Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Next Article in Journal
The Short-Term Effects of Prescribed Burning on the Root Biomass and Soil Carbon Dynamics of Larch Plantations
Previous Article in Journal
Spring’s Signal: Can Bud Burst Timing Enhance Resistance to Ash Dieback in Europe?
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Forests
Volume 16
Issue 1
10.3390/f16010142
forests-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Development of Fractional Vegetation Cover Change and Driving Forces in the Min River Basin on the Eastern Margin of the Tibetan Plateau

by
Shuyuan Liu
1,2,
Li Zhou
1,
Huan Wang
2,*,
Jin Lin
2,
Yuduo Huang
2,
Peng Zhuo
2 and
Tianqi Ao
1,*
1
State Key Laboratory of Hydraulics and Mountain River Engineering, College of Water Resource & Hydropower, Sichuan University, No. 24 South Section 1, Yihuan Road, Chengdu 610065, China
2
Institute of Hydrology and Water Resources, Nanjing Hydraulic Research Institute, No. 225, Guangzhou Road, Nanjing 210029, China
*
Authors to whom correspondence should be addressed.
Forests 2025, 16(1), 142; https://doi.org/10.3390/f16010142
Submission received: 16 November 2024 / Revised: 25 December 2024 / Accepted: 13 January 2025 / Published: 14 January 2025
(This article belongs to the Section Forest Inventory, Modeling and Remote Sensing)
Browse Figures
Versions Notes

Abstract

:
Fractional vegetation cover (FVC) is an important indicator of regional ecological environment change, and quantitative research on the spatial and temporal distribution of FVC and the trend of change is of great significance to the monitoring, evaluation, protection, and restoration of regional ecology. This study estimates the FVC of the eastern Tibetan Plateau margin from 2000 to 2020 using the image element dichotomous model based on the Google Earth Engine platform using MODIS-NDVI images. It also investigates the temporal and spatial changes of the FVC in this region and its drivers using the Theil–Sen and Mann–Kendall trend tests, spatial autocorrelation analysis, geodetector, and machine learning approaches impact. The results of this study indicated a generally erratic rising tendency, with the Min River Basin (MRB) near the eastern tip of the Tibetan Plateau having an annual average FVC of 0.67 and an annual growth rate of 0.16%. The percentage of places with better vegetation reached 60.37%. The regional FVC showed significant positive spatial autocorrelation and was clustered. Driver analyses showed that soil type, DEM, temperature, potential evapotranspiration, and land use type were the main drivers influencing FVC on the eastern margin of the Tibetan Plateau. In addition, the random forest (RF) model outperformed the support vector machine (SVM), backpropagation neural network (BP), and long short-term memory network (LSTM) in FVC regression fitting. In summary, this study shows that the overall FVC in the eastern margin of the Tibetan Plateau is on an upward trend, and the regional ecological environment has improved significantly over the past two decades.

1. Introduction

Vegetation is a critical component of terrestrial ecosystems, playing a key role in regulating regional climate, conserving biodiversity, and maintaining ecosystems in terms of material cycling and energy exchange [1,2,3]. The regional FVC, as an important indicator of the quality of ecosystem development, is defined as the ratio of the surface area covered by vegetation to the total surface area [4,5,6,7]. The FVC not only reflects the extent of surface vegetation cover and growth but can also be used to monitor dynamic changes in vegetation cover. In recent years, researchers have increasingly focused on FVC as a key basis for assessing the health of regional ecosystems [8,9]. Exploring the trend and spatial differentiation of FVC and the intrinsic mechanism of its spatial and temporal evolution is crucial for maintaining ecosystem balance and enhancing environmental quality [10,11]. This study can reveal the distribution patterns and density changes of vegetation on the surface, which is scientifically important for the assessment of ecosystem health, land use types, and climate change [12,13]. Vegetation is significantly influenced by various elements of the natural environment, especially the changes in climatic drivers that are closely related to vegetation changes. At the regional scale, there are obvious differences in the effects of climatic drivers on vegetation changes in different regions [14]. Currently, most studies have focused on vegetation response to temperature and precipitation [15,16]. However, more research is required to understand the temporal and geographical patterns of vegetation changes and the mechanisms behind them at both global and local scales, particularly in high-altitude areas that are particularly vulnerable to climate change or climatic transition zones. According to recent research, the way various environmental elements interact greatly affects the kind of vegetation and its geographical distribution [17,18]. Through their effects on photosynthesis and vegetation respiration, these components contribute to the evolution of ecosystem structure and function. Vegetation has also been significantly impacted by human activities, such as land use change, urbanization, overharvesting, and agricultural development, which have changed the natural environment’s circumstances and made vegetation change even worse [19,20]. Correlation and regression analyses are commonly used to explore the relationship between climatic conditions and vegetation change [21,22]. Differences in topographic features and soil types lead to variations in moisture, heat, and nutrient conditions, and how these drivers, in conjunction with climate change and human activities, affect regional vegetation changes and their interactions still needs further study.
A geodetector is used in quantitative analysis to identify the geographical variability influencing vegetation and the ways in which elements interact to uncover the forces behind vegetation change [23,24]. For example, a quantitative analysis of the drivers of vegetation change in the Yellow River Basin by applying a geodetector showed that climatic drivers, especially annual precipitation, were the main drivers and had the greatest impact on vegetation change [25]. Temperature and soil texture have been identified as the primary causes of geographical differences in geodetector investigations regarding vegetation cover change in mountainous settings [26,27]. The geodetector, as a robust and straightforward method, is effective in investigating the spatial heterogeneity of specific geographic phenomena and revealing the strength of contextual drivers and their interactions [28]. The approach is popular for investigating the drivers influencing regional land expansion, ecosystem health, and vegetation change since it does not depend on intricate parameter settings or the presumptions of traditional linear statistical approaches [29,30]. As a vital water recharge area and a wealth of natural resources, the MRB on the Tibetan Plateau region has been identified as an essential ecological safety zone [31]. At the same time, the cultural diversity of the region reflects the complex interweaving of the natural environment and the historical background. The ethnic, linguistic, religious, and artistic diversity of this region constitute a unique cultural landscape. The ecosystem of the MRB is significantly influenced by topography and climate, making it particularly important to study ecological changes in the region. Nevertheless, the majority of earlier research has examined the relationship between vegetation and environmental drivers using the conventional linear approach, which may have overlooked the intricate relationships between vegetation and environmental drivers and failed to quantitatively evaluate the distinct effects of each environmental factor [32]. As a result, a thorough and quantitative analysis of the intricate relationships between vegetation and environmental forces in the MRB is crucial.
Using a variety of techniques, this study examines the driving causes and spatiotemporal fluctuations of FVC in the MRB, which is situated on the eastern edge of the Tibetan Plateau. The research workflow includes (1) statistically analyzing the spatial distribution of vegetation cover in the study area; (2) employing spatial autocorrelation analysis to identify spatial heterogeneity and clustering patterns; (3) applying geographic detectors to evaluate the influence of climate, human activities, topography, and other drivers on FVC; and (4) identifying the major environmental drivers that affect vegetation growth.

2. Study Area and Data

2.1. Study Area

The MRB, located on the eastern edge of the Tibetan Plateau in Sichuan Province, China, is a significant tributary of the Yangtze River. It encompasses multiple cities and counties along the river, from the upper to the lower reaches, including Dujiangyan and Mount Qingcheng. The geographical coordinates of the basin range from 99°30′ E to 104°50′ E and from 29°20′ N to 34°00′ N, covering an area of approximately 135,881 km2 (Figure 1). The basin is bordered to the east by the Chengdu Plain, to the west by mountainous regions, to the south by Yunnan Province, and to the north by Shaanxi Province. The topography is characterized by a mix of high mountains, hills, and plains, with the western region having a higher elevation compared to the lower-elevation eastern areas, with altitudes ranging from 300 to 5000 m. With average annual temperatures between 12 °C and 18 °C and annual precipitation ranging from 800 mm to 1200 mm, the MRB is categorized as having a subtropical humid monsoon climate [33]. The basin is characterized by a well-developed water system, consisting mainly of the MRB and its tributaries, which form a dense water network that supports a variety of functions such as agricultural irrigation, water supply, and flood control. Soil types are dominated by moist red and grayish soils. The vegetation is diverse, ranging from alpine coniferous forests to humid broadleaf forests in the plains. In 2020, the MRB will occupy a significant position in the total economic output within Sichuan Province and will be a key support area for the province’s economic development.

