Assessment and Spatial-Temporal Evolution Analysis of Land Use Conflict within Urban Spatial Zoning: Case of the Su-Xi-Chang Region

Abstract

As China rapidly urbanizes, land resources tend to deplete. This paper aims to identify and propose a resolution of land use conflicts to promote sustainable land use and coordinate the interaction between humans and the environment in urban areas. The methodology of land use conflict assessment within spatial zoning of the Su–Xi–Chang region was evaluated. Taking into consideration the intensity of human activities and the background condition of the natural environment, we divided the study area into a few regions. Furthermore, we developed a methodology by calculating landscape complexity, fragility, and stability based on spatial zoning so as to derive the spatio-temporal characteristics of the land use conflict index (LUCI) in the Su–Xi–Chang region. The results indicate the following: (1) According to the urban spatial attribute index (USAI) statistics, we delineate the core, fringe, and suburban zones of the Su–Xi–Chang region, which accounted for 9.61%, 11.06%, and 79.33% of urban agglomerations respectively; (2) from 1990 to 2018, the fragility index (FI) and LUCI showed S-shaped curves, while the complexity (CI) and the stability indices (SI) exhibited minor fluctuations in the fringe and suburb zones; and (3) intensive and severe conflict is concentrated in core areas due to quite intense human activities and in fringe areas due to an increased interaction between humans and land, while moderate conflict is mainly found in rural and suburban areas that pose as a natural environmental space. The results can serve as a theoretical framework for an effective identification of the LUCI in an urban agglomeration and coordination of the optimal allocation of regional land resources.

1. Introduction

Within the context of rapid urbanization and industrialization, land use conflicts pose a serious challenge to the sustainability of the land system and regional coordinated development. Since attention is constantly directed towards the construction of an ecological civilization, studies on land use conflict have gradually become a hot topic for scholars [1,2]. In reference to studies on social, resource, eco-environmental, and spatial conflicts [3], land use conflicts are believed to occur when different land use stakeholders pursue incompatible interests or when various land use methods and the environment are contradicting [4]. Due to the development of global urbanization, the identification of land use conflicts has become an important decision-making problem for the urban environment [5,6]. Intense land use conflict is often accompanied by the disharmonious interaction between humans and the environment and the deterioration of environmental quality and ecosystem health, thus increasing ecological risk [7,8]. Therefore, it is useful to develop a feasible model to map land use conflict patterns in urban areas and determine conflict levels within a spatial division of urban agglomerations.

The past decade witnessed a rapid increase in the number of studies about different research perspectives on land use conflict [9,10]. Topics of such studies mainly include the mapping of the land use conflict pattern [11,12], identification and intensity diagnosis [13,14], and evolution and driving mechanisms [8,15]. As for evaluation methods, three of them are mainly involved. Firstly, the multiple criteria evaluation method based on land suitability evaluation is used to identify the land use conflict [16,17,18]. The calculated result is accurate to some extent. However, it is also affected by the selected indicators of collected data. At the same time, the social and economic data and environmental variables used in the process of evaluating land suitability are difficult to accurately quantify for certain scales, such as the convenient living conditions of residents and the cultivated land reclamation conditions. Secondly, some scholars refer to rough numbers to determine land use intensity and evaluate land use conflict levels by calculating the difference in intensity between the actual and most adequate land use [19,20]. This method is suitable for land use conflict identification and intensity measurement of large agricultural and dense forest land, rather than rapidly urbanizing areas. Lastly, other scholars have attempted to measure land use conflicts with a linear calculation model so as to calculate land use conflict in urban areas [21,22,23] while considering the spatial type, spatial structure, and spatial conflict process index of landscape units. Additionally, they tried to enrich the connotation of land use conflict with land use structure, land use transformation and landscape pattern conflicts. Nevertheless, previous studies still have several limitations. The distribution and evolution of land use conflict is rarely investigated within different urbanization levels. Land use conflicts are not only small social phenomena, but also geographical phenomena that show spatial heterogeneity. Therefore, it is crucial to investigate the spatial characteristics of land use conflicts at different levels of urbanization and scientifically guide their regulation and mitigation.

Due to the Chinese economic reform and the rapid growth of China’s social economy, the urbanization process is quickly advancing, and the urban agglomeration is becoming an important growth pole for China’s national economy [24,25]. With the development of urban agglomerations worldwide, studies on land use conflicts in urban agglomerations are gradually being emphasized [26,27]. The Su–Xi–Chang region is a typical representation of a developed economic agglomeration. Rapid advances in industrialization and urbanization cause severe changes in land use coverage [28,29,30,31,32]. Hence, exploring the spatio-temporal pattern of land use conflicts in the Su–Xi–Chang region provides scientific guidance for the implementation of “three space and three Lines” in urban agglomerations and points out to new ideas for resolving the contradiction between culture and nature.

