Monitoring Vegetation Dynamics and Driving Forces in the Baijiu Golden Triangle Using Multi-Decadal Landsat NDVI and Geodetector Modeling

Abstract

The China Baijiu Golden Triangle (BGT) serves as the core production hub of China’s Baijiu industry, where the ecological environment plays a pivotal role in ensuring the industry’s sustainable development. However, urbanization, industrial expansion, and climate change pose potential threats to the region’s vegetation dynamics. Utilizing Landsat remote sensing data from 2002 to 2022, this study integrates Theil–Sen trend analysis, the Mann–Kendall (MK) test, coefficient of variation (CV) analysis, and the Geodetector model (GD model) to investigate the spatiotemporal evolution of the Normalized Difference Vegetation Index (NDVI) and its underlying driving mechanisms within the BGT. The findings reveal an overall upward trend in vegetation NDVI, with the annual mean NDVI increasing from 0.45 to 0.67, corresponding to a growth rate of 0.49%. Spatially, areas of high vegetation cover are predominantly located in mountainous forest zones with favorable ecological conditions, whereas regions of low vegetation cover are concentrated in zones of urban expansion. Precipitation and topographic factors (elevation and slope) emerge as the primary natural drivers of vegetation change, while land use change and the night-time light index stand out as the most influential human-induced factors. Further analysis uncovers a nonlinear interactive enhancement effect between natural and anthropogenic factors, with the interaction between the night-time light index and precipitation being particularly pronounced. This suggests that urbanization not only directly impacts vegetation but may also exert indirect effects on the ecosystem by altering regional hydrological and climatic processes. The results indicate that ecological protection policies in the BGT have yielded some success; however, vegetation fragmentation and ecological pressures stemming from urban expansion remain significant challenges. Moving forward, optimizing land use policies and promoting eco-friendly development models will be essential to achieving ecosystem stability and sustaining industrial growth.

1. Introduction

Chinese Baijiu, as a quintessential representative of distilled spirits worldwide, is deeply integrated with its unique brewing process and regional ecological conditions, forming a quality spirit characterized by the fourfold synergy of “water, soil, air, and life” [1,2]. From a global perspective, the decisive role of regional ecology in terms of alcohol quality has been proven in several world-renowned alcohol regions. For instance, the wines of Bordeaux, France, are profoundly influenced by the local climate, soil, and microbial environment [3,4]; Scotch whisky from the Scottish Highlands relies on pristine water sources and unique climatic conditions [5,6]; and Niigata sake in Japan is renowned for its high-quality rice and distinctive water quality [7,8]. These international experiences demonstrate that a superior ecological environment not only shapes the unique flavor of alcoholic beverages but also determines their market competitiveness and potential for sustainable development. As the transition to China’s new “dual circulation” development pattern continues to be driven by domestic and international economic cycles that create growing consumer demand for liquor consumption, the high-quality development of the Baijiu industry has gradually become an important part of the national strategy. However, the high quality of the Baijiu industry is largely dependent on the stability of the ecological environment in the Baijiu production area, including the diversity of the microbial community, the purity of the water, and the quality of the raw materials used in the production of Baijiu, which together determine the flavor and quality of the Baijiu [9,10]. If the sustainable management of the ecological environment is neglected, the “regional scarcity” advantage that Baijiu production relies on may face irreversible degradation risks [11]. Therefore, conducting a systematic assessment of the spatiotemporal evolution of ecological environmental quality in Baijiu-producing regions and thoroughly analyzing its driving mechanisms is not only a fundamental prerequisite for ensuring Baijiu quality but also an essential requirement for enhancing the industry’s long-term competitiveness and achieving sustainable development.

In recent years, scholars have conducted valuable explorations into the ecological environment of Baijiu-producing regions at a micro-scale, utilizing approaches such as soil microbiology [12,13,14,15], water chemistry analysis [16,17,18], and microclimate monitoring [19,20,21]. These studies have revealed the critical role of the brewing microenvironment in the formation of Baijiu’s flavor compounds. However, the existing research exhibits two major limitations. First, few studies have systematically evaluated the overall ecological environment quality of Baijiu-producing regions from a macro perspective, particularly lacking long-term spatiotemporal analyses. This research gap may result in insufficient scientific support for environmental management and policy-making in the Baijiu industry. Second, there is a lack of systematic investigations into the combined effects of socioeconomic factors and natural conditions on the ecological environment. In particular, the complex impact of socioeconomic activities on ecological changes has not been fully explored, leading to an incomplete understanding and weak scientific foundation for macro-level environmental governance.

In addition, recent studies on the drivers of ecological quality in Baijiu production areas have mainly focused on the effects of individual factors while ignoring the combined effects of multiple interacting factors, a limitation that constrains us from comprehensively revealing the ecological evolutionary processes in these areas. The Normalized Difference Vegetation Index (NDVI), as a key indicator for evaluating ecological quality, has been widely used in macro-environmental monitoring and evaluation [22]. Many scholars have used the NDVI to conduct in-depth macro-scale analyses of ecological changes across various regions [23,24,25,26,27], providing scientific evidence and valuable insights for environmental governance.

Building on this foundation, previous studies have employed the Geodetector model (GD model) to explore the factors influencing ecological quality [28,29,30,31]. The GD model effectively compensates for deficiencies in traditional multi-factor analysis, particularly in elucidating the combined effects of natural conditions and socioeconomic activities on the ecological environment. This approach not only identifies the relative contributions of different factors to ecological changes but also reveals the interactions between them, thereby offering a more scientific and precise basis for macro-level governance decisions [32,33]. These studies have established a solid theoretical foundation for employing the NDVI and the GD model to investigate the spatiotemporal variations and driving factors of ecological quality in Baijiu-producing regions from a macro perspective [34,35]. Therefore, utilizing the NDVI and the GD model to assess the ecological quality of Baijiu-producing regions and its driving factors from a macro and dynamic perspective holds significant theoretical and practical value. By utilizing the NDVI for macro-monitoring of vegetation cover changes in the study area and combining it with the GD model to quantify the relative contributions and interactions of the influencing factors, this method provides a more scientific and refined basis for the ecological protection and sustainable development of the Baijiu industry. Furthermore, the GD model demonstrates significant advantages in identifying driving factors as it quantitatively assesses the impacts of both natural and socioeconomic factors on the ecological environment [36]. This capability effectively addresses the limitations of traditional multi-factor analysis and unveils the interactions among driving factors, thereby offering robust scientific evidence for understanding the mechanisms underlying ecological changes in Baijiu-producing regions.