2.2. Data Sources

The NDVI MOD13Q1 data product, which was made available by the NASA Data Center in the United States of America, had the remotely sensed data. It had a time series from 2000 to 2020 with a spatial resolution of 250 m and a temporal resolution of 16 d. For greater sensitivity to vegetation, the data have been processed using atmospheric revision, cloud detection, and radiometric calibration [34,35]. The National Earth System Science Data Center (https://www.geodata.cn/ (accessed on 5 January 2024)) provided the meteorological information, which comprised temperature, precipitation, potential evapotranspiration, humidity index, and aridity index. Data on population density, nighttime lighting, vegetation type, soil type, and GDP were gathered from the Chinese Academy of Sciences Resource and Environmental Science Data Center (http://www.resdc.cn/ (accessed on 9 January 2024)) [36]. DEM data with a spatial resolution of 30 m were obtained from the Geospatial Data Cloud (http://www.gscloud.cn/ (accessed on 10 January 2024)) [37]. Based on the random forest categorization and with a 250 m spatial resolution, the CLCD (China Land Cover Dataset) dataset provided the land use information [38]. Based on 14 potential influencing drivers, this study utilizes geodetectors to identify key drivers affecting FVC changes. Taking the 2015 data as an example, the spatial distribution of the drivers is shown in Figure 2. Since the geodetector requires the input variables to be discrete data, this study used ArcGIS to reclassify all the influencing drivers and resampled them to the same 1 × 1 km resolution as the NDVI data to ensure the consistency and comparability of the data.

3. Methodology

3.1. Calculation of Fractional Vegetation Cover

Large-scale surface FVC estimation can be performed using remote sensing, and currently, the image dichotomy model is often applied for FVC estimation. The image dichotomous model assumes that the information of an image element on the image consists of only two parts: vegetation and no vegetation cover [39]. The calculation formula is as follows:
f c = N D V I − N D V I s o i l N D V I v e g − N D V I s o i l
where N D V I s o i l is the N D V I value of the area without vegetation cover and N D V I v e g is the N D V I value of the area with pure vegetation cover. In the absence of measured data, the maximum value of N D V I max and the minimum value of N D V I min within the 95% confidence interval are taken as N D V I v e g and N D V I s o i l , respectively. Referring to the criteria for classifying vegetation cover in the Soil Erosion Classification Standards, the MRB vegetation cover is classified into five classes. Class I is low vegetation cover ( f c < 30%), Class II is medium-low vegetation cover (30% ≤ f c < 45%), Class III is medium vegetation cover (45% ≤ f c < 60%), Class IV is medium-high vegetation cover (60% ≤ f c < 75%), and Class V is high vegetation cover ( f c ≥ 75%) [40].

3.2. Sen + MK Trend Analysis

The linear regression method is sensitive to noise and requires that the time series data follow a normal distribution. The Theil–Sen regression, a non-parametric estimation method, effectively reduces the impact of noise and is widely used in environmental time series analysis [41]. The trend of FVC changes is examined in this study using the Theil–Sen technique, and significance is assessed using the Mann–Kendall test. The formula for calculating the time series trend is as follows:
β = M e d i a n ( x j − x i j − i ) , ∀ j > i
where x i and x j represent the FVC for time series i and j , and β represents the time series trend.
The Mann–Kendall trend test statistic is shown as follows:
Z = S − 1 V A R ( S ) ( S > 0 ) 0 ( S = 0 ) S + 1 V A R ( S ) ( S   <   0 )
S = ∑ i = 1 n − 1 ∑ j = i + 1 n sgn ( x j − x i )
sgn ( x j − x i ) = 1 ( x j − x i   >   0 ) 0 ( x j − x i = 0 ) − 1 ( x j − x i   <   0 )
V A R ( S ) = n ( n − 1 ) ( 2 n + 5 ) 18
When β > 0 , the time series shows an upward trend and, conversely, a downward trend [42]. When the trend is tested for significance using the MK method, a bilateral test is employed. The trend passes the significance test at the 0.05 and 0.01 levels, respectively, when the absolute value of Z is larger than 1.96 and 2.58 [43].

3.3. Spatial Autocorrelation Analysis

Through statistical analysis, the geographical distribution of characteristic values across spatial units may be examined using spatial autocorrelation [44]. The formula below is used to obtain the global Moran’s I.
G l o b a l   M o r a n   I = ∑ i = 1 n ∑ j = 1 n ( x i − x − ) ( x j − x − ) S 2 ∑ i = 1 n ∑ j = 1 n w i j
where x i and x j are the observations in regions i and j ; n is the number of region units; x − is the mean; S 2 is the variance of the sample; and w i j is the spatial weight matrix [45]. The global Moran’s I takes values in the range of [ − 1 ,   1 ] ; greater than 0 means that the observations are spatially characterized by a clustered distribution, and less than 0 means that the observations are spatially characterized by a discrete distribution [46].
Using the spatial correlation localization index, the correlation between each adjacent spatial unit in the MRB at the eastern edge of the Tibetan Plateau was examined in order to determine any spatial dependency or differences between spatial units [47]. The formula is as follows:
L o c a l   M o r a n   I i = ( x i − x − ) S i 2 ∑ i = 1 , j ≠ 1 n w i j ( x i − x − )
The LISA clustering map produced by the local spatial autocorrelation study has five different types of spatial distributions: non-significant, high–high aggregation, low–low aggregation, high–low anomaly, and low–high anomaly.

3.4. Geodetectors

Geodetectors, a geographic statistical technique for identifying spatial heterogeneity and measuring the influence of drivers, are frequently employed in the analysis of multifactor interactions, drivers, and impacts of a variety of phenomena [48,49]. Factor detection is used to analyze the spatial dissimilarity of the dependent variable Y and the extent to which the respective variable X influences FVC. The calculation formula is as follows:
q = 1 − ∑ h = 1 L N h σ h 2 N σ 2
where N h and N are the number of units in stratum h and the number of units in the entire area, respectively; σ h 2 and σ 2 are the stratum h variance and the variance of the Y value in the entire region; L is the stratification of the dependent variable FVC or the influencing factor X ; and q is the explanatory power of each influencing factor for Y in the range of [0, 1] [50].
To identify interactions between various impacts on factor X, interaction detection is employed. It evaluates whether two drivers acting alone have an impact on Y or if their combined effect increases or decreases the explanatory power. Table 1 shows the five different kinds of factor interactions.
While risk tests are intended to identify the proper range or type, ecological tests are used to establish whether there is a substantial difference between the elements; the greater the FVC score, the better the indicator matches. In this study, we selected typical data from 2000, 2005, 2010, 2015, and 2020 for analysis.

4. Results

4.1. Spatial and Temporal Variations in FVC

The general trend of long-term vegetation change in the study area was assessed by analyzing the temporal changes in FVC between 2000 and 2020 along the eastern margin of the Tibetan Plateau. First, the FVC values of the watershed were divided into five categories (Figure 3a), 0–0.2, 0.2–0.4, 0.4–0.6, 0.6–0.8, and 0.8–1, which corresponded to low, low-moderate, moderate, moderate-high, and high FVC. From 2000 to 2012, the annual average proportion of high FVC was 27.31%, and after 2012, this proportion increased to 35.69%. Meanwhile, the percentage of medium-high FVC was 33.51 percent during this period, but it decreased to 28.29 percent after 2012. In certain locations, this suggests a transition from medium-high to high FVC, while the percentage of low and medium-low coverage did not change much. According to Figure 3b, the FVC in the study area showed some volatility, but overall, it showed a small upward trend, with a mean value of 0.67. However, the FVC has been trending lower since 2016. This is the outcome of both human activity and the consequences of climate change.
It is evident that there has been a general improvement in FVC in the area based on the geographical distribution of the average FVC at the eastern margin of the Tibetan Plateau in 2000, 2005, 2010, 2015, and 2020, as well as the multi-year averages from 2000 to 2020 (Figure 4). Based on the multi-year averages, it shows that the medium, medium-high, and high FVC categories occupy 78.6% of the region, with higher FVC in the southeast and lower in the northwest and central regions. The findings indicate that the regions with low FVC are mostly found in the high-elevation grasslands in the basin’s center and the urbanized Chengdu Plain, while the regions with high and medium-high FVC are mostly found in the middle and lower reaches of the basin with relatively low elevation. Overall, from 2000 to 2020, the FVC in the study area did not change much, but the high FVC gradually increased. Among them, low and medium-low FVC decreased and then increased from 15.6% in 2000 to 20.6% in 2020 but decreased to 13.23% in 2015. Medium FVC and medium-high FVC had the highest percentage and were widely distributed in most parts of the watershed, and their percentage decreased from 55.89% in 2000 to 51.24% in 2020. Meanwhile, the share of high FVC gradually increased, slowly growing from 27.52% in 2000 to 28.74% in 2020 but reaching 38.47% in 2015, indicating that high FVC decreased between 2015 and 2020 and was mainly distributed in the Liangshan, Ya’an, Leshan, and Yibin regions. This trend suggests that although there is an improvement in FVC in general, there are still significant fluctuations and declines in some specific regions.

