Coordination Analysis between the Development of Urban Built-Up Areas and Urban Environmental Factors through Remote Sensing of Nighttime Lights: A Case Study in Nanjing, China

Abstract

The fast development of urban built-up areas in China is causing many problems, such as pollution, congestion, etc. How to effectively evaluate the coordination between urban areas and environmental problems has been attracting many scholars worldwide. This paper intends to discover this “secretary” through investigating the built-up areas and their accompanied economic and environmental factors over almost 30 years (1992 to 2020) in Nanjing, China. DMSP/OLS nighttime lights images from 1992 to 2013 and the NPP/VIIRS nighttime lights images from 2012 to 2022 are used for extraction of built-up areas. A spatiotemporal evolution model is established to evaluate whether the built-up areas have developed in coordination and the relationship between urban built-up areas and various factors, including compactness, the fractal dimension, boundary and shape changes, exhaust emissions, and the production of general industrial solid waste, which was further investigated to ascertain whether there was coordination or not. The investigated results discovered that Nanjing’s built-up areas had maintained continuous growth from 1992 to 2020, with the compactness of built-up areas gradually decreasing from 0.42 to 0.23 and the built-up differentiation dimension changing from 1.31 to 1.39, demonstrating that built-up areas had gradually moved from a loose pattern to a compact pattern and from irregular development to balanced development in all directions. The macro model of the coordination index change trend is 0.847 from 1995 to 2020, which indicates that the coordination between urban built-up areas of development and their environments has been improving; however, the reduction in urban green space, the increase in waste emissions, and the increased production of general industrial solid waste has raised questions regarding sustainable development.

1. Introduction

With the sharp increase in urban built-up areas, a large amount of capital and a significant portion of the population have rapidly gathered in cities, resulting in problems such as a decrease in the amount of cultivated land, population congestion, traffic congestion, energy waste, and urban pollution. Many scholars have investigated the deep relationship between built-up areas and the various economic and social parameters in megacities, finding that significantly built-up areas had expanded alongside associated environmental effects. With the gradual finalization of megacities, built-up areas were no longer expanding, and more and more medium-sized and large cities were being included in the research scope. In this situation, analyzing iconic cities can help with future planning strategies and achieve effective urban expansion paths, which is crucial for ensuring rapid economic growth while protecting the ecological environment.

Numerous studies have analyzed urbanization and environmental change through the use of qualitative and quantitative methods. Lin [1] analyzed the impact of urbanization on the ecological environment by using a comprehensive assessment system of urbanization quality that includes multiple indicators, such as urban population and urban economy. Hong [2] used the city’s fuel-related statistical energy data to calculate spatiotemporal carbon emissions to answer the scientific question concerning whether CO₂ emissions could be offset by ecosystem CO₂ sequestration in the HCUA under China’s rapid urban expansion. However, existing research has seldom focused on the interaction mechanisms between urban environment and urban expansion.

Nighttime lights(NTL) data are widely used to analyze the relationship between urban expansion and various urban parameters. In Lu’s [3] study, the urban areas of the West Taiwan Strait Urban Agglomeration (WTSUA) were extracted using NTL imagery from 1992 to 2013. The spatiotemporal characteristics and pattern of urban sprawl were quantitatively analyzed by combining an urban expansion rate index and a standard deviation ellipse model. Yang [4] proposed a universal way of depicting the relationship between land cover and NTL data within an administrative city and then developed a Spatial-Socioeconomic Urban Development Curve to monitor urban development status. Wang [5] used NTL data to structure a research framework for urban agglomerations, which was based on four major urban agglomerations to explore their spatiotemporal characteristics and offers insights for government urban planning. Zhao [6] used the NPP-VIIRS-like NTL dataset for the period of 2001 to 2020 to spatialize GDP data and to quantify the pixel-level spatiotemporal patterns and trends in the spatialization of these data in Henan Province. Zhou [7] established a back propagation neural network based on genetic algorithm optimization (GABP) coupled with NPP/VIIRS NTL datasets to estimate the CO₂ emissions of China’s three major urban agglomerations at 500 m resolution from 2014 to 2019. Zuo [8] regressed the NTL data of more than 30 provinces in China with total carbon emissions and found a positive correlation between NTL data and carbon emissions. Xia [9] proposed a neural network method that fits NTL data with carbon dioxide emissions and evaluates and analyzes the spatiotemporal changes in carbon emissions.

Existing research mostly focuses on analyzing the relationship between the expansion of built-up areas and single atmospheric pollutants such as CO and CO₂ (for example, Zhou [10,11]) while ignoring other pollution indicators such as urban wastewater discharge and the production of general industrial solid waste. Similarly, urban expansion is not only an expansion of the dimensions of built-up areas, but also an expansion of dimensions such as infrastructure construction, economy, and population. Therefore, this article analyzed the coordination between parameters related to the expansion of built-up areas and parameters associated with changes in environmental parameters to evaluate the coordination of urban development by establishing an environmental parameter coordination index. Taking Nanjing as an example, validation was conducted using nighttime lighting data and statistical data. Through the analysis results of this article, it was possible to evaluate the coordinated development of urban expansion and environmental parameters in Nanjing, help identify shortcomings in development, and provide a reference for the development of other cities.

2. Study Area and Data Preprocessing

2.1. Study Area

Nanjing, located at 31°14′–32°36′N and 118°22′–119°14′E, is an important central city in the Yangtze River Delta urban agglomeration (Figure 1). It has 11 districts (Xuanwu, Qinhuai, Jianye, Gulou, Yuhuatai, Qixia, Jiangning, Pukou, Liuhe, Lishui, and Gaochun) and an administrative area of 6587 km.

2.2. Datasets and Data Preprocessing

2.2.1. Datasets

DMSP/OLS nighttime lights data: The defense meteorological satellite program (DMSP) of the United States is operated by the space and missile systems center of the United States Air Force.

DMSP/OLS night light images can reflect comprehensive information. A total of 34 images from 1992 to 2013 were obtained from the National Geographic Information Center (NOAA National Centers for Environmental Information (NCEI)).

NPP/VIIRS nighttime lights data: The NPP (National Polar-Orbiting Partnership) satellite system is jointly developed by NASA and the NOAA and is mainly used for environmental monitoring. The NPP satellite system includes five sensors (ATMs, CRLs, OMPs, Ceres, and VIIRS). A total of nine images from 2012 to 2020 were obtained from the Colorado School of Mines (VIIRS Nighttime Light (mines.edu (accessed on 1 December 2022))).

Landsat data: The United States Land Satellite (LANDSAT) series of satellites is jointly managed by NASA and the United States Geological Survey (USGS). Since 1972, eight satellites have been launched successively. The sensor carried by Landsat 1–5 is the MSS, Landsat 7 carries ETM + sensors, and Landsat 8 possesses OLI and TIRS sensors. This article uses the Landsat 5 (1992–2013) dataset and the Landsat 8 (2013–2020) dataset for analysis.

Statistical yearbook data: The Jiangsu Province Bureau of Statistics (http://tj.jiangsu.gov.cn/ (accessed on 1 December 2022)) provided access to Nanjing data from 1992 to 2020.

The specifications of satellite parameters are listed in

Table 1. DMSP/OLS satellite and NPP/VIIRS satellite parameters.

Parameter	DMSP/OLS	NPP/VIIRS
Time (Year)	1992–2013	2012–Now
Spatial resolution	2.7 km	740 m
Imaging width	3000 km	3000 km
Spectral range	0.5–0.9 μm	0.5–0.9 μm
Spectral resolution	6 bits	14 bits

.

2.2.2. Preprocessing of Nighttime Lights Data

This article preprocessed data according to the steps in Figure 2. The major preprocessing steps are shown below.

DMSP/OLS data correction:

Duo to discontinuity and saturation of the pixel light intensity values in DMSP/OLS datasets [12,13,14,15], this article preprocessed data through the following steps.

Projection transformation and regional clipping: DMSP/OLS nighttime lights data were converted into Lambert projection nighttime lights and resampled with a resolution of 1 KM × 1 KM.

Extraction of stable bright light area: By assuming that F182013 nighttime lights data were stable bright, we extracted and analyzed the data of the previous year according to this data, and thus obtained a long time series of regional stable bright pixel data in China.

Mutual calibration between sensors: In this experiment, Hegang City of Heilongjiang Province was selected as the pseudo invariant target area. The 34-phase stable bright metadata were extracted and the DN value was accumulated using F16 as the standard sensor to conduct data correction for other sensors.

The specific steps of mutual correction are as follows:

DN_FA = a × DN²_FB + b × DN_FB + c

(1)

where DN_FA is the sum of metadata for FA sensor data and DN_FB is the sum of metadata for FB sensor data. a, b, and c are the parameters of the quadratic equation in one variable.

Continuity correction: In order to ensure the continuity and uniqueness of data, the following corrections were made.

The calculation formula is:

$$DNF=\frac{(DNFA+DNFB)}{2}$$

(2)

where DNFA and DNFB are the image DN values obtained by two different sensors in the same year and DNF is the DN value of the year after correction.

In practice, the light region and light value of the last year should not be less than that of the previous year on the nighttime lights image. The other data were corrected by the following equation.

$${DN}_{n+1}=\left\{\begin{array}{c}{DN}_n, {DN}_n>{DN}_{n+1}\\ {DN}_{n+1}, others\end{array}\right.$$

(3)

NPP/VIIRS data correction:

This paper chooses annual VNL V2 datasets, which were corrected with the steps below.

Projection transformation and region clipping: The NPP/VIIRS nighttime lights data projection was converted into Lambert projection and resampled with a resolution of 1 KM × 1 KM.

Extraction of unstable bright areas: The downtown of the study area was taken as the threshold area, and all areas with pixel values greater than this area were extracted and compared with urban Landsat images. At the same time, the DN values of these areas were modified to pixel values of built-up areas.

Building a DMSP/ORS-like image dataset: In order to make NPP/VIIRS data closer to DMSP/OLS data, the NPP/VIIRS data smoothed using a low-pass filter were imported into the fitting model and fitted with DMSP/OLS data. The analysis results showed that the logarithmic model had the highest fitting degree, so the logarithmic model was used to calculate the NPP data.

The calculation formula is:

y = 5.0185ln(x) + 33.061

(4)

After correction, the determination coefficient (R²) between the NPP/VIIRS image DN value and the DMSP/ORS image DN value was increased from 0.67 to 0.80, greatly improving the fit and enhancing the continuity of the image. Nighttime lights data before and after correction are shown in Figure 3.

2.2.3. Determination of the Expansion of Built-Up Areas

According to the characteristics of nighttime lights data, the higher the DN value in the data, the greater the probability of belonging to a built-up area [16,17,18,19]. This paper set up a DN threshold value, used dichotomy to continuously change the threshold value, and compared it with statistical yearbook data to calculate the difference. The minimum difference was obtained after several iterations, with the corresponding one being the threshold value. After continuous adjustment, the accuracy of the final extraction of built-up areas was 88.55%, which could thus be used for actual analysis.

3. Research Methods

The methods related to the formulas used for analysis in this article are as follows.

3.1. Urban Areas Expansion Index

To objectively describe the speed and morphological changes in urban expansion from a spatial perspective, we used the following four indicators for calculation, which may be expressed as follows:

Expansion speed ($R_v$) (Zhou [20]):

$$R_v=\frac{U_b-U_a}{U_a}\times \frac{1}{T}\times 100\mathrm{\%}$$

(5)

Expansion intensity ($R_m$) (Iu [21]):

$$R_m=\frac{U_b-U_a}{U}\times \frac{1}{T}\times 100\mathrm{\%}$$

(6)

Compactness (CI) (Wang [22]):

$$CI=\frac{2\sqrt{S\pi }}{C}$$

(7)

Fractal dimension (DI) (Zhang [23]):

$$DI=\frac{2(lnC-ln4)}{lnS}$$

(8)

Ua and Ub represent the area of built-up areas corresponding to time a and b. T represents the years between time a and time b. S represents the area of built-up areas. C represents the perimeter of urban areas.

3.2. The Change in Gravity Centers and Standard Deviational Ellipse

The change in gravity centers and the standard deviational ellipse can be used to analyze the morphological characteristics of the distribution of built-up areas, such as the scope, direction, and centroid; it can more intuitively describe the trend of urban expansion and determine the patterns and processes of changes in built-up areas.

Analysis of the change in the gravity center of built-up areas (Li [24,25]):

$$\mathrm{G}=\sqrt{{(X_a-X_b)}^2-{(Y_a-Y_b)}^2}$$

(9)

Standard deviational ellipse analyses of built-up areas (Yang [26]):

$$\mathrm{S}=\left(\genfrac{}{}{0pt}{}{var\left(x\right)cov\left(x,y\right)}{cov\left(y,x\right)var\left(y\right)}\right)=\frac{1}{n}\left(\genfrac{}{}{0pt}{}{\sum _{i=1}^n\widetilde{x_i^2}\sum _{i=1}^n\widetilde{x_i}\widetilde{y_i}}{\sum _{i=1}^n\widetilde{x_i}\widetilde{y_i}\sum _{i=1}^n\widetilde{y_i^2}}\right)$$

(10)

Xa,b and Ya,b represent the coordinates of point a and b. $\widetilde{x_i}$ and $\widetilde{y_i}$ are the average center and coordinate difference of xy.

3.3. The Spearman Correlation Coefficient

The Spearman correlation coefficient can effectively measure the relationship between two variables and has been widely used in economics, ecology, and social science research.

$$R=\frac{\sum _{i=1}^n\left({X1}_i-\stackrel{-}{X1}\right)({X2}_i-\stackrel{-}{X2})}{\sqrt{{\sum _{i=1}^n({X1}_i-\stackrel{-}{X1})}^2}\sqrt{{\sum _{i=1}^n({X2}_i-\stackrel{-}{X2})}^2}}$$

(11)

3.4. The Entropy Weight Method

The entropy weight method is an objective weighting method. In the specific use process, based on the degree of dispersion for the data for each indicator, the entropy weight of each indicator is calculated using information entropy, and the entropy weight is then modified according to each indicator to obtain a more objective indicator weight.

$$y_{ij}=\frac{z_{ij}}{\sum _{i=1}^n{x}_{ij}},e_j=k{\sum }_{i=1}^n{y}_{ij}{lny}_{ij}$$

(12)

$$w_j=\frac{1-e_j}{\sum _{i=1}^m(1-e_j)}$$

(13)

3.5. Analysis of Changes in Coordination Degree

The coordination index trend model can be used to analyze the coordination degree and change trend of two variables, and, by adjusting the timeline, different change patterns at different time scales can be explored.

Coordination index trend model (Sun [27]):

$$C=\frac{F_i}{F_j}÷\frac{E_i}{E_j}$$

(14)

E represents the comprehensive score of urban development. F represents the comprehensive score of urban environmental pollution. i and j represent the year of study.

3.6. Obstacle Factor Analysis

The obstacle model can be used to measure risk factors in order to conduct pathological diagnosis of regional environmental risks. The least square error model uses variance to reflect the effect of changes in sample data around the average value of the sample. The minimum deviation between the actual distribution and the theoretical distribution is defined as the least square, and category attribution analysis is performed on the sample data. By introducing the least square error model into the obstacle model, the obstacle score can be used to divide resistance patterns into subsystems in the indicator system and thus achieve analysis of resistance types.

Obstacle model (Zhang [28]):

$$O_{ij}=1-X_{ij},I_j=\frac{O_{ij}·w_j}{\sum _{j=1}^n{O}_{ij}·w_j}$$

(15)

Oij represents the gap between a single factor indicator and the system’s development goals, while Wj represents the weight of a single factor.

Least Square Error model (OECD [29,30]):

$$S^2=\frac{1}{n}{\sum }_{i=1}^n{\left(x_i-\stackrel{-}{x}\right)}^2$$

(16)

S² represents the variance, x_i represents the sample data in the calculation, and $\stackrel{-}{x}$ represents the average value.

3.7. Linear Propensity Estimation (Slope)

The linear propensity estimation method can obtain the propensity values of different years in the time series and analyze the change trend in samples.

$$S=\frac{n\sum _{i=1}^n{t}_i{E}_i-\sum _{i=1}^n{t}_i{\sum }_{i=1}^n{E}_i}{n\sum _{i=1}^n{E}_i^2-{\left(\sum _{i=1}^n{E}_i\right)}^2}$$

(17)

where n is the total number of years, E_i is the CO₂ emissions in year i, and x_i is year i.

4. Analysis and Results

4.1. Analysis of the Expansion of Built-Up Areas

4.1.1. Evaluation Index of the Expansion of Built-Up Areas

The extraction results and related parameters of the built-up areas are shown in

Table 2. Extracted results of the built-up area in the main years.

Year	Built-Up Area	DN Value	Extracted Perimeter/km	Extraction Area/km2
1992	148	42	98	134
1995	151	43	110	145
2000	201	50	144	227
2005	513	49	218	501
2010	619	53	260	588
2013	713	15	415	838
2015	755	14	444	909
2020	868	18	502	1041

and Figure 4 and Figure 5.

To explore the speed and intensity of urban expansion and evaluate its coordination [31,32,33,34], four factors were used: expansion rate, expansion intensity, compaction degree, and fractal dimension. The results of these factors are shown in Figure 6.

Results:

It can be seen from Figure 6 that from 1992 to 1995, the built-up area grew slowly, with an expansion rate of 2.74 and an expansion intensity of 0.06. From 1995 to 2000, Nanjing built-up areas entered a stage of rapid growth, with an expansion rate of 313% and an expansion intensity of 342%. From 2000 to 2005, the constructed area entered a stage of rapid growth. The expansion rate increased by 113% compared with the previous stage, and the expansion intensity increased by 235%. From 2005 to 2010, the growth rate of built-up areas slowed down; the expansion rate was 86% lower than that of the previous stage, and the expansion intensity was 68% lower. From 2010 to 2013, Nanjing built-up areas entered a rapid growth stage again, with an expansion rate of 308% and an expansion intensity of 379% compared with the previous stage. From 2013 to 2015, the growth rate of built-up areas slowed down again, with the expansion rate reduced by 70% and the expansion intensity reduced by 57% compared with the previous stage. From 2015 to 2020, the growth rate of built-up areas was stable, with an expansion rate of 2.90 and expansion intensity of 0.40, similar to the initial stage of urban expansion.

The density of built-up areas in Nanjing gradually decreased from 0.42 to 0.23, and the differentiation dimension of built-up areas slightly changed from 1.31 to 1.39.

Discussion:

From 1992 to 2020, a high growth rate for built-up areas was maintained in Nanjing, and expansion speed and intensity were at a high level; expansion changed from irregular and disorderly to comprehensive and balanced development, reaching a coordinated and orderly state. This coincided with Nanjing’s policy of promoting urban–rural integration, improving urban construction and strengthening regional ties.

4.1.2. Analysis of Spatiotemporal Changes in Built-Up Areas

The center of gravity model was applied to investigate the spatiotemporal changes of built-up areas in Nanjing [31,32,33,34]. The results are shown in

Table 3. Spatiotemporal changes in built-up areas of Nanjing.

Year	X Coordinates (E)	Y Coordinates (N)	Area/km2	The Offset Direction	Offset Distance/m
1992	118°46′12.59″	32° 4′30.93″	116	--	--
1995	118°46′24.35″	32°4′35.07″	148	East	412.11
2000	118°46′44.42″	32°3′33.33″	163	South	1942.11
2005	118°46′57.64″	32°3′19.08″	223	Southeast	550.58
2010	118°46′55.57″	32°3′17.38″	240	Southwest	74.43
2013	118°47′48.12″	32°1′8.27″	883	South	4143.07
2015	118°47′44.88″	32°1′0.77″	873	South	242.31
2020	118°47′36.22″	32°0′58.82″	888	West	230.74

, and the corresponding distributions are shown in Figure 7.

Results:

The whole city developed eastward from 1992 to 1995, developed southward from 1995 to 2000, developed southward from 2000 to 2005, developed southwest from 2005 to 2010, developed significantly southwest from 2010 to 2013, developed southward from 2013 to 2015, and developed less westward from 2015 to 2020.

As observed in

Table 4. Nighttime lights parameters and city parameters.

	1995	2000	2005	2010	2015	2020	Correlation
Nighttime light DN Value	54,725	71,010	94,285	134,406	130,600	188,035
Urban population	266	290	525	629	746	809	0.943 **
Urban non-agricultural GDP	540	1016	2376	5056	9783	14521	0.943 **
Urban compactness	0.39	0.37	0.36	0.33	0.24	0.23	−0.943 **
Urban electricity consumption	604,800	1,377,395	2,466,661	3,736,638	4,951,753	6,329,425	0.943 **
Passenger transport capacity	10,068	15,294	20,537	39,104	15,929	11,375	0.371429
Road length	1271	1802	6132	5599	7771	9335	0.829 *
Drainage facilities	1014	1370	3380	4948	8308	10,863	0.943 **
Household consumption of LPG	83,670	80,272	81,903	76,409	50,481	23,422	−0.886 *
Wastewater discharge	9	6	5	3	2	1	−0.943 **
Exhaust emissions	1602	2155	3754	5738	8782	9340	0.943 **
Soot emissions	5	5	5	3	8	2	−0.57682
Production of general industrial solid waste	542	652	1159	1657	1475	1888	1.000 **
Number of key pollution control projects	196	299	135	77	76	1056	−0.02857
Sewage treatment rate	37.4	63.6	81.2	88.8	95.7	97.9	0.943 **
Garbage treatment rate	100	85.76	87.46	78.74	100	100	−0.03036
Parks and green areas	1798	2250	6139	6773	9328	10981	0.943 **

** At the 0.01 level (double-tailed), the correlation is significant. * At the 0.05 level (double-tailed), the correlation is significant.

, it can be seen that the center of gravity of built-up areas moved 412.11 m to the east from 1992 to 1995, moved 1942.11 m to the south from 1995 to 2000, moved 550.58 m to the south from 2000 to 2005, moved 74.43 m to the southwest from 2005 to 2010, moved 4143.07 m to the south from 2010 to 2013, moved 242.31 m to the south from 2013 to 2015, and moved 230.74 m to the west from 2015 to 2020.

The standard deviation ellipse area of built-up areas increased by 32 km² with an annual growth rate of 9.2% from 1992 to 1995, increased by 15 km² with an annual growth rate of 2.0% from 1995 to 2000, increased by 60 km² with an annual growth rate of 7.36% from 2000 to 2005, increased by 17 km² with an annual growth rate of 1.52% from 2005 to 2010, increased by 643 km² with an annual growth rate of 89.31% from 2010 to 2013, decreased by 10 km² with an annual growth rate of −0.57% from 2013 to 2015, and increased by 15 km² with an annual growth rate of 0.34% from 2015 to 2020.

From 1992 to 2020, the overall gravity center of Nanjing built-up areas moved from the northwest to the southeast, and the standard deviation ellipse of built-up areas also showed an increasing trend from northwest to southeast.

Discussion:

From 1995 to 2010, Nanjing put forward an urban construction plan for cross-river development, and the Pukou area developed rapidly as a result. From 2010 to 2020, Nanjing proposed a construction plan for the Southern New Area, and the Jiangning, Lishui, and Gaochun regions have developed rapidly as a consequence. In general, the direction of Nanjing’s urban gravity center and standard deviation ellipse was the same as that of urban planning policy; the urban development focus of Nanjing was developing from north to south, and the development area was gradually shifting from Pukou to Jiangning, Lishui, and Gaochun.

Overall, the direction of Nanjing’s urban gravity center and standard deviation ellipse was consistent with the direction of urban planning policies. From 1995 to 2010, Nanjing proposed the urban construction plan for cross-river development, and the Pukou area developed rapidly. During the same period, the standard deviation ellipse of Nanjing expanded violently towards the north. From 2010 to 2020, Nanjing proposed a construction plan for the Southern New Area, with rapid development in the Jiangning, Lishui, and Gaochun regions. During the same period, Nanjing’s urban gravity center shifted significantly southward, and the standard deviation ellipse expanded violently towards the south.

4.2. Analysis of the Expansion of Built-Up Areas vs. Environmental Parameters

Nighttime lightsnighttime lights

The development and expansion of built-up areas is a systematic process affected by many uncertain factors. Scholars [35] point out that the dynamic factors of urban spatial expansion include natural conditions [36], urban economic development [37], urban population growth [38], urban transportation network development [39], government policy intervention, etc., and the expansion of built-up areas in Nanjing is also driven and restricted by these factors. Along with our country in the 21st century, the development of science and technology and improvements to national planning regulations [40] have led to the natural conditions of urban construction direction gradually becoming smaller while the influence of urban economy, urban population, and government planning and policy factors that affect the direction of urban construction has increased.

The Spearman correlation coefficient can effectively measure the relationship between two variables and has been widely used in economics, ecology, and social science research, such as in the studies of Wang [41] and Zheng [42]. In order to unveil the correlations between nighttime lights data and the urban parameters in Nanjing, this article used nighttime lighting data as the independent variable and other urban parameters as the dependent variable and analyzed them separately. The formula for calculating the Spearman correlation coefficient is shown in Section 3.3, and the results are shown in Table 4.

According to the above analysis of urban social parameters, the correlation coefficient between the urban population, urban non-agricultural GDP, urban electricity consumption, drainage facilities, and nighttime lights was close to 1, and their changing trends were the same, thus jointly affecting urban expansion. The correlation coefficient between urban compactness and nighttime lights was close to −1, and their changing trend was opposite, thus affecting urban expansion in the opposite direction. The correlation coefficient between passenger transport capacity and nighttime lights was close to 0 and their trend similarity was low, which had a small impact on urban expansion.

In urban environmental parameters, the correlation coefficient between the production of general industrial solid waste and nighttime lighting was 1 and their changing trends were the same, thus jointly affecting urban expansion. The correlation coefficient between exhaust emissions, sewage treatment rate, parks and green areas, and nighttime lights was close to 1 and their changing trends were the same, thus jointly affecting urban expansion. The correlation coefficient between household consumption of LPG, wastewater discharge, and nighttime lights was close to −1 and their changing trend was opposite, thus affecting urban expansion in the opposite direction. The correlation coefficient between the number of key pollution control projects and nighttime lights was close to −0 and their trend similarity was low, which had a small impact on urban expansion.

In order to explore the degree of coordination in urban development, this article selected 17 indicators possessing a high correlation coefficient with nighttime lights from two aspects for analysis, explored the trends in changes between urban expansion and the urban environment, and provided suggestions for urban development.

4.2.1. Weight Analysis

The remote sensing images applied in this paper for extraction of built-up areas, combined with urban parameter indicators, build an evaluation index system for compositing the degree of coordination in urban expansion and the urban environment; the entropy method was applied in the calculation of a standardized matrix [43,44,45,46], information entropy, and feature weight. The formula for calculating the entropy method is shown in Section 3.4. The calculated feature weights are listed in

Table 5. Index weight values for urban expansion and environmental parameters.

Indicator	Feature Weight	Indicator	Feature Weight
Compactness	0.19	Average noise value	0.11
Population density in built-up areas	0.09	Wastewater discharge	0.12
GDP per square kilometer of built-up areas	0.21	Exhaust emissions	0.18
Road density	0.13	Soot emissions	0.11
Student density in urban built-up areas	0.13	Production of general industrial solid waste	0.17
Electricity consumption	0.10	Sewage treatment rate	0.11
Drainage facilities	0.15	Proportion of green spaces in built-up areas	0.19

.

As shown in Table 5, it can be seen that among the urban development factors, GDP per square kilometer of built-up areas had the highest weight, accounting for 21%, thus indicating it to be the most important factor affecting urban development. Second was compactness and third was drainage facilities, followed by road density, student density in urban built-up areas, and electricity consumption. The factor with the lowest proportion was population density in built-up areas, which was only 7%, thus having a relatively small impact on urban development.

Among urban environmental factors, the proportion of green spaces in built-up areas had the highest weight, accounting for 18.8%, thus indicating that it was the most important factor affecting the urban environment. Second was exhaust emissions and third was the production of general industrial solid waste, followed by wastewater emissions, average noise value, and sewage treatment rate. The minimum weight proportion was found for soot emissions, which was only 11%, thus having a relatively small impact on the urban environment.

4.2.2. Trend Analysis

To explore changes in the degree of coordination over time, this article used analysis of changes in the degree of coordination to analyze trends [47,48]. The formula for the coordination index trend model is shown in Section 3.5. The results are shown in

Table 6. Trends in urban development and environmental parameters.

	1995	2000	2005	2010	2015	2020
Environmental pollution	1.10	1.30	1.10	1.187	0.93	2.294
Urban development	1.19	1.27	1.29	0.98	1.11	2.71
Microscopic	0.93	1.03	0.85	1.21	0.84	0.93
Macroscopic	1	0.96948	0.93896	0.90844	0.87792	0.84739

.

From 1995 to 2020, the macro and micro changes in the coordination index between urban development and environmental parameters showed a downward trend, which indicates that the coordination between urban development and environmental parameters had been improving. At the same time as urban development, environmental parameters had been relatively effectively controlled. Compared with the long-term coordination curve, the short-term coordination curve exhibits more frequent and significant changes. There was no long-term planning for urban development and environmental parameters, and environmental parameters were arbitrary and lacked supervision.

From 1995 to 2020, the trend in changes according to the macro model of the coordination index was 0.847, indicating that the degree of coordination between urban development and environmental pollution in 2020 was 84.7% compared to 1995, indicating an overall improvement in the degree of coordination.

From 1995 to 2000, the trend in changes according to the micro model of the coordination index was 0.932, indicating an improvement in the degree of coordination. From 2000 to 2005, the trend for changes in the coordination index was 1.027, indicating a deterioration in coordination. From 2005 to 2010, the trend for changes in the coordination index was 0.85444, indicating an improvement in coordination. From 2010 to 2015, the trend for changes in the coordination index in Nanjing was 1.213, indicating a deterioration in coordination. From 2015 to 2020, the trend for changes in the coordination index was 0.843, indicating an improvement in the degree of coordination.

4.2.3. Impact Factor Analysis

In order to explore the factors that have the greatest impact on environmental change, this article used the obstacle model and the minimum variance model to calculate the factors that have the greatest impact on the environment during each period [49,50,51,52]; the formula for obstacle factor analysis is shown in Section 3.6, and the results are shown in Figure 8.

Comparing the number of occurrences for the seven indicators, sewage treatment rate only became a source of obstacles twice and had the smallest impact on co-scheduling. Average noise value, wastewater discharge, and soot emissions each became obstacle sources three times, with a moderate impact on coordinated scheduling. The amount of exhaust emissions and the production of general industrial solid waste each became obstacles four times, which had a significant impact on coordinated scheduling. The proportion of green spaces in built-up areas became a source of obstacles five times, thus having the greatest impact on coordinated scheduling and highlighting it as an important factor causing disharmony.

In 1995, three indicators became obstacles to urban environmental development, with average noise value being the largest indicator, followed by wastewater discharge and sewage treatment rate.

In 2000, five indicators became obstacles to urban environmental development, with the proportion of green spaces in built-up areas being the largest indicator, followed by wastewater discharge, sewage treatment rate, average noise value, and soot emissions. Comparing the five obstacle indicators for the year 1995, the proportion of green spaces in built-up areas had decreased by 20%, wastewater discharge had decreased by 27%, sewage treatment rate had increased by 70%, the average noise value had decreased by 2%, and soot emissions had increased by 3%.

In 2005, xix indicators became obstacles to urban environmental development, with the production of general industrial solid waste being the largest indicator, followed by average noise, wastewater discharge, exhaust emissions, soot emissions, and the proportion of green spaces in built-up areas. Comparing the five obstacle indicators for the year 2000, the production of general industrial solid waste had increased by 77%, the average noise value had increased by 2%, wastewater discharge had decreased by 27%, exhaust emissions had increased by 74%, soot emissions had decreased by 7%, and the proportion of green spaces in built-up areas had increased by 23%.

In 2010, three indicators became obstacles to urban environmental development, with the production of general industrial solid waste being the largest indicator, followed by exhaust emissions and the proportion of green spaces in built-up areas. Comparing the three obstacle indicators in 2005, the production of general industrial solid waste had increased by 42%, exhaust emissions had increased by 52%, and the proportion of green spaces in built-up areas had decreased by 5%.

In 2015, four indicators became obstacles to urban environmental development, with exhaust emissions being the largest indicator, followed by the proportion of green spaces in built-up areas, the production of general industrial solid waste, and soot emissions. Comparing the four obstacle indicators in 2010, exhaust emissions had increased by 53%, the proportion of green spaces in built-up areas had decreased by 10%, the production of general industrial solid waste had decreased by 10%, and soot emissions had increased by 149%.

In 2020, three indicators became obstacles to urban environmental development, with exhaust emissions being the largest indicator, followed by the production of general industrial solid waste and the proportion of green spaces in built-up areas. Comparing the three obstacle indicators in 2015, exhaust emissions had increased by 6%, the production of general industrial solid waste had increased by 8%, and the proportion of green spaces in built-up areas had increased by 2%.

