Research on the Spatial Pattern and Driving Mechanism of Urban Agglomeration in the Upper Reaches of the Yellow River: A Perspective of Integrated Development

Abstract

As an important economic growth pole and ecological security barrier in western China, the urban agglomeration in the upper Yellow River has significant strategic significance for regional coordinated development. This research combines the exploratory geographic analysis method, stepwise regression analysis and geographic detector model to reveal the spatial pattern and driving mechanism of the linear river valley urban agglomeration and the band-shaped overall pattern of the urban agglomeration in the upper Yellow River from the perspective of integrated development. The research shows that the integrated development of the urban agglomeration in the upper reaches of the Yellow River presents the spatial characteristics of multi-point linkage and point-axis superposition. The core cities such as Lanzhou, Xining, and Yinchuan drive the multi-dimensional linkage of regional economy, transportation, and ecology, and promote the coordinated development of economy and ecology in the region. High-value areas are concentrated in provincial capitals and areas rich in ecological resources, while edge city cities are in urgent need of policy support due to insufficient infrastructure and resource endowment and low level of development. Progressive regression analysis and geographic detector analysis show that seven factors, such as resource endowment, natural conditions and traffic accessibility, are the key driving forces of regional integrated development, and the research results provide theoretical basis and empirical support for promoting the sustainable development of regional ecology and economy in western China.

1. Introduction

As highly integrated spatial and economic entities, urban agglomeration has become crucial platforms for enhancing global competitiveness, promoting coordinated regional development, and achieving sustainable growth [1,2]. With the acceleration of the urbanization process, how to balance economic growth and ecological protection in the region has become the key topic of the current research [2]. The integrated development of urban agglomeration in the upper reaches of the Yellow River is a key strategy to promote regional coordination, ecological security, and economic growth. The Lan-Xi urban agglomeration is increasingly demonstrating its scale advantages and is steadily evolving into a highly integrated metropolitan region centered on Lanzhou and Xining, connected by the Lan-Xi high-speed railway as its central axis [3]. According to the theory of integrated development, which emphasizes the synergistic evolution of economy, ecology, and society under spatial interdependence, coordinated regional development has become the foundation for sustainable growth [4]. The integrated development of the urban agglomeration in the upper reaches of the Yellow River holds significant strategic importance for safeguarding national ecological security, supporting the Belt and Road Initiative, and promoting the modernization of Northwest China. The urban agglomeration in the upper reaches of the Yellow River has two functions: it is not only an important ecological barrier in the Yellow River basin, but also a potential support pole for regional economic growth [5], in the context of national strategies such as the “Belt and Road” initiative and the construction of new land and sea corridors in western China. The integrated development of the urban agglomeration in the upper reaches of the Yellow River has created new opportunities for regional coordination and sustainable growth [6]. However, how to effectively integrate the diversified resources, industries, and ecological functions inside and outside the region, and promote the coordinated development of regions, is still a key problem to be solved [7].

Existing research has explored various aspects of urban agglomeration development, ranging from ecological–economic coordination to infrastructure and resource allocation, but few have addressed their interrelations. Ecological protection in urban agglomeration can not only guarantee environmental quality, but also provide support for long-term economic growth [8]. To explore the relationship between transportation infrastructure and regional development, infrastructure construction is recognized as crucial to promoting regional coordination and economic integration [9]. The study also explores the dynamic coupling relationship between transportation network and urban vitality, and emphasizes the role of transportation infrastructure in promoting the economic integration of urban agglomeration [10]. The issue of resource allocation within urban agglomerations has also attracted widespread attention. Optimizing the allocation of resource elements and realizing their effective flow is considered to be the basis for promoting regional integrated development [11]. Several studies have also demonstrated that the relationship between regional coordination, ecological protection, and industrial integration has become increasingly interrelated [12]. For instance, several studies have examined the spatio-temporal dynamics of resource and environmental carrying capacity [13], the quantitative characteristics of construction land [14], and spatial patterns of water use efficiency through the Human Green Development Index [15].

Although existing studies provide valuable insights into the integrated development of the urban agglomeration in the upper reaches of the Yellow River, most of them analyze ecological, economic, or infrastructural factors separately. What remains insufficiently understood is how these factors interact and jointly influence the integrated development of a loosely connected urban agglomeration in western China. This constitutes the central scientific problem that this study seeks to address. Accordingly, this paper focuses on three specific scientific questions: (1) what is the spatial structure of the upper Yellow River urban agglomeration?, (2) what driving mechanisms determine the degree and pattern of regional integration?, (3) how do economic, ecological, and social subsystems interact to shape coordinated development when interpreted through fusion development theory? By addressing these scientific questions, this study not only deepens the understanding of integrated development dynamics in western China’s emerging urban agglomerations, but also offers insights that are applicable to other ecologically sensitive and spatially fragmented regions worldwide.

2. Materials and Methods

2.1. Research Area

The Yellow River basin originates from the Bayan Kala Mountains in Qinghai Province, China, with a total length of 5464 km and an area of 795,000 square kilometers (including an inland area of 42,000 square kilometers). The main stream of the Yellow River is 3472 km long, with a basin area of 428,000 square kilometers. It is an important ecological barrier and a multi-ethnic integration area in China, but its ecological resource endowment is relatively fragile.