4.2. FVC Change Trend Analysis

FVC experienced significant changes between 2000 and 2020. In this paper, the transfer of each class of FVC was analyzed for 2000–2010 (Figure 5a) and 2010–2020 (Figure 5b), respectively, with I, II, III, IV, and V representing low, medium-low, medium, medium-high, and high FVC, respectively. During the period of 2000–2010, the main transfer of low FVC was to medium-low coverage, with a percentage of 13.18%, which was mainly located in the basin parts of the upper basin; the main direction of change for low and medium FVC was low and medium, with a 5.61% shift to medium and a 16.9% shift to low, and these changes were mainly distributed in the lower basin area. The main direction of change for medium FVC was low and medium-high, with 9.32% and 21.72%, respectively, which were mainly located in the central region of the basin. The medium-high level of FVC shifted mainly to medium and high levels, where the proportion of conversion to medium was 15.04% higher than the proportion of conversion to high. High levels of FVC change primarily to medium and high levels, and their overall change remains largely consistent with the previous period. Between 2010 and 2020, there was a 24 percent shift from low FVC to medium-low coverage, mostly in the grassland regions of the watershed’s upper and middle reaches. Medium-low FVC, on the other hand, shifted primarily toward medium coverage, with a 46.99% share, with the area of change being located in the upper reaches of the watershed. Medium FVC mainly changed to medium-low and medium-high levels, with a high percentage of 36.68% converting to medium-high coverage, mainly located in the middle and lower reaches of the watershed. Medium-high FVC mainly changed to high and medium levels, with the proportion converted to high and medium as high as 18.37% and 13.75%, respectively, and the area of change was mainly located in the southwestern part of the basin. The overall high FVC showed a decreasing trend, with 8.01% less area than before. In general, 73.69% of the region’s FVC had an upward trend. However, in the upper part of the watershed and the Chengdu Plain, FVC showed a decreasing trend.
Between 2000 and 2020, the FVC at the eastern margin of the Tibetan Plateau showed a trend of −0.051 to 0.023 (Figure 6a). About 60.37% of the study area showed an increasing trend in FVC, while 39.63% showed a decreasing trend. During this period, areas such as Chengdu Plain and Aba Tibetan Autonomous Prefecture had faster urban development, and there was a greater concentration in the regions where FVC had significantly decreased. According to the results of the significance test (Figure 6b), 37.32% of the areas showed significant improvement in NDVI values, of which 15.85% showed insignificant improvement. Vegetation degradation in the Chengdu Plain, especially near urban plots, was obvious or insignificant, accounting for 2.8% and 2.6%, respectively. These areas mainly consisted of building sites and unutilized land. About 35.31% of the areas were not significantly degraded, mainly in the upstream and southwestern areas. Overall, there are more areas with significant improvement relative to the areas with declining FVC.

4.3. FVC Spatial Autocorrelation Analysis

Geoda 1.16 software was used to determine the worldwide Moran’s I index of FVC on the Tibetan Plateau’s eastern boundary between 2000 and 2020 (Figure 7), and the results showed that the values were 0.460, 0.445, 0.419, 0.446, and 0.450, respectively, which were greater than 0 (p < 0.01), indicating the existence of a significant positive spatial autocorrelation. This suggests that there is an aggregated distribution of vegetation in the research region. Between 2000 and 2020, the FVC of the eastern edge of the Tibetan Plateau was analyzed by Moran’s I, which showed that it gradually decreased from 0.460 in 2000 to 0.419 in 2010 and then rebounded to 0.450 in 2020, which indicates that the FVC of the area in 2000 showed a stronger spatial aggregation effect and was higher than that of other years. Over time, the FVC in the watershed shows a trend of weakening in spatial correlation, and its clustering effect is significantly reduced, which is reflected in the decline of the global Moran’s I index. Despite this trend, the FVC in the eastern edge of the Tibetan Plateau maintained a relatively stable spatial aggregation pattern with relatively small changes during the two decades from 2000 to 2020.
Further localized spatial autocorrelation analyses were conducted for the vegetation cover of the eastern margin of the Tibetan Plateau (Figure 8). The results showed that there were significant overall spatial variations in vegetation cover in the study area over the five years between 2000 and 2020. The areas of high–high aggregation were consistent with the areas of improved vegetation cover shown in Figure 6. On the contrary, low–low aggregation was mainly distributed in the upper part of the watershed and the Chengdu Plain, which was consistent with the distribution of low vegetation cover areas in Figure 4. There are relatively few low–high anomalies, while there are more high–low anomalies, which are mainly distributed in some areas in the middle-upper and lower reaches of the watershed. This suggests that the vegetation cover is more pronounced in certain places. The local spatial autocorrelation analysis of different years in the study area showed that in 2000, the high–high aggregation in the study area was sparse, mainly concentrated in the central and northeastern parts of the basin. By 2005, the local spatial autocorrelation results changed significantly, with a slight increase in high–high aggregation in the central region and high–low aggregation in the lower part of the watershed, which was previously insignificant. In 2010, high–high aggregation significantly increased from 2000 and was distributed in Ya’an City, Leshan City, and Liangshan Prefecture, which were areas that in 2000 and 2005 showed insignificant and high–low aggregation. In contrast, the northeastern part of the basin showed high–high aggregation in 2000, 2010, and 2015, while it was non-significant in 2005 and 2020. The northwestern region, which was non-significant in 2000, 2005, and 2010, changed to low–low aggregation in 2015 and returned to non-significant in 2020. Most of the lower portion of the watershed shifted from high–low aggregation to high–high aggregation in 2020 compared to 2005, a trend that was also seen in 2010 and 2015.

4.4. Geographical Detection of Drivers Influencing FVC

4.4.1. Factor Detection

Factor detection analysis was conducted to assess the impact of 14 drivers, including natural and human activities, on the FVC of the eastern Qinghai–Tibet Plateau. The q-values obtained from the factor detection provide an explanation of the impact of each driving factor on the change in vegetation cover. The results are shown in Figure 9, which indicates that all 14 drivers have a highly significant impact on FVC variation. Soil type, vegetation type, DEM, precipitation, temperature, potential evapotranspiration, land use type, and drought index consistently exhibit strong explanatory power for FVC across the years, with land use type having the greatest influence, with an average annual q-value of 0.561. The next most influential drivers are temperature, DEM, potential evapotranspiration, and soil type, with average annual q-values of 0.559, 0.552, 0.471, and 0.455, respectively. The drought index, vegetation type, precipitation, relative humidity, population density, slope, nighttime lights, GDP, and aspect show progressively weaker influences, with q-values of 0.282, 0.273, 0.172, 0.066, 0.034, 0.018, 0.015, 0.011, and 0.0004, respectively. The variation in q-values across different driving drivers exhibits certain fluctuations; for instance, the drought index decreased from 0.396 in 2000 to 0.217 in 2020. Temperature, DEM, plant type, soil type, and potential evapotranspiration all had declining q-values from 2000 to 2010 but rising q-values from 2010 to 2020. In conclusion, soil type, DEM, temperature, potential evapotranspiration, and land use type are the primary drivers influencing FVC on the eastern Qinghai–Tibet Plateau. Furthermore, the trends from 2010 to 2020 suggest that the influence of these drivers on FVC has gradually strengthened over time.

4.4.2. Factor Interaction Detection

Assessing whether the combined effect of the two drivers increases or decreases the influence on FVC is the main purpose of interaction probing. The interaction between the drivers in various years is shown to have explanatory power in Figure 10. The findings indicate that the influence on FVC was often amplified by the interaction between the various components. Comparing the results of the single-factor probes, the effects of slope, slope direction, relative humidity, population density, GDP, and nighttime lighting are small. However, when they interacted with other drivers, their explanatory power was significantly enhanced. Specifically, as Figure 10d illustrates, the most substantial impact on FVC in various years is caused by the interaction of land use with other variables, with an annual average explanatory power ranging from 0.62 to 0.72. The predominant vegetation cover interactions on the eastern edge of the Tibetan Plateau, as seen in several years, were those between land use and the aridity index, land use and temperature, and DEM. In addition, the interactions of DEM, temperature, and potential evapotranspiration also had strong effects on FVC, with annual average explanatory power above 0.5, which further indicates that the three drivers of land use, temperature, and potential evapotranspiration have the strongest effects on the MRB. As society has developed, the influence of human activities on the FVC has gradually increased to a certain extent. This is demonstrated by the decreasing trend in the fluctuation of the q-values of the interactions between various drivers, particularly the interaction between natural drivers, and the increasing trend in the fluctuation of the interactions between land use conversion type and population density, nighttime lighting, and GDP.

