural boundary regions. Consequently, variations in the Land Continuity Index clearly demonstrate that Zhengzhou’s urban–rural boundaries are evolving from distinct to blurred. This trend not only reflects integration in terms of spatial morphology but also establishes a foundation for further analyzing the depth of urban–rural integrated development across the dimensions of population distribution and functional structure.

3.2. Analysis of Urban–Rural Boundaries Based on POI Data

Based on POI data classified by functional category, Figure 6 illustrates the spatial distribution of the Shannon Index in Zhengzhou for the years 2013, 2017, 2021, and 2025. Overall, the POI diversity indices across all years exhibit a decreasing gradient from the urban center toward the periphery. Specifically, regions with the highest functional diversity cluster within the main urban area, whereas outer suburban rural areas are dominated by singular living service facilities and demonstrate lower functional complexity. This distribution pattern aligns with the laws of urban spatial organization, reflecting the high agglomeration of urban functions and the dense spatial layout of service facilities. However, time-series analysis reveals an overall upward trend in the POI Diversity Index for Zhengzhou from 2013 to 2025. Both central urban areas and outer peripheral rural areas demonstrate a consistent increase in the number of functional categories and the degree of balance, reflecting the continuous spatial diffusion of urban functional elements. This growth is particularly pronounced in the transition zones connecting urban areas and rural areas, where the functional diversity index exhibits the most significant rise, with certain regions approaching levels observed at the urban fringe.

Secondly, functional typologies within rural areas and the urban–rural fringe exhibit rapid expansion. A significant influx of facilities—encompassing commercial services, education, healthcare, logistics, public administration, and leisure venues—is progressively permeating peripheral rural zones. This trend drives the urbanization of the rural functional system, transforming it from a structure dominated by agricultural production and basic services. Consequently, the enhancement of functional diversity indicates that rural areas are acquiring urban service capabilities, thereby narrowing the functional disparity between urban areas and rural areas. Further scrutiny of spatial evolution characteristics reveals a concentric expansion of high functional diversity spreading from the center outward. The functional composite core, originally situated within the main urban area, now extends toward county centers, transport corridors, industrial parks, and new urban districts. This diffusion dissolves the rigid functional segregation between urban areas and rural areas. As the rural functional structure transitions from a single-function to a composite model, it exhibits distinct characteristics of urban function sinking and rural spatial restructuring, further evidencing that the urban–rural boundary is becoming increasingly blurred.

3.3. Analysis of Urban–Rural Boundaries Based on LandScan Data

Utilizing LandScan population distribution data, Figure 7 presents the spatial population gradients for Zhengzhou in 2013, 2017, 2021, and 2025, obtained through Gaussian smoothing. The Gaussian Smoothing Index captures the macro-level trends in population density transition from urban areas to rural areas. Specifically, a larger magnitude of change indicates a steeper variation in population distribution, denoting a distinct urban–rural boundary. Conversely, a flattening population gradient signifies diminished spatial disparities between urban areas and rural areas, implying that the urban–rural boundary is gradually becoming blurred.

From a holistic spatial perspective, the Gaussian Smoothing Index consistently exhibits a distinct core–periphery gradient across all years, characterized by high values in the city center that gradually diminish toward the outskirts. This distribution highlights a sharp contrast in population density: the central urban area maintains high population agglomeration, whereas the outer suburbs remain relatively sparsely populated. Nevertheless, akin to the Land Continuity Index and the functional diversity index, this spatial configuration displays significant dynamic trends throughout the period from 2013 to 2025. First, the magnitude of population gradient variations in the central urban area has progressively diminished over the years. The Gaussian Smoothing Index within the urban core exhibits a flattening trend, indicating that spatial disparities in intra-urban population density are narrowing. This convergence is driven by two concurrent processes: on one hand, initiatives such as urban renewal, the renovation of aging neighborhoods, and the optimization of public services have fostered a more balanced population distribution within the central city; on the other hand, sustained population growth in adjacent sub-centers and new urban districts has mitigated the sharp, “cliff-like” density contrast between the core and its periphery, thereby causing the previously distinct population boundaries to gradually fade. Secondly, the population gradient within the urban–rural transition zone has shifted from a steep decline to a smoother profile, significantly intensifying boundary fuzziness. In 2013, demographic changes at the periphery of the main urban area were characterized by a precipitous drop, creating a distinct spatial fracture between urban areas and rural areas. However, this gradient has progressively flattened over time; since 2017, improvements in transportation, industrial transfer, and urbanization initiatives have driven population agglomeration in certain rural sectors, effectively urbanizing the demographic structure of the transition zone. By the 2021–2025 period, the distinct breakpoints in the population gradient had largely dissipated, giving way to a continuous, graduated spatial pattern of urban–rural population distribution.

