• Dryness index (*Idry*)

The dryness index refers to the quantification of soil desiccation, which is a condition detrimental to the ecological environment. As most urban construction land is located in our study area, the dryness index can be represented by combining the bare soil index (SI) and the built-up index (IBI) into a normalized building–bare-soil index (NDBSI) [29]. We proposed extracting the bare soil and building area by setting an appropriate threshold and, subsequently, calculating the NDBSI as a weighted average and employing the area ratio as the weight.

$$NDBSI = (SI + IBI)/2\tag{10}$$

$$SI = \left[ (\rho 5 + \rho 3) - (\rho 4 + \rho 1) \right] / \left[ (\rho 5 + \rho 3) + (\rho 4 + \rho 1) \right] \tag{11}$$

$$IBI = \begin{bmatrix} \frac{2\rho 5}{\rho^5 + \rho 4} - \left(\frac{\rho 4}{\rho^4 + \rho 5} + \frac{\rho 2}{\rho^2 + \rho 5}\right) \end{bmatrix}$$

$$\begin{bmatrix} \frac{2\rho 5}{\rho^5 + \rho 4} + \left(\frac{\rho 4}{\rho^4 + \rho 5} + \frac{\rho^2}{\rho^2 + \rho 5}\right) \end{bmatrix} \tag{12}$$

where *ρ*1, *ρ*2, *ρ*3, *ρ*4 and *ρ*5 have been defined earlier in the context of the humidity index.

2.2.3. Water Mask and Standardization

The humidity index reflects the moisture of the vegetation and soil. The area covered by water in the study area occupies a large proportion of the *Iwet*, which reduces the advantage of vegetation and soil in *Iwet*. Therefore, the calculated *Iwet* is not a true reflection of the vegetation and soil moisture, and it is necessary to mask the water bodies present in the study area. We use a modified normalized difference water index (MNDWI) to mask these water bodies. The formula is:

$$MNDWI = (\rho\_{\text{Green}} - \rho\_{MIR}) / (\rho\_{\text{Green}} + \rho\_{MIR}) \tag{13}$$

where *ρGreen* represents the reflectance of the near-infrared band and *ρMIR* represents the reflectance of the red band.

2.2.4. Construction of the Improved Remote Sensing Ecological Index (IRSEI) Evaluation Model

First, we obtain the primary remote sensing ecological index based on PCA. The four indices are standardized to the range [0–1] and PCA is used to combine these indices. PCA1 is obtained from the four RSEIs to build a preliminary assessment model. Generally, the first PCA collects most of the information on the four indicators, and PC1 can be used to represent the characteristics of the regional ecological environment. Therefore, we use only one PC in further analyses. To facilitate index measurement and comparison, the initial RSEI is standardized, as follows:

$$RSEI\_{PCA} = 1 - f\left(I\_{\text{wet}} \mid I\_{\text{ndvi}} \mid I\_{\text{heat}} \mid I\_{\text{dry}}\right) \tag{14}$$

$$f = \sum\_{i=1}^{4} (e\_i \times PC1) \tag{15}$$

where *Indvi* represents the green component; *Iwet* represents the humidity component; *Iheat* represents heat; *Idry* represents dryness; and PC1 is the first principal component. The obtained RSEI value is within the [0–1] range. *ei* is the characteristic value contribution rate of the index corresponding to PC1. The closer RSEI is to 1, the better the UEQ of the region. The first principal component analysis index values are listed in Table 2. A detailed description of the calculation steps is available in the relevant literature [11,22,24,29].

**Year PC1 Eigenvalue Contribution/% Accumulation/%** 1995 NDVI 0.0441 88.6768 88.6768 WET 0.0048 9.5508 98.2276 NDBSI 0.0002 0.4187 98.6463 LST 0.0006 1.3537 100 2005 NDVI 0.0464 81.3557 81.3557 WET 0.0071 12.5046 93.8603 NDBSI 0.0003 0.5021 94.3624 LST 0.0032 5.6376 100 2015 NDVI 0.0476 96.3065 96.3065 WET 0.0012 2.3826 98.6891 NDBSI 0.0001 0.131 98.8201 LST 0.0006 1.1799 100 2020 NDVI 0.04 97.4195 97.4195 WET 0.0007 1.7021 99.1216 NDBSI 0.0001 0.035 99.1566 LST 0.0003 0.8434 100

**Table 2.** Principal component analysis index and eigenvalue.

Second, we introduce the entropy value method, which determines the weight of each index according to the information provided by the observed values of each index [38,39]. The evaluation index system includes N indices (NDVI, WET, NDBSI and LST). This is a problem that consists of m samples (cell) and uses N indicators for comprehensive evaluation. The initial data matrix A of the evaluation system is formed and *Xij* is the value in *i* cell of the *j* remote sensing ecological indicator. The detailed procedures of the entropy method are described as follows [22,38,40]:

$$A = \left( \begin{array}{cccc} \mathbf{X}\_{11} & \cdots & \mathbf{X}\_{1m} \\ \vdots & \vdots & \vdots \\ \mathbf{X}\_{m1} & \cdots & \mathbf{X}\_{mm} \end{array} \right)\_{m \times m}$$

1. Proportion of the value in *i* cell of the indicator *j*.

$$P\_{ij} = \frac{X\_{ij}}{\sum\_{i=1}^{n} X\_{ij}} \quad (j = 1, 2, \cdots, m) \tag{16}$$

2. Entropy value of the *j* th index.

$$e\_j = -\frac{1}{\text{lnm}} \times \sum\_{i=1}^{n} P\_{ij} \ln(P\_{ij}) \; k > 0,\\ e\_j \ge 0, 0 \le e\_j \le 1$$

3. Difference coefficient of the first index.

For the *j* th index, the more significant the difference is in the index value *Xij*, the greater the effect on the scheme evaluation and the smaller the entropy value.

$$\mathbf{g}\_{j} = 1 - e\_{j}$$

The larger the *gj* value, the more critical the indicator.

4. Weight.

$$\mathcal{W}\_{\dot{l}} = \begin{array}{c} \mathcal{S}\_{\dot{l}} \\ \frac{\sum\_{j=1}^{m} \mathcal{S}\_{\dot{j}}}{\sum\_{j=1}^{m} \mathcal{S}\_{\dot{j}}} \end{array}, \dot{f} = 1, 2 \cdot \cdots \cdot m \tag{17}$$

5. Ecological index score based on the entropy method.

$$RSEI\_{EW} = \sum\_{j=1}^{m} W\_j \times P\_{i\bar{j}} \ (i = 1, 2, \dots, n) \tag{18}$$

The PCA effectively removes redundant information between bands and compresses multiband image information into a few independent bands that are more effective than the original band. The entropy method can effectively remove deficiencies caused by a lack of PCA information. The weights for all of the indicators are listed in Table 3.

**Table 3.** Weights of indicators.


Finally, the IRSEI integrates humidity, greenness, heat and dryness through PCA and EW, which is calculated according to Equation (19):

$$IRSEI \, = \, \left( RSEI\_{\rm PCA} + RSEI\_{\rm EW} \right) / 2 \,\tag{19}$$

In this formula, *RSEIPCA* is the main component, *RSEIEW* is the weighted result of the entropy method and the final IRSEI is calculated as their arithmetic average. The IRSEI for each year has to be standardized to accurately compare the remote sensing images of different time frames. The closer IRSEI is to 1, the better the UEQ (and vice versa). The IRSEI for the four years is classified into five groups employing the ArcGIS software (Esri, USA). Referring to previous studies [22–24,29,41], these groups are labeled "Excellent, Good, Moderate, Fair, and Poor" and they facilitate comparisons across the study area (Table 4).

**Table 4.** Grades of ecological indicators.


#### 2.2.5. Spatial Autocorrelation Analysis of IRSEI

Global spatial autocorrelation (SA) measures the average correlation, spatial distribution pattern and significance of all of the objects in the entire study area. SA visualizes spatial aggregations and exceptions to the IRSEI. The Moran's index is commonly used to calculate SA [42]. The main calculation indices for spatial autocorrelation are the global Moran's index and the local Moran's index. We analyze both the "global" spatial clustering and the "local" spatial clustering of the IRSEI. The formula for calculating the global Moran's index is:

$$I = \frac{\sum\_{i=1}^{n} \sum\_{j=1}^{n} \mathcal{W}\_{ij} (\mathbf{x}\_i - \overline{\mathbf{x}}) \left(\mathbf{x}\_j - \overline{\mathbf{x}}\right)}{S^2 \times \sum\_{i=1}^{n} \sum\_{j=1}^{n} \mathcal{W}\_{ij}} \text{ (SA)}\tag{20}$$

where *n* is the total number of grid cells in the study area (500 m × 500 m); *Wij* represents the spatial weight of elements *i* and *j*; *xi* and *xj* are the attribute values of cell *i* and cell *j*, respectively; *x* represents the average value of the attributes across all cells; and *S*<sup>2</sup> is the sample variance.

The value of the global Moran index *I* varies between −1 and 1, where *I* > 0 indicates positive SA, i.e., a high value corresponds to high-value clusters, whereas a low value corresponds to low-value clusters. The closer I is to 1, the smaller the overall spatial difference. When *I* < 0, there is negative SA, i.e., there is significant spatial difference between a cell and its surrounding cells. The closer *I* is to −1, the greater the overall spatial difference. When *I* = 0, there is no SA.

Due to the fact that the global Moran's index describes the overall aggregation situation, it cannot accurately determine where the place of aggregation is located and is unable to indicate the hot spots and cold spots of the entire region. Accordingly, we use the local indicator of SA to measure local SA and determine hot and cold spots. The formula for the local Moran's *Ii* for cell *i* is:

$$I\_i = \frac{(\boldsymbol{\pi}\_i - \boldsymbol{\pi})}{S^2} \sum\_{j=1}^n \mathcal{W}\_{ij} (\boldsymbol{\pi}\_j - \boldsymbol{\pi}) \tag{21}$$

When the local Moran index *Ii* > 0, the spatial difference between the cell and its surrounding cells is minor. When the local Moran index *Ii* < 0, the spatial difference between cell *i* and its surrounding cells is significant. When the local Moran index *Ii* = 0, there is no spatial difference between cell *i* and its surrounding cells. In this study, we use the software GeoDA to calculate and obtain the global and local Moran's indices.

#### **3. Results**

#### *3.1. Attributing Factors*

A comparison of the spatial distributions of the four ecological factors in the study area (Figure 3) shows high levels of land surface moisture close to and alongside the Yangtze River, which extends in the central part of Wuhan from west to east. The NDVI is high on the northeast side, along the Yangtze and Han rivers, the central part of Wuhan and in patches in the south and east. Comparing the NDVI, LST and moisture maps shows that moderate temperature and moisture are the most favorable conditions for vegetation growth, whereas extreme weather conditions can damage plant vitality. The temperature and moisture conditions are moderate in the study area and the NDVI is remarkably high. A high LST is detected in the southern part of Wuhan, with some patches in the north and

east. A moderate LST is detected in the central part of Wuhan. The NDBSI does not display much variation, as most of the study area is covered by agricultural land (Figure 3).

**Figure 3.** Spatial distribution of ecological indicators, 1995–2020. (**a**–**d**) indicators 1995, (**e**–**h**) indicators 2005, (**i**–**l**) indicators 2015, (**m**–**p**) indicators 2020.

To test the representativeness of the index IRSEI, we calculate the correlation coefficients among IRSEI, WET, NDVI, NDSI and LST in the same period (Table S1, Supplementary Materials), and test the applicability of the model through average correlations. From 1995 to 2015, the average correlation of IRSEI with the other variables is the highest, ranging from 0.60 to 0.70. The mean correlation of IRSEI over this period was 0.64, which indicates that IRSEI integrates most of the information embodied in all four indicators. It is more representative than any single indicator and can better reflect the ecological situation.

#### *3.2. Spatial and Temporal Distribution of UEQ in Wuhan*

Generally, higher IRSEI values are associated with higher levels of greenness and moisture, whereas lower IRSEI values are directly proportional to dryness and temperature. This implies that high IRSEI values represent positive ecological conditions.

As shown in Figure 4, the IRSEI increases from 0.79 to 0.98 from 2010 to 2015, indicating improved ecological conditions. However, from the second half of 2015 up to 2020, its value drops to 0.82, indicating deterioration. Comparing the values from 2010 to 2020 indicates overall improved conditions, as the IRSEI increases from 0.79 to 0.82. However, the maximum values (1.09, 1.03 and 0.96) decline continuously, indicating that high-quality IRSEI conditions are declining continuously. Further, low-quality IRSEI conditions improve in the first half of the study period (1995 to 2005); however, in the second half (2005 to 2020), these conditions decline and reach their previous stage. Our findings also show maximal variation in the median IRSEI values, i.e., indicating the recovery of favorable conditions (moderate to high temperature, moderate to low moisture and higher vegetation) for all factors during the study period.

**Figure 4.** Changing trend of the IRSEI in each district of Wuhan from 1995 to 2020.

The mean IRSEI value and the area and percentage of each evaluation grade in Wuhan from 1995 to 2020 are displayed on Figures 4 and 5. Overall, the proportion of areas with average and good IRSEI ratings is the highest during the study period (>57%). The proportions of average and above average regions are 82.33%, 87.14%, 74.21% and 57.34%, indicating that the ecological environment of Wuhan was unstable from 1995 to 2020, with ecological conditions first improving and subsequently deteriorating. The UEQ of the Xinzhou, Hanyang, Qiaokou, Huangpi and Caidian districts show the most obvious decline, with reduction rates of 32.32%, 30.18%, 27.84%, 27.67% and 27.24%, respectively.

From the perspective of a single year (see Table 5), the area share of good ecological environment in 1995 was the highest, reaching 43.67% of the total area. The area share of poor ecological environment was the lowest in 1995, comprising an area of only 371 km2, or less than 5% of the total area. The share of poor ecological environment was approximately 12% of the total area. The area with a good ecological environment rating in 2005 was larger than that of 1995 and accounted for the highest proportion (45.57%), comprising an area of 3494 km2. The percentage of area rated excellent was the smallest (6.23%) after 1995. In 2020, the poor ecological environment generally accounted for the highest

proportion (38.92%), comprising an area of 2984 km2. The good ecological environment rating accounted for only 18.32%.

**Figure 5.** Changing area proportion of the IRSEI in Wuhan from 1995 to 2020.



The changing trend during the research period shows that the mean IRSEI values in 1995, 2005, 2015 and 2020 decreased year by year (0.60, 0.67, 0.58 and 0.47, respectively). The declining values indicate that the ecological environment of Wuhan has deteriorated continuously, probably owing to the rapid economic development of the city. According to the Wuhan Municipal Bureau of Statistics, the gross domestic product (GDP) increased from CNY 3.991 billion in 1978 to CNY 134.10 billion in 2017. The permanent resident population increased from 8.58 million people in 2004 to 10.33 million people in 2014. Ecological problems ascribed to human activities, such as vegetation damage and soil pollution, have become increasingly prominent.

As governments and social organizations have become increasingly aware of environmental protection, Wuhan has strengthened its enforcement of ecologically relevant laws and regulations, effectively halting the trend of environmental deterioration. This is reflected in the varying ecological evaluation grades. The differences in rating reflect an increase in area from 643 km<sup>2</sup> in 2005 to 1684 km2 in 2015 (area expansion of 14%) to 2221 km2 in 2020 (area expansion of 7%).

The spatial distribution (Table 6 and Figure 6) shows that areas with a good ecological environment are distributed mainly in the surrounding urban areas of Wuhan. These areas have a relatively weak economy and the land-use types are mainly cultivated land and woodland, with rich vegetation and high biodiversity levels. The areas with poor ecological environments are concentrated in Hongshan, Hanyang, Wuchang and Qingshan. According to the different functions of each administrative region of Wuhan, Hongshan

is based mainly on the education industry. Several colleges and universities are located in the area, and it is densely populated. Qingshan, Hanyang and Wuchang are primarily industrial areas. Heavy industrial companies, such as Wuhan Iron & Steel Co., Ltd., Wushi Chemical Co., Ltd. and Dongfeng Motor Co., Ltd., are located in these areas. Industrial production and human economic activities have a direct detrimental effect on the environment of these areas.

**Figure 6.** Grading map of UEQ from 1995 to 2020 in Wuhan city. (**a**) IRSEI 1995, (**b**) IRSEI 2005, (**c**) IRSEI 2015 and (**d**) IRSEI 2020.

