The YREB (Figure 1) is one of China’s key economic zones, playing a significant role in the country’s overall development (Song et al., 2022). It spans China’s eastern, central, and western regions, serving as a vital link between these areas. The YREB, which makes up over 20% of China’s geographical area and accounts for 45% of the country’s GDP, is essential to economic growth. The YREB is divided into three regions (Dong et al., 2020): the lower reaches (Shanghai, Zhejiang, Jiangsu, and Anhui), covering 17.1% of the belt (353,300 km²); the middle reaches (Jiangxi, Hubei, and Hunan), covering 27.5% of the belt (564,600 km²); and the upstream region (Chongqing, Guizhou, Sichuan, and Yunnan), covering 55.4% of the belt (1,137,400 km²). The region is rich in resources of natural origin and has significant growth potential.

The land use data were sourced from Resource and Environmental Science Data Platform.¹ This dataset covers five time periods: 2000, 2005, 2010, 2015, and 2020. To meet the research needs, the original 30 m resolution data is resampled to 300 m. In the study area, the land was divided into six primary varieties, including grassland, woodlands, cropland, water, construction land, and unused land. The nighttime light data came from the corrected Chinese long-term dataset spanning 2000–2020 (Zhong et al., 2022), whereas the energy data came from the China Energy Statistical Yearbook (2001–2021). Socioeconomic data primarily originated from the China County Statistical Yearbook and district/county reports on national economic and social development statistics (2001–2021). Population data were retrieved from the World Population database.² Prefectural scale economic data were obtained through conversions from corresponding county scale data to ensure consistency and accuracy, thereby eliminating errors arising from statistical discrepancies.

Economic development (X1), government intervention (X2), population size (X3), industrial structure (X4), carbon emission intensity (X5), and land use degree (X6) were chosen for this article based on available literature and relevant studies (Li et al., 2021; Rahaman et al., 2022;Cai and Li, 2024; Fan et al., 2024; Wang et al., 2024). Carbon dioxide emissions are greatly influenced by economic development; fast economic expansion frequently results in higher energy consumption and, consequently, higher carbon dioxide emissions, as measured by GDP (Shi et al., 2019). Government intervention is a pivotal factor in impacting economic growth, industrial composition, and technological advancement and, consequently, carbon emissions. The ratio of fiscal expenditure to GDP is frequently used to evaluate the degree of government intervention (Ma and Liu, 2021). The industrial sector is a major source of CO₂ emissions, and changes in industrial structure can have a large influence on carbon emissions (Shi et al., 2019). Population has a considerable impact on carbon dioxide emissions, as there is a tight link between the two. The continuing rise of the population not only directly contributes to increasing energy consumption but also to higher carbon dioxide emissions (Li et al., 2017). The industrial structure is assessed by calculating the ratio of the output value of the tertiary industry to that of the secondary industry (Li et al., 2023). It is anticipated that carbon emission intensity will continue to decline as technology develops, which is essential for reducing total CO₂ emissions (Li et al., 2021). The quantity of carbon emissions for each unit of economic production is shown by the carbon intensity decrease, which is computed as the ratio of total carbon emissions to GDP. A higher land use degree usually leads to larger construction land areas and greater carbon emissions. However, efficient and economical construction land use can reduce carbon dioxide emissions. The percentage of construction land to the entire land area is one metric used to evaluate the extent of land utilization (Song et al., 2022).

Significant variance was seen in land use types and change patterns over the whole YREB from 2000 to 2020. The most common land use categories were grassland, cropland, and woodland (Figure 2a). In general, cropland with grassland consistently decreased, while woodlands, water bodies, construction land, and unused land exhibited upward trends. Notably, among these changes, the shifts in cropland and construction land were particularly noteworthy. Cropland decreased from 638,558.68 to 607,027.86 km², marking a total reduction of 31,530.83 km². Conversely, construction land expanded from 52,107.04 to 84,604.74 km², representing a gain of 32,497.7 km². Woodlands, water, and unused land increased by 661.14, 2,142.44, and 269.34 km², respectively, while the grassland area remained under 4,039.80 km². According to Sankey’s diagram (Figure 2b) illustrating land use transitions over different periods, cropland, primarily converted into construction land, emerged as the most significant land use transformation. This implies that a significant loss of cropland resources resulted from the YREB’s fast urbanization.

From 2000 to 2020, the LUCE’s global Moran’s I values at the three administrative scales in the YREB are continuously positive (Table 1). The Z-value exceeds 1.96, except for the provincial scales in 2005, 2010, and 2015, with the p-value at the 5% level being significant. These findings suggest a significant positive spatial correlation in the LUCE at all three administrative scales in the YREB. Between 2000 and 2020, there were significant fluctuations in the global Moran’s I values for LUCE across different administrative scales. Specifically, at the provincial scale, the global Moran’s I grew from 0.201 to 0.211, indicating a clear upward trend. Similarly, at the prefectural scale, the global Moran’s I also showed an increasing trend, rising from 0.136 in 2000 to 0.161 in 2020. It is noteworthy that the increase in Moran’s I at the provincial scale was more pronounced than that at the prefectural scale. Conversely, at the county scale, the global Moran’s I showed a decreasing trend, declining from 0.561 in 2000 to 0.522 in 2020. This phenomenon may be attributed to the diversification of land use patterns among counties and the differences in policy implementation and development positioning. Some counties have significantly reduced carbon emissions due to the promotion of ecological protection and restoration projects, while others have experienced increased carbon emissions driven by urbanization and the expansion of construction land. This heterogeneity has weakened the spatial agglomeration at the county level. Meanwhile, during rapid urbanization, the relatively slower growth of carbon emissions in economically underdeveloped counties further diminishes the agglomeration effect. In contrast, the increase in the global Moran’s index at the provincial and prefectural levels reflects the uniformity of macro-level policies and the convergence of industrial planning. At these scales, the development of low-carbon industries and regional coordination planning significantly enhance the spatial correlation of carbon emissions. Larger spatial units effectively smooth out local differences, resulting in a stronger spatial aggregation effect over broader areas. Overall, the spatiotemporal characteristics of land use carbon emissions exhibit significant scale effects, with the county level highlighting local heterogeneity more prominently.

Concerning local spatial autocorrelation, it is observed that at the provincial scale, the LUCE in the YREB lacks spatial agglomeration features. At the prefectural scale, there is no significant alteration in the spatial clustering pattern of the LUCE in the YREB (Figure 7). This pattern is characterized by High–High (H–H) and Low–Low (L–L) regional clustering, along with sporadic occurrences in High–Low (H–L) and Low–High (L–H) regions. In general, the H–H agglomeration regions are concentrated in the lower Yangtze River region, encompassing Shanghai, Jiangsu, and Zhejiang provinces. L–L agglomeration regions are primarily situated in Sichuan and Yunnan Provinces. L–H regions are mostly located in Huzhou, Zhoushan, and Ziyang. These cities are mainly concentrated around cities with high LUCE, indicating that their LUCE is low but affected by the positive spillover effect of high-carbon areas. The main regions with H–L characteristics are Chengdu and Chongqing, clustered predominantly around L–L agglomeration areas, leading to significant carbon emissions that spill over into the low-value regions. The prefectural scale LUCE displays more pronounced regional agglomeration and spatial heterogeneity compared to the provincial level.

At the county level (Figure 8), the spatial clustering of the LUCE in the YREB exhibits no significant variation. They are characterized by H–H and L–L regional agglomeration, with sporadic instances of H–L and L–H distribution. From 2000 to 2020, regions with H–H agglomeration were mostly concentrated in Zhejiang, Jiangsu, and Shanghai, and their numbers continued to grow. H–H agglomeration regions were detected in the districts and counties of Wuhan, Changsha, Chongqing, and Chengdu. L–L agglomeration areas were mostly found in Jiangxi, Yunnan, and Sichuan. Additionally, they were sparsely distributed in the central parts of Hubei and Hunan, with occasional appearances in the lower Yangtze River portions of Zhejiang. L–H agglomeration regions are characterized by low LUCE values, although they are impacted by the favorable spillover effects of high-carbon areas. H–L agglomeration regions are located around the L–L agglomeration regions in the central reaches of Hunan, Hubei, and Jiangxi Provinces, and the upstream regions of Sichuan and Yunnan Provinces. Their high LUCE scales have a spillover impact on the L–L agglomeration regions to some extent. The findings of the global Moran’s I analysis are consistent with this, indicating that the spatial correlation of LUCE at the county level decreased throughout this time. In summary, the county level exhibits greater spatial heterogeneity and regional agglomeration characteristics compared to the prefectural scale.

Due to data availability and the limited number of provincial-scale samples, this study focuses exclusively on examining the factors influencing LUCE at the prefectural and county scales within the YREB for the years 2000, 2005, 2010, 2015, and 2020.

The YREB’s considerable increase in LUCE from 2000 to 2020 was mostly caused by the expansion of construction land that contributes to carbon emissions. The period between 2000 and 2010 saw rapid growth (Figure 3), fueled by an extensive development model and the expansion of construction land. However, growth slowed from 2010 to 2020, influenced by the 2012 ecological civilization construction plan and subsequent sustainable development policies (Lin, 2018). Despite the slowdown, urbanization and construction land expansion persist, leading to a continued rise in LUCE.

The spatial distribution of the LUCE in the YREB can be summarized as higher in the lower reaches and lower in the upper reaches. The downstream provinces in the YREB, such as Zhejiang, Jiangsu, and Shanghai, are economically developed and experience rapid infrastructure development, accompanied by significant human activity, substantial energy consumption, and the ongoing growth of construction land (Wang and Shao, 2023). Simultaneously, the ongoing reduction in cropland, woodlands, and grasslands leads to the LUCE’s growth. In contrast, the upstream region, with slower urbanization and larger woodlands, grasslands, and carbon sequestration capacity (Luo et al., 2022), exhibits lower LUCE. However, the middle and upper reaches experience increasing LUCE, attributed to “Western development,” “One Belt, One Road” initiatives, and the relocation of high-emission enterprises (Tian et al., 2019).

County-scale LUCE in the YREB demonstrates stronger spatial autocorrelation, indicating that smaller administrative units are prone to prominent spatial clustering patterns for LUCE (Shi et al., 2019). Local spatial autocorrelation analysis indicates that the provincial administrative scale lacks significant spatial agglomeration, possibly due to a limited number of samples. The prefectural scale exhibits positive spatial agglomeration, while the county administrative scale shows even more significant positive spatial agglomeration than the prefectural scale. Overall, the county scale in the YREB demonstrates stronger spatial heterogeneity and regional agglomeration features compared to the other two administrative scales. The primary reasons for the observed results lie in the differences in spatial unit division and socio-economic driving factors across various administrative scales. At the provincial scale, the relatively large administrative unit size often conceals internal variations in land use. The prefectural scale, situated between the provincial and county scales, provides a better representation of land use differences and their connections to regional economic activities, while also revealing heterogeneity that remains undetected at the provincial scale. In contrast, the county scale focuses on smaller spatial units, capturing local characteristics of land use changes, population distribution, and economic activities more comprehensively. This allows for a more pronounced observation of spatial heterogeneity and agglomeration effects. Consequently, land use carbon emissions exhibit distinct spatial heterogeneity across different administrative scales.

Economic development is the primary determinant of LUCE at the prefectural and county levels in the YREB, according to the findings of the geographic detector study. This result is in line with previous studies showing that the main driver of the rise in carbon emissions in Chinese cities has always been economic development (Guan et al., 2018; Zheng et al., 2020). The main cause for this result is that rapid local economic growth mainly stems from infrastructure construction and industrial upgrading. Meanwhile, infrastructure development and industrial upgrading also affect regional energy consumption and carbon emissions. With the advancement of urbanization, the urban agglomeration effect further strengthens the driving force behind local economic development, creating a virtuous cycle. This cyclical mechanism leads to the continuous interaction between economic development, industrial upgrading, and urbanization, promoting rapid regional economic growth. However, as economic activities increase, carbon emissions also show a noticeable accumulation trend, particularly in regions with rapid economic development, where the growth in carbon emissions is especially prominent.

Apart from economic development, other influencing factors also show different characteristics at different administrative scales. Specifically, at the prefectural scale, the importance of influencing factors is ranked as follows: population, land use degree, government intervention, industrial structure, and carbon emission intensity. At the county scale, the ranking is: land use degree, population, carbon emission intensity, government intervention, and industrial structure. The differences in the intensity ranking of influencing factors across scales highlight the impact of scale effects on carbon emissions. The differences in the ranking of carbon emission influencing factors across administrative scales essentially reflect the scale effect, meaning that the importance of these factors varies with scale due to differences in spatial and social contexts. This effect highlights the regional heterogeneity and the necessity of tailoring policies to local conditions, indicating that emission reduction strategies should be precisely designed according to the specific circumstances of each region. An examination of the two-factor interaction reveals that it takes the form of two-factor enhancement and nonlinear enhancement at both scales. This implies that the combined activity of several factors influences LUCE, with the two-factor interaction having a greater effect than the effects of the individual elements. A thorough analysis of the interplay between variables should be taken into account while developing emission reduction strategies.

The MGWR model is utilized to investigate the influencing variables of LUCE at the prefectural and county scales in the YREB in 2020 in order to better assess the local characteristics of the influencing factors on LUCE while taking spatial aspects into account. Applying the MGWR model in 2020 at the prefectural and county scales, economic development, population size, industrial structure, and carbon emission intensity exhibit similar bandwidths with pronounced spatial heterogeneity. Land use intensity followed, while government intervention manifested as a global feature, with minimal spatial heterogeneity. However, at the county scale, the bandwidths of impact factors demonstrated certain spatial heterogeneity, with no single dominant global factor. The degree of spatial heterogeneity decreased in the order of economic development > carbon intensity > population size > government intervention > land use intensity > industrial structure. This indicates that different impact factors show different spatial heterogeneity on the same research scale, and the same impact factors would also show different spatial heterogeneity due to different research scales. Comparing impact factor coefficient ranges between prefectural and county scales reveals that the county scale exhibits greater variability, emphasizing its stronger spatial heterogeneity in contrast with the prefectural scale. Comparing the coefficient ranges of the same impact factors between the prefectural and county scales, we found that the county scale showed larger variations than the prefectural scale. This is mainly because, in comparison to the prefectural scale, different counties have significant geographic, economic, and social differences, which, at the county scale, lead to a noticeable regional variability in the LUCE affecting factors.

The study’s findings are critical for the government in developing more accurate emission reduction plans and measures for the YREB. Based on this, the following three recommendations are made in this study. First, county-scale carbon emission management: because the spatial heterogeneity of factors affecting LUCE at the county scale is high, the government should prioritize county-scale carbon emission control. To address the carbon emission challenges effectively, tailored policies and measures must be developed, taking into account the unique characteristics and issues specific to various county-scale regions. This may involve the implementation of strategies like promoting low-carbon agriculture, enhancing ecological preservation efforts, and advancing renewable energy initiatives in high-carbon emission counties. Additionally, providing comprehensive support through carbon emission monitoring and technological assistance is essential. Second, urban planning and carbon emission reduction should be integrated. As economic growth drives LUCE, the government must integrate carbon emission reduction into urban planning. This requires adopting low-carbon urban planning, setting carbon reduction targets, and fostering economic development while coordinating carbon emissions effectively. This might involve encouraging the use of clean energy, encouraging sustainable mobility, enhancing urban land planning, and enhancing the monitoring and control of carbon emissions. By combining urban planning and carbon emission management, a win-win scenario of economic expansion and carbon emission reduction may be realized. Finally, interregional cooperation should be strengthened. LUCE at different scales in the YREB have differing spatial heterogeneity, implying that interregional differences must be addressed and resolved. To jointly address carbon emission challenges, governments can promote interregional collaboration and cooperation as well as share experience and technology through cross-regional cooperation mechanisms. Additionally, the government can enhance interregional policy coordination and formulate integrated carbon emission management policies. This will facilitate coordinated development and optimize carbon emission reduction across diverse YREB regions.

In this paper, we examined the spatiotemporal traits of LUCE in YREB provinces, prefectural cities, and counties. We also evaluated the LUCE-affecting elements at the county and prefectural levels, which led to important findings. Nevertheless, the following restrictions are present: (1) Because most energy data at the prefectural and county scales are unavailable, the energy carbon emission data estimated at the prefectural and county scales must be checked and examined further. (2) As impact factors, only economic development, industrial structure, government interference, population size, land use degree, and carbon emission intensity are examined. However, because LUCE is a complicated process with numerous impact components, other driving forces and internal processes need to be studied further.