This study refers to the basin division standards of the Yellow River Conservancy Commission and, in combination with natural geographic features and economic linkages, defines the area above Hekou Town in Tuoketuo County as the upper reaches of the Yellow River. Specifically, it includes the Lanzhou–Xining urban agglomeration and the Ningxia Along–Yellow River urban agglomeration. The delineation is based on the guiding documents Development Plan for the Lanzhou–Xining Urban Agglomeration [16] and Development Plan for the Ningxia Along–Yellow River Urban Belt [17]. The study area included 3 provinces, 13 prefecture-level cities, and 52 counties (districts) (Figure 1).

2.2. Data Sources

The statistical data used in this study are from the following yearbooks: 2021 China City Statistical Yearbook, China County Statistical Yearbook, Urban Statistical Yearbook of Gansu, Qinghai, and Ningxia provinces, and urban statistical Yearbook of prefecture-level and county-level cities.

Remote sensing data: 2020 Net Primary Productivity of Visible Infrared Imaging Radiometer Suite (NPP-VIIRS) night light data, from the National Aeronautics and Space Administration (NASA) website. We selected the “stable_lights.avg_vis” data, which was used to determine the night light brightness values. Normalized difference vegetation index (NDVI) data, from MODIS13Q1 remote sensing data of NASA, the data is a 16-day synthetic product, used to calculate the vegetation coverage of the upper reaches of the Yellow River in 2018. Elevation and slope data were obtained from the Resources and Environment Data Cloud of the Chinese Academy of Sciences, with a spatial resolution of 30 m and a time period of 2020. Road data are obtained from the Open Street Map (OSM) platform. Raster data of land use status are from Globe Land30 data set in 2020, and nine data types include cultivated land, woodland, grassland, shrub, land, wetland, water body, tundra, man-made surface, and bare land (

Table 1. List of data sources.

Data Name	Data Accuracy	Data Sources	Time
The data set of net primary productivity of vegetation (NPP)	30 m	The results of the remote sensing survey and assessment of ecological status in China	2000–2021
The data sets of meteorology	—	The Data Center for Environmental Sciences, Chinese Academy of Sciences. http://www.resdc.cn (accessed on 25 January 2025).	2010–2020
The data of net primary productivity of visible infrared imaging radiometer suite (NPP-VIIRS)	500 m	The National Aeronautics and Space Administration. https://www.nasa.gov (accessed on 3 February 2025)	2020
The raster data of land use	30 m	The Data Center for Environmental Sciences, Chinese Academy of Sciences. http://www.resdc.cn (accessed on 23 February 2025)	2020
The data of normalized difference vegetation index (NDVI)	30 m	The Data Center for Environmental Sciences, Chinese Academy of Sciences. http://www.resdc.cn (accessed on 3 March 2025)	2018
The data set of soil	1:106	Soil Science Database, China. http://www.tpdc.ac.cn/zh-hans (accessed on 20 February 2025)	2017
Population density	—	The Data Center for Environmental Sciences, Chinese Academy of Sciences. http://www.resdc.cn (accessed on 19 January 2025)	2020
DEM, slope	30 m	The geographical space data from Computer Network Information Center, Chinese Academy of Science. The website of cloud image. http://www.gscloud.cn (accessed on 6 January 2025)	2020
The distances to the Yellow River and roads	—	Euclidean distance analysis and calculation on ArcGIS 10.8 platform	2020

).

The ecosystem service value was evaluated based on the methodology and findings presented in a previously published study by the author [18,19]. The area with the same attributes in the grid is extracted, and transformed into vector graphics of polygons, and superposition the range of 52 counties (districts) to calculate the ecosystem service value data in the research area.

2.3. Study Methods

2.3.1. Entropy Method

This study uses the entropy value method to measure the integrated development of the urban agglomeration in the upper reaches of the Yellow River.

(1)

Standardization of data. To eliminate the effect of different dimensional metrics and ensure that the normalized data are non-zero, the standardized data are translated with the minimal values. The formula for normalization is as follows [20]:

$$r\left(x_{\mathrm{i}\mathrm{j}}\right)={\displaystyle \frac{x_{ij}-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_j\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left(x_j\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_j\right)}}$$

(1)

where x_ij is the original value of the index data for the i-th sample under the j-th indicator, r_ij is the normalized value of the i-th sample for the j-th indicator, max(x_j) and min(x_j) are the minimum and maximum values of the j-th indicator, respectively. All index values after normalization are in the range [0, 1].

(2)

Calculation of normalized index. For each normalized index, the formula is as follows:

$$r_{ij}=x_{ij}∕{\sum }_{i=1}^n{x}_{ij}$$

(2)

where $r_{ij}$ is the normalized value of the i-th sample under the j-th index, $x_{ij}$ is the original value of the i-th sample under the j-th index, n is the number of samples, ${\sum }_{i=1}^n{x}_{ij}$ is the sum of all values of the j-th index.