The Baijiu Golden Triangle (BGT) region is one of the most significant Baijiu-producing areas in China. Renowned for its exceptional natural conditions and long-standing brewing history, this region holds a central position in China’s Baijiu production and market share, making it a crucial area for advancing the high-quality development of the Baijiu industry. However, in recent years, with the increasing frequency of industrial development, urban expansion, agricultural production and other human activities, the quality of the ecological environment in the BGT region has gradually declined. This is evidenced by the significant decline in vegetation cover, serious soil erosion, and ecosystem degradation, and the declining quality of the ecological environment poses a great threat to the high-quality and sustainable development of the Baijiu industry [37,38].

Therefore, this study takes BGT as the study area, and dynamically reveals the spatial and temporal characteristics and drivers of vegetation cover changes in the study area from a macro perspective based on the NDVI. Specifically, this study investigates the effects of natural conditions and economic factors on NDVI changes. The results of this study not only help to understand the ecological changes in the BGT, but also provide scientific support for the management of the ecological environment and the sustainable development of the Baijiu industry in the study area. In addition, this study provides a theoretical basis for the formulation of ecological protection policies, the optimization of Baijiu industry layout, and the realization of coordinated ecological and economic development in the study area, which is of great practical significance.

2. Materials and Methods

2.1. Study Area

The BGT is located in Southwest China and primarily consists of three major cities: Yibin, Luzhou, and Zunyi. The geographic coordinates of this region range from 105°10′ E to 108°00′ E longitude and 27°00′ N to 29°30′ N latitude, covering a total area of approximately 68,000 km². The region is characterized by complex topography, with higher elevations in the northwest gradually decreasing towards the southeast. The landforms within BGT include mountains, hills, and plains (Figure 1). The BGT region falls within the subtropical monsoon climate zone, featuring a mild and humid climate. The annual average temperature ranges between 16 °C and 19 °C, while annual precipitation varies between 800 mm and 1200 mm. The region boasts diverse vegetation types, including evergreen broadleaf forests, deciduous broadleaf forests, and coniferous forests, with natural vegetation exhibiting a distinct vertical zonation pattern. The BGT region has a well-developed hydrological network, with the Yangtze River and its tributaries flowing across the area, providing abundant water resources. Yibin and Luzhou are situated along the Yangtze River, while Zunyi is covered by the Chishui River Basin. These rivers play a crucial role in supporting local agriculture and the Baijiu brewing industry. Moreover, the region is characterized by fertile soils, primarily red soil, yellow soil, and purple soil, which are highly favorable for agriculture, particularly the cultivation of high-quality raw materials essential for Baijiu production.

2.2. Overview of the BGT

As both a significant Baijiu production hub and a representative ecosystem, the BGT holds substantial research value. The cities of Yibin, Luzhou, and Zunyi constitute China’s three core Baijiu-producing regions, specializing in strong-aroma and sauce-aroma Baijiu, respectively. In 2022, the Baijiu industry in Yibin achieved a total revenue of CNY 180.84 billion, marking a 10.6% year-on-year growth and accounting for 52.5% of Sichuan Province’s total Baijiu revenue. Among them, Wuliangye Group alone contributed CNY 73.97 billion, playing a pivotal role in Sichuan’s Baijiu economy. Yibin was designated as the first pilot city for a world-class Baijiu industrial cluster, with its strong-aroma Baijiu production capacity accounting for 70% of the national total. In 2022, Sichuan Province produced 3.489 million kiloliters of Baijiu, representing 51.9% of the national total, with an industry revenue of CNY 344.72 billion (45.5% of the national total) and total profits reaching CNY 75.32 billion (56.3% of the national total). Luzhou Laojiao, a leading enterprise in Luzhou, generated CNY 25.12 billion in revenue, accounting for 68% of Luzhou’s total Baijiu revenue. Luzhou, with Luzhou Laojiao as its dominant player, holds a significant position within Sichuan’s Baijiu industry, collectively shaping the national market structure for strong-aroma Baijiu alongside Yibin. Meanwhile, Zunyi, with Moutai as its core, aims to expand its Baijiu production capacity to 600,000 kiloliters by 2025 (including Moutai’s expansion projects). The city also targets an increase in sauce-aroma Baijiu’s national market share to 55%, further solidifying its dominance in this category. The rapid development of Baijiu industries across these three regions not only reflects a high-end transformation of local economies but also highlights China’s Baijiu industry’s unique competitive edge in the global market.

2.3. Data Sources

2.3.1. NDVI Data

This study utilizes Landsat series imagery from 2002 to 2022, including Landsat 5 TM, Landsat 7 ETM+, and Landsat 8 OLI/TIRS, obtained from the Google Earth Engine (GEE) platform. The datasets used include “LANDSAT/LT05/C02/T1_L2” (Landsat 5), “LANDSAT/LE07/C02/T1_L2” (Landsat 7), and “LANDSAT/LC08/C02/T1_L2” (Landsat 8), with a spatial resolution of 30 m and a temporal resolution of 16 days. The NDVI calculation is based on the standard red and near-infrared bands of Landsat 5, 7, and 8 imageries. For Landsat 5 and 7, bands SR_B3 and SR_B4 are used, while for Landsat 8, bands SR_B4 and SR_B5 are employed. These bands are processed through band renaming and a standardized difference function on the GEE platform to generate consistent NDVI layers, ensuring methodological uniformity across different time periods [39]. The raw data underwent radiometric correction, atmospheric correction (based on the MODTRAN atmospheric radiative transfer model), and terrain correction (assisted by SRTM DEM) through GEE’s built-in algorithms. To remove cloud and cloud shadow interference, the CFMask algorithm was applied using the cloud mask thresholding method. The Maximum Value Composite (MVC) method was further employed to generate an annual NDVI dataset, mitigating short-term atmospheric fluctuations [40,41]. Additionally, to address the striping artifacts in Landsat 7 SLC-off data, the adaptive local regression interpolation method was used for data gap restoration [42,43].

2.3.2. Other Data

The data used in this study to investigate the factors influencing NDVI changes are mainly categorized into two groups: natural conditions and human activities. Natural conditions encompass meteorological data, soil data, and topographic data, while human activities include land use, population density, and night-time lighting data. For a detailed description of these datasets, please refer to

Table 1. Data sources and descriptions.

Data Type	Driving Factors	Year	Data Source	Data Accuracy
Climate	mean annual temperature	2002~2022	National Earth System Science Data Center (https://www.geodata.cn (accessed on 5 March 2024))	1 km
	annual precipitation	2002~2022
Terrain	slope	2009	SRTM Dataset (https://openmaptiles.org/languages/zh/ (accessed on 25 May 2024))	500 m
	aspect	2009
	elevation	2009
Soil	soil data	2011	HWSD v1.2: World Soil Database HWSD v1.2 (http://webarchive.iiasa.ac.at/Research/LUC/External-World-soil-database (accessed on 20 April 2024))	500 m
Land use	land use	2002~2022	CLCD (China Land Cover Dataset) (https://zenodo.org/records/8176941 (accessed on 20 March 2024))	500 m
Human activities	population density	2002~2022	LandScan Global Population Dataset (https://landscan.ornl.gov/ (accessed on 13 May 2024))	500 m
	night-time lights	2002~2022	Night-time Lights Dataset (http://geodata.nnu.edu.cn (accessed on 20 May 2024))	500 m

.

2.3.3. Data-Processing Platform

All spatial data underwent format conversion, projection standardization to the WGS84/UTM Zone 48N coordinate system, and spatial matching. These processes were performed using the Google Earth Engine platform and QGIS version 3.28, respectively. QGIS, an open-source software compliant with the GPL license, offers robust data processing and map generation capabilities, making it well suited for academic research applications.

2.3.4. Technical Roadmap

The technical workflow of this study is depicted in Figure 2 and comprises the following steps: 1. Data acquisition and preprocessing: Satellite imagery from Landsat 5, 7, and 8, spanning 2002 to 2022, was acquired and subjected to essential preprocessing steps, including atmospheric correction and georeferencing. The maximum value composite method, alongside additional preprocessing techniques, was applied to derive a clear NDVI from multi-temporal images. This approach minimized cloud cover interference and enhanced data quality. 2. Spatiotemporal change analysis: NDVI data were analyzed spatiotemporally to evaluate trends in vegetation cover changes within the BGT region. The Theil–Sen median trend analysis was employed to quantify NDVI change trends, while the Mann–Kendall (MK) test assessed the statistical significance of these trends. Additionally, the coefficient of variation (CV) was used to characterize the spatiotemporal variability in the NDVI, providing insights into the stability and heterogeneity of vegetation cover. 3. Selection of driving factors: drawing on prior research, nine representative variables were selected as driving factors to investigate their influence on vegetation changes. 4. Analysis of driving factors: The GD model was utilized to dissect the roles of these driving factors. Specifically, the factor detector quantified the individual contribution of each factor to vegetation changes. The interaction detector evaluated pairwise interactions between factors. The ecological detector elucidated the combined ecological impacts of these factors. 5. Quantitative analysis and summary of driving factors: The results from the above analyses were statistically evaluated to quantify the influence of each driving factor on vegetation changes across varying temporal and spatial scales. This enabled the identification and summarization of the primary drivers of vegetation change in the BGT region.

2.4. Research Methods

2.4.1. Theil–Sen Median Trend Analysis and MK Test

In order to investigate the trend of the Normalized Vegetation Index (NDVI) in the BGT, we first employed the Theil–Sen median trend estimation method, followed by the MK test to assess the statistical significance of the observed trend. The Theil–Sen median trend analysis is a nonparametric statistical method with several advantages: it does not require the sample data to conform to a specific distribution, it exhibits high computational efficiency, and it remains robust against outliers without sacrificing accuracy. This method demonstrates a strong capacity to mitigate the impacts of measurement errors or anomalous data [44]. Comparative analyses with various linear regression models further reveal that this approach provides distinct advantages, particularly when applied to small sample sizes [45].

The Theil–Sen Median method is a robust nonparametric trend calculation technique. It is computationally efficient and well adapted for analyzing trends in long-term time series data [46]. The method is expressed by the following formula:

$$\beta =Median\left({\displaystyle \frac{x_j-x_i}{j-i}}\right),\forall j>i$$

(1)

where $\beta$ denotes the trend of vegetation change; Median represents the median function; and $x_i$ and $x_j$ are the NDVI values for the i-th and j-th years, respectively. A positive $\beta (\beta$> 0) indicates an increasing NDVI trend, whereas a negative $\beta$ suggests a decreasing trend.

The MK test is another nonparametric method for assessing the significance of time series trends. This test does not require prior assumptions about the statistical distribution of the data. The normalized test statistic Z is used to evaluate both the direction and significance of the trend, with a threshold of |Z| ≥ 1.96 generally accepted as indicating statistical significance at the 95% confidence level [47]. The formula for the MK test is provided as follows:

$$S={\displaystyle {\sum }_{i=1}^{t-1}{\displaystyle {\sum }_{j=i+1}^{t}sign}}\left(NDV{I}_j-NDV{I}_i\right)$$

(2)

$$sign\left(NDV{I}_i-NDV{I}_j\right)=\left\{\begin{array}{ll}-1\hfill & \mathrm{if}\left(NDV{I}_i-NDV{I}_j\right)<0\hfill \\ 0\hfill & \mathrm{if}\left(NDV{I}_i-NDV{I}_j\right)=0\hfill \\ 1\hfill & \mathrm{if}\left(NDV{I}_i-NDV{I}_j\right)>0\end{array}\right.$$

(3)

$$\mathrm{var}(S)={\displaystyle \frac{t(t-1)(2t+5)}{18}}$$

(4)

$$Z=\left\{\begin{array}{ll}{\displaystyle \frac{S-1}{\sqrt{\mathrm{var}(S)}}}\hfill & S>0\hfill \\ 0\hfill & S=0\hfill \\ {\displaystyle \frac{S+1}{\sqrt{\mathrm{var}(S)}}}\hfill & S<0\hfill \end{array}\right.$$

(5)

2.4.2. Pixel-Based Calculation of the CV for Vegetation NDVI

The CV is a widely used metric in vegetation studies, primarily employed to characterize the degree of data dispersion and interannual fluctuations in vegetation [48]. By calculating the CV, temporal variability in vegetation NDVI can be observed, effectively quantifying the dispersion and mean level of NDVI data across time series. This metric is commonly applied to evaluate data stability [49]. The formula for the CV is expressed as follows:

$$C_v={\displaystyle \frac{\sqrt{\frac{1}{n-1}{\displaystyle {\sum }_{i=1}^n{(NDV{I}_i-\overline{NDVI})}^2}}}{\overline{NDVI}}}$$

(6)

2.4.3. GD Model

The GD model, developed by Wang Jinfeng and colleagues [50], is a statistical framework designed to detect spatial stratified heterogeneity and uncover its underlying driving forces. Its minimal dependence on distributional assumptions makes it particularly suitable for analyzing mixed-type data, contributing to its widespread use in studies of natural sciences and socio-economic factors. The model comprises four modules: factor detection, interaction detection, risk zone detection, and ecological detection. This study primarily utilizes the factor detector and interaction detector modules.

