Spatiotemporal Analysis of Eco-Geological Environment Using the RAGA-PP Model in Zigui County, China

Abstract

The Three Gorges Reservoir Area in China presents a critical conflict between industrial development and ecological conservation. It functions as a key hub for water management, energy production, and shipping, while also serving as a vital zone for ecological and environmental protection. Focusing on Zigui County, this study developed a 16-indicator evaluation system integrating geological, ecological, and socioeconomic factors. It utilized the Analytic Hierarchy Process (AHP), coefficient of variation (CV), and the Real-Coded Accelerating Genetic Algorithm-Projection Pursuit (RAGA-PP) model for evaluation, the latter of which optimizes the projection direction and utilizes PP to transform high-dimensional data into a low-dimensional space, thereby obtaining the values of the projection indices. The findings indicate the following: (1) The RAGA-PP model outperforms conventional AHP-CV methods in assessing Zigui County’s eco-geological environment, showing superior accuracy (higher Moran’s I) and spatial consistency. (2) Hotspot analysis confirms these results, revealing distinct spatial patterns. (3) From 2000 to 2020, “bad” quality areas decreased from 17.31% to 12.33%, while “moderate” or “better” zones expanded. (4) This improvement reflects favorable natural conditions and reduced human impacts. These trends underscore the effectiveness of China’s ecological civilization policies, which have prioritized sustainable development through targeted environmental governance, afforestation initiatives, and stringent regulations on industrial activities.

1. Introduction

The Three Gorges Reservoir Area represents one of China’s most ecologically critical yet vulnerable regions, serving as both a strategic hub for water resources, energy production, and navigation and a vital ecological barrier for the Yangtze River Basin. However, rapid industrialization, coupled with complex geomorphology and intense anthropogenic pressures, has exacerbated environmental degradation, including soil erosion, habitat fragmentation, and geological instability [1]. This dual role—economic engine versus ecological safeguard—creates an urgent need for advanced, quantitative assessments to balance development with conservation. Remote sensing technologies, combined with innovative modeling approaches, are pivotal in addressing these challenges by enabling large-scale, dynamic monitoring of eco-geological conditions [2].

Recent advances in eco-geological evaluations have shifted from single-attribute analyses to integrated frameworks that incorporate socioeconomic and environmental dimensions [3]. Establishing a scientifically sound and reasonable evaluation index system is fundamental to research on the eco-geological environment [4], and the appropriateness of the selected indicators will directly influence the accuracy and reliability of the results in subsequent research. Scholars, both domestically and internationally, have begun to shift their focus towards the evaluation methods of eco-geological environment quality, specifically concentrating on the development of evaluation systems and models [5]. The increasing integration of information technology and scientific modeling, along with advancements in interdisciplinary research, has led to the incorporation of fundamental theories and mathematical methods into the evaluation of eco-geological environments. Commonly employed research methodologies include the ecological footprint method [6], system modeling approach [7], hierarchical analysis technique [8], set-pair analysis method [9], fuzzy synthesis approach [10], state-space analysis [11], principal component analysis, comprehensive evaluation technique [12], and projection tracing method [13].

Traditional weighting methods like Analytic Hierarchy Process (AHP), Entropy Weighting [14], and coefficient of variation (CV) have been extensively applied in eco-geological assessments due to their respective strengths. AHP effectively incorporates expert knowledge for structured decision-making, while CV provides objective, data-driven weights based on indicator variability. However, these approaches present notable limitations: AHP’s reliance on subjective judgments may introduce bias, and CV’s pure statistical approach often overlooks critical spatial relationships in environmental systems. Emerging techniques, such as projection pursuit models, address these gaps by optimizing data dimensionality reduction while preserving critical spatial patterns [15]. Notably, the Real-Coded Accelerating Genetic Algorithm-Projection Pursuit (RAGA-PP) model offers superior objectivity by iteratively refining projection directions using evolutionary algorithms, thus minimizing human bias. However, current applications in fragile ecosystems like the Three Gorges Reservoir Area remain limited, particularly in reconciling remote sensing-derived indicators (e.g., NDVI or land subsidence) with ground-truth geological data.

The current study focuses on developing an integrated assessment framework for the eco-geological environment of Zigui County in the Three Gorges Reservoir Area, combining remote sensing data with conventional geological and socioeconomic indicators. Given the region’s complex geological conditions and prevalent environmental hazards, this research contributes to ongoing efforts toward sustainable development in reservoir areas. The proposed methodology demonstrates the potential of multi-source remote sensing observations when combined with field-based data for supporting land-use planning and ecological protection initiatives. While certain limitations remain to be addressed, the findings may provide a foundation for future studies seeking to enhance the integration of geospatial technologies with eco-geological assessments.