Taking the Su–Xi–Chang region as the area of research, the study mainly aims to: (1) construct a model of urban spatial characteristics based on multivariate logistic regression in order to identify and analyze spatial zoning in the study area; (2) explore the impacts of various urbanization levels on the structures and spatial distribution of land use conflicts; and (3) apply the hot spot analysis to identify the typical conflict zones.

2. The Study Area

The Su–Xi–Chang region (30°46′–32°04′ N, 119°08′–121°15′ E) is an urban agglomeration consisting of three cities: Suzhou, Wuxi, and Changzhou. It is located in the Yangtze River floodplain, where Shanghai city is to its east. The region is comprised of 22 districts and counties (Figure 1) [33]. As the core area of the Yangtze River Delta, Su–Xi–Chang region has flat terrain, a superior geographical position, and a complex ecosystem [34]. Numerous river systems flow through the region, most of which are river channels, especially in the Taihu plain to the east. Not only is the region a typical plain river network area, but it is also a traditional high-yield one, with a large high-yield farmland area. Forests are mainly found in the west Maoshan and southern Tianmu mountains near the periphery of the accumulation. The urban agglomeration has a subtropical monsoon climate with an annual average temperature of 15.3 °C and an annual average rainfall of 1092.4 mm.

In recent years, the Su–Xi–Chang region’s economy has rapidly developed. As one of the three major urban agglomerations in Jiangsu Province, the Su–Xi–Chang metropolitan area accounts for about 17% of the province’s land area and yields about 40% of the GDP and local government revenue. By the end of 2018, more than 22 million permanent residents lived in the region, with an average population density about eight times higher than in China. The proportion of urban land increased from 7.52% to 27.60% from 1990 to 2018. The structure and spatial distribution of regional land use have greatly changed, as well. However, land urbanization and utilization in some regions threaten the construction of an ecological civilization and sustainable development of the province.

3. Materials and Methods

3.1. Research Framework

The research framework of modeling the spatial zoning and the assessment of the LUCI is shown in Figure 2.

Step 1: Gradually introduce the NTL, GPP, Road maps, and LST data to establish an optimal model for spatial zoning of the Su–Xi–Chang region.

Step 2: Model the spatial zoning using multivariate logistic regression on the basis of 17,446 sample points selected with an equal distance of 500 m.

Step 3: Construct a spatial measurement model of land use conflict by three dimensions: complexity, fragility, and stability.

Step 4: Explore the land use conflicts variation of structure and spatial patterns based on the spatial zoning of the Su–Xi–Chang region.

3.2. Data Collection and Preprocessing

The main data included in this research are the NPP-VIIRS nighttime light (NTL) data, gross primary productivity (GPP) data, road maps data, land surface temperature (LST) data, and land cover data (

Table 1. Data sources and descriptions.

Data Name	Data Type	Time Period	Data Source
Land cover data	Grid	1990, 2000, 2010, 2018	Resource and Environment Science and Data Center
NTL data	Grid	2018	National Oceanic and Atmospheric Administration
GPP data	Grid	2018	United States Geological Survey (USGS)
LST data	Grid	2018	United States Geological Survey (USGS)
Road maps	Vector	2018	National Catalogue Service for Geographic Information

).

Land cover data were used to calculate assessment indexes, which include complexity, fragility, and fragmentation obtained from the Resource and Environment Science and Data Center of the Chinese Academy of Sciences (https://www.resdc.cn, accessed on 31 December 2019). Afterwards, the indexes were validated using nationwide field surveys [35,36]. The vector data from 1990, 2000, 2010, and 2018 were interpreted by Landsat TM and ETM remote sensing data with spatial resolution of 30 m and the interpretation precision reached 93%. According to the LUCC classification system of China and regional characteristics, land use types were divided into six categories: cultivated land, forest land, urban land, grassland, water, and saltern [37].

NPP-VIIRS NTL data from 2018 were downloaded from the National Oceanic and Atmospheric Administration (https://www.ngdc.noaa.gov/eog/viirs/download_ut_mos.html, accessed on 31 December 2018). Data were rigorously processed to remove the effects of clouds [38].

GPP data and LST data from 2018 were obtained from the United States Geological Survey (https://earthexplorer.usgs.gov/, accessed on 31 December 2018). Furthermore, MOD17A2 and MYD11A2 products from the Moderate resolution imaging spectroradiometer (MODIS) of National Aeronautics and Space Administration (NASA) were used, respectively.