Factor Detection:

The factor detection module quantifies the influence of a driving factor X on changes in vegetation NDVI, expressed through the q-statistic:

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

(7)

$$SSW={\displaystyle \sum _{h=1}^L{N}_h{\sigma }_h^2,SST=N{\sigma }^2}$$

(8)

Here, h = 1, 2…, L denotes the strata (categories or partitions) of the dependent variable Y (NDVI) or the driving factor X; N and N_h represent the total number of units across the entire region and within stratum h, respectively; ${\sigma }_h^2$ and ${\sigma }^2$ are the variances in Y within stratum h and across the entire region, respectively; SSW is the sum of within-stratum variances, and SST is the total variance in the region. The q-value ranges from 0 to 1, with higher values indicating a stronger influence of the driving factor X on vegetation NDVI.

Interaction Detection:

The interaction detection module assesses the combined effects of driving factors, such as X₁ and X₂, on vegetation NDVI. It determines whether their joint influence enhances, weakens, or remains independent of their individual effects. Interaction relationships are categorized into five types, as outlined in

Table 2. Interaction types between two factors.

Determination	Interaction Effect
q (X1 ∩ X2) < min [q (X1), q (X2)]	Nonlinear attenuation
min [q (X1), q (X2)] < q (X1 ∩ X2) < max [q (X1), q (X2)]	Single-factor nonlinear attenuation
q (X1 ∩ X2) > max [q (X1), q (X2)]	Two-factor synergistic enhancement
q (X1 ∩ X2)= q (X1) + q (X2)	Independence
q (X1 ∩ X2) > q (X1) + q (X2)	Nonlinear amplification

.

Ecological Detection:

The ecological detection module utilizes the F-statistic to evaluate whether the spatial heterogeneity of the NDVI, as influenced by two distinct factors, exhibits statistically significant differences. When a significant difference is identified, it is denoted as “Y”; otherwise, it is denoted as “N”. This analysis enables the identification of which factor exerts a greater influence on NDVI variability.

Selection of Driving Factors:

The spatiotemporal dynamics of vegetation cover are shaped by the combined effects of natural geographical processes and human activities. Considering the ecological vulnerability of the BGT and the sustainable development needs of its liquor production bases, this study builds upon previous research [51,52,53,54] and integrates the specific conditions of the BGT. A total of nine variables (

Table 3. Driving factors of the NDVI.

Driving Factors	Codes	Units
mean annual temperature	X1	°C
annual precipitation	X2	mm
slope	X3	°
aspect	X4	°
elevation	X5	m
soil data	X6	/
land use	X7	/
population density	X8	person/km2
night-time lights	X9	lx

) encompassing both natural and anthropogenic driving factors were comprehensively selected. Natural factors include climatic conditions (X1: annual mean temperature, X2: annual precipitation), topographic gradients (X3: slope, X4: aspect, X5: elevation), and soil properties (X6: soil type). Climatic factors, particularly hydrothermal combinations, influence the NDVI dynamics by regulating vegetation photosynthetic thresholds and phenological cycles. Topographic factors, on the other hand, constrain vegetation spatial differentiation through mechanisms such as slope-dependent surface runoff retention efficiency, aspect-driven variations in solar radiation, and elevation-mediated hydrothermal redistribution. Anthropogenic factors include (X7: Land use—reflecting land use change intensity, X8: Population density grid data—indicating population spatial agglomeration, X9: NPP-VIIRS night-time light index—representing economic development intensity). These factors, respectively, characterize land cover fragmentation resulting from the expansion of the liquor industry, encroachment of urbanization and population pressure on ecological spaces, and nonlinear disturbances imposed by industrial park construction on vegetation resilience. The selection of these driving factors is informed by studies on vegetation response mechanisms in karst ecological zones and research on eco-industrial synergy in the Chishui River Basin. To address the distinctive characteristics of the liquor production region, indicator representation was optimized. For instance, the night-time light index (X9) effectively captures high-density industrial activities and infrastructure expansion trajectories within liquor enterprise clusters. The spatial distribution of these nine driving factors is illustrated in Figure 3.

3. Results

3.1. Spatiotemporal Variation Characteristics of NDVI

3.1.1. Temporal Variation Characteristics of NDVI

In the BGT region, the temporal variation in the NDVI over the past two decades (Figure 4) shows an overall upward trend from 2002 to 2022. Specifically, the NDVI started at 0.469 in 2002 and demonstrated periodic fluctuations, reaching notable peaks in 2007 (0.508) and 2016 (0.572). Post-2016, the NDVI values consistently remained above 0.50, indicating sustained favorable vegetation cover conditions. Nevertheless, the index displayed some variability, with minor declines observed in 2011 and 2014, followed by a marked increase in 2021, ultimately reaching 0.5306 in 2022. This trajectory reflects a long-term improvement in vegetation conditions within the study area despite intermittent influences from annual environmental factors. Overall, the NDVI in the BGT region tends to increase steadily from 2002 to 2022, with long-term enhancement remaining evident notwithstanding short-term fluctuations.

To comprehensively analyze and understand the spatiotemporal variation characteristics of the NDVI in the BGT region, this study draws upon existing literature and employs the equal interval method in ArcGIS to classify the NDVI into four distinct levels [52,55]: low coverage (<0.2), medium–low coverage (0.2–0.4), medium–high coverage (0.4–0.6), and high coverage (≥0.6) (

Table 4. Statistical proportion of NDVI-graded area in China’s BGT.

NDVI	Vegetation Condition Classification	Area Proportion of NDVI Classification/%
		2002	2009	2015	2022	Multi-Year Average
<0.2	Low Coverage	0.41	0.32	0.29	0.26	0.32
0.2~0.4	Moderate–Low Coverage	8.99	2.66	3.30	3.28	4.56
0.4~0.6	Moderate–High Coverage	88.92	89.84	86.79	81.96	86.88
>0.6	High Coverage	1.68	7.18	9.62	14.50	8.25

). The results reveal that, over the period from 2002 to 2022, the NDVI in the BGT region exhibited a phased, fluctuating upward trend. Over these 20 years, the mean NDVI increased by 9.80%, corresponding to an average annual growth rate of 0.49%, indicative of a sustained improvement in vegetation coverage within the study area.

Data presented in Table 4 demonstrate that the proportion of areas with low vegetation coverage (<0.2) and medium–low vegetation coverage (0.2–0.4) decreased progressively over the study period. Specifically, the area classified as low coverage declined from 0.41% in 2002 to 0.26% in 2022, while the medium–low-coverage area decreased from 8.99% to 3.28%. In contrast, the proportion of areas with medium–high coverage (0.4–0.6) and high coverage (>0.6) exhibited a consistent increase. Notably, the high coverage area expanded substantially, rising from 1.68% in 2002 to 14.50% in 2022. These findings collectively suggest a positive trend in vegetation growth across the BGT region.