This shift highlights the intensifying diffusive potential of the urban population. Concurrently, rural areas increasingly accommodate the demands of the urban spillover population, emerging as new agglomeration spaces for residence and employment. Ultimately, this evolution signifies that the urban–rural population system is transitioning from a state of binary separation to one of integrated flow.

3.4. Analysis of Urban–Rural Boundaries Based on Urban–Rural Fuzziness

Building upon calculations from the Land Continuity Index, POI Diversity Index, and Population Gaussian Smoothing Index, this study employs the FT-Transformer model to develop a comprehensive index of urban–rural fuzziness. This composite metric integrates three critical elements—land, population, and function—to reflect their convergence within the urban–rural space. In contrast to single-dimensional boundary identification approaches, this comprehensive index detects the evolution of the urban–rural boundary with greater sensitivity and holistic coverage. Therefore, it serves as a crucial indicator for quantifying the spatial implications of urban–rural integrated development.

In this study, urban–rural boundary clarity is not a numerical indicator directly calculated by a separate formula. Instead, it is a comprehensive interpretation concept of the urban–rural boundary state, based on the spatial distribution characteristics of the URFI. The URFI is used to characterize the continuity and mixing degree of urban–rural attributes in space. A higher URFI value indicates that urban and rural characteristics are more difficult to distinguish within that spatial unit, and the spatial transition is more gradual. Correspondingly, “urban–rural boundary clarity” reflects whether the change in urban–rural attributes is rapid and whether the boundary is concentrated during the transition from urban to rural areas. Specifically, when URFI values show steep gradient changes in space and a narrow transition zone between high- and low-value areas, it indicates a rapid transformation of urban–rural attributes and a relatively clear urban–rural boundary. Conversely, when URFI values change continuously over a large spatial area, forming a wide intermediate transition zone, it indicates a higher degree of mixing of urban–rural attributes and a blurred urban–rural boundary. Therefore, in the comparative analysis of this study, the “clarity” of the urban–rural boundary obtained from different data sources (land use, POI, LandScan, and multi-source fusion) essentially reflects the differences in their corresponding indicators’ ability to identify the width of the urban–rural transition zone and gradient change characteristics in space, rather than repeated calculations of the same boundary under different definitions.

A comparative analysis of the trends across four key indicators for the years 2013, 2017, 2021, and 2025 (Figure 8) reveals a consistent directional pattern: the urban–rural boundary has persistently trended towards urban–rural fuzziness throughout the study period. Specifically, the increasing complexity of land-use structures, the diversification of functional types, and the smoothing of population gradients collectively demonstrate that the inherent spatial disparities between urban areas and rural areas are diminishing, leading to a gradual expansion of the boundary transition zone.

However, an analysis of the magnitude of change reveals distinct disparities among the indices. Most notably, the comprehensive index of urban–rural fuzziness exhibits the most significant decline. Driven by the superposition of trends in population flow, land-use transformation, and functional diffusion, this composite index decreases at a rate significantly exceeding that of any single indicator. These findings suggest that the blurring of boundaries is most pronounced when viewed through a multi-dimensional lens, underscoring the structural and comprehensive nature of the ongoing urban–rural integrated development. Following the comprehensive index, the Population Gaussian Smoothing Index exhibits the next most significant downward trend. The spatial smoothing of population gradients manifests distinctly from 2013 to 2025, driven by rapid population spillover into the urban–rural transition zone, which accentuates the transitional nature of the boundary. Consequently, population metrics prove more sensitive than land indicators in capturing the trend of boundary weakening, making this the most prominent single indicator after the composite index. Conversely, the Land Continuity Index records the smallest magnitude of change. Although land-use patterns actively evolve toward composite urban functions, they demonstrate a marked lag compared to population and functional shifts. Since land development remains constrained by planning controls, construction cycles, and policy restrictions, short-term variations are limited, resulting in a relatively slower pace of urban–rural boundary weakening within the land dimension.