(3)

Calculation of the entropy value. The entropy value for each indicator is calculated as follows:

$$e_j={\displaystyle \frac{1}{\mathit{ln}n}}{\sum }_{i=1}^n{p}_{ij}\ln (p_{ij})$$

(3)

where $e_j$ is the entropy value of the j-th index, $p_{ij}=r_{ij}∕{\sum }_{i=1}^n{r}_{ij}$ is the normalized proportion of the i-th sample under the j-th index, n is the number of samples, ln is the natural logarithm.

Note that 0 < $e_j$ < 1, and the entropy value reflects the uncertainty or information content in the index j.

(4)

Calculation of the redundancy. The redundancy of the entropy value is calculated as follows:

$$d_i=1-e_j$$

(4)

where $d_i$ is the redundancy of the j-th index, $e_j$ is the entropy value of the j-th index.

(5)

Calculation of the weight of each index. The weight of each index $w_i$ is calculated as follows:

$$w_j=d_j∕{\sum }_{j=1}^m{d}_j$$

(5)

where w_j is the weight of item j-th index, $d_j$ is the redundancy of the j-th index, m is the total number of indicators.

2.3.2. The Spatial Autocorrelation

(1)

The global spatial autocorrelation analysis method mainly measures the degree of average correlation between cities. The common Moran index (Moran I) measures the spatial correlation of the integrated development of urban agglomeration in the upper reaches of the Yellow River. The specific calculation formula is as follows [21]:

$$I={\displaystyle \frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}={\displaystyle \frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{W}_{ij}{Z}_i{Z}_j}{{\sum }_i{Z}_i^2}}$$

(6)

where n is the total number of cities in the study area, is the observed value of the study index of the i city in the study area, is the spatial weight matrix, (i, j) elements represent the spatial distance difference between cities i and j, and Z is the distance difference. Moran I index reflects spatial autocorrelation: value less than 0 indicates negative spatial autocorrelation, large difference and discretization; value greater than 0 indicates positive spatial autocorrelation, close relationship and strong aggregation; index of 0 indicates no spatial autocorrelation and random distribution. $w_{ij}$ is the element of the spatial weight matrix $w$, which represents the spatial adjacency relationship between cities i and j; $w$ was constructed based on the Contiguity-based spatial weight matrix, meaning that two prefecture-level cities are considered neighbors if they share a common boundary or vertex. The weight matrix was row-standardized to ensure comparability.

(2)

The local spatial autocorrelation analysis method mainly measures the degree of spatial correlation between a certain city and neighboring cities [21]. The spatial correlation of the integrated development of the urban agglomeration in the upper reaches of the Yellow River was measured using Moran’s I.

(3)

In addition, to test the statistical significance and robustness of the spatial clustering results, we referred to the comparative analytical framework of Akbar et al. (2025) [22], which provides a methodological reference for quantitative spatial analysis under complex environmental conditions.

2.3.3. Kernel Density Analysis

Kernel density analysis is a spatial analysis method to estimate the concentration degree of the points with a regular moving square. The calculation formula is as follows [23]:

$$f\left(s\right)={\sum }_{i=1}^n{\displaystyle \frac{1}{h^2}}k\left({\displaystyle \frac{s-c_i}{h}}\right)$$

(7)

where f (s) is the kernel density calculation function of regional spatial position s; h is the distance decay threshold; n is the number of element points whose distance from spatial position s is less than or equal to h; k function is a function of spatial weight.

2.3.4. Stepwise Regression Model

Through stepwise regression analysis, the independent variables with significant effects were screened one by one to establish the best regression equation and study the influence factor. The calculation formula is as follows [24]:

$$y=b_0+b_1{x}_1+b_2{x}_2+\cdots +b_m{x}_m$$

(8)

where y is the dependent variable, x₁, x₂, … x_n is an independent variable, b₀ is a constant term.

2.3.5. Geographic Detector

(1)

Factor detection

The geographic detector reveals the influence of spatial heterogeneity by calculating the explanatory force (q-value) of the driver factor on the target variable, as expressed in Formula [25]:

$$q=1-{\displaystyle \frac{{\sum }_{h=1}^{\mathrm{L}}{N}_h{\sigma }_h^2}{N{\sigma }^2}}$$

(9)

where q is the influence of the spatial connection in the study area; N is the number of cities in the city cluster; N_h is the sample number of sub-level cities; L is the number of secondary cities; σ is the variance of spatial connection in the study area; and σ_h is the variance of spatial connection in the study area. The range of q is [0, 1], the higher the value indicates that the stronger the explanatory power of different independent variables X on the dependent variable Y, and the weaker the opposite.

(2)

Interactive detection

It can quantitatively represent whether the explanatory power of two dependent factors is enhanced or decreased, or independent of each other [24]. The detection method is to calculate the q-values of the two variables X_i and X_j in the study area, q (X_i) and q (X_j), and calculate the q-value, q(X_i∩X_j), and compare them.

In this study, both regression analysis and the GeoDetector model were employed to comprehensively examine the driving mechanisms of regional development. Regression analysis was used to quantify the direction and magnitude of the relationships among variables under parametric assumptions, while the GeoDetector model was applied to detect spatial heterogeneity and assess the explanatory power of each factor without assuming linearity. The combination of these two approaches allows for complementary insights—linking statistical inference with spatial pattern detection—to ensure a more robust and integrated methodological framework.

3. Results

3.1. Establish the Index System

The upper reaches of the Yellow River are not only the ecological barrier of the country, but also the potential support pole of regional economic growth. The construction of the index system should be based on policy orientation, regional characteristics, and scientific principles [26,27], and reflect the core concept of ecological priority and sustainable development. Therefore, this study comprehensively considers the four dimensions of ecology, economy, society, and resources, and constructs a multi-level index system (

Table 2. Evaluation index of the integrated development of urban agglomeration in the upper reaches of the Yellow River.

Target Layer	Criterion Layer	Specific Indicator	Unit	Attribute	Weight
Yellow River Upper Reaches Urban Agglomeration	Resource Endowment	Urban Population Proportion	%	Positive	0.035
		Built-up Area	km2	Positive	0.073
		Arable Land Area	Hectares	Positive	0.010
		Output Intensity per Unit of Built-up Area	Ten thousand yuan/km2	Positive	0.005
		Input Intensity per Unit of Built-up Area	Ten thousand yuan/km2	Positive	0.032
	Economic Development	Secondary and Tertiary Industry Share of GDP	%	Positive	0.023
		Nighttime Lighting Area Proportion	%	Positive	0.073
		Per Capita Regional GDP	Yuan	Positive	0.016
		Total Retail Sales of Consumer Goods	Ten thousand yuan	Positive	0.025
		Urbanization Rate of Permanent Population	%	Positive	0.081
		Fixed Asset Investment	Ten thousand yuan	Positive	0.033
	Social Welfare	Public Financial Expenditure	Ten thousand yuan	Positive	0.020
		Industrial Output Value above Designated Size	Ten thousand yuan	Positive	0.029
		Number of Doctors per 1000 People	People	Positive	0.012
		Number of Teachers per 10,000 People	People	Positive	0.015
		Per Capita Disposable Income of Urban Residents	Yuan	Positive	0.024
		Per Capita Disposable Income of Rural Residents	Yuan	Positive	0.192
		Social Security Level	-	Positive	0.045
	Ecological Environment	Per Capita Water Resources	m3/capita	Positive	0.156
		Grassland Coverage Rate	%	Positive	0.009
		Wetland Protection Rate	%	Positive	0.030
		Vegetation Coverage	-	Positive	0.030
		Ecosystem Service Value	-	Positive	0.030

), aiming to comprehensively evaluate the integrated development level and synergistic effect of the urban agglomeration in the upper reaches of the Yellow River, and provide theoretical support and practical guidance for regional coordinated development.

3.2. Level of Integrated Development

The calculation results of the entropy weight method (Formulas (1)–(5)) in the upper reaches of the Yellow River are shown in Table 1. The weights of resource endowment, economic development, social people’s livelihood, and ecological environment were 0.156, 0.251, 0.338, and 0.255, respectively. Therefore, the integrated development level of the urban agglomeration in the upper reaches of the Yellow River was evaluated accordingly. We performed a spatial analysis of the results, and Figure 2 shows the final results. It is divided into five categories by the natural fracture method.

The results show that the upper reaches of the Yellow River show the spatial characteristics of “multi-point linkage” and “point-axis” superposition. The three provincial capitals—Lanzhou, Xining, and Yinchuan—have played a leading role in promoting regional economic and ecological development and driving growth in the upper reaches of the Yellow River. High-level cities such as Republic County and Haiyan County promote green development through the efficient use of ecological resources. However, some remote cities, such as Tongren City and Xunhua County, still face the challenges of weak economic foundation and insufficient infrastructure. In terms of resource endowment, the upper reaches of the Yellow River have formed a stable agricultural industrial economic belt in the core areas of Gansu, Ningxia, and Qinghai provinces, promoting the combination of modern agriculture and ecological protection. High-level cities improve the efficiency of resource utilization through agricultural resources and ecological advantages. Higher level cities such as Hualong County and Gonghe County represent the development mode of circular agriculture and animal husbandry in the surrounding area of Qinghai, which further expands the influence of areas with high-resource endowment. The level of economic development has been significantly improved, and the provincial capitals Lanzhou, Yinchuan, and Xining have become the core of the multi-center economic pattern. Gradually through the radiation effect of Lanbai metropolitan circle, Yinchuan and Xining economic circle through the central city, promoting the linkage and integrated development of surrounding cities. However, some remote areas, such as Dongxiang County and Hualong County, are still weak in transportation and industrial foundation in a relatively backward state. In terms of people’s livelihood, the upper reaches of the Yellow River present a multi-core development pattern. High-level cities are mainly concentrated in the provincial capitals of Lanzhou, Yinchuan, and Xining and their surrounding areas. The number of cities of higher level accounted for 26.92%, forming a people’s livelihood development belt along the upper reaches of the Yellow River. However, some remote areas in Qinghai, especially Haiyan and Republican counties, still face great challenges in terms of infrastructure and social security. There are obvious spatial differences in the ecological environment level. High-level cities such as Republic County and Haiyan County rely on high-quality ecological resources to promote green development. Some medium-level cities, such as Guide, Yongdeng, and Pingluo counties, have achieved good results in ecological protection and resource restoration, forming an ecological restoration belt along the Yellow River. However, the economically developed cities, such as Chengguan District, Anning District, and Qilihe District, face great ecological pressure in the process of resource development and urban construction, and the ecological environment problems are relatively severe.

