Assessment of Ecological Environment Quality and Analysis of Its Driving Forces in the Dabie Mountain Area of Anhui Province Based on the Improved Remote Sensing Ecological Index

Abstract

The Dabie Mountain area in Anhui Province is an essential ecological security barrier and a critical protected area in East China. It is very important to assess its ecological environment quality and identify its key driving forces. Five indicators, including Greenness, Wetness, Dryness, Heat, and Biological Richness, were used to construct an improved remote sensing ecological Index (IRSEI) to assess ecological environment quality. The weights of the five indicators were determined by coupling the analytic hierarchy process (AHP) and the entropy weight method (EWM). The optimal parameters-based geographical detector (OPGD) was used to recognize driving factors. The main conclusions were as follows: (1) the overall rank of ecological environment quality was mainly good and excellent. The ecological quality of forest land was excellent, that of farmland was good, and that of built-up areas was poor. (2) The change in ecological environment quality was mainly stable from 2000 to 2020. The ecological quality of some forests and farmlands improved, with a deteriorating trend in the built-up areas. (3) The Moran’s Index of ecological quality ranged from 0.77 to 0.85, indicating high spatial agglomeration. (4) The OPGD indicated that the DEM had the most explanatory power for ecological quality, and the interactive relationship between the DEM and population density had the most significant impact. (5) In comparison to the conventional remote sensing ecological Index (RSEI), the IRSEI exhibited higher congruence with observed circumstances and improved ecological interpretability.

1. Introduction

The ecological environment is vital for human existence and growth, and its quality influences global sustainability and survival [1,2]. The ecological environment endures significant stresses from global climate change and urbanization [3]. Current research on ecological environments include ecological environment quality evaluation, ecological safety assessment [4], ecological risk identification [5], and so on.

Dabie Mountain area in Anhui, located in the southwestern part of Anhui Province, safeguard East China’s ecological security through their abundant natural resources and unique ecosystems [6,7]. At present, research on the Anhui Dabie Mountain area focuses on assessing the function of specific services provided by ecosystems to human beings. Huang et al. [8] analyzed the spatial and temporal evolution of ecosystem services and their coupling with residents’ well-being in Shucheng county, which was located in Dabie Mountain area in Anhui Province from 2005 to 2020, based on the InVEST model. Quan et al. [9] analyzed the evolution of ecosystem service values and their drivers in the Anhui Dabie Mountain area through land-use change and ecosystem service value assessment models. In summary, there are fewer studies on evaluating the ecological environment quality of the Anhui Dabie Mountain area.

In 2020, China promulgated the Overall Plan for Major Projects for the Protection and Restoration of Nationally Important Ecosystems (2021–2035), which proposed to promote the ecological protection and restoration of key projects, including those in the Dabie Mountain area. The plan proposed that an integrated natural ecological monitoring and supervision network of Space-Air-Ground would be established by 2025. Consequently, a precise evaluation of ecological environment quality (EEQ) and a comprehensive analysis of driving factors will be crucial for effective ecological preservation and sustainable development.

Presently, methodologies for assessing ecological environment quality include fuzzy evaluation [10,11], ecological footprint [12,13], environmental index (EI) [14,15], and grey relational analysis [16]. The selection of indices and the derivation of weights used with EI have been subjective, and certain indices have been difficult to obtain [17]. Greenness, Wetness, Dryness, and Heat were utilized by Xu [18] to construct the remote sensing ecological index (RSEI) using principal component analysis (PCA). It was widely used [19,20].

However, there are uncertainties in the application of the RSEI model. For example, for areas with extreme ecological conditions, such as desert areas and land degradation areas, the application of the RSEI was adversely impacted [21]. The principal component analysis method also has some limitations. The ecological significance represented by the high value part of its transformation is ambiguous [22]. The direction of eigenvectors is often ambiguous [23], and the process of dimensionality reduction leads to a loss of information [24]. In this regard, the researchers improved the RSEI model in a number of ways. For example, the RSEI model was optimized by adding or replacing metrics to make the evaluation results more accurate and consistent with actual ecological scenarios [25,26]. Different weight assignment methods were used to improve the RSEI and address the limitations of the PCA [27,28]. The improved model solved the problems of spatial inconsistency and interpretability of the RSEI.

Wang et al. [29] proposed a geographical detector for measuring spatial heterogeneity, detecting explanatory factors and analyzing interactions between variables. Nonetheless, this method can influence geographic data discretization and the selection of statistical units, hence diminishing the accuracy of conclusions. The optimal parameters-based geographical detector (OPGD) is proposed to enhance the accuracy and reliability of spatial heterogeneity analysis [30]. Zhao et al. [31] employed the optimal parameters-based geographical detector model to estimate vegetation cover in the urban agglomeration of South Sichuan. He et al. [32] employed it to examine the determinants of urban development across various urban growth patterns.

In summary, this study achieves the followings: (1) Introduces the biological richness index to construct the IRSEI model, which is conducive to assessing the sustainability of the ecosystems in the study area and capturing ecological defects more accurately. (2) Hierarchical analysis and entropy weighting are used to construct the combined weights, effectively overcoming the defects of the single-assignment method and realizing synergistic optimization between expert knowledge-driven and data law-driven results. (3) The drivers of ecological environment quality are identified using the optimal parameters-based geographical detector.

2. Materials and Methods

2.1. Study Area

The Dabie Mountain Area is situated in the confluence of Hubei, Henan, and Anhui, serving as the watershed between the Yangtze River and the Huaihe River, and is recognized as one of the 25 significant ecological regions in China [33]. The Anhui Dabie Mountain region is situated in the western section of Anhui Province, with geographic coordinates ranging from approximately 115°22′ to 116°46′ E and 30°9′ to 31°48′ N and a total area of around 12,085 km². It is situated in the transition zone between subtropical monsoon climate and temperate monsoon climate, characterized by four distinct seasons, a warm climate, ample rainfall, an average annual temperature ranging from 14 to 16 °C, and an annual precipitation of between 1000 to 1500 mm. The second national remote sensing survey on soil erosion indicates that ecological issues, including water and soil erosion, are more pronounced in this region, which is characterized as ecologically vulnerable [34]. The study area includes Jinzhai County, Huoshan County, Yuexi County, Taihu County, and Qianshan City within the Anhui Dabie Mountain region (Figure 1).

2.2. Materials and Processing

The research for this study employed land remote sensing satellite (Landsat) imagery from 2000, 2006, 2011, 2015, and 2020. To guarantee data quality, images were selected from August to October, with cloud cover lower than 10%. Radiometric calibration, atmospheric correction, and cropping were performed using ENVI 5.6 software for the subsequent IRSEI computation.

To analyze the impact determinants of ecological environment quality in the Anhui Dabie Mountain, referring to the literatures [35,36], IRESI was used as the dependent variable, and the digital elevation model (DEM), slope, aspect, precipitation, temperature, population density, and gross domestic product (GDP) were selected as independent variables.

To facilitate the use of the OPGD, the data were preprocessed using GIS tools. Firstly, the slope and aspect data were calculated on the basis of the elevation data. Then all data were standardized to the coordinate system “GCS_WGS_1984” and cropped to match the spatial extent of the study area. Finally, the data were resampled using nearest neighbor interpolation to achieve a uniform resolution of 1000 × 1000 m. Information on the data used for this study is presented in

Table 1. Data used for the study.

Data Type	Spatial Resolution	Period	Data Sources
Landsat 5, 7, 8	30 m	2000, 2006, 2011, 2015, 2020	Geospatial data cloud (https://www.gscloud.cn/)
Land use data	30 m	2000, 2006, 2011, 2015, 2020	The 30m annual land cover datasets and their dynamics in China from 1985 to 2023 [37]
GDP data(X1)	1 km	2000, 2006, 2011, 2015, 2020	Resource and Environmental Science data platform (https://www.resdc.cn/)
Aspect data(X2)	30 m	—	—
Slope data(X3)	30 m	—	—
Digital elevation data(X4)	30 m	—	Geospatial data cloud (https://www.gscloud.cn/)
Temperature data(X5)	1 km	2000, 2006, 2011, 2015, 2020	Third Pole Environment Data Center (https://data.tpdc.ac.cn)
Population density data(X6)	1 km	2000, 2006, 2011, 2015, 2020	Third Pole Environment Data Center (https://data.tpdc.ac.cn)
Precipitation data(X7)	1 km	2000, 2006, 2011, 2015, 2020	Third Pole Environment Data Center (https://data.tpdc.ac.cn)

.

2.3. Methodology

The detailed workflow of this study is shown in Figure 2. First, the study calculated five indicators: WET, NDVI, NDBSI, LST, and AI. By coupling AHP and EWM, the weights of the five indicators were obtained and then used to construct the Improved remote sensing ecological index (IRSEI) to generate ecological quality maps for the years 2000, 2006, 2011, 2015, and 2020. The study used transfer matrices and spatial autocorrelation to analyze these data. Then, the OPGD model was used to analyze the driving factors of ecological environment quality from 2000 to 2020.

2.3.1. Calculation of Indicators

Utilizing ENVI software, the IRSEI was developed by combining indices of Greenness, Wetness, Dryness, Heat, and Biological Richness.

1.

Greenness: The normalized difference vegetation index (NDVI) accurately represented the growth condition of vegetation and provided biomass information [38]. Therefore, NDVI was used to represent greenness.

$$\mathit{NDVI}=\left({\rho }_{\mathit{NIR}}-{\rho }_{\mathit{red}}\right)/\left({\rho }_{\mathit{NIR}}+{\rho }_{\mathit{red}}\right)$$

(1)

${\rho }_{NIR}$—Reflectance in the near-infrared band, ${\rho }_{Red}$—Reflectance in the red band.

2.

Wetness: The wet indicator (WET) can reflect the moisture content in soil and vegetation, which was an important indicator of the regional water cycle, climate regulation, and ecosystem stability [39]. The specific formula is as follows [18,40].

$${WET}_{\left(\mathit{TM}\right)}=0.031 5{\rho }_{\mathit{blue}}+0.202 1{\rho }_{\mathit{green}}+0.310 2{\rho }_{\mathit{red}}+0.159 4{\rho }_{\mathit{NIR}}-0.680 6{\rho }_{\mathit{SWIR}1}-0.610 9{\rho }_{\mathit{SWIR}2}$$

(2)

$${WET}_{\left(\mathit{OIL}\right)}=0.151 1{\rho }_{\mathit{blue}}+0.197 3{\rho }_{\mathit{green}}+0.328 3{\rho }_{\mathit{red}}+0.340 7{\rho }_{\mathit{NIR}}-0.711 7{\rho }_{\mathit{SWIR}1}-0.455 9{\rho }_{\mathit{SWIR}2}$$

(3)

${\rho }_{blue}$—Reflectance in the blue band, ${\rho }_{green}$—Reflectance in the green band, ${\rho }_{SWIR1}$—Reflectance in the short-wave infrared 1 band, ${\rho }_{SWIR2}$—Reflectance in the short-wave infrared 2 band.

3.

Dryness: The normalized differential build-up and bare soil index (NDBSI) was calculated from the soil index (SI) and index-based build-up index (BIL), The specific formula is as follows [41].

$$NDBSI=\left(I_{\mathit{SI}}+I_{\mathit{IBL}}\right)/2$$

(4)

$$I_{SI}={\displaystyle \frac{\left[\left({\rho }_{\mathit{SWIR}1}+{\rho }_{\mathit{red}}\right)-\left({\rho }_{\mathit{blue}}+{\rho }_{\mathit{NIR}}\right)\right]}{\left[\left({\rho }_{\mathit{SWIR}1}+{\rho }_{\mathit{red}}\right)+\left({\rho }_{\mathit{blue}}+{\rho }_{\mathit{NIR}}\right)\right]}}$$

(5)

$$I_{IBL} = {\displaystyle \frac{\frac{2{\rho }_{SWIR1}}{{\rho }_{SWIR1} + {\rho }_{NIR}} - \frac{{\rho }_{NIR}}{{\rho }_{NIR} + {\rho }_{red}} + \frac{{\rho }_{green}}{{\rho }_{green} + {\rho }_{SWIR1}}}{\frac{2{\rho }_{SWIR1}}{{\rho }_{SWIR1}+ {\rho }_{NIR}} + \frac{{\rho }_{NIR}}{{\rho }_{NIR} + {\rho }_{red}} + \frac{{\rho }_{green}}{{\rho }_{green} + {\rho }_{SWIR1}}}}$$

(6)

$I_{SI}$—Value of the soil index, $I_{IBL}$—Value of the index-based build-up index.

4.

The land surface temperature (LST) was calculated with the help of ENVI software for radiometric calibration of the near infrared band and converting the units of LST to Celsius. The specific formula is as follows.

$$T={\displaystyle \frac{K_2}{\mathit{ln}\left(\frac{K_1}{L_{\lambda }}+1\right)}}$$

(7)

$$LST=T-273.15$$

(8)

T—Image brightness temperature (K), $K_1,K_2$—Constants preset before the satellite launch, $L_{\lambda }$—Intensity of radiation received by remote sensors.

5.

Biological richness: This index indirectly reveals the rich or poor status of biological abundance in the assessed area by quantifying the differences in the number of species between different ecosystems within a unit area [42]. It is calculated based on land-use type data [37]. Refer to the “Technical Standards for Ecosystem State Assessment” for the calculation formula [43].

$$AI=\left(0.35\times Forest+0.21\times Grassland+0.28\times Water+0.11\times \mathit{Farmland}+0.04\times \mathit{Built}+0.01\times Unused\right)/Area$$

(9)

Forest—raster units of forest, grassland—raster units of grassland, water—raster units of water, farmland—raster units of farmland, build—raster units of build, unused—raster units of unused, area—study area raster data.

2.3.2. Construction of the Improved RSEI

The analytic hierarchy process (AHP) effectively captured subjective requirements in ecological assessments and guaranteed that essential indicators were not overlooked [44]. The entropy weight method (EWM) determined the weights of the factors based on the degree of dispersion of the data, which was a relatively objective weighting method [45]. The SPSSPRO platform was utilized to compute the weights of the two models and obtained the final combined weights for the construction of IRSEI.

1.

Weights of AHP

The first step was to establish the hierarchical model. Then, the judgment matrix (pairwise comparison) was constructed, and the square root method was used to calculate the approximation value of the matrix eigenvectors. Finally, the consistency of the judgment matrix (CI) was examined. The formula for calculating the eigenvectors of the judgment matrix is as follows [44,46].

$$M_i=\sqrt[n]{{\displaystyle \prod _{j=1}^n}{\alpha }_{ij}}$$

(10)

$$W_i^{AHP}={\displaystyle \frac{M_i}{{\sum }_{i=1}^n{M}_i}}$$

(11)

$$\lambda ={\displaystyle \sum _{i=1}^n}{\displaystyle \frac{{\left(Aw\right)}_i}{n{w}_i}}$$

(12)

$M_i$—The n-th root of each row element of a matrix, $W_i^{AHP}$—Normalized AHP method weights, $\lambda$—Judgment matrix maximum eigenvalue.

2.

Weights of EWM

In order to ensure the comparability and consistency of the data for each index, the data for the five indicators were normalized so that the range of the data was unified between 0 and 1. The specific formulas are as follows.

$$X_{ij}^{-}={\displaystyle \frac{X_{max}-X}{X_{max}-X_{min}}}$$

(13)

$$X_{ij}^{+}={\displaystyle \frac{X-X_{min}}{X_{max}-X_{min}}}$$

(14)

$X_{ij}^{-}$—Normalization of negative indicators (NDBSI, LST), $X_{ij}^{+}$—Normalization of positive indicators (NDVI, AI, WET).

The study area was segmented into 1000 m × 1000 m units using ArcGIS 10.8 software, and the average values of the five criteria for each zone were tabulated. The entropy weights were computed using the subsequent formula [45,47].

$$e_j=-k{\displaystyle \sum _{i=1}^n}{p}_{ij}ln\left(p_{ij}\right),j=1,\dots ,m$$

(15)

$$d_j=1-e_j,j=1,\dots ,m$$

(16)

$$w_j^{Entropy}={\displaystyle \frac{d_j}{{\sum }_{j=1}^m{d}_j}},j=1,\dots ,m$$

(17)

$e_j$—the entropy of the Jth index, $d_j$—information entropy redundancy (variance), $w_j^{Entropy}$—weighting of indicators.

3.

Construction of IRSEI by coupling weights

The AHP considered the actual ecological significance and expert wisdom of the indexes, while the EWM relied entirely on data and was objective [48]. The following formula was used to couple the two weights [49]. It avoided the arbitrariness of subjective assignment and the lack of ecological significance that might be caused by objective assignment.

$$w_i^{Final}={\displaystyle \frac{w_i^{AHP}\times w_j^{Entropy}}{{\sum }_{i=1}^{m_1}\left(w_i^{AHP}\times w_j^{Entropy}\right)}}$$

(18)

$$IRSEI={\displaystyle \sum _{i=1}^{m_1}}{w}_i^{Final}\cdot Z_i$$

(19)

$Z_i$—values of raster of indicators.

2.3.3. Spatial Autocorrelation Analysis

Spatial autocorrelation analysis was used to study the correlation between the values of variables and the neighboring values for geographic or spatial data [50]. Global Moran’s index and Local Moran’s index were used in this study. The Global Moran’s index ranged from −1 to 1, with positive values being aggregated, negative values being discrete, and near to 0 being random [50,51]. The Local Moran’s index was used to analyze the local association pattern of each spatial unit [52]. Positive values indicated high–high agglomeration or low–low agglomeration, and negative values indicated the presence of high–low outliers or low–high outliers [51]. This study used GeoDa 1.20 software to calculate the two indicators. The specific formulas were given in the literature [53].

2.3.4. Transition Matrix

The transition matrix was widely utilized to characterize the comprehensive framework and magnitude of land-use change transfer [54]. The change pattern of the IRSEI in the Anhui Dabie Mountain region from 2000 to 2020 was analyzed using the transition matrix.

2.3.5. Optimal Parameters-Based Geographical Detector (OPGD)

The OPGD model can identify the ideal combination of spatial data discretization techniques, spatial layers, and spatial scale parameters to enhance the exploration of the driving forces behind geographic phenomena and increase the identification of spatial dissimilarity and variable relationships [30,31]. It uses algorithms to automatically search for optimal parameter combinations with high computational efficiency, reduced subjectivity through optimization, and more robust results. It also avoids the defects of the geographical detector’s parameter combinations relying on manual experience, low computational efficiency, and possible bias of the results due to subjective stratification [30]. With the help of ArcGIS, this study employed a fishnet tool to generate 1000 × 1000 m grid cells to analyze the driving factors of changes in ecological environmental quality in the Anhui Dabie Mountain area and extracted both dependent and independent variable values to grid points for further analysis. The OPGD operations were performed utilizing the geographical detector package for R.

3. Results

3.1. Results of Indicator Weights

The weights of the five indices were determined by coupling the weights of the hierarchical analysis method and the weights of the entropy weighting method (

Table 2. Combined weights of indicators.

Weight	AI	NDVI	WET	NDBSI	LST
2000	0.509	0.192	0.089	0.098	0.112
2006	0.609	0.244	0.052	0.077	0.018
2011	0.491	0.202	0.070	0.088	0.149
2015	0.507	0.187	0.064	0.090	0.153
2020	0.526	0.216	0.061	0.090	0.106

). The biological richness index had the largest weight, which was about 50%. It was followed by NDVI index and LST index. The WET index and the NDBSI index accounted for a lower percentage. This showed that biological richness had a great influence on the ecological environment quality in the study area.

3.2. Spatiotemporal Distribution of EEQ

Based on the combined weights, a raster calculator was used to obtain the IRSEI values. By the pertinent standards [18], the ecological environment quality of the research region was classified into five categories as delineated in

Table 3. Range of IRSEI grade.

Level	Poor	Fair	Moderate	Good	Excellent
Interval	[0,0.2]	(0.2,0.4]	(0.4,0.6]	(0.6,0.8]	(0.8,1]

.

The mean values of the IRSEI were 0.835, 0.886, 0.867, 0.850, and 0.857 in 2000, 2006, 2011, 2015, and 2020, respectively. IRSEI values showed a fluctuating changing trend of the “N” type. The mean value of the IRSEI fluctuated from 0.835 in 2000 to 0.857 in 2020. It showed that the ecological quality of the Dabie Mountain area of Anhui Province generally increased slightly during the study period, and the regional ecological environment improved. Specifically, the IRSEI increased from 0.835 in 2000 to 0.886 in 2006, dropped to 0.849 in 2015, then increased to 0.857 in 2020.

The ecological environment quality in the Anhui Dabie Mountain area was mainly excellent from 2000 to 2020 (Figure 3). Specifically, poor and fair ecological quality levels were concentrated in the northern, northeastern, and southeastern parts of the study area, which were dominated by extensive built-up areas and high population density. Moderate and good ecological quality levels were concentrated in the southeastern part of the study area, which were dominated by farmland. The excellent ecological quality level was distributed in the central part of the study area, which was characterized by forested and grassland areas with relatively abundant precipitation and high vegetation cover.

In terms of the percentage of area at different ecological quality levels (Figure 4), the excellent level covered the largest area at more than 75%. The good and moderate levels covered about 20% of the area. The area covered by the fair and poor levels was very small, which was less than 3%. Generally speaking, the ecological environment quality in the Anhui Dabie Mountain area was mainly at the good and excellent levels between 2000 and 2020.

3.3. Dynamic Changes in EEQ from 2000 to 2020

In order to explore the changes in the ecological quality of each ecological level, the study used a transfer matrix to present a detailed evolution of the ecological quality from 2000 to 2020 (Figure 5). The area of the excellent grade increased between 2000 and 2006. During this time period, the government adopted the Grain for Green Project. This transformed farmlands into forests and grasslands, which then enhanced the forest cover. After 2006, there was a small decrease. The area of the good grade decreased slowly between 2000 and 2011 and increased slowly between 2011 and 2020. The area of the poor grade increased between 2006 and 2011, decreased between 2011 and 2015, and remained stable for the rest of the period.

This study used a spatial analysis method to investigate the spatial and temporal changes in ecological quality between 2000 and 2020 based on the transfer matrix (Figure 6). The improvement area of ecological quality was mainly in the northern and southern parts of the study area from 2000 to 2006. The unchanged area dominated the ecological quality from 2006 to 2015. There was a deterioration trend in a very small area in the northern part of the study area from 2015 to 2020, which was dominated by the built-up area. Overall, the ecological environment quality of most forests and grasslands in the Anhui Dabie Mountain area remained unchanged from 2000 to 2020, the ecological environment quality of some farmland improved, and the ecological environment quality of a small part of the built-up area deteriorated.

3.4. Spatial Autocorrelation Pattern of EEQ

To analyze the spatial autocorrelation of the IRSEI from 2000 to 2020, the study employed Moran’s I index, significance mapping, and LISA clustering to analyze spatial patterns (Figure 7). The results showed that the Moran’s I values were consistently maintained between 0.77 and 0.85, indicating that the spatial distribution of ecological environment quality in the study area was characterized by highly positive spatial clustering (Figure 7a). The significance map showed that the non-significant areas were mainly distributed in the central-north and central-south regions of the study area, which were the transition areas between built-up areas, farmland, and forested land, implying that the environmental conditions in these areas were more complicated (Figure 7b). At the 95% confidence level, the “High–High” agglomeration was widely distributed in the central part of the study area and had good ecological environment quality, as well as high ecological environment quality in the neighboring areas. The area was dominated by forest and grassland, with high elevation, high vegetation cover, and good environmental background values. The “low–low” clusters were concentrated in the farmland in the south and the built-up area in the north. The ecological environment quality of these areas was poor, as was that of the adjacent areas. The region had a gentle topography and high impact of human activities. This suggested that the ecological quality of the Anhui Dabie Mountain area showed a polarized pattern.

3.5. Analysis of the Driving Factors

3.5.1. Results of the Single-Factor Analysis

The OPGD was used to calculate the independent effects of the influencing factors (

Table 4. Results of q-value probes for the driving factors.

Order	2000		2006		2011		2015		2020
	Factor	q-Value	Factor	q-Value	Factor	q-Value	Factor	q-Value	Factor	q-Value
1	DEM	0.335	DEM	0.289	DEM	0.286	DEM	0.301	DEM	0.284
2	TEMP	0.303	TEMP	0.255	TEMP	0.252	TEMP	0.272	POP	0.243
3	POP	0.196	POP	0.18	POP	0.169	POP	0.185	TEMP	0.232
4	Slope	0.186	Slope	0.164	Slope	0.159	Slope	0.168	Slope	0.158
5	PCPN	0.147	PCPN	0.16	GDP	0.093	GDP	0.115	PCPN	0.136
6	GDP	0.113	GDP	0.129	PCPN	0.066	PCPN	0.083	GDP	0.13
7	Aspect	0.017	Aspect	0.016	Aspect	0.021	Aspect	0.019	Aspect	0.002

Note: TEMP represents temperature, POP represents population density, and PCPN represents precipitation.

). Each factor had a significant effect on ecological quality (p < 0.01). The q-value of the DEM was stable at about 0.3. This was closely related to the geographic location of the region. The Dabie Mountain area of Anhui is in the transition zone between north and south of China, with great topographic undulations and high vegetation cover. Therefore, the DEM had the strongest explanatory power for ecological quality during the study period. The temperature was the second most important factor between 2000 and 2015, and population density was the second most influential factor in 2020. The rest of the factors had a significant effect on ecological quality, but with relatively low q-statistics.

3.5.2. Results of the Interactive Detection

Figure 8 showed the results of the interaction test of factors. The interaction of the factors all had higher explanatory power for ecological environment quality than any single factor. Specifically, the interaction between the DEM and population density was the strongest from 2000 to 2020, with q-values ranging from 0.313 to 0.358. In addition to this, the interactions between the DEM and all other factors showed a two-factor enhancement, with q-statistics values ranging from 0.288 to 0.353.

4. Discussion

4.1. Spatiotemporal Variations in EEQ

Quantifying ecological environment quality is crucial for formulating ecological protection measures. Generally, the ecological environment quality of the Dabie Mountain area of Anhui showed a gradual improvement trend during the study period. The ecological environment quality improved significantly from 2000 to 2006, declined from 2006 to 2015, and slowly increased from 2015 to 2020. From 2000 to 2006, the government started the Grain for Green Project and included the Dabie Mountain area of Anhui Province into the Natural Forest Protection Project. These policies effectively increased the forest cover and protected the ecology of natural forests in the study area. Between 2006 and 2015, there was rapid economic development, which might have led to a decrease in vegetation cover, irrational land use, and the fast development of tourism [55,56]. In 2015, new Environmental Protection Law was implemented, which was known as the strictest environmental protection law. The ecological environment quality was improved in the study area.

In terms of spatial distribution, areas of good ecological environment quality were concentrated in the central part of the study area, while poor areas were distributed in the southeastern and a small part of the northern part of the study area. This distribution pattern was highly consistent with the regional vegetation cover status. Areas of poor ecological quality were mainly built-up areas, while areas of good ecological quality were covered by forest, grassland, and scrub.

The Moran’s I values of the IRSEI were above 0.77 from 2000 to 2020, showing a high degree of spatial clustering. The LISA clustering map was dominated by the “High–High” and “Low–Low” types. The “Low–Low” aggregation zone was a localized spatial positive correlation. The ecological quality of the southeastern and northern regions formed a continuous band of low values in the study area, which might be related to the development of the regional economy and the intensity of human activities. These areas need to be prioritized for environmental protection and the formulation of targeted environmental protection strategies. Huang et al. [57] also found that the habitat quality of the Anhui Dabie Mountain followed a similar spatial distribution.

4.2. Factors Affecting the Ecological Environment Quality

Identifying the main factors affecting the changes in ecological environment quality can provide useful references for decision makers. In this study, the analysis of OPGD showed that the DEM was the dominant factor. Altitude affects temperature and precipitation, creating vertical climate zones. This determines the potential vegetation types and ecosystem primary productivity baselines for different elevation zones. Some studies showed a decreasing trend in vegetation cover in the Dabie Mountain from 100 to 600 m [58,59]. Habitat quality was positively proportional to elevation, and substantially increased with elevation [60]. The DEM (elevation of terrain) affects land use and human activities. In areas with gentle slopes and low elevations, human activities were strong, the level of socio-economic development was high, and the quality of the ecological environment was poor. In the higher altitude areas, the land was covered with vegetation, mostly forest and grassland, and the ecological environment was of good quality. At the same time, slope and aspects derived from the DEM also influence the sensitivity of ecosystems to disturbances and the successional paths and rates after damage. Zheng et al. [60] also confirmed that the DEM was the main factor influencing the habitat quality in the Dabie Mountain area. Other factors also had significant impacts on ecological environment quality, but to different degrees. Because of the rapid economic development, the influence of GDP increased in 2011 and 2015, as did population density in 2020.

In this study, it was found that changes in ecological environment quality were influenced by different factors. The interaction between the DEM and population density had the greatest effect on ecological environment quality. This might be due to the fact that higher elevations limited human development and thus ensured ecological quality. The gentler terrain was more favorable for human activities, leading to high population densities and thus posing a threat to the ecosystem. This is similar to the results of Zheng et al. [60] in which the DEM and land-use intensity had the largest effect.

4.3. Adaptive Analysis of the IRSEI Model

In order to validate the IRSEI, this study was based on the calculation methods of the IRSEI and RSEI to obtain the ecological environment quality of the study area in 2020. The standard deviation of the IRSEI and RSEI models were 0.4029 and 0.3873, respectively. According to the cited research [43,61], it can be seen that the larger the standard deviation, the more it can indicate the degree of data dispersion. Therefore, the IRSEI model better reflected the regional variability. The results of the IRSEI model were used as a reference benchmark, and the results of the RSEI model were used as the standard to be evaluated. With the help of GeoSOS-FLUS_V2.4 software, the Kappa coefficient was used to compare the accuracies of the IRSEI and RSEI. The value of Kappa coefficient was 0.285. It showed that there was a significant difference in the results of the two models. Three local details from Landsat images, the RSEI, and the IRSEI models were selected for comparison (Figure 9). Combined with the land-use type map, it could be seen that the three areas A, B, and C were all urban construction areas, which should belong to poor ecological environment quality. The results based on the RSEI model were moderate and fair; in comparison, the results of the IRSEI model were fair and poor. It can be seen that the evaluation results based on the IRSEI model were better than those of the RSEI model, which were more in line with the actual situation and better adapted to the regional evaluation.

4.4. Uncertainty and Prospects

Compared with the similar studies [19,21], there were still some uncertainties in this study. Firstly, Landsat had a 16-day revisit period and narrow imaging strips, which made it more susceptible to interference caused by cloud cover and image mosaics. Other images may be used for subsequent studies. Secondly, remote sensing images from August to October were selected as data sources for this study. The study area had relatively low cloud cover during this period, and the imagery was easily accessible. However, this period was susceptible to human activities, and there was some risk of seasonal bias. Therefore, in the follow-up study, it was recommended to obtain multi-temporal images covering the key seasonal periods and construct a time series for a more comprehensive assessment of the ecological environment quality.

5. Conclusions

This study used an IRSEI model to monitor the spatial and temporal changes in ecological environment quality in the Anhui Dabie Mountain area from 2000 to 2020 and quantified the influencing factors of ecological environment quality using the OPGD. The analysis showed that:

• Temporally, the mean values of the IRSEI were 0.835, 0.886, 0.867, 0.850, and 0.857 for 2000, 2006, 2011, 2015 and 2020, respectively. The ecological environment quality experienced a dynamic process of significant improvement in the early stage, deterioration in the middle stage, and slow recovery in the late stage from 2000 to 2020. This suggests that late-stage policy had a positive impact on the ecological environment.
• Spatially, the overall ecological quality of the study area was mainly excellent. The central part was dominated by forests and grasslands, with good ecological quality, and belonged to the “H–H” agglomeration area. The southeastern part was dominated by farmland with good ecological quality and belonged to “H–H” agglomeration. The built-up area in the northern part was of poor ecological quality, which was an “L–L” agglomeration. Based on this, it was possible to implement policies in zones and protect them precisely. For the ecological high-value agglomeration areas, it should strengthen the protection of ecological sources and build an ecological corridor network. For the ecological low-value areas, it should implement multi-functional management of farmland ecosystems and conduct ecological red line control of farmland. For ecological degradation zones, urban ecological restoration projects were conducted.
• DEM factors had the highest explanatory power for the ecological environment quality. The interaction between the DEM and population density had the largest effect on the ecological environment quality. The ecological protection of the study area can strengthen the protection of topographically sensitive areas and implement topographically appropriate ecological restoration. In low-altitude areas, urban sprawl should be strictly controlled, and ecological isolation zones should be delineated to alleviate the ecological squeeze caused by population concentration.
• The IRSEI model was more able to reflect regional differences than the traditional RSEI model. The evaluation results were more in line with the actual situation and well adapted to the regional evaluation. It is applicable to the evaluation of integrated environmental quality status of natural and semi-natural ecosystems. Moreover, it is of typical significance for regions with complex ecological elements and strong data heterogeneity.