A comprehensive analysis across the dimensions of land-use structure, functional diversity, and population spatial distribution reveals a definitive pattern. Since the proposal of the urban–rural integrated development strategy, the urban–rural boundary in Zhengzhou exhibits a significant shift from clarity to blurriness. Specifically, the Land Continuity Index demonstrates that the internal spatial structure becomes increasingly complex and improves in mixing degree. Furthermore, the Shannon Index reveals that urban functions continuously penetrate into rural areas, and urbanization characteristics within rural spaces emerge gradually. Meanwhile, the Population Gaussian Smoothing Index indicates that the urban–rural population gradient flattens rapidly, which significantly narrows the population difference between urban areas and rural areas. Although these three indicators differ in the magnitude of change, they all point to the same trend across the time dimension. Consequently, the original structural distinction between urban areas and rural areas weakens, and the “urban–rural fracture zone” in the traditional sense disappears.

Furthermore, the constructed comprehensive index of urban–rural fuzziness demonstrates that the blurring of the urban–rural boundary is significantly more pronounced when multi-dimensional factors are superimposed, surpassing the trends indicated by any single metric. The comprehensive index exhibits the most substantial decline, suggesting that urban–rural integrated development is no longer confined to unidimensional changes. Rather, it manifests as a systemic coupling of population, land, and function, representing a holistic process of boundary dissolution. Therefore, since the inception of urban–rural integrated development, the urban–rural boundary has not sharpened; rather, under the synergistic influence of multiple factors, it manifests significant fuzziness, weakening, and expansion. This trend signifies a shift in urban–rural regional dynamics from a binary dichotomy to deep integration. Consequently, the distinction between urban areas and rural areas is no longer defined by sharp spatial, demographic, or functional disparities. Instead, the boundary transforms into a multi-dimensionally interwoven, continuous, and highly composite transitional zone, reflecting the fundamental direction of urban–rural integrated development in the new era.

4. Discussion

Existing scholarship on the urban–rural boundary can generally be categorized into two distinct streams. The first focuses on spatial processes, anchored by concepts such as the “urban–rural gradient,” “urban–rural continuum,” and “urban fringe/transition zone,” underscoring the inherent dynamism and fuzziness of these boundaries [69,70,71]. The second centers on institutional and planning perspectives, contextualized by the “urban–rural dual structure” and “urban–rural coordination,” with a primary focus on interface governance and the restructuring of management boundaries [72,73]. While aligning with the overarching consensus that rapid urbanization and urban–rural integrated development progressively blur the distinction between urban and rural areas, this study offers significant theoretical and methodological advancements. Specifically, it diverges from and extends previous work through a novel research perspective, a refined technical framework, and the design of innovative indicators [74].

From a research perspective, extensive literature employs indicators such as population density, nighttime lights, and land use to identify and delineate the boundary between urban areas and rural areas [75]. The primary focus of these studies is to determine the precise location of the dividing line and the spatial extent of the transition zone [76,77]. While theoretical frameworks often emphasize the existence of a continuous gradient or “urban–rural continuum,” they typically remain limited to qualitative assertions regarding its “fuzzy existence” [78]. This study distinguishes itself by elevating the increasing blurriness of the boundary to a core metric for measuring the effectiveness of urban–rural integrated development. By explicitly quantifying this dynamic into a comparative URFI, the research redefines the urban–rural boundary not as a mere subsidiary phenomenon of urban expansion, but as a critical result variable for evaluating integration success—a methodological approach that remains relatively rare in the current domestic and international literature [79]. Domestic and international research on the urban fringe and peri-urban areas broadly concurs that rapid urbanization induces a spatial blurring of the urban–rural boundary [80]. Consequently, the traditional urban–rural dichotomy is increasingly being replaced by an urban–rural continuum and transition zones [81]. The empirical findings of this study align with this consensus, revealing that from 2013 to 2025, Zhengzhou exhibits a marked trend toward urban–rural fuzziness. During this period, the transition zone has continuously expanded, effectively transforming the boundary from a distinct line into a broad band. However, distinguishing itself from the majority of studies that remain limited to descriptive confirmation [82], this research employs a comparative time-series analysis of four distinct indices: land, population, function, and comprehensive urban–rural fuzziness. The findings reveal that the functional and population dimensions are the most sensitive drivers of boundary blurring, exhibiting the most rapid evolution, whereas changes in the land dimension are relatively lagged. Notably, the comprehensive fuzziness index experienced the most substantial decline among all indicators. This leads to the conclusion that the spatial efficacy of urban–rural integrated development can only be accurately characterized through the superposition of multi-dimensional factors [83]. Consequently, the blurring of urban–rural boundaries functions as a comprehensive spatial manifestation of urban–rural integrated development, rather than merely a byproduct of urban sprawl. In this context, this study transcends the simple binary question of whether boundaries are becoming indistinct; instead, it posits that boundary fuzziness itself serves as a critical metric to quantify and evaluate the actual efficacy of urban–rural integrated development policies [84,85].