3.3. Spatial Pattern of Integrated Development

3.3.1. Global Autocorrelation

Using Formula (6), the ArcGIS10.8 was used to calculate Moran’s I index, Z-score, and p value for the integrated development level of the urban agglomeration in the upper reaches of the Yellow River. The results are shown in Figure 3. Moran’s I values for overall development, resource endowment, economic growth, social people’s livelihood, and ecological environment are all positive, and their p value are below 0.1. This indicates a statistically significant global positive spatial autocorrelation at the 90% confidence level.

Resource endowment and economic development: The advantages of resource endowment and economic development are concentrated in core cities such as Lanzhou, Yinchuan, and Xining, and their surrounding areas. These cities exhibit a strong agglomeration effect that promotes efficient resource allocation and regional economic clustering. Economic activities are increasingly concentrated spatially, with nearby cities gradually integrating into the radiation zone of the economic core.

Social people’s livelihood and ecological environment: With the improvement of infrastructure and public services, disparities in living standards among cities have narrowed, forming a more concentrated pattern of social people’s livelihood. In contrast, the ecological environment shows a relatively dispersed spatial distribution. High-value ecological areas—such as Haiyan County and Gonghe County in Qinghai Province—are spatially clustered, but overall distribution remains uneven. Some cities face great ecological pressure, reflecting the imbalance between ecological protection and resource development.

3.3.2. Local Autocorrelation

Based on Moran scatter plot analysis of GeoDA software (version 1.20), the spatial autocorrelation in the upper reaches of the Yellow River showed significant agglomeration distribution characteristics, as shown in Figure 4. The integrated development and resource endowment of the upper reaches of the Yellow River show an obvious trend of spatial agglomeration.

The core cities (Lanzhou, Xining, Yinchuan) and surrounding areas form high-value gathering areas with prominent advantages in resource endowment, which promote the effective allocation and development of resources. Despite the relative concentration of high-value areas, they still show some dispersion overall, and some low-value areas are scattered. Economic activities show the agglomeration effect of provincial capitals, and Lanzhou, Xining, and Yinchuan promote the economic growth of surrounding cities through the radiation effect, forming a centralized economic space model. The ecological environment shows a discrete distribution, and the high-value ecological areas, such as Republican counties and Haiyan counties, are concentrated, but the overall distribution is relatively scattered, reflecting the imbalance between resource development and ecological protection. The social people’s livelihood level shows a gathering trend, and the social welfare of provincial capitals and their surrounding areas is relatively high, while remote cities (such as parts of Qinghai and Gansu) have a large gap in people’s livelihood, which is manifested as scattered spatial distribution.

3.3.3. Kernel Density Distribution Analysis

The agglomeration and distribution characteristics of the upper reaches of the Yellow River are calculated by the kernel density (Formula (7)). Figure 5 shows the use of the natural fracture method to classify the results into five categories.

(1)

Spatial agglomeration and multi-center development. The urban agglomeration in the upper reaches of the Yellow River presents a spatial structure of “multi-center agglomeration”. Lanzhou, Xining, and Yinchuan, as the core cities, promote the integration and agglomeration of the surrounding areas. Through the “point-axis” mode, the core city and the surrounding cities form different linkage effects. The high-density area of Gansu revolves around Lanzhou, Linxia, and Baiyin, forming a diamond-shaped spatial structure. In Qinghai, the high-density area is centered on Xining and presents a linear (“one-shaped”) spatial distribution pattern. The agglomeration area of Ningxia surrounds Yinchuan with a “back character” structure. These high-density areas concentrate resources, economic and social functions, and promote the coordinated development of the urban agglomeration in the upper reaches of the Yellow River.

(2)

Formation of resource endowment and agricultural economic belt. The spatial distribution of resource endowment in the upper reaches of the Yellow River shows a significant agglomeration effect. Datong County, Huzhu County, and Ledu District in Qinghai Province have formed an efficient development area where agriculture and ecological resources gather. Especially under the mode of “integrated planting and breeding”, the output of grain and cash crops has been significantly increased. Relying on the advantages of agricultural industry and resources, Lanzhou, Linxia and Baiyin in Gansu province have promoted the win-win development of modern agriculture and ecological protection, forming a green agricultural economic belt along the Yellow River. The “inverted U-shaped” and “diamond-shaped” structures in Qinghai and Gansu provinces strengthen the agglomeration effect of high-resource endowment areas, improve the efficiency of resource utilization, and promote the construction of a green and efficient agricultural system.

(3)

Economic development and regional integration. The economic development of the urban agglomeration in the upper reaches of the Yellow River shows the strong agglomeration effect of Lanzhou, Xining, and Yinchuan. These core cities drive the economic growth of surrounding areas through a spillover effect, forming a multi-core economic circle centered on the provincial capitals and gradually expanding outward. Lanbai metropolitan area, Yinchuan, Xining economic circle, and other regions, by promoting the city scale economy and radiation effect, promote the integration and coordinated development of regional economy.