2. Study Area and Data Sources

2.1. Study Area

Zigui County, located in the Three Gorges Reservoir Area in Yichang City, Hubei Province, lies between 30°38′N and 31°11′N and 110°18′E and 111°0′E (Figure 1). The study area exhibits notable topographical variation, with elevations declining from the southwest to the northeast. The elevation ranges significantly, from a low of −48 m to a maximum peak of 2028 m, creating a relative elevation difference of 2076 m. The region is characterized by a distinct monsoon climate, with an annual average temperature ranging from 17 to 22 °C, which contributes to abundant and well-distributed precipitation throughout the year. The majority of rainfall occurs between April and October, with frequent heavy rain events during the wet season. Geologically, Zigui County extends from the western edge of the Zigui syncline to the Yellow Acid anticline, spanning west to east, with the southern part of the county marked by the Xianglu Mountain anticline. The county’s geology is complex, featuring a comprehensive and complete stratigraphic succession. At the end of 2022, the county’s registered population totaled 361,900, reflecting a slight decrease of 0.64% from the previous year. Economically, Zigui County achieved an annual GDP of CNY 20.504 billion, representing an annual growth rate of 5.6%. In terms of industrial structure, the distribution across sectors was 19.0% for primary industries, 26.7% for secondary industries, and 54.3% for tertiary industries.

2.2. Data Sources

The spatial information data in this study mainly came from a geospatial data cloud, which mainly included DEM images and Landsat8 data, and four 30 m resolution digital elevation models (DEMs) were provided by the geospatial data cloud for extracting the slope and topographic relief. Basic geological conditions and related vector data were obtained from regional geological maps and hydro-geological maps. Ecological and environmental data, including NDVI, NPP, and precipitation, were mainly derived from the Resources and Environmental Sciences and Data Center of the Chinese Academy of Sciences. Social and environmental data were mainly generated based on the collected population data and GDP data interpolation. Global land cover data with 30 m spatial resolution were collected from the Global Land Cover 2000 database, covering land type information in the study area. The ecological hazard point data were provided by the Guidance Center for Monitoring and Early Warning of Geological Disasters in the Three Gorges Reservoir Area for extracting the geological hazard index. The road data and river data were obtained from OpenStreetMap.

Table 1. Data sources.

Data Type	Data Sources	Times
STRM30 m DEM	http://www.gscloud.cn/ (accessed on 13 March 2023)	2022
Disaster point data	http://www.yichang.cgs.gov.cn/ (accessed on 13 March 2023)	2022
Landsat8 data	http://www.gscloud.cn/ (accessed on 13 March 2023)	2022
Geological maps	http://www.ngac.cn/ (accessed on 15 May 2023)	2023
Ecological and environmental data	http://ids.ceode.ac.cn/ (accessed on 15 May 2023)	2023
Social and environmental data	https://www.stats.gov.cn/ (accessed on 23 May 2023)	2000–2020
Land cover	http://www.globallandcover.com/ (accessed on 23 May 2023)	2000–2020
Road and river data	https://www.openstreetmap.org/ (accessed on 23 May 2023)	2000–2020

tabulates the data sources.

2.3. Selection of Evaluation Factors

Based on the background and characteristics of the eco-geological environment of Zigui County, the evaluation index system of the eco-geological environment of Zigui County was established. It includes three subsystems: geological environment, ecological environment, and social economy, and the selection factors of each subsystem are shown in

Table 2. Construction of eco-geological environment evaluation index system.

Factors	Indicators	Number	Types
Geological environment	DEM	1	Negative
	Slope	2	Negative
	Surface roughness	3	Positive
	Surface cutting depth	4	Negative
	Structural density	5	Negative
	Disaster site density	6	Negative
	Engineering geological strata	7	/
Ecological environment	NDVI	8	Positive
	NPP	9	Positive
	Drainage distance	10	/
	Annual precipitation	11	Positive
	Land-use	12	/
Socioeconomy	Road density	13	Negative
	Population density	14	Negative
	GDP	15	Negative
	Night-light data	16	Negative

. The positive index can indicate that it plays a supporting role in the eco-geological environment, while the negative index indicates that it has an inhibitory effect on the eco-geological environment.

(1)

Geological environment

As the basic condition of eco-geological environment assessment, the geological environment background is the most critical part of the evaluation process, and the ground elevation, topographic slope, surface roughness, surface cutting depth, tectonic density, disaster point density, and engineering rock group indices are mainly selected for analysis. DEM refers to absolute elevation, which is the elevation from a point on the ground to a certain horizontal plane. With increasing elevation, temperature and precipitation decrease, leading to lower vegetation coverage and a more fragile eco-geological environment. In contrast, lower-altitude areas exhibit suitable temperature and precipitation levels, higher vegetation coverage, and a more stable eco-geological environment. Therefore, elevation is considered a negative indicator. The slope, surface roughness, and surface cutting depth are generated by the DEM. Steeper slopes exhibit poorer soil retention and less favorable vegetation growth conditions; thus, slope is classified as a negative indicator. Terrain roughness index (TRI) is an index used to express the degree of unevenness of the ground surface, which is a major indicator to describe the surface microtopography, and is defined as the ratio of surface area to projected area, higher surface roughness increases resistance to soil movement, whereas smoother slopes facilitate landsliding. Therefore, TRI is considered a positive indicator.

