Remote Sensing-Based Assessment of Eco-Environmental Quality Dynamics and Driving Forces in the Anhui Section of the Yangtze-to-Huaihe Water Diversion Project (2015–2024)

Abstract

The water source protection areas of the Yangtze-to-Huaihe Water Diversion Project (YHWDP) in Anhui Province serve as crucial ecological barriers to water quality protection. Quantifying their eco-environmental quality (EEQ) dynamics and driving mechanisms is critical for sustainable management. This paper calculated the Remote Sensing Ecological Index (RSEI) for the study area using Landsat satellite data (2015–2024). Temporal and spatial variation characteristics were analyzed using the Theil–Sen estimator, Mann–Kendall test, and coefficient of variation. Future trends were predicted using the Hurst exponent. Finally, the Geodetector model was applied to assess the impact of driving factors. EEQ exhibited a declining trend (p < 0.05), with significant intra-regional heterogeneity. Mean RSEI values ranked as follows: (1) Yangtze River–Huaihe River Connection < Yangtze River Water Northward Conveyance < Yangtze River–Chaohu Lake Water Diversion. (2) From 2015 to 2024, eco-environmental quality improved significantly, showing a spatial pattern of “south > north, east > west.” (3) Overall EEQ changes were characterized by slight to moderate fluctuations. Stability rankings: Yangtze River–Huaihe River Connection > Yangtze River–Chaohu Lake Water Diversion > Yangtze River Water Northward Conveyance. (4) Geodetector analysis identified precipitation, impervious area, and vegetation coverage as the primary factors influencing EEQ in the YHWDP’s water source protection areas. This study reveals ecological changes in the YHWDP region and validates the effectiveness of the comprehensive evaluation method. The findings provide actionable insights for ecological protection in large-scale water diversion projects.

1. Introduction

A favorable ecological environment provides advantageous natural conditions for human survival and development [1]. The rapid economic development has led to increasing conflicts between humans and nature, gradually disrupting the ecological balance. In recent years, a series of ecological issues, such as air pollution and land degradation, have garnered widespread attention among scholars. A comprehensive assessment of regional ecological conditions, coupled with analysis of ecological quality trends and their driving mechanisms, provides a crucial foundation for regional ecological conservation and sustainable socio-economic development [2,3,4].

The Normalized Difference Vegetation Index (NDVI) has become an indispensable analytical tool in the comprehensive assessment of spatiotemporal variations in vegetation and analysis of environmental changes [5,6]. Methods such as the Pressure–State–Response (PSR) [7] and the Ecological Index (EI) [8] have also been applied to evaluate the evolutionary status of regional natural conditions. However, these research methods are relatively limited and have not been able to integrate natural and social factors in a single evaluation. The Remote Sensing Ecological Index (RSEI) [9,10] offers distinct advantages over conventional single-indicator methods and statistical evaluation approaches. By systematically integrating multiple ecological indicators and automatically weighting them according to their principal component contributions, the RSEI provides a more robust and objective framework for assessing spatiotemporal variations in ecological quality. Thus, the RSEI has become the dominant assessment model in recent ecological studies [11,12]. Chinese scholars have adopted the RSEI to assess ecological changes in various climatically representative regions, including the Yellow River Basin in the north [13], the Dianchi Lake Basin in the south [14], and the Lanxi urban agglomeration in the east [15], all yielding highly accurate results. At the same time, there are also good application examples in urban environmental assessment in Japan [16] and the eastern region of India [17]. This method has been widely applied around the world.

Previous studies have primarily focused on ecological assessments within individual river basins, while comprehensive evaluations of ecological impacts in inter-basin water diversion areas have been scarce. The Yangtze-to-Huaihe Water Diversion Project (YHWDP) region serves as a critical corridor connecting eastern and central China. Local water scarcity, coupled with high population density, has intensified the conflict between human development and the natural environment [18,19]. The purpose of this study is to evaluate the spatiotemporal changes in the ecological environment of the inter-basin area using the RSEI, explore the driving factors behind them, and predict future trends. This provides a scientific basis for the local government in the YHWDP region to implement targeted policies for protecting vegetation and the ecological environment

2. Study Area and Datasets

2.1. Study Area

The YHWDP is a major inter-catchment hydraulic engineering initiative that establishes hydrological connectivity between the Yangtze and Huaihe River basins [20]. It provides water supplies to 55 counties and districts across 15 cities in Anhui and Henan provinces. Construction of the YHWDP commenced in December 2016, with the trial operation of Phase I in the Anhui section initiated in December 2023. This project optimizes regional water resource allocation, alleviates water scarcity in parts of the Yangtze River Delta in Anhui, and enhances regional soil–water ecosystems and ecological functions [21].

The Anhui section of the Yangtze-to-Huaihe River Water Diversion Project is systematically divided into three hydraulically and geographically distinct segments (Figure 1), reflecting spatial heterogeneity in water resource allocation and engineering design principles. The Yangtze River–Chaohu Lake Water Diversion spans 7676.16 km², encompassing critical nodes such as Tongcheng City (Anqing), Zongyang County (Tongling), and Lujiang County (Hefei). Adjacent to this, the Yangtze River–Huaihe River Connection covers 12,394.50 km², linking Feixi County and Chaohu City (Hefei) with Huainan’s Tianjia’an District, Xiejiaji District, and Shou County. The Yangtze River Water Northward Conveyance extends across 12,437.29 km², integrating water delivery infrastructure across Bozhou’s Qiaocheng District, Guoyang/Lixin/Mengcheng Counties, Fuyang’s Taihe County/Jieshou City/Linquan County/Yingquan–Yingdong Districts, and Huainan’s Fengtai County/Maoji Experimental Zone/Panji District. This tripartite division optimizes hydraulic efficiency through terrain-adaptive conduit geometries and flow parameter calibration, balancing large-area coverage with localized hydraulic stability while maintaining water transfer capacity across varying topographic gradients.

2.2. Datasets

The remote sensing data utilized in this study comprises Landsat 8 OLI satellite imagery procured from the Geospatial Data Cloud (https://www.gscloud.cn) spanning the vegetation growing season (April to July) from 2015 to 2024. Elevation data were obtained from the ASTER GDEM, which has a 30 m resolution. Using ENVI (5.6 version) software, preprocessing steps such as cloud removal, radiometric correction, atmospheric correction, and cropping were conducted. Additionally, water bodies were masked using the Modified Normalized Difference Water Index (MNDWI). All data were georeferenced to WGS 1984 UTM Zone 49 N with a spatial resolution of 30 m × 30 m to ensure consistency in spatial resolution and projection. Socioeconomic data, collected at the county/district level, were derived from statistical yearbooks and water resources bulletins of Anhui Province and its cities.

3. Methods

This study employs the Sen slope, Mann–Kendall trend analysis, and Hurst exponent to assess spatiotemporal variations in RSEI from 2015 to 2024 and predict future ecological trends. The geographical detector method was applied to quantitatively identify the spatial heterogeneity and underlying driving mechanisms of RSEI variations (Figure 2).

3.1. RSEI Model

As a comprehensive ecosystem assessment tool, the RSEI algorithm combines four fundamental environmental components: greenness, wetness, dryness, and heat, all of which are directly measurable through spectral analysis of satellite imagery. NDVI serves as an indicator of vegetation greenness and is utilized to demonstrate the state of the environment. WET stands for wetness. The Normalized Difference Built-up and Soil Index (NDBSI) characterizes land surface dryness. Land Surface Temperature (LST) reflects the local climate variations induced by environmental changes.

(1)

Retrieval of vegetation

NDVI serves as a widely adopted remote sensing metric for quantifying and monitoring vegetation growth and coverage status, mathematically defined as follows:

$$NDVI={\displaystyle \frac{{\rho }_{NIR}{-\rho }_R}{{\rho }_{NIR}{+\rho }_R}}$$

(1)

where ${\mathsf{\rho }}_{\mathrm{N}\mathrm{I}\mathrm{R}}$ and ${\mathsf{\rho }}_{\mathrm{R}}$ represent the reflectance values in the near-infrared and red spectral bands, respectively.

(2)

Retrieval of land surface moisture

The wetness component serves as a spectral indicator of hydrological conditions, quantitatively characterizing moisture levels in three key surface features: soil matrix water retention, liquid water bodies, and vegetation water content. For Landsat OLI sensors, this component is derived through the following tasseled cap transformation formula:

$${WET}_{OLI}=0.151{\rho }_B+{0.1973\rho }_G+{0.3283\rho }_R+{0.3407\rho }_{NIR}-{0.7117\rho }_{SWIR1}-{0.4559\rho }_{SWIR2}$$

(2)

(3)

Retrieval of land surface moisture

The accelerated urbanization process and intensive anthropogenic disturbances have triggered land surface modifications. Impervious surfaces and exposed soils progressively supplant natural vegetation cover, which induces ecological degradation. Hu and Xu developed NDBSI through spectral feature fusion, integrating the Index-based Built-up Index (IBI) and Soil Index (SI) as follows:

$$SI={\displaystyle \frac{\left({\rho }_{SWIR1}{+\rho }_R\right)-\left({\rho }_{NIR}{+\rho }_B\right)}{\left({\rho }_{SWIR1}{+\rho }_R\right)+\left({\rho }_{NIR}{+\rho }_B\right)}}$$

(3)

$$IBI={\displaystyle \frac{2\times {\rho }_{SWIR1}/({\rho }_{SWIR1}+{\rho }_{NIR})-\left[{\rho }_{NIR}/\left({\rho }_{NIR}{+\rho }_R\right)+{\rho }_G/({\rho }_G+{\rho }_{SWIR1})\right]}{2\times {\rho }_{SWIR1}/({\rho }_{SWIR1}+{\rho }_{NIR})+\left({\rho }_{NIR}/\left({\rho }_{NIR}{+\rho }_R\right)+{\rho }_G/({\rho }_G+{\rho }_{SWIR1})\right)}}$$

(4)

$$NDBSI={\displaystyle \frac{SI-IBI}{2}}$$

(5)

(4)

Retrieval of land surface temperature

LST was evaluated as follows:

$$L=gain\times DN+bias$$

(6)

$$T_b={\displaystyle \frac{K_2}{ln\left(\frac{K_1}{L}-1\right)}}$$

(7)

$$LST={\displaystyle \frac{T}{1+\frac{\lambda T}{\rho }ln\epsilon }}$$

(8)

λ denotes the spectral wavelength of terrestrial emitted radiance, ρ represents the first radiation constant (1.438 × 10^{− 2} m·K), $\mathsf{\epsilon }$ characterizes the dimensionless land surface emissivity ranging from 0 to 1, and T indicates the at-satellite brightness temperature (K) measured through atmospheric window channels.

(5)

Acquisition of RSEI

Principal component analysis (PCA) was performed on the four components of the RSEI, namely the NDVI, WET, LST, and NDBSI, and the first principal component was defined as the RSEI₀. The RSEI₀ was then standardized to obtain the RSEI to facilitate the measurement as follows:

$${RSEI}_0=1-\left\{{PC}_1\left[f\left(NDVI,WET,NDBSI,LST\right)\right]\right\}$$

(9)

After normalization, RSEI is obtained, which is a unitless value between 0 and 1. The larger the value, the better the ecological quality of the region. The formula is as follows:

$$RSEI={\displaystyle \frac{{RSEI}_0-{RSEI}_{0,min}}{{RSEI}_{0,max}-{RSEI}_{0,min}}}$$

(10)

where RSEI₀ is the first principal component of the four indicators; RSEI_0,min is the minimum value of the RSEI₀; and RSEI_0,max is the maximum value of the RSEI₀.

RSEI adopts a normalized scale ranging from 0 (degraded) to 1(optimal). The numerical magnitude exhibits a positive correlation with ecosystem health status. With an interval of 0.2, the levels of RSEI were classified into five groups: poor (0–0.2), fair (0.2–0.4), moderate (0.4–0.6), good (0.6–0.8), and excellent (0.8–1.0).

3.2. Combined T–S Estimator with M–K Test

The combined application of the Theil–Sen median trend analysis and the Mann–Kendall test provides a robust method for assessing trend changes in time series data. This method exhibits stronger resistance to errors and outliers in the data, thereby enhancing analytical accuracy. It evaluates trends by calculating the median of slopes between all pairwise data points, with the mathematical formula expressed as follows:

$$\beta =Median\left({\displaystyle \frac{{RSEI}_j-{RSEI}_i}{j-i}}\right), 2015\le \mathrm{i}<\mathrm{j}\le 2024$$

(11)

where median denotes the median value; β is the slope of RSEI change. When the coefficient β is greater than zero, it indicates an upward (improving) trend in RSEI; conversely, a negative β value (β < 0) suggests a downward trend.

The Mann–Kendall test is a non-parametric statistical method that can be applied to determine the significance of trends in time series data. It imposes no distributional assumptions on the data and is robust against outliers. The test statistic is calculated as follows:

$$Z=\left\{\begin{array}{c}\left(S-1\right)/\sqrt{var\left(S\right)},S>0\\ \left(S+1\right)/\sqrt{var\left(S\right)},S<0\end{array}\right.$$

(12)

$$S={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^{n}sgn\left({RSEI}_j-{RSEI}_i\right)$$

(13)

$$\mathrm{s}\mathrm{g}\mathrm{n}\left({RSEI}_j-{RSEI}_i\right)=\left\{\begin{array}{c}1,{RSEI}_j-{RSEI}_i>0\\ 0,{RSEI}_j-{RSEI}_i=0\\ -1,{RSEI}_j-{RSEI}_i<0\end{array}\right.$$

(14)

$$var\left(S\right)=n\left(n-1\left)\right(2n+5\right)/18$$

(15)

where n is the length of the time series, sgn () denotes the sign function; Z is the significance test statistic, whose value ranges from (−∞, +∞).

The trend categories for the Theil–Sen (T–S) estimator and Mann–Kendall (M–K) test are summarized in

Table 1. Trend categories of the Theil–Sen estimator and Mann–Kendall method.

$\mathit{\beta }$	Z	Trend Type	Trend Features
<0	≤−1.96	SD	Significant decrease
<0	−1.96 to 1.96	NSD	No significant decrease
0	−1.96 to 1.96	NC	No change
>0	−1.96 to 1.96	NSI	No significant increase
>0	≥1.96	SI	significant increase

NOTE: β is the T–S slope coefficient, which quantifies the magnitude of the temporal trend in RSEI values; Z is the M–K standardized test statistic, which represents the significance level of detected trends following M–K hypothesis testing.

.

3.3. Coefficient of Variation (CV)

The coefficient of variation (CV) reflects the relative variation in geographic data and serves as a critical indicator for measuring the stability of a time series. By calculating the ratio of the standard deviation to the mean, it quantifies the relative variability of a variable, making it widely applicable for assessing the degree of variation in time series data. Therefore, the CV is employed to analyze the stability of the RSEI. The calculation formula is as follows:

$$\mathrm{C}\mathrm{V}={\displaystyle \frac{\sigma }{\mu }}$$

(16)

CV denotes the coefficient of variation, $\sigma$ represents the standard deviation of the RSEI time series, and $\mu$ represents the average value of the RSEI time series. A lower CV value indicates more stable interannual variations in the RSEI, while a higher CV value suggests greater volatility in its year-to-year fluctuations.

3.4. Hurst Exponent

The Hurst exponent (H), obtained through rescaled range analysis, serves as a metric for characterizing the long-term dependence and persistence of time series data. The Hurst exponent (H, 0 < H < 1) is generally divided into three cases: 0.5 < H < 1 indicates long-term memory effects, meaning the future trend will align with the past; H = 0.5 indicates that the time series is an independent random sequence; and 0 < H < 0.5 exhibits anti-persistent dynamics, suggesting that the future trend will likely reverse past patterns (

Table 2. Classification of RSEI change trends.

Hurst	Sen’s Slope	Future Trends
0.5 < H < 1	β > 0	Improvement
	β < 0	Degradation
H = 0.5	-	Uncertain
0 < H < 0.5	β > 0	Improvement
	β < 0	Degradation

).

3.5. Quantitative Analysis Methods for Geospatial Differentiation

The Geographical Detector (GeoDetector) represents a spatially explicit statistical framework designed to quantify spatial heterogeneity and identify its underlying driving factors. The formula for spatial differentiation and driving force detection is expressed mathematically as follows:

$$q=1-{\displaystyle \frac{{\sum }_{h=1}^L{N}_h{\sigma }_h^2}{N{\sigma }^2}}$$

(17)

L denotes the stratification (or classification/partitioning) of either variable Y or explanatory factor X, N_h represents the sample size within stratum h and N corresponds to the total population size, respectively. The terms σh² and σ² indicate the variances of Y-values within stratum h and across the entire study region, respectively. The q-statistic spans a range from 0 to 1, where a higher value indicates a greater degree of spatial heterogeneity in Y. If the stratification is based on the independent variable X, elevated q-values further imply greater explanatory capacity of X regarding Y’s spatial distribution. Detailed interaction effects are presented in

Table 3. Interaction relation.

Criteria	Interaction
q(X1 ∩ X2) < Min (q(X1), q(X2))	Nonlinear Weakening
Min(q(X1), q(X2)) < q(X1 ∩ X2) < Max(q(X1), q(X2))	Single-Factor Nonlinear Weakening
q(X1 ∩ X2) > Max(q(X1), q(X2))	Two-Factor Enhancement
q(X1 ∩ X2) = q(X1) + q(X2)	Mutual Independence
q(X1 ∩ X2) > q(X1) + q(X2)	Nonlinear Enhancement

.

4. Results

4.1. RSEI Model Applicability Validation

PCA was employed to reduce the dimensionality of the dataset while retaining the most significant variance among the ecological indicators, facilitating a comprehensive assessment of the RSEI framework. The first principal component (PC₁) accounts for 78.0% to 84% of the total variance, as shown in

Table 4. Mean values of each indicator and the contribution of PC₁ from 2015 to 2024.

Year	RSEI Value	NDVI Value	WET Value	NDBSI Value	LST Value	PC1 Value	PC1 Contribution Rate
2015	0.6232	0.4412	0.5248	0.6984	0.6095	0.2863	79.40%
2020	0.5374	0.5625	0.4815	0.6563	0.5333	0.2605	83.74%
2024	0.5247	0.3879	0.4481	0.6952	0.4757	0.2369	78.17%

, indicating that it effectively captures the dominant features of the four indicators.

4.2. Spatial Distribution of EEQ

The RSEI values were categorized into five distinct ecological quality tiers using the Jenks optimization method, with the following categorical designations: Poor (0–0.2), Fair (0.2–0.4), Moderate (0.4–0.6), Good (0.6–0.8), and Excellent (0.8–1.0). The RSEI in the water source protection areas along the Anhui section of the YHWDP exhibits a spatial distribution pattern characterized by lower values in the western and northern regions and higher values in the eastern and southern regions (Figure 3).

Significant differences and dynamic trends are observed among the three segments: Yangtze River–Chaohu Lake Water Diversion, the Yangtze River–Huaihe River Connection, and the Yangtze River Water Northward Conveyance. According to

Table 5. RSEI classification area and proportion.

Class	Range	RSEI (2015)		RSEI (2020)		RSEI (2024)
		Area (km2)	Proportion (%)	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)
Excellent	0.8–1.0	10924.20	37.44	3526.90	12.42	3637.24	12.43
Good	0.6–0.8	6474.50	22.19	7118.85	23.13	7608.52	26.34
Moderate	0.4–0.6	4637.73	15.89	8837.25	30.31	9018.88	30.82
Fair	0.2–0.4	4838.39	16.58	6766.31	23.47	6444.21	21.97
Poor	0.0–0.2	2305.91	7.9	2930.69	10.67	2471.84	8.45

, the proportions of poor and fair RSEI zones in 2024 are 8.45% and 21.97%, respectively, while the combined area of Grade 1–3 zones (representing better ecological conditions) accounts for 65.98%. The mean RSEI values across grades initially increased and then declined over time. For example, Grade 4 (moderate ecological status) shows a decreasing trend: 0.6232 (2015), 0.5374 (2020), and 0.5247 (2024).

The ecological quality assessed by RSEI remains at a median quality level (RSEI = 0.4–0.6), with mean values of 0.4026 in 2015, 0.6917 in 2020, and 0.6917 in 2024. However, the Yangtze River–Chaohu Lake Water Diversion shows an overall improving trend in RSEI-based ecological quality (Figure 4). Spatially, high RSEI values are concentrated along the YHWDP channels, the Yangtze River banks, the southern shore of Chaohu Lake, and surrounding water bodies.

As shown in Figure 5, the ecological quality assessed by RSEI remains at a poor level, though the area with significantly low RSEI in the Yangtze River–Huaihe River Connection shows a shrinking trend. The mean RSEI values for the Yangtze River–Huaihe River Connection are 0.4821 (2015), 0.5147 (2020), and 0.4789 (2024). Notably, Wuwei County, Shushan District, and Xiejiaji District exhibit a yearly decline in RSEI, indicating a persistent deterioration in their ecological environments. Spatially, RSEI degradation is concentrated in the southern regions. For example, the Binhu New District shows fragmented ecological quality distribution (marked by red patches), likely linked to lake reclamation or over-tourism development. In contrast, areas with high ecological quality are predominantly clustered along the southern shore of Lake Wabu.

As shown in Figure 6, the RSEI index reflects an overall declining trend in regional ecological quality. The mean RSEI values are 0.6552 (2015), 0.5577 (2020), and 0.4842 (2024), indicating progressive deterioration. However, certain areas still exhibit significantly poor ecological conditions. JiaoCheng District, Wuoyang County, and Yingshang County experienced a transition from “good” to “moderate” and finally to “poor” RSEI levels, signaling sustained degradation in ecological quality. Panji District shows a year-on-year decline in RSEI, with its ecological environment worsening markedly.

4.3. Spatial Change Characteristics of EEQ

A combined non-parametric approach integrating the T–S estimator and M–K test was systematically applied to investigate spatiotemporal variations in Ecological Environment Quality (EEQ) along the Anhui segment of the YHWDP during the period 2015–2024 (Figure 7,

Table 6. Grading table of RSEI change amplitude in the Yangtze-to-Huaihe Water Diversion Project.

Segment	Trend Type	2015–2024
		Area/km2	Proportion/%
Yangtze River–Chaohu Lake Water Diversion	No change	1968.36	25.64
	No significant increase	2936.97	38.26
	Significant increase	57.33	0.75
	No significant decrease	2057.56	26.80
	Significant decrease	26.93	0.35
Yangtze River–Huaihe River Connection	No change	1896.21	15.30
	No significant increase	4633.03	37.38
	Significant increase	34.06	0.27
	No significant decrease	5146.40	41.52
	Significant decrease	41.88	0.34
Yangtze River Water Northward Conveyance	No change	895.43	7.20
	No significant increase	8366.28	67.27
	Significant increase	202.44	1.63
	No significant decrease	2056.50	16.53
	Significant decrease	91.76	0.74

).

4.4. Spatial Stability of the RSEI

The EEQ fluctuations in the water source protection areas along the Anhui segment of the YHWDP main water conveyance route were analyzed using the coefficient of variation (CV) of the RSEI. The variability was categorized into five levels via the natural breaks method (Jenks): Stability (CV ≤ 0.30); Slight changes (0.30 < CV ≤ 0.60); Moderate variation (0.60 < CV ≤ 0.90); Significant changes (0.90 < CV ≤ 1.20); and Drastic changes (1.20 < CV ≤ 1.50). This classification procedure yielded a comprehensive spatial-temporal visualization of RSEI CV patterns throughout the study region during the period 2015–2024 (Figure 8).

The average coefficient of variation (CV) for RSEI changes was 0.4281, indicating that slight fluctuation (66.57%) and moderate fluctuation (9.24%) dominated the overall variability, collectively accounting for 75.81% of the study area (

Table 7. Statistics of RSEI coefficient of variation (CV) area and proportion in the study area from 2015 to 2024.

CV	Yangtze River–Chaohu Lake Water Diversion		Yangtze River–Huaihe River Connection		Yangtze River Water Northward Conveyance
	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)
Stability	1141.11	14.87	862.64	6.96	5928.66	47.67
Slight changes	5109.67	66.57	8215.37	66.28	116.11	0.93
Moderate variation	709.46	9.24	2507.26	20.23	3705.16	29.79
Significant changes	68.21	0.89	151.61	1.22	1486.79	11.95
Drastic changes	15.58	0.20	5.51	0.04	282.68	2.27

). Extreme fluctuation zones (CV > 1.2) represented 0.2% of the total area and were concentrated near Caizi Lake, water conveyance channels, and the north bank of the Yangtze River, particularly in the paddy fields highlighted in Figure 8a.

As shown in Figure 8b, the mean coefficient of variation (CV) for RSEI changes was 0.4991, indicating slight fluctuation (66.28%) as the dominant variability category, followed by moderate fluctuation (20.23%), collectively accounting for 86.51% of the study area. Pronounced RSEI variability was primarily distributed in lake-, river-, and pond-dense regions, particularly around Hefei City. These patterns likely stem from the interplay of topographic complexity, climatic variability, and intense human activities (e.g., urban sprawl).

As shown in Figure 8c, the average coefficient of variation (CV) for RSEI changes was 0.1732, with slight fluctuation, moderate fluctuation, and higher fluctuation collectively accounting for 41.74% of the study region. Pronounced RSEI variability was predominantly found in the northern Huaihe River Basin, where intensive agricultural activities have led to significant anthropogenic disturbances.

The ecological environment quality across the study region predominantly exhibited slight to moderate fluctuations, with coefficient of variation (CV) values of 0.4281 for the Yangtze River–Chaohu Lake Water Diversion, 0.4991 for the Yangtze River–Huaihe River Connection, and 0.1732 for the Yangtze River Water Northward Conveyance. By aggregating stable and slightly fluctuating zones, the stability ranking was: Yangtze River–Huaihe River Connection (86.51%) > Yangtze River–Chaohu Lake Water Diversion (78.81%) > Yangtze River Water Northward Conveyance (29.79%).

4.5. Prediction of Future Trends in Ecological Environment Quality

Yangtze River–Chaohu Lake Water Diversion: The RSEI exhibited an average Hurst index of 0.5319, indicating a moderate persistence in ecological quality trends. It suggests that 27.78% of the study area is likely to experience enhanced ecological conditions, while 38.38% may face sustained deterioration (Figure 9).

Yangtze River–Huaihe River Connection: The RSEI analysis reveals an average Hurst index value of 0.5428. Approximately 30.02% of the study area is anticipated to undergo ecological improvement, while a more substantial proportion (49.49%) is expected to experience continued degradation (

Table 8. Statistics of Sen–Hurst coefficient area and proportion for RSEI in the study area (2015–2024).

CV	Yangtze River–Chaohu Lake Water Diversion		Yangtze River–Huaihe River Connection		Yangtze River Water Northward Conveyance
	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)	Area (km2)	Proportion (%)
Uncertain	1968.36	25.64	1896.21	15.30	895.43	7.20
Improvement	2132.66	27.78	3720.89	30.02	1754.32	14.11
Degradation	2946.12	38.38	6134.14	49.49	8962.36	72.06

).

Yangtze River Water Northward Conveyance: The average Hurst index of RSEI is 0.6327, with only 14.11% of areas likely to improve, while 72.06% are expected to worsen persistently. Significant RSEI variations are concentrated in the northern Huaihe, where intensive farming leads to pronounced human disturbance.

Overall Trend: The future ecological environment quality of the study region is generally pessimistic (Figure 10). The average Hurst indices are 0.5319 for Yangtze River–Chaohu Lake Water Diversion, 0.5428 for Yangtze River–Huaihe River Connection, and 0.6327 for Yangtze River Water Northward Conveyance. The projected deterioration follows this order: Yangtze River Water Northward Conveyance (72.06%) > Yangtze River–Huaihe River Connection (49.49%) > Yangtze River–Chaohu Lake Water Diversion (38.38%).

4.6. Quantitative Attribution Analysis of Driving Factors for Spatial Differentiation of RSEI

Integrating the natural and socioeconomic conditions of the water source conservation areas along the Jianghuai Water Diversion Project’s Anhui Section, this study selected ten variables—including population, GDP (Gross Domestic Product), built-up area, precipitation, and mean NDVI—as influencing factors for correlation analysis with the RSEI. The Geodetector method was employed to analyze the driving mechanisms behind dynamic changes in EEQ. The independent variables (X) include the following: X1: Annual mean temperature (°C); X2: Annual mean precipitation (mm); X3: Vegetation coverage (%); X4: Mean NDVI; X5: Elevation (m); X6: Slope (°); X7: Built-up area; X8: Land Surface Temperature (LST); X9: GDP per unit area (10,000 yuan/km²); X10: Population density (persons/km²).

The analysis of critical topographic parameters across the three major water diversion segments reveals distinct geomorphological characteristics (

Table 9. Parameters of driving factors influencing RSEI changes.

Driving Factors		Variable	Yangtze River–Chaohu Lake Water Diversion			Yangtze River–Huaihe River Connection			Yangtze River Water Northward Conveyance
			2015	2020	2024	2015	2020	2024	2015	2020	2024
natural factors	climatic factors	X1	18.20	17.60	17.40	17.50	16.30	17.40	16.00	16.60	14.80
		X2	1421.30	1877.90	1776.30	1234.60	1634.20	1293.10	808.60	1020.90	829.30
	vegetation factors	X3	24.86	24.86	24.86	23.43	23.43	23.43	26.91	26.91	26.91
		X4	0.50	0.56	0.21	0.35	0.34	0.22	0.26	0.56	0.29
	topographic factors	X5	49.32	49.32	49.32	38.20	38.20	38.20	30.68	30.68	30.68
		X6	5.16	5.16	5.16	2.77	2.77	2.77	2.02	2.02	2.02
	surface disturbance factors	X7	619.87	863.75	1015.04	1174.01	1443.37	2100.10	1343.59	1502.63	1623.43
		X8	0.44	0.48	0.75	0.51	0.48	0.54	0.60	0.56	0.60
socio-economic	economic factors	X9	360.00	288.00	279.00	323.00	319.00	318.00	466.00	502.00	445.00
	population size	X10	972.77	1471.97	2348.36	270.74	1824.07	2580.59	1047.92	1742.43	2495.18

). The Yangtze River–Chaohu Lake Water Diversion exhibits the highest mean elevation (49.32 m) and steepest mean slope (5.16°), indicating significant terrain complexity that may influence hydraulic dynamics and sediment transport patterns. In contrast, the Yangtze River–Huaihe River Connection features a moderate elevation (38.20 m) and gentler slope (2.77°), favoring efficient water conveyance over long distances while requiring vigilance against anthropogenic disturbances in low-lying zones. The Yangtze River Water Northward Conveyance, with the lowest elevation (30.68 m) and minimal slope (2.02°), presents a flat terrain profile that heightens flood susceptibility, necessitating engineered safeguards to maintain operational reliability.

The single factor detection results indicated that ecological environment quality (as dependent variable Y) was input into the optimal parameter geographical detector model. A higher q-value signifies a more robust explanatory capacity of the factor influencing ecological environment quality (

Table 10. Single-factor influence contribution.

	X1	X2	X3	X4	X5	X6	X7	X8	X9	X10
q statistic	0.37	0.76	0.06	0.08	0.17	0.30	0.34	0.64	0.00	0.50
p value	1.00	0.56	0.97	0.95	0.93	0.78	0.49	0.77	1.00	0.91

). The q-values of factors ranked in descending order were: annual precipitation (0.76), land surface temperature (LST) (0.64), population density (0.50), mean annual temperature (0.37), built-up area (0.34), slope (0.30), elevation (0.17), NDVI mean (0.08), vegetation coverage (0.06), and GDP per unit area (0.00). This ranking demonstrates that annual precipitation and land surface temperature-LST serve as the primary drivers influencing the ecological environment quality in water source conservation areas along the YHWDP Route.

The interaction effects between factors demonstrated significant explanatory power for ecological environment quality. Notably, the interaction between NDVI and elevation, slope, and built-up area yielded the highest synergistic q-value (0.95), indicating their dominant combined influence on ecological conditions. Secondarily, interactions involving land surface temperature (LST) with mean annual temperature, precipitation, vegetation coverage, elevation, and slope all exhibited q-values exceeding 0.91. Other factor interactions consistently showed q-values above 0.5, representing substantial improvements over single-factor q-values. These findings emphasize that ecological restoration strategies must systematically integrate multi-factor spatial distribution patterns, prioritizing precipitation regimes, vegetation dynamics, and thermal conditions (LST) to optimize governance frameworks and enhance intervention efficacy.

As illustrated in Figure 11, ecological drivers operate through synergistic interactions rather than isolated effects. The dominant interaction patterns—nonlinear enhancement (e.g., NDVI × elevation) and bilinear enhancement (e.g., LST × slope)—reveal that spatial heterogeneity in ecosystem quality arises from compounded multi-factor processes. Particularly, vegetation-terrain interactions (NDVI × elevation/slope) demonstrated the strongest explanatory power for RSEI variations, with q-values surpassing 0.9. In contrast, anthropogenic impacts exhibited weaker direct effects, primarily manifesting indirectly through land use/cover modifications.