This study addresses the pivotal question of whether urban–rural boundaries become “clearer” or more “blurred” in the context of urban–rural integrated development. To resolve this, it establishes a multi-dimensional evaluation framework that systematically reflects the implementation effects of integration strategies. Distinct from existing studies that prioritize institutional designs, policy paths, or factor flow processes, this research contributes a multi-dimensional analysis system centered on population, land, and function [86]. This framework systematically characterizes the dynamic evolution of boundaries across spatial structures, functional complexity, and population gradients, thereby overcoming the limitations of traditional single-perspective identification. Crucially, the study proposes the URFI as a novel metric to quantify the clarity—or fuzziness—of boundaries. By capturing the coupling relationships among diverse elements, this index provides a robust, replicable, and generalizable methodology for assessing the efficacy of urban–rural integrated development.

While this study establishes a multi-dimensional evaluation system for urban–rural fuzziness based on population, land, and function, and employs comprehensive indices to reveal boundary evolution trends, certain limitations persist that require further refinement in future research. First, despite the comprehensive nature of the methodological model, it may oversimplify the multifaceted complexity of urban–rural integrated development. Since urban–rural integrated development functions as a complex system driven by policy, industry, transportation, and social behavior, relying solely on spatial data regarding land, population, and function is insufficient to fully capture its dynamic characteristics [87]. Second, the spatial scope of the findings is constrained by the specific selection of the study area. Although Zhengzhou serves as a typical example of rapid urbanization where urban–rural fuzziness is significant, this trend may not be universally representative. In other contexts—such as mountainous counties, areas experiencing depopulation, or rural zones with weak industrial foundations—the evolution of urban–rural boundaries likely demonstrates distinct patterns [88]. Therefore, future research should further validate the theoretical framework established in this study through comparative analyses across multiple cities and diverse regional types.

5. Conclusions

This study constructs an evaluation framework for urban–rural fuzziness across the dimensions of population, land, and function. Utilizing the FT-Transformer to integrate multi-source data, it systematically characterizes the evolutionary trajectory of urban–rural boundaries in Zhengzhou from 2013 to 2025. The core findings reveal a continuous trend of boundary blurring, which serves as direct spatial evidence of urban–rural integrated development. Specifically, the mixing of land use reflected by reduced land continuity, the functional urbanization evidenced by increased POI diversity, and the population diffusion toward urban–rural transition zones characterized by smoothed population gradients consistently support this conclusion. Collectively, these indicators demonstrate that the traditional spatial pattern of a distinct urban–rural fracture is dissolving, marking a transition from clear to blurred boundaries. Second, functional dynamics and population shifts serve as the primary drivers of boundary blurring, whereas land-use transformation exhibits a relative lag. A comparative analysis of the three basic indices reveals that the POI Diversity Index and the Population Gaussian Smoothing Index decline most rapidly. This trend indicates that urban service factors, employment opportunities, and residential populations transcend administrative and morphological boundaries to diffuse toward rural areas ahead of physical changes. Conversely, the Land Continuity Index displays the most gradual rate of change, demonstrating that institutional and physical constraints on land-use adjustment result in the slowest response within the process of urban–rural integrated development. Third, the comprehensive URFI provides the most accurate reflection of the overall effectiveness of urban–rural integrated development. Integrating multi-dimensional factors, this comprehensive index exhibits the most significant magnitude of decline, substantially exceeding that of single-dimension indicators. This demonstrates that the depth and breadth of urban–rural integrated development are fully revealed only when population flow, land use, and functional evolution are considered simultaneously. Consequently, boundary blurring represents a systemic process driven by multi-factor coupling and mutual reinforcement, rather than an isolated manifestation of a single element.

Overall, the blurring of urban–rural boundaries constitutes not merely a spatial phenomenon but a pivotal manifestation of urban–rural integrated development. The significant trend of boundary weakening observed in Zhengzhou from 2013 to 2025 demonstrates that urban function spillover, rural functional restructuring, and bidirectional population flows collectively drive the transformation of urban–rural space from a binary division toward deep integration. Accordingly, this provides a novel quantitative perspective for evaluating the effectiveness of urban–rural integration policies, offering a valuable reference for future coordinated development.