Spatiotemporal Decoupling of Urban Expansion Intensity and Land Use Efficiency in Arid Oasis Agglomerations

Abstract

Rapid and uncoordinated urban expansion in arid oasis city clusters intensifies land use conflicts and ecological pressure, threatening regional sustainability. This study investigates the Urban Agglomeration on the Northern Slopes of the Tianshan Mountains (UANSTM) in Xinjiang, northwestern China—an arid region urban cluster. A multi-source spatial data framework was established to delineate urban built-up areas and to construct land use efficiency (LUE) indicators, thereby facilitating an integrated analysis of the spatial coupling between urban expansion intensity (UEI) and LUE from 2000 to 2020. The results indicate that: (1) The urban built-up area expanded from 322 km² to 1096 km², shifting northward and northwestward, producing fragmented and decentralized patterns; (2) LUE improved but exhibited clear spatial disparities. Core cities like Urumqi showed strong synergy between rapid expansion and rising efficiency, whereas peripheral cities such as Wusu expanded quickly without corresponding efficiency gains, reflecting evident trade-offs; (3) The relationship between UEI and LUE exhibited a nonlinear evolution—trade-offs dominated during 2000–2005, synergy strengthened from 2005 to 2015, and trade-offs resurged again after 2015.These findings reveal the cyclical vulnerability of arid region urbanization and highlight the effectiveness of the proposed framework for diagnosing spatial mismatches and guiding compact, efficiency-oriented urban development toward long-term sustainability.

1. Introduction

Urbanization, as one of the most significant human activities of contemporary times, is profoundly transforming the Earth’s surface and human social structures [1]. According to United Nations statistics, the global urbanization rate has increased from 30% in 1950 to 56.2% in 2020 and is projected to reach 68% by 2050. This process has not only driven population migration from rural to urban areas but has also triggered largescale urban land expansion [2]. Particularly in China, urbanization has progressed exceptionally rapidly, with urban built-up areas expanding by nearly 300% since 2000, while population growth has been relatively slow, forming a phenomenon of “expansion outpacing population growth” that is prevalent globally [3]. This uncontrolled expansion of construction land is often accompanied by inefficient land use, triggering a series of challenges including environmental degradation, food security concerns, and climate change [4]. Inefficient construction land typically manifests as low population density and weak socioeconomic activities, failing to fully leverage agglomeration effects [5]. Therefore, accurately assessing construction land use efficiency and deploying appropriate mitigation strategies is of critical importance [6].

Accurate identification of urban built-up area boundaries is a prerequisite for evaluating land use efficiency [7,8]. Traditional methods rely on administrative boundaries and statistical data, making it difficult to achieve dynamic spatial monitoring [9]. Remote sensing technology, with its large-scale spatial coverage and cost advantages, provides a unique perspective for monitoring urban expansion. Among these technologies, nighttime light (NTL) data, which captures the nighttime light intensity of the Earth’s surface, demonstrates significant advantages in delineating urban spatial morphology and reflecting socioeconomic activity intensity [7,10,11]. Currently, the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP/OLS) and the National Polar-Orbiting Partnership’s Visible Infrared Imaging Radiometer Suite (NPP/VIIRS) are the most widely applied NTL data sources [10].

The main methods for identifying built-up areas using NTL data include threshold segmentation, supervised classification, and edge detection [12,13,14]. However, NTL data suffers from limitations such as light spillover effects and relatively low spatial resolution. To address these issues, the integration of multi-source remote sensing data has been proven to effectively improve extraction accuracy. Early research introduced the Vegetation Adjusted NTL Urban Index (VANUI) [15], which combines NDVI (Normalized Difference Vegetation Index) with NTL data to mitigate saturation effects and enhance spatial variability at the global scale. Subsequently, the Vegetation and Building-Adjusted NTL Urban Index (VBANUI) was developed, performing dual spectral corrections by incorporating NDBI (Normalized Difference Built-up Index) to suppress light spillover in suburban areas while enhancing signal strength in urban cores [16]. However, in economically underdeveloped regions, particularly arid oasis urban agglomerations with sparse vegetation and unique surface characteristics, single spectral correction methods cannot completely eliminate systematic errors in NTL data [17]. To overcome this limitation, this study developed the Synergistically Corrected VBANUI (SCVBANUI), which integrates the Global Urban Boundaries (GUB) product, VBANUI, and statistical yearbook data through three dimensions—spectral correction, spatial constraints, and statistical verification—to significantly improve built-up area extraction accuracy in arid regions.

After identifying urban built-up area boundaries, evaluating LUE becomes crucial for determining whether urban expansion is rational [18]. LUE is a comprehensive indicator measuring construction land resource allocation, economic vitality, and human activity intensity. Traditional research is primarily based on panel data, employing indicators such as per capita built-up area, economic output per unit land, and population density for measurement [19,20]. However, panel data suffers from temporal lag [19]. The development of remote sensing technology provides new approaches to overcome these issues [21]. NTL exhibits a stable positive correlation with population expansion and economic dynamics; land surface temperature data (LST) reflects the urban heat island effect and is closely related to human activity intensity; gridded population (POP) and gross domestic product (GDP) data can directly characterize the spatial distribution of socioeconomic activities. Based on the comprehensive integration of these multi-source remote sensing data, fine-scale spatial expression and temporal dynamic monitoring of urban LUE can be achieved.