**Table 6.** Area statistics of UEQ evaluation grade from 1995 to 2020 in Wuhan city (unit: km2, %).



**Table 6.** *Cont.*

#### *3.3. Dynamic Monitoring of UEQ in Wuhan*

Based on the IRSEI grade classification, the detected changes were divided further into nine levels and seven classes. The range for the levels of detected changes was −4 to +4, with a positive value indicating that the UEQ had improved, 0 indicating no change and a negative value indicating deterioration. For the classes with no detected changes, level 0 was classified as unchanged, level −4 as significantly worse and levels −2 and −3 as worse; level −1 as slightly worse; level 1 as slightly better; levels 2 and 3 as better; and level 4 as significantly better (Table 7).


**Table 7.** Change in the ecological index grade.

Table 8 presents the ecological changes in Wuhan from 1995 to 2020. The size of the area representing both UEQ and ecological deterioration (obviously worse and slightly worse) is 3636 km2, accounting for the highest proportion (39.44%) over 2015–2020. The size of the area with the same UEQ (no change) is 2984 km2, accounting for 35.56% of the total area. Among the areas with deteriorating UEQ, most (69.51%) deteriorated by one grade. Deterioration in UEQ accounted for 25.60%. Most of the areas showing improved environmental conditions improved by one grade, accounting for 79.41% of the entire improved area. The areas improving by two grades account for 18%. The areas representing levels 3 or 4 are relatively small, indicating gradual changes. The areas with significant changes are related to direct economic activities, such as the transformation of cultivated land and woodland into construction and industrial land. The spatial distribution of UEQ (Figure 7) shows that the deteriorating areas are located mainly around cities and most water bodies. The deterioration of the ecological environment around water bodies is related to a leakage of urban domestic sewage and enterprise wastewater and a rise in aquaculture in recent years. Moreover, the areas with a deteriorating ecological environment are expanding along both sides of the Yangtze and Han rivers. Except for the water area, the UEQ in the central metropolitan area remains mainly unchanged and several areas show signs of improvement. This result indicates that environmental governance in the main urban area of Wuhan has played a positive role in recent years.

**Table 8.** Change in the ecological index grade from 1995 to 2020.


**Figure 7.** Spatial transfer distributions of the ecological levels of the IRSEI in Wuhan from 1995 to 2020.

#### *3.4. Spatial Autocorrelation Analysis*

We explore the spatial autocorrelation (SA) of the IRSEI at a grid cell scale of 500 m × 500 m and our results indicate the existence of SA. The Moran's I was 0.568 in 1995, and 0.535 in 2020. All four IRSEI maps (1995, 2005, 2015 and 2020) display an extremely low probability (*p*-value < 0.01) of completely random spatial distribution. Therefore, the statistical significance test shows that SA exists for all of the ecological factors. The IRSEI increased in places where spatial distribution was favorable to the UEQ. In 1995, high-value clustering of the IRSEI in Wuhan was distributed mainly in the south and north of the study area, whereas low-value clustering was concentrated in the middle of the study area. In 2005, high IRSEI values started gathering gradually in the southern region, and low IRSEI values became more concentrated in the clustering distribution. By 2015, the high/high clustering and low/low clustering of the IRSEI in the study area became more dispersed and tended to spread in every direction. In 2020, low/low clusters had spread from the middle to the east and west, whereas high/high clusters were concentrated mainly in the south and north of Wuhan City.

The Moran's I scatter graph is divided into four quadrants, corresponding to four different spatial distribution types (Figure 8). The first quadrant represents high/high clustering, the second quadrant low value and high-value aggregation, the third quadrant low/low aggregation and the fourth quadrant high-value and low-value aggregation. The IRSEI of Wuhan is concentrated mainly in the first and third quadrants. This result indicates that the IRSEI spatial distribution in Wuhan represents positive spatial autocorrelation, and high IRSEI agglomeration zones are mainly distributed in outer suburban areas, mainly in the north and southeast Wuhan.

**Figure 8.** Spatial correlation and Moran index scatterplot of IRSEI from 1995 to 2020 in Wuhan city. (**a**) SA with 1995, (**b**) SA with 2005, (**c**) SA with 2015 and (**d**) SA with 2020. (Note: SA—Spatial autocorrelation).

#### **4. Discussion**

#### *4.1. Literature, Policy and Practice*

We have reviewed previous studies and demonstrated that it is feasible to evaluate the quality of the urban ecological environment through remote sensing. This research proposes a feasible method. Other remote sensing images could also have been used as data in this research, such as Tiangong-2 WIS images [11]. In terms of method improvement, we mainly improved the integration of quantitative factors. A related similar index, RSUSEI, has primarily increased remote sensing ecological factors by adding the impervious surface cover (ISC) [15]. ISC is also one of the most important factors that distinguish different

types of land use/land cover characteristics in urban environments, and has a strong impact on UEQ. However, in our study we also consider the dryness index (NDBSI), the bare soil index (SI), the building index (IBI) and the normalized buildings–bare-soil index. However, there are strong correlations between the impervious surface, bare soil and building indices. Previous studies have found that the relationship between ISC and LST has the form of an exponential function, rather than a simple linear function, as commonly believed [43]. This exponential relationship has been confirmed by many subsequent studies [44,45]. Our IRSEI index takes into account the bare soil, building index and surface temperature. We suggest that the correlations of remote sensing ecological indicators affecting the regional ecological environment should be introduced into comprehensive indicators, or different indicators should be set according to the characteristics of the study region.

There are few high-quality ecological environment patches in Wuhan (IRSEI > 0.8), with close to zero over the past five years, and most of the patches are in the center of the ecological environment. Therefore, we propose a policy whereby Wuhan would focus on protecting forest land and gardens, build high-quality ecological corridors and coordinate the management of rivers in the future, so as to guide sustainable urban development and achieve sustainable development goals (such as SDG 11, sustainable cities and communities). Lake and wetland protection and ecological restoration and management will optimize the pattern of ecological security. Further analyses of the results indicate that there was a negative correlation between LSI, NDBSI and urban ecological quality. The ecological environment in areas with a high surface temperature, such as the Wuhan downtown area and coastal area around the Yangtze River, has tended to deteriorate; however, the humidity indices in these areas were also relatively high, which is conducive to ecological protection. Low vegetation index values in the central urban area also affect the quality of the ecological environment of Wuhan to a certain extent. The IRSEI can macro-evaluate the quality of the regional ecological environment, which is more convenient and efficient. In the future, higher precision can be introduced at the block level. Data, such as Google Street View data, could be used with machine learning algorithms to further identify the proportion of regional urban green space, trees, etc., and improve the accuracy of ecological environment assessment. The index has a high ability to distinguish between different land cover uses. The framework can also be easily extended to a global scale or to map other gridded socio-economic variables (such as GDP and population) to monitor and assess progress towards the SDGs [25]. The assessment and modelling of uses is critical to supporting sustainability assessment in achieving Sustainable Development Goals (SDGs), such as sustainable cities and communities. Therefore, IRSEI can be used to assess the spatial and temporal sustainability of cities.

#### *4.2. Analysis of the Factors Affecting the UEQ*

The regression least squares method (OLS) can be used to quantitatively describe the relationship between the ecological index and natural, economic and social factors in Wuhan. The data include temperature, precipitation, elevation, slope and DMSP as explanatory variables. The night light variable reflects the human footprint and fundamentally affects the urban ecological environment. Impervious surfaces and roads and a high population density are not conducive to UEQ. The regression coefficients represent the contribution of six independent variables to the dependent variable. The regression coefficients of precipitation and elevation are equal to 0.522 and 0.441, respectively, indicating that precipitation and elevation positively contribute to the IRSEI.

In contrast, the regression coefficients of night light and slope are negative, indicating that these variables contribute negatively to the IRSEI. The night light variable has a regression coefficient of –0.619, indicating a negative effect. The R2 is 0.901 and *p* < 0.05, indicating that climate, precipitation, elevation, slope and night light data account for 90% of the variations of the IRSEI.

The regression equation between the IRSEI and the independent variables is as follows:

IRSEI = 0.926 + 0.148 × Temperature + 0.522 × Precipitation + 0.441 × Slope − 0.001 × Elevation − 0.619 × DMSP (R = 0.901).

#### *4.3. Method Framework and Validation Analysis*

Weighting is an important process in the development of aggregated ecological indices that help promote sustainability. Different weighting methods have different characteristics, and the method employed could reflect the subjectivity of the decision makers. However, such methods combined with remote sensing index data can facilitate decisions and reduce the calculations required.