3.1.2. Spatial Variation Characteristics of NDVI

Further exploration of the NDVI variation characteristics within the study area reveals distinct spatial patterns. As illustrated in Figure 5, in 2002, regions with high and medium–high vegetation cover in the BGT were predominantly located in Pingshan County of Yibin, Gulin County of Luzhou, the central part of Chishui City, the central part of Tongzi County, and the northern part of Xishui County in Zunyi. These areas, typically distant from urban centers, exhibited robust vegetation cover. Conversely, regions with low and medium–low vegetation cover were concentrated in Longmatan District, Luxian County, and Jiangyang District of Luzhou, as well as Honghuagang District and Bozhou District of Zunyi, displaying a distribution pattern centered around major urban hubs.

From 2002 to 2022, the extent of high vegetation cover expanded significantly, particularly in the Yibin and Luzhou regions of Sichuan Province. Notable examples include Xuzhou District, Changning County, Xingwen County, Pingshan County, and Gong County of Yibin, as well as Hejiang County and Xuyong County of Luzhou, all of which demonstrated marked increases in vegetation cover. Similar growth trends were observed in Chishui City and Wuchuan Gelao and Miao Autonomous County in Zunyi. Concurrently, areas with low and medium–low vegetation cover diminished, especially in Luzhou, where spatial contraction indicates improved vegetation conditions in peri-urban zones. Overall, the expansion of high vegetation cover is predominantly concentrated in Sichuan Province, underscoring a regional enhancement in vegetation cover.

3.2. Sustainability of Vegetation NDVI Changes

Utilizing the Sen + MK trend analysis method, we generated a significance map of the annual mean NDVI change trends in the BGT from 2002 to 2022 (Figure 6). The analysis reveals that vegetation in the majority of the BGT exhibits an improving trend, with the magnitude of increase far surpassing that of decrease. According to the trend analysis (

Table 5. MK test trend categories.

$\mathit{\beta }$	Z	Trend Characteristics	Area Proportion/%
$\beta <0$	2.58 < Z	Extremely significant decrease	0.58
	1.96 < Z < 2.58	Highly significant decrease	0.41
	1.65 < Z < 1.96	Moderately significant decrease	0.21
	Z < 1.65	Non-significant decrease	2.93
$\beta =0$	Z	No significant change	0.04
$\beta >0$	Z < 1.65	Non-significant increase	19.14
	1.65 < Z < 1.96	Marginally significant increase	6.85
	1.96 < Z < 2.58	Significant increase	18.15
	2.58 < Z	Extremely significant increase	51.68

), the most pronounced improvement is a highly significant increase, accounting for 51.68% of the area, primarily concentrated in Sichuan Province. In Guizhou Province, this trend is more dispersed, mainly observed in Wuchuan Gelao and Miao Autonomous County, the northern and central parts of Tongzi County, and regions bordering Sichuan Province. Areas showing non-significant, marginally significant, and significant increases in vegetation cover account for 19.14%, 6.85%, and 18.15%, respectively. Vegetation cover degradation, by contrast, is minimal, comprising only 4.17% of the total area, and is predominantly localized around the urban centers of Yibin, Luzhou, and Zunyi, with degradation patterns radiating outward from these cities.

3.3. Volatility of Vegetation NDVI

In this study, the CV of the NDVI in the BGT region from 2002 to 2022 was calculated (Figure 7), revealing the fluctuation characteristics of vegetation cover in this area. For further quantitative analysis, based on established classification principles [56,57,58], the CV was categorized into five levels: high volatility (≥0.20), relatively high volatility (0.15–0.20), medium volatility (0.10–0.15), relatively low volatility (0.05–0.10), and low volatility (0–0.05). Nearly half of the BGT region exhibits medium volatility, while 47.81% of the area shows low to relatively low volatility in vegetation cover, indicating overall high stability of the NDVI in the study area from 2002 to 2022.

Analyzing the spatial distribution of CV in the BGT, over the past 20 years, areas with relatively high and high volatility in vegetation cover are primarily concentrated in Pingshan County, Cuiping District, Nanxi District, Xuzhou District, Jiang’an County, and Gao County of Yibin; Longmatan District, Jiangyang District, and Hejiang County of Luzhou; and Honghuagang District, Bozhou District, Meitan County, and Yuqing County of Zunyi. In contrast, areas with low volatility are mainly found in Jiangyang District and Hejiang County of Luzhou, and Gong County and Xingwen County of Yibin.

3.4. Analysis of Driving Factors Influencing Vegetation NDVI Changes

3.4.1. Factor Detection

This study aims to elucidate the roles of various factors in the spatial differentiation of vegetation NDVI in BGT. We selected nine variables from the years 2002, 2007, 2012, 2017, and 2022, comprising natural factors—annual mean temperature (X1), annual precipitation (X2), slope (X3), aspect (X4), elevation (X5), and soil type (X6)—and human activity factors—land use (X7), population density (X8), and night-time light (X9)—to conduct a factor detection analysis.

As illustrated in Figure 8, between 2002 and 2022, the influence of annual mean temperature (X1) on vegetation NDVI in BGT progressively weakened, with its q-value decreasing from 0.1433 to 0.0155. This decline suggests a diminishing contribution over time. Conversely, the impact of annual precipitation (X2) increased steadily, with its q-value rising from 0.0368 in 2002 to 0.1110 in 2022, underscoring its growing role as a driver of vegetation changes. In comparison, the effects of slope (X3), aspect (X4), elevation (X5), and soil type (X6) remained relatively stable and minor, exhibiting limited fluctuations in their q-values. This indicates a consistent but modest influence on the spatial differentiation of vegetation. Overall, while precipitation has assumed an increasingly prominent role in NDVI changes in BGT, the influence of average annual temperature has waned, with other natural factors maintaining steady contributions.

Regarding human activity factors, the influence of land use type (X7) on vegetation NDVI in BGT displayed variability over the study period. Its q-value decreased from 0.1928 in 2002 to 0.1202 before rebounding to 0.1610 by 2022, reflecting an unstable impact on vegetation dynamics. The q-value for population density (X8) rose consistently from 0.0414 to 0.0616, indicating a strengthening influence on vegetation changes, particularly amid accelerating urbanization in BGT. Most notably, the q-value for night-time light (X9) surged from 0.0145 to 0.1333, with a marked increase between 2017 and 2022. This pronounced rise highlights a substantial escalation in its impact on vegetation, reflecting the intensifying pressures of urbanization and human activities in BGT. Collectively, these findings demonstrate that human activity factors, especially night-time light, have progressively emerged as critical drivers of vegetation changes in the region.

3.4.2. Interaction Detection

In this study, we analyzed the interactive effects of various factors influencing the NDVI in the BGT region (see Figure 9). Interactions among natural factors, such as annual mean temperature (X1) and annual precipitation (X2), or slope (X3) and elevation (X5), predominantly exhibited nonlinear enhancement. This suggests that the combined explanatory power of multiple factors on geospatial differentiation significantly exceeds that of individual factors, manifesting as a nonlinear superposition effect. For instance, the interaction between annual mean temperature (X1) and annual precipitation (X2), denoted as X1 ∩ X2, consistently displayed an “Enhance, nonlinear-” pattern from 2002 to 2022. This highlights the complex nonlinear characteristics of climatic factor combinations in shaping ecological patterns within BGT.

Interactions between natural and human activity factors—such as elevation (X5) and land use (X7), or soil type (X6) and population density (X8)—demonstrated a rising proportion of nonlinear enhancement over time. For example, the interaction between elevation (X5) and land use (X7), denoted as X5 ∩ X7, shifted from “Enhance, bi-” to “Enhance, nonlinear-” after 2017. This transition indicates that the breakthrough effect of human activities on topographic constraints has progressively strengthened in BGT.

Other interaction types, including linear enhancement or weakening, were also observed. Interactions among human activity factors, such as land use (X7) and population density (X8), or population density (X8) and night-time light (X9), typically exhibited bidirectional linear enhancement (“Enhance, bi-”). For instance, the interaction between population density (X8) and night-time light (X9), denoted as X8 ∩ X9, remained consistently stable as a bidirectional synergistic enhancement from 2002 to 2022. This stability reflects the strong correlation between population agglomeration and economic activities during urbanization in BGT. Conversely, the enhancement effect of certain natural factor interactions—such as aspect (X4) and soil type (X6), or annual precipitation (X2) and aspect (X4)—weakened over time, potentially due to influences from climate change or human intervention.

As regards spatiotemporal evolution trends, from 2002 to 2012, natural factors dominated the interactions, with nonlinear enhancement primarily concentrated in combinations of climatic factors (X1, X2) and topographic factors (X3, X5). During this period, the interactive effects of human activity factors remained relatively weak. However, between 2017 and 2022, nonlinear interactions between human activity factors (X7, X8, X9) and natural factors markedly increased. For example, the interaction between land use (X7) and annual mean temperature (X1), denoted as X1 ∩ X7, transitioned to nonlinear enhancement in 2022. This shift suggests that the capacity of human activities to transform natural systems in BGT has progressively surpassed simple linear relationships.

3.4.3. Ecological Detection

In conducting ecological detection of vegetation NDVI changes in the BGT region, we established a relationship matrix incorporating ecological and risk detection factors (X1 to X9). The associations between these factors were assessed using significance testing (F-test) at a 0.05 significance level, corresponding to a 95% confidence interval. Pairs of factors passing the significance test were considered to exhibit significant spatial distribution differences, denoted as “Y” in

Table 6. Ecological detection results.

Driving Factor	X1	X2	X3	X4	X5	X6	X7	X8
X2	Y
X3	Y	N
X4	N	Y	Y
X5	N	Y	Y	N
X6	N	Y	N	N	N
X7	Y	Y	Y	Y	Y	Y
X8	N	N	N	Y	N	N	Y
X9	Y	N	Y	Y	Y	Y	N	Y

Note: Y indicates a significant association, N indicates no significant association.

, while “N” indicates insufficient evidence to support an association.

Ecological detection revealed marked differences in factor interactions over the 20-year study period. In 2002, significant interactions were observed between the annual mean temperature (X1) and both annual precipitation (X2) and night-time light (X9), whereas the association between slope (X3) and soil data (X6) remained weak. Over time, the complexity of factor interactions increased: the relationships between annual mean temperature (X1) and aspect (X4), as well as night-time light (X9), strengthened, and land use (X7) consistently exhibited strong associations with other factors across both periods. By 2022, interactions involving population density (X8) displayed a more differentiated pattern, underscoring the increase in the impact of human activities on vegetation ecology.

Overall, the driving mechanisms of the ecosystem in the BGT region are shifting from being predominantly governed by natural factors to a combined influence of natural and anthropogenic drivers. This transition highlights the increasingly critical roles of climate change and human activities in shaping the vegetation ecosystem over the past two decades.

4. Discussion

This study, based on remote sensing monitoring and the GD model, systematically analyzed the spatiotemporal variation characteristics of vegetation NDVI and its driving factors in the BGT region from 2002 to 2022. The results indicate that the NDVI in this area has generally increased, albeit with significant spatial heterogeneity, driven by the complex interaction between natural environmental factors and human activities. This section discusses the dynamic change characteristics of the NDVI, the mechanisms of driving factors, and their ecological implications, aiming to deepen the understanding of the change mechanisms in the BGT ecosystem and provide a scientific basis for regional sustainable development.

4.1. Spatiotemporal Evolution of Vegetation NDVI and Its Ecological Significance

This study shows that from 2002 to 2022, the NDVI in the BGT region generally shows an increasing trend, with the average annual NDVI growing from 0.45 to 0.67, with an average annual growth rate of 0.49%. This trend aligns with ecological restoration patterns observed in the upper Yangtze River and southwestern karst regions. For instance, Gao et al. (2024) demonstrated that reforestation efforts in karst areas of the upper Yangtze River have significantly enhanced vegetation cover since 2000 [59]. Similarly, He et al. (2022) used Google Earth Engine to show that ecological compensation policies in southwest China have driven forest recovery in karst landscapes [60]. Huang et al. (2024) further highlighted the role of environmental governance, such as the Grain for Green program, in improving vegetation along the Yangtze River shoreline [61]. These studies collectively underscore the positive impact of ecological restoration initiatives, which our findings corroborate in the BGT context. However, NDVI changes were not uniformly progressive, with notable declines in 2006 and 2011. These fluctuations are likely linked to extreme climatic events and anthropogenic disturbances. Zhu et al. (2023) found that drought events in northeastern China significantly reduced the NDVI in karst ecosystems, emphasizing the role of hydrothermal variability [62]. Similarly, Lu et al. (2025) reported that flood events in southwestern China disrupted vegetation recovery, particularly in areas with high human activity [63]. Our study supports these findings, suggesting that extreme weather, combined with urban expansion and agricultural development, contributed to NDVI declines in the BGT during these years.

The spatial analysis results reveal that high-NDVI areas within the BGT region are predominantly situated in ecologically favorable mountainous forest zones, such as Pingshan County in Yibin, Gulin County in Luzhou, and Chishui City in Zunyi. These areas benefit from abundant precipitation, high forest coverage, and minimal human interference, contributing to stable vegetation conditions. A focused analysis of Figure 6a–c provides a clearer visualization of low-NDVI areas, which are concentrated in regions experiencing rapid urbanization, including Longmatan District and Jiangyang District in Luzhou (Figure 6b) and Honghuagang District in Zunyi (Figure 6c). These patterns highlight the pronounced impact of urban expansion on regional ecosystems.

Existing research indicates that urbanization processes—marked by increased impervious surfaces, intensified heat island effects, and shifts in land use—can drive localized ecological degradation. Yao et al. (2020) demonstrated that urbanization in Beijing increased impervious surfaces, reducing vegetation cover [64]. Seifollahi-Aghmiuni et al. (2022) reported similar trends in southern Europe, where urban expansion led to land degradation in peri-urban areas [65]. Coseo and Larsen (2019) further noted that urban heat island effects exacerbate vegetation loss in urban environments [66]. The findings of this study reinforce this conclusion, demonstrating that urban sprawl has become a major driver of spatial variability in vegetation cover across the BGT region.

Furthermore, NDVI variations reflect a competitive interplay between ecological restoration and human interference. Ecological restoration policies, such as the Grain for Green program, have demonstrably enhanced vegetation cover. However, the effects of urbanization, agricultural expansion, and industrial development remain significant. For example, certain areas exhibit reduced vegetation cover due to industrial and infrastructure development, with the positive impacts of ecological restoration partially suppressed in high-density population zones. While ecological restoration policies have enhanced vegetation cover, urbanization and industrial development pose significant challenges. Liu and Zhang (2024) found that urbanization in the Guizhou karst region suppressed the benefits of ecological restoration in densely populated areas [67]. Huang et al. (2024) similarly noted that industrial expansion along the Yangtze River counteracted vegetation recovery efforts [61]. Our findings align with these studies, highlighting the need to balance economic development with environmental protection to mitigate the ecological threats posed by overdevelopment in the BGT.

4.2. Driving Mechanisms of Vegetation NDVI: Interaction Between Natural Environment and Human Activities

4.2.1. Dominant Role of Natural Factors

This study identifies climatic and topographic factors as foundational drivers of vegetation changes in the BGT region, with precipitation emerging as the most critical natural factor (q-value = 0.111). Situated within a subtropical monsoon climate zone, the BGT region is directly influenced by the spatiotemporal distribution of precipitation, which affects soil moisture and vegetation growth conditions. Previous studies have demonstrated that vegetation dynamics in the upper Yangtze River region are strongly regulated by precipitation variability, with significant recovery during years of ample rainfall and declines in the NDVI during drought periods [68,69]. These findings are corroborated by our study, which reveals pronounced NDVI fluctuations in the BGT region during years with anomalous precipitation. In Sichuan Province, Ning et al. (2023) employed the GD model to analyze the driving factors of vegetation changes from 2000 to 2020, identifying elevation as the primary driver (q = 0.417), with precipitation also exerting a significant influence (q > 0.05). This aligns with our observation that precipitation is a key natural factor in the BGT region, including Yibin and Luzhou [70]. In Guizhou Province, Chen and Xu (2023) investigated the spatial heterogeneity of human activities and their driving factors, including NDVI changes. Using the GD model, they found that while human activities had the greatest impact, natural factors such as slope, precipitation, elevation, and temperature also played significant roles in shaping NDVI distribution. This supports our conclusion that precipitation and topographic factors are critical drivers in the BGT region, particularly in Zunyi [71]. In the Qinba Mountains, which include parts of Sichuan, Zhang et al. (2022) applied the Geodetector method to examine vegetation dynamics, identifying geomorphological type, aridity index, humidity index, and annual precipitation as primary drivers of NDVI changes, with precipitation exhibiting high explanatory power (q > 0.19). This further validates the dominant role of precipitation in the BGT region [72].

Additionally, topographic factors, including elevation and slope, impose significant constraints on the spatial distribution of vegetation cover in the BGT. High-altitude areas, characterized by a humid climate, forest-dominated vegetation, and limited human interference, sustain consistently high NDVI levels over the long term. In contrast, low-altitude areas experience reductions in vegetation cover due to pressures from agricultural development and urbanization. In the adjacent Chengdu–Chongqing region, Zhang et al. (2023) found that elevation significantly influences the NDVI, highlighting the importance of topographic factors in vegetation dynamics, consistent with our findings [73].

4.2.2. Enhanced Role of Human Activities

Among factors related to human activities, land use change and the night-time light index demonstrated the highest explanatory power (q-values of 0.161 and 0.133, respectively), suggesting that urban expansion and industrial activities exert an increasingly pronounced influence on vegetation cover changes in the BGT region. Previous studies have identified industrial development and infrastructure construction as critical drivers affecting regional ecological environments. Gao et al. (2024) highlighted that industrial development in karst areas of the upper Yangtze River accelerates land use change, reducing vegetation cover [59]. Li and Zhang (2022) showed that industrial growth in the Yangtze River Economic Belt exerts significant ecological pressure, impacting the NDVI [74]. Xu et al. (2022) used machine learning to demonstrate that infrastructure construction in karst basins degrades vegetation [75]. The present study further reveals a significant increase in the influence of the night-time light index over the period from 2002 to 2022, highlighting industrial development and urban expansion as pivotal factors shaping vegetation cover dynamics. This effect is particularly evident in peri-urban areas, such as Yibin, Luzhou, and Zunyi—core zones of the liquor industry—where a notable negative correlation exists between rising urban light pollution and vegetation decline.

Furthermore, interaction detection analysis indicates a significant nonlinear enhancement effect between natural factors and human activities. For example, the interplay between night-time light and precipitation exhibits “nonlinear enhancement,” implying that economic activities not only directly impact vegetation cover but may also indirectly affect vegetation growth by modifying local hydrological cycles. Xia et al. (2021) found that night-time light in Southeast Asia correlates with vegetation loss by modifying local climates [76]. Yi et al. (2022) reported that urbanization in the Yangtze River Basin disrupts hydrological processes, impacting the NDVI [70]. These studies support our finding that ecosystem evolution in the BGT is a coupled system driven by both natural and anthropogenic factors. Additionally, Guo et al. (2021) demonstrated that climatic and human factors interactively shape NDVI dynamics in the Mongolian Plateau [77], while Xin et al. (2008) noted similar interactions on the Chinese Loess Plateau [78]. To further explore interaction effects, Wang et al. (2022) used the GD model to show that combined climatic and anthropogenic factors drive ecological quality in northern Anhui, reinforcing the utility of the GD approach [24]. Guo et al. (2024) also applied the GD model to reveal multi-factor interactions affecting vegetation in the Yongding River Basin, supporting our methodology [30].

4.3. Limitations of This Study

4.3.1. Limitations of Data Accuracy

The current study encounters two primary limitations in terms of data precision. Firstly, due to constraints in spatial resolution, Landsat data may exhibit mixed pixel effects when extracting NDVI values in suburban peripheries, fragmented croplands, and areas of mixed land use [79]. This limitation impedes the ability to capture ecological heterogeneity at local scales, particularly in urban–rural transition zones. Secondly, the findings predominantly rely on remote sensing estimates and publicly available datasets. In the absence of integrated field survey data, it remains challenging to validate NDVI estimates across diverse land use types or to evaluate the applicability of driving factors identified by the Geographical Detector at microscales. This constraint somewhat undermines the external reliability and policy adaptability of this study’s conclusions.

To address these shortcomings, future research should incorporate multi-temporal field control points to validate NDVI monitoring and land use classification accuracy. It is recommended to establish sampling plots in representative land types, combining ground observations with high-resolution drone imagery for cross-validation of multi-source data. This approach will bolster the empirical robustness and practical applicability of the research outcomes. Furthermore, this process will support the development of a cohesive “remote sensing monitoring—ground observation—mechanism identification” framework, offering more compelling evidence to inform decision-making for the coordinated development of regional ecology and industry.

4.3.2. Incomplete Collection of Influencing Factors

The current analysis primarily focused on factors such as climate, topography, and land use in assessing their effects on vegetation while omitting socio-economic variables such as economic development and policy interventions. These unaccounted-for factors may exert significant influence on vegetation changes in the BGT region across varying temporal scales. Future research could incorporate socio-economic data and policy analysis to develop a more comprehensive framework of driving factors, thereby improving this study’s overall robustness.

5. Conclusions and Policy Recommendations

5.1. Conclusions

This study, utilizing remote sensing data from 2002 to 2022, employed Theil–Sen trend analysis, the MK test, CV analysis, and the GD model to examine the spatiotemporal evolution characteristics and driving mechanisms of vegetation NDVI in the BGT region of China. The key findings are summarized as follows:

(1)

Overall upward trend in vegetation NDVI with phased fluctuations and spatial heterogeneity: Between 2002 and 2022, the mean NDVI in the BGT region increased from 0.45 to 0.67, reflecting a general improvement in vegetation conditions. However, transient declines observed in 2006 and 2011 may be attributable to extreme climatic events or shifts in land use. High vegetation coverage was predominantly concentrated in ecologically advantageous high-altitude forested areas, whereas reductions in vegetation were more pronounced in zones undergoing urban expansion.

(2)

Precipitation and topography as dominant natural drivers of vegetation dynamics: Precipitation emerged as the most influential factor affecting NDVI variations. High-altitude zones, characterized by extensive forest cover and favorable climatic conditions, exhibited relatively stable NDVI values. In contrast, low-altitude areas, subject to intensified human interference, displayed greater vegetation fluctuations.

(3)

Increasing impact of human activities, with land use and urbanization as pivotal factors: Land use change demonstrated the strongest explanatory power for NDVI variations. The expansion of urban areas, evidenced by increased land conversion and a rising night-time light index, highlights the growing influence of economic development on vegetation cover in the BGT region. This impact was particularly notable in industrial agglomeration zones, where the positive effects of vegetation recovery were partially constrained.

(4)

Interactive effects of natural and anthropogenic factors shaping regional vegetation patterns: A nonlinear interaction between night-time light and precipitation indicates that urban expansion not only directly impacts vegetation but may also indirectly modify ecosystems by altering hydrological and climatic processes. This interaction exacerbates the spatial differentiation of vegetation across the region.

In conclusion, while vegetation conditions in the BGT region have shown improvement over the past two decades, the ecological pressures stemming from urbanization and expansion remain substantial. Future strategies should prioritize the optimization of land use policies and the adoption of ecologically sustainable development models to ensure the harmonious advancement of the regional ecological environment and industrial growth.

5.2. Policy Suggestions

Based on a systematic analysis of the spatiotemporal evolution characteristics and driving mechanisms of vegetation NDVI in the BGT region from 2002 to 2022, the following targeted policy suggestions are proposed. These recommendations aim to provide decision-making references for regional ecological environment governance and the sustainable development of the Baijiu industry:

(1) Establish Differentiated Ecological Control Zones Based on NDVI Dynamic Monitoring.

This study reveals that areas exhibiting significant NDVI increases are predominantly located in high-altitude mountainous forest regions, such as Pingshan County in Yibin City and Gulin County in Luzhou City. These areas demonstrate high ecological stability and relatively low levels of human disturbance, making them suitable for designation as “ecological function core zones”. To preserve the integrity and stability of the regional ecosystem, strict ecological protection measures should be enforced in these zones, including restrictions on high-intensity industrial development activities.

In contrast, areas with significant NDVI decreases or pronounced fluctuations are primarily concentrated in urban expansion frontier zones, such as the peripheries of Luzhou’s central urban area and the southern urban area of Zunyi City. These regions experience markedly increased pressure on ecological carrying capacity. It is recommended that such areas be incorporated into “ecological restoration priority zones”. To restore and enhance regional ecological service functions and carrying capacity, measures such as returning farmland to forest, implementing land ecological restoration, and constructing ecological infrastructure should be prioritized. These interventions aim to alleviate the conflict between urban expansion and ecological protection.

(2) Integrate Key Human Activity Intensity Indicators into Industrial Layout Constraints.

This study utilized the Geographical Detector method to quantitatively identify land use change (q = 0.161) and the night-time light index (q = 0.133) as the predominant anthropogenic drivers of vegetation ecosystem changes. The findings demonstrate that industrial expansion and urbanization processes impose significant negative impacts on the regional ecological environment. Consequently, it is proposed that an ecological environmental impact assessment system be established during the planning of site selection for Baijiu industrial parks and the spatial layout of new urban areas. This system should incorporate dynamic NDVI monitoring data alongside human activity intensity indicators. By establishing ecological protection baselines and early-warning thresholds, the scale of new construction land should be rigorously controlled while promoting the ecological transformation and efficiency enhancement of existing industrial land. Moreover, industrial spatial layouts should be progressively redirected toward regions with high ecological carrying capacity, thereby fostering a synergistic balance between industrial development and ecological preservation.

(3) Construct a Collaborative Ecological Regulation Mechanism Coupling Natural and Human Activities.

This research further elucidates a pronounced nonlinear interactive enhancement between natural environmental factors (e.g., precipitation) and human activities (e.g., night-time light index), with particularly marked effects on vegetation ecosystems in regions undergoing rapid urban expansion. Based on these insights, it is recommended to explore the development of an “ecological-industrial risk coupling regulation model”. This model should integrate spatiotemporal NDVI variation monitoring data with key driving factors identified through the Geographical Detector method, enabling regional ecosystem resilience assessments prior to major project approvals and planning decisions. Additionally, by leveraging real-time ecological monitoring data, a dynamic adjustment mechanism for industrial intensity under ecological capacity constraints should be instituted. This mechanism would facilitate adaptive alignment between the scale of industrial development and ecological environment capacity, fundamentally enhancing the coordinated development potential and sustainability of the regional ecological–economic system.