4.2.4. Comparative Analysis of Cities

The research results obtained in this article were similar to Chen’s [53] research showing that the coordination between urban development and environment parameters in Nanjing had improved. However, the current environmental parameters still posed a challenge for achieving the goals of energy conservation, emissions reductions, and carbon peaking in Nanjing by 2030. The results obtained in this article were similar to Lu’s [54] and Lei’s [55] research, where solid waste and green spaces were the main obstacles to environmental development.

In order to conduct a deeper analysis of the two environmental obstacle sources in Nanjing, identify the differences between Nanjing and other cities, and provide suggestions for future development, this article selected 15 cities with similar urban scale levels to Nanjing and compared the production of general industrial solid waste and the proportion of green spaces in built-up areas for these cities in a time series. The comparison results are shown in Figure 9, Figure 10, Figure 11 and Figure 12.

As shown in Figure 9, according to the 25-year average for the production of general industrial solid waste, Nanjing was the highest. In each analysis period, except for 2015, Nanjing ranked at the highest level in terms of the production of general industrial solid waste.

SLOPE analysis was conducted on changes in the production of general industrial solid waste in the 16 cities; the formula for linear propensity estimation is shown in Section 3.7, and the analysis results are shown in Figure 10.

As shown in Figure 10, according to the results of trend analysis, Nanjing had the highest growth trend for the production of general industrial solid waste among the 16 cities. Compared to other cities, Nanjing had a higher growth rate for the production of general industrial solid waste and a larger growth volume.

As shown in Figure 11, among the 16 cities, the proportion of green spaces in built-up areas in Nanjing is at an average level, ranked sixth on average over the past 25 years. In 1995, the proportion of green spaces in built-up areas in Nanjing ranked third, while in 2020 it ranked second from last.

As shown in Figure 12, according to the trend analysis results, Nanjing has the lowest trend among the 16 cities for the proportion of green spaces in built-up areas. Compared to other cities, the proportion of green spaces is increasing over time, while Nanjing shows a decreasing trend.

In general, compared to other cities of the same level, Nanjing had more production of general industrial solid waste and a larger growth trend. In terms of the proportion of green spaces in built-up areas, Nanjing did not occupy a leading position and had a slow decreasing trend. They were all obstacles to the healthy development of Nanjing city.

5. Conclusions

This article used low-pass filtering and logarithmic models to process the NPP/VIIRS data, constructed a long-term series night scene lighting image dataset, and used this data to investigate and analyze the relationship between the rapid urban expansion of Nanjing City in China and its socio-economic factors. Based on the analyzed results, the following conclusions can be drawn.

(1) The amount of built-up areas in Nanjing maintained continuous growth from 1992 to 2020, while the compactness of built-up areas gradually decreased from 0.42 to 0.23 and the differentiation dimension for built-up areas changed from 1.31 to 1.39. This meant that built-up areas had gradually moved from a loose pattern to a compact pattern and from irregular development to balanced development in all directions. The center of gravity for Nanjing city migrated from a northwest to southeast direction, and the long axis of the standard deviation of Nanjing’s noumenal ellipse expanded from 9893 m to 26,733 m; the entirety of urban construction developed outward along the northwest–southeast axis.

(2) From 1995 to 2020, the macro model for the change trend of the coordination index was 0.847, which indicates that the degree of coordination between urban development and the environment had been improving. From 1995 to 2000, the micro model for the change trend of the coordination index change was 0.932, while from 2000 to 2005 it was 1.027, from 2005 to 2010 it was 0.854, from 2010 to 2015 it was 1.213, and from 2015 to 2020 it was 0.843, indicating an improvement in the degree of coordination. Compared with the long-term coordination curve, the short-term coordination curve showed more frequent and significant changes. There was no long-term plan for urban development and environmental pollution, and pollution emissions were arbitrary and lacked supervision.

(3) From 1995 to 2020, the sewage treatment rate only became a source of obstacles twice and had the smallest impact on coordinated scheduling. Average noise value, wastewater discharge, and soot emissions each became obstacle sources three times, thus having a moderate impact on coordinated scheduling. Exhaust emissions and the production of general industrial solid waste each became obstacles four times, which had a significant impact on coordinated scheduling. The proportion of green spaces in built-up areas became a source of obstacles five times, thus having the greatest impact on coordinated scheduling. Compared to other cities of the same level, Nanjing had more production of general industrial solid waste and a lower proportion of green spaces in built-up areas.

In general, the degree of coordination between urban development and the environment in Nanjing has been improving; however, the reduction in urban green space, the increase in waste emissions, and the increase in the production of general industrial solid waste have cast a shadow over the future development of Nanjing and have raised questions regarding sustainable development.

This article uses convolutional fitting models to construct DMSP/OLS-like nighttime lights images. Though processed image fitting accuracy (R²) reaches 0.80, there is still room to improve accuracy. In future research, convolutional neural networks can be considered for fitting to achieve better fitting results.

For future research, adding more parameter indicators and refining the development of various aspects of the city into various small areas for analysis should be considered.