PCA is widely used in the evaluation of the RSEI. In several studies, the three principal components obtained after dimensionality reduction did not show any obvious effects (contribution was below 15%). However, including all of the pixels in extensive data calculations is a time-consuming process. The RSEI employs a covariance-based (unstandardized) PCA to determine the importance of each indicator involved. The weight of each indicator can be assigned objectively and automatically based on the load (contribution) of each indicator to PC1. In this study, we used PCA and EW to comprehensively calculate the IRSEI. After improvement, the combined method was able to reflect the degree of change in the index, and the calculation was quick and uncomplicated. The spatial distribution of the UEQ over the study period (1995–2020) is consistent with the information in the bulletin on the eco-environmental situation in China in that year. The current, more popular assessment method is based on habitat quality (HQ) [46–50]. In further research, we intend to include HQ in this quantitative assessment.

#### *4.4. Limitations and Future Prospects*

The proposed UEQ evaluation model is feasible and straightforward, providing a new idea for ecological protection and comprehensively reflecting the changes in UEQ in Wuhan. From 1995 to 2020, the UEQ of Wuhan declined overall, probably owing to a combination of natural factors and human activities. However, the ecological level in the eastern and southeastern mountainous areas has increased because of the influence of forest resource protection, desertification land management and the warm and humid climate. In contrast, the regional ecological level has declined, owing to the overexploitation and overgrazing of lake resources in the northwest and southwest of Wuhan. The constantly rising levels of urbanization and construction over nearly 20 years have resulted in a downward trend in the UEQ. Overall, the ecology of Wuhan is in a fragile state. In 2020, the proportion of areas with poor ecological environment grades remained high, accounting for 42.68% of the total area.

In future social and economic development, we should follow the laws of nature, prioritize protection and rationally develop and utilize natural resources. The IRSEI effectively revealed the spatial distribution of and change in the UEQ in Wuhan, based on remote sensing images. Although four types of ecological factors closely related to the ecological environment were selected in the calculation process, the ecological environment is a complex and comprehensive variable. Areas with a deteriorating ecological environment tend to be spread along the Yangtze and Han rivers and around the central urban area. Urban expansion has damaged the ecological environment, and urban planning should integrate more ecological concepts to promote a harmonious coexistence and sustainable development for humans, nature and society.

Comprehensive quantitative evaluation requires selecting several impact factors that reflect the actual situation in the study area. We aimed to conduct the UEQ evaluation by employing a scientific, objective and feasible method. Nevertheless, choosing the UEQ evaluation index remains exploratory work. Determining the index weight affects the accuracy of the evaluation results. Accordingly, expanding research to a more scientific multifunction performance index system and determining the index weights require further work. Furthermore, the limited availability of data and a lack of longitudinal comparison of urban data have affected the scientific nature of our research results. In addition, our next step will be exploring how the IRSEI changes at different spatial scales. The rapid development of cities will inevitably lead to a series of ecological and environmental problems, and the deterioration of the ecological environment may further affect the surrounding environment, forming a cycle and harming urban sustainability. This study also demonstrates that IRSEI is characterized by spatial heterogeneity; that is, the poor UEQ patches will focus on areas where the ecological environment is poor and the urbanization is also highest.

### **5. Conclusions**

In this study, the IRSEI model was used to evaluate and monitor the ecological environment in Wuhan from 1995 to 2020. The IRSEI is an ecological environmental quality assessment method based on remote sensing technology. The method has many advantages, such as the ease of obtaining parameters, a large time sequence span and a wide evaluation range. The UEQ method employing remote sensing technology is feasible and simple, and provides a new tool for territorial spatial control and spatiotemporal urban sustainable development. Our proposed UEQ assessment framework can also help to develop potentially relevant additional sub-indicators, which could help to address one of the current challenges in SDG monitoring, namely how to implement SDG indicators. We have implemented the proposed workflow in this study based on an open-source platform and free satellite data, making it an appealing option that is applicable in almost all countries.

The main conclusions from the results of this study are:


**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/rs13214440/s1, Table S1: Correlation matrix of IRSEI and four factors.

**Author Contributions:** Conceptualization, J.L. and J.G.; methodology, J.L. and J.G.; validation, J.L., J.Y. and J.-M.G.; data curation, J.L. and J.Y.; writing—original draft preparation, J.L.; writing—review and editing, J.L. and J.-M.G.; funding acquisition, J.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China, grant number 41871172 and 41701228 and supported by the Fundamental Research Funds for National Universities, China University of Geosciences (Wuhan).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The satellite images used in this study are obtained from http:// earthexplorer.usgs.gov, accessed on 15 May 2021.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