Road maps and county administrative boundaries were derived from the database of the National Basic Geographic Information Center (http://www.ngcc.cn/ngcc/, accessed on 31 December 2018). Additionally, socioeconomic data on a county level, including gross domestic production and population density, were collected from the China Statistical Yearbook (1990–2018) published by the Municipal Statistics Bureau of each city. Finally, all spatial data were georeferenced in Albers Conical Equal Area projection depending on the different source data.

3.3. Multivariate Logistic Regression

The logistic regression model is one of the classical machine learning methods and is commonly used in simulating urban expansion [39]. In comparison to the support vector machine and the neural network model, the logistic regression model has great advantages in training and recognition time [40]. The driving forces of spatial zoning were selected from previous studies that assessed the independent variables of urban expansion in the context similar to our own [39,40,41]. Three driving forces—the intensity of human activities, the human–environment interaction effect, and the physical geography and ecological backgrounds—are identified and indicted by four factors: NTL, GPP, Road map, and LST. The mathematical expression of multivariate logistic regression is as follows:

$$\lg P=\mathrm{Z}=\alpha +{\displaystyle \sum _{i=1}^4{\beta }_{\mathrm{i}}{X}_{\mathrm{i}}}$$

(1)

$$P=\frac{{\mathrm{e}}^Z}{1+{\mathrm{e}}^Z}$$

(2)

where P is the urban spatial attribute index (USAI), α represents the error value of urban spatial development probability without the selected factors, β_i is the logistic regression coefficient, and X_i represents independent variables.

The NPP-VIIRS Nighttime Light remote sensing image dataset includes Raw Data Records (RDR), Sensor Data Records (SDR), and Environment Data Records (EDR), i.e., three levels of processing data products. The SDR product of the NPP-VIIRS monthly night light remote sensing image dataset from January 2018 to December 2018 was adopted with a spatial resolution of 500 m and sorted into the total brightness data of nighttime light in the Su–Xi–Chang region.

MODIS MOD17A2 is an 8-day synthesis of GPP products. Each MOD17A2 data set represents the average of the corresponding 8-day GPP data. Therefore, the collation formula of the MOD17A2 data set from 2018 is as follows:

$$GP{P}_{2018}=5\times Collec{t}_{46}+{\displaystyle \sum _{i=1}^{45}8\times Collec{t}_{\mathrm{i}}}$$

(3)

where GPP₂₀₁₈ is the total value of GPP from 2018 on each pixel, Collect_i represents the GPP value of each period from phase 1 to 45, and Collect₄₆ is the GPP value for phase 46. Finally, the total value of GPP in the Su–Xi–Chang region for 365 days was calculated.

Roads of various levels in the traffic network were sorted out and weighted as shown in

Table 2. The traffic network information statistical table.

Road	Length (km)	Weight
Expressway	2721.64	9
National highway	1080.08	7
Provincial highway	3417.09	7
Urban main road	9178.48	5
County highway	3174.17	3
Urban road	39,980.66	3
Country road	1531.86	1

. By using the ArcGIS linear density method, the traffic network density of the Su–Xi–Chang region was calculated as follows:

$$\mathit{Road}\mathit{Density}=\frac{L_1\times V_1+L_2\times V_2+\dots +L_{\mathrm{n}}\times V_{\mathrm{n}}}{S_{\mathit{circle}}}$$

(4)

where L₁, L₂, and L_n are the lengths of each linear element in a certain distance region of per space unit respectively. V₁, V₂, and V_n are corresponding weight values, respectively. S_circle represents the area of the space unit.

LST data were processed using the MODIS Conversion Toolkit (MCTK) by ENVI so as to carry out the coordinate system transformation and numerical sorting. Since the MYD11A2 dataset takes 8 days and each period of the MYD11A2 dataset is the average value of the LST data, the collation formula of the MOD17A2 dataset from 2018 is:

$$LS{T}_{2018}=\frac{5\times Collec{t}_{46}+{\displaystyle \sum _{\mathrm{i}=1}^{45}8\times Collec{t}_i}}{365}$$

(5)

where LST₂₀₁₈ is the average value of LST from 2018 on each pixel, Collect_i represents the LST value of each period from phase 1 to 45, Collect₄₆ is the LST value for phase 46. Finally, the average value of GPP in the Su–Xi–Chang region for 365 days was calculated.