$$TRI=1/\cos \beta$$

(1)

The degree of surface cut is a quantitative indicator of the difference in elevation between a point on the ground and its neighborhood in a specific area, and is calculated as the difference between the average elevation and the lowest elevation of the neighborhood of that point, higher surface dissection (greater elevation differences) indicates more severe erosion, making it also a negative indicator. Structural density is generated by density analysis of tectonic layers, and the higher the tectonic density, the more fragile the eco-geological environment of the area; thus, structural density is classified as a negative indicator. Disaster point data uses historical disaster points; disaster site density refers to the number of various types of geologic hazard points distributed per unit area, which indicates the distribution of geologic hazards in the region, and the types of geologic hazards in the region are dominated by landslides, avalanches, and ground subsidence. Disaster site density is a negative indicator, as a higher number of geohazard points per unit area correlates with elevated risk and degraded eco-geological conditions. The engineered geological strata data from the National Geologic Library, which determines fundamental rock properties and serves as the material basis for geohazards, were classified and graded according to their characteristics.

(2)

Ecological environment

NDVI, derived from Landsat 8 data, serves as a positive indicator since higher values reflect denser vegetation coverage, which typically correlates with improved ecosystem stability and enhanced geological environment protection. Similarly, NPP—representing the net amount of organic carbon accumulated by vegetation after accounting for respiratory losses (i.e., gross photosynthesis minus autotrophic respiration)—functions as a positive indicator, as greater NPP values indicate more vigorous vegetation growth, stronger carbon sequestration capacity, and consequently, healthier ecological conditions. Considering the action distance of the water system, the buffer zones of 1000 m, 3000 m, 5000 m, and 7000 m were extracted and graded by buffering the water system surface. Annual precipitation is a major limiting factor for vegetation growth, affecting vegetation cover and soil and water conservation capacity, which is positively correlated with the carrying capacity of the ecological–geological environment.

(3)

Socioeconomy

Economic and social factors include road density, population density, GDP per capita, and night-light data. Road density is the ratio of the total mileage of the road network to the area of a given region. Population density refers to the distribution of population within 1 km² of each street, which can reflect the spatial variability of the regional economy. GDP per capita is a better indicator of a country’s or region’s affluence and economic development than total GDP. Night-light detection remote sensing technology is an advanced optical remote sensing technology that focuses on capturing information on weak light sources at night, which is difficult to capture by daytime remote sensing technology. Considering that artificial light sources are the main cause of nighttime brightness in cities, nighttime light detection remote sensing images can more directly reveal the differences in the spatial distribution of nighttime human activities. Higher values of these indicators typically imply stronger human impacts, exacerbating ecological and geological vulnerabilities. Therefore, these are all negative indicators.

3. Methods

The research meticulously gathered multi-source remote sensing data and integrated GIS and RS technologies to devise an eco-geological environment evaluation index system encompassing geological environment, ecological conditions, and socioeconomic factors. This comprehensive framework was then utilized to assess the eco-geological environment of Zigui County in the Three Gorges Reservoir Area, leveraging techniques such as the AHP, CV, and RAGA-PP methodology. The technical road map of this study (Figure 2) can be systematically organized into four pivotal stages: data acquisition, formulation of the evaluation index system, determination of weights, and ultimately, the evaluation and analysis of the eco-geological environment. This undertaking holds paramount significance in furnishing both technical underpinnings and theoretical foundations for infrastructure development, ecological safeguarding, and sustainable growth in Zigui County.

The evolution of indicator weighting methods has progressed from subjective qualitative approaches to more sophisticated quantitative techniques. In this study, both the AHP and CV methods were selected for their complementary strengths in addressing different aspects of eco-geological assessment. AHP provides a structured framework for incorporating expert knowledge, while CV offers objective data-driven weighting based on indicator variability. The integration of these methods through the minimum information entropy principle represents a balanced approach to weight determination, particularly suitable for the multi-dimensional nature of eco-geological systems. However, recognizing that different weighting approaches may be optimal for specific analytical needs, the study further employs the RAGA-PP model. This advanced technique provides an alternative projection-based solution that effectively handles high-dimensional data while minimizing subjective bias.