4.4.3. Differences Between FVC Drivers

The results of the ecological probes for five typical years between 2000 and 2020 were analyzed using the F-test method (Figure 11). The results showed no significant differences between population density and nighttime light intensity and other drivers. On the contrary, soil type, vegetation type, DEM, temperature, precipitation, potential evapotranspiration, and land use were significantly different among the drivers. Relative humidity with slope and slope orientation, precipitation with slope and population density, potential evapotranspiration with soil type, land use with temperature, GDP with DEM, and aridity index with vegetation type, relative humidity, and precipitation showed significant correlation in a given year.

4.4.4. Appropriate Ranges or Types of FVC Driving Drivers

Due to the small q-values for slope, slope orientation, relative humidity, population density, and nighttime light intensity, as well as uncertainty about the significance of ecological detection, they had weak effects on FVC. Risk identification using t-tests (p < 0.05) identified appropriate ranges or types of other drivers, which are displayed in Table 2 and Figure 12.
Among the effects of climatic drivers, FVC increases with increasing temperature between −15.2 and 12 °C, and the average FVC reaches the highest at 6.69–12 °C (Figure 12d), which is 0.789, whereas potential evapotranspiration increases with increasing potential evapotranspiration up to 982 mm, and the highest FVC is found when the potential evapotranspiration is in the interval of 859–982 mm (Figure 12f), which reached 0.766, indicating that the relationship between temperature and potential evapotranspiration and FVC is not such that higher temperatures and potential evapotranspiration lead to higher FVC. The average FVC was highest when the precipitation was in the range of 890–1040 mm (Figure 12e), which had the highest FVC of 0.733. On the other hand, the average FVC increased with the dryness index, which increased with the increase in the dryness index. When the dryness index was in the range of 1.47–1.84, the maximum mean FVC was 0.811, indicating that higher FVC promotes vegetation growth. In addition to other environmental and topographic drivers, the mean FVC showed an increasing trend at elevations less than 3517 m (Figure 12c), while a decrease in FVC was observed when the elevation was greater than 3517 m. The maximum FVC was 0.801 in the range of 2782–3517 mm; the soil type and vegetation type were eluvial and broadleaved, respectively. Forest reached the highest FVCs of 0.778 and 0.703, respectively, while the lowest FVCs of 0.282 and 0.442 were recorded for glaciers and other vegetation types, respectively. In human activities (Figure 12g), the FVC maximum was at 0.879 when the land use type was a forest, and it was 0.784 for crops and 0.879 when the land use type was other, which includes built-up land, unutilized land, and water bodies. The FVC is at its lowest at 0.193.

4.4.5. FVC Regression Fitting Based on Machine Learning

In this study, the data with multiple independent variables were homogenized, and the latest data were used for those that could not be processed. As shown in Figure 13, this study constructed SVM, BP, LSTM, and RF models for the FVC of the eastern margin of the Tibetan Plateau using all the drivers as independent variables and calculated the fitting effectiveness indexes R2, RMSE, and MAE of these models, respectively. As shown by the results in Figure 10, the R2 of the SVM model was 0.82, the RMSE was 0.0877, and the MAE was 0.0678. The BP model has an R2 of 0.81, an RMSE of 0.0878, and an MAE of 0.0664. The LSTM model has an R2 of 0.88, an RMSE of 0.0697, and an MAE of 0.0476, while the RF model achieves an R2 of 0.97, an RMSE of 0.0329, and an MAE of 0.0249. The R2 of the RF model is significantly higher than that of the SVM, BP, and LSTM models, while its RMSE and MAE are significantly lower than those of the other models, which fully indicates that the RF model is more suitable for regression fitting of FVC.

5. Discussion

5.1. Characterization of Spatial and Temporal Changes in FVC

The eastern margin of the Tibetan Plateau encompasses diverse topographic and climatic types, ranging from alpine regions to basin zones and from humid climates to arid zones, resulting in diverse ecosystems, including forests, grasslands, and wetlands [51]. However, human activities have caused far-reaching impacts on the ecological environment, making the ecological management and protection of the region a serious challenge. In this study, we analyzed the overall trend of improvement in the MRB over the past 21 years from 2000 to 2020, with the higher FVC and high FVC areas gradually expanding from northwest to southeast. The vegetation growth areas are mainly concentrated in the southeast, which is basically consistent with the results of Sun et al. [52]. In terms of temporal distribution, high FVC and medium-high FVC increased significantly from 2013 onwards and reached the highest level in 2016. However, a decreasing trend began to appear subsequently, and this change may be related to climate change as well as the impact of human activities, reflecting the annual variability of FVC. With an overall distribution pattern of “high in the east, low in the west, high in the south, and low in the north,” FVC in the MRB at the eastern margin of the Tibetan Plateau demonstrated clear diversity in terms of spatial distribution. This is due to the rich vegetation types in the alpine zone in the eastern part of the MRB and the humid region in the southern part of the basin, which are dominated by alpine meadows and forests [53], and the relatively high precipitation in these areas, which results in a higher FVC. The low FVC and lower FVC areas in the MRB are mainly concentrated in the upper part of the basin, including some high and arid mountainous areas. These areas have low precipitation and relatively high evapotranspiration, resulting in arid and infertile land, which is not conducive to vegetation growth [54]. In addition, the vegetation in these areas is dominated by alpine grassland and desert vegetation, resulting in lower FVC [55].

5.2. Drivers Influencing FVC

The eastern edge of the Tibetan Plateau is characterized by significant differences in natural resource endowment, and its natural drivers have a significant impact on vegetation growth, while anthropogenic drivers have a lesser impact on its FVC [56]. This study emphasized the importance of natural drivers on vegetation growth in the MRB. Possible explanations for this include the MRB’s mild, humid environment, which lessens vegetation’s vulnerability to climate change, and the minor regional variation in its climatic features between 2000 and 2020 [54,57]. Furthermore, manmade impacts—such as ecological engineering and enclosure measures—have improved degraded land, aided in vegetation regeneration, and lessened the influence of precipitation on vegetation. However, these effects have been more restricted throughout the MRB [58]. The type of land use is a major factor influencing changes in vegetation cover, and in the Chengdu Plain region, urbanization usually results in the replacement of natural vegetation by buildings, roads, and other infrastructure [59]. Such changes reduce natural vegetation cover and affect ecosystem functions, such as water cycling and biological habitats. In addition, the existing results have also shown that economic development can also promote the increase in vegetation cover [60], and the effective implementation of ecological protection policies can realize the coordinated development of the economy and environment, which is verified by the enhancement of FVC in Chengdu Plain after 2010 [61]. Apart from land use, watershed FVC was less affected by other anthropogenic variables including GDP, population density, and nighttime illumination [62]. This was mostly because the MRB had a lower total population density or degree of economic activity, which lessened the strain on the vegetation.
In the eastern margin of the Tibetan Plateau, different soil types significantly affect the soil’s ability to store and excrete water, which in turn has a significant impact on the water availability and growth of vegetation [63]. Specifically, soil types in the watershed, such as yellow and brown soils, determine the nutrient content of the soil, which directly affects plant growth and cover. In addition, the mountainous soils of the MRB are susceptible to erosion, and this erosion process may lead to soil erosion, which further affects vegetation cover [64]. The effect of altitude on FVC in the MRB shows significant spatial differentiation. In the high-elevation areas of the northwestern part of the basin, vegetation growth in these areas is less restricted by altitude due to the complexity of the topography and abundant hydrothermal conditions [65]. Specifically, high-elevation areas usually have impacts on vegetation types and distribution due to changes in climatic conditions, such as decreases in temperature and changes in precipitation [66]. However, the diversity of topography and the complexity of the local climate in the northwestern part of the MRB make the vegetation in these high-elevation areas more adaptable, and thus the effect of elevation on vegetation growth is relatively small in these areas [67,68]. This phenomenon implies that topographic characteristics and local climate, in addition to elevation variables, have a significant role in determining FVC and growth patterns. Human activities also have an impact on the amount of vegetation, influencing its extent and intensity [68]. This implies further that the impacts of each component are not independent and that it is especially crucial to examine how the elements interact.
The influence of climatic drivers on FVC is particularly significant in the MRB. Among them, temperature has the most prominent effect on FVC. Temperature has a significant impact on the growth cycle and pace of plants at high elevations [69]. Suitable temperatures can promote plant growth, while lower temperatures may lead to slow growth and lagging phenological periods, thus seriously affecting the FVC of vegetation. In contrast, in plains areas where watershed elevations are lower, high temperatures can inhibit plant growth and cause heat stress [70]. Climatic characteristics of the MRB make the effect of temperature change on FVC particularly pronounced, especially during the dry season, when the increase in temperature exerts significant pressure on the water demand and growth status of vegetation. Precipitation responds relatively poorly to changes in FVC [71]. This may be due to the richness of vegetation types in the watershed, including forests, grasslands, and agricultural lands, which are mainly dominated by forests and grasslands. Whereas grasslands are more susceptible to fluctuations in precipitation, forests react to these changes more steadily [72]. Furthermore, the FVC could not alter much through time, and precipitation in the MRB stays rather constant. Although seasonal or annual precipitation changes can have some impact on vegetation, stable precipitation over a long period of time can keep FVC at a relatively stable level. Potential evapotranspiration is a more accurate representation of actual water demand and consumption during plant growth than precipitation and thus has a more direct impact on FVC [73]. The MRB is located in a complex climatic region, and increased potential evapotranspiration during seasons or periods of higher temperatures may result in increased plant water demand, which can significantly affect FVC [74]. Topographic and climatic differences make the effect of potential evapotranspiration on vegetation in different regions uneven. For example, climatic conditions in alpine regions may result in higher evapotranspiration demand, while they may be lower in plains regions [75]. This difference makes the effect of potential evapotranspiration on FVC more significant. In recent years, the drought index has shown a significant decrease after 2010. This phenomenon may be related to improvements in water management and regulation measures in recent years, which have mitigated the direct effects of drought on vegetation [76]. At the same time, climate change has led to changes in precipitation patterns in the basin, reducing the frequency of extreme drought events, thus further mitigating the impact of drought on vegetation [77]. Furthermore, it was discovered that the combination of natural and anthropogenic components provided a stronger explanation for FVC, indicating that modifications to human activity boosted the impact of other drivers on FVC [78]. Therefore, when formulating vegetation management policies for the MRB on the eastern edge of the Tibetan Plateau, the effects of multiple drivers need to be considered in an integrated manner in order to provide a valuable governance model.