To obtain the model samples of multiple logistic regression analysis in the study area, the normalization results of various factors and the distribution of the urban spatial types (urban land = 1, non-urban land = 0) were taken as independent and dependent variables respectively. A total of 17,446 sample points were evenly selected with an equal distance of 500 m, and the values of each data were selected as independent variables. Finally, multivariate logistic regression analysis was conducted in IBM SPSS Statistics 24.

3.4. Measuring Land Use Conflicts

Land use conflict and the landscape ecological risk are closely related, so the evaluation dimensions of the two can correspond to each other [8,42]. Based on existing studies [8,13,23,43], we constructed the spatial conflict index (LUCI) to measure the regional and use conflicts while considering three dimensions, i.e., complexity (risk sources), fragility (risk receptors), and stability (risk effects) of the land system.

From the perspective of the evaluation dimension of land use conflicts, the effective diagnosis and evaluation of land use conflict require a characterization of pressure from human disturbance and the deterioration of natural conditions. Likewise, it demands an evaluation of the ability of land resources themselves to withstand the pressure of conflicts and an indication of the stability of the land system [23]. The pressure from conflict reflects the impact and pressure of human activities on the land system and can be represented by the landscape complexity index. The areas with lower intensity of landscape disturbance are less pressed by human demands, so the corresponding conflict pressure is smaller. The intensity of land use conflict is also related to the attributes of land resources, which can be expressed through the vulnerability of the landscape. Different types of land resource use have different functions and their ability to withstand conflicts is different as well. The stability of land use under conflict stress is also an important aspect of land use conflict assessment. The more fragmented the landscape, the poorer the stability of landscape components and the more intense the conflicts. Areas of increased landscape fragmentation, complexity, and vulnerability instigate ecological risk and land use conflict.

Therefore, the complexity, vulnerability, and fragmentation of the landscape were considered as three indices that reflect the risk sources, risk receptors, and risk effects of the ecological risk, respectively. The mathematical expression can be expressed as follows:

$$LUCI=CI+FI-SI$$

(6)

where the LUCI is the index of land use spatial conflict; the CI, FI, and SI are the spatial complexity, spatial fragility, and spatial stability indices respectively.

Firstly, landscape complexity is an effective indicator of risk sources. It is used to reflect the spatial complexity of landscape patches and the interference degree of neighborhood landscape on the current spatial landscape units [44] which are defined by the area-weighted average patch fractal index [45]. The equation for the CI can be represented as:

$$CI={\displaystyle \sum _{i=1}^m{\displaystyle \sum _{j=1}^n\left[\frac{2\ln \left(0.25P_{ij}\right)}{\ln \left(a_{ij}\right)}\left(\frac{a_{ij}}{A}\right)\right]}}$$

(7)

where P_ij is the patch perimeter; a_ij is the patch area; A is the grid size of the fishnet; m is the total number of patches in the grid; n is the number of land use types.

Next, landscape fragility reflects the exposure status of the evaluation unit and the carrying capacity of risk receptors [46]. There are great differences in the resistance of land use types to external disturbance in different stages, such as maintaining ecosystem stability, protecting biodiversity, and improving the overall structure of the ecosystem [8]. To reflect the difference between the responses of land use spatial units to external pressures and land use processes, the landscape fragility index (FI) is used [23]. The equation for the FI is shown in the following formula:

$$FI={\displaystyle \sum _{i=1}^n{F}_i}\times \frac{a_i}{S}n=6$$

(8)

where F_i is the fragility degree of the land use type i; a_i is the area of various landscapes in a unit; S is the total area of space units in the landscape; n is the total number of spatial land use types. The fragility degree of six land use types is ranked in ascending order: built-up land (1), forest (2), grassland (3), arable land (4), water (5), and unused land (6).

Lastly, landscape stability is a representative indicator of the risk effect. As the main impact of land use conflict on regional landscape patterns, landscape stability fragments landscape patches. Here we defined the landscape fragmentation index (LFI) by patch density to show the degree of landscape stability in a certain region [47]. The stability index (SI) of land use type can be expressed as:

$$SI=1-LFI,LFI=\frac{n_i}{A}$$

(9)

where n_i is the patch number of a landscape unit i. A is the grid size of the fishnet. The greater the patch density, the lower the degree of landscape stability in the region.

Considering the scale of the study area and data availability, a 2.5 km fishnet was selected as the basic spatial analysis unit [46], while the cell size of all raster data was set to be 30 m × 30 m. In order to enable the aggregation of various indices, the calculation results of the LUCI, CI, FI, and SI were normalized to the range from 0 to 1.