3.1. AHP-CV

AHP is a systematic approach designed to tackle complex multi-objective decision-making problems [16]. This method involves breaking down the overall objective into several criteria or sub-objectives and further decomposing these into multiple indicators across different levels. By employing qualitative index fuzzy quantization techniques, AHP facilitates the calculation of hierarchical single rankings (weights) as well as an overall ranking. The CV method serves as a statistically grounded evaluation tool specifically designed to assess the degree of variation in each indicator of the system. This method ensures objectivity in weight assignments by thoroughly analyzing the information associated with each indicator. When an indicator demonstrates significant fluctuations and can effectively differentiate between various evaluation objects, it indicates that the indicator carries considerable identification information. Conversely, when all the evaluation objects converge on a specific value, it is essential to allocate a higher weight to the indicator [17]. The fundamental steps in AHP are as follows:

(1)

Construct a judgment matrix

The construction of a judgment comparison matrix involves using a comparison scale to express the comparison results of the i factor relative to the j factor under certain comparison criteria when the evaluation target is affected by n factors so that the judgment matrix can be obtained by one comparison.

(2)

Calculate the relative weight coefficient

The single ranking weight of each index hierarchy can be obtained by solving the eigenvector of the eigenvalue.

$$\mathrm{A}\mathrm{W}={\lambda }_{\mathrm{m}\mathrm{a}\mathrm{x}}\mathrm{W}$$

(2)

where A is the judgment matrix; ${\lambda }_{max}$ determines the maximum eigenvalue of matrix A; W determines the eigenvector corresponding to matrix A; and $Wi$ is each element that makes up the feature vector that is the weight value of the required hierarchical single sort.

The eigenvector W and the maximum eigenvalue of the judgment matrix are usually evaluated by the root method, power method, and weighted summation method. In this study, the weighted summation method is used for calculation.

(3)

Consistency test

When the judgment matrix A satisfies positive reciprocity and transitivity, it is called A consistency matrix. However, the actual evaluation object is often quite complex, and the judgment matrix A may not have consistency. This is because there will be value judgments of different scales in the pairwise comparison of index factors, so it is necessary to carry out a consistency test on the judgment matrix A. The consistency index CI of judgment matrix A is defined as CI = (−n)/(n − 1). When Cl = 0, it indicates that the judgment matrix is consistent; the larger the value of Cl, the more serious the degree of inconsistency of the matrix; conversely, the smaller the value of Cl, the better the degree of consistency of the judgment matrix.

The CV is an important indicator to describe the trend of deviation, reflecting the degree of numerical variability, which was used to calculate the objective weight in this paper. The calculation steps are as follows:

(4)

Original data collection

Assume that there are m indicators and n samples to be evaluated in a set of data, i.e., there exists an n × m matrix such that its value is X, which then denotes the data in the ith row and jth column.

$$\mathbf{X}=\left(\begin{array}{ccc}{x}_{11}& \dots & x_{1m}\\ ⋮& \ddots & ⋮\\ x_{n1}& \dots & x_{nm}\end{array}\right)$$

(3)

(5)

Indicator Data Normalization

The main goal of indicator normalization is to convert all indicators into positive indicators.

For positive indicators, keep their original data unchanged.

$$x_{ij}^{\prime }=x_{ij}$$

(4)

For negative indicators,

$$x_{ij}^{\prime }={\displaystyle \frac{1}{k+\max \left|x_j\right|+x_{ij}}}$$

(5)

where k is a specified arbitrary coefficient, the value of which can be 0.1, 0.2, etc., which indicates the maximum value of the absolute value of the jth column of data (indicator).

(6)

Data standardization (elimination of scales)

Given that the units of data for different indicators vary, direct calculations can lead to distortion. The core objective of data standardization is to unify the scale and eliminate the influence of unit differences so that all indicator data can be calculated using uniform standards and methods. Let the standardized data matrix be R.

$$r_{ij}={\displaystyle \frac{x_{ij}^{\prime }}{\sqrt{\sum _{i=1}^n{x}_{ij}^{\prime 2}}}}$$

(6)

(7)

Calculate the CV

Calculate the mean value of each indicator:

$$A_j={\displaystyle \frac{1}{n}}\sum _{i=1}^n{r}_{ij}$$

(7)

Calculate the standard deviation (mean square error) for each indicator:

$$S_j=\sqrt{{\displaystyle \frac{1}{n}}\sum _{i=1}^n{(r_{ij}-A_j)}^2}$$

(8)

Since the standard deviation is a value that can measure the degree of dispersion of the data, reveal the fluctuation of the value of a certain indicator, and reflect the discriminatory ability of the indicator, the standard deviation can be used to define and quantify the weight of each indicator. Calculate the CV for each indicator:

$$V_j={\displaystyle \frac{S_j}{A_j}}$$

(9)

(8)

Combining the advantages of the subjective assignment AHP method and the objective assignment CV method, the combined weights are obtained through two models for determining the weights in order to make the results more reasonable and accurate:

$$W=\alpha w_i^a+(1-\alpha )w_i^b$$

(10)

where W is the combination weight, $\alpha$ is the coefficient sought, $w_i^a$ is the weight determined by AHP, and $w_i^b$ is the weight determined by CV. To minimize the sum of squares of the deviations of W from both the $w_i^a$ and $w_i^b$,

$$minW=\sum _{i=1}^n[{\left(w_i-w_i^a\right)}^2+{\left(w_i-w_i^b\right)}^2]$$

(11)