Existing research has made significant progress in measuring the relationship between urban expansion and land use efficiency, but most studies employ traditional statistical methods such as correlation analysis [22,23,24], regression models, or coupling coordination degree models, primarily revealing the overall correlation between the two at the macro scale, making it difficult to precisely characterize their spatial heterogeneity and dynamic interaction types. Specifically, coupling coordination degree models focus on measuring the overall coordination level of two systems but cannot distinguish the spatial differences between positive synergy and negative trade-offs; meanwhile, the decoupling index can identify relative change trends but lacks characterization of absolute intensity and spatial patterns [25]. This ambiguity in understanding relationships constrains the formulation of precise, differentiated urban spatial governance policies.

To break through this bottleneck, this study draws upon the mature “trade-off-synergy” analytical framework from ecosystem services research and innovatively applies it to urban systems [26]. The theoretical foundation of this framework is built upon resource competition theory and synergy effect theory: according to resource competition theory, when urban expansion occupies high-quality land resources without bringing corresponding improvements in socioeconomic benefits, a competitive relationship exists in resource allocation between the two, manifesting as a trade-off; according to synergy effect theory, when expansion promotes agglomeration economic effects and improves land output efficiency, a positive feedback mechanism forms between the two, manifesting as synergy [27]. Through spatial autocorrelation analysis and bivariate local spatial association indicators, this framework can precisely identify the interaction types of different geographical units and their evolutionary trajectories, providing spatial targeting basis for precise policy interventions [28,29]. Compared with traditional coupling/decoupling frameworks, this method can simultaneously characterize the directionality, intensity, and spatial heterogeneity of relationships, demonstrating significant methodological advantages.

The UANSTM, located in the arid zone of northwestern China, faces severe ecological constraints, water scarcity, and uneven socioeconomic development. Rapid urbanization in this fragile region has intensified land use conflicts and environmental stress, posing significant challenges to sustainable development. Although previous studies have examined urban expansion and land use dynamics in major Chinese cities, relatively few have focused on arid oasis agglomerations, where ecological fragility and resource limitations make the mechanisms of urban growth distinctive. Therefore, this study takes the UANSTM as a representative case of arid region urban clusters to (1) accurately extract urban built-up area boundaries from 2000 to 2020 using the SCVBANUI method and quantify the spatiotemporal evolution of urban expansion intensity; (2) construct an entropy-weighted LUE index by integrating multi-source remote sensing data (NTL, LST, POP, GDP) to reveal its spatial and temporal patterns; and (3) apply a trade-off–synergy analytical framework to examine the interactions between urban expansion intensity and land use efficiency. This research provides a scientific basis for sustainable urban planning in arid oasis cities and offers a replicable methodological framework for regional land-resource management.

2. Region and Methods

2.1. Study Area

The UANSTM is situated in the heartland of the Eurasian continent, representing an emerging urban agglomeration within the arid inland region of Northwestern China (42°78′–45°59′ N, 84°33′–90°32′ E). The urban agglomeration covers an area of 2.15 × 10⁶ km², constituting 13% of the total land area of Xinjiang. It encompasses 18 key counties and cities, with core urban centers including Urumqi as the regional capital, Karamay as a vital petroleum base, and Shihezi as a major agricultural and industrial hub. In terms of natural conditions, the topography of the urban agglomeration is primarily composed of deserts, Gobi, grasslands, and oases. Notably, deserts cover more than 60% of the total area, marking a significant distinction from other urban agglomerations and representing a characteristic phenomenon unique to arid regions. Based on the data from 2020, the urban population of the UANSTM accounted for 47.22% of the whole of Xinjiang, and its GDP contributes to 55.75% of the region’s total. This area stands out as the most economically developed in Xinjiang. Serving as a crucial overland hub connecting China and Eurasia, it plays an irreplaceable role in facilitating communication and exerting a regional radiating effect. However, due to the UANSTM’s location in the continental interior and the vast surrounding deserts and Gobi, the distances between cities are considerable. Therefore, the urban built-up area accounts for a relatively small proportion of the administrative divisions of within the urban agglomeration (Figure 1). To better illustrate the land use characteristics of the UANSTM, Figure 1 includes aerial photographs (d), (e), and (f), which were taken during field investigations. These subfigures represent key productive, living, and ecological lands within the region, such as urban areas, surrounding agricultural lands, and industrial parks. These images offer valuable visual context to the study, allowing readers to better understand the spatial structure and urbanization patterns of the UANSTM.

As a rare arid urban agglomeration with considerable development potential, the UANSTM will provide a reference for achieving a dynamic relationship between urbanization and Land use efficiency for urban agglomerations in other arid regions worldwide [30].

2.2. Materials and Pre-Processing

Spatial data of urban built-up areas are primarily derived from nighttime light data. The DMSP-OLS and NPP-VIIRS datasets provide nighttime lighting products for the periods 1992–2013 and 2012–present, respectively [31]. However, these two datasets differ in spatial resolution and sensor types, so the raw NTL data cannot be directly used to extract urban built-up areas. Consequently, the raw NTL data must undergo calibration and integration to create a consistent NTL dataset. Initially, a stepwise calibration method is applied to the DMSP/OLS data, which fully utilizes the characteristics of the raw images and minimizes alterations to the original Digital Number (DN) values [32,33]. Next, the monthly NPP/VIIRS data are averaged to obtain the annual composite product. It is important to note that negative NTL values, which usually indicate anomalous image data, were excluded from the analysis [34]. To ensure consistency in spatial resolution, the NPP/VIIRS data (500 m resolution) were resampled using the nearest-neighbor method to match the 1 km resolution of the DMSP/OLS data [35]. After resampling, a cubic polynomial regression mode was applied to fit the non-zero values of the 2012 and 2013 datasets. Finally, the NTL data were corrected by combining NDVI and NDBI data, and reasonable thresholds were determined based on built-up area data from statistical yearbooks and GUB urban boundary data to further extract the urban built-up area boundaries. This method effectively identifies the expansion boundaries of urban built-up areas for each year, providing high-precision spatial data support for analyzing the urbanization dynamics [31].

The GUB dataset is a comprehensive collection of global urban boundary data, offering precise and detailed boundary information for major cities worldwide [36]. Surface temperature information was derived from the Moderate Resolution Imaging Spectroradiometer (MODIS), which provides thermal measurements at 1 km spatial resolution. The analysis employed multi-year median daytime land surface temperature values to ensure representative thermal characterization. Population distribution data was primarily obtained from the Oak Ridge National Laboratory’s LandScan dataset, which provides demographic data at a 1 km spatial resolution. Due to discrepancies between the LandScan population data and GDP data and those reported in the statistical yearbooks, the population and economic data from the yearbooks were used to calibrate the POP and GDP datasets, ensuring closer alignment with actual values [37]. Detailed processing procedures and data validation comparisons results are provided in the Supplementary Materials (Text S1 and Table S1). Socioeconomic metrics were gathered from authoritative governmental sources, specifically the China Statistical Yearbook, China City Yearbook, and Xinjiang Statistical Yearbook. To ensure analytical consistency, all datasets were uniformly resampled to establish a consistent 1 km grid resolution.

Table 1. Summary of the primary data.

Data	Source	Link	Date of Access	Resolution
DMSP-OLS	NOAA’s National Centers for Environmental Information (NCEI)	https://www.ngdc.noaa.gov/eog/download.html	21 May 2024	1 km
NPP-VIIRS	NOAA’s National Centers for Environmental Information (NCEI)	https://www.ngdc.noaa.gov/eog/download.html	21 May 2024	500 m
GUB	World Bank Global Urban Expansion Program	https://datacatalog.worldbank.org/search/dataset/0038272/World-Bank-Official-Boundaries	27 March 2025	1 km
NDVI	Calculated from Landsat imagery via Google Earth Engine	https://code.earthengine.google.com	27 March 2025	30 m
NDBI	Calculated from Landsat imagery via Google Earth Engine	https://code.earthengine.google.com	27 March 2025	30 m
LST	NASA’s LP DAAC (MODIS product MYD11A2.006)	https://lpdaac.usgs.gov/products/myd11a2v006/	28 March 2025	1 km
GDP	Global gridded GDP dataset consistent with the SSPs (DOI: 10.5281/zenodo.5880037)	https://zenodo.org/records/5880037	23 June 2025	1 km
POP	LandScan Global Population Database (Oak Ridge National Laboratory)	https://landscan.ornl.gov	23 June 2025	1 km
DEM	Shuttle Radar Topography Mission (SRTM) via Geospatial Data Cloud	https://www.gscloud.cn	21 June 2025	90 m
Socio-economy statistics data	Xinjiang Statistical Yearbook	/		/

presents a comprehensive overview of all datasets used in this study.

3. Methods

This investigation aims to examine the spatiotemporal characteristics of UEI and LUE across the UANSTM during the 2000–2020 period, while investigating the dynamic interactions between these two variables. The study utilizes corrected nighttime light data (VBANUI) to extract urban built-up areas and construct the UEI index. Furthermore, various geospatial datasets—encompassing VBANUI, LST, POP, and GDP—were integrated to derive the LUE metric. A unified scale comparison of UEI and LUE was conducted, and a trade-off synergy model is developed. By introducing the Trade-off Synergy Intensity Index (TSI), the study aims to provide a more nuanced analysis of the dynamic relationship between these two indicators (Figure 2).

3.1. Urban Expansion Indicators

In this study, we developed an indicator system to quantitatively analyze the expansion characteristics of built-up areas within the UANSTM. To extract the built-up areas, nighttime light data were employed, and thresholds were established based on the government-reported built-up area and GUB urban boundaries, thereby delineating the spatial distribution of built-up areas in the region [38]. Detailed processing procedures and data validation comparisons are provided in the Supplementary Materials (Texts S2 and S3, Table S2 and Figures S1–S5). Subsequently, we comprehensively evaluate the urban expansion process by constructing indicators for expansion intensity and speed. In addition, the evolution of urban morphology during the expansion is quantitatively analyzed using the fractal dimension index and compactness index [39,40,41]. Fractal dimension can reflect the complexity and fragmentation degree of urban space, while the compactness reveals whether urban expansion is intensive and compactness [42]. These methods provide a comprehensive understanding of the spatial characteristics and evolution patterns of urban expansion, offering a scientific basis for sustainable planning and management.

$$\mathsf{\Delta }\mathrm{U}\mathrm{E}{\mathrm{I}}_{\mathrm{i}}={\displaystyle \frac{(\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}2}-\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}1})}{\mathrm{G}\mathrm{Z}\mathrm{V}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}1}}}\times {\displaystyle \frac{1}{\mathrm{T}}}\times 100\%$$

(1)

$$\mathrm{V}\mathrm{i}={\displaystyle \frac{(\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}2}-\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}1})}{\mathrm{T}}}$$

(2)

$${\mathrm{F}}_{\mathrm{D}}={\displaystyle \frac{2\ln (0.25{\mathrm{L}}_{\mathrm{i}})}{\ln (\mathrm{G}\mathrm{Z}\mathrm{V}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}})}}$$

(3)

$${\mathrm{C}}_{\mathrm{I}}={\displaystyle \frac{\sqrt{\mathsf{\pi }\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}}}}{{\mathrm{L}}_{\mathrm{i}}}}$$

(4)

In the equation, $\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}}$ represents the urban built-up area extracted based on nighttime light data at time points $\mathrm{t}2$ and $\mathrm{t}1$, denoted as $\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}2}$ and $\mathrm{S}\mathrm{C}\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i},\mathrm{t}1}$, respectively. $\mathsf{\Delta }\mathrm{U}\mathrm{E}{\mathrm{I}}_{\mathrm{i}}$ indicates the urban expansion intensity between $\mathrm{t}2$ and $\mathrm{t}1$;$\mathrm{L}\mathrm{i}$ represents the perimeter of the built-up area; $\mathrm{T}$ denotes the time interval; $\mathrm{V}\mathrm{i}$ corresponds to the urban expansion rate; ${\mathrm{C}}_{\mathrm{I}}$ measures the compactness of urban expansion; and ${\mathrm{F}}_{\mathrm{D}}$ represents the fractal dimension of the urban form.

3.2. Land Use Efficiency Index

In this study, LUE index was computed using a normalized weighting approach [43]. To ensure the objectivity of the weight coefficients, the entropy weight method (EWM)—a widely accepted technique for weight determination—is employed [44]. Within the framework of the EWM, the entropy value reflects the dispersion of each indicator; a smaller entropy value indicates greater dispersion and, consequently, a higher weight (i.e., influence) of that indicator. Based on these evaluation weights, a normalized weighted integration formula for the LUE index is established [45,46]. Weighting coefficients for individual indicators were computed across five temporal intervals: 2000, 2005, 2010, 2015, and 2020. The results reveal that the variations in weight values across these periods are minimal, differing only in the fourth decimal place, which exerted a negligible overall effect. Therefore, this study adopts fixed weight values rounded to four decimal places for constructing the LUE indicators.

$$\mathrm{L}\mathrm{U}\mathrm{E}=0.25\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}+0.24\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}+0.25\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}+0.26\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}$$

(5)

The $\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}$ represents nighttime light index; the $\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}$ denotes he land surface temperature index; the $\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}$ signifies the GDP index; and the $\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}$ corresponds to population index.

All indicators underwent normalization procedures to mitigate the influence of varying dimensional scales.

$$\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}={\displaystyle \frac{\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}}-\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(6)

$$\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}={\displaystyle \frac{\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{i}}-\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(7)

$$\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}={\displaystyle \frac{\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{i}}-\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(8)

$$\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x}}={\displaystyle \frac{\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{i}}-\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(9)

In the equation, $\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{i}}$, $\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{i}}$, $\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{i}}$, $\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{i}}$ respectively represent the nighttime light index ($\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}\mathrm{I}$), land surface temperature ($\mathrm{L}\mathrm{S}\mathrm{T}$), population ($\mathrm{P}\mathrm{O}\mathrm{P}$), gross domestic product ($\mathrm{G}\mathrm{D}\mathrm{P}$) of a given pixel. The maximum values of $\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}\mathrm{I}$ ($\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$), $\mathrm{L}\mathrm{S}\mathrm{T}$ ($\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$), $\mathrm{G}\mathrm{D}\mathrm{P}$ ($\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$), $\mathrm{P}\mathrm{O}\mathrm{P}$ ($\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$) correspond to the highest values of nighttime light index, land surface temperature, population, gross domestic product data among all pixels in the urban agglomeration of the northern slope of the Tianshan Mountains. Similarly, the minimum values of $\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}\mathrm{I}$ ($\mathrm{V}\mathrm{B}\mathrm{A}\mathrm{N}\mathrm{U}{\mathrm{I}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$), $\mathrm{L}\mathrm{S}\mathrm{T}$ ($\mathrm{L}\mathrm{S}{\mathrm{T}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$), $\mathrm{G}\mathrm{D}\mathrm{P}$ ($\mathrm{G}\mathrm{D}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$), $\mathrm{P}\mathrm{O}\mathrm{P}$ ($\mathrm{P}\mathrm{O}{\mathrm{P}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$) represent the lowest values of these variables within the same region.

3.3. Trade-Offs/Synergies Relationships and Strength Identification

To comprehensively examine the spatial associations between urban expansion intensity and land use efficiency within the UANSTM, this investigation initially conducts a qualitative assessment of trade-off and synergistic interactions between these variables, followed by quantitative relationship evaluation through differential comparative analysis. Throughout the analytical timeframe, synergistic associations are identified when the product of the variable changes is greater than zero, trade-off dynamics occur when the product is less than zero, and negligible correlations are observed when the product equals zero [47,48]. Based on this analysis, we introduce the TSI is introduced to quantify the strength of the relationship during the process of urban development. The TSI ranges from 0 to 1, with higher values representing a greater degree of trade-off/synergy, to facilitate comparison in terms of time and space, the TSI values are divided into three grades by equal interval method, namely low (0–0.33), medium (0.34–0.66), and high (0.67–1). The calculation formula is as follows:

$$LU{E}_{j,t2}-LU{E}_{j,t1}=\Delta LU{E}_j$$

(10)

$$\left\{\begin{array}{cc}\Delta UE{I}_i\times \Delta LU{E}_j>0& \left(\mathrm{Synergies}\right)\\ \Delta UE{I}_i\times \Delta LU{E}_j<0& \left(\mathrm{Trade}\text{-}\mathrm{offs}\right)\end{array}\right.$$

(11)

$$TSI=1-\left|\left|\Delta UE{I}_i\right|-\left|\Delta LU{E}_j\right|\right|$$

(12)

In the equation, $\mathrm{L}\mathrm{U}{\mathrm{E}}_{\mathrm{j},\mathrm{t}2}$, $\mathrm{L}\mathrm{U}{\mathrm{E}}_{\mathrm{j},\mathrm{t}1}$ represents the urban land use efficiency at times $\mathrm{t}2$ and $t1$, with $\mathsf{\Delta }\mathrm{L}\mathrm{U}{\mathrm{E}}_{\mathrm{j}}$ denoting the change in urban land use efficiency for city j between $\mathrm{t}2$ and $\mathrm{t}1$.

4. Results

4.1. Space-Time Variation of Built-Up Areas

The results of the urban built-up area extracted based on nightlight data are resented in Figure 3. The threshold for defining the urban built-up area was determined according to the total built-up area of the UANSTM urban agglomeration. Therefore, in the early stages of the UANSTM urban agglomeration. Consequently, in the early stages of development, several cities did not meet the overall threshold and therefore were not represented spatially. However, with the rapid growth of urbanization, additional cities gradually emerged in later periods.

The urban built-up area of the UANSTM has shown a significant growth trend between 2000 and 2020, particularly in the cities of Karamay (Region I), Urumqi (Region IV), and Shihezi (Region III). According to changes in the built-up area over the years, urbanization in Karamay and Urumqi has been particularly prominent, with Urumqi experiencing a dramatic increase in built-up area by 2020, primarily expanding along major transportation corridors. Karamay displays a more concentrated urban expansion pattern, with construction primarily focused on the core area and gradually extending to surrounding areas. Kuitun-Dushanzi-Wusu (KDW) urban region (Region II) exhibits slower urban growth compared to other cities, although its urbanization process gradually accelerated after 2005, driven by infrastructure development. Shihezi’s urban expansion is relatively balanced, with growth primarily concentrated in the city center and surrounding areas. Overall, the spatial heterogeneity of urbanization among these cities reflects the combined influence of geographical location, resource endowment, and transportation accessibility. As the regional hub, Urumqi exerts a strong radiative effect on the sur-rounding cities, serving as a central driver of urban growth across the UANSTM.

The urban expansion of the UANSTM urban agglomeration between 2000 and 2020 generally occurred in a northwestward direction, with the newly added area also expanding in this direction. The built-up area of the UANSTM increased from 322 km² in 2000 to 1096 km² in 2020, an expansion of 744 km², approximately 3.4 times the original area (Figure 3). As expansion intensity increased, the fractal dimension steadily rose, while compactness continuously declined. This indicates that urban expansion led to the fragmentation of urban forms, making the spatial structure more complex and irregular, with a tendency toward decentralized growth and the continuous outward extension of urban boundaries. Specifically, between 2000 and 2005, the intensity of urban expansion was relatively low, at only 0.87%. This suggests that, during this phase, urban expansion was mainly concentrated on the early stages of infrastructure development, with a relatively slow expansion rate. As urbanization progressed, particularly between 2005 and 2010, the expansion intensity rose sharply to 11.37%, reflecting an accelerated urban growth trend. The high expansion rate during this period (38.2 km²/a) indicates concentrated and rapid urban development. However, from 2010 to 2015, urban expansion intensity decreased to 6.03%, and the expansion rate fell to 31.8 km²/year, indicating a slowdown in urban growth. Despite this slowdown, the fractal dimension increased from 1.57 to 1.69, suggesting that urban expansion became more complex and irregular during this period, with a significant increase in the fragmentation of urban forms. Finally, between 2015 and 2020, the intensity of urban expansion rose again to 11.95%, with an expansion rate increasing to 82.0 km²/a, marking the peak of urban expansion (Figure 3). Both expansion rate and intensity reached new highs during this period, further intensifying the decentralization of urban space and showing a notable characteristic of rapid growth.

4.2. Urban Expansion Intensity Index

Based on the figure (Figure 4), the spatial distribution of Urban Expansion Intensity (UEI) in the UANSTM urban agglomeration shows significant variations across different time periods. The four regions (I, II, III, IV) represent Karamay (Region I), the KDW area (Region II), Shihezi (Region III), and Urumqi (Region IV). The intensity of urban expansion in each region is categorized into three levels: low (green), medium (yellow), and high (red). Each time period depicted in the figure shows the changes in UEI within these regions, reflecting the spatial dynamics of urban growth over time.

Between 2000 and 2020, the urban expansion of the UANSTM exhibited a clear acceleration, particularly in Urumqi and Karamay. The spatial distribution of UEI revealed a shift from low to high expansion intensity, particularly in Urumqi, which experienced the most rapid growth. In the early stages (2000–2005), expansion intensity was relatively low, with most areas displaying medium to low expansion levels, particularly in the KDW region and Shihezi. This reflects the initial phase of infrastructure development, where urban growth was slow. From 2005 to 2010, expansion intensified, especially around Urumqi, where high-intensity growth became prominent. This acceleration continued between 2010 and 2015, although the expansion in Karamay and Shihezi slowed down slightly, with medium-intensity growth dominating these regions. From 2015 to 2020, urban expansion reached its peak, with Urumqi and Karamay experiencing high expansion intensity. This period marked a significant phase of rapid urbanization, characterized by decentralized growth and intensified fragmentation of urban forms. Overall, the urban expansion of the UANSTM urban agglomeration was marked by spatial and temporal variations, with the central cities of Urumqi and Karamay experiencing the most significant and rapid urbanization, while other regions expanded at a relatively slower pace.

4.3. Space-Time Variation of Land Use Efficiency Index

Between 2000 and 2020, the urban expansion area of the UANSTM increased significantly, with an average annual expansion rate of 38.7 km²/year, reflecting the accelerated pace of urbanization. The maximum values of the LUE index for the UANSTM in 2000, 2005, 2010, 2015, and 2020 were 0.58, 0.56, 0.59, 0.54, and 0.566, respectively. The average values were 0.28, 0.27, 0.26, 0.26, and 0.25, and the minimum values were 0.21, 0.21, 0.20, 0.22, and 0.20, respectively. Despite the substantial increase in the expansion area, the average LUE for the UANSTM exhibited a downward trend. This phenomenon is depicted in Figure 5.

The average LUE decreased from 0.28 in 2000 to 0.25 in 2020, a reduction of 10.7%. This indicates that despite the large-scale urban expansion, particularly in Urumqi and Karamay, the newly developed areas were predominantly located in low-density, poorly planned regions, leading to a decline in land use efficiency. Notably, the KDW area experienced rapid urbanization, yet its LUE showed only a modest improvement, highlighting imbalances and inefficiencies in urban planning and resource utilization.

From the spatial analysis, during the 2000–2005 period, the LUE change was relatively small across all regions, particularly in Karamay and the KDW area, where the LUE change primarily occurred in the negative or low positive range. The KDW area, in particular, saw slow LUE improvements, indicating low land use efficiency. Between 2005 and 2010, the LUE change slightly increased, especially in Urumqi, where LUE saw notable improvement. As the core city of the UANSTM, Urumqi’s rapid economic development and infrastructure improvements led to increased land use efficiency. However, the KDW area and Shihezi still exhibited relatively low LUE increases, suggesting that these regions did not achieve significant improvements in land use efficiency despite ongoing urbanization. From 2010 to 2015, Urumqi continued to show positive LUE changes, especially in the city center, where land use efficiency significantly improved. However, the LUE changes in the KDW area and Karamay remained slow. Even though urban construction in these areas accelerated, the LUE increase was limited, indicating low efficiency in land development. From 2015 to 2020, although Urumqi continued to experience some LUE improvements, the overall LUE changes across the region were relatively flat. In particular, the KDW area saw a decline in LUE changes, suggesting that while urban expansion accelerated in this region, land use efficiency did not improve in parallel. This could be attributed to the area’s reliance on low-density expansion and the imbalanced allocation of resources.

In summary, while urban expansion in the UANSTM increased significantly from 2000 to 2020, the overall land use efficiency showed a downward trend, primarily due to the expansion in low-density areas and inefficient urban planning. Despite improvements in cities like Urumqi, other areas like the KDW region and Karamay showed minimal LUE improvements, underscoring the challenges of achieving sustainable urban growth and efficient land use in rapidly urbanizing regions.

4.4. Relationship Between UEI and LUE

4.4.1. Spatial Analysis of the Trade-Off/Synergy Relationships

The spatiotemporal relationship between UEI and LUE in the UANSTM from 2000 to 2020 exhibited pronounced temporal variation and regional heterogeneity (Figure 6). Quantitative statistics reveal four distinct evolutionary stages of the trade-off/synergy relationship.

During 2000–2005, trade-offs dominated the overall pattern, accounting for 71.16%, while synergies represented only 28.84% (Figure 6e). High trade-off (HT) and medium trade-off (MT) zones were primarily concentrated in Regions I (Karamay) and II (KDW), occupying over 60% of the total area. These regions displayed scattered, low-density expansion with weak spatial clustering of synergies. Between 2005–2010, the share of trade-offs dropped to 43.30%, while synergies increased to 56.70%, indicating a 27.86 percentage point improvement in coordination between UEI and LUE. Spatially, synergy clusters expanded substantially in Region IV (Urumqi) and emerged in several parts of Region II. The radar chart (Figure 6f) shows a clear shift from low trade-offs (LT) and low synergies (LS) toward medium and high synergy (MS, HS) levels. From 2010–2015, synergy remained dominant for 56.70%, while trade-offs accounted for 43.30%. HS and MS zones became more contiguous in Regions I and II, and LS patches decreased, suggesting more spatially compact development. However, during 2015–2020, the trend reversed: trade-offs increased to 56.40%, whereas synergies declined to 43.60%. HT zones reappeared in Regions I and II, and LS areas expanded along the urban peripheries of Region IV. Overall, the 20-year period presented a nonlinear fluctuation: an initial trade-off-dominant stage (2000–2005), a synergy-enhancement stage (2005–2015), and a later trade-off resurgence (2015–2020).

4.4.2. Trade-Offs/Synergies Between UEI and LUE

The relationship between UEI and LUE across the 18 cities and counties in the UANSTM during the periods of 2000–2005, 2005–2010, 2010–2015, and 2015–2020 is illustrated in Figure 7. Each of the 18 regions (labeled 1–18) is represented by a scatter plot. Figure 7a–d show the changes in LUE of the urban built-up areas in the UANSTM across the respective periods. Figure 7e–h depict the distribution of UEI and LUE in the UANSTM’s urban built-up areas during these time frames. Finally, Figure 7i–l illustrate the trade-off/synergy relationships between UEI and LUE for each county and city in the UANSTM (Table S3).

Based on the threshold-adjusted analysis of night-time light data, only cities that met the built-up area threshold in each period were included in the evaluation. During 2000–2005, only seven cities met the threshold, with one city (14.3%) exhibiting synergy and six (85.7%) showing trade-offs. These trade-off relationships were mainly observed in Changji (1), Fukang (2), and Kuitun (8), where UEI values were positive but LUE changes remained negative, reflecting inefficient early expansion. In contrast, Urumqi (16) was the only city showing synergy, indicating relatively compact growth and a higher degree of coordination between UEI and LUE even in the initial phase. In 2005–2010, the number of observable cities increased to nine, and the synergy share rose sharply to 77.8%, while trade-offs declined to 22.2%. The majority of newly added synergy cities were Karamay (7), Hutubi (4), and Shihezi (14), whose QH values became positive for the first time, suggesting a transition toward more efficient urban development. Spatially, synergy clusters appeared along the Urumqi–Kuitun–Shihezi corridor, forming the first continuous belt of coordinated development in the UANSTM. Between 2010–2015, ten cities met the threshold, maintaining a high synergy level of 70.0%. Karamay, Kuitun, and Hutubi exhibited the highest positive QH values, while only Qitai (11) and Shanshan (13) still displayed trade-offs. This stage represents the peak phase of synergy, characterized by intensified land use efficiency and stable built-up expansion in most subregions. In 2015–2020, all sixteen cities were represented, indicating full spatial coverage. However, the synergy ratio declined to 50.0%, and trade-offs rose to 50.0%, suggesting a reemergence of uncoordinated growth. While Urumqi (16) and Manas (9) maintained positive coordination, negative QH values appeared in Changji (1), Karamay (7), and Shihezi (14), reflecting the spatial divergence between urban expansion and land efficiency in the late period. This spatial differentiation highlights the varying developmental pace among subregions within the UANSTM. Overall, after accounting for visibility bias, the relationship between UEI and LUE followed a nonlinear evolution: initial dominance of trade-offs (2000–2005), a rapid increase in synergy (2005–2015), and a partial reversal after 2015. The transition of synergy centers—from Urumqi to the northwestern sub-corridor (Kuitun–Shihezi–Karamay)—further confirms the spatial heterogeneity in coordinated urban growth across the UANSTM.

5. Discussion

5.1. Data Reliability and Comparative Validation

The integration of multi-source data significantly enhances the robustness and credibility of this study’s findings. Comparative validation against Landsat-based studies in arid regions confirms the consistency and reliability of the extracted urban boundaries [49,50]. Moreover, the temporal evolution trends derived here align with independent estimates of urban expansion rates across northwestern China [51]. These cross-comparisons reinforce the robustness of the dataset and affirm the methodological soundness of using NTL–GUB–Yearbook integration for long-term spatiotemporal analysis.

The results reveal that from 2000 to 2020, the UANSTM experienced pronounced spatial expansion and fluctuating land use efficiency, with a temporal pattern shifting from trade-off (2000–2005) to synergy (2005–2015), followed by renewed trade-off (2015–2020). These findings are consistent with [52,53]. Similar cyclical patterns were also reported in studies of the Yangtze River Delta and Shandong Peninsula, where rapid urban growth initially suppressed efficiency but later benefited from policy and infrastructure improvements [54,55]. Recent research on high-speed transport corridors in coastal China confirms that transport infrastructure can significantly enhance LUE through agglomeration and accessibility effects. The spatiotemporal variations observed in UANSTM, where synergy peaks align with major policy and investment cycles (around 2010–2015), thus resonate with the broader empirical evidence on regional policy–infrastructure–efficiency linkages.

It is also noteworthy that the synergy–trade-off model, originally developed to assess ecological service interactions, was innovatively adapted in this study to evaluate the coupling between urban expansion intensity and land use efficiency. This methodological adaptation provides a valuable framework for quantifying urbanization dynamics in resource-constrained landscapes.

5.2. Advantages, Limitations, and Future Directions

The methodological innovation of this study lies in its integration of VBANUI, GUB, and multi-source socioeconomic datasets, enabling fine-grained monitoring of urban expansion and efficiency in an ecologically fragile oasis context. The use of a Trade-off–Synergy Intensity model extends existing frameworks [56,57], allowing both temporal transitions and spatial heterogeneity to be quantitatively compared. Furthermore, combining fractal dimension and compactness indicators bridges the gap between morphological evolution and functional efficiency—a dimension often overlooked in previous studies.