4. Results

4.1. Spatial Zoning in the Su–Xi–Chang Region

4.1.1. The Collate Partitioned Data

The year 2018 is selected as the research period in order to avoid inaccuracies caused by inconsistent statistical coverage. Figure 3 represents the spatial distribution of four factors, while Figure 3a is the NPP-VIIRS annual total DN value. Overall, the circular layer characteristic of the center-transition-periphery structure is very prominent. Patches with a large number of high-value nighttime light areas are concentrated in six cities. Additionally, the connection between the patches is strong, exhibiting obvious spatial aggregation characteristics. This distribution coincides with the characteristics of the regional urbanization mode of the urban agglomeration.

The total value of GPP shows that the Su–Xi–Chang region has an excellent environment background. Apart from water and construction land, GPP in most regions is at a medium-high level. The red area represents the high value of GPP and is mainly located in hilly areas in the southwest and west of the urban agglomeration. Forest and grass vegetation at the periphery have a high level of GPP, while water areas have a 0 GPP. The edge area with intertwining cultivated and construction lands is located at the intermediate transition zone. The core area of the urban agglomeration is located in the inner region.

As shown in the map of traffic network density, the spatial distribution of the traffic road network density exhibits a multi-core, indicating that the Su–Xi–Chang region is connected as an accessible whole. In addition, Suzhou, Wuxi, Kunshan, and Changzhou together are seen as the peak traffic area, while road network density in other areas is low. Overall, spatial distribution is hierarchical and extends from the peak area to the lowest area. The blue area represents the area with the lowest traffic network density, namely, the natural ecological one. Medium density represented in yellow is distributed in a ring band around the peak area and has become the transition zone between the core area of the urban agglomeration and the natural ecological area. This “ring band” characteristic conforms to the traditional definition of an urban edge area.

In 2018, the highest annual LST was about 36.46 °C, the lowest was about 18.46 °C, and the average about 20.33 °C in the Su–Xi–Chang region. In combination with the spatial distribution diagram of LST, it is evident that the spatial distribution characteristics of LST in the study area are high in the plains and low in mountains and rivers. In addition, the spatial distribution of LST has formed an enormous and continuous integrated LST patch of high value. The low-value LST areas are concentrated in the hilly and mountainous areas in the southwest and the Yangtze River and Tai Lake. Land cover types in these areas are mainly forest vegetation, rivers, and lakes with high specific heat capacity, less human activity, and less anthropogenic heat emissions. As a result, these factors together form the spatial agglomeration of low-value LST areas.

4.1.2. Spatial Zoning of the Su–Xi–Chang Region

In the spatial multivariate logistic regression model of the Su–Xi–Chang region, the urban spatial attribute index (USAI) is the main indicator of spatial zoning. The Wald index is used to measure the major effects of various factors on urban spatial characteristics. The value 0.05 is taken as the evaluation standard for the confidence of the significance level. Moreover, the p value is chosen as 0.05 to represent the sampling and explain that the 5% chance of the relationship between this factor and the USAI is random. Therefore, the smaller the p value, the higher the credibility degree [48].

According to

Table 3. Output results of the multiple logistic regression model.

Variable	Regression Coefficient	Standard Error	Wald	Degree of Freedom	Significance
NTL (X1)	−25.49	1.74	215.49	1	<0.001
GPP (X2)	6.01	0.48	153.53	1	<0.001
Road (X3)	−10.61	1.13	88.80	1	<0.001
LST (X4)	−5.16	0.56	85.58	1	<0.001
intercept	5.06	0.29	308.41	1	<0.001

, the factors that have the greatest influence on urban spatial characteristics in the Su–Xi–Chang region are NTL and Road. The regression coefficient shows that the factors of nighttime light intensity, LST, and traffic network density are negative. Subsequently, this indicates that the normalization of the evaluation value of these three factors is involved in the promotion of urban spatial characteristics. Moreover, GPP is positive, which indicates that the normalized evaluation value has a negative effect on urban spatial characteristics.

According to results in Table 3, the multiple logistic regression model of the urban space in the Su–Xi–Chang region is:

$$\lg P=5.06-25.49\times X_1+6.01\times X_2-10.61\times X_3-5.16\times X_4$$

(10)

$$P=\frac{1}{1{+\mathrm{e}}^{5.06-25.49\times X_1+6.01\times X_2-10.61\times X_3-5.16\times X_4}}$$

(11)

As shown in Figure 4a, the Raster Calculator in ArcGIS is used to calculate the USAI with the factors of NTL, GPP, LST, and traffic network density. The spatial distribution of the USAI has a significant circle effect. The high-value distribution regions that show the USAI in jujube color are concentrated in Suzhou, Wuxi, Changzhou, and Kunshan. Furthermore, the high-value areas exhibiting urban spatial characteristics is particularly prominent and obvious. The high-value cluster areas of Suzhou and Kunshan have almost connected so as to form an entire high-value patch. Moreover, the median distribution region of the USAI in yellow and light blue is located in the periphery of the high-value distribution region and the zone between several high-value distribution regions. The transitional spatial characteristics are obvious and conform to the characteristics of the transition zone of the edge region. Finally, the distribution of the low value in dark blue is located in the peripheral area and water where the degree of human activity is low, which conforms to the characteristics of the ecological area.

The USAI in the Su–Xi–Chang region is concentrated in the high section. The quantitative change rate of spatial features in the (0.08, 0.09) and (0.42, 0.43) sections exhibit obvious mutation points. Therefore, the boundary thresholds are set to 0.085 and 0.425 respectively. Due to the large number and serious fragmentation of landscape in the Su–Xi–Chang region contradict with the principle of continuity and integrity of the spatial zoning of urban agglomerations, 0.25 km² was taken as the smallest patch unit for the mentioned zoning results. Additionally, patch consolidation was carried out for each fragmented patch (Figure 4b).

Table 4. Area and proportion of each partitioned space.

Type	Area (km2)	Percentage (%)
Core Zone	2113	9.61
Fringe Zone	2432	11.06
Suburb Zone	17,443	79.33
Total	21,988

lists the area and proportion of each space partition.

4.1.3. Model Accuracy of Spatial Zoning

The USAI is used in our study to divide the city into functional zones. We adopted the density of construction land within a certain neighborhood so as to characterize the intensity of human development activities. Theoretically, the higher the intensity of human development activities, the more prominent the urban spatial characteristics of urban agglomerations. In addition, a positive correlation between the intensity and the characteristics is obvious in numerical statistics.

Furthermore, we adopted the construction land ratio (%) of each grid unit through spatial superposition analysis of the 500 m × 500 m grid unit and land use data of the Su–Xi–Chang region from 2018. Based on this, we calculated the ratio of construction land within a 5 × 5 neighborhood of each grid unit. In order to verify the quality of the USAI, 17,446 sample points evenly distributed in the Su–Xi–Chang region were selected with 500 m taken as an equal distance. Using IBM SPSS Statistics 24, we conducted a cubic polynomial regression analysis on the USAI and the intensity of human development activities. Regression results obtained through the calculation of the sample point data are shown in Figure 5.

The intensity of human development activities is presented as X, while the urban spatial characteristic of the USAI is shown as Y. Polynomial regression analysis was performed for both and the obtained fitting model is shown in the following formula.

$$Y=1.786+0.779X-0.023X^2+0.00086X^3$$

(12)

correlation coefficient r² is 0.516, sig = 0.000 < 0.01. Rejecting the nihilistic hypothesis, the regression analysis fitting model has significant nonlinear characteristics. This model shows that the urban spatial characteristics of the Su–Xi–Chang region significantly correlate with the intensity of human development activities, so the fitting effect of the model is applicable. Therefore, the USAI model is suitable for representing the urban spatial characteristics of the urban agglomerations.

4.2. Spatiotemporal Characteristics of the Land Use Conflict

4.2.1. Changes in the Structure of the Land Use Conflict

The land use conflict index (LUCI) is composed of the CI, FI, and SI. Between 1990 and 2018, the FI and LUCI exhibited S-shaped curves of different forms, while the CI and SI presented relatively stable patterns, except in the core zone (Figure 6). From 2000 to 2010, the CI and SI in the core zone experienced significant changes, which is consistent with findings from previous studies [49,50]. The normalized value of CI decreased from 0.4415 to 0.27, while the SI increased from 0.7703 to 0.8742. In the fringe and suburban zones, minor fluctuations of the CI and SI were most noticeable during the study period. The FI and LUCI continued to rise from 1990 to 2010 but began to fall and stabilize from 2010 to 2018.

In our research, the LUCI is divided into four types using the inverted “U” classification method, i.e., mild conflicts [0, 0.35], moderate conflicts [0.35, 0.7], intense conflicts [0.7, 0.9], and severe conflicts [0.9, 1]. Our results indicate that the intensity of the spatial conflict in the Su–Xi–Chang region shows an overall upward trend. However, it is still within the basic controllable range (

Table 5. Types of land use conflict from 1990 to 2018.

Degree	Classification	Unit Number				Proportion/%
		1990	2000	2010	2018	1990	2000	2010	2018
Mild	0.0–0.35	2417	2165	1757	1677	81.24	72.77	59.06	56.37
Moderate	0.35–0.7	434	591	737	794	14.59	19.87	24.77	26.69
Intense	0.7–0.9	101	179	420	438	3.39	6.02	14.12	14.72
Severe	0.9–1.0	23	40	61	68	0.77	1.34	2.05	2.29
Total		2975	2975	2975	2975	100	100	100	100
Average		0.269	0.318	0.407	0.418

). The mean values of LUCI in 1990, 2000, 2010, and 2018 were 0.269, 0318, 0.407, and 0.418, respectively. Although the proportion of mild conflicts showed a decreasing trend, it remained a dominant proportion with values between 56.37% and 81.24% during the study period. The proportion of intense and severe conflicts demonstrated an overall upward trend. Likewise, the number of severe conflicts units increased from 23 in 1990 to 68 in 2018, indicating that the land use conflict in the Su–Xi–Chang region gradually intensified.

4.2.2. Spatial Patterns of the Land Use Conflict

During the study period, the conflicts in the Su–Xi–Chang region have intensified (Figure 7). The spatial distribution characteristics of the conflict types are obviously different. From 1990 to 2018, the spatial conflict index of the fringe zone was high and its level from intense gradually turned to severe, while the conflict level in the core zone was moderate. With the expansion of the core into the fringe zone, its land use conflict type began to change from a moderate to an intense one. Mild conflicts were clustered in two types of areas: water (Tai Lake, Yangcheng Lake, etc.) and large tracts of forest or grass areas. Severe conflicts were scattered in ecotones of arable land, urban land, and water, which were located in the suburban zone around Suzhou–Kunshan and Changzhou–Lianyungang. Due to the rapid expansion of urban areas and the influence of Shanghai, moderate conflict units in Kunshan, Taicang, and Changshu spread rapidly and exhibited the tendency to converge in urban centers of Suzhou, Wuxi, and Changzhou. Overall, land use conflict in the Su–Xi–Chang region generally showed a significant increasing trend, especially in areas closest to urban ones. The higher increase rate of conflict levels in the fringe zone can be explained by the rise in complexity and fragmentation caused by abrupt changes in land use.

In our research, spatial distribution characteristics of LUCI from 1990 to 2018 were obtained by employing the method of Global Moran’ I and hot spot analysis (Figure 8) [51,52]. During this period, the Global Moran’ I of LUCI in the study area was 0.188, 0.181, 0.262, and 0.284 (p < 0.001), indicating that spatial units show a certain tendency to cluster. Hot spot analysis demonstrates that both hot spots and cold spots of the LUCI show the spatial pattern of a cluster and belt agglomeration. On one hand, hot spots present clusters of intense and severe conflict located in two typical conflict zones, i.e., the core zone of built-up areas, including urban ones in Suzhou, Wuxi, Changzhou, and other major cities and the Su–Xi–Chang urban–rural interface, which includes metropolitan suburbs of Suzhou and the districts and counties of Kunshan and Taicang closer to Shanghai. On the other, cold spots represent clusters of mild and moderate conflict. They are concentrated in two types of areas, i.e., the contiguous water and lake areas, including Tai, Yangcheng, and Jingji Lake and forests and grassland where development and their utilization is prohibited, mainly in the southwest of the Su–Xi–Chang region. Overall, land use conflict in the Su–Xi–Chang region shows intensified trends and the intensification or alleviation of conflicts is relatively evenly distributed. However, it is worth mentioning that a ring-shaped mitigating area located between the urban–rural interface of major cities was present over the study period.

5. Discussion

5.1. Analysis of the Results of the LUCI in the Su–Xi–Chang Region

Due to the global development of urban agglomerations, more attention is being drawn to research on land use conflict in urban areas [7,10]. Because the field of land use conflict research is still developing and improving, studies are more concerned about the identification and measurement of the conflicts [9,18,21]. However, they often neglect the relationship between the conflicts and the level of urbanization. Therefore, our research constructs an effective spatial measurement model of land use conflict based on the landscape risk assessment method and tries to implement the urban spatial attribute index so as to validate the identified conflict and its spatial feature. Consequently, it provides a good reference point for coordinating the sustainable use of regional land resources. Similar to previous research results on spatio-temporal changes of coordinated land use allocation and the landscape pattern in urban agglomerations [8,46,53], our model supports the hypothesis that land use conflict is stronger in urban areas with high land use tension and rapid change in the frequency of land use. The results also imply that our model is accurate and applicable to a certain extent. Since land use conflict is quantified by the landscape risk index, we adopt the sampling method of 2500 × 2500 m and combine the requirements of a landscape scale in order to introduce innovative perspectives of spatial statistical analysis of model calculation results. The analysis results not only focus on the degree and spatial distribution of land use conflict, but also demonstrate the spatial dynamic characteristics of land use conflict in more detail through hot spot analysis.

5.2. Limitations and Prospects

Despite the results, our research still has some limitations. Based on the landscape index, the regional land use conflict is weighted and summed in accordance with three perspectives, complexity, fragility, and stability. The operation method is relatively simple, and it effectively reflects the spatial dynamic change characteristics of the land use conflict in urban agglomerations. During the index construction, however, the spatial measurement is only carried out from the perspective of landscape ecology. Factors such as economy and society are not considered. Since land use conflict is not only a spatial conflict, but also a complex social, ecological, and economical system, constructing the evaluation criterion with multiple factors by referring to various mathematical models needs to be further improved [8,54,55,56,57]. Another limitation is that the influence of a sudden government policy on land use conflicts is not taken into consideration. As policies are the catalyst of land use conflicts, we should explore the ways in which best to include policy factors in our models. At the same time, our research expounds on land use conflict from the perspective of spatial statistical analysis. As it is difficult to quantify social and economic data reasonably, our study lacks in major factors that affect land use. In future studies, aim to extensively collect socio-economic variables and geographical characteristics such as the NPP, precipitation, and the average daily temperature at urban and even county levels in order to better reflect the multi-level driving mechanism of the land use conflict.

6. Conclusions and Policy Implications

While including multivariate logistic regression, this study also gradually introduces the NTL, GPP, Road maps, and LST data to establish an optimal model for spatial zoning of the Su–Xi–Chang region. In the context of rapid urbanization and industrialization, we construct a spatial measurement model of land use conflict for rapidly urbanizing areas to quantitatively analyze the spatio-temporal characteristics in Su–Xi–Chang region for the years 1990 to 2018. The findings show that the core, fringe, and suburban zones of the Su–Xi–Chang region accounted for 9.61%, 11.06%, and 79.33% of the urban agglomerations, respectively. In combination with the result of polynomial fitting, the USAI effectively represents the intensity of human activities and identifies the layer structure and spatial partition of urban agglomerations. The LUCI shows an overall adverse trend from 1990 to 2018. Its structure the CI and SI exhibit minor fluctuations in the fringe and suburb zones, while the FI shows S-shaped curves during the study period. Land use conflicts are especially prominent in the fringe zone and areas experiencing rapid urban-rural transformation and terrain transition. Moreover, two typical conflict zones are identified through hotspot analysis, i.e., the Su–Xi–Chang urban–rural interface and the terrain transition areas. Despite the limitations, the results of this study can be applied to countries and regions experiencing similar problems as they can provide scientific guidance to combat land use conflict and help coordinate land management at a national level or local level.

Given the above findings and regional characteristics of the Su–Xi–Chang region, we propose the following policies to promote the rational use of natural land capital and the decoupling of land capital occupation from the urbanization process based on the partition perspective.

(1)

Considering that land use competition is fierce, the rapid urban expansion and frequent changes in land use patterns, such as cultivated land, forests, and grassland, are more likely to cause landscape fragmentation in the fringe zone and surrounding areas of central cities with fragile ecological environments (Wujiang, South of Wuxi City and Huishan). Measures related to the cultivated and forest land protection such as the Grain for Green Project and cropland held-replenish balance should be strictly implemented to improve the carrying capacity of the environment. Likewise, later monitoring and management are crucial for the maintenance of ecological restoration achievements due to the rapid expansion of urban construction land.

(2)

In the core zone that features a higher level of economic development and urbanization, construction land has high comparative advantage over cultivated and ecological ones due to the development stage of China. Therefore, it is necessary to delineate the boundary line of urban development so as to reasonably control the extent of urban expansion.

(3)

At the same time, reasonably controlling the boundary of urban space expansion and optimizing the population structure can give autonomy to the powerful driving effect of talents to the economy instead the overuse of natural resources.

(4)

In the suburban zone, the areas with intense and severe conflict levels are widespread and fragmented due to the high spatial competition between cultivated land and forests or grassland and rapid urban expansion. Our results could well serve the national strategy to better coordinate the ecological, production, and urban spaces through rational division of the basic farmland protection line, urban development boundary, and the ecological protection red line [58]. For ecological lands such as forests, grassland, and water, which have lower competition than other production lands, it is necessary to prevent their use as cultivated and construction lands in order to maintain their ecosystem structure and functions.