Combining the above two equations gives $\alpha =0.5$.

$$W=0.5w_i^a+0.5w_i^b$$

(12)

3.2. RAGA-PP

The projection tracing model compresses multi-dimensional evaluation indices into one-dimensional projection data, while the optimized genetic algorithm controls the projection index function to find the optimal projection direction. By examining the differences in the best projection directions, the influence degree of each index can be accurately judged, allowing for an objective evaluation of the eco-geological environment based on the corresponding projection values. This approach minimizes subjective bias from manual assignments and effectively addresses potential information redundancy among indicator variables. Additionally, it does not require prior assumptions about data distribution patterns, thus ensuring the objectivity and accuracy of the evaluation results [18].

Traditional genetic algorithms often suffer from slow convergence and weak global optimization capabilities. This project proposes combining quantum genetic algorithms with a projection-seeking model to address high-dimensional, complex nonlinear function optimization problems. This project proposes to use this model to construct the projection objective function to calculate the projection tracing value. The main steps of the model construction are as follows (Figure 3):

(1)

Dimensionless processing

In order to eliminate the interference of different measures on the evaluation results, it is necessary to standardize each indicator and perform dimensionless processing.

For positive indicators,

$$x^{\prime }\left(i,j\right)={\displaystyle \frac{x\left(i,j\right)-x_{min}\left(j\right)}{x_{max}\left(j\right)-x_{min}\left(j\right)}}$$

(13)

For negative indicators,

$$x^{\prime }\left(i,j\right)={\displaystyle \frac{x_{max}\left(j\right)-x\left(i,j\right)}{x_{max}\left(j\right)-x_{min}\left(j\right)}}$$

(14)

(2)

Construct the projection objective function

Let $a=\{a\left(1\right),a\left(2\right),\dots ,a(p\left)\right\}$ be the ecological safety evaluation projection direction vector; the one-dimensional projection value of the sample in this direction is as follows:

$$z\left(i\right)=\sum _{j=1}^{p}a\left(j\right)\cdot x\left(i,j\right). i=\mathrm{1,2},\dots ,n$$

(15)

In the above equation, a denotes the unit length vector, and the projection indicator function is as follows:

$$Q\left(a\right)=S_z\times D_z$$

(16)

In the above equation, $Q\left(a\right)$ denotes the projection index function; $S_z,D_z$ denote the standard deviation of the safety projection value and the local density (Equations (16) and (17)), which are calculated, respectively, by the following formulas:

$$S_z=\sqrt{{\displaystyle \frac{\sum _{i=1}^n{(z\left(i\right)-E(z\left)\right)}^2}{n-1}}}$$

(17)

$$D_z=\sum _{i=1}^n\sum _{j=1}^n[R-r\left(i,j\right)\times u(R-r\left(i,j\right)\left)\right]$$

(18)

(3)

Optimization of projection index function

After obtaining the ecological safety evaluation index value, the objective function $Q\left(a\right)$ will change with the change in projection direction; according to the calculation results of maximizing the projection objective function, the optimization model is used to measure the best projection direction of a:

$$\left\{\begin{array}{c}maxQ\left(a\right)=S_z\times D_z\\ s.t.{\displaystyle \sum _{j=1}^p}{a}^2\left(j\right)=1\end{array}\right.$$

(19)

The best projection direction is brought into the equation to obtain the projection seeking value, and the absolute value of the best projection direction vector element value is the weight of the evaluation index.

The eco-geological assessment framework incorporates 16 standardized indicators analyzed through an optimized RAGA-PP model implementation. The computational parameters were carefully selected based on preliminary sensitivity analysis: initial population size (N = 400), crossover probability (Pc = 0.8), and mutation probability (Pm = 0.2), with mutation direction randomization (n = 10) and 7 acceleration iterations to ensure convergence efficiency.

The model generated optimal projection direction vectors for longitudinal evaluation, yielding the following weight distribution across indicators (presented in original order): [0.3780, 0.2554, 0.0115, 0.2938, 0.3538, 0.1224, 0.3719, 0.0373, 0.1385, 0.0283, 0.0206, 0.3290, 0.0413, 0.2336, 0.3085, 0.3816]. These vectors were subsequently normalized to produce the final composite evaluation scores, enabling a comprehensive spatial assessment of Zigui County’s eco-geological conditions.

4. Results

4.1. Eco-Geological Environment Assessment Factors

The data preprocessing included coordinate transformation, spatial cropping, land cover classification, and buffer analysis. These processed data layers were then integrated to generate the final factor distribution map (Figure 4). All the preprocessing steps were performed using standard GIS tools with consistent parameters across the study area.

Given the substantial variations in scale and value range among the indicators, it is imperative to grade these indicators to ensure data consistency and operational feasibility. To achieve this, the natural break point method was employed, which facilitated the categorization of the data into five distinct classes, each assigned a corresponding score ranging from 1 to 5 (

Table 3. Grading of evaluation indicators.

