Vegetation Dynamics and Responses to Climate Variations and Human Activities in the Basin of the Yarlung Tsangpo, Lhasa, and Nianchu Rivers in the Tibetan Plateau

Abstract

Terrestrial ecosystem vegetation are vulnerable to the joint impacts of human activities and climate change, particularly in ecologically fragile areas such as the Tibetan Plateau. Identifying vegetation cover changes and distinguishing their driving factors are crucial for ecological conservation in this region. This study utilized MODIS normalized difference vegetation index (NDVI) data from 2000 to 2019, combined with trend analysis (univariate linear regression and the Mann–Kendall test), partial correlation analysis, and residual analysis methods, to investigate the spatial and temporal dynamics of vegetation cover and its responses to climate change and human activities in the Yarlung Tsangpo River, Lhasa River, and Nianchu River Basin (YLN Basin) on the Tibetan Plateau. The results revealed significant differences in vegetation dynamics both in summer and the growing season: the average summer NDVI showed a significant decreasing trend during the study period, whereas the growing season NDVI exhibited no significant overall temporal trend, which highlighted the necessity of assessing vegetation dynamics seasonally to accurately capture their interannual complexity. Partial correlation analysis indicated that precipitation was the key limiting climatic factor for vegetation growth in this region, with its positive influence covering over 90% of the land area during summer and over 60% during the growing season. The residual analysis further indicated the dual and spatially heterogeneous roles of human activities: on the one hand, positive impacts, primarily from vegetation restoration projects, promoted NDVI increases in some areas; on the other hand, negative impacts, such as continuous grazing pressure, population growth, and associated land use changes, inhibited vegetation development in other areas. This study quantitatively assessed the combined effects of climate variability and complex human activities on the vegetation NDVI in the YLN Basin, emphasizing that the development of adaptive management and effective vegetation restoration strategies must fully consider seasonal differences, the key climatic limiting factor (water availability), and the spatial heterogeneity of human impacts to promote sustainable development in this ecologically fragile region.

1. Introduction

Vegetation, as one of the primary components of terrestrial ecosystems, is essential for material and energy cycling and serves as a link among ecosystems, human activities, and climate change [1,2]. Climatic patterns and human activities are considered to be two fundamental drivers influencing the spatial and temporal distribution patterns of vegetation [3,4]. Climate controls how vegetation grows over time, influences its geographical distribution, and determines the type and extent of regional ecosystem services. However, human activities can significantly shift the spatial extent and quality of vegetation growth, which has considerable impacts on the services and functions of surrounding ecosystems [5,6]. Therefore, understanding the interactions among vegetation dynamics, climatic conditions, and human activities is a major focus of global change research [7]. Particularly in highly sensitive ecological regions such as the Tibetan Plateau, an in-depth investigation into the mechanisms, processes, and consequences of these interactions is critically important [8,9,10].

Vegetation cover is a key indicator of changes in regional vegetation and the ecosystem, as it can characterize vegetation growth, identify environmental changes, and quantify ecological quality to some extent [11]. Therefore, variations in vegetation cover can reflect changes in terrestrial ecosystems [12]. In recent years, satellite remote sensing imagery has become an important source of information for monitoring large-scale vegetation dynamics. Photosynthetic capacity, leaf area index (LAI), biomass, and net primary productivity (NPP) exhibit strong correlations with the normalized difference vegetation index (NDVI) obtained from sources such as GIMMS or MODIS. Furthermore, NDVI can more accurately reflect changes in plant growth and is highly sensitive to the dynamic changes of surface vegetation [13]. NDVