Vegetation Dynamics and Influencing Mechanisms in Zhejiang Province, a Typical Subtropical Region of China

Vegetation Dynamics and Influencing Mechanisms in Zhejiang Province, a Typical Subtropical Region of China | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Vegetation Dynamics and Influencing Mechanisms in Zhejiang Province, a Typical Subtropical Region of China Ke Wang, Hongwen Yao, Wei Jin, Nan Li, Jun Chen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7593091/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Vegetation cover plays a fundamental role in maintaining ecosystem structure and function. Understanding its spatial and temporal variability, along with its driving factors, is critical for advancing environmental studies. This research targets the subtropical Zhejiang region in southeastern China, utilizing MODIS-derived NDVI data covering 2001 to 2020. By integrating Sen’s slope estimator, Mann–Kendall trend analysis, spatial autocorrelation (Moran’s I), and the Geodetector framework, we assessed trends, patterns, and primary influencing factors of vegetation change. Our findings include: (1) A statistically significant upward trend in NDVI across 59.4% of the study area (Sen’s slope = 0.0025, p < 0.01), reflecting ongoing ecological improvement; (2) Notable spatial clustering of NDVI values, with high NDVI zones located in southwestern forested areas and low NDVI zones in expanding urban regions; (3) Elevation, slope, land use/cover, and nighttime lights were identified as major contributors to NDVI spatial variation, with notable interaction effects such as a nonlinear synergy between land use and light; (4) High-risk zones, associated with dense populations and intense urban development, coincided with lower NDVI values. These results deepen our understanding of vegetation dynamics in subtropical zones and provide insights for sustainable ecosystem and land management. Biological sciences/Ecology Earth and environmental sciences/Ecology Earth and environmental sciences/Environmental sciences NDVI spatiotemporal variation Sen–MK test Moran’s I Geodetector Zhejiang Province Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 1. Introduction As a fundamental element of land-based ecosystems, vegetation contributes critically to regulating climate, driving water cycles, and enabling carbon sequestration, while also supporting ecological balance.(Yong Xu et al. 2024 ). Vegetation cover not only reflects the structural and functional attributes of ecosystems, but is also widely regarded as a key indicator of ecological health and environmental quality. In recent years, accelerated urbanization, shifting land use patterns, and intensified impacts of climate change have led to notable spatiotemporal changes in vegetation cover at the global scale(Cumming et al. 2014). A systematic understanding of vegetation dynamics and their dominant driving factors is therefore essential for elucidating human–environment interactions, supporting regional ecological management, and advancing sustainable development goals(Dong et al. 2025 ; Gong et al. 2025 ). Subtropical regions are characterized by humid climates, rich biodiversity, and high ecosystem sensitivity, and thus represent critical areas for global ecological security(X. Xu and Chen 2024 ; L. Zhang et al. 2024 ; Qiu et al. 2025 ). However, these regions often face multiple pressures, including rapid urban expansion and industrial restructuring, which increasingly challenge ecosystem stability(Ziyi Wang et al. 2024 ). Against the backdrop of global change, investigating the spatiotemporal heterogeneity and driving mechanisms of vegetation dynamics in subtropical zones is of great importance for enhancing ecosystem resilience and regulatory capacity. In recent years, advances in remote sensing and spatial analysis have enabled extensive investigations of vegetation dynamics at both global and regional scales(Beck et al. 2006 ; Tang et al. 2021 ; L. Hu et al. 2023 ). Among these, the Normalised Difference Vegetation Index (NDVI) has been widely employed to monitor vegetation changes (Lin et al. 2022 ). Sen’s slope estimation and the Mann–Kendall test are commonly used to identify long-term trends and their significance(Yanxin Xu et al. 2023 ), Moran’s I is applied to evaluate spatial autocorrelation and clustering patterns(M. Liu et al. 2021 ), while the Geodetector model has recently been adopted to quantify the influence and interaction of natural and anthropogenic factors (Y. Zhang et al. 2023 ). Although previous studies have produced a wealth of findings, further exploration is needed in typical subtropical regions to understand how multifactorial interactions drive vegetation pattern evolution, especially under conditions of spatial heterogeneity. Located in southeastern coastal China, Zhejiang Province features a typical subtropical monsoon climate, complex topography, and diverse ecosystem types(Mao et al. 2022 ). As a densely populated, economically developed, and ecologically sensitive region, it exhibits a dual pattern of urban expansion and ecological restoration in recent years, driven by rapid socioeconomic development and ongoing ecological civilization initiatives(Wu et al. 2022 ). This interplay between disturbance and recovery makes Zhejiang a representative case for investigating the spatiotemporal dynamics of vegetation and their driving forces in subtropical regions(Zheng et al. 2019 ). In this study, we use Zhejiang Province as the study area and employ MODIS NDVI remote sensing data from 2001 to 2020. An integrated analytical framework combining Sen’s slope trend estimation, Mann–Kendall significance testing, Moran’s I spatial autocorrelation, and the Geodetector model is developed to assess vegetation change across three dimensions: temporal evolution, spatial distribution, and driving mechanisms. The objectives of this study are to: (1) identify long-term trends in NDVI and their significance; (2) reveal spatial clustering and heterogeneity patterns; and (3) quantify the dominant effects and interactions of natural and anthropogenic drivers. The findings aim to provide a scientific basis for ecological conservation and sustainable land use management in subtropical regions. 2. Materials and methods 2.1 Study area Zhejiang Province lies along China’s southeastern coastline, stretching between 118°01′–123°10′ E and 27°02′–31°11′ N(Huang et al. 2024 ). It is bordered by the East China Sea to the east, Fujian to the south, with Jiangxi and Anhui bordering its west, and Shanghai and Jiangsu to its north. The province covers a territorial coverage of about 105,500 km², featuring predominantly hilly and mountainous terrain, occupying nearly 70% of the province’s total surface. Plains are primarily found in the northern coastal belt and near key rivers. Zhejiang features a representative subtropical monsoon climate, featuring marked by clearly defined seasonal changes, warm and humid conditions, with mean annual temperatures between roughly 15°C and 18°C, and yearly rainfall ranging from 1100 mm to 2000 mm. The synchrony of rainfall and heat provides favorable conditions for the growth of various vegetation types. The province possesses abundant natural resources and supports a diverse range of ecosystems, including forests, wetlands, rivers, lakes, and coastal zones. In recent decades, fast-paced urban and industrial development has significantly altered the ecological landscape, with urban expansion and land-use changes imposing considerable pressure on natural ecosystems. Nevertheless, Zhejiang has placed great emphasis on ecological civilization, promoting the philosophy that “ clear waters and verdant mountains are priceless treasures”(N. Wang et al. 2022 ). A series of ecological protection and restoration projects have been implemented, offering strong support for vegetation recovery and ecosystem quality improvement. Given its complex topography, diverse climatic conditions, and pronounced human activity, Zhejiang Province serves as a representative and well-founded case for studying the spatiotemporal dynamics of vegetation cover and its driving mechanisms in subtropical regions. 2.2. Data Sources 2.1.2 Land Cover Data Land use/land cover (LULC) data were obtained from the China CLCD product(Yang and Huang 2021). The dataset spans from 2001 to 2020 with a spatial resolution of 30 m. Within GEE, the data were clipped to the extent of Zhejiang Province and resampled to 1000 m to ensure spatial resolution consistency with other variables. 2.1.3 Climate Data Temperature (TEM) together with precipitation (PRE) data were obtained from the National Tibetan Plateau Scientific Data Center ( http://data.tpdc.ac.cn , accessed 8 November 2023)(Peng et al. 2019 ), with a spatial resolution of 1 km. The final datasets include precipitation by month (unit: 0.1 mm) and average monthly temperature (unit: 0.1°C) at 1 km resolution for 2001–2020. These data have been validated against 496 independent meteorological stations and demonstrate high accuracy and reliability. We calculated the mean annual temperature and the cumulative yearly precipitation for the study area to represent climatic variables. Photosynthetically Active Radiation (PAR) data were acquired via the TerraClimate dataset(Y. Chen et al. 2024 ) and processed through clipping and resampling to a spatial resolution of 1 km for the period 2001–2020. 2.1.4 Nighttime Light Data Nighttime light (NTL) data were obtained from Version 4 of the DMSP-OLS time series, which provides cloud-free composites based on all available archived data with smoothed resolution for each calendar year. These datasets were sourced from the U.S. Air Force Weather Agency and subsequently processed by NOAA’s National Centers for Environmental Information (NCEI). We extracted annual average NTL values for Zhejiang Province from 2001 to 2020 to represent human activity intensity( https://www.noaa.gov/ ). 2.1.5 Population and GDP Data Population (POP) data were obtained from the WorldPop global population grid dataset( https://www.worldpop.org/ ), and annual values were extracted using GEE to calculate yearly averages at 100 m resolution. The Gross Domestic Product (GDP) data were sourced from the global 1 km × 1 km revised real GDP dataset, which addresses previous issues such as spatial–temporal discontinuity and overestimated growth. This dataset adopts a top-down estimation approach. Annual mean GDP values (unit: yuan/km²) were extracted for Zhejiang Province and resampled to a 1000 m resolution for analysis as economic drivers. 2.1.6 Topographic Variables Topographic elevation data were obtained from the SRTM dataset provided by the United States Geological Survey (USGS), with a spatial resolution of 30 m( https://www.earthdata.nasa.gov/data/instruments/srtm ). Slope (SLO) and aspect (ASP) were derived from the elevation (ELE) data. All topographic variables were clipped to the study area and resampled to a 1000 m resolution. 2.1.7 Soil Type Data Soil type (SOIL) data were obtained from the FAO HWSD_V2 dataset ( https://www.fao.org/land-water/databases-and-software/hwsd/zh/ ). The FAO90 layer was selected to represent major soil mapping units. The data were clipped to the boundary of Zhejiang Province and resampled to ensure consistency with the resolution of other factors. 3. Methods This study focused on Zhejiang Province and utilized MODIS NDVI remote sensing data spanning 2001 to 2020. A combination of spatial and geographical analysis methods was employed to examine the spatiotemporal patterns and driving factors behind subtropical vegetation cover across three key perspectives: temporal trends, spatial patterns, and driving mechanisms. First, Sen’s slope estimation together with the Mann–Kendall test were jointly applied on the NDVI time series to assess both the direction and statistical significance of vegetation change over time. Second, global and local Moran’s I indices served to measure spatial autocorrelation characteristics and identify clustering patterns of vegetation variation. Finally, the Geodetector model was applied to quantify the explanatory power of both natural environmental and anthropogenic factors on vegetation dynamics, as well as to explore their interaction effects. By integrating these analytical approaches, this research seeks to offer a thorough understanding of the spatiotemporal heterogeneity of vegetation change and its dominant drivers in Zhejiang Province, thereby offering scientific support for regional ecological management and policy formulation. The specific methods are described as follows. 3.1. Sen’s Slope Estimation Sen’s slope estimation is a nonparametric approach commonly applied for assessing monotonic trends within time series data (Bikeko and E. 2024). It is especially effective for examining long-term variations in remote sensing-derived vegetation indices, such as NDVI (Yan et al. 2022 ). This method derives the trend slope as the median value among all pairwise rates of change between different time points. It is robust to outliers and does not require the assumption of data normality, making it effective for reliably detecting consistent upward or downward trends present in the data. The fundamental computation procedure is as follows: \(\:{\beta\:}_{ij}=\frac{{X}_{j}-{X}_{i}}{j-i}\) (1) Where \(\:{X}_{i}\) ​ denotes the NDVI value in year i; n is the length of the time series; \(\:{\beta\:}_{ij}\) refers to the rate of NDVI change from year i to year j; (j − i) is the time interval; and \(\:{X}_{j}-{X}_{i}\) ​ is the NDVI change. The Sen’s Slope value is calculated as the median across all computed slopes, that is: \(\:\beta\:=\text{m}\text{e}\text{d}\text{i}\text{a}\text{n}\left({\beta\:}_{ij}\right)\) (2) Where \(\:\beta\:\) represents the trend of the variable over the entire time series. In this research, the Sen’s Slope approach was employed to quantitatively assess the trend of vegetation cover change in Zhejiang Province based on MODIS NDVI products from 2001 to 2020. This was combined with the Mann-Kendall significance test to identify regions with significant changes, aiming to reveal the dynamic evolution characteristics of the ecosystem under the combined influence of natural and human factors. 3.2. Mann–Kendall (MK) Trend Test To further evaluate the statistical significance of vegetation change trends identified through Sen’s slope estimation, the Mann–Kendall (MK) trend test was employed in this study(Almazroui and Şen 2020). The MK test is a widely used non-parametric statistical technique applied in environmental and climate change studies(Guan et al. 2020 ). It is specifically designed to determine whether a time series demonstrates a statistically significant monotonic increase or decrease. The MK test operates without imposing distributional assumptions on the data, and it offers several advantages, including broad applicability, computational efficiency, and robustness to outliers(Şan et al. 2021 ). The test works by comparing the values of all possible pairs in the time series and evaluating their relative magnitudes to construct a test statistic 𝑆, which is defined as follows: \(\:S=\underset{i=1}{\overset{n-1}{?}}?\underset{j=i+1}{\overset{n}{?}}?\text{s}\text{g}\text{n}?({x}_{j}-{x}_{i})\) (3) Where the sign function \(\:\text{s}\text{g}\text{n}?({x}_{j}-{x}_{i})\) ) is defined as follows: \(\:\text{s}\text{g}\text{n}?({x}_{j}-{x}_{i})=\left\{\begin{array}{c}1,\:\:\:\:{x}_{j}-{x}_{i}>0\\\:0,\:\:\:\:{x}_{j}-{x}_{i}=0\\\:-1,\:\:\:\:{x}_{j}-{x}_{i}<0\:\:\:\:\end{array}\right.\) (4) When the sample size is large (e.g., n ≥ 8), the test statistic S approximately follows a normal distribution and can be used to test the significance level. The calculation formula is as follows: \(\:Z=\left\{\begin{array}{c}\frac{S-1}{\sqrt{\text{V}\text{a}\text{r}?\left(S\right)}},\:\:\:\:S>0\\\:0,\:\:\:\:S=0\\\:\frac{S+1}{\sqrt{\text{V}\text{a}\text{r}?\left(S\right)}},\:\:\:\:S<0\:\:\:\:\end{array}\right.\) (5) Where \(\:\text{V}\text{a}\text{r}?\left(S\right)\) is the variance of the test statistic S. 3.3. Spatial Autocorrelation Analysis Using Moran’s I To further investigate the spatial distribution and dependency of NDVI variation, this study employed Moran’s I statistic for spatial autocorrelation analysis(Phung et al. 2015 ). Moran’s I is a widely recognized indicator used to measure the spatial similarity of a variable across a geographic region. It includes two forms: global Moran’s I and local Moran’s I, which describe overall spatial autocorrelation patterns and localized clustering (C. Liu et al. 2021 ). Global Moran’s I is used to assess whether a variable exhibits spatial clustering across the entire study area. It is calculated using the following formula: \(\:I=\frac{n}{W}\cdot\:\frac{\sum\:_{i=1}^{n}\:\sum\:_{j=1}^{n}\:{w}_{ij}({x}_{i}-\stackrel{-}{x})({x}_{j}-\stackrel{-}{x})}{\sum\:_{i=1}^{n}\:({x}_{i}-\stackrel{-}{x}{)}^{2}}\) (6) Where n is the number of samples, x i and xⱼ represent the variable values at units i and j, respectively, x̄ is the mean of the variable, and w i ⱼ is the spatial weight matrix (commonly represented by adjacency or inverse distance). \(\:W=\underset{i=1}{\overset{n}{?}}?\underset{j=1}{\overset{n}{?}}?{w}_{ij\:}\) When I > 0, it indicates that high or low values tend to cluster spatially (positive autocorrelation); when I < 0, it reflects that dissimilar values are adjacent (negative autocorrelation); and when I ≈ 0, it denotes a random spatial distribution. To further reveal the specific locations and types of spatial clusters, this study also introduces the Local Moran’s I, which is defined as: \(\:{I}_{i}=({x}_{i}-\stackrel{-}{x})\sum\:_{j}\:{w}_{ij}({x}_{j}-\stackrel{-}{x})\) (7) This indicator also enables the identification of local spatial clustering types, such as high–high and low–low aggregations, along with high–low and low–high anomalies. These patterns are typically visualized using Local Indicators of Spatial Association (LISA) cluster maps. The Moran’s I method is valued for its intuitive results and strong capacity to detect spatial patterns. It finds broad application in spatial structure analysis of geographical phenomena and plays a significant role in research areas such as vegetation cover, land use, and urban expansion. 3.4. Geodetector To determine the primary driving factors and how they interact in influencing vegetation cover change in Zhejiang Province, this study employed the Geodetector model(Y. Guo et al. 2024 ). Geodetector serves as a statistical approach for identifying spatial stratified heterogeneity and uncovering underlying driving forces. The core concept is that if an explanatory variable significantly influences a dependent variable, their spatial patterns should display strong similarity. The framework comprises four core modules: factor detector, interaction detector, ecological detector, and risk detector(H. Chen et al. 2024 ). This study mainly applied the factor and interaction detectors. The factor detector was applied to evaluate the explanatory strength of a single factor in explaining spatial variation in NDVI(Zhu et al. 2020 ). This is measured using a statistic known as the 𝑞-value, which ranges from 0 to 1. A higher 𝑞-value indicates a stronger explanatory power of the factor. The calculation is given by: \(\:q=1-\frac{\sum\:_{h=1}^{L}\:{N}_{h}\cdot\:{\sigma\:}_{h}^{2}}{N\cdot\:{\sigma\:}^{2}}\) (8) Where h denotes the h-th subregion, L is the number of subregions, and N represent the number of samples in the subregion and the entire region, respectively, and and are the variances of NDVI within the subregion and the entire region, respectively. A q value of 0 indicates that the factor is unrelated to NDVI variation, while q = 1 means the factor fully explains the NDVI variation. The interaction detector is used to analyze the type of interaction effect between two factors on NDVI variation. The interaction detector is used to assess how combinations of two factors influence NDVI variation, identifying types of interaction such as enhancement (the combined effect exceeds that of either factor alone), nonlinear enhancement (the combined effect far exceeds individual contributions), or independence (the combined effect approximates the stronger single factor)(Sun et al. 2024 ). Geodetector offers several advantages, including the absence of a linearity assumption, compatibility with heterogeneous data sources, and the ability to handle both categorical and continuous variables. It has been widely applied in studies on changes in land use, ecosystem dynamics, and human–environment interactions (Zhao et al. 2020 ). Within this research, the model was utilized to quantitatively determine the dominant roles and interaction effects of both natural factors (e.g., precipitation, temperature, topography) and socio-economic factors (e.g., GDP, population density, land use intensity) on vegetation change, thereby identifying the spatial variability of its driving processes. The ecological detector evaluates whether the means of NDVI differ significantly across various strata of a specific factor (H. Guo et al. 2023 ). Fundamentally, it functions as an analysis of variance (ANOVA), using F-tests to assess statistical differences between groups. This is particularly effective for detecting NDVI response differences among environmental or administrative zones. The risk detector identifies categories of an explanatory variable associated with significantly higher or lower values of the dependent variable (e.g., high or low NDVI), enabling the detection of high-risk or high-probability zones(Qiao et al. 2022 ). This is valuable for identifying sensitive areas or priority zones for ecological monitoring. Overall, the Geodetector model serves as a robust approach for exploring spatial mechanisms by which natural and human factors influence ecological variables. It is suitable for stratified or categorical spatial data, does not require linear assumptions, accommodates multiple explanatory variables, and is robust to multicollinearity. 4. Results 4.1. Temporal Variation of NDVI in Zhejiang Province Figure 3 illustrates the temporal variation and annual rate of change in the Normalized Difference Vegetation Index (NDVI) for Zhejiang Province from 2001 to 2020. Overall, the NDVI exhibits a statistically significant increasing trend. The linear regression results yield a slope of 0.0025 with a coefficient of determination 𝑅2 = 0.6658, indicating a sustained improvement in vegetation cover over the study period. The green-shaded area in the figure represents the 95% confidence interval, further confirming the statistical significance of the upward trend. In terms of annual rate of change, most years show positive NDVI growth, with particularly notable increases in 2002 (+ 4.9%), 2006 (+ 3.8%), and 2011 (+ 6.1%). However, several years also experienced declines in NDVI, including 2004 (–1.2%), 2007 (–0.9%), 2010 (–4.6%), and 2014 (–1.5%), suggesting potential impacts from climatic anomalies or human disturbances during those periods. Overall, the NDVI in Zhejiang Province showed a steady upward trajectory throughout the study period, reflecting gradual improvements in regional ecological and environmental quality. 4.2. Spatial Distribution and Temporal Segmentation of NDVI Trends in Zhejiang Province Figure 4 presents the spatial distribution of NDVI trends and classified trend types in Zhejiang Province from 2001 to 2020, based on Sen’s slope estimation and the Mann–Kendall (MK) significance test. Both trend maps and pie chart statistics across five distinct time periods (2001–2005, 2006–2010, 2011–2015, 2016–2020, and the full period 2001–2020) reveal pronounced spatio-temporal heterogeneity in NDVI trends. During the four consecutive 5-year periods, most areas were characterized by either ‘slight improvement’ or ‘slight degradation’, indicating frequent NDVI fluctuations and alternating patterns of improvement and decline. From 2001 to 2005, 63.0% of the area showed slight improvement, while 30.0% exhibited slight degradation. In 2006–2010, the proportions were nearly equal, with 46.4% and 47.4% respectively. Between 2011 and 2015, slight improvement increased to 56.2%, whereas in 2016–2020, slight degradation rose to 56.8%, suggesting a slowdown in NDVI improvement and partial ecological deterioration in some areas. Over the long-term period (2001–2020), the trends were more pronounced: regions with significant improvement accounted for 59.4%, and those with slight improvement made up 21.8%, together comprising over 80% of the total area. This indicates a stable and substantial upward trend in NDVI across most of Zhejiang Province, reflecting continued ecological enhancement. Meanwhile, 5.9% of the area experienced significant degradation, primarily located in urban fringes, industrial zones, or topographically fragmented regions, likely due to intense anthropogenic disturbances or drastic land cover changes. In summary, vegetation cover across Zhejiang Province exhibited a predominantly "significant improvement" spatial pattern during the study period, suggesting sustained ecological progress. However, localized areas of degradation warrant attention in future ecological management and policy planning. 4.3. Spatial Autocorrelation and Clustering Patterns of NDVI in Zhejiang Province Figure 5 illustrates the spatial autocorrelation characteristics of NDVI in Zhejiang Province for the years 2001, 2005, 2010, 2015, and 2020. By combining LISA cluster maps, Moran scatter plots, and pie chart statistics, the spatial clustering patterns and their temporal evolution are systematically revealed. Moran's I values across all years indicate a statistically significant and positive spatial autocorrelation (p < 0.01), suggesting a strong and consistent spatial clustering of NDVI. Specifically, Moran’s I values were 0.7892 (2001), 0.8005 (2005), 0.8055 (2010), 0.8114 (2015), and 0.8191 (2020), showing a gradual upward trend. This indicates a strengthening of spatial aggregation and increasing spatial stability of vegetation cover over time in Zhejiang Province. In terms of LISA clustering types, “High–High” and “Low–Low” clusters were dominant. “High–High” areas were mainly located in mountainous or ecologically functional zones with good vegetation continuity, such as the western and southern hilly regions. In contrast, “Low–Low” areas were typically concentrated in zones of urban expansion, industrial parks, or intensively developed plains, including the core areas of the Yangtze River Delta urban agglomeration. In 2001, “High–High” clusters accounted for 35.37% of the province, and by 2020 remained relatively stable at approximately 35.68%, suggesting a spatially stable distribution of high-quality ecological areas. “Low–Low” clusters consistently accounted for around 23%, with slight fluctuations, remaining the second most common cluster type. In addition, pie charts indicate that “Not Significant” regions accounted for 39%–41% of the area in each year, implying that NDVI variation in these zones lacked clear spatial structure—potentially due to micro-scale disturbances or heterogeneous topography. Heterogeneous clusters such as “High–Low” and “Low–High” were minor (each < 1%) but may signal localized human disturbances or ecological edge zones, which warrant further investigation at finer spatial scales. Overall, NDVI in Zhejiang Province from 2001 to 2020 exhibited a steadily strengthening pattern of positive spatial autocorrelation. While the overall clustering structure remained stable, some local heterogeneity persisted, reflecting an overall improving ecological environment accompanied by emerging risks of local degradation and uneven ecological development. 4.4. Driving Factors of NDVI Spatial Differentiation in Zhejiang Province Figure 6 presents the results of the factor detector module of the Geodetector model, illustrating the explanatory capacity (Q-values) of different driving factors on the spatial differentiation of NDVI in Zhejiang Province. Overall, the Q-values varied considerably among the factors, indicating differing levels of influence on NDVI spatial distribution. ELE exhibited the highest Q-value (0.64), indicating that topography played a key role in determining vegetation spatial distribution. This was closely followed by NTL and SLO, with Q-values of 0.63 and 0.57, respectively. These results highlight the combined influence of human activity intensity and terrain variability on NDVI distribution. LULC also demonstrated strong explanatory power, with a Q-value of 0.56, indicating that changes in land use structure significantly influenced vegetation cover. Socio-economic factors such as GDP (0.49) and population density (POP, 0.39) further contributed to NDVI variation, underscoring the considerable role of anthropogenic activities in modulating vegetation patterns. In contrast, climatic factors such as precipitation (PRE, 0.25) and temperature (TEM, 0.28) showed relatively low Q-values, suggesting that in this subtropical region with generally favorable hydrothermal conditions, climate variability had a comparatively limited impact on NDVI spatial differentiation. Among all factors, aspect (ASP, 0.04) and photosynthetically active radiation (PAR, 0.05) had the weakest explanatory power, indicating a minimal correlation with NDVI spatial distribution. In summary, topographic conditions—particularly elevation and slope—alongside indicators of human activity such as night-time lights and land use types, were identified as the primary drivers of NDVI spatial variation in Zhejiang Province. The statistical significance of all Q-values was verified using F-tests, with p-values equal to 0, indicating that the explanatory powers of all driving factors were statistically significant at the 0.01 level. Figure 7 illustrates the results of the interaction detector module of the Geodetector model, highlighting the joint explanatory power of different factor combinations on the spatial differentiation of NDVI in Zhejiang Province. In the figure, higher Q-values indicate stronger explanatory strength, with a color gradient ranging from light yellow to deep red representing increasing interaction intensity. Overall, the majority of factor combinations exhibited higher Q-values than their respective individual contributions, indicating the presence of interaction effects—either enhancement or nonlinear enhancement—between variables. Among all combinations, the interaction between NTL and LULC was the most prominent, with a Q-value of 0.72. This greatly exceeded the Q-values of NTL (0.63) and LULC (0.56) when considered independently, suggesting a strong nonlinear enhancement. This result indicates that the combined influence of human activity intensity and land use structure exerts a substantial effect on the spatial pattern of NDVI. Similarly, high Q-values were observed for other combinations, such as NTL and SLO, POP and LULC, GDP and ELE, and SOIL and LULC—all around 0.70. These findings further demonstrate the crucial role played by interactions between natural topographic features and socio-economic activities in shaping vegetation distribution. In contrast, combinations involving ASP and PAR exhibited relatively low Q-values, indicating weak interaction effects and limited joint explanatory power for NDVI spatial variability. In summary, the results underscore the significant compound effects of natural and anthropogenic factors in shaping NDVI spatial patterns in this subtropical region of human–environment interaction. These findings highlight the importance of integrated, multi-factor approaches in ecological environment management and land use planning. Figure 8 presents the results of the ecological detector module, which tests whether the impacts of different driving factors on NDVI vary significantly. In the figure, an “N” denotes that no significant difference (p ≥ 0.05) was detected between any pair of factors at the 95% confidence level. As shown, none of the factor pairs passed the significance test, indicating that the differences in their explanatory power for NDVI spatial differentiation were statistically insignificant. Although the Q-values of individual factors varied (with ELE and NTL ranking higher and ASP and PAR lower), these differences were not statistically significant. This outcome may be attributed to factors such as classification granularity, spatial scale, or sample distribution within the study area. It also suggests that in a region like Zhejiang—characterized by both natural and anthropogenic influences—the explanatory strength of different factor types tends to be relatively balanced. That is, the impact of individual factors may not differ significantly on their own. These findings support the conclusions of the interaction detector analysis, which highlighted that the spatial heterogeneity of NDVI is primarily driven not by the significance of individual factors, but by their synergistic and compounded interactions. Thus, the ecological detector results reinforce the importance of considering multi-factor interactions in explaining vegetation dynamics. Figure 9 presents the results of the risk detector analysis, illustrating the mean NDVI values across different strata of each driving factor. This analysis enables the identification of areas with high vegetation cover as well as zones potentially at risk of degradation. Overall, natural factors such as ELE, SLO, and PRE exhibit a significant positive correlation with NDVI. As the categorical level of these factors increases, the mean NDVI value consistently rises. For instance, NDVI increases from 0.41 to 0.72 across ascending elevation classes, indicating that higher-elevation regions tend to have better vegetation cover. Similar trends are observed for slope and precipitation, suggesting that favorable topographic and moisture conditions play a critical role in sustaining vegetation. In contrast, socio-economic factors including GDP, POP, and NTL generally show a negative correlation with NDVI. Higher levels of human activity correspond to lower NDVI values. For example, as GDP categories increase from low to high, the mean NDVI declines from 0.70 to 0.41; NTL levels show an even sharper decrease from 0.70 to 0.27, highlighting the high ecological degradation risk in areas of intense economic development and urbanization. Moreover, LULC shows substantial differentiation: high NDVI values are concentrated in categories 2 and 3 (e.g. forest and cropland), while built-up areas and bare land are associated with significantly lower NDVI values (approximately 0.30–0.40). ASP and PAR display only weak associations with NDVI, showing minimal variation across their respective strata. In summary, the risk detector results reveal clear NDVI response patterns across different factor categories and highlight that high-risk zones with low vegetation cover are primarily concentrated in areas subject to intense anthropogenic disturbance. These regions should be prioritised in ecological protection and restoration initiatives. 5. Discussion 5.1. Vegetation Greening Patterns and Spatial Persistence in Zhejiang This study identified a significant greening trend in Zhejiang Province between 2001 and 2020, with NDVI increasing at a rate of 0.0025/year (Fig. 3 a), and over 80% of the area showing either slight or significant improvement in vegetation cover (Fig. 4 e). The most notable increases were observed in the early years of the study period, such as 2002 and 2011 (Fig. 3 b), likely reflecting the cumulative effects of afforestation policies and rural reforestation programs(D. Zhang 2019 ). Spatially, persistent “High–High” NDVI clusters were concentrated in the forested southwest (e.g., Lishui, Quzhou), while “Low–Low” clusters remained in urbanized areas such as Hangjiahu Plain (Fig. 5 ), illustrating enduring urban–rural ecological disparities. These spatial clusters became increasingly stable over time, as evidenced by the steady rise in Moran’s I values from 0.79 to 0.82 (Fig. 5 ), suggesting spatial reinforcement of vegetation patterns. While these results support the effectiveness of national ecological policies and regional urban greening efforts(L. Wang 2010 ), years with NDVI decline (e.g., 2010, 2014) hint at vulnerability to short-term climate extremes or construction-induced land conversion(Abed et al. 2024 ). Similar regional trends have been reported in eastern China and the Yangtze River Delta, where ecological improvement coexists with urban expansion(Li et al. 2022 ). Our findings extend this understanding by explicitly linking NDVI gains to their spatial structures and dynamics. 5.2. Interactive Drivers of NDVI Differentiation and Risk Distribution NDVI spatial heterogeneity was jointly shaped by terrain and anthropogenic factors, with elevation (Q = 0.64), slope (Q = 0.57), and LULC type (Q = 0.56) being the dominant natural variables(B. Guo et al. 2018 ), and nighttime lights (Q = 0.63) and GDP (Q = 0.49) standing out among human indicators (Fig. 6 ). This indicates that even in ecologically favorable regions, urban development intensity can offset vegetation potential(Zhifang Wang et al. 2021 ). Importantly, the interaction detector revealed nonlinear enhancement in explanatory power for combinations like NTL × LULC (Q = 0.72) and NTL × slope (Fig. 7 ), showing that vegetation change arises from complex coupling between topographic constraints and socio-economic pressures. The risk detector results reinforce this interpretation, showing that high NDVI values are concentrated in high-elevation, low-disturbance areas(Nolè et al. 2018 ), while the lowest values are found in zones of intensive human activity (Fig. 9 ). Notably, NDVI declined from 0.70 to 0.27 across increasing NTL levels, indicating severe degradation risk in urbanized areas. Similar spatial interaction effects have been reported in studies on urban fringe forests(Y. Hu et al. 2009 ), emphasizing that vegetation in urbanizing subtropical landscapes is shaped more by compound drivers than by single factors. Our results provide further evidence that ecological responses are highly sensitive to the configuration of human–natural interactions rather than to isolated landscape elements. 5.3. Model Limitations and Future Research Directions While this study offers a robust spatial diagnostic of vegetation dynamics using NDVI and Geodetector-based modeling, several limitations remain. NDVI, though widely used, is subject to saturation in dense canopies(Ni Guo et al. 2007 ) and may fail to capture degradation beneath greening trends. Moreover, this analysis focuses on ecosystem supply without addressing service demand, stakeholder heterogeneity, or land use conflicts(D. Wang et al. 2022 ). The absence of temporal lag modeling may also obscure short-term effects of policy or land cover change. Future research should integrate multiple vegetation indicators (e.g., EVI, SIF)(Leng et al. 2022 ) with finer spatial and temporal resolution to enhance detection accuracy(Howey et al. 2014 ). Coupling remote sensing data with policy timelines, population shifts, and land use projections will better capture system dynamics. Additionally, the use of spatially explicit models to evaluate trade-offs and synergies among ecosystem services—particularly at multiple spatial scales—can offer actionable insights into ecological zoning and adaptive governance(Obiang Ndong et al. 2021 ). As vegetation dynamics in Zhejiang reflect a complex balance between conservation and development, future work must move toward a socio-ecological systems perspective to inform sustainable land management. 6. Conclusion This study focused on Zhejiang Province, a typical subtropical area in southeastern China, aiming to systematically assess the spatiotemporal dynamics and primary driving forces of vegetation cover from 2001 to 2020. Based on MODIS NDVI data and an integrated methodological framework—including Sen’s Slope trend estimation, Mann–Kendall significance test, Moran’s I spatial autocorrelation analysis, and the Geodetector model—the following key conclusions were drawn: (1) During the two decades of study, the annual mean NDVI in Zhejiang Province exhibited a statistically significant increasing trend (Sen’s slope = 0.0025, p < 0.01), with positive annual growth rates in most years, reflecting a favorable trend in vegetation recovery. Long-term trend analysis revealed that areas classified as “significantly improved” accounted for 59.4%, indicating an overall positive trajectory in ecosystem quality. (2) The Moran’s I index consistently stayed above 0.79 throughout the entire study period, demonstrating a strong level of spatial autocorrelation. LISA cluster maps further revealed that “High–High” clusters were mainly located in forested mountainous regions, whereas “Low–Low” clusters were chiefly found in highly urbanized zones. Although the overall spatial pattern remained largely stable, localized degradation hotspots were also identified. (3) The factor detector results showed that elevation, slope, land use type, and nighttime light possessed strong explanatory power (Q > 0.55) for NDVI spatial differentiation. These results emphasize the dominant influence of natural terrain and human activity intensity in shaping vegetation patterns. In contrast, factors such as aspect and PAR had weak explanatory power, while the impact of climatic factors was relatively limited due to climatic homogeneity in the region. (4) Most combinations of driving factors exhibited enhancement or nonlinear enhancement effects. Interactions such as NTL × LULC and SLO × NTL had Q-values exceeding 0.70, indicating that the combined effects of natural and socio-economic factors were considerably stronger than any single factor alone. (5) The risk detector revealed that NDVI values declined significantly in high categories of population density, GDP, and nighttime light, identifying urban fringe areas and industrial clusters as potential ecological degradation hotspots. It is recommended to implement region-specific ecological protection policies, strengthen dynamic monitoring, and carry out targeted restoration efforts in these high-risk areas. In conclusion, this study elucidated the spatiotemporal dynamics and spatial structure of NDVI in Zhejiang Province, while identifying key driving forces and their interaction mechanisms. The findings offer important theoretical insights and practical implications. Future research should incorporate multi-scale spatiotemporal datasets, integrate remote sensing with in situ observations, and adopt advanced approaches such as structural equation modeling to further explore the coupling mechanisms of human–environment systems and support scientific decision-making for sustainable regional ecological development. Declarations Conflict of interest The authors declare no competing interests. Funding The authors did not receive support from any organization for the submitted work. Author Contribution K.W. conceived and designed the study, performed methodology, formal analysis, and drafted the original manuscript.H.Y. supervised the work and revised the manuscript critically for important intellectual content.W.J. contributed to methodology and investigation, and assisted in manuscript review and editing.N.L. curated the data and prepared the visualizations.J.C. provided resources and contributed to visualization.All authors read and approved the final manuscript. Data Availability The datasets analyzed during the current study are publicly available in the NASA LP DAAC (https://lpdaac.usgs.gov/) and the China Meteorological Data Service Center (http://data.cma.cn/). References Abed, Salwan Ali, Bijay Halder, and Zaher Mundher Yaseen. 2024. Investigation of the decadal unplanned urban expansion influenced surface urban heat island study in the Mosul metropolis. Urban Climate 54: 101845. https://doi.org/10.1016/j.uclim.2024.101845. Almazroui, Mansour, and Zekâi Şen. 2020. Trend Analyses Methodologies in Hydro-meteorological Records. Earth Systems and Environment 4: 713–738. https://doi.org/10.1007/s41748-020-00190-6. Beck, Pieter S.A., Clement Atzberger, Kjell Arild Høgda, Bernt Johansen, and Andrew K. Skidmore. 2006. Improved monitoring of vegetation dynamics at very high latitudes: A new method using MODIS NDVI. Remote Sensing of Environment 100: 321–334. https://doi.org/10.1016/j.rse.2005.10.021. Bikeko, Samuel Shibeshi, and Venkatesham E. 2024. Land use land cover change as a casual factor for climate variability and trends in the Bilate Watershed, Ethiopia. Environmental Monitoring and Assessment 196: 1250. https://doi.org/10.1007/s10661-024-13435-y. Chen, Hui, Hongxing Chen, Xiaoyun Huang, Song Zhang, Tengbing He, and Zhenran Gao. 2024. Landscape ecological risk assessment and driving factor analysis in southwest china. Scientific Reports 14: 23208. https://doi.org/10.1038/s41598-024-74506-1. Chen, Yun, Peter Taylor, Susan Cuddy, Shahriar Wahid, Dave Penton, and Fazlul Karim. 2024. Inferring vegetation response to drought at multiscale from long-term satellite imagery and meteorological data in Afghanistan. Ecological Indicators 158: 111567. https://doi.org/10.1016/j.ecolind.2024.111567. Cumming, Graeme S., Andreas Buerkert, Ellen M. Hoffmann, Eva Schlecht, Stephan von Cramon-Taubadel, and Teja Tscharntke. 2014. Implications of agricultural transitions and urbanization for ecosystem services. Nature 515. Nature Publishing Group: 50–57. https://doi.org/10.1038/nature13945. Dong, Dejin, Jianbo Shen, Daohong Gong, Tianxu Sun, Jiahe Chen, and Yuichiro Fujioka. 2025. Research Trends in Vegetation Spatiotemporal Dynamics and Driving Forces: A Bibliometric Analysis (1987–2024). Forests 16. Multidisciplinary Digital Publishing Institute: 588. https://doi.org/10.3390/f16040588. Gong, Daohong, Dejin Dong, Huaqiang Du, Yufeng Zhou, Sang Fu, and Yuichiro Fujioka. 2025. Spatiotemporal coupling mechanisms and driving forces of ecosystem services and human activity from a multidimensional perspective. International Journal of Digital Earth 18: 2512061. https://doi.org/10.1080/17538947.2025.2512061. Guan, Yanlong, Hongwei Lu, Chuang Yin, Yuxuan Xue, Yelin Jiang, Yu Kang, Li He, and Janne Heiskanen. 2020. Vegetation response to climate zone dynamics and its impacts on surface soil water content and albedo in China. Science of The Total Environment 747: 141537. https://doi.org/10.1016/j.scitotenv.2020.141537. Guo, Bing, Fang Han, and Lin Jiang. 2018. An Improved Dimidiated Pixel Model for Vegetation Fraction in the Yarlung Zangbo River Basin of Qinghai-Tibet Plateau. Journal of the Indian Society of Remote Sensing 46: 219–231. https://doi.org/10.1007/s12524-017-0692-8. Guo, Han, Yingjun Sun, Qi Wang, Xvlu Wang, and Liguo Zhang. 2023. Construction of Greenspace Landscape Ecological Network Based on Resistance Analysis of GeoDetector in Jinan. Stochastic Environmental Research and Risk Assessment 37: 651–663. https://doi.org/10.1007/s00477-022-02296-x. Guo, Yujing, Lirong Cheng, Aizhong Ding, Yumin Yuan, Zhengyan Li, Yizhe Hou, Liangsuo Ren, and Shurong Zhang. 2024. Geodetector model-based quantitative analysis of vegetation change characteristics and driving forces: A case study in the Yongding River basin in China. Int. J. Appl. Earth Obs. Geoinformation 132: 104027. https://doi.org/10.1016/j.jag.2024.104027. Howey, Meghan C.L., Michael Palace, Crystal H. McMichael, and Bobby Braswell. 2014. Moderate-Resolution Remote Sensing and Geospatial Analyses of Microclimates, Mounds, and Maize in the Northern Great Lakes. Advances in Archaeological Practice 2: 195–207. https://doi.org/10.7183/2326-3768.2.3.195. Hu, Lulu, Xiaojun Xu, Juzhong Wang, and Huaixing Xu. 2023. Individual tree crown width detection from unmanned aerial vehicle images using a revised local transect method. Ecological Informatics 75: 102086. https://doi.org/10.1016/j.ecoinf.2023.102086. Hu, Yonghong, Gensuo Jia, and Huadong Guo. 2009. Linking primary production, climate and land use along an urban–wildland transect: a satellite view. Environmental Research Letters 4: 044009. https://doi.org/10.1088/1748-9326/4/4/044009. Huang, Zihao, Huaqiang Du, Fangjie Mao, Xuejian Li, Guomo Zhou, Jiaqian Sun, Yanxin Xu, Jie Xuan, Yagang Lu, and Lei Huang. 2024. Assessing the impact of land use and cover change on above-ground carbon storage in subtropical forests: a case study of Zhejiang Province, China. Geo-spatial Information Science . Taylor & Francis: 1–27. Leng, Song, Alfredo Huete, Jamie Cleverly, Sicong Gao, Qiang Yu, Xianyong Meng, Junyu Qi, Rongrong Zhang, and Qianfeng Wang. 2022. Assessing the Impact of Extreme Droughts on Dryland Vegetation by Multi-Satellite Solar-Induced Chlorophyll Fluorescence. Remote Sensing 14: 1581. https://doi.org/10.3390/rs14071581. Li, Xin, Bin Fang, Mengru Yin, Tao Jin, and Xin Xu. 2022. Multi-Dimensional Urbanization Coordinated Evolution Process and Ecological Risk Response in the Yangtze River Delta. Land 11: 723. https://doi.org/10.3390/land11050723. Lin, Min, Lizhu Hou, Zhiming Qi, and Li Wan. 2022. Impacts of climate change and human activities on vegetation NDVI in China’s Mu Us Sandy Land during 2000–2019. Ecological Indicators 142: 109164. https://doi.org/10.1016/j.ecolind.2022.109164. Liu, Chenli, Wenlong Li, Wenying Wang, Huakun Zhou, Tiangang Liang, Fujiang Hou, Jing Xu, and Pengfei Xue. 2021. Quantitative spatial analysis of vegetation dynamics and potential driving factors in a typical alpine region on the northeastern Tibetan Plateau using the Google Earth Engine. CATENA 206: 105500. https://doi.org/10.1016/j.catena.2021.105500. Liu, Meiling, Xiangnan Liu, Ling Wu, Yibo Tang, Yu Li, Yaqi Zhang, Lu Ye, and Biyao Zhang. 2021. Establishing forest resilience indicators in the hilly red soil region of southern China from vegetation greenness and landscape metrics using dense Landsat time series. Ecological Indicators 121: 106985. https://doi.org/10.1016/j.ecolind.2020.106985. Mao, Fangjie, Huaqiang Du, Guomo Zhou, Junlong Zheng, Xuejian Li, Yanxin Xu, Zihao Huang, and Shiyan Yin. 2022. Simulated net ecosystem productivity of subtropical forests and its response to climate change in Zhejiang Province, China. Science of The Total Environment 838: 155993. https://doi.org/10.1016/j.scitotenv.2022.155993. Ni Guo, Xiaoping Wang, Dihua Cai, and Jia Yang. 2007. Comparison and evaluation between MODIS vegetation indices in Northwest China. In 2007 IEEE International Geoscience and Remote Sensing Symposium , 3366–3369. Barcelona, Spain: IEEE. https://doi.org/10.1109/IGARSS.2007.4423566. Nolè, Angelo, Angelo Rita, Agostino Maria Silvio Ferrara, and Marco Borghetti. 2018. Effects of a large-scale late spring frost on a beech (Fagus sylvatica L.) dominated Mediterranean mountain forest derived from the spatio-temporal variations of NDVI. Annals of Forest Science 75: 83. https://doi.org/10.1007/s13595-018-0763-1. Obiang Ndong, Gregory, Jean Villerd, Isabelle Cousin, and Olivier Therond. 2021. Using a multivariate regression tree to analyze trade-offs between ecosystem services: Application to the main cropping area in France. Science of The Total Environment 764: 142815. https://doi.org/10.1016/j.scitotenv.2020.142815. Peng, Shouzhang, Yongxia Ding, Wenzhao Liu, and Zhi Li. 2019. 1km monthly temperature and precipitation dataset for China from 1901 to 2017. Earth System Science Data 11. Copernicus GmbH: 1931–1946. https://doi.org/10.5194/essd-11-1931-2019. Phung, D., C. Huang, S. Rutherford, C. Chu, X. Wang, M. Nguyen, N. H. Nguyen, C. M. Do, and T. H. Nguyen. 2015. Temporal and spatial patterns of diarrhoea in the Mekong Delta area, Vietnam. Epidemiology and Infection 143: 3488–3497. https://doi.org/10.1017/S0950268815000709. Qiao, Yu, Xueqiu Wang, Zhixuan Han, Mi Tian, Qiang Wang, Hui Wu, and Futian Liu. 2022. Geodetector based identification of influencing factors on spatial distribution patterns of heavy metals in soil: A case in the upper reaches of the Yangtze River, China. Applied Geochemistry 146: 105459. https://doi.org/10.1016/j.apgeochem.2022.105459. Qiu, Zhengyang, Daohong Gong, Mingxing Zhao, and Dejin Dong. 2025. Spatiotemporal Dynamics and Driving Mechanisms of Soil Conservation Services (SCS) in Zhejiang Province, China: Insights from InVEST Modeling and Machine Learning. Remote Sensing 17: 2865. https://doi.org/10.3390/rs17162865. Şan, Murat, Fatma Akçay, Nguyen Thi Thuy Linh, Murat Kankal, and Quoc Bao Pham. 2021. Innovative and polygonal trend analyses applications for rainfall data in Vietnam. Theoretical and Applied Climatology 144: 809–822. https://doi.org/10.1007/s00704-021-03574-4. Sun, Yiwen, Huiwen Zhan, Chao Gao, Hang Li, and Xianhua Guo. 2024. Spatial Syntactic Analysis and Revitalization Strategies for Rural Settlements in Ethnic Minority Areas: A Case Study of Shuanglang Town, China. Buildings 14: 2531. https://doi.org/10.3390/buildings14082531. Tang, Yan, Xiaojun Xu, Zhongsheng Zhou, Yiling Qu, and Yue Sun. 2021. Estimating global maximum gross primary productivity of vegetation based on the combination of MODIS greenness and temperature data. Ecological Informatics 63: 101307. https://doi.org/10.1016/j.ecoinf.2021.101307. Wang, Dan, Youjia Liang, Shouzhang Peng, Zhangcai Yin, and Jiejun Huang. 2022. Integrated assessment of the supply–demand relationship of ecosystem services in the Loess Plateau during 1992–2015. Ecosystem Health and Sustainability 8: 2130093. https://doi.org/10.1080/20964129.2022.2130093. Wang, Lijun. 2010. The changes of China’s environmental policies in the latest 30 years. Procedia Environmental Sciences 2: 1206–1212. https://doi.org/10.1016/j.proenv.2010.10.131. Wang, Ning, Caiyao Xu, and Fanbin Kong. 2022. Value Realization and Optimization Path of Forest Ecological Products—Case Study from Zhejiang Province, China. International Journal of Environmental Research and Public Health 19: 7538. https://doi.org/10.3390/ijerph19127538. Wang, Zhifang, Min Xu, Haowen Lin, Salman Qureshi, Ankang Cao, and Yujing Ma. 2021. Understanding the dynamics and factors affecting cultural ecosystem services during urbanization through spatial pattern analysis and a mixed-methods approach. Journal of Cleaner Production 279: 123422. https://doi.org/10.1016/j.jclepro.2020.123422. Wang, Ziyi, Haolong Chen, and Prasanna Divigalpitiya. 2024. How does ICT development in resource-exhausted cities promote the urban green transformation efficiency? Evidence from China. Sustainable Cities and Society 115. Elsevier BV: 105835. https://doi.org/10.1016/j.scs.2024.105835. Wu, Yue, Zexu Han, Auwalu Faisal Koko, Siyuan Zhang, Nan Ding, and Jiayang Luo. 2022. Analyzing the Spatio-Temporal Dynamics of Urban Land Use Expansion and Its Influencing Factors in Zhejiang Province, China. International Journal of Environmental Research and Public Health 19: 16580. https://doi.org/10.3390/ijerph192416580. Xu, Xiaojun, and Danna Chen. 2024. Estimating global annual gross primary production based on satellite-derived phenology and maximal carbon uptake capacity. Environmental Research 252: 119063. https://doi.org/10.1016/j.envres.2024.119063. Xu, Yanxin, Xuejian Li, Huaqiang Du, Fangjie Mao, Guomo Zhou, Zihao Huang, Di’en Zhu, Qi Chen, Chi Ni, and Keruo Guo. 2023. Spatiotemporal variation in bamboo Solar-induced chlorophyll fluorescence (SIF) in China based on the global Orbiting Carbon Observatory-2 (OCO-2) carbon satellite and study on the response to climate and terrain. GIScience & Remote Sensing 60: 2253952. https://doi.org/10.1080/15481603.2023.2253952. Xu, Yong, Yun-Gui Lu, Bin Zou, Ming Xu, and Yu-Xi Feng. 2024. Unraveling the enigma of NPP variation in Chinese vegetation ecosystems: The interplay of climate change and land use change. Science of The Total Environment 912: 169023. https://doi.org/10.1016/j.scitotenv.2023.169023. Yan, Junjie, Guangpeng Zhang, Hongbo Ling, and Feifei Han. 2022. Comparison of time-integrated NDVI and annual maximum NDVI for assessing grassland dynamics. Ecological Indicators 136: 108611. https://doi.org/10.1016/j.ecolind.2022.108611. Yang, Jie, and Xin Huang. 2021. The 30m annual land cover dataset and its dynamics in China from 1990 to 2019. Earth System Science Data 13: 3907–3925. https://doi.org/10.5194/essd-13-3907-2021. Zhang, Daowei. 2019. China’s forest expansion in the last three plus decades: Why and how? Forest Policy and Economics 98: 75–81. https://doi.org/10.1016/j.forpol.2018.07.006. Zhang, Lu, Jianxia Chang, Aijun Guo, Kai Zhou, Guibin Yang, and Dongjing Zou. 2024. Ecological drought evolution characteristics under different climatic regions in the Yangtze River basin. Journal of Hydrology 629: 130573. https://doi.org/10.1016/j.jhydrol.2023.130573. Zhang, Yan, Luoqi Zhang, Junyi Wang, Gaocheng Dong, and Yali Wei. 2023. Quantitative analysis of NDVI driving factors based on the geographical detector model in the Chengdu-Chongqing region, China. Ecological Indicators 155: 110978. https://doi.org/10.1016/j.ecolind.2023.110978. Zhao, Rui, Liping Zhan, Mingxing Yao, and Linchuan Yang. 2020. A geographically weighted regression model augmented by Geodetector analysis and principal component analysis for the spatial distribution of PM2.5. Sustainable Cities and Society 56: 102106. https://doi.org/10.1016/j.scs.2020.102106. Zheng, Junlong, Fangjie Mao, Huaqiang Du, Xuejian Li, Guomo Zhou, Luofan Dong, Meng Zhang, Ning Han, Tengyan Liu, and Luqi Xing. 2019. Spatiotemporal Simulation of Net Ecosystem Productivity and Its Response to Climate Change in Subtropical Forests. Forests 10: 708. https://doi.org/10.3390/f10080708. Zhu, Lijun, Jijun Meng, and Likai Zhu. 2020. Applying Geodetector to disentangle the contributions of natural and anthropogenic factors to NDVI variations in the middle reaches of the Heihe River Basin. Ecological Indicators 117: 106545. https://doi.org/10.1016/j.ecolind.2020.106545. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7593091","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":523903403,"identity":"1deff68f-83c0-49d7-84fc-5334ec08c9f4","order_by":0,"name":"Ke Wang","email":"","orcid":"","institution":"Forest Resources Monitoring Center for Zhejiang Province","correspondingAuthor":false,"prefix":"","firstName":"Ke","middleName":"","lastName":"Wang","suffix":""},{"id":523903404,"identity":"d1205a21-2a0f-45af-93e1-03641aa78288","order_by":1,"name":"Hongwen Yao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAsUlEQVRIiWNgGAWjYBACfv7mgw8/8Ejw8BOtRXLGsWRjCRkLGckGYrUYHMgxk+CxqbAxOEC0LQ1njA0kciR4jI8nb2D4UbGNsBZ+5rbCBwVnJHjMzjwrYOw5c5sYWw5vNpDsAWq5kWPAzNhGhBaDAwlmErz/gA6bQbyWFKD3gYEM9BCRWiCBDNQiAfTLQaL8Ao3KOnv+9uSND35UEKEFCSQQHzUILaTqGAWjYBSMghECACwFOrorIaInAAAAAElFTkSuQmCC","orcid":"","institution":"Forest Resources Monitoring Center for Zhejiang Province","correspondingAuthor":true,"prefix":"","firstName":"Hongwen","middleName":"","lastName":"Yao","suffix":""},{"id":523903405,"identity":"bd908cce-3e93-456c-8130-32254b60bf25","order_by":2,"name":"Wei Jin","email":"","orcid":"","institution":"Forest Resources Monitoring Center for Zhejiang Province","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Jin","suffix":""},{"id":523903406,"identity":"11b1632f-b357-4488-800c-238777b5e759","order_by":3,"name":"Nan Li","email":"","orcid":"","institution":"Forest Resources Monitoring Center for Zhejiang Province","correspondingAuthor":false,"prefix":"","firstName":"Nan","middleName":"","lastName":"Li","suffix":""},{"id":523903407,"identity":"10baa9dc-1d9f-44ad-8a49-29ebd8eb1bdd","order_by":4,"name":"Jun Chen","email":"","orcid":"","institution":"Forest Resources Monitoring Center for Zhejiang Province","correspondingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2025-09-11 14:53:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7593091/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7593091/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92941468,"identity":"8098b9dd-8f64-4547-a4e0-d7d059b65bd5","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18943488,"visible":true,"origin":"","legend":"","description":"","filename":"6manuscript.doc","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/7318886688955e8164901c5e.doc"},{"id":92941456,"identity":"f797732b-bacf-4211-9ddf-39946c5d65db","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6697,"visible":true,"origin":"","legend":"","description":"","filename":"e84098332a8249468eae478a5da35453.json","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/a9b6213d43f17ca6553615f8.json"},{"id":92942778,"identity":"c5012539-e708-4eef-82f3-c3861b171914","added_by":"auto","created_at":"2025-10-07 11:50:38","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":140029,"visible":true,"origin":"","legend":"","description":"","filename":"e84098332a8249468eae478a5da354531enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/66a5c8452f2c62811c6c8d5b.xml"},{"id":92941464,"identity":"e74e6d74-8ceb-4982-b7c5-3e0001949bc0","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4127380,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/0f90fc9858d9849efcabe86a.png"},{"id":92942351,"identity":"c3ef8cef-92d1-4f0d-a283-32bdf8e824da","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":622499,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/e9919c888ed9efabd47cda4e.png"},{"id":92942350,"identity":"0c9b40d2-f3d3-4a8c-96ac-804159e219f7","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":923199,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/5b891837b896bcd23a25ee9c.png"},{"id":92941480,"identity":"1a37df34-5d93-456e-8b26-6e0b4c8a7304","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":315998,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/889866bc358c857d3e667fcd.jpeg"},{"id":92942353,"identity":"12461eab-51c5-493c-ae14-d2ad33f76df8","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":493768,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/690a0b9e478591b260d8485c.png"},{"id":92941474,"identity":"6bbdb097-a7ec-4692-a836-aad97294f3ff","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7374403,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/d328a0966a079baa6a47fbcb.png"},{"id":92941483,"identity":"3ef26831-5c85-45df-b294-03d47005c040","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3702731,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/e7cc26762b9be19c2e14fd12.png"},{"id":92941473,"identity":"95fb392e-bc7c-4b7c-91c9-b6deb3a5f916","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":329319,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/631f5102ccdf780c1783e201.png"},{"id":92942365,"identity":"6646f145-d41a-4594-81d8-7051b80d9bb0","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":448119,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/153392ada9aa94dc43a481e0.png"},{"id":92941476,"identity":"a62819fc-7e56-4641-8df8-c5b69a5d9473","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":209817,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/6caffd80b421e7cd0554621f.png"},{"id":92942356,"identity":"d720d13f-c0bf-4981-9f57-126ce911465a","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":567234,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/0b3f7cf654cb14d1b6f845ad.png"},{"id":92941470,"identity":"321e4753-79dd-458e-b3a4-ab3ceca6afc4","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":253810,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/54e2a57cc5cbb73e3119e338.png"},{"id":92941478,"identity":"2fcd4335-bf4e-400b-9b62-5cd33d75b65b","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":244841,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/829507235f98020181f9025c.png"},{"id":92941469,"identity":"34ccd1cc-eb81-4fea-8863-43a5512f4524","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41802,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/c69d404fca170a3db260ff31.png"},{"id":92942354,"identity":"a010cf8b-3fd8-4f3a-a40c-fdc0a23f4f0a","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":142840,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/35a0b53111dba458304e8eec.png"},{"id":92942357,"identity":"08d29b2e-88a4-47db-96cc-b1096af7a1c5","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1717160,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/fe8c932088b3dd5b10f1bc7c.png"},{"id":92941484,"identity":"d94c3322-ec24-4041-b6c6-6702d005e90a","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":715162,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/612640eaae6571b15fdf2115.png"},{"id":92941472,"identity":"50589b2c-4568-4556-ba62-fba48fc17745","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":181666,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/16cc7b977b0f7b22a3b1acdf.png"},{"id":92941479,"identity":"3b568d28-5360-4495-9f3f-0dc4d9995c2c","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":179220,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/eaeb03a0c17da6cc938ad6ea.png"},{"id":92942355,"identity":"20857b56-b223-49d8-8e88-abb805e60283","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":139170,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/9a7543c1135eb80e98f4bda2.png"},{"id":92942358,"identity":"b1e8ff37-4b14-4e41-9a4f-bf77075ff242","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"xml","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":137839,"visible":true,"origin":"","legend":"","description":"","filename":"e84098332a8249468eae478a5da354531structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/6177c1bf154a16d150176ddd.xml"},{"id":92941482,"identity":"9bebf5e4-0e3a-43e5-8967-bdc257e2cf07","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"html","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":149592,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/be52ae13eb3582a59f86459e.html"},{"id":92941454,"identity":"f1968dbc-46b7-4320-8d57-82b4a4068a75","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":188004,"visible":true,"origin":"","legend":"\u003cp\u003eLocation and Geographical Characteristics of Zhejiang Province. (a) Geographic location of Zhejiang Province in China; (b) Elevation distribution map; (c) Land use and land cover (LULC) map of Zhejiang Province.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/bf972044974bac3ffb14f3ea.jpg"},{"id":92941453,"identity":"703ab6b4-5983-487a-9b93-c62a31059634","added_by":"auto","created_at":"2025-10-07 11:34:37","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":221602,"visible":true,"origin":"","legend":"\u003cp\u003eFramework of NDVI Spatiotemporal Analysis and Driving Mechanism Detection in Zhejiang Province (2001–2020).\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/ae4fbd41cf7e0a04715573a9.jpg"},{"id":92941455,"identity":"f362bc97-5bd1-4ae4-9d71-47a59bb390f7","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":104865,"visible":true,"origin":"","legend":"\u003cp\u003eTemporal Dynamics of NDVI in Zhejiang Province from 2001 to 2020. (a) Interannual trend of mean NDVI with linear regression and 95% confidence interval. (b) Annual NDVI change rate showing interannual fluctuations, including years of increase and decline.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/5df9ae6f36fa671e01db5951.jpg"},{"id":92942776,"identity":"6341449e-b135-4e87-9f8d-1a3569c0e892","added_by":"auto","created_at":"2025-10-07 11:50:38","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":202383,"visible":true,"origin":"","legend":"\u003cp\u003eNDVI Trend Classification and Proportions in Zhejiang Province during 2001–2020. (a–e) Spatial classification of NDVI trends in five periods: 2001–2005, 2006–2010, 2011–2015, 2016–2020, and the entire period of 2001–2020. (f–j) Corresponding proportions of NDVI trend categories, including significant improvement, slight improvement, stability, slight degradation, and severe degradation.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/1e2a2051623bd8922d487adf.jpg"},{"id":92943571,"identity":"34e01426-fc79-4c10-a35d-28e63149755f","added_by":"auto","created_at":"2025-10-07 11:58:38","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1740521,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial Autocorrelation Analysis of NDVI in Zhejiang Province from 2001 to 2020. The figure includes (a–e) LISA cluster maps, (f–j) Moran’s scatter plots, and (k–o) LISA pie charts for the years 2001, 2005, 2010, 2015, and 2020. The LISA clusters include High-High, Low-Low, High-Low, Low-High, and Not Significant, indicating different types of local spatial associations.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/06152613541f61cf7ec77b61.jpg"},{"id":92942347,"identity":"7bf37255-88a8-4768-89cd-033a3e9c1dba","added_by":"auto","created_at":"2025-10-07 11:42:38","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":75292,"visible":true,"origin":"","legend":"\u003cp\u003eExplanatory Power of Different Driving Factors on NDVI Spatial Differentiation Based on the Geodetector Factor Detector.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/81e8cf542c836ba9bbd27b7d.jpg"},{"id":92941460,"identity":"e42db825-3064-4498-bb8d-3dee37966d95","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":92942,"visible":true,"origin":"","legend":"\u003cp\u003eInteraction Effects of Driving Factors on the Spatial Differentiation of NDVI Based on GeoDetector Q Values.\u003c/p\u003e","description":"","filename":"7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/c8aa9d62b7a6aa01bd4f90b0.jpg"},{"id":92941462,"identity":"8f8aa56c-0ae1-4bba-9747-2cc8ed6f048b","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":92703,"visible":true,"origin":"","legend":"\u003cp\u003eStatistical Significance of Differences in NDVI Impact among Driving Factors Based on Ecological Detector Results.\u003c/p\u003e","description":"","filename":"8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/e38758c103b045d0ff9e83f2.jpg"},{"id":92941466,"identity":"1a8e6fff-9afe-4dbb-9e6a-dd5953875ba7","added_by":"auto","created_at":"2025-10-07 11:34:38","extension":"jpg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":167906,"visible":true,"origin":"","legend":"\u003cp\u003eMean NDVI Values under Different Categories of Driving Factors.\u003c/p\u003e","description":"","filename":"9.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/5da10de78cb8375ef0cdacdb.jpg"},{"id":102564060,"identity":"ae151f6f-048f-4f9f-9017-d5d21d0f2a51","added_by":"auto","created_at":"2026-02-13 05:10:32","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3739338,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7593091/v1/63267bd4-51be-45d3-a425-cd935814f828.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Vegetation Dynamics and Influencing Mechanisms in Zhejiang Province, a Typical Subtropical Region of China","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eAs a fundamental element of land-based ecosystems, vegetation contributes critically to regulating climate, driving water cycles, and enabling carbon sequestration, while also supporting ecological balance.(Yong Xu et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Vegetation cover not only reflects the structural and functional attributes of ecosystems, but is also widely regarded as a key indicator of ecological health and environmental quality. In recent years, accelerated urbanization, shifting land use patterns, and intensified impacts of climate change have led to notable spatiotemporal changes in vegetation cover at the global scale(Cumming et al. 2014). A systematic understanding of vegetation dynamics and their dominant driving factors is therefore essential for elucidating human\u0026ndash;environment interactions, supporting regional ecological management, and advancing sustainable development goals(Dong et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Gong et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eSubtropical regions are characterized by humid climates, rich biodiversity, and high ecosystem sensitivity, and thus represent critical areas for global ecological security(X. Xu and Chen \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; L. Zhang et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Qiu et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, these regions often face multiple pressures, including rapid urban expansion and industrial restructuring, which increasingly challenge ecosystem stability(Ziyi Wang et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Against the backdrop of global change, investigating the spatiotemporal heterogeneity and driving mechanisms of vegetation dynamics in subtropical zones is of great importance for enhancing ecosystem resilience and regulatory capacity.\u003c/p\u003e\u003cp\u003eIn recent years, advances in remote sensing and spatial analysis have enabled extensive investigations of vegetation dynamics at both global and regional scales(Beck et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Tang et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; L. Hu et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Among these, the Normalised Difference Vegetation Index (NDVI) has been widely employed to monitor vegetation changes (Lin et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Sen\u0026rsquo;s slope estimation and the Mann\u0026ndash;Kendall test are commonly used to identify long-term trends and their significance(Yanxin Xu et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), Moran\u0026rsquo;s I is applied to evaluate spatial autocorrelation and clustering patterns(M. Liu et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), while the Geodetector model has recently been adopted to quantify the influence and interaction of natural and anthropogenic factors (Y. Zhang et al. \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Although previous studies have produced a wealth of findings, further exploration is needed in typical subtropical regions to understand how multifactorial interactions drive vegetation pattern evolution, especially under conditions of spatial heterogeneity.\u003c/p\u003e\u003cp\u003eLocated in southeastern coastal China, Zhejiang Province features a typical subtropical monsoon climate, complex topography, and diverse ecosystem types(Mao et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). As a densely populated, economically developed, and ecologically sensitive region, it exhibits a dual pattern of urban expansion and ecological restoration in recent years, driven by rapid socioeconomic development and ongoing ecological civilization initiatives(Wu et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This interplay between disturbance and recovery makes Zhejiang a representative case for investigating the spatiotemporal dynamics of vegetation and their driving forces in subtropical regions(Zheng et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eIn this study, we use Zhejiang Province as the study area and employ MODIS NDVI remote sensing data from 2001 to 2020. An integrated analytical framework combining Sen\u0026rsquo;s slope trend estimation, Mann\u0026ndash;Kendall significance testing, Moran\u0026rsquo;s I spatial autocorrelation, and the Geodetector model is developed to assess vegetation change across three dimensions: temporal evolution, spatial distribution, and driving mechanisms. The objectives of this study are to: (1) identify long-term trends in NDVI and their significance; (2) reveal spatial clustering and heterogeneity patterns; and (3) quantify the dominant effects and interactions of natural and anthropogenic drivers. The findings aim to provide a scientific basis for ecological conservation and sustainable land use management in subtropical regions.\u003c/p\u003e"},{"header":"2. Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Study area\u003c/h2\u003e\u003cp\u003eZhejiang Province lies along China\u0026rsquo;s southeastern coastline, stretching between 118\u0026deg;01\u0026prime;\u0026ndash;123\u0026deg;10\u0026prime; E and 27\u0026deg;02\u0026prime;\u0026ndash;31\u0026deg;11\u0026prime; N(Huang et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). It is bordered by the East China Sea to the east, Fujian to the south, with Jiangxi and Anhui bordering its west, and Shanghai and Jiangsu to its north. The province covers a territorial coverage of about 105,500 km\u0026sup2;, featuring predominantly hilly and mountainous terrain, occupying nearly 70% of the province\u0026rsquo;s total surface. Plains are primarily found in the northern coastal belt and near key rivers. Zhejiang features a representative subtropical monsoon climate, featuring marked by clearly defined seasonal changes, warm and humid conditions, with mean annual temperatures between roughly 15\u0026deg;C and 18\u0026deg;C, and yearly rainfall ranging from 1100 mm to 2000 mm. The synchrony of rainfall and heat provides favorable conditions for the growth of various vegetation types. The province possesses abundant natural resources and supports a diverse range of ecosystems, including forests, wetlands, rivers, lakes, and coastal zones. In recent decades, fast-paced urban and industrial development has significantly altered the ecological landscape, with urban expansion and land-use changes imposing considerable pressure on natural ecosystems. Nevertheless, Zhejiang has placed great emphasis on ecological civilization, promoting the philosophy that \u0026ldquo; clear waters and verdant mountains are priceless treasures\u0026rdquo;(N. Wang et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). A series of ecological protection and restoration projects have been implemented, offering strong support for vegetation recovery and ecosystem quality improvement. Given its complex topography, diverse climatic conditions, and pronounced human activity, Zhejiang Province serves as a representative and well-founded case for studying the spatiotemporal dynamics of vegetation cover and its driving mechanisms in subtropical regions.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Data Sources\u003c/h2\u003e\u003cdiv id=\"Sec5\" class=\"Section3\"\u003e\u003ch2\u003e2.1.2 Land Cover Data\u003c/h2\u003e\u003cp\u003eLand use/land cover (LULC) data were obtained from the China CLCD product(Yang and Huang 2021). The dataset spans from 2001 to 2020 with a spatial resolution of 30 m. Within GEE, the data were clipped to the extent of Zhejiang Province and resampled to 1000 m to ensure spatial resolution consistency with other variables.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section3\"\u003e\u003ch2\u003e2.1.3 Climate Data\u003c/h2\u003e\u003cp\u003eTemperature (TEM) together with precipitation (PRE) data were obtained from the National Tibetan Plateau Scientific Data Center (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://data.tpdc.ac.cn\u003c/span\u003e\u003cspan address=\"http://data.tpdc.ac.cn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, accessed 8 November 2023)(Peng et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), with a spatial resolution of 1 km. The final datasets include precipitation by month (unit: 0.1 mm) and average monthly temperature (unit: 0.1\u0026deg;C) at 1 km resolution for 2001\u0026ndash;2020. These data have been validated against 496 independent meteorological stations and demonstrate high accuracy and reliability. We calculated the mean annual temperature and the cumulative yearly precipitation for the study area to represent climatic variables. Photosynthetically Active Radiation (PAR) data were acquired via the TerraClimate dataset(Y. Chen et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and processed through clipping and resampling to a spatial resolution of 1 km for the period 2001\u0026ndash;2020.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\u003ch2\u003e2.1.4 Nighttime Light Data\u003c/h2\u003e\u003cp\u003eNighttime light (NTL) data were obtained from Version 4 of the DMSP-OLS time series, which provides cloud-free composites based on all available archived data with smoothed resolution for each calendar year. These datasets were sourced from the U.S. Air Force Weather Agency and subsequently processed by NOAA\u0026rsquo;s National Centers for Environmental Information (NCEI). We extracted annual average NTL values for Zhejiang Province from 2001 to 2020 to represent human activity intensity(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.noaa.gov/\u003c/span\u003e\u003cspan address=\"https://www.noaa.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\u003ch2\u003e2.1.5 Population and GDP Data\u003c/h2\u003e\u003cp\u003ePopulation (POP) data were obtained from the WorldPop global population grid dataset(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.worldpop.org/\u003c/span\u003e\u003cspan address=\"https://www.worldpop.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and annual values were extracted using GEE to calculate yearly averages at 100 m resolution.\u003c/p\u003e\u003cp\u003eThe Gross Domestic Product (GDP) data were sourced from the global 1 km \u0026times; 1 km revised real GDP dataset, which addresses previous issues such as spatial\u0026ndash;temporal discontinuity and overestimated growth. This dataset adopts a top-down estimation approach. Annual mean GDP values (unit: yuan/km\u0026sup2;) were extracted for Zhejiang Province and resampled to a 1000 m resolution for analysis as economic drivers.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\u003ch2\u003e2.1.6 Topographic Variables\u003c/h2\u003e\u003cp\u003eTopographic elevation data were obtained from the SRTM dataset provided by the United States Geological Survey (USGS), with a spatial resolution of 30 m(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.earthdata.nasa.gov/data/instruments/srtm\u003c/span\u003e\u003cspan address=\"https://www.earthdata.nasa.gov/data/instruments/srtm\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Slope (SLO) and aspect (ASP) were derived from the elevation (ELE) data. All topographic variables were clipped to the study area and resampled to a 1000 m resolution.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec10\" class=\"Section3\"\u003e\u003ch2\u003e2.1.7 Soil Type Data\u003c/h2\u003e\u003cp\u003eSoil type (SOIL) data were obtained from the FAO HWSD_V2 dataset (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.fao.org/land-water/databases-and-software/hwsd/zh/\u003c/span\u003e\u003cspan address=\"https://www.fao.org/land-water/databases-and-software/hwsd/zh/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The FAO90 layer was selected to represent major soil mapping units. The data were clipped to the boundary of Zhejiang Province and resampled to ensure consistency with the resolution of other factors.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"3. Methods","content":"\u003cp\u003eThis study focused on Zhejiang Province and utilized MODIS NDVI remote sensing data spanning 2001 to 2020. A combination of spatial and geographical analysis methods was employed to examine the spatiotemporal patterns and driving factors behind subtropical vegetation cover across three key perspectives: temporal trends, spatial patterns, and driving mechanisms. First, Sen\u0026rsquo;s slope estimation together with the Mann\u0026ndash;Kendall test were jointly applied on the NDVI time series to assess both the direction and statistical significance of vegetation change over time. Second, global and local Moran\u0026rsquo;s I indices served to measure spatial autocorrelation characteristics and identify clustering patterns of vegetation variation. Finally, the Geodetector model was applied to quantify the explanatory power of both natural environmental and anthropogenic factors on vegetation dynamics, as well as to explore their interaction effects. By integrating these analytical approaches, this research seeks to offer a thorough understanding of the spatiotemporal heterogeneity of vegetation change and its dominant drivers in Zhejiang Province, thereby offering scientific support for regional ecological management and policy formulation. The specific methods are described as follows.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.1. Sen\u0026rsquo;s Slope Estimation\u003c/h2\u003e\u003cp\u003eSen\u0026rsquo;s slope estimation is a nonparametric approach commonly applied for assessing monotonic trends within time series data (Bikeko and E. 2024). It is especially effective for examining long-term variations in remote sensing-derived vegetation indices, such as NDVI (Yan et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This method derives the trend slope as the median value among all pairwise rates of change between different time points. It is robust to outliers and does not require the assumption of data normality, making it effective for reliably detecting consistent upward or downward trends present in the data. The fundamental computation procedure is as follows:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{ij}=\\frac{{X}_{j}-{X}_{i}}{j-i}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(1)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}_{i}\\)\u003c/span\u003e\u003c/span\u003e​ denotes the NDVI value in year i; n is the length of the time series; \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{ij}\\)\u003c/span\u003e\u003c/span\u003e refers to the rate of NDVI change from year i to year j; (j\u0026thinsp;\u0026minus;\u0026thinsp;i) is the time interval; and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{X}_{j}-{X}_{i}\\)\u003c/span\u003e\u003c/span\u003e​ is the NDVI change.\u003c/p\u003e\u003cp\u003eThe Sen\u0026rsquo;s Slope value is calculated as the median across all computed slopes, that is:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabb\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\beta\\:=\\text{m}\\text{e}\\text{d}\\text{i}\\text{a}\\text{n}\\left({\\beta\\:}_{ij}\\right)\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(2)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\beta\\:\\)\u003c/span\u003e\u003c/span\u003e represents the trend of the variable over the entire time series. In this research, the Sen\u0026rsquo;s Slope approach was employed to quantitatively assess the trend of vegetation cover change in Zhejiang Province based on MODIS NDVI products from 2001 to 2020. This was combined with the Mann-Kendall significance test to identify regions with significant changes, aiming to reveal the dynamic evolution characteristics of the ecosystem under the combined influence of natural and human factors.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.2. Mann\u0026ndash;Kendall (MK) Trend Test\u003c/h2\u003e\u003cp\u003eTo further evaluate the statistical significance of vegetation change trends identified through Sen\u0026rsquo;s slope estimation, the Mann\u0026ndash;Kendall (MK) trend test was employed in this study(Almazroui and Şen 2020). The MK test is a widely used non-parametric statistical technique applied in environmental and climate change studies(Guan et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). It is specifically designed to determine whether a time series demonstrates a statistically significant monotonic increase or decrease. The MK test operates without imposing distributional assumptions on the data, and it offers several advantages, including broad applicability, computational efficiency, and robustness to outliers(Şan et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The test works by comparing the values of all possible pairs in the time series and evaluating their relative magnitudes to construct a test statistic \u0026#119878;, which is defined as follows:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabc\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:S=\\underset{i=1}{\\overset{n-1}{?}}?\\underset{j=i+1}{\\overset{n}{?}}?\\text{s}\\text{g}\\text{n}?({x}_{j}-{x}_{i})\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(3)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere the sign function \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{s}\\text{g}\\text{n}?({x}_{j}-{x}_{i})\\)\u003c/span\u003e\u003c/span\u003e ) is defined as follows:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabd\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{s}\\text{g}\\text{n}?({x}_{j}-{x}_{i})=\\left\\{\\begin{array}{c}1,\\:\\:\\:\\:{x}_{j}-{x}_{i}\u0026gt;0\\\\\\:0,\\:\\:\\:\\:{x}_{j}-{x}_{i}=0\\\\\\:-1,\\:\\:\\:\\:{x}_{j}-{x}_{i}\u0026lt;0\\:\\:\\:\\:\\end{array}\\right.\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(4)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhen the sample size is large (e.g., n\u0026thinsp;\u0026ge;\u0026thinsp;8), the test statistic S approximately follows a normal distribution and can be used to test the significance level. The calculation formula is as follows:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabe\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Z=\\left\\{\\begin{array}{c}\\frac{S-1}{\\sqrt{\\text{V}\\text{a}\\text{r}?\\left(S\\right)}},\\:\\:\\:\\:S\u0026gt;0\\\\\\:0,\\:\\:\\:\\:S=0\\\\\\:\\frac{S+1}{\\sqrt{\\text{V}\\text{a}\\text{r}?\\left(S\\right)}},\\:\\:\\:\\:S\u0026lt;0\\:\\:\\:\\:\\end{array}\\right.\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(5)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{V}\\text{a}\\text{r}?\\left(S\\right)\\)\u003c/span\u003e\u003c/span\u003e is the variance of the test statistic S.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Spatial Autocorrelation Analysis Using Moran\u0026rsquo;s I\u003c/h2\u003e\u003cp\u003eTo further investigate the spatial distribution and dependency of NDVI variation, this study employed Moran\u0026rsquo;s I statistic for spatial autocorrelation analysis(Phung et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Moran\u0026rsquo;s I is a widely recognized indicator used to measure the spatial similarity of a variable across a geographic region. It includes two forms: global Moran\u0026rsquo;s I and local Moran\u0026rsquo;s I, which describe overall spatial autocorrelation patterns and localized clustering (C. Liu et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Global Moran\u0026rsquo;s I is used to assess whether a variable exhibits spatial clustering across the entire study area. It is calculated using the following formula:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabf\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:I=\\frac{n}{W}\\cdot\\:\\frac{\\sum\\:_{i=1}^{n}\\:\\sum\\:_{j=1}^{n}\\:{w}_{ij}({x}_{i}-\\stackrel{-}{x})({x}_{j}-\\stackrel{-}{x})}{\\sum\\:_{i=1}^{n}\\:({x}_{i}-\\stackrel{-}{x}{)}^{2}}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(6)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere n is the number of samples, x\u003csub\u003ei\u003c/sub\u003e and xⱼ represent the variable values at units i and j, respectively, x̄ is the mean of the variable, and w\u003csub\u003ei\u003c/sub\u003eⱼ is the spatial weight matrix (commonly represented by adjacency or inverse distance). \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:W=\\underset{i=1}{\\overset{n}{?}}?\\underset{j=1}{\\overset{n}{?}}?{w}_{ij\\:}\\)\u003c/span\u003e\u003c/span\u003eWhen I\u0026thinsp;\u0026gt;\u0026thinsp;0, it indicates that high or low values tend to cluster spatially (positive autocorrelation); when I\u0026thinsp;\u0026lt;\u0026thinsp;0, it reflects that dissimilar values are adjacent (negative autocorrelation); and when I\u0026thinsp;\u0026asymp;\u0026thinsp;0, it denotes a random spatial distribution. To further reveal the specific locations and types of spatial clusters, this study also introduces the Local Moran\u0026rsquo;s I, which is defined as:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabg\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{I}_{i}=({x}_{i}-\\stackrel{-}{x})\\sum\\:_{j}\\:{w}_{ij}({x}_{j}-\\stackrel{-}{x})\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(7)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eThis indicator also enables the identification of local spatial clustering types, such as high\u0026ndash;high and low\u0026ndash;low aggregations, along with high\u0026ndash;low and low\u0026ndash;high anomalies. These patterns are typically visualized using Local Indicators of Spatial Association (LISA) cluster maps. The Moran\u0026rsquo;s I method is valued for its intuitive results and strong capacity to detect spatial patterns. It finds broad application in spatial structure analysis of geographical phenomena and plays a significant role in research areas such as vegetation cover, land use, and urban expansion.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Geodetector\u003c/h2\u003e\u003cp\u003eTo determine the primary driving factors and how they interact in influencing vegetation cover change in Zhejiang Province, this study employed the Geodetector model(Y. Guo et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Geodetector serves as a statistical approach for identifying spatial stratified heterogeneity and uncovering underlying driving forces. The core concept is that if an explanatory variable significantly influences a dependent variable, their spatial patterns should display strong similarity. The framework comprises four core modules: factor detector, interaction detector, ecological detector, and risk detector(H. Chen et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This study mainly applied the factor and interaction detectors. The factor detector was applied to evaluate the explanatory strength of a single factor in explaining spatial variation in NDVI(Zhu et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). This is measured using a statistic known as the \u0026#119902;-value, which ranges from 0 to 1. A higher \u0026#119902;-value indicates a stronger explanatory power of the factor. The calculation is given by:\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Tabh\" border=\"1\"\u003e\u003ccolgroup cols=\"2\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=1-\\frac{\\sum\\:_{h=1}^{L}\\:{N}_{h}\\cdot\\:{\\sigma\\:}_{h}^{2}}{N\\cdot\\:{\\sigma\\:}^{2}}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e(8)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWhere h denotes the h-th subregion, L is the number of subregions, and N represent the number of samples in the subregion and the entire region, respectively, and and are the variances of NDVI within the subregion and the entire region, respectively. A q value of 0 indicates that the factor is unrelated to NDVI variation, while q\u0026thinsp;=\u0026thinsp;1 means the factor fully explains the NDVI variation. The interaction detector is used to analyze the type of interaction effect between two factors on NDVI variation.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe interaction detector is used to assess how combinations of two factors influence NDVI variation, identifying types of interaction such as enhancement (the combined effect exceeds that of either factor alone), nonlinear enhancement (the combined effect far exceeds individual contributions), or independence (the combined effect approximates the stronger single factor)(Sun et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Geodetector offers several advantages, including the absence of a linearity assumption, compatibility with heterogeneous data sources, and the ability to handle both categorical and continuous variables. It has been widely applied in studies on changes in land use, ecosystem dynamics, and human\u0026ndash;environment interactions (Zhao et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Within this research, the model was utilized to quantitatively determine the dominant roles and interaction effects of both natural factors (e.g., precipitation, temperature, topography) and socio-economic factors (e.g., GDP, population density, land use intensity) on vegetation change, thereby identifying the spatial variability of its driving processes.\u003c/p\u003e\u003cp\u003eThe ecological detector evaluates whether the means of NDVI differ significantly across various strata of a specific factor (H. Guo et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Fundamentally, it functions as an analysis of variance (ANOVA), using F-tests to assess statistical differences between groups. This is particularly effective for detecting NDVI response differences among environmental or administrative zones.\u003c/p\u003e\u003cp\u003eThe risk detector identifies categories of an explanatory variable associated with significantly higher or lower values of the dependent variable (e.g., high or low NDVI), enabling the detection of high-risk or high-probability zones(Qiao et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This is valuable for identifying sensitive areas or priority zones for ecological monitoring. Overall, the Geodetector model serves as a robust approach for exploring spatial mechanisms by which natural and human factors influence ecological variables. It is suitable for stratified or categorical spatial data, does not require linear assumptions, accommodates multiple explanatory variables, and is robust to multicollinearity.\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Results","content":"\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e4.1. Temporal Variation of NDVI in Zhejiang Province\u003c/h2\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the temporal variation and annual rate of change in the Normalized Difference Vegetation Index (NDVI) for Zhejiang Province from 2001 to 2020. Overall, the NDVI exhibits a statistically significant increasing trend. The linear regression results yield a slope of 0.0025 with a coefficient of determination \u0026#119877;2\u0026thinsp;=\u0026thinsp;0.6658, indicating a sustained improvement in vegetation cover over the study period. The green-shaded area in the figure represents the 95% confidence interval, further confirming the statistical significance of the upward trend. In terms of annual rate of change, most years show positive NDVI growth, with particularly notable increases in 2002 (+\u0026thinsp;4.9%), 2006 (+\u0026thinsp;3.8%), and 2011 (+\u0026thinsp;6.1%). However, several years also experienced declines in NDVI, including 2004 (\u0026ndash;1.2%), 2007 (\u0026ndash;0.9%), 2010 (\u0026ndash;4.6%), and 2014 (\u0026ndash;1.5%), suggesting potential impacts from climatic anomalies or human disturbances during those periods. Overall, the NDVI in Zhejiang Province showed a steady upward trajectory throughout the study period, reflecting gradual improvements in regional ecological and environmental quality.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003e4.2. Spatial Distribution and Temporal Segmentation of NDVI Trends in Zhejiang Province\u003c/h2\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents the spatial distribution of NDVI trends and classified trend types in Zhejiang Province from 2001 to 2020, based on Sen\u0026rsquo;s slope estimation and the Mann\u0026ndash;Kendall (MK) significance test. Both trend maps and pie chart statistics across five distinct time periods (2001\u0026ndash;2005, 2006\u0026ndash;2010, 2011\u0026ndash;2015, 2016\u0026ndash;2020, and the full period 2001\u0026ndash;2020) reveal pronounced spatio-temporal heterogeneity in NDVI trends.\u003c/p\u003e\u003cp\u003eDuring the four consecutive 5-year periods, most areas were characterized by either \u0026lsquo;slight improvement\u0026rsquo; or \u0026lsquo;slight degradation\u0026rsquo;, indicating frequent NDVI fluctuations and alternating patterns of improvement and decline. From 2001 to 2005, 63.0% of the area showed slight improvement, while 30.0% exhibited slight degradation. In 2006\u0026ndash;2010, the proportions were nearly equal, with 46.4% and 47.4% respectively. Between 2011 and 2015, slight improvement increased to 56.2%, whereas in 2016\u0026ndash;2020, slight degradation rose to 56.8%, suggesting a slowdown in NDVI improvement and partial ecological deterioration in some areas.\u003c/p\u003e\u003cp\u003eOver the long-term period (2001\u0026ndash;2020), the trends were more pronounced: regions with significant improvement accounted for 59.4%, and those with slight improvement made up 21.8%, together comprising over 80% of the total area. This indicates a stable and substantial upward trend in NDVI across most of Zhejiang Province, reflecting continued ecological enhancement. Meanwhile, 5.9% of the area experienced significant degradation, primarily located in urban fringes, industrial zones, or topographically fragmented regions, likely due to intense anthropogenic disturbances or drastic land cover changes.\u003c/p\u003e\u003cp\u003eIn summary, vegetation cover across Zhejiang Province exhibited a predominantly \"significant improvement\" spatial pattern during the study period, suggesting sustained ecological progress. However, localized areas of degradation warrant attention in future ecological management and policy planning.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003e4.3. Spatial Autocorrelation and Clustering Patterns of NDVI in Zhejiang Province\u003c/h2\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e illustrates the spatial autocorrelation characteristics of NDVI in Zhejiang Province for the years 2001, 2005, 2010, 2015, and 2020. By combining LISA cluster maps, Moran scatter plots, and pie chart statistics, the spatial clustering patterns and their temporal evolution are systematically revealed.\u003c/p\u003e\u003cp\u003eMoran's I values across all years indicate a statistically significant and positive spatial autocorrelation (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), suggesting a strong and consistent spatial clustering of NDVI. Specifically, Moran\u0026rsquo;s I values were 0.7892 (2001), 0.8005 (2005), 0.8055 (2010), 0.8114 (2015), and 0.8191 (2020), showing a gradual upward trend. This indicates a strengthening of spatial aggregation and increasing spatial stability of vegetation cover over time in Zhejiang Province.\u003c/p\u003e\u003cp\u003eIn terms of LISA clustering types, \u0026ldquo;High\u0026ndash;High\u0026rdquo; and \u0026ldquo;Low\u0026ndash;Low\u0026rdquo; clusters were dominant. \u0026ldquo;High\u0026ndash;High\u0026rdquo; areas were mainly located in mountainous or ecologically functional zones with good vegetation continuity, such as the western and southern hilly regions. In contrast, \u0026ldquo;Low\u0026ndash;Low\u0026rdquo; areas were typically concentrated in zones of urban expansion, industrial parks, or intensively developed plains, including the core areas of the Yangtze River Delta urban agglomeration. In 2001, \u0026ldquo;High\u0026ndash;High\u0026rdquo; clusters accounted for 35.37% of the province, and by 2020 remained relatively stable at approximately 35.68%, suggesting a spatially stable distribution of high-quality ecological areas. \u0026ldquo;Low\u0026ndash;Low\u0026rdquo; clusters consistently accounted for around 23%, with slight fluctuations, remaining the second most common cluster type.\u003c/p\u003e\u003cp\u003eIn addition, pie charts indicate that \u0026ldquo;Not Significant\u0026rdquo; regions accounted for 39%\u0026ndash;41% of the area in each year, implying that NDVI variation in these zones lacked clear spatial structure\u0026mdash;potentially due to micro-scale disturbances or heterogeneous topography. Heterogeneous clusters such as \u0026ldquo;High\u0026ndash;Low\u0026rdquo; and \u0026ldquo;Low\u0026ndash;High\u0026rdquo; were minor (each \u0026lt;\u0026thinsp;1%) but may signal localized human disturbances or ecological edge zones, which warrant further investigation at finer spatial scales.\u003c/p\u003e\u003cp\u003eOverall, NDVI in Zhejiang Province from 2001 to 2020 exhibited a steadily strengthening pattern of positive spatial autocorrelation. While the overall clustering structure remained stable, some local heterogeneity persisted, reflecting an overall improving ecological environment accompanied by emerging risks of local degradation and uneven ecological development.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003e4.4. Driving Factors of NDVI Spatial Differentiation in Zhejiang Province\u003c/h2\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e presents the results of the factor detector module of the Geodetector model, illustrating the explanatory capacity (Q-values) of different driving factors on the spatial differentiation of NDVI in Zhejiang Province. Overall, the Q-values varied considerably among the factors, indicating differing levels of influence on NDVI spatial distribution. ELE exhibited the highest Q-value (0.64), indicating that topography played a key role in determining vegetation spatial distribution. This was closely followed by NTL and SLO, with Q-values of 0.63 and 0.57, respectively.\u003c/p\u003e\u003cp\u003eThese results highlight the combined influence of human activity intensity and terrain variability on NDVI distribution. LULC also demonstrated strong explanatory power, with a Q-value of 0.56, indicating that changes in land use structure significantly influenced vegetation cover. Socio-economic factors such as GDP (0.49) and population density (POP, 0.39) further contributed to NDVI variation, underscoring the considerable role of anthropogenic activities in modulating vegetation patterns. In contrast, climatic factors such as precipitation (PRE, 0.25) and temperature (TEM, 0.28) showed relatively low Q-values, suggesting that in this subtropical region with generally favorable hydrothermal conditions, climate variability had a comparatively limited impact on NDVI spatial differentiation.\u003c/p\u003e\u003cp\u003eAmong all factors, aspect (ASP, 0.04) and photosynthetically active radiation (PAR, 0.05) had the weakest explanatory power, indicating a minimal correlation with NDVI spatial distribution. In summary, topographic conditions\u0026mdash;particularly elevation and slope\u0026mdash;alongside indicators of human activity such as night-time lights and land use types, were identified as the primary drivers of NDVI spatial variation in Zhejiang Province. The statistical significance of all Q-values was verified using F-tests, with p-values equal to 0, indicating that the explanatory powers of all driving factors were statistically significant at the 0.01 level.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e illustrates the results of the interaction detector module of the Geodetector model, highlighting the joint explanatory power of different factor combinations on the spatial differentiation of NDVI in Zhejiang Province. In the figure, higher Q-values indicate stronger explanatory strength, with a color gradient ranging from light yellow to deep red representing increasing interaction intensity. Overall, the majority of factor combinations exhibited higher Q-values than their respective individual contributions, indicating the presence of interaction effects\u0026mdash;either enhancement or nonlinear enhancement\u0026mdash;between variables.\u003c/p\u003e\u003cp\u003eAmong all combinations, the interaction between NTL and LULC was the most prominent, with a Q-value of 0.72. This greatly exceeded the Q-values of NTL (0.63) and LULC (0.56) when considered independently, suggesting a strong nonlinear enhancement. This result indicates that the combined influence of human activity intensity and land use structure exerts a substantial effect on the spatial pattern of NDVI. Similarly, high Q-values were observed for other combinations, such as NTL and SLO, POP and LULC, GDP and ELE, and SOIL and LULC\u0026mdash;all around 0.70. These findings further demonstrate the crucial role played by interactions between natural topographic features and socio-economic activities in shaping vegetation distribution.\u003c/p\u003e\u003cp\u003eIn contrast, combinations involving ASP and PAR exhibited relatively low Q-values, indicating weak interaction effects and limited joint explanatory power for NDVI spatial variability.\u003c/p\u003e\u003cp\u003eIn summary, the results underscore the significant compound effects of natural and anthropogenic factors in shaping NDVI spatial patterns in this subtropical region of human\u0026ndash;environment interaction. These findings highlight the importance of integrated, multi-factor approaches in ecological environment management and land use planning.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e presents the results of the ecological detector module, which tests whether the impacts of different driving factors on NDVI vary significantly. In the figure, an \u0026ldquo;N\u0026rdquo; denotes that no significant difference (p\u0026thinsp;\u0026ge;\u0026thinsp;0.05) was detected between any pair of factors at the 95% confidence level. As shown, none of the factor pairs passed the significance test, indicating that the differences in their explanatory power for NDVI spatial differentiation were statistically insignificant.\u003c/p\u003e\u003cp\u003eAlthough the Q-values of individual factors varied (with ELE and NTL ranking higher and ASP and PAR lower), these differences were not statistically significant. This outcome may be attributed to factors such as classification granularity, spatial scale, or sample distribution within the study area. It also suggests that in a region like Zhejiang\u0026mdash;characterized by both natural and anthropogenic influences\u0026mdash;the explanatory strength of different factor types tends to be relatively balanced. That is, the impact of individual factors may not differ significantly on their own.\u003c/p\u003e\u003cp\u003eThese findings support the conclusions of the interaction detector analysis, which highlighted that the spatial heterogeneity of NDVI is primarily driven not by the significance of individual factors, but by their synergistic and compounded interactions. Thus, the ecological detector results reinforce the importance of considering multi-factor interactions in explaining vegetation dynamics.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e presents the results of the risk detector analysis, illustrating the mean NDVI values across different strata of each driving factor. This analysis enables the identification of areas with high vegetation cover as well as zones potentially at risk of degradation.\u003c/p\u003e\u003cp\u003eOverall, natural factors such as ELE, SLO, and PRE exhibit a significant positive correlation with NDVI. As the categorical level of these factors increases, the mean NDVI value consistently rises. For instance, NDVI increases from 0.41 to 0.72 across ascending elevation classes, indicating that higher-elevation regions tend to have better vegetation cover. Similar trends are observed for slope and precipitation, suggesting that favorable topographic and moisture conditions play a critical role in sustaining vegetation.\u003c/p\u003e\u003cp\u003eIn contrast, socio-economic factors including GDP, POP, and NTL generally show a negative correlation with NDVI. Higher levels of human activity correspond to lower NDVI values. For example, as GDP categories increase from low to high, the mean NDVI declines from 0.70 to 0.41; NTL levels show an even sharper decrease from 0.70 to 0.27, highlighting the high ecological degradation risk in areas of intense economic development and urbanization.\u003c/p\u003e\u003cp\u003eMoreover, LULC shows substantial differentiation: high NDVI values are concentrated in categories 2 and 3 (e.g. forest and cropland), while built-up areas and bare land are associated with significantly lower NDVI values (approximately 0.30\u0026ndash;0.40). ASP and PAR display only weak associations with NDVI, showing minimal variation across their respective strata.\u003c/p\u003e\u003cp\u003eIn summary, the risk detector results reveal clear NDVI response patterns across different factor categories and highlight that high-risk zones with low vegetation cover are primarily concentrated in areas subject to intense anthropogenic disturbance. These regions should be prioritised in ecological protection and restoration initiatives.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003ch2\u003e5.1. Vegetation Greening Patterns and Spatial Persistence in Zhejiang\u003c/h2\u003e\u003cp\u003eThis study identified a significant greening trend in Zhejiang Province between 2001 and 2020, with NDVI increasing at a rate of 0.0025/year (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), and over 80% of the area showing either slight or significant improvement in vegetation cover (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). The most notable increases were observed in the early years of the study period, such as 2002 and 2011 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), likely reflecting the cumulative effects of afforestation policies and rural reforestation programs(D. Zhang \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Spatially, persistent \u0026ldquo;High\u0026ndash;High\u0026rdquo; NDVI clusters were concentrated in the forested southwest (e.g., Lishui, Quzhou), while \u0026ldquo;Low\u0026ndash;Low\u0026rdquo; clusters remained in urbanized areas such as Hangjiahu Plain (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), illustrating enduring urban\u0026ndash;rural ecological disparities.\u003c/p\u003e\u003cp\u003eThese spatial clusters became increasingly stable over time, as evidenced by the steady rise in Moran\u0026rsquo;s I values from 0.79 to 0.82 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), suggesting spatial reinforcement of vegetation patterns. While these results support the effectiveness of national ecological policies and regional urban greening efforts(L. Wang \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2010\u003c/span\u003e), years with NDVI decline (e.g., 2010, 2014) hint at vulnerability to short-term climate extremes or construction-induced land conversion(Abed et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Similar regional trends have been reported in eastern China and the Yangtze River Delta, where ecological improvement coexists with urban expansion(Li et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Our findings extend this understanding by explicitly linking NDVI gains to their spatial structures and dynamics.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e\u003ch2\u003e5.2. Interactive Drivers of NDVI Differentiation and Risk Distribution\u003c/h2\u003e\u003cp\u003eNDVI spatial heterogeneity was jointly shaped by terrain and anthropogenic factors, with elevation (Q\u0026thinsp;=\u0026thinsp;0.64), slope (Q\u0026thinsp;=\u0026thinsp;0.57), and LULC type (Q\u0026thinsp;=\u0026thinsp;0.56) being the dominant natural variables(B. Guo et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), and nighttime lights (Q\u0026thinsp;=\u0026thinsp;0.63) and GDP (Q\u0026thinsp;=\u0026thinsp;0.49) standing out among human indicators (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). This indicates that even in ecologically favorable regions, urban development intensity can offset vegetation potential(Zhifang Wang et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Importantly, the interaction detector revealed nonlinear enhancement in explanatory power for combinations like NTL \u0026times; LULC (Q\u0026thinsp;=\u0026thinsp;0.72) and NTL \u0026times; slope (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e), showing that vegetation change arises from complex coupling between topographic constraints and socio-economic pressures.\u003c/p\u003e\u003cp\u003eThe risk detector results reinforce this interpretation, showing that high NDVI values are concentrated in high-elevation, low-disturbance areas(Nol\u0026egrave; et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), while the lowest values are found in zones of intensive human activity (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e). Notably, NDVI declined from 0.70 to 0.27 across increasing NTL levels, indicating severe degradation risk in urbanized areas. Similar spatial interaction effects have been reported in studies on urban fringe forests(Y. Hu et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2009\u003c/span\u003e), emphasizing that vegetation in urbanizing subtropical landscapes is shaped more by compound drivers than by single factors. Our results provide further evidence that ecological responses are highly sensitive to the configuration of human\u0026ndash;natural interactions rather than to isolated landscape elements.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\u003ch2\u003e5.3. Model Limitations and Future Research Directions\u003c/h2\u003e\u003cp\u003eWhile this study offers a robust spatial diagnostic of vegetation dynamics using NDVI and Geodetector-based modeling, several limitations remain. NDVI, though widely used, is subject to saturation in dense canopies(Ni Guo et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2007\u003c/span\u003e) and may fail to capture degradation beneath greening trends. Moreover, this analysis focuses on ecosystem supply without addressing service demand, stakeholder heterogeneity, or land use conflicts(D. Wang et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The absence of temporal lag modeling may also obscure short-term effects of policy or land cover change.\u003c/p\u003e\u003cp\u003eFuture research should integrate multiple vegetation indicators (e.g., EVI, SIF)(Leng et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) with finer spatial and temporal resolution to enhance detection accuracy(Howey et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Coupling remote sensing data with policy timelines, population shifts, and land use projections will better capture system dynamics. Additionally, the use of spatially explicit models to evaluate trade-offs and synergies among ecosystem services\u0026mdash;particularly at multiple spatial scales\u0026mdash;can offer actionable insights into ecological zoning and adaptive governance(Obiang Ndong et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). As vegetation dynamics in Zhejiang reflect a complex balance between conservation and development, future work must move toward a socio-ecological systems perspective to inform sustainable land management.\u003c/p\u003e\u003c/div\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eThis study focused on Zhejiang Province, a typical subtropical area in southeastern China, aiming to systematically assess the spatiotemporal dynamics and primary driving forces of vegetation cover from 2001 to 2020. Based on MODIS NDVI data and an integrated methodological framework\u0026mdash;including Sen\u0026rsquo;s Slope trend estimation, Mann\u0026ndash;Kendall significance test, Moran\u0026rsquo;s I spatial autocorrelation analysis, and the Geodetector model\u0026mdash;the following key conclusions were drawn:\u003c/p\u003e\u003cp\u003e(1) During the two decades of study, the annual mean NDVI in Zhejiang Province exhibited a statistically significant increasing trend (Sen\u0026rsquo;s slope\u0026thinsp;=\u0026thinsp;0.0025, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), with positive annual growth rates in most years, reflecting a favorable trend in vegetation recovery. Long-term trend analysis revealed that areas classified as \u0026ldquo;significantly improved\u0026rdquo; accounted for 59.4%, indicating an overall positive trajectory in ecosystem quality.\u003c/p\u003e\u003cp\u003e(2) The Moran\u0026rsquo;s I index consistently stayed above 0.79 throughout the entire study period, demonstrating a strong level of spatial autocorrelation. LISA cluster maps further revealed that \u0026ldquo;High\u0026ndash;High\u0026rdquo; clusters were mainly located in forested mountainous regions, whereas \u0026ldquo;Low\u0026ndash;Low\u0026rdquo; clusters were chiefly found in highly urbanized zones. Although the overall spatial pattern remained largely stable, localized degradation hotspots were also identified.\u003c/p\u003e\u003cp\u003e(3) The factor detector results showed that elevation, slope, land use type, and nighttime light possessed strong explanatory power (Q\u0026thinsp;\u0026gt;\u0026thinsp;0.55) for NDVI spatial differentiation. These results emphasize the dominant influence of natural terrain and human activity intensity in shaping vegetation patterns. In contrast, factors such as aspect and PAR had weak explanatory power, while the impact of climatic factors was relatively limited due to climatic homogeneity in the region.\u003c/p\u003e\u003cp\u003e(4) Most combinations of driving factors exhibited enhancement or nonlinear enhancement effects. Interactions such as NTL \u0026times; LULC and SLO \u0026times; NTL had Q-values exceeding 0.70, indicating that the combined effects of natural and socio-economic factors were considerably stronger than any single factor alone.\u003c/p\u003e\u003cp\u003e(5) The risk detector revealed that NDVI values declined significantly in high categories of population density, GDP, and nighttime light, identifying urban fringe areas and industrial clusters as potential ecological degradation hotspots. It is recommended to implement region-specific ecological protection policies, strengthen dynamic monitoring, and carry out targeted restoration efforts in these high-risk areas.\u003c/p\u003e\u003cp\u003eIn conclusion, this study elucidated the spatiotemporal dynamics and spatial structure of NDVI in Zhejiang Province, while identifying key driving forces and their interaction mechanisms. The findings offer important theoretical insights and practical implications. Future research should incorporate multi-scale spatiotemporal datasets, integrate remote sensing with in situ observations, and adopt advanced approaches such as structural equation modeling to further explore the coupling mechanisms of human\u0026ndash;environment systems and support scientific decision-making for sustainable regional ecological development.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflict of interest\u003c/strong\u003e\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eThe authors did not receive support from any organization for the submitted work.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eK.W. conceived and designed the study, performed methodology, formal analysis, and drafted the original manuscript.H.Y. supervised the work and revised the manuscript critically for important intellectual content.W.J. contributed to methodology and investigation, and assisted in manuscript review and editing.N.L. curated the data and prepared the visualizations.J.C. provided resources and contributed to visualization.All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets analyzed during the current study are publicly available in the NASA LP DAAC (https://lpdaac.usgs.gov/) and the China Meteorological Data Service Center (http://data.cma.cn/).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbed, Salwan Ali, Bijay Halder, and Zaher Mundher Yaseen. 2024. Investigation of the decadal unplanned urban expansion influenced surface urban heat island study in the Mosul metropolis. \u003cem\u003eUrban Climate\u003c/em\u003e 54: 101845. https://doi.org/10.1016/j.uclim.2024.101845.\u003c/li\u003e\n\u003cli\u003eAlmazroui, Mansour, and Zek\u0026acirc;i Şen. 2020. Trend Analyses Methodologies in Hydro-meteorological Records. \u003cem\u003eEarth Systems and Environment\u003c/em\u003e 4: 713\u0026ndash;738. https://doi.org/10.1007/s41748-020-00190-6.\u003c/li\u003e\n\u003cli\u003eBeck, Pieter S.A., Clement Atzberger, Kjell Arild H\u0026oslash;gda, Bernt Johansen, and Andrew K. Skidmore. 2006. Improved monitoring of vegetation dynamics at very high latitudes: A new method using MODIS NDVI. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e 100: 321\u0026ndash;334. https://doi.org/10.1016/j.rse.2005.10.021.\u003c/li\u003e\n\u003cli\u003eBikeko, Samuel Shibeshi, and Venkatesham E. 2024. Land use land cover change as a casual factor for climate variability and trends in the Bilate Watershed, Ethiopia. \u003cem\u003eEnvironmental Monitoring and Assessment\u003c/em\u003e 196: 1250. https://doi.org/10.1007/s10661-024-13435-y.\u003c/li\u003e\n\u003cli\u003eChen, Hui, Hongxing Chen, Xiaoyun Huang, Song Zhang, Tengbing He, and Zhenran Gao. 2024. Landscape ecological risk assessment and driving factor analysis in southwest china. \u003cem\u003eScientific Reports\u003c/em\u003e 14: 23208. https://doi.org/10.1038/s41598-024-74506-1.\u003c/li\u003e\n\u003cli\u003eChen, Yun, Peter Taylor, Susan Cuddy, Shahriar Wahid, Dave Penton, and Fazlul Karim. 2024. Inferring vegetation response to drought at multiscale from long-term satellite imagery and meteorological data in Afghanistan. \u003cem\u003eEcological Indicators\u003c/em\u003e 158: 111567. https://doi.org/10.1016/j.ecolind.2024.111567.\u003c/li\u003e\n\u003cli\u003eCumming, Graeme S., Andreas Buerkert, Ellen M. Hoffmann, Eva Schlecht, Stephan von Cramon-Taubadel, and Teja Tscharntke. 2014. Implications of agricultural transitions and urbanization for ecosystem services. \u003cem\u003eNature\u003c/em\u003e 515. Nature Publishing Group: 50\u0026ndash;57. https://doi.org/10.1038/nature13945.\u003c/li\u003e\n\u003cli\u003eDong, Dejin, Jianbo Shen, Daohong Gong, Tianxu Sun, Jiahe Chen, and Yuichiro Fujioka. 2025. Research Trends in Vegetation Spatiotemporal Dynamics and Driving Forces: A Bibliometric Analysis (1987\u0026ndash;2024). \u003cem\u003eForests\u003c/em\u003e 16. Multidisciplinary Digital Publishing Institute: 588. https://doi.org/10.3390/f16040588.\u003c/li\u003e\n\u003cli\u003eGong, Daohong, Dejin Dong, Huaqiang Du, Yufeng Zhou, Sang Fu, and Yuichiro Fujioka. 2025. Spatiotemporal coupling mechanisms and driving forces of ecosystem services and human activity from a multidimensional perspective. \u003cem\u003eInternational Journal of Digital Earth\u003c/em\u003e 18: 2512061. https://doi.org/10.1080/17538947.2025.2512061.\u003c/li\u003e\n\u003cli\u003eGuan, Yanlong, Hongwei Lu, Chuang Yin, Yuxuan Xue, Yelin Jiang, Yu Kang, Li He, and Janne Heiskanen. 2020. Vegetation response to climate zone dynamics and its impacts on surface soil water content and albedo in China. \u003cem\u003eScience of The Total Environment\u003c/em\u003e 747: 141537. https://doi.org/10.1016/j.scitotenv.2020.141537.\u003c/li\u003e\n\u003cli\u003eGuo, Bing, Fang Han, and Lin Jiang. 2018. An Improved Dimidiated Pixel Model for Vegetation Fraction in the Yarlung Zangbo River Basin of Qinghai-Tibet Plateau. \u003cem\u003eJournal of the Indian Society of Remote Sensing\u003c/em\u003e 46: 219\u0026ndash;231. https://doi.org/10.1007/s12524-017-0692-8.\u003c/li\u003e\n\u003cli\u003eGuo, Han, Yingjun Sun, Qi Wang, Xvlu Wang, and Liguo Zhang. 2023. Construction of Greenspace Landscape Ecological Network Based on Resistance Analysis of GeoDetector in Jinan. \u003cem\u003eStochastic Environmental Research and Risk Assessment\u003c/em\u003e 37: 651\u0026ndash;663. https://doi.org/10.1007/s00477-022-02296-x.\u003c/li\u003e\n\u003cli\u003eGuo, Yujing, Lirong Cheng, Aizhong Ding, Yumin Yuan, Zhengyan Li, Yizhe Hou, Liangsuo Ren, and Shurong Zhang. 2024. Geodetector model-based quantitative analysis of vegetation change characteristics and driving forces: A case study in the Yongding River basin in China. \u003cem\u003eInt. J. Appl. Earth Obs. Geoinformation\u003c/em\u003e 132: 104027. https://doi.org/10.1016/j.jag.2024.104027.\u003c/li\u003e\n\u003cli\u003eHowey, Meghan C.L., Michael Palace, Crystal H. McMichael, and Bobby Braswell. 2014. Moderate-Resolution Remote Sensing and Geospatial Analyses of Microclimates, Mounds, and Maize in the Northern Great Lakes. \u003cem\u003eAdvances in Archaeological Practice\u003c/em\u003e 2: 195\u0026ndash;207. https://doi.org/10.7183/2326-3768.2.3.195.\u003c/li\u003e\n\u003cli\u003eHu, Lulu, Xiaojun Xu, Juzhong Wang, and Huaixing Xu. 2023. Individual tree crown width detection from unmanned aerial vehicle images using a revised local transect method. \u003cem\u003eEcological Informatics\u003c/em\u003e 75: 102086. https://doi.org/10.1016/j.ecoinf.2023.102086.\u003c/li\u003e\n\u003cli\u003eHu, Yonghong, Gensuo Jia, and Huadong Guo. 2009. Linking primary production, climate and land use along an urban\u0026ndash;wildland transect: a satellite view. \u003cem\u003eEnvironmental Research Letters\u003c/em\u003e 4: 044009. https://doi.org/10.1088/1748-9326/4/4/044009.\u003c/li\u003e\n\u003cli\u003eHuang, Zihao, Huaqiang Du, Fangjie Mao, Xuejian Li, Guomo Zhou, Jiaqian Sun, Yanxin Xu, Jie Xuan, Yagang Lu, and Lei Huang. 2024. Assessing the impact of land use and cover change on above-ground carbon storage in subtropical forests: a case study of Zhejiang Province, China. \u003cem\u003eGeo-spatial Information Science\u003c/em\u003e. Taylor \u0026amp; Francis: 1\u0026ndash;27.\u003c/li\u003e\n\u003cli\u003eLeng, Song, Alfredo Huete, Jamie Cleverly, Sicong Gao, Qiang Yu, Xianyong Meng, Junyu Qi, Rongrong Zhang, and Qianfeng Wang. 2022. Assessing the Impact of Extreme Droughts on Dryland Vegetation by Multi-Satellite Solar-Induced Chlorophyll Fluorescence. \u003cem\u003eRemote Sensing\u003c/em\u003e 14: 1581. https://doi.org/10.3390/rs14071581.\u003c/li\u003e\n\u003cli\u003eLi, Xin, Bin Fang, Mengru Yin, Tao Jin, and Xin Xu. 2022. Multi-Dimensional Urbanization Coordinated Evolution Process and Ecological Risk Response in the Yangtze River Delta. \u003cem\u003eLand\u003c/em\u003e 11: 723. https://doi.org/10.3390/land11050723.\u003c/li\u003e\n\u003cli\u003eLin, Min, Lizhu Hou, Zhiming Qi, and Li Wan. 2022. Impacts of climate change and human activities on vegetation NDVI in China\u0026rsquo;s Mu Us Sandy Land during 2000\u0026ndash;2019. \u003cem\u003eEcological Indicators\u003c/em\u003e 142: 109164. https://doi.org/10.1016/j.ecolind.2022.109164.\u003c/li\u003e\n\u003cli\u003eLiu, Chenli, Wenlong Li, Wenying Wang, Huakun Zhou, Tiangang Liang, Fujiang Hou, Jing Xu, and Pengfei Xue. 2021. Quantitative spatial analysis of vegetation dynamics and potential driving factors in a typical alpine region on the northeastern Tibetan Plateau using the Google Earth Engine. \u003cem\u003eCATENA\u003c/em\u003e 206: 105500. https://doi.org/10.1016/j.catena.2021.105500.\u003c/li\u003e\n\u003cli\u003eLiu, Meiling, Xiangnan Liu, Ling Wu, Yibo Tang, Yu Li, Yaqi Zhang, Lu Ye, and Biyao Zhang. 2021. Establishing forest resilience indicators in the hilly red soil region of southern China from vegetation greenness and landscape metrics using dense Landsat time series. \u003cem\u003eEcological Indicators\u003c/em\u003e 121: 106985. https://doi.org/10.1016/j.ecolind.2020.106985.\u003c/li\u003e\n\u003cli\u003eMao, Fangjie, Huaqiang Du, Guomo Zhou, Junlong Zheng, Xuejian Li, Yanxin Xu, Zihao Huang, and Shiyan Yin. 2022. Simulated net ecosystem productivity of subtropical forests and its response to climate change in Zhejiang Province, China. \u003cem\u003eScience of The Total Environment\u003c/em\u003e 838: 155993. https://doi.org/10.1016/j.scitotenv.2022.155993.\u003c/li\u003e\n\u003cli\u003eNi Guo, Xiaoping Wang, Dihua Cai, and Jia Yang. 2007. Comparison and evaluation between MODIS vegetation indices in Northwest China. In \u003cem\u003e2007 IEEE International Geoscience and Remote Sensing Symposium\u003c/em\u003e, 3366\u0026ndash;3369. Barcelona, Spain: IEEE. https://doi.org/10.1109/IGARSS.2007.4423566.\u003c/li\u003e\n\u003cli\u003eNol\u0026egrave;, Angelo, Angelo Rita, Agostino Maria Silvio Ferrara, and Marco Borghetti. 2018. Effects of a large-scale late spring frost on a beech (Fagus sylvatica L.) dominated Mediterranean mountain forest derived from the spatio-temporal variations of NDVI. \u003cem\u003eAnnals of Forest Science\u003c/em\u003e 75: 83. https://doi.org/10.1007/s13595-018-0763-1.\u003c/li\u003e\n\u003cli\u003eObiang Ndong, Gregory, Jean Villerd, Isabelle Cousin, and Olivier Therond. 2021. Using a multivariate regression tree to analyze trade-offs between ecosystem services: Application to the main cropping area in France. \u003cem\u003eScience of The Total Environment\u003c/em\u003e 764: 142815. https://doi.org/10.1016/j.scitotenv.2020.142815.\u003c/li\u003e\n\u003cli\u003ePeng, Shouzhang, Yongxia Ding, Wenzhao Liu, and Zhi Li. 2019. 1km monthly temperature and precipitation dataset for China from 1901 to 2017. \u003cem\u003eEarth System Science Data\u003c/em\u003e 11. Copernicus GmbH: 1931\u0026ndash;1946. https://doi.org/10.5194/essd-11-1931-2019.\u003c/li\u003e\n\u003cli\u003ePhung, D., C. Huang, S. Rutherford, C. Chu, X. Wang, M. Nguyen, N. H. Nguyen, C. M. Do, and T. H. Nguyen. 2015. Temporal and spatial patterns of diarrhoea in the Mekong Delta area, Vietnam. \u003cem\u003eEpidemiology and Infection\u003c/em\u003e 143: 3488\u0026ndash;3497. https://doi.org/10.1017/S0950268815000709.\u003c/li\u003e\n\u003cli\u003eQiao, Yu, Xueqiu Wang, Zhixuan Han, Mi Tian, Qiang Wang, Hui Wu, and Futian Liu. 2022. Geodetector based identification of influencing factors on spatial distribution patterns of heavy metals in soil: A case in the upper reaches of the Yangtze River, China. \u003cem\u003eApplied Geochemistry\u003c/em\u003e 146: 105459. https://doi.org/10.1016/j.apgeochem.2022.105459.\u003c/li\u003e\n\u003cli\u003eQiu, Zhengyang, Daohong Gong, Mingxing Zhao, and Dejin Dong. 2025. Spatiotemporal Dynamics and Driving Mechanisms of Soil Conservation Services (SCS) in Zhejiang Province, China: Insights from InVEST Modeling and Machine Learning. \u003cem\u003eRemote Sensing\u003c/em\u003e 17: 2865. https://doi.org/10.3390/rs17162865.\u003c/li\u003e\n\u003cli\u003eŞan, Murat, Fatma Ak\u0026ccedil;ay, Nguyen Thi Thuy Linh, Murat Kankal, and Quoc Bao Pham. 2021. Innovative and polygonal trend analyses applications for rainfall data in Vietnam. \u003cem\u003eTheoretical and Applied Climatology\u003c/em\u003e 144: 809\u0026ndash;822. https://doi.org/10.1007/s00704-021-03574-4.\u003c/li\u003e\n\u003cli\u003eSun, Yiwen, Huiwen Zhan, Chao Gao, Hang Li, and Xianhua Guo. 2024. Spatial Syntactic Analysis and Revitalization Strategies for Rural Settlements in Ethnic Minority Areas: A Case Study of Shuanglang Town, China. \u003cem\u003eBuildings\u003c/em\u003e 14: 2531. https://doi.org/10.3390/buildings14082531.\u003c/li\u003e\n\u003cli\u003eTang, Yan, Xiaojun Xu, Zhongsheng Zhou, Yiling Qu, and Yue Sun. 2021. Estimating global maximum gross primary productivity of vegetation based on the combination of MODIS greenness and temperature data. \u003cem\u003eEcological Informatics\u003c/em\u003e 63: 101307. https://doi.org/10.1016/j.ecoinf.2021.101307.\u003c/li\u003e\n\u003cli\u003eWang, Dan, Youjia Liang, Shouzhang Peng, Zhangcai Yin, and Jiejun Huang. 2022. Integrated assessment of the supply\u0026ndash;demand relationship of ecosystem services in the Loess Plateau during 1992\u0026ndash;2015. \u003cem\u003eEcosystem Health and Sustainability\u003c/em\u003e 8: 2130093. https://doi.org/10.1080/20964129.2022.2130093.\u003c/li\u003e\n\u003cli\u003eWang, Lijun. 2010. The changes of China\u0026rsquo;s environmental policies in the latest 30 years. \u003cem\u003eProcedia Environmental Sciences\u003c/em\u003e 2: 1206\u0026ndash;1212. https://doi.org/10.1016/j.proenv.2010.10.131.\u003c/li\u003e\n\u003cli\u003eWang, Ning, Caiyao Xu, and Fanbin Kong. 2022. Value Realization and Optimization Path of Forest Ecological Products\u0026mdash;Case Study from Zhejiang Province, China. \u003cem\u003eInternational Journal of Environmental Research and Public Health\u003c/em\u003e 19: 7538. https://doi.org/10.3390/ijerph19127538.\u003c/li\u003e\n\u003cli\u003eWang, Zhifang, Min Xu, Haowen Lin, Salman Qureshi, Ankang Cao, and Yujing Ma. 2021. Understanding the dynamics and factors affecting cultural ecosystem services during urbanization through spatial pattern analysis and a mixed-methods approach. \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e 279: 123422. https://doi.org/10.1016/j.jclepro.2020.123422.\u003c/li\u003e\n\u003cli\u003eWang, Ziyi, Haolong Chen, and Prasanna Divigalpitiya. 2024. How does ICT development in resource-exhausted cities promote the urban green transformation efficiency? Evidence from China. \u003cem\u003eSustainable Cities and Society\u003c/em\u003e 115. Elsevier BV: 105835. https://doi.org/10.1016/j.scs.2024.105835.\u003c/li\u003e\n\u003cli\u003eWu, Yue, Zexu Han, Auwalu Faisal Koko, Siyuan Zhang, Nan Ding, and Jiayang Luo. 2022. Analyzing the Spatio-Temporal Dynamics of Urban Land Use Expansion and Its Influencing Factors in Zhejiang Province, China. \u003cem\u003eInternational Journal of Environmental Research and Public Health\u003c/em\u003e 19: 16580. https://doi.org/10.3390/ijerph192416580.\u003c/li\u003e\n\u003cli\u003eXu, Xiaojun, and Danna Chen. 2024. Estimating global annual gross primary production based on satellite-derived phenology and maximal carbon uptake capacity. \u003cem\u003eEnvironmental Research\u003c/em\u003e 252: 119063. https://doi.org/10.1016/j.envres.2024.119063.\u003c/li\u003e\n\u003cli\u003eXu, Yanxin, Xuejian Li, Huaqiang Du, Fangjie Mao, Guomo Zhou, Zihao Huang, Di\u0026rsquo;en Zhu, Qi Chen, Chi Ni, and Keruo Guo. 2023. Spatiotemporal variation in bamboo Solar-induced chlorophyll fluorescence (SIF) in China based on the global Orbiting Carbon Observatory-2 (OCO-2) carbon satellite and study on the response to climate and terrain. \u003cem\u003eGIScience \u0026amp; Remote Sensing\u003c/em\u003e 60: 2253952. https://doi.org/10.1080/15481603.2023.2253952.\u003c/li\u003e\n\u003cli\u003eXu, Yong, Yun-Gui Lu, Bin Zou, Ming Xu, and Yu-Xi Feng. 2024. Unraveling the enigma of NPP variation in Chinese vegetation ecosystems: The interplay of climate change and land use change. \u003cem\u003eScience of The Total Environment\u003c/em\u003e 912: 169023. https://doi.org/10.1016/j.scitotenv.2023.169023.\u003c/li\u003e\n\u003cli\u003eYan, Junjie, Guangpeng Zhang, Hongbo Ling, and Feifei Han. 2022. Comparison of time-integrated NDVI and annual maximum NDVI for assessing grassland dynamics. \u003cem\u003eEcological Indicators\u003c/em\u003e 136: 108611. https://doi.org/10.1016/j.ecolind.2022.108611.\u003c/li\u003e\n\u003cli\u003eYang, Jie, and Xin Huang. 2021. The 30m annual land cover dataset and its dynamics in China from 1990 to 2019. \u003cem\u003eEarth System Science Data\u003c/em\u003e 13: 3907\u0026ndash;3925. https://doi.org/10.5194/essd-13-3907-2021.\u003c/li\u003e\n\u003cli\u003eZhang, Daowei. 2019. China\u0026rsquo;s forest expansion in the last three plus decades: Why and how? \u003cem\u003eForest Policy and Economics\u003c/em\u003e 98: 75\u0026ndash;81. https://doi.org/10.1016/j.forpol.2018.07.006.\u003c/li\u003e\n\u003cli\u003eZhang, Lu, Jianxia Chang, Aijun Guo, Kai Zhou, Guibin Yang, and Dongjing Zou. 2024. Ecological drought evolution characteristics under different climatic regions in the Yangtze River basin. \u003cem\u003eJournal of Hydrology\u003c/em\u003e 629: 130573. https://doi.org/10.1016/j.jhydrol.2023.130573.\u003c/li\u003e\n\u003cli\u003eZhang, Yan, Luoqi Zhang, Junyi Wang, Gaocheng Dong, and Yali Wei. 2023. Quantitative analysis of NDVI driving factors based on the geographical detector model in the Chengdu-Chongqing region, China. \u003cem\u003eEcological Indicators\u003c/em\u003e 155: 110978. https://doi.org/10.1016/j.ecolind.2023.110978.\u003c/li\u003e\n\u003cli\u003eZhao, Rui, Liping Zhan, Mingxing Yao, and Linchuan Yang. 2020. A geographically weighted regression model augmented by Geodetector analysis and principal component analysis for the spatial distribution of PM2.5. \u003cem\u003eSustainable Cities and Society\u003c/em\u003e 56: 102106. https://doi.org/10.1016/j.scs.2020.102106.\u003c/li\u003e\n\u003cli\u003eZheng, Junlong, Fangjie Mao, Huaqiang Du, Xuejian Li, Guomo Zhou, Luofan Dong, Meng Zhang, Ning Han, Tengyan Liu, and Luqi Xing. 2019. Spatiotemporal Simulation of Net Ecosystem Productivity and Its Response to Climate Change in Subtropical Forests. \u003cem\u003eForests\u003c/em\u003e 10: 708. https://doi.org/10.3390/f10080708.\u003c/li\u003e\n\u003cli\u003eZhu, Lijun, Jijun Meng, and Likai Zhu. 2020. Applying Geodetector to disentangle the contributions of natural and anthropogenic factors to NDVI variations in the middle reaches of the Heihe River Basin. \u003cem\u003eEcological Indicators\u003c/em\u003e 117: 106545. https://doi.org/10.1016/j.ecolind.2020.106545.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"NDVI, spatiotemporal variation, Sen–MK test, Moran’s I, Geodetector, Zhejiang Province","lastPublishedDoi":"10.21203/rs.3.rs-7593091/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7593091/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eVegetation cover plays a fundamental role in maintaining ecosystem structure and function. Understanding its spatial and temporal variability, along with its driving factors, is critical for advancing environmental studies. This research targets the subtropical Zhejiang region in southeastern China, utilizing MODIS-derived NDVI data covering 2001 to 2020. By integrating Sen\u0026rsquo;s slope estimator, Mann\u0026ndash;Kendall trend analysis, spatial autocorrelation (Moran\u0026rsquo;s I), and the Geodetector framework, we assessed trends, patterns, and primary influencing factors of vegetation change. Our findings include: (1) A statistically significant upward trend in NDVI across 59.4% of the study area (Sen\u0026rsquo;s slope\u0026thinsp;=\u0026thinsp;0.0025, p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), reflecting ongoing ecological improvement; (2) Notable spatial clustering of NDVI values, with high NDVI zones located in southwestern forested areas and low NDVI zones in expanding urban regions; (3) Elevation, slope, land use/cover, and nighttime lights were identified as major contributors to NDVI spatial variation, with notable interaction effects such as a nonlinear synergy between land use and light; (4) High-risk zones, associated with dense populations and intense urban development, coincided with lower NDVI values. These results deepen our understanding of vegetation dynamics in subtropical zones and provide insights for sustainable ecosystem and land management.\u003c/p\u003e","manuscriptTitle":"Vegetation Dynamics and Influencing Mechanisms in Zhejiang Province, a Typical Subtropical Region of China","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-07 11:34:33","doi":"10.21203/rs.3.rs-7593091/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"59179121-8a78-4b54-89e8-2d624b0b5108","owner":[],"postedDate":"October 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":55683287,"name":"Biological sciences/Ecology"},{"id":55683288,"name":"Earth and environmental sciences/Ecology"},{"id":55683289,"name":"Earth and environmental sciences/Environmental sciences"}],"tags":[],"updatedAt":"2026-02-13T05:09:42+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-07 11:34:33","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7593091","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7593091","identity":"rs-7593091","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}