Indicator	Grading of Evaluation Indicators
	Bad	Poor	Moderate	Good	Better
DEM (m)	≥1334	1022~1334	730~1022	428~730	≤428
Slope (°)	≥42.84	31.67~42.84	22.35~31.67	13.35~22.35	≤13.35
Surface roughness	≤1.10	1.10~1.25	1.25~1.49	1.49~1.90	≥1.90
Surface cutting depth (m)	≥323.31	235.37~323.31	168.12~235.37	103.46~168.12	≤103.46
Structural density (km/km2)	≥0.35	0.26~0.35	0.19~0.26	0.12~0.19	≤0.12
Disaster site density (per/km2)	≥0.36	0.18~0.36	0.09~0.18	0~0.09	0
Engineering geological strata	flexible bulk structure	weak and thinly bedded mud and shale formations	soft—harder thinly bedded moderately thickly laminated sand and mudstone formations	soft and hard medium—thick laminated sandstone and mudstone formations	harder-harder moderately- thickly bedded sandstone plus sandstone groups
NDVI	≤−0.02	−0.02~0.30	0.30~0.51	0.51~0.66	≥0.66
NPP	1569~7975	7975~12,305	12,305~19,457	19,457~27,411	27,411~32,767
Drainage distance (m)	≤1000	1000~3000	3000~5000	5000~7000	≥7000
Annual precipitation (mm)	≤61.60	61.60~62.23	62.23~62.94	62.94~64.38	≥64.38
Land-use	buildings	plow land	waters, wetlands	woodland, shrubs, grassland	deserts
Road density (km/km2)	≥1.08	0.76~1.08	0.51~0.76	0.31~0.51	≤0.31
Population density (per/km2)	≥2459	1176~2459	469~1176	135~469	≤135
GDP per capita (CNY)	≥437.75	245.62~437.75	94.63~245.62	17.77~94.63	≤17.77
Night-light data	≥31	21~31	12~21	0~12	0

).

4.2. Eco-Geological Environment Evaluation Results

The eco-geological assessment of Zigui County, integrating geological, ecological, and socioeconomic subsystems, reveals the proportional distribution of each evaluation level, as summarized in

Table 4. The percentage of area in the evaluation level sub-area of four methods.

Methods		Vulnerability Classes
		Better	Good	Moderate	Poor	Bad
AHP	Area/km2	250.31	539.43	451.04	434.86	392.02
	Proportion/%	9.72	25.55	24.76	22.67	17.31
CV	Area/km2	182.279	555.96	508.23	449.62	371.56
	Proportion/%	8.45	24.45	25.13	23.75	18.21
AHP-CV	Area/km2	193.11	540.64	486.31	462.73	384.87
	Proportion/%	12.72	29.29	26.19	19.82	11.97
RAGA-PP	Area/km2	317.14	629.49	479.16	386.88	255.00
	Proportion/%	15.34	30.44	23.17	18.71	12.33

, derived from diverse assessment methods.

The evaluation results indicate that the overall geological environment performance ranges from medium to high. Figure 5 illustrates the importance values of different indicators. The area percentages of evaluation grades calculated are generally consistent, with the lowest proportion in the “bad” grade and the highest in the “moderate” grade. As the evaluation level changes, the trends in percentage variations across the methods are consistent. In the “moderate” grade, all four methods account for over 23% of the total, with the AHP-CV method showing the highest proportion at 26.19%.

The eco-geological environment evaluation grades in the study area, as depicted in Figure 6, indicate that regions classified as “bad” and “poor” are predominantly located in the northwest and southeast corners. According to AHP, elevation is the most significant influencing factor, followed closely by structural density. Zigui County exhibits a mountainous topography, with higher elevations in the southwest and lower elevations in the northeast. The eastern region is characterized by the Huangling anticline slope, while the western region comprises the Zigui syncline, part of the Three Gorges landscape. In contrast, the evaluation grades derived from the CV method identify “bad” and “poor” conditions primarily in the northwest and south, with the most critical factors being the engineered rock group, annual average precipitation, and elevation. Similarly, the evaluation grades obtained from the RAGA-PP method highlight “bad” and “poor” areas concentrated in the northwest, associated with higher elevations, significant surface cutting depth, and increased structural density, which contribute to conditions prone to geological hazards.

4.3. Eco-Geological Environment of Township Evaluation Results

Based on the eco-geological environment evaluation outcomes derived from the RAGA-PP model, the areal distribution of evaluation grades across 12 townships in Zigui County in 2020 has been tallied (

Table 5. Statistics on the results of the eco-geological environment evaluation of townships.

