Land-Use Transfer and Its Ecological Effects in Rapidly Urbanizing Areas: A Case Study of Nanjing, China

Abstract

Rapid transformations in land use and land cover (LULC) serve as critical drivers influencing the eco-environmental quality in swiftly urbanizing areas. This study aims to assess and analyze the land-use transfer in Nanjing, China from 2003 to 2023 and its effects on ecological environment quality, utilizing the land expansion intensity (E_n), the land use composite index (LUCI), the remote sensing ecological index (RSEI), and other models. The results show that (1) farmland and forestland decreased significantly, with forestland showing the largest decrease (−20.65%), while construction land showed the largest increase (98.70%), mainly from farmland; (2) the overall RSEI level in Nanjing was relatively high, with a slight decline in fluctuation in the past 20 years. The RSEI values of forest land (0.8786) and farmland (0.8318) were higher, while the construction land (0.3790) and unused land (0.3701) were lower; (3) there was significant correlation (p < 0.05) and spatial autocorrelation between land-use changes and RSEI changes. The RSEI of rural areas was better than that of urban centers. There was a significant negative linear correlation between RSEI and LUCI (R² = 0.711), a significant positive correlation with the area proportion of farmland, forest land and water, and a significant negative correlation with the area proportion of construction land. (4) Predictions indicate continued declines in farmland and forestland, accompanied by further expansion of construction areas, and the RSEI will continue to decline. It is suggested that forestland and farmland should be protected in the future, the expansion of construction land should be restrained, urban construction and ecological protection should be balanced, and the sustainable and high-quality development of rapid urbanization areas should be maintained by accurate land-use planning.

1. Introduction

The rapid advancement of urbanization has been significantly driven by economic development. Over the past two decades, the urbanization rate in most regions of China has surged from less than 40% to nearly 80% [1]. This trend is particularly pronounced in the more developed coastal areas, which have spearheaded urbanization efforts nationwide [2]. The escalation in urbanization levels is intrinsically linked to alterations in land use, as extensive tracts of vegetative cover lands have been transferred into construction land to accommodate urban expansion. This urban growth has precipitated large-scale deforestation, the reclamation of lakes and wetlands, and a reduction in arable land, resulting in a myriad of ecological challenges, including air pollution, the urban heat island effect, soil degradation, and biodiversity loss. These issues have begun to exert a profound impact on the quality of the ecological environment, thereby affecting human habitation [3,4,5].

High-quality development is a central objective of China’s 14th Five-Year Plan, representing a transition from a previous emphasis on rapid economic growth to a new paradigm that prioritizes green and ecologically sustainable development. As one of the world’s most populous nations, China’s commitment to “protection of arable land and ecological preservation” constitutes a fundamental national policy closely linked to public welfare and national security. This commitment inevitably engenders a conflict in land use between urban expansion and environmental protection. Consequently, in rapidly urbanizing regions, it is imperative to accurately assess the dynamic changes in land use and ecological quality, balance various land-use metrics, and align urban development with the ecological integrity of the living environment. The formulation of scientific land-use plans that prioritize ecological considerations and promote low-carbon, sustainable development holds significant theoretical and practical value.

Land types dominated by vegetation cover, such as farmland, forests, shrubs, and grasslands, provide significant ecological benefits, including carbon dioxide absorption, cooling, humidification, and air purification [6,7,8]. Water bodies also serve essential ecological functions by cooling, humidifying, and storing carbon [9,10]. Construction land exacerbates the heat island effect and fragments natural landscapes, negatively impacting regional ecosystem services [11]. Rapid urbanization, which converts large areas of farmland and forests into construction land, inevitably impacts the ecological quality of the environment [12].

In recent years, advances in satellite remote sensing technology have rendered high-resolution land-use data widely accessible for the investigation of environmental and ecological quality [13]. As research has evolved, land-use changes have been identified as a critical factor influencing regional eco-environmental quality [11]. These changes, in conjunction with local ecosystem functions, exhibit significant spatial heterogeneity and directly impact environmental quality and human well-being [14]. Amid rapid urbanization, the competition between construction land and ecological lands has intensified, exacerbating the confrontation between economic development and environmental protection [2].

Previous studies have demonstrated that rapid urban expansion gives rise to a range of ecological challenges that pose severe threats to human survival and development [15,16]. The effects of land-use changes predominantly impact atmospheric conditions, water bodies, and soil quality [8,17,18]. Furthermore, socioeconomic transformations drive alterations in land-use patterns and structures, thereby affecting ecosystem services and their efficiency [19]. The mechanisms underlying these impacts are intricate, and research into the ecological consequences of land-use changes remains an evolving field.

Early studies predominantly assessed land-use value by calculating ecosystem service value coefficients [20]. For instance, Tatiana and colleagues employed an agent-based model (ABM) to elucidate, from an economic perspective, how the land market influences coastal ecosystems through property rights and pricing mechanisms [21]. Qi et al. (2019) analyzed time-series changes in LULC in China since 1980, identifying population growth, economic expansion, urbanization, and policy regulations as the primary drivers of land-use transformations [22]. Overall, the previous research placed less emphasis on the comprehensive impacts of land-use changes on ecological quality. However, with advancements in remote sensing technology and the growing significance of ecological issues stemming from urbanization, researchers have increasingly utilized multi-temporal remote sensing images to investigate large-scale land-use changes [23,24,25]. Nonetheless, the limited resolution of remote sensing images has resulted in a relative scarcity of research focused on small- and medium-scale land use.

In terms of methodology, most current studies build evaluation indicator systems and comprehensive index models to quantitatively research the patterns of LULC and ecological quality variations in different regions [26,27,28]. In the realm of ecological quality assessment, Xu (2013) introduced a remote-sensing-based ecological index (RSEI) model that utilizes indicators such as humidity index—WET, greenness index—normalized difference vegetation index (NDVI), heat index—land surface temperature (LST), and dryness index—normalized difference soil and built-up index (NDSBI) [29]. While this model is primarily designed to quantify ecological quality in urban environments, it has limitations regarding its consideration of water bodies. As a result, subsequent researchers have enhanced the model by integrating additional indices, including atmospheric, water, and desertification indices, to more comprehensively assess the impacts of multiple factors on ecological quality [30,31]. Other notable models in this domain include the eco-environmental index (EI) [32] and the eco-environmental quality index (EQI) [33]. The scope of research has evolved from a narrow focus on specific land types, such as lakes, rivers, and deserts, to a more holistic examination of comprehensive land-use spaces, including urban, rural, and watershed areas [34,35].

In rapidly urbanizing regions, characterized by intense economic growth and escalating demand for urban expansion, competition for land use has become increasingly pronounced. The rapid development of economy and the rapid expansion of cities have seriously affected the environmental quality of cities [34]. Objectively, we should accurately and timely evaluate and monitor the changes in eco-environmental quality, especially the dynamic changes in land use and its ecological effects in different scales and different periods. It can provide technical support for optimizing the land-use strategy, accurately regulating the land-use structure, guiding land and urban planning, and avoiding environmental problems caused by improper land use. It is the key guarantee for the future high-quality and sustainable development in rapidly urbanizing areas, which need to be further studied.