5.3. Implications for Vegetation Management

The findings of this study have important implications for vegetation management in the MRB. A comprehensive understanding of the interactions between climatic, topographic, and anthropogenic factors can inform targeted interventions to enhance vegetation resilience. For instance, soil conservation measures in erosion-prone areas, reforestation in degraded zones, and sustainable land use planning in urban areas can collectively contribute to the sustainable development of the region. Policymakers should adopt adaptive management strategies that account for the spatial heterogeneity of FVC and prioritize interventions that address the most pressing environmental challenges in each subregion.

5.4. Shortcomings and Perspective

Geodetectors are a useful tool for determining the effects of these elements and how they interact, as well as for identifying and measuring important drivers influencing plant change in the MRB on the eastern edge of the Tibetan Plateau. Geodetectors can more accurately determine the function of influencing drivers and their interactions on FVC than other methods. However, geodetector models show significant advantages in identifying relationships among multiple drivers, but their criteria for discretizing continuous data are not clear. This study adopted the natural breakpoint method, which is commonly used by most scholars, but different classification methods may have an impact on the model results. Therefore, the classification methods and their number still need to be further explored.
Since policy drivers and public awareness of ecological protection have not yet been quantified by current research, more thorough studies of FVC changes and their dynamic processes using standardized classification methods are required in order to support sustainable socioeconomic development and provide scientific evidence for the sustainable use of natural resources. Furthermore, the effects of climatic extremes, drought, and variations in CO2 concentration on FVC should be considered. Lastly, in light of global warming, future research utilizing CMIP6 data in conjunction with deep learning techniques should examine the accelerated effects of climate change on vegetation in ecologically vulnerable areas.

6. Conclusions

The conclusions of this study are as follows:
(1) According to the findings, the research area’s overall FVC was high between 2000 and 2020, with an average FVC of 0.67. The medium, medium-high, and high FVC categories made up 78.6% of the area, with the center region having higher vegetation values and the northwest having lower FVC.
(2) During the previous 21 years, FVC has improved tremendously and now makes up 60.37 percent of the overall area. But, 39.63% of the region continues to exhibit a declining tendency, mostly as a result of the fast urbanization process. It is especially noteworthy that FVC is declining in the Chengdu Plain region.
(3) According to the spatial autocorrelation analysis, the overall FVC in the MRB shows significant positive spatial autocorrelation, i.e., the FVC in the study area is in the state of aggregation; the rise in the high–high aggregation phenomena in the center of the research area and the rise in the low–low aggregation phenomenon in the Chengdu Plain region are the primary manifestations of the varied local spatial autocorrelation analysis results in different years.
(4) Temperature and potential evapotranspiration are the primary climatic elements influencing the change in FVC, according to the geodetector model results. Among the topographic drivers, DEM significantly affected FVC more than slope and aspect. Soil and vegetation types also have a significant influence on environmental drivers, while land use type is the main anthropogenic driver.
(5) There were two-factor nonlinearly enhanced interactions of the drivers on FVC, and the effects of these interactions were greater than those of the individual drivers. The appropriate temperature range was 6.69–12 °C, the potential evapotranspiration was 859–982 mm, the precipitation was 890–1040 mm, the aridity index was 1.47–1.84, and the elevation range was 2782–3517 mm. Woodland land use, broadleaf forest vegetation, and alluvial soil were the best types of land use and vegetation, respectively.
(6) Regression fitting of 14 drivers using four machine learning methods, SVM, BP, RF, and LSTM, found that the 14 drivers in RF have very high applicability in the regression fitting of FVC in the MRB, and the RF model is more suitable for the regression fitting of FVC in the MRB compared with other models.

Author Contributions

Methodology, S.L., L.Z., Y.H. and T.A.; Software, S.L., H.W. and T.A.; Validation, J.L. and P.Z.; Formal analysis, S.L., L.Z. and H.W.; Investigation, H.W., J.L. and P.Z.; Resources, S.L. and P.Z.; Data curation, J.L. and Y.H.; Writing—original draft, S.L.; Writing—review & editing, L.Z., Y.H., P.Z. and T.A.; Visualization, J.L.; Supervision, Y.H.; Funding acquisition, H.W. and T.A. All authors have read and agreed to the published version of the manuscript.

Funding

We gratefully acknowledge the National Key R&D Program of China (2022YFC3002902) and the Key R&D Project (XZ202101ZY0007G) from the Science and Technology Department of Tibet.