Townships	Better	Good	Moderate	Poor	Bad
	Area/km2	Area/km2	Area/km2	Area/km2	Area/km2
Yanglinqiao	14.61	83.05	79.92	37.70	6.80
Moping	64.56	48.93	6.75	0.20	0.00
Meijiahe	13.41	22.58	19.25	18.55	3.93
Xietan	0.35	25.35	40.72	35.73	18.44
Quyuan	30.94	83.76	55.99	26.36	8.92
Guizhou	0.00	5.59	19.30	27.26	35.23
Maoping	19.25	42.53	29.33	32.45	44.45
Lianghekou	79.83	78.77	29.13	12.40	2.12
Jiuwanxi	39.41	95.90	51.76	26.86	9.32
Shuitianba	3.02	41.63	50.85	50.55	43.04
Shazhenxi	0.30	22.27	53.32	63.65	43.69
Guojiaba	51.60	78.41	49.29	55.38	39.76

). The statistical analysis reveals that Moping Township is predominantly characterized by “better”-rated areas, with virtually no presence of “poor” or “bad” zones. Areas rated as “moderate” and above encompass approximately 100% of Moping’s territory. Conversely, Guizhou Township is largely dominated by “bad” and “poor” areas, with no “better”-rated areas identified, and “moderate” and lower-rated areas constituting 93.6% of its total area. According to Figure 7, Yanglinqiao, Meijiahe, Xietan, and Jiuwanxi exhibit a relatively large proportion of “moderate”-rated areas, suggesting a comparatively smoother eco-geological environment. Notably, Quyuan and Lianghekou have a substantial percentage of “better”-rated areas, indicating a superior ecological environment in these townships. Conversely, Shuitianba and Shazhenxi exhibit a poorer ecological status, necessitating the implementation of specific management measures to prevent further deterioration.

These evaluation results serve as a valuable tool for the local government to devise targeted prevention and control measures for eco-geological environmental protection. Furthermore, they provide a solid foundation for the development planning and sustainable growth of Zigui County, as well as the implementation of corresponding ecological restoration policies. Consequently, these measures aim to enhance the overall ecological environment of the study area.

4.4. Spatiotemporal Evolution Results

Based on the above results, RAGA-PP was selected to calculate the weights of eco-geological environment evaluation indicators of Zigui County in 2000, 2005, 2010, and 2015, and the results obtained are shown in

Table 6. Temporal variation in percentage of area by evaluation level sub-area (2000-2020).

Methods		Vulnerability Classes
		Better	Good	Moderate	Poor	Bad
2000	Area/km2	199.01	523.40	507.13	464.29	354.53
	Proportion/%	9.72	25.55	24.76	22.67	17.31
2005	Area/km2	174.17	503.75	517.71	489.24	375.14
	Proportion/%	8.45	24.45	25.13	23.75	18.21
2010	Area/km2	260.99	601.01	537.36	406.74	245.68
	Proportion/%	12.72	29.29	26.19	19.82	11.97
2015	Area/km2	329.53	671.21	499.21	348.68	205.76
	Proportion/%	16.04	32.67	24.30	16.97	10.02
2020	Area/km2	317.14	629.49	479.16	386.88	255.00
	Proportion/%	15.34	30.44	23.17	18.71	12.33

.

Elevation indicators consistently maintained a significant weight share across the five years analyzed (Figure 8). In 2000, slope was the most influential indicator, followed by surface cutting depth and elevation. By 2005, slope remained the dominant factor, with land-use type and elevation as secondary influences. In 2010, the primary indicators shifted to elevation, surface cutting depth, and engineered rock group. By 2015, elevation emerged as the leading factor, followed by NPP and engineered rock group, with land-use type also contributing a weight exceeding 0.1. In 2020, elevation, structural density, engineered rock group, and night-light indicators became the primary determinants. Despite fluctuations in weight values, the geo-environmental subsystem predominantly governed the overall influence throughout the study period.

As shown in Table 4, the eco-geological environment of Zigui County exhibited a gradual improvement from 2000 to 2020 (Figure 9). The area classified as “bad” decreased from 354.53 km² to 255 km², with its proportion declining from 17.31% to 12.33%. The area rated as “poor” displayed an unstable downward trend, peaking at 489.24 km² in 2005, accounting for 23.75% of the total area. In contrast, the area classified as “moderate” and above showed a consistent increase, with its total proportion rising from 60.03% to 68.95%, and the total area expanding from 1229.54 km² to 1425.79 km².

5. Discussion

5.1. Spatial Auto-Correlation Analyses

Spatial auto-correlation, as a basic spatial statistical method in geography, is able to reflect the potential interdependence between spatial data and contains global and local spatial auto-correlation, which are used to characterize the spatial distributions of the whole and the localized study area, respectively. In this study, Moran’s I was used to measure the spatial auto-correlation of the ecological and geological environment in Zigui County. Moran’s I assesses the correlation between attribute values at each location and those of neighboring locations, employing a spatial contiguity matrix [19]. It involves the covariance of attribute values and the spatial weight matrix, providing a standardized value between −1 and 1. Moreover, a negative value indicates a spatial dispersion or dissimilar values close to each other, and a value close to zero suggests a random spatial pattern. The statistical significance of Moran’s I index is computed with the Z-score method, assuming a random distribution with a mean equal to zero and a variance of one. A positive Z-score indicates that neighboring features have similar values, whereas a negative Z-score implies that the feature is surrounded by dissimilar values. The results of Moran’s index for different methods are shown in

Table 7. Moran’s I analysis result.