Nanjing, a major provincial capital in China’s economically developed coastal region and a key city within the Yangtze River Delta metropolitan area, has undergone significant urban expansion over the past two decades. The urbanization rate increased from 74.2% (2003) to 87.2% (2023), exceeding the national average. According to Nanjing’s 14th Five-Year Plan, this rate is projected to surpass 88% within the next five years [36].

This study focuses on Nanjing for two primary reasons. First, it serves as a representative region exemplifying rapid urbanization. Second, as a prominent provincial capital in the Yangtze River Delta, Nanjing possesses considerable influence, significant development potential, and a strong demand for construction land, making it a highly relevant case for analysis. The aim of this study is to investigate the land-use transfer and its ecological effects at county and township scales utilizing high-resolution, long-term, and remote sensing data. To accomplish this objective, the study will address the following key issues: (1) Analyze the dynamic change characteristics of land use transfer in Nanjing from 2003 to 2023 and calculate the development trend in the future; (2) Assess the spatio-temporal changes in eco-environmental quality in Nanjing during the same period; (3) Examine the regression relation and spatial autocorrelation between LUCI and ecological quality in Nanjing. Based on the results, this study aims to offer suggestions for optimizing land-use patterns in rapidly urbanizing areas like Nanjing, thereby fostering high-quality development with an emphasis on ecological sustainability.

2. Area Studied and Data Sources

2.1. Area Studied

Nanjing is the capital city of Jiangsu Province, located in the lower reaches of the Yangtze River, geographically positioned between 31°14′–32°37′ N and 118°22′–119°14′ E. Nanjing governs 11 districts and counties, covering a gross area of 6587.02 km², with an urban built-up area of 868.28 km² (Figure 1). As of 2023, Nanjing had a population of 9.49 million, with 8.26 million residing in urban areas, resulting in an urbanization rate of 87.20%. The city’s GDP reached 1.74 trillion RMB. Water bodies and low hills account for 11.4% and 64.5% of the city’s total area, respectively.

As an economically developed city in China’s eastern coastal region, Nanjing has experienced rapid urban expansion over the past 20 years. Rapid urbanizing has driven massive demand for various types of construction land.

2.2. Data Sources and Data Processed

2.2.1. Data Sources

The LULC data for this study were downloaded from the CLCD data set from Wuhan University with a 30-m resolution (https://zenodo.org/records/8176941, accessed on 6 August 2024) [37]. The data were selected for five time points spanning two decades: 2003, 2008, 2013, 2018, and 2023. Based on the land-use classification scheme made by the Chinese Academy of Sciences, the dataset was reclassified into 6 categories: farmland, forestland, shrubs, grassland, water bodies, unused land, and construction land [38].

High-quality remote sensing data for monitoring the eco-environmental quality of Nanjing were sourced from the official website of the China Geo-spatial Data Cloud (http://www.gscloud.cn/search, accessed on 23 August 2024) [39], specifically using Landsat 5-TM and Landsat 8-OLI imagery. To minimize the effects of seasonal differences in vegetation growth, remote sensing data for five years: 2003, 2008, 2013, 2018, and 2023, were selected. Additionally, to reduce cloud interference, all selected images had cloud cover of less than 5%. In order to keep the land-use data consistent with the satellite remote sensing data in time, the time of the two types of data was selected from April to September, which was the most vigorous month for plant growth, and the data were taken as the average value (

Table 1. Information of data sources.

Type	Sensor Type	Path/Row	Date	Data Source
LULC data	Land-use datasets (CLCD-30)	N50_30	2003, 2008, 2013, 2018, 2023	http://www.resdc.cn https://zenodo.org/records/8176941 (accessed on 6 August 2024)
Satellite remote sensing data	Landsat_5 TM	120/038	20 August 2003	http://www.gscloud.cn/search (accessed on 6 August 2024)
		120/038	9 August 2008
		120/038	11 August 2013
	Landsat_8 OLI/TIRS	120/038	6 August 2018
		120/038	7 August 2023
SHP data	administrative boundary data	/	2023	http://www.webmap.cn/ (accessed on 6 August 2024) National Geomatics Center of China

).

Other social data were obtained from the Nanjing Municipal Statistics Bureau’s official website (http://tjj.nanjing.gov.cn/material/, accessed on 9 October 2024) [36] and corresponding editions of the Nanjing Yearbook and district statistical yearbooks.

2.2.2. Data Processed

The satellite remote sensing data of Nanjing were processed using ArcGIS 10.8 and ENVI 5.5 software, which included steps such as radiometric calibration, atmospheric correction, and projection transformation. All remote sensing data were transformed to the WGS_1984_UTM_Zone_50N coordinate system. The vector data for Nanjing’s administrative boundaries were then used to mask and clip the images for subsequent extraction and inversion of environmental and ecological indicators. Given that water bodies can significantly affect inversion calculations, the modified normalized difference water index (MNDWI) was employed to extract and exclude water bodies [25]. Because the reflectivity of water is the strongest in the near infrared and mid infrared bands, therefore, the ratio of green band (B3) to mid infrared band (B5) can suppress the vegetation information to the greatest extent and highlight the water information in satellite images. In this study, the R language code of the water mask module in GEE platform is used to mask the water body. The threshold of water extraction is 0.3, and the binary image of the water mask is generated.

For LULC data processed, according to the current situation of land use in Nanjing, using the reclassification function of ArcGIS 10.8 software, the original cropland, forest, shrub, grassland, water, snow/ice, barren, impervious, and wetland in the CLCD data from Wuhan University were reclassified into 6 land types, namely, farmland (cropland), forestland, grassland (shrub and grassland), water body (water and wetland), construction land (from impervious), and unused land (from barren). farmland, forestland, shrubs, and grassland were grouped as vegetation-covered land [27]. Due to the limitations of the data resolution, some water surfaces with abundant aquatic vegetation were mistakenly classified as vegetation-covered land, while relatively narrow or small green spaces in construction land were often misclassified as construction land. Therefore, manual visual interpretation and corrections based on high-resolution satellite images were necessary in the post-processing stage to improve accuracy.

3. Methodology

The research framework of this study is shown in Figure 2. First, LULC data and remote sensing images of Nanjing from 2003 to 2023 were collected to analyze the transfer of land use and calculate the land expansion intensity and utilization degree for various land types. At the same time, RSEI was used to assess the eco-environmental quality over the same period. By overlaying land-use data with RSEI data, regression models were established between eco-environmental quality and land-use indicators such as the land-use comprehensive index (LUCI), the area and percentage of different land-use types, in order to explore the effects by changes in land use.