Data Availability Statement

Data are contained within the article.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Galidaki, G.; Zianis, D.; Gitas, I.; Radoglou, K.; Karathanassi, V.; Tsakiri–Strati, M.; Woodhouse, I.; Mallinis, G. Vegetation biomass estimation with remote sensing: Focus on forest and other wooded land over the Mediterranean ecosystem. Int. J. Remote Sens. 2017, 38, 1940–1966. [Google Scholar] [CrossRef]
  2. Li, W.; Buitenwerf, R.; Munk, M.; Bøcher, P.K.; Svenning, J.-C. Deep-learning based high-resolution mapping shows woody vegetation densification in greater Maasai Mara ecosystem. Remote Sens. Environ. 2020, 247, 111953. [Google Scholar] [CrossRef]
  3. Ma, S.; Qiao, Y.-P.; Wang, L.-J.; Zhang, J.-C. Terrain gradient variations in ecosystem services of different vegetation types in mountainous regions: Vegetation resource conservation and sustainable development. For. Ecol. Manag. 2021, 482, 118856. [Google Scholar] [CrossRef]
  4. Chu, D. Fractional vegetation cover. In Remote Sensing of Land Use and Land Cover in Mountain Region: A Comprehensive Study at the Central Tibetan Plateau; Springer: Singapore, 2020; pp. 195–207. [Google Scholar]
  5. Wu, D.; Wu, H.; Zhao, X.; Zhou, T.; Tang, B.; Zhao, W.; Jia, K. Evaluation of spatiotemporal variations of global fractional vegetation cover based on GIMMS NDVI data from 1982 to 2011. Remote Sens. 2014, 6, 4217–4239. [Google Scholar] [CrossRef]
  6. Mao, P.; Zhang, J.; Li, M.; Liu, Y.; Wang, X.; Yan, R.; Shen, B.; Zhang, X.; Shen, J.; Zhu, X. Spatial and temporal variations in fractional vegetation cover and its driving factors in the Hulun Lake region. Ecol. Indic. 2022, 135, 108490. [Google Scholar] [CrossRef]
  7. Kong, Z.; Ling, H.; Deng, M.; Han, F.; Yan, J.; Deng, X.; Wang, Z.; Ma, Y.; Wang, W. Past and projected future patterns of fractional vegetation coverage in China. Sci. Total Environ. 2023, 902, 166133. [Google Scholar] [CrossRef] [PubMed]
  8. Bao, Z.; Shifaw, E.; Deng, C.; Sha, J.; Li, X.; Hanchiso, T.; Yang, W. Remote sensing-based assessment of ecosystem health by optimizing vigor-organization-resilience model: A case study in Fuzhou City, China. Ecol. Inform. 2022, 72, 101889. [Google Scholar] [CrossRef]
  9. Xu, W.; Yang, H.; Chen, Z.; Shi, R.; Liu, Y.; Chen, J. Enhancing bird habitat networks in metropolitan areas: Resilience assessment and improvement strategies—A case study from Shanghai. Sci. Total Environ. 2024, 954, 176316. [Google Scholar] [CrossRef] [PubMed]
  10. Zou, S.; Qian, J.; Xu, B.; Tu, Z.; Zhang, W.; Ma, X.; Liang, Y. Spatiotemporal changes of ecosystem health and their driving mechanisms in alpine regions on the northeastern Tibetan Plateau. Ecol. Indic. 2022, 143, 109396. [Google Scholar] [CrossRef]
  11. Feng, S.; Liu, X.; Zhao, W.; Yao, Y.; Zhou, A.; Liu, X.; Pereira, P. Key areas of ecological restoration in Inner Mongolia based on ecosystem vulnerability and ecosystem service. Remote Sens. 2022, 14, 2729. [Google Scholar] [CrossRef]
  12. Chen, S.; Zhang, Q.; Chen, Y.; Zhou, H.; Xiang, Y.; Liu, Z.; Hou, Y. Vegetation change and eco-environmental quality evaluation in the Loess Plateau of China from 2000 to 2020. Remote Sens. 2023, 15, 424. [Google Scholar] [CrossRef]
  13. Wang, S.; Chen, H.; Su, W.; Cui, S.; Xu, Y.; Zhou, Z. Research on habitat quality assessment and decision-making based on Semi-supervised Ensemble Learning method—Daxia River Basin, China. Ecol. Indic. 2023, 156, 111153. [Google Scholar] [CrossRef]
  14. Jiang, L.; Bao, A.; Guo, H.; Ndayisaba, F. Vegetation dynamics and responses to climate change and human activities in Central Asia. Sci. Total Environ. 2017, 599, 967–980. [Google Scholar] [CrossRef] [PubMed]
  15. Wu, D.; Zhao, X.; Liang, S.; Zhou, T.; Huang, K.; Tang, B.; Zhao, W. Time-lag effects of global vegetation responses to climate change. Glob. Change Biol. 2015, 21, 3520–3531. [Google Scholar] [CrossRef]
  16. Chang, C.-T.; Wang, S.-F.; Vadeboncoeur, M.A.; Lin, T.-C. Relating vegetation dynamics to temperature and precipitation at monthly and annual timescales in Taiwan using MODIS vegetation indices. Int. J. Remote Sens. 2014, 35, 598–620. [Google Scholar] [CrossRef]
  17. Liu, C.; Li, W.; Wang, W.; Zhou, H.; Liang, T.; Hou, F.; Xu, J.; Xue, P. 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 2021, 206, 105500. [Google Scholar] [CrossRef]
  18. Ren, Z.; Tian, Z.; Wei, H.; Liu, Y.; Yu, Y. Spatiotemporal evolution and driving mechanisms of vegetation in the Yellow River Basin, China during 2000–2020. Ecol. Indic. 2022, 138, 108832. [Google Scholar] [CrossRef]
  19. Vlek, P.L.; Khamzina, A.; Tamene, L.D. Land Degradation and the Sustainable Development Goals: Threats and Potential Remedies; International Center for Tropical Agriculture (CIAT): Nairobi, Kenya, 2017. [Google Scholar]
  20. Billah, M.M.; Rahman, M.M.; Abedin, J.; Akter, H. Land cover change and its impact on human–elephant conflict: A case from Fashiakhali forest reserve in Bangladesh. SN Appl. Sci. 2021, 3, 649. [Google Scholar] [CrossRef]
  21. Watson, J.E.; Evans, T.; Venter, O.; Williams, B.; Tulloch, A.; Stewart, C.; Thompson, I.; Ray, J.C.; Murray, K.; Salazar, A. The exceptional value of intact forest ecosystems. Nat. Ecol. Evol. 2018, 2, 599–610. [Google Scholar] [CrossRef] [PubMed]
  22. Jenny, J.-P.; Anneville, O.; Arnaud, F.; Baulaz, Y.; Bouffard, D.; Domaizon, I.; Bocaniov, S.A.; Chèvre, N.; Dittrich, M.; Dorioz, J.-M. Scientists’ warning to humanity: Rapid degradation of the world’s large lakes. J. Great Lakes Res. 2020, 46, 686–702. [Google Scholar] [CrossRef]
  23. Wang, H.; Qin, F.; Xu, C.; Li, B.; Guo, L.; Wang, Z. Evaluating the suitability of urban development land with a Geodetector. Ecol. Indic. 2021, 123, 107339. [Google Scholar] [CrossRef]
  24. Wang, W.; Yang, Y. Spatial-temporal differentiation characteristics and driving factors of China’s energy eco-efficiency based on geographical detector model. J. Clean. Prod. 2024, 434, 140153. [Google Scholar] [CrossRef]
  25. Chen, Y.; Guo, D.; Cao, W.; Li, Y. Changes in Net Primary Productivity and Factor Detection in China’s Yellow River Basin from 2000 to 2019. Remote Sens. 2023, 15, 2798. [Google Scholar] [CrossRef]
  26. Zhang, W.-C.; Wu, W.; Li, J.-W.; Liu, H.-B. Climate and topography controls on soil water-stable aggregates at regional scale: Independent and interactive effects. Catena 2023, 228, 107170. [Google Scholar] [CrossRef]
  27. Zeng, W.; Wan, X.; Lei, M.; Gu, G.; Chen, T. Influencing factors and prediction of arsenic concentration in Pteris vittata: A combination of geodetector and empirical models. Environ. Pollut. 2022, 292, 118240. [Google Scholar] [CrossRef]
  28. Xia, H.; Yuan, S.; Prishchepov, A.V. Spatial-temporal heterogeneity of ecosystem service interactions and their social-ecological drivers: Implications for spatial planning and management. Resour. Conserv. Recycl. 2023, 189, 106767. [Google Scholar] [CrossRef]
  29. He, J.; Pan, Z.; Liu, D.; Guo, X. Exploring the regional differences of ecosystem health and its driving factors in China. Sci. Total Environ. 2019, 673, 553–564. [Google Scholar] [CrossRef]
  30. Berihun, M.L.; Tsunekawa, A.; Haregeweyn, N.; Meshesha, D.T.; Adgo, E.; Tsubo, M.; Masunaga, T.; Fenta, A.A.; Sultan, D.; Yibeltal, M. Exploring land use/land cover changes, drivers and their implications in contrasting agro-ecological environments of Ethiopia. Land Use Policy 2019, 87, 104052. [Google Scholar] [CrossRef]
  31. Yang, K.; Chen, T.; Ao, T.; Zhang, X.; Zhou, L.; Gao, D. Response of runoff in the upper reaches of the Minjiang River to climate change. J. Water Clim. Chang. 2022, 13, 260–273. [Google Scholar] [CrossRef]
  32. Meng, X.; Gao, X.; Li, S.; Lei, J. Spatial and temporal characteristics of vegetation NDVI changes and the driving forces in Mongolia during 1982–2015. Remote Sens. 2020, 12, 603. [Google Scholar] [CrossRef]