	Moran I	z	p
AHP	0.776	522.71	0.00
CV	0.841	566.46	0.00
AHP-CV	0.792	533.80	0.00
RAGA-PP	0.815	548.78	0.00

.

Based on the computed Moran’s I index, it is evident that the values obtained through various methodologies lie within the range of 0 to 1, indicating a positive spatial correlation among geographic entities. This correlation intensifies as the spatial distribution of these entities becomes more clustered. With a z-score exceeding 1.65, the presence of spatial auto-correlation is confirmed. Furthermore, a p-value below 0.05 signifies the significance of the Moran’s I index, thereby indicating that the experimental results are highly reliable and consistent with the underlying spatial distribution patterns (Figure 10).

The spatial auto-correlation analysis demonstrates that the RAGA-PP model achieves superior performance in capturing spatial dependence patterns (Moran’s I = 0.815) compared to conventional methods (AHP-CV: 0.792). This optimal balance reflects RAGA-PP’s unique capability to (1) preserve meaningful spatial clusters while avoiding over-smoothing artifacts common in expert-based approaches, and (2) maintain ecological gradient continuity better than purely statistical methods. The model’s enhanced spatial auto-correlation performance is particularly evident in transitional eco-geological zones, where it reduces boundary discontinuities by 23% compared to AHP-CV.

5.2. Hot Spot Analysis

Hot spot analysis (Getis-Ord Gi*) is a technique used in spatial data analysis that determines the distribution of hot and cold spots by calculating a local index (Gi value) for each element (e.g., point, line, or area on a map). Hot spots represent areas where high values are clustered, while cold spots represent areas where low values are clustered.

The results are shown in Figure 11, with red hues typically denoting hot spots and blue hues signifying cold spots. Specifically, the darkest shades of red and blue represent the most intense clustering of high and low values, respectively.

Analysis of hot spot maps (2000–2020) reveals stable spatial aggregation patterns (Figure 12). Northwestern areas consistently emerged as hot spots, though confidence levels declined from 99% to 95% between 2000 and 2005. This shift likely reflects ecological impacts from the 2003 Three Gorges Reservoir impoundment, which altered local hydrology, biodiversity, and geomorphology. The stabilization post-2005 suggests ecosystem adaptation to new hydrological regimes.

5.3. Eco-Geological Environment Protection and Development Suggestions

The evaluation results reveal significant spatial differentiation in Zigui County’s eco-geological conditions, with Moping, Quyuan, and Lianghekou townships exhibiting superior environmental quality due to favorable topography and vegetation cover, while Guizhou, Shazhenxi, and Shuitianba face serious ecological challenges [20]. These findings provide critical spatial references for implementing tiered conservation strategies. For high-performing areas, establishing continuous remote sensing monitoring systems utilizing Sentinel-2 and Landsat data to track vegetation dynamics and surface changes, complemented by strict controls on construction activities, is recommended [21]. Transitional zones require integrated slope stabilization measures combining InSAR deformation monitoring with ecological engineering interventions. The most degraded areas demand immediate restoration efforts focusing on landslide prevention and soil conservation, guided by high-resolution UAV surveys and LiDAR terrain analysis.

The local government could incorporate these spatial evaluation results into land-use planning decisions, prioritizing ecological protection in vulnerable areas while permitting controlled development in stable zones. Future management should emphasize cross-sectoral coordination between geological disaster prevention and ecological restoration programs, supported by regular remote sensing-based environmental assessments to monitor intervention effectiveness.

6. Conclusions

This study successfully establishes and applies an evaluation system for the eco-geological environment of Zigui County, utilizing AHP-CV and RAGA-PP models to comprehensively assess its conditions, providing valuable results. The integration of these two methods offers a robust framework for determining indicator weights and evaluating eco-geological conditions. Notably, the RAGA-PP model demonstrates superior alignment with influencing factors and actual conditions, making it the preferred method for calculating indicator weights and deriving evaluation results for the years 2000, 2005, 2010, and 2015. Over the past two decades, the area classified as “better” in the eco-geological environment has shown a consistent expansion, while the proportion of “good” areas has exhibited an upward trend, albeit with some fluctuations. These findings highlight the significant recovery and improvement in the eco-geological environment of Zigui County, underscoring the effectiveness of ecological management efforts in the region. The innovative application of RAGA-PP in this context provides a reliable and accurate approach for eco-geological assessments, contributing to the advancement of methodologies in environmental evaluation and sustainable management practices.