(4)

Social people’s livelihood and unbalanced development. The high-density areas of the upper reaches of the urban agglomeration of the Yellow River have formed a multi-center radiation of social people’s livelihood promotion belt, mainly concentrated in provincial capitals and peripheral areas. However, remote cities, such as Haiyan County in Qinghai Province and Dongxiang County in Gansu Province, despite their good resource endowment, still face challenges from infrastructure and social security, which are reflected in the low level of social people’s livelihood and scattered spatial distribution.

(5)

The spatial distribution of the ecological environment is strongly discrete in Xining City, Haidong City, and Huangnan Prefecture, especially in Huangzhong District, Haiyan County, and Guide County. The ecological environment quality has been significantly improved, forming the ecological “green belt” in the upper reaches of the Yellow River. Through the agglomeration effect of ecological restoration zones, these regions not only provide important support for ecological protection, but also ensure the sustainable management of the water source of the Yellow River. The Hehuang Valley and the Pan-Gonghe Basin, as comprehensive treatment areas, have promoted the construction of a green ecological corridor in the Yellow River basin and realized the restoration of the ecological environment. The river ecological belt of the Yellow River is formed in Ningxia and Gansu provinces, connecting Huinong District and Zhongning County, forming the ecological security barrier of Helan Mountain extending to the north and south, providing a key guarantee for the ecological security of the Yellow River basin.

3.4. Influencing Factors

The urban agglomeration in the upper reaches of the Yellow River has unique geographical, ecological, and economic characteristics, faces challenges such as water shortage and ecological fragility, and also has rich agricultural resources and geographical advantages. The integrated development of the region needs to find a balance between promoting economic growth and ecological protection, and make full use of the resource advantages and industrial potential in the region. For this reason, referring to relevant documents [28,29], this study selected key impact factors including natural resource utilization, ecological and environmental protection, economic structure optimization, infrastructure connectivity, and government coordination ability (

Table 3. Index system of influencing factors.

Factor	Order Number	Metric	Unit	Indicator Meaning
Agricultural resource	X1	The area sown to the crops	Ten thousand mu	Reflect the level of resource endowment
Water resource	X2	Water resources	m3	Reflect the water resources level
Green development	X3	Excellent proportion of ambient air quality	%	Reflect the green environment level
Ecological condition	X4	Wetland coverage	%	Reflect the ecological endowment situation
Economic level	X5	Total output value	100 million	Reflect support for economic and social development
Industrial structure	X6	The proportion of secondary and tertiary industries in GDP	%	Reflect the support for industrial development
Infrastructure	X7	Highway network	m	Reflect the convenience of urban connections
Market growth	X8	Total retail sales of social consumer goods	100 million	Reflect the support of the market mechanism
Government capacity	X9	Fiscal expenditure	yuan	It reflects the government’s financial input in the process of integrated development
Land resources	X10	agricultural acreage	Hectare	Reflect the land resource level
External development	X11	External contact intensity	-	Reflect the external radiation capacity
Central location	X12	Provincial administrative area	-	Reflects the development level of the central city
Geo conditions	X13	Interact along the yellow field	-	Reflect the development level of the advantageous cities along the Yellow River

) to deeply analyze how to achieve efficient integration and sustainable development under resource endowment and environmental pressure.

3.4.1. Analysis of the Influencing Factors Based on the Stepwise Regression Model

(1)

Correlation Test

The correlation test results were obtained by using the city group level (y) in the upper reaches of the Yellow River as the dependent variable and 13 influencing factors (X) as the independent variables (

Table 4. Correlation analysis between the integrated development index and its influencing factors.

Metric	Correlation Coefficient	p-Value
X1	−0.492 **	0.000
X2	−0.140	0.323
X3	−0.053	0.71
X4	−0.166	0.239
X5	0.895 ***	0.000
X6	0.397 **	0.004
X7	−0.151	0.287
X8	0.846 **	0.000
X9	0.122	0.388
X10	−0.474 **	0.000
X11	0.810 **	0.000
X12	0.610 **	0.000
X13	0.059	0.678

Note: ***, **, are significant at the 0.001 and 0.01 levels, respectively.

).

The results show that the integrated development level of the upper reaches of the Yellow River was positively correlated with X₅, X₈, X₁₁, X₁₂, and X₆, and the correlation coefficient was 0.895, 0.846, 0.810, 0.610, and 0.397, respectively, and all reached the significance level of 0.01. However, there was a significant negative correlation with X₁ and X₁₀ with correlation coefficients of −0.492 and −0.474, and a significance level of 0.01. There was no significant correlation with X₂, X₃, X₄, X₇, X₉, and X₁₃, and the correlation coefficient was close to 0, and the p-values were all greater than 0.05.

(2)

Stepwise regression analysis

The dependent variables and independent variables that passed the correlation test were sorted out, and the gradual regression analysis was conducted using Statistical Package for the Social Science (SPSS) software (version 19.0) platform to select the regression equations with high goodness of fit and good model interpretation power. The results are shown in

Table 5. Summary of stepwise regression models.

Model	R2	Adjust R2	F	p-Value	Durbin-Watson Value (D-W Value)
1	0.921	0.911	F (6,45) = 87.889	0.000	1.812

.

According to the model treatment, the X₁, X₅, X₈, X₁₁, X₁₃, by explaining 89.3% of the changes in the upper reaches of the Yellow River, the R square value was 0.893, and the model passed the F test (F = 76.466, p = 0.000 < 0.05).