3.1. Land-Use Change Analysis

3.1.1. Land Expansion Intensity

The intensity of land expansion refers to the average annual change in the area of various types of land within a certain region. It can reflect the transfer speed and change magnitude of 6 land types:

$$E_n={\displaystyle \frac{A_{t2}-A_{t1}}{A_{t1}}}\times {\displaystyle \frac{1}{t_2-t_1}}\times 100\%$$

(1)

where E_n is the expansion intensity of a certain land-use type, and A_t1 and A_t2 are the areas (km²) of the land-use type in year t₁ and year t₂, respectively.

3.1.2. Land-Use Degree

The degree of land use refers to the level at which a certain type of land is developed and utilized by human activities. This study introduces the concept of the land use comprehensive index (LUCI), which can reflect the impact of human activities on land use. Commonly used land types are divided into five levels, each assigned different values (0–4), to quantify the degree of land use [40]. The greater the impact of human activities on land, the higher the LUCI value. The formula of LUCI is

$$LUCI=100\times \sum _{i=1}^n{P}_i\times G_i$$

(2)

where G_i is the land type grade value i (when i corresponds to unused land, grassland, water, forestland, farmland, and construction land, G_i is recorded as 0, 1, 2, 2, 3, and 4, respectively [40]); P_i represents the percentage of the area of level i land use in the total area. LUCI is the sum of all types of land use, with a value ranging from 100 to 400.

$$\mathrm{\Delta }LUCI={LUCI}_{t1}-{LUCI}_{t2}$$

(3)

where LUCI_t1 and LUCI_t2 represent the values of the LUCI for t₁ and t₂, respectively.

ΔLUCI can measure the degree of land transfer in a certain period. When ΔLUCI > 0, it indicates that the comprehensive degree of land use is in the development stage; When ΔLUCI < 0, it is in the recession stage.

3.1.3. Transfer Matrix of Land Use

The LULC transfer matrix can comprehensively reflect the area and the direction of regional various land transformation and the spatial structure changes of LULC. Using the “Overlay Analysis” tool in ArcGIS 10.5, a land-use transfer matrix for the 20-year period from 2003 to 2023 was obtained.

3.2. Evaluation of Eco-Environment Quality: RSEI

RSEI is a method for evaluating surface ecological quality using satellite remote sensing data. This approach is not constrained by on-site surveys or statistical data, making it convenient and accessible for data acquisition. The evaluation results are highly consistent with the EI model used by China’s Ministry of Environmental Protection [41], effectively reflecting local spatial ecological quality. In the RESI model, principal component analysis (PCA) was utilized to reduce dimensionality and integrated numerous indicators to four major indicators: NDVI, WET, NDBSI, and LST (Equation (4)) [29].

$$RSEI=f\left(NDVI,WET,NDSBI,LST\right)$$

(4)

where NDVI is the vegetation coverage, which represents the impact of vegetation on ecological quality; WET represents the moisture of soil; NDBSI represents the amount of building and impermeable floor in construction land and unused land, which is composed of soil index (SI) and building index (BI) [42], and can reflect the construction level of the city; LST represents the surface temperature of regional vegetation and soil, which is an important climate indicator affecting vegetation growth [19]. These indicators are inversed using the atmospheric correction method through thermal infrared Landsat data. The specific calculation formulas are shown in reference [29].

Additionally, large water bodies such as Xuanwu Lake and the Yangtze River exist within Nanjing’s boundaries. The WET index may be affected by these water bodies, leading to potentially inaccurate RSEI results for these areas [43]. Therefore, during the remote sensing image processing, the water bodies should be masked.

3.3. Spatial Analysis

Spatial autocorrelation can measure the spatial correlation between various types of land use and the quality of the ecological environment (RSEI) in Nanjing. The evaluation indicators include global spatial autocorrelation (global Moran’s I) and local spatial autocorrelation (local Moran’s I) [44]. The formula for calculating the Global Moran’s I is as follows:

$$GlobalMoran’sI={\displaystyle \frac{n\sum _{i=1}^n\sum _{j=1}^n{M}_{ij}(x_i-\overline{x})(x_j-\overline{x})}{\left(\sum _{i=1}^n\sum _{j=1}^n{M}_{ij}\right)\sum _{i=1}^n{(x_i-\overline{x})}^2}}$$

(5)

where Mij is the spatial matrix; n is the number of regional units; x_i is the observed value of the ith unit; $\overline{x}$ is the average value of the observed quantity. In this article, the adjacent method is used to construct the spatial weight matrix, and the expected value of the global Moran index is calculated as follows:

E(Global Moran’s I) = −1/(n − 1)

(6)

If the calculation result is positive, it indicates that the ecological quality of Nanjing is significantly clustered in space. If the result is negative, it indicates that the spatial differences in ecological quality are significant. The larger the absolute value of the calculation result, the stronger the significance.

The local Moran’s index can be used to reveal the local spatial distribution characteristics of the ecological environment quality in Nanjing. This article uses the local Moran’s index LISA method for calculation [44], with the following formula:

$$LocalMoran’sI={\displaystyle \frac{n(x_i-\overline{x})\sum _{j=1}^n{W}_{ij}(x_j-\overline{x})}{\sum _{i=1}^n{(x_i-\overline{x})}^2}}$$

(7)

where the parameters represent the same meaning as before.

3.4. Data Statistical Analysis Method

This study employed the geographically weighted regression (GWR) model within the spatial analysis tools of ArcGIS 10.5 to perform an overlay analysis of land-use data and RSEI raster data for Nanjing. Additionally, land-use data for 171 towns and 11 districts of Nanjing from 2003 to 2023 were statistically analyzed, including the area of various land-use types. A regression equation was established between the RSEI and key land-use variables such as land-use area and the percentage of total area by SPSS 27.

4. Results

4.1. Spatial-Temporal Distribution of Land Use in Nanjing

The land use raster data of Nanjing were collected for five years: 2003, 2008, 2013, 2018, and 2023, spanning a total of 20 years. After reclassification using ArcGIS 10.5 software, land use was categorized into five major types (grassland is ignored due to its small area), as shown in Figure 3.

4.1.1. Area of Various Types of Land and E_n

Farmland dominates the land use in Nanjing, accounting for approximately 73%, mainly located in rural and suburban areas. Between 2003 and 2023, the farmland area continuously decreased, with a total reduction of 14.16% and an average annual expansion intensity of −0.71%. The decline was faster before 2013 and gradually slowed afterward. Since 2018, due to the influence of national farmland protection policies, the rate of reduction has significantly slowed. Among other major land-use types, construction land saw the highest growth, increasing by 98.70% between 2003 and 2023, with an average annual expansion rate of 4.93%. Notably, during 2008–2013, the annual E_n was 4.77%. Between 2018 and 2023, the E_n slowed, with 1.89%, closely matching the trend in urbanization during the same period. Forest land in Nanjing exhibited a “V”-shaped trend, initially decreasing and then increasing. The largest decline occurred between 2013 and 2018, followed by a recovery after 2018. Over the 20-year period, forest land decreased by 20.65%. Water bodies in Nanjing are abundant, and their area increased 17.82% from 2003 to 2023. Unused land accounted for the smallest proportion (except grassland), and although its area fluctuated the most, the absolute area is too small to be significant (

Table 2. Area and E_n of land uses from 2003 to 2023.