5. Discussion

During the period 2015–2024, the ecological environment quality of water source conservation areas along the YHWDP route in Anhui Province exhibited a fluctuating yet overall deteriorating trend, necessitating heightened focus on future ecological restoration efforts. Notably, the spatial extent of ecologically degraded zones gradually decreased over this period, reflecting a progressive mitigation of environmental stress. Yangtze River–Huaihe River Connection demonstrated the most unstable Remote Sensing Ecological Index (RSEI) dynamics (mean RSEI = 0.4789), indicating heightened ecological vulnerability. Yangtze River Water Northward Conveyance ranked second in ecological fragility (RSEI = 0.4842), while the Yangtze River–Chaohu Lake Water Diversion showed relatively better conditions (RSEI = 0.6917). The spatial distribution of RSEI exhibited a pronounced latitudinal and longitudinal gradient, with elevated values distinctly clustered in southern and eastern regions, while lower magnitudes prevailed in northern and western areas. Areas with ”significant degradation” dominated the ecological quality classification, particularly in urbanized regions like Hefei City, where vegetation cover was sparse. Higher-quality ecosystems clustered in densely vegetated zones, confirming vegetation coverage as a critical stabilizing factor.

This study classified the ecological environment quality of water source conservation areas along the Anhui section of the YHWDP corridor into five levels from 2015 to 2024. Statistical analysis demonstrated that both vegetation vigor (NDVI) and atmospheric moisture content (WET) significantly enhanced ecological quality indicators, while thermal stress (LST) and dryness (NDBSI) showed significant negative impacts. Single-factor analysis demonstrated the following explanatory power ranking for ecological drivers: LST > NDBSI > NDVI > WET, with LST achieving the highest q-value, highlighting its dominant role in explaining the spatial heterogeneity of the Remote Sensing Ecological Index (RSEI). Dual-factor interaction effects further indicated that LST × NDBSI had the strongest combined influence on RSEI variability, whereas NDVI × WET interactions were comparatively weaker. Overall, natural factors—including precipitation patterns, temperature gradients, and topographic elevation—were identified as primary drivers of spatial differentiation in ecological quality. In contrast, anthropogenic activities indirectly impacted the environment through land use/cover modifications, such as urbanization-driven increases in LST and expansion of built-up areas [16,17]. Meanwhile, the increase in NDVI contributed to the improvement of regional ecological environment quality [22,23]. Therefore, in the process of urban expansion and construction, attention should be paid to the reasonable protection of local natural resources. At the same time, ecological protection policies such as afforestation and water purification should be implemented, as they can help alleviate a series of environmental problems. This study is consistent with the conclusions of previous literature. These findings emphasize the need to integrate natural climatic–topographic dynamics with targeted land-use regulations to mitigate human-induced pressures and enhance ecological resilience in large-scale water diversion systems.

6. Conclusions

This research conducted a comprehensive investigation into the spatiotemporal dynamics and underlying driving mechanisms of the RSEI within water source conservation areas along the Anhui segment of the YHWDP corridor from 2015 to 2024. Key findings include the following:

(1)

The RSEI in these areas exhibited a fluctuating upward trend followed by a decline, with increasing spatial heterogeneity in ecological quality. The ecological quality ranking across subregions was: Yangtze River–Huaihe River Connection (86.51%) < Yangtze River Water Northward Conveyance (78.81%) < Yangtze River–Chaohu Lake Water Diversion (29.79%). Over the nine-year period, areas classified as “Good” and “Moderate” ecological quality significantly expanded, while “Poor” and “Very Poor” categories decreased, indicating overall ecological improvement. However, spatial imbalances persist, with pronounced regional disparities in stability: Yangtze River–Huaihe River Connection (40.66% stable/slight fluctuation) > Yangtze River–Chaohu Lake Water Diversion (30.19%) > Yangtze River Water Northward Conveyance (25.02%).

(2)

Yangtze River Water Northward Conveyance demonstrated relatively high ecological quality but greater RSEI volatility, attributed to vegetation dynamics in cropland protection zones. In contrast, the Yangtze River–Huaihe River Connection exhibited lower baseline ecological quality but milder fluctuations (slight to moderate), with intense ecological changes observed near Hefei City.

(3)

Annual precipitation, impervious surface area, and vegetation coverage emerged as primary drivers of ecological quality, interacting synergistically with annual average temperature and land cover to significantly enhance explanatory power over RSEI variations. Human activities, particularly land-use changes, markedly amplified RSEI dynamics, demonstrating strong collaborative effects with natural factors. All factor interactions exhibited nonlinear enhancement, emphasizing the critical role of coupled natural-anthropogenic processes in shaping ecological patterns.

This study calculated the RSEI for the YHWDP region and analyzed the temporal and spatial characteristics of changes, as well as future trends. Subsequently, the impact of different driving factors was evaluated, providing guidance for future environmental protection policy formulation. These approaches will empower watershed management departments to enhance the precision and scientific rigor of strategy formulation, effectively coordinating farmland protection, urban expansion control, and vegetation restoration. Such coordinated efforts aim to address ecological vulnerabilities in specific areas within large-scale water diversion systems. Due to space constraints, this study did not conduct a detailed analysis of the impact of different spatial scales of Landsat remote sensing data on the computational results. Future research should conduct comparative studies at different scales to better implement the comprehensive ecological environment assessment method developed in this study, and thus apply it to ecological protection in other watersheds and cross-basin water diversion work.