The final model formula of the stepwise regression is y = 0.095 − 0.001 $\times$ X₁ + 0.001 $\times$ X₅ + 0.000 $\times$ X₈ − 0.000 $\times$ X₁₁ + 0.048 $\times$ X₁₂, as shown in

Table 6. Stepwise regression model results.

	Non-Standardized Coefficients		Standardization Coefficient	t	p-Value	Variance Inflation Factor (VIF)
	B	Standard Error	Beta
Constant	0.095	0.019	-	4.93	0.000 **	-
X1	−0.001	0	−0.213	−4.072	0.000 **	1.167
X5	0.001	0	0.606	5.133	0.000 ***	5.963
X8	0	0	0.624	3.072	0.004 **	17.654
X11	0	0	−0.451	−2.171	0.035 *	18.438
X12	0.048	0.021	0.136	2.288	0.027 *	1.521

Note: ***, **, * are significant at the 0.001, 0.01, and 0.05 levels, respectively.

.

The multicollinearity test indicates that while most VIF values are below 10, variables X₈ and X₁₁ exceed this threshold (17.654 and 18.438), revealing potential multicollinearity. Stepwise regression, though effective in identifying key predictors, may introduce overfitting and collinearity risks. Considering both statistical and theoretical criteria, the final model is deemed acceptable, with the D-W value (−2) suggesting no autocorrelation. X₁ and X₁₁ negatively influence the spatial differentiation of the upper Yellow River urban agglomeration, while X₅, X₈, and X₁₂ show positive impacts.

The results show that X₁ has a significant negative effect (β = −0.001, t = −4.072, p = 0.000); X₅, X₈, and X1₂ are 0.001 (p = 0.001), 0.000 (t = 3.072, p = 0.004), and 0.048 (t = 2.0288, p = 0.027); X₁₁ has a significant negative effect (β = −0.000, and t = −2.171, p = 0.035).

3.4.2. Analysis of the Influencing Factors Based on the GeoDetector Model

The level of integrated development was used as the independent variable. Since the raw data are all numerical quantity, the combination of expert scoring method and natural fracture method is adopted to change the type-quantity. The results are summarized below, and Figure 6 presents the factor detection outcomes.

The results show that the highest q-values were observed for X₅ (0.927), X₁₁ (0.871), and X₈ (0.892). Among them, X₅ representing the economic level, serves as the core driving force of integration, playing the most significant role. In the upper reaches of the Yellow River, where water and agricultural resources are abundant, industrial diversification and the improvement of market mechanisms promote factor mobility and optimize resource allocation. X₁₁ (q = 0.871) reflects the degree of openness. Under the framework of the Belt and Road Initiative, it has generated a strong radiation effect, promoting regional cooperation and international development. X₂ (q = 0.727) and X₁ (q = 0.516) provide material support for water-intensive industries and agricultural production, thereby improving the efficiency of water resource utilization. X₇ (q = 0.427) enhances spatial connectivity and transportation mobility between cities. However, X₃ (q = 0.518) and X₄ (q = 0.221) exert relatively weak driving effects. This suggests that balancing the development of resource-intensive industries with ecological sustainability remains a major challenge for the region.

Figure 7 shows the interaction detection results. Results: It can be seen that among all the influencing factors (independent variables) of the integrated development of the upper reaches of the Yellow River, the interaction between the two independent variables and the spatial pattern of the integrated development level is greater than the separate effect of a single independent variable. In particular, the four factors X₂, X₅, X₈, and X₁₁ showed significant enhancement effects under both isolation and interaction.

From the perspective of interaction types, first, there is a strong linear correlation between X₅ and X₁, X₈, X₇, and X₁₁, reflecting the close relationship between Gross Domestic Product (GDP) and economy and transportation. The interaction has an enhancement effect, indicating that the economic aggregate is the key factor to promote the level of urban agglomeration in the upper reaches of the Yellow River, emphasizing the synergistic effect of food security, agricultural ecological modernization, and transportation infrastructure construction, and providing empirical support for policy making. Second, the linear correlation between X₁₁ and X₇, X₁₂, and X₁₁ shows that the traffic highway network promotes the linkage development of regional economy by optimizing the relationship between time efficiency and spatial distance. At the same time, strengthening the economic radiation effect of provincial capital cities can accelerate the expansion of provincial foreign economy, and then improve the overall integrated development level of the upper reaches of the Yellow River urban agglomeration.

3.4.3. Comparison of the Results

Through gradual regression analysis and different methods of geographic detector, the results of the driving mechanism are compared. There are five common independent variables, namely, X₁, X₅, X₈, X₁₁, and X₁₂, which are also the main factors affecting the level change in the urban agglomeration in the upper reaches of the Yellow River. The results of the GeoDetector heterogeneity analysis further indicate that X₂, X₁₃, X₁₀, X₆, and X₇ also serve as important driving factors influencing the level of integrated development in the upper reaches of the Yellow River. These variables act as complementary mechanisms to the main driving factors identified earlier, jointly shaping the regional development pattern. In total, ten key driving factors were identified, which can be classified into seven major categories: resource endowment, natural conditions, ecological–economic dynamics, consumer demand, transportation accessibility, policy impetus, and locational potential (Figure 8).

The integrated development of the urban agglomeration in the upper reaches of the Yellow River is driven by multiple factors. Resource endowment optimizes agricultural industrial structure, promotes spatial differentiation and forming industrial clusters. Natural ecological conditions, especially the allocation of water resources, support land use, ecological protection, and agricultural production. Economic industry promotes gross production, external economy, and industrial structure. Industrial adjustment through consumer demand. Traffic accessibility enhances the urban linkage and agglomeration effect through highway network construction. Policy and location pattern promote the interaction of provincial administrative regions and central cities along the Yellow River. These driving factors cooperate to promote regional integration and economic and efficient allocation, providing theoretical support and empirical basis.

4. Discussion

(1)

This study reveals the spatial structure of “multi-core, axial, and ribbon” organization in the upper reaches of the Yellow River, highlighting a development model led by Lanzhou as the core and Xining and Yinchuan as sub-centers. Unlike the compact polycentric systems of the Yangtze River Delta or the Beijing–Tianjin–Hebei region [30], this configuration demonstrates a looser but ecologically interdependent structure. Such differentiation suggests that integration here is driven not only by economic agglomeration but also by ecological constraints and resource complementarity, expanding the theoretical understanding of coordinated regional development.

(2)

Compared with previous studies emphasizing transportation and industrial specialization as integration drivers [31,32], this study identifies a dual mechanism of ecological interdependence and economic linkage. The Lanbai metropolitan area functions as an economic growth pole, while ecological corridors such as the Pan-Republican Basin and Helan Mountain enhance spatial connectivity. This model illustrates how urban agglomeration in the upper reaches of the Yellow River can achieve integration through ecological–economic synergy rather than pure market density, offering a new conceptual framework for sustainable regional development.

(3)

Sub-central cities such as Xining and Yinchuan serve as mediating hubs balancing ecological protection and socioeconomic growth. This pattern contrasts with coastal agglomerations dominated by economic efficiency, suggesting that integration under ecological constraints requires coordination between environmental carrying capacity and industrial upgrading. From an integration perspective, ecological and economic systems in inland areas can evolve jointly rather than competitively.

(4)

Policy implications. Economic level (X₅, q = 0.927), openness (X₁₁, q = 0.871), and innovation capacity (X₈, q = 0.892) are key drivers. Policies should promote industrial upgrading, innovation, and openness in core cities. Resource endowment (X₁, q = 0.516) and water resources (X₂, q = 0.727) call for integrating ecological corridors such as Helan Mountain and the Pan-Republican Basin into planning. Transportation (X₇, q = 0.427) and location (X₁₂) highlight the need to improve infrastructure in peripheral areas.

(5)

Overall, the findings support innovation-led, eco-coordinated, and well-connected regional development. This case study also offers insights for other ecologically constrained and spatially dispersed urban regions worldwide, showing that integrated development can be achieved through ecological–economic coupling, corridor-based spatial linkages, and coordinated governance approaches. Nevertheless, the study has limitations: it relies on regional-scale data that may not fully reflect micro-level interactions, employs spatial models that emphasize static patterns, and applies a theoretical framework that could be complemented by more dynamic or multi-scalar analytical tools. Future research should incorporate finer-scale ecological and socioeconomic data, utilize dynamic modeling to explore long-term integration trajectories, and conduct comparative analyses across different global ecological regions to further test the applicability of the mechanisms identified in this study.

5. Conclusions

This study aimed to explore the spatial structure, influencing factors, and driving mechanisms of the integrated development of the urban agglomeration in the upper reaches of the Yellow River, by combining spatial autocorrelation analysis, regression analysis, and the GeoDetector model. The results provide empirical insights into the spatial coordination and sustainable development of northwestern China.

(1)

The integrated development level of the urban agglomeration in the upper reaches of the Yellow River exhibits a spatial pattern characterized by “multi-point linkage” and “point–axis superposition,” reflecting economic, ecological, and social agglomeration. High-value areas are concentrated in Lanzhou, Xining, Yinchuan, and ecologically rich regions such as Gonghe and Haiyan. Marginal areas such as Tongren, Xunhua, and Jishishan require targeted policy and infrastructure support.

(2)

The spatial structure exhibits a “multi-center and axial agglomeration” pattern. Provincial capitals serve as regional growth poles, forming industrial, transport, and ecological linkages through radiation effects. Lanzhou drives Dingxi and Baiyin, Xining extends through the Hehuang Valley, and Yinchuan develops along the Yellow River corridor. The Pan-Republican Basin and Helan Mountain corridors act as key ecological and green development nodes.

(3)

Regression and GeoDetector analyses jointly identify the key driving factors of integrated development. Regression results highlight the importance of economic and ecological variables, while GeoDetector results emphasize natural conditions, resource endowment, and transport accessibility. Together, they reveal the complementary roles of economic growth and ecological protection.

(4)

Overall, integrated development is driven by seven potential forces: resource endowment, natural conditions, ecological–economic linkages, consumer demand, transportation accessibility, policy support, and locational advantage. These factors jointly enhance regional integration, improve resource allocation efficiency, and strengthen economic connectivity, offering practical insights for regional policy and planning.