  33. Zhai, H.; Wang, M.; Shen, D.; Hu, B.; Li, Y. Analysis of runoff variation and driving factors in the Minjiang River Basin over the last 60 years. J. Water Clim. Chang. 2022, 13, 3675–3691. [Google Scholar] [CrossRef]
  34. Daham, A.; Han, D.; Rico-Ramirez, M.; Marsh, A. Analysis of NVDI variability in response to precipitation and air temperature in different regions of Iraq, using MODIS vegetation indices. Environ. Earth Sci. 2018, 77, 389. [Google Scholar] [CrossRef]
  35. Xiong, C.; Ma, H.; Liang, S.; He, T.; Zhang, Y.; Zhang, G.; Xu, J. Improved global 250 m 8-day NDVI and EVI products from 2000–2021 using the LSTM model. Sci. Data 2023, 10, 800. [Google Scholar] [CrossRef]
  36. Zhong, S.; Zhang, K.; Bagheri, M.; Burken, J.G.; Gu, A.; Li, B.; Ma, X.; Marrone, B.L.; Ren, Z.J.; Schrier, J. Machine learning: New ideas and tools in environmental science and engineering. Environ. Sci. Technol. 2021, 55, 12741–12754. [Google Scholar] [CrossRef] [PubMed]
  37. Carrera-Hernandez, J.J. Not all DEMs are equal: An evaluation of six globally available 30 m resolution DEMs with geodetic benchmarks and LiDAR in Mexico. Remote Sens. Environ. 2021, 261, 112474. [Google Scholar] [CrossRef]
  38. Hao, X.; Qiu, Y.; Jia, G.; Menenti, M.; Ma, J.; Jiang, Z. Evaluation of global land use–land cover data products in Guangxi, China. Remote Sens. 2023, 15, 1291. [Google Scholar] [CrossRef]
  39. Anees, S.A.; Zhang, X.; Shakeel, M.; Al-Kahtani, M.A.; Khan, K.A.; Akram, M.; Ghramh, H.A. Estimation of fractional vegetation cover dynamics based on satellite remote sensing in Pakistan: A comprehensive study on the FVC and its drivers. J. King Saud Univ. Sci. 2022, 34, 101848. [Google Scholar] [CrossRef]
  40. Wang, C. Vegetation coverage and its correlation with topographic factors in upstream watershed of Minjiang River. Bull. Water Conserv. 2018, 38, 69–75. [Google Scholar]
  41. Zhou, J.; Deitch, M.J.; Grunwald, S.; Screaton, E. Do the Mann-Kendall test and Theil-Sen slope fail to inform trend significance and magnitude in hydrology? Hydrol. Sci. J. 2023, 68, 1241–1249. [Google Scholar] [CrossRef]
  42. Blewitt, G.; Kreemer, C.; Hammond, W.C.; Gazeaux, J. MIDAS robust trend estimator for accurate GPS station velocities without step detection. J. Geophys. Res. Solid Earth 2016, 121, 2054–2068. [Google Scholar] [CrossRef] [PubMed]
  43. Yacoub, E.; Tayfur, G. Trend analysis of temperature and precipitation in Trarza region of Mauritania. J. Water Clim. Change 2019, 10, 484–493. [Google Scholar] [CrossRef]
  44. Li, R.; Chen, N.; Zhang, X.; Zeng, L.; Wang, X.; Tang, S.; Li, D.; Niyogi, D. Quantitative analysis of agricultural drought propagation process in the Yangtze River Basin by using cross wavelet analysis and spatial autocorrelation. Agric. For. Meteorol. 2020, 280, 107809. [Google Scholar] [CrossRef]
  45. Uyan, M.; Ertunç, E. Investigating the impact of urban growth on land use using spatial autocorrelation methods in Konya/Türkiye. Environ. Monit. Assess. 2024, 196, 740. [Google Scholar] [CrossRef] [PubMed]
  46. Hu, X.; Ma, C.; Huang, P.; Guo, X. Ecological vulnerability assessment based on AHP-PSR method and analysis of its single parameter sensitivity and spatial autocorrelation for ecological protection—A case of Weifang City, China. Ecol. Indic. 2021, 125, 107464. [Google Scholar] [CrossRef]
  47. Sergeev, A.; Buevich, A.; Baglaeva, E.; Shichkin, A. Combining spatial autocorrelation with machine learning increases prediction accuracy of soil heavy metals. Catena 2019, 174, 425–435. [Google Scholar] [CrossRef]
  48. Chen, J.; Ma, H.; Yang, S.; Zhou, Z.; Huang, J.; Chen, L. Assessment of Urban Resilience and Detection of Impact Factors Based on Spatial Autocorrelation Analysis and GeoDetector Model: A Case of Hunan Province. ISPRS Int. J. Geo-Inf. 2023, 12, 391. [Google Scholar] [CrossRef]
  49. Zhu, L.; Meng, J.; Zhu, L. Applying Geodetector to disentangle the contributions of natural and anthropogenic factors to NDVI variations in the middle reaches of the Heihe River Basin. Ecol. Indic. 2020, 117, 106545. [Google Scholar] [CrossRef]
  50. Ren, D.; Cao, A. Analysis of the heterogeneity of landscape risk evolution and driving factors based on a combined GeoDa and Geodetector model. Ecol. Indic. 2022, 144, 109568. [Google Scholar] [CrossRef]
  51. He, Y.; Pan, H.; Wang, R.; Yao, C.; Cheng, J.; Zhang, T. Research on the cumulative effect of multiscale ecological compensation in river basins: A case study of the Minjiang River Basin, China. Ecol. Indic. 2023, 154, 110605. [Google Scholar] [CrossRef]
  52. Sun, W.; Tian, Y.; Mu, X.; Zhai, J.; Gao, P.; Zhao, G. Loess landslide inventory map based on GF-1 satellite imagery. Remote Sens. 2017, 9, 314. [Google Scholar] [CrossRef]
  53. Xiong, Y.; Li, Y.; Xiong, S.; Wu, G.; Deng, O. Multi-scale spatial correlation between vegetation index and terrain attributes in a small watershed of the upper Minjiang River. Ecol. Indic. 2021, 126, 107610. [Google Scholar] [CrossRef]
  54. Ma, F.; Jiang, Q.o.; Xu, L.; Lv, K.; Chang, G. Processes, potential, and duration of vegetation restoration under different modes in the eastern margin ecotone of Qinghai-Tibet Plateau. Ecol. Indic. 2021, 132, 108267. [Google Scholar] [CrossRef]
  55. Wang, H.; Peng, P.; Kong, X.; Zhang, T.; Yi, G. Vegetation dynamic analysis based on multisource remote sensing data in the east margin of the Qinghai-Tibet Plateau, China. PeerJ 2019, 7, e8223. [Google Scholar] [CrossRef] [PubMed]
  56. Hu, Y.; Yu, Z.; Fang, X.; Zhang, W.; Liu, J.; Zhao, F. Influence of mining and vegetation restoration on soil properties in the eastern margin of the Qinghai-Tibet Plateau. Int. J. Environ. Res. Public Health 2020, 17, 4288. [Google Scholar] [CrossRef] [PubMed]
  57. Gao, H.; Liu, F.; Yan, T.; Qin, L.; Li, Z. Drainage density and its controlling factors on the eastern margin of the Qinghai–Tibet Plateau. Front. Earth Sci. 2022, 9, 755197. [Google Scholar] [CrossRef]
  58. Liu, S.; Gu, Y.; Wang, H.; Lin, J.; Zhuo, P.; Ao, T. Response of Vegetation Coverage to Climate Drivers in the Min-Jiang River Basin along the Eastern Margin of the Tibetan Plat-Eau, 2000–2022. Forests 2024, 15, 1093. [Google Scholar] [CrossRef]
  59. Wei, Y.; Zhou, P.; Zhang, L.; Zhang, Y. Spatio-temporal evolution analysis of land use change and landscape ecological risks in rapidly urbanizing areas based on Multi-Situation simulation—A case study of Chengdu Plain. Ecol. Indic. 2024, 166, 112245. [Google Scholar] [CrossRef]
  60. Yu, E.-x.; Zhang, M.-f.; Jiang, Z.-w.; Xu, Y.-l.; Deng, S.-y. Spatiotemporal dynamics of soil erosion and associated influencing factors in the upper Minjiang River Watershed. Res. Soil Water Conserv. 2023, 30, 1–10. [Google Scholar]
  61. Wu, Y.; Pu, W.; Dong, J.; Dai, W.; Wang, Y. Analysis of Land Use Gravity Center Change and Carbon Emission Impact in Chengdu Plain of China from 2006 to 2022. Land 2024, 13, 873. [Google Scholar] [CrossRef]
  62. Zhang, L.; Peng, W.; Zhang, J. Assessment of Land Ecological Security from 2000 to 2020 in the Chengdu Plain Region of China. Land 2023, 12, 1448. [Google Scholar] [CrossRef]
  63. Yu, E.; Zhang, M.; Hou, Y.; Jiang, Z.; Deng, L.; Zhang, S.; Yang, C.; Xu, Y.; Deng, S. Assessing Environmental Quality Dynamics and its Response to Vegetation Change in the Upper Minjiang River Watershed by Modis and Spot Products. In Proceedings of the 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS, Brussels, Belgium, 11–16 July 2021; pp. 6472–6475. [Google Scholar]
  64. Jiang, S.; Chen, X.; Smettem, K.; Wang, T. Climate and land use influences on changing spatiotemporal patterns of mountain vegetation cover in southwest China. Ecol. Indic. 2021, 121, 107193. [Google Scholar] [CrossRef]
  65. Xu, H.; Li, T.-B.; Chen, J.-N.; Liu, C.-N.; Zhou, X.-h.; Xia, L. Characteristics and applications of ecological soil substrate for rocky slope vegetation in cold and high-altitude areas. Sci. Total Environ. 2017, 609, 446–455. [Google Scholar] [CrossRef] [PubMed]