Nevertheless, several limitations should be acknowledged. First, despite the rigorous multi-source calibration, the spatial fragmentation and uneven urbanization pattern of UANSTM result in partial underestimation of built-up areas, especially for smaller or newly emerging oasis towns in the early years (Figure 8). This limitation reflects the intrinsic challenges of remote sensing in sparsely illuminated desert regions. Second, while the TSI model successfully captured the dynamic interactions between urban expansion and efficiency, it remains a first attempt to transfer an ecological coupling framework into urban land system analysis. Future research could further refine this model by integrating transportation accessibility, industrial structure, and water–energy constraints into the analytical framework. Third, although field verification and high-resolution imagery support the overall data accuracy, quantitative uncertainty assessments—such as sensitivity analysis or Bayesian error propagation—should be incorporated in future studies to strengthen empirical robustness.

Looking forward, expanding the framework to include the more socioeconomic and ecological feedback mechanisms could enhance the interpretation of spatial heterogeneity in urban efficiency. Moreover, comparative studies with other arid agglomerations (e.g., Central Asia, the Arabian Peninsula) would enable broader generalization of the observed patterns, contributing to a more global understanding of sustainable urban development in water-scarce environments.

5.3. Policy Implications and Practical Significance

The findings underscore the necessity of differentiated governance strategies to achieve coordinated expansion and efficiency in arid urban agglomerations. Policymakers should prioritize compact development in core cities while strengthening spatial control and infrastructure integration in peripheral areas [58]. Empirical evidence from other Chinese regions shows that synchronizing land expansion with transport and service investments yields long-term efficiency gains [54,59]. For regions like UANSTM, where ecological and hydrological constraints are pronounced, policies should incorporate ecological redline management and water availability assessments into urban planning [60]. Moreover, policy timing is critical—synergy peaks observed during 2010–2015 suggest that coordinated investments made in the early stages of expansion can generate lasting benefits. Integrating these insights can guide adaptive planning strategies that balance economic growth, spatial efficiency, and environmental sustainability in western China’s arid cities.

Beyond western China, the methodological and empirical insights of this study also hold broader reference value for arid urban agglomerations worldwide. Many inland regions in Central Asia, the Arabian Peninsula, and North Africa face similar constraints, including severe water scarcity, fragile ecosystems, and spatially discontinuous patterns of urbanization. For instance, Sietz et al. (2011) identified seven typical vulnerability patterns across global drylands, emphasizing that poverty, water stress, and land degradation are key determinants of sustainability [61]. Studies in Central Asia, such as that of Zhang et al. (2019) on the Aral Sea Basin, similarly reveal strong spatial disparities in land and water resource efficiency [62]. Evidence from the Negev Desert shows that compact urbanization is more effective than agricultural expansion in reducing desertification risks while maintaining economic viability [63]. Meanwhile, research on the Phoenix Metropolitan Area demonstrates the utility of remote-sensing indicators and land-configuration metrics for assessing fragmentation and urban heat island effects in dryland cities [64,65].

Collectively, these findings indicate that the SCVBANUI–based framework and the trade-off–synergy analytical approach developed in this study can be extended to monitor urban expansion, evaluate land use efficiency, and assess sustainability trade-offs in other resource-constrained regions. Therefore, this research not only provides policy implications for arid cities in western China but also offers a transferable analytical framework for global dryland urban systems, contributing to comparative studies on sustainable urban development in water-scarce environments.

6. Conclusions

This study proposed a multi-source spatial data framework for extracting built-up areas in arid regions and constructing land use efficiency metrics, with the aim of examining the spatial coupling between UEI and LUE. The framework addresses the spatial mismatch between land input and utilization efficiency, thereby supporting sustainable urban development. The results reveal that the region experienced rapid yet spatially uneven urban expansion, accompanied by gradual but unstable improvements in land use efficiency. Overall, urban growth within this arid inland system has transitioned from compact concentration to dispersed diffusion, reflecting a profound transformation of urban spatial structure under resource and ecological constraints.

Although land use efficiency improved in most cities, its growth was primarily structural rather than extensive. The enhancement of LUE was driven mainly by land use optimization, industrial upgrading, and the agglomeration of productive sectors in core cities such as Urumqi and Karamay. In contrast, peripheral oasis towns remained locked into land-intensive and resource-dependent development patterns, resulting in efficiency stagnation. This divergence underscores a key constraint on urbanization in arid regions: once the marginal benefits of spatial expansion diminish, additional land supply no longer guarantees economic returns. Therefore, sustainable efficiency improvement depends on the internal restructuring and functional intensification of existing urban land, rather than on continued outward expansion.

The synergy–trade-off dynamics between UEI and LUE demonstrate that the relationship is inherently cyclical and context-dependent. The oscillation between trade-off- and synergy-dominated stages reflects the feedback between development intensity, policy orientation, and ecological carrying capacity. Periods of synergy emerged when compact development and coordinated planning improved land productivity, whereas trade-offs reappeared when rapid spatial growth outpaced infrastructural and ecological thresholds. This cyclical interaction reveals a fundamental vulnerability of arid urban systems: excessive reliance on spatial expansion triggers diminishing efficiency returns. Therefore, maintaining long-term synergy requires a transition from extensive, land-driven urbanization toward an endogenous growth model emphasizing spatial compactness, efficiency-oriented governance, and resilience-based planning.