Land Use	Area (km2)					En (%)
	2003	2008	2013	2018	2023	2003–2008	2008–2013	2013–2018	2018–2023	2003–2023
FA	4805.47	4517.86	4275.61	4158.00	4125.16	−1.20	−1.07	−0.55	−0.16	−0.71
FO	506.28	520.86	485.62	391.96	401.74	0.58	−1.35	−3.86	0.50	−1.03
WB	582.45	704.29	780.81	781.62	686.21	4.18	2.17	0.02	−2.44	0.89
CO	690.00	841.20	1042.02	1252.51	1371.05	4.38	4.77	4.04	1.89	4.93
BA	0.03	0.02	0.17	0.14	0.07	−6.67	150.00	−3.53	−10.00	6.67

Note: the unit of land conversion area in the table is km². FA: farmland, FO: forest land, WB: water body, CO: construction land, BA: unused land.

).

4.1.2. LUCI and ΔLUCI

The LUCI values for Nanjing in 2003, 2008, 2013, 2018, and 2023 were 325.05, 329.24, 334.37, 339.02, and 340.76, respectively, indicating a high overall level of LUCI. The ΔLUCI value has consistently remained positive, indicating a steady increase in LUCI with an average annual growth rate of 0.52. Over the past two decades, the highest ΔLUCI value was observed from 2008 to 2013, with an increase of 5.12. Conversely, the lowest ΔLUCI increase was from 2018 to 2023, with only a 1.74 increase. This trend aligns with the E_n of construction land, and there is a significant correlation between the degree of land use and the E_n of construction land (p < 0.01). The aforementioned results suggest that the degree of land use in Nanjing has been steadily increasing over the past 20 years, with the continuous growth of construction land area being the primary factor driving the increase in LUCI. Another significant factor influencing the degree of land use is land-use policy. In 2017, China initiated a policy to protect the red line of cultivated land, restricting land circulation. Since then, the expansion intensity of construction land has begun to slow down, and the growth rate of land-use intensity was the lowest in the five-year period from 2018 to 2023. The E_n of construction land not only enhances the LUCI but also simultaneously occupies ecological spaces such as farmland, water bodies, and forests, posing a risk of exacerbating the conflict between land use and ecological environment quality (Figure 4).

4.1.3. Transfer of Land Use

In

Table 3. Land-use Transition Matrix of Nanjing from 2003 to 2023.

Years	Land Type	FA	FO	GR	WB	BA	CO	Land Transition Chart
2003–2008	FA	5074.60	47.67	1.20	176.20	0.00	166.00
	FO	29.74	542.82	0.03	0.00	0.00	2.61
	GR	0.18	0.04	0.54	0.01	0.00	0.14
	WB	34.06	0.20	0.01	621.38	0.01	6.70
	BA	0.00	0.00	0.01	0.00	0.01	0.02
	CO	0.08	0.00	0.00	3.05	0.00	781.43
2008–2013	FA	4760.99	22.51	2.51	132.12	0.03	220.50
	FO	61.50	525.96	0.01	0.04	0.00	3.23
	GR	0.14	0.02	1.09	0.00	0.15	0.39
	WB	40.11	0.31	0.06	752.95	0.01	7.21
	BA	0.00	0.00	0.00	0.01	0.01	0.01
	CO	0.07	0.00	0.00	2.73	0.00	954.11
2013–2018	FA	4574.30	8.90	0.17	54.62	0.00	224.81
	FO	109.90	436.26	0.00	0.01	0.00	2.62
	GR	0.23	0.01	0.52	0.00	0.02	2.89
	WB	44.41	0.00	0.00	830.88	0.04	12.51
	BA	0.01	0.00	0.00	0.00	0.09	0.09
	CO	0.04	0.00	0.00	3.10	0.00	1182.31
2018–2023	FA	4537.51	44.59	0.01	20.00	0.00	126.79
	FO	32.69	412.10	0.00	0.01	0.00	0.38
	GR	0.22	0.01	0.24	0.00	0.02	0.20
	WB	121.23	0.02	0.00	757.68	0.01	9.67
	BA	0.06	0.00	0.01	0.00	0.05	0.04
	CO	0.08	0.00	0.00	2.42	0.00	1422.73

Note: the unit of land conversion area in the table is km². FA: Farmland, FO: forest land, WB: Water body, CO: Construction land, GR: Grass land, BA: Unused land. In the land transfer charts, red represents FA, blue represents FO, purple represents WB, and green represents CO.

and Figure 5, From 2003 to 2008, the main land-use changes were the conversion of farmland to construction land (166.00 km²) and farmland to water bodies (176.20 km²). During this time, in addition to the urbanization-driven expansion of construction land, some areas, particularly Gaochun County and Lishui County, developed the aquaculture industry, leading to a portion of farmland being converted into fish ponds. Other changes included a small amount of forest land being converted to farmland (29.74 km²), farmland converted to forest land (47.67 km²), and water bodies converted to farmland (34.06 km²). The proportion of unused land in Nanjing has always been relatively small and experienced little change.

From 2008 to 2013, influenced by policies such as “returning forest to farmland” and urban development, significant land-use changes included 109.90 km² of forestland being transferred to farmland and 224.81 km² of farmland being converted to construction land. There were also minor reciprocal changes between water bodies and farmland, with a small amount converted into construction land. Other land-use types remained relatively stable. This indicates that between 2008 and 2013, apart from the intense conversion between farmland and construction land, competition between farmland and forest land also intensified, while water bodies became an important supplement for construction land.

From 2013 to 2018, the main land-use conversions were the transformation of farmland to construction land (224.81 km²) and forest land to farmland (109.90 km²). Reciprocal changes between water bodies and farmland also occurred, with areas of 44.41 km² and 54.62 km² being exchanged, respectively.

Between 2018 and 2023, with the implementation of the “farmland protection” policy, the intensity of farmland conversion to construction land significantly weakened. Simultaneously, much of the farmland that had been previously used for aquaculture returned to cultivation, with 121.23 km² of water bodies being converted back to farmland. This suggests that the pace of urban development slowed during this period, particularly as the farmland protection policy influenced land-use transfers.

In summary, from 2003 to 2023, land-use conversions in Nanjing were mainly driven by urbanization and land policies. Farmland consistently shifted towards construction land, while reciprocal conversions occurred between ecological land types such as forest land and water bodies with farmland. Each period exhibited distinct characteristics that reflected the land policies of the time.

4.2. Spatio-Temporal Changes in Ecological Quality

4.2.1. The Eco-Environmental Quality (RSEI) in Nanjing from 2003 to 2023

This study employs a five-level classification method to categorize the RSEI value to five levels: poor (0–0.2), inferior (0.2–0.4), medium (0.4–0.6), good (0.6–0.8), and excellent (0.8–1). A map depicting the RSEI in Nanjing is presented in Figure 6, along with a statistical analysis of the area and proportion of each ecological level over four time periods (

Table 4. Proportion of RSEI Areas at Various Levels in Nanjing from 2003 to 2023.

RSEI Levels	2003		2008		2013		2018		2023
	Area	%	Area	%	Area	%	Area	%	Area	%
Poor (0–0.2)	103.36	1.58	326.75	4.98	528.01	8.05	215.76	3.29	541.34	8.25
Inferior (0.2–0.4)	693.67	10.57	880.28	13.42	988.15	15.06	835.75	12.74	1033.40	15.75
Medium (0.4–0.6)	2118.65	32.29	1591.03	24.25	1538.50	23.45	1448.58	22.08	1358.32	20.70
Good (0.6–0.8)	2587.17	39.43	2706.28	41.25	2439.50	37.18	2462.55	37.53	1985.87	30.27
Excellent (0.8–1)	1058.33	16.13	1056.82	16.11	1056.37	16.10	1598.53	24.36	1642.25	25.03

).

Comparing the RSEI levels from 2003 to 2023, the areas with the poorest ecological quality (RSEI) are concentrated in the central urban areas of Nanjing and spread outward from the core urban zone. This indicates that as Nanjing’s urban area expands, the ecological quality within the urban area—except for densely vegetated regions like Zijin Mountain and Niushou Mountain, which maintain excellent level—exhibits poor levels in construction land areas compared to suburban and rural regions. In terms of RSEI spatial distribution, areas of excellent levels are mainly found in forested regions, while areas with poor level are concentrated in central urban construction land and unused land. The largest area with relatively good RSEI levels is concentrated in farmland and small grassland areas, as shown in Figure 6.

The area proportion of the excellent and poor levels has been increasing, reflecting a polarization trend, while the proportions of good, inferior, and poor levels experienced fluctuations. Notably, the area in the higher category reached its peak, then declined; the medium and poor categories showed relatively stable changes with slight variations (see Figure 6). Starting in 2003, Nanjing’s urbanization accelerated, leading to significant increases in construction land over the next decade, which likely contributed to the decline in overall ecological quality. Beginning in 2013, Nanjing initiated the construction of an ecological garden city, developing high-quality urban parks and rural green belts, which the overall ecological quality was gradually improving. The changes in RSEI closely align with land-use changes during the same period.

4.2.2. Spatio-Temporal Changes in RSEI in Nanjing from 2003 to 2023

Based on the RSEI model calculations, the average RSEI values for Nanjing from 2003 to 2023 were 0.8049 (2003), 0.7901 (2008), 0.7741 (2013), 0.7931 (2018), and 0.7908 (2023). This indicates that the overall ecological quality of Nanjing remained relatively good over the 20 years, showing a trend of decline followed by recovery, with the lowest point in 2013.

Analyzing the changes in RSEI in Nanjing from 2003 to 2023 reveals a trend of both improvement and deterioration. The proportion of deteriorated areas increased year by year, while the proportion of improved areas decreased, which led to an overall decline in Nanjing’s RSEI levels. Shown in

Table 5. Change detection of RSEI levels in Nanjing in 2003–2023.

Category	Change Level	2003–2008		2008–2013		2013–2018		2018–2023
		Ratio	Subtotal Ratio	Ratio	Subtotal Ratio	Ratio	Subtotal Ratio	Ratio	Subtotal Ratio
Deteriorated	Deteriorated rapidly	3.69%	18.12%	4.18%	19.84%	3.91%	25.27%	4.87%	25.02%
	deteriorated slowly	14.42%		15.65%		21.36%		20.14%
Unchanged	Unchanged	33.80%	33.80%	42.58%	42.58%	37.95%	37.95%	47.30%	47.30%
Improved	Improved slowly	37.48%	48.08%	33.14%	37.58%	27.42%	36.78%	25.01%	27.69%
	Improved rapidly	10.60%		4.45%		9.37%		2.67%

, from 2003 to 2013, the area with improved ecological quality decreased from 48.08% to 37.58%, a reduction of 10.50%, while the area with deteriorating quality increased from 18.12% to 19.84%, an increase of 1.72%. Although the improved area was larger than the deteriorated area, the areas with the worst and poorer ecological quality increased the most (by 6.47% and 4.49%, respectively), while the area of moderate quality decreased the most (by 8.84%). Therefore, from 2003 to 2013, Nanjing’s ecological quality continuously declined, with the RSEI decreasing from 0.8049 (2003) to 0.7741 (2013). From 2013 to 2023, the improved ecological quality area further decreased from 36.78% to 27.69%, a reduction of 9.09%. Simultaneously, the area of deteriorated ecological quality slightly decreased, while the area with excellent level increased by 8.93%, leading to a slight improvement in overall ecological quality. Thus, ecological environment quality is influenced by both the area and the level of ecological quality.

The spatial distribution of RSEI changes in Nanjing from 2003 to 2023, at five-year intervals, is illustrated in Figure 7. Between 2003 and 2008, there were significant regional ecological environment changes. Areas with significant improvement (blue areas) were mainly loaded in the more remote regions such as Gaochun, Liuhe, and Pukou. In contrast, the central urban area experienced noticeable deterioration (red areas). Compared to the significant changes in the previous five years, from 2008 to 2013, the environmental quality of urban agglomerations continued to deteriorate, although the overall change in RSEI slowed and the affected area expanded. From 2013 to 2018, environmental quality in the main urban area significantly improved, and most areas, except for a few rural towns, became stable. From 2015 to 2020, apart from several rapidly developing areas such as the Jiangbei core development zone and the Hexi South development zone, the overall environmental quality remained stable.

At the administrative district level, central districts such as Gulou, Jianye, and Qinhuai had lower overall RSEI levels than peripheral districts like Lishui, Liuhe, Jiangning, and Pukou. Among them, Jiangning, Xuanwu, and Yuhuatai districts had a higher average RSEI level due to the abundance of forest land.

To further compare the changes in ecological environment quality in local space, this study used spatial autocorrelation analysis to express the spatial aggregation and differences of ecological quality in Nanjing. The results showed that the local spatial aggregation and differences in ecological quality (RSEI) in Nanjing were significant. Among them, the areas of high–high clustering were mainly concentrated in the forest land area (blue); the low–low clustering areas were mainly concentrated in the construction land area (purple); a few local areas would only have high–low clustering (red) and low–high clustering (green); most areas of farmland had no significant differences (gray), as shown in Figure 8.

4.3. Effects of Land-Use Changes Affecting Eco-Environment Quality

4.3.1. RSEI Differences of Land Types and PCA of RSEI

The results of the PCA of the four comprehensive indicators of RSEI indicate that the contribution rate of PC1 in all five years exceeds 60%, indicating that PC1 is representative. Among PC1, NDVI and WET are positive indicators, while NDBSI and LST are negative indicators. This result is similar to other research findings [27,45]. From the contribution of the four indicators to PC1, NDVI has the greatest impact on ecological quality than other indicators, indicating that vegetation-covered lands (mainly including farmland, grassland, forest land, etc.) are the main positive factor affecting the ecological environment in Nanjing, while construction land dominated by impervious surfaces is the main negative indicator. The degree of influence and dynamic changes of each indicator on the principal component of RSEI can provide a reference for local governments to plan a more targeted and scientific land-use index and protect the ecological environment (

Table 6. Principal Component Table of RSEI in Nanjing from 2003 to 2023.

Indicators	PC1 of RSEI
	2003	2008	2013	2018	2023
NDVI	0.824	0.7801	0.7536	0.8083	0.818
WET	0.3914	0.5043	0.571	0.4076	0.339
LST	−0.281	−0.2373	−0.166	−0.297	−0.2573
NDBSI	−0.298	−0.284	−0.2799	−0.3036	−0.3868
Characteristic value	0.2129	0.2326	0.2527	0.2816	0.3564
Contribution rate	67.54%	63.75%	62.45%	60.77%	66.68%

).

4.3.2. Correlation Analysis Between RSEI and Area of Land Types

Land-use data are discrete, making it impossible to directly establish a geographically weighted regression relationship with the continuous RSEI data. This study extracted the area, proportion of area, and average RSEI value data for different land use types from 171 towns and 11 counties in Nanjing between 2003 and 2023 and conducted regression analysis. The results show that forestland had the highest RSEI, at 0.8786, followed by farmland at 0.8318, while construction land had the lowest at 0.3790. Grassland and unused land were close, with values of 0.3822 and 0.3701, respectively. Due to potential anomalies in water body data during inversion, these were masked and excluded. According to other relevant studies [41,46,47], the ecological level of water bodies is similar to that of farmland. In this study, the RSEI value of the water area refers to farmland.

During the regression analysis of RSEI with land-use area at the township level, some townships had too little area for certain land-use types, limiting the fit of the RSEI model for land types other than farmland. However, the fit (R²) of the regression model between land use area and RSEI was higher at the county level than at the township level, indicating that area size needs to reach a certain threshold to impact environmental quality. As shown in

Table 7. Correlation analysis between RSEI and land area.

Correlation	RSEI	Area/km2
		Farmland	Forest Land	Grassland	Water	Unused Land	Construction Land	Total Area
RSEI	/
Farmland	0.656 **	/
Forestland	0.582 **	0.735 **	/
Grassland	0.138	0.422 **	0.658 **	/
Water	0.518 **	0.615 **	0.362 **	0.137	/
Unused land	0.195	0.406 **	0.621 **	0.775 **	0.133	/
Construction land	0.464 **	0.771 **	0.576 **	0.445 **	0.321 *	0.604 **	/
Total area	0.669 **	0.993 **	0.765 **	0.454 **	0.656 **	0.460 **	0.803 **	/

Note: *: p < 0.05, **: p < 0.01.

, RSEI had a significant correlation with farmland, forest land, water bodies, construction land, and total area (p < 0.01). Among land types, the total area and farmland area had the highest correlation, followed by construction land area. In addition to the total area, farmland area had the highest correlation with construction land area, followed by forest land area. Forest land area had the highest correlation with farmland, followed by grassland. Water body area had a significant correlation with farmland, and construction land area had a high correlation with farmland, followed by unused land area (Table 7). These results are consistent with the land-use transition matrix in Table 3.

4.3.3. Regression Analysis Between RSEI and the Area of Various Land-Use Types

The optimal regression fit between land-use area and RSEI is a logarithmic relationship. As land-use area increases, RSEI grows along a logarithmic curve. There is a threshold effect, where below the threshold, increasing land-use area rapidly improves ecological quality. Beyond the threshold, further increases in land-use area lead to a gradual leveling off of improvements in ecological quality. The regression fit results are shown in Figure 9,

Table 8. Results of the regression analysis between percentage of land-use area and RSEI.

Independent Variable (x)	Dependent Variable (y)	R2	Sig.	Threshold Value/km2
Farmland area	y = 0.0255 ln(x) + 0.6333	0.6138	<0.001	400–500
Forestland area	y = 0.0144 ln(x) + 0.7299	0.5494	<0.001	20–30
Water area	y = 0.0264 ln(x) + 0.6674	0.4028	<0.001	100–120
Construction land area	y = 0.0472 ln(x) + 0.5524	0.2591	<0.001	120–150
Total area	y = 0.0364 ln(x) + 0.5473	0.5595	<0.001	600–800

Note: Dependent variable y = RSEI.

. These findings are similar to other research results [48,49,50].

4.3.4. Regression Analysis Between RSEI and the Area Proportion of Various

Land-Use Types

This study analyzed the area percentage of various land-use types across different towns and counties in Nanjing over a 10-year period and established a correlation and regression model with RSEI.

As shown in

Table 9. Correlation analysis between RSEI and percentage of lands.

Correlation	RSEI	Percentage of Land Area/%
		Farmland	Forestland	Grassland	Water	Unused Land	Construction Land
RSEI	--
Farmland area%	0.763 **	--
Forestland area%	0.438 **	0.052	--
Grassland area%	−0.385	−0.229	−0.006	--
Water area%	−0.078	0.039	−0.382 **	−0.177	--
Unused land area%	0.011	−0.052	0.056	0.461 **	−0.075	--
Construction land area%	−0.805 **	−0.946 **	−0.228	0.262	−0.219	0.053	--

Note: **: p < 0.01.

, among individual land-use types, unused land and grassland were excluded from consideration due to their very low area proportions and insignificant correlation with RSEI. Water bodies were also excluded as they were masked during the RSEI analysis, resulting in no significant correlation and statistically insignificant regression results (p = 0.5030).

A significant linear regression relationship was found between the percentages of land area and RSEI. As indicated in

Table 10. Results of OLS and ridge regression.

Percentage of Land Area	OLS Model (k = 0)				Model 2 (k = 0.10)
	B	Beta	SE	VIF	B	Beta	SE	VIF
farmland area%	−0.146 *	−0.464	0.057	70.64	0.096 ***	0.306	0.005	0.563
forestland area%	−0.022	−0.027	0.056	10.24	0.195 ***	0.241	0.016	0.800
Water area%	−0.302 **	−0.469	0.056	16.42	−0.058 ***	−0.09	0.012	0.759
construction land area%	−0.365 **	−1.319	0.056	90.33	−0.113 ***	−0.409	0.004	0.477
_cons.	0.987 **	--	0.056		0.742 ***	--	0.004	--
R2	0.674				0.663
Adjust R2	0.672				0.661
MSE	0.056				0.010

Note: *: p < 0.05, **: p < 0.01, ***: p < 0.001. Dependent variable y = RSEI, Ridge regression can solve the multicollinearity problem of indicators.

, Model 1 is the result of a multivariate linear regression using ordinary least squares (OLS). However, it can be observed that high multicollinearity exists among the independent variables such as the area proportions of farmland, forest land, water bodies, and built-up land (VIF > 10), making OLS multivariate regression unsuitable. To address this issue, this study applied ridge regression to fit Model 2. Ridge regression is a nonlinear regression method designed to handle data with high multicollinearity. It not only mitigates multicollinearity among variables but also retains as much information from each variable as possible, making the model more practical and reliable. In comparing the two models, Model 2 shows lower multicollinearity and a better fit. It reflects the impact of the area proportions of farmland, forestland, water, and construction land on RSEI. The area proportions of farmland and forest land positively influence ecological quality, while the area proportions of water bodies and built-up land have negative impacts. After standardizing the coefficients, the area proportion of built-up land was found to have the largest negative impact on ecological quality (−0.409), followed by farmland (0.309), forest land having the smallest positive impact (0.241), and water bodies exerting the least influence (−0.09) (Table 10).

A possible explanation is that both farmland and forestland are vegetated areas, and an increase in vegetative cover leads to higher NDVI and WET values, thereby improving ecological quality. On the other hand, with increasing urbanization, the expansion of built-up land raises NDBSI and LST values, leading to the degradation of ecological quality. Water bodies, having been masked during the RSEI calculation, show an insignificant correlation with RSEI (−0.078, p > 0.05), with a weak explanatory power in the regression (R² = 0.0006) and a minimal impact on RSEI (−0.09) (Figure 10,

Table 11. Results of the percentage of land-use area and RSEI regression.

Independent Variable (x)	Dependent Variable (y)	R2	Sig.	SE
Percentage of farmland area	y = 0.2236 x + 0.6459	0.5032	<0.001	0.066
Percentage of forestland area	y = 0.3365 x + 0.7266	0.1728	<0.001	0.085
Percentage of water area	y = −0.0162 x + 0.7488	0.0006	0.5030	0.094
Percentage of construction land area	y = −0.2104 x + 0.8293	0.5774	<0.001	0.061

Note: Dependent variable y = RSEI.

).

Grassland and unused land were excluded from the analysis due to their minimal area proportions and weak influence on RSEI values.

In summary, the regression analysis between land-use area proportions and RSEI indicates that urban development and the E_n pose a risk of overall ecological quality degradation.

4.3.5. Regression Analysis Between RSEI and LUCI

LUCI reflects the comprehensive impact of human activities on various land-use types and their utilization intensity. It is the result of the combined effects of human activities and natural factors on different land-use types. To further clarify the quantitative relationship between RSEI and LUCI, this study extracted RSEI and LUCI data from 171 towns across 11 administrative districts in Nanjing for the period between 2003 and 2023. A multiple regression equation was established (Equation (8)). After adjustment, the model’s R² was 0.711, indicating a good fit of the regression equation.

RSEI = −0.0022 LUCI + 1.4716 (R² = 0.711)

(8)

Based on the scatter plot and fitting equation (Equation (8)), it is evident that there is a significant negative linear correlation between RSEI and LUCI. As urbanization progresses and the proportion of built-up land gradually increases, land-use intensity also rises, leading to a continuous decline in ecological quality.

To validate this fitting model, the study estimated the land area for 2023 based on the E_n of land types between 2018 and 2023. The LUCI for 2023 was then calculated and substituted into Equation (8) to estimate the 2023 RSEI values. These were compared with the actual RSEI values measured for 2023, as shown in

Table 12. Validation and prediction of RSEI and LUCI fitting models.

Year	Percentage of Area/%				En/%				LUCI	RSEI Observed Value	RSEI Model Value
	FA	FO	WB	CO	FA	FO	WB	CO
2003	72.99	7.68	8.85	10.48	−1.20	0.55	4.18	4.38	293.92	0.8049	0.7909
2008	68.62	7.89	10.70	12.78	−1.07	−1.42	2.17	4.77	294.12	0.7901	0.7906
2013	64.94	7.33	11.86	15.83	−0.55	−3.77	0.02	4.04	296.49	0.7741	0.7866
2018	63.15	5.94	11.87	19.02	−0.16	0.52	−2.44	1.89	301.18	0.8110	0.7786
2023	62.65	6.10	10.42	20.82	−0.16	0.52	−2.44	1.89	304.29	0.8108	0.7733
2028	62.54	6.13	10.17	21.21	−0.16	0.52	−2.44	1.89	305.07	/	0.7720

Note: FA: farmland, FO: forestland, WB: water body, CO: construction land.

. The predicted and observed RSEI values from 2003 to 2023 show over 95% consistency, indicating that the model has good accuracy.

This model holds potential value for predicting future land use and ecological quality. However, in terms of prediction time span, the land-use changes from 2003 to 2023 reveal frequent shifts in land use in rapidly urbanizing areas, with many influencing factors. If the time span is too long, it may affect the prediction’s accuracy [51]. In this study, a 5-year interval is used to predict that the ecological quality in Nanjing will slightly decline in the future.

Based on the expansion intensity of the various land-use types mentioned above, the future trends in land-use changes were predicted. Using the regression model between LUCI and RSEI, the future RSEI values for Nanjing were forecasted. The predictions indicate that by 2028, Nanjing’s LUCI will reach 305.07, and the RSEI will be 0.7720. Compared to 2023, land-use intensity will have slightly increased, while ecological quality will have slightly declined (Table 12).

5. Discussion

5.1. Measurement and Inversion of Eco-Environmental Quality

The quantitative assessment of ecological environmental quality holds significant theoretical and practical value for the protection of human living environments and the prevention of ecological risks. Early evaluation models faced challenges such as difficulties in data acquisition and insufficient accuracy of remote sensing data, leading to various shortcomings [41]. Among these, RSEI is an effective model for quantitatively evaluating ecological environmental quality. This index model utilizes high-resolution remote sensing imagery and employs principal component analysis (PCA) for dimensionality reduction, extracting four primary factors: NDVI, WET, NDBSI, and LST, which effectively reflect the ecological quality of terrestrial regions. Compared to other index models, RSEI boasts advantages such as ease of data acquisition, high accuracy of results, and broad applicability. However, a notable limitation of this model is its inability to assess the ecological quality of water bodies [33]. Although subsequent researchers proposed the water benefit ecological index (WBEI) model suitable for aquatic areas, this model overemphasizes the influence of water bodies, resulting in lower values for terrestrial regions, which still requires further refinement [49,51]. Considering that most areas of Nanjing are predominantly land, with a relatively small water body area (11.40%), this study still opts to use RSEI to invert the ecological environmental quality of Nanjing. The four principal factors of RSEI indicate that NDVI represents vegetation coverage, directly corresponding to land covered by vegetation, and is a key factor affecting regional ecological quality. WET and LST represent the main climatic factors influencing vegetation growth, while NDBSI indicates the intensity of human activities, corresponding to constructed land, which mainly includes buildings and impervious surfaces. This study shows that the RSEI values across different land use types follow the order: forest > farmland > grassland > constructed land, with vegetation coverage being a critical driver of RSEI values. Ecological environmental quality exhibits significant correlations with the area and proportion of various land types. RSEI is significantly positively correlated with the proportions of forest, farmland, and grassland and significantly negatively correlated with constructed land, which is consistent with findings from other similar studies [43,46,51].

5.2. Response of Ecological Environmental Quality to Land-Use Transfer

Research on the ecological effects of land use has largely focused on ecological services, ecological risks, and the urban heat island effect [32,52,53]; however, studies specifically addressing the changes in ecological quality caused by land-use transfer in rapidly urbanizing areas are relatively scarce. The results of this study indicate a high correlation between the spatiotemporal variations of RSEI and concurrent changes in land use. As urbanization progresses, urban areas continuously expand, significantly increasing the area of constructed land, most of which is converted from ecological land such as farmland and forests. In Nanjing, the increase in construction land indirectly leads to the reduction of ecological land. While local fluctuations in ecological environmental quality occur due to factors such as land policies, there is an overall declining trend, consistent with findings from other studies [53,54]. The driving mechanisms of land-use change on ecological environmental quality are often influenced by urbanization and land-use policies. Since 2000, rapid urbanization and urban–rural integration in Nanjing have accelerated the expansion of urban construction land. The swift development of real estate projects in several urban new districts, such as Hexi New Area, Jiangpu New Area, and Zidong New Area, has led to increased land-use intensity, peaking in 2013. In 2015, with the implementation of high-level green development strategies and farmland protection policies, the conversion of farmland and forests into construction land was restricted, slowing the growth of the real estate sector and urban expansion, which has allowed for a slight recovery in ecological environmental quality, although a general downward trend remains. Thus, government land policies dictate the pace of land-use conversion, and changes in land use lead to alterations in the area of ecological vegetation growth, with vegetation change being the core driver of shifts in ecological environmental quality [40]. Although land-use data are discrete and cannot be directly regressed with RSEI, this study conducts regression analyses using data on land-use proportions and intensity from 171 townships in Nanjing over the past decade, alongside corresponding RSEI data, providing valuable insights into the impact mechanisms of land-use changes on ecological environmental quality.

5.3. Optimization Suggestions for Land-Use Planning and Management in Nanjing

Based on this study’s predictions, the decline in ecological environmental quality in Nanjing is expected to slow over the next five years. This may be due to the city having undergone rapid development, resulting in a largely established urban structure, with construction land expansion tapering off. Additionally, recent policies aimed at farmland protection and curbing the rapid growth of the real estate sector have noticeably slowed the development of construction land, creating conditions favorable for ecological recovery. As a major economic hub in the Yangtze River Delta, Nanjing’s future urban development must consider the constraints of farmland and ecological protection boundaries, limit urban development peripheries, strengthen land-use planning and management, and maintain regional ecological quality for sustainable, high-quality growth. The following recommendations are proposed: (1) Protect forest vegetation and reduce farmland loss. In Nanjing, restore damaged natural forests through ecological restoration and afforestation, improve vegetation systems, delineate ecological protection zones, and safeguard ecological boundaries; while reducing farmland loss and protecting arable land, enhance protective forest networks around farmland to improve environmental quality. (2) Increase riparian wetlands, green spaces, greenways, and protective forests along water bodies to enhance vegetation coverage and improve ecological quality. (3) Clearly define urban development boundaries to prevent excessive expansion of construction land. Within these boundaries, implement rooftop gardens, vertical greening, and increase three-dimensional greening to boost green land ratios, thereby enhancing vegetation coverage and overall environmental quality.

5.4. Limitations

This study focuses on Nanjing’s land transfer and its effects on environmental ecological quality, aiming to provide a theoretical basis and reference for optimizing land-use planning and management to enhance regional ecological quality. However, this research has several limitations:

(1)

Due to data source constraints, the 30 m land-use data, while capable of distinguishing most land-use types, does not effectively separate green spaces within constructed areas in urban centers, in particular the lack of fine-grained classification of urban green spaces, which affects the accuracy of data extraction. Future research could benefit from employing machine learning methods to classify land-use data with 1 m and sub-meter accuracy. In particular, fine-grained classification of urban green space has important value for obtaining accurate urban green space data and is an important research direction in the future [55].

(2)

The inversion of RSEI shows significant seasonal variations. Although this study attempts to use remote sensing data from the same or similar months, data source limitations affect consistency. Additionally, because RSEI cannot assess the ecological quality of water bodies, which significantly impacts overall ecological quality, the inversion model has inherent limitations and may incur errors. Future studies should focus on optimizing RSEI to address the inability to assess water bodies.

(3)

Regarding future predictions of land-use changes, this study’s projections are based on fitted models and trends from the past five years. These predictions may be influenced by other factors, potentially reducing accuracy. Future research should incorporate comprehensive land-use forecasting models such as FLUS and PLUS to enhance the robustness of predictions in subsequent studies [56,57].

6. Conclusions

This study utilizes Landsat 8 remote sensing images from 2003, 2008, 2013, 2018, and 2023 to derive the remote sensing ecological index (RSEI) for Nanjing, integrating corresponding land-use data. It investigates how changes in land use affect ecological quality amidst rapid urbanization, aiming to inform land-use planning and management to balance economic development with ecological protection. Key conclusions include the following:

(1)

From 2003 to 2023, Nanjing experienced significant changes in land area and utilization. Construction land saw the largest increase, while agricultural land faced the most substantial reduction. LUCI steadily increased, while RSEI continued to decrease.

(2)

RSEI can well reflect the eco-environmental quality of rapid urbanization areas. Significant correlations (p < 0.05) and spatial autocorrelations were observed between land use changes and RSEI fluctuations. High ecological quality primarily clusters in forested regions, while low-quality areas are mainly in built-up zones; rural and suburban areas exhibit better ecological quality than urban centers. This conclusion is similar to other studies [39,47,49]

(3)

Under the policy background of building ecological civilization and green high-quality development, the monitoring of land-use transfer and ecological environment quality is particularly important for the construction of digital cities in rapidly urbanizing areas. On the one hand, they can monitor the direction and quantity of all kinds of land transfer in real time so as to scientifically formulate the land-use indicators of all kinds of land and provide a theoretical basis for protecting cultivated land, limiting the excessive expansion of construction land and realizing sustainable development; On the other hand, using remote sensing technology to retrieve the regional ecological quality is of great practical value to discover the weak areas of ecological quality in time and eliminate ecological risks, which is worthy of promotion and application by local governments in daily land management.