  66. Yang, J.; El-Kassaby, Y.A.; Guan, W. The effect of slope aspect on vegetation attributes in a mountainous dry valley, Southwest China. Sci. Rep. 2020, 10, 16465. [Google Scholar] [CrossRef] [PubMed]
  67. Wang, R.; Wang, Y.; Yan, F. Vegetation growth status and topographic effects in frozen soil regions on the Qinghai–Tibet plateau. Remote Sens. 2022, 14, 4830. [Google Scholar] [CrossRef]
  68. Al Balasmeh, O.I.; Karmaker, T. Effect of temperature and precipitation on the vegetation dynamics of high and moderate altitude natural forests in India. J. Indian Soc. Remote Sens. 2020, 48, 121–144. [Google Scholar] [CrossRef]
  69. Özkan, U.; Gökbulak, F. Effect of vegetation change from forest to herbaceous vegetation cover on soil moisture and temperature regimes and soil water chemistry. Catena 2017, 149, 158–166. [Google Scholar] [CrossRef]
  70. Zou, L.; Tian, F.; Liang, T.; Eklundh, L.; Tong, X.; Tagesson, T.; Dou, Y.; He, T.; Liang, S.; Fensholt, R. Assessing the upper elevational limits of vegetation growth in global high-mountains. Remote Sens. Environ. 2023, 286, 113423. [Google Scholar] [CrossRef]
  71. Sekar, K.C.; Thapliyal, N.; Pandey, A.; Joshi, B.; Mukherjee, S.; Bhojak, P.; Bisht, M.; Bhatt, D.; Singh, S.; Bahukhandi, A. Plant species diversity and density patterns along altitude gradient covering high-altitude alpine regions of west Himalaya, India. Geol. Ecol. Landsc. 2024, 8, 559–573. [Google Scholar] [CrossRef]
  72. Liu, Y.; Huang, T.; Qiu, Z.; Guan, Z.; Ma, X. Effects of precipitation changes on fractional vegetation cover in the Jinghe River basin from 1998 to 2019. Ecol. Inform. 2024, 80, 102505. [Google Scholar] [CrossRef]
  73. Anwar, S.A.; Mamadou, O.; Diallo, I.; Sylla, M.B. On the influence of vegetation cover changes and vegetation-runoff systems on the simulated summer potential evapotranspiration of tropical Africa using RegCM4. Earth Syst. Environ. 2021, 5, 883–897. [Google Scholar] [CrossRef]
  74. Liu, X.; Yang, W.; Zhao, H.; Wang, Y.; Wang, G. Effects of the freeze-thaw cycle on potential evapotranspiration in the permafrost regions of the Qinghai-Tibet Plateau, China. Sci. Total Environ. 2019, 687, 257–266. [Google Scholar] [CrossRef] [PubMed]
  75. Wang, Y.; Li, Z.; Xue, J.; Si, L.; Xu, C. Moisture conditions are limiting evapotranspiration changes of alpine mountains of Qilian Mountains. Hydrol. Process. 2024, 38, e15256. [Google Scholar] [CrossRef]
  76. Li, X.; Pan, Y. The Impacts of Drought Changes on Alpine Vegetation during the Growing Season over the Tibetan Plateau in 1982–2018. Remote Sens. 2024, 16, 1909. [Google Scholar] [CrossRef]
  77. Huang, K.; Zhang, Y.; Zhu, J.; Liu, Y.; Zu, J.; Zhang, J. The influences of climate change and human activities on vegetation dynamics in the Qinghai-Tibet Plateau. Remote Sens. 2016, 8, 876. [Google Scholar] [CrossRef]
  78. Wei, Y.; Lu, H.; Wang, J.; Wang, X.; Sun, J. Dual influence of climate change and anthropogenic activities on the spatiotemporal vegetation dynamics over the Qinghai-Tibetan plateau from 1981 to 2015. Earths Future 2022, 10, e2021EF002566. [Google Scholar] [CrossRef]
Forests 16 00142 g001
Figure 1. Location of the MRB study area, (a) Specific location on the Tibetan Plateau, (b) DEM, (c) Land use.
Figure 1. Location of the MRB study area, (a) Specific location on the Tibetan Plateau, (b) DEM, (c) Land use.
Forests 16 00142 g001
Forests 16 00142 g002
Figure 2. Spatial distribution of the drivers in 2015.
Figure 2. Spatial distribution of the drivers in 2015.
Forests 16 00142 g002
Forests 16 00142 g003
Figure 3. (a) Proportion of each FVC type from 2000 to 2020; (b) temporal trend of FVC variation.
Figure 3. (a) Proportion of each FVC type from 2000 to 2020; (b) temporal trend of FVC variation.
Forests 16 00142 g003
Forests 16 00142 g004
Figure 4. Spatial pattern of different classes of FVC on the eastern margin of the Tibetan Plateau, 2000–2020.
Figure 4. Spatial pattern of different classes of FVC on the eastern margin of the Tibetan Plateau, 2000–2020.
Forests 16 00142 g004
Forests 16 00142 g005
Figure 5. FVC spatial transfer area distribution (a) from 2000 to 2010; (b) from 2010 to 2020.
Figure 5. FVC spatial transfer area distribution (a) from 2000 to 2010; (b) from 2010 to 2020.
Forests 16 00142 g005
Forests 16 00142 g006
Figure 6. Trends in FVC and their significance.
Figure 6. Trends in FVC and their significance.
Forests 16 00142 g006
Forests 16 00142 g007
Figure 7. FVC global spatial autocorrelation.
Figure 7. FVC global spatial autocorrelation.
Forests 16 00142 g007
Forests 16 00142 g008
Figure 8. FVC localized spatial autocorrelation LISA aggregation distribution.
Figure 8. FVC localized spatial autocorrelation LISA aggregation distribution.
Forests 16 00142 g008
Forests 16 00142 g009
Figure 9. FVC factor detection results.
Figure 9. FVC factor detection results.
Forests 16 00142 g009
Forests 16 00142 g010
Figure 10. Interaction test results of vegetation cover drivers in different years (NE indicates nonlinear enhancement, BE indicates two-factor enhancement).
Figure 10. Interaction test results of vegetation cover drivers in different years (NE indicates nonlinear enhancement, BE indicates two-factor enhancement).
Forests 16 00142 g010
Forests 16 00142 g011
Figure 11. Significance statistics for differences in the impact of each driver.
Figure 11. Significance statistics for differences in the impact of each driver.
Forests 16 00142 g011
Forests 16 00142 g012
Figure 12. Statistical findings for various FVC types or ranges for every factor.
Figure 12. Statistical findings for various FVC types or ranges for every factor.
Forests 16 00142 g012
Forests 16 00142 g013
Figure 13. Comparison of true and regression values of FVC: (a) SVM, (b) BP, (c) LSTM, (d) RF.
Figure 13. Comparison of true and regression values of FVC: (a) SVM, (b) BP, (c) LSTM, (d) RF.
Forests 16 00142 g013
Table 1. Types of factor interactions.
Table 1. Types of factor interactions.
Types of InteractionsInteraction Types
Nonlinear weakened q ( X 1 ∩ X 2 ) < M i n ( q ( X 1 ) , q ( X 2 ) )
Univariate weakened M i n ( q ( X 1 ) , q ( X 2 ) ) < q ( X 1 ∩ X 2 ) < M ax ( q ( X 1 ) , q ( X 2 ) )
Bivariate enhanced q ( X 1 ∩ X 2 ) > M a x ( q ( X 1 ) , q ( X 2 ) )
Independent q ( X 1 ∩ X 2 ) = q ( X 1 ) + q ( X 2 )
Nonlinear enhanced q ( X 1 ∩ X 2 ) > q ( X 1 ) + q ( X 2 )
Table 2. Appropriate range or type of each driver.
Table 2. Appropriate range or type of each driver.
Driving DriversAppropriate Range or TypeAverage FVC
Soil type (X3)Eluvial0.778
Vegetation type (X4)Broadleaved forest0.703
DEM (X5)2782–35170.801
Temperature (X8)6.69–120.789
Precipitation (X9)890–10400.733
Potential evapotranspiration (X10)859–9820.766
Land use type (X11)Forest0.879
Drought index (X14)1.47–1.840.811
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Liu, S.; Zhou, L.; Wang, H.; Lin, J.; Huang, Y.; Zhuo, P.; Ao, T. Development of Fractional Vegetation Cover Change and Driving Forces in the Min River Basin on the Eastern Margin of the Tibetan Plateau. Forests 2025, 16, 142. https://doi.org/10.3390/f16010142

AMA Style

Liu S, Zhou L, Wang H, Lin J, Huang Y, Zhuo P, Ao T. Development of Fractional Vegetation Cover Change and Driving Forces in the Min River Basin on the Eastern Margin of the Tibetan Plateau. Forests. 2025; 16(1):142. https://doi.org/10.3390/f16010142

Chicago/Turabian Style

Liu, Shuyuan, Li Zhou, Huan Wang, Jin Lin, Yuduo Huang, Peng Zhuo, and Tianqi Ao. 2025. "Development of Fractional Vegetation Cover Change and Driving Forces in the Min River Basin on the Eastern Margin of the Tibetan Plateau" Forests 16, no. 1: 142. https://doi.org/10.3390/f16010142

APA Style

Liu, S., Zhou, L., Wang, H., Lin, J., Huang, Y., Zhuo, P., & Ao, T. (2025). Development of Fractional Vegetation Cover Change and Driving Forces in the Min River Basin on the Eastern Margin of the Tibetan Plateau. Forests, 16(1), 142. https://doi.org/10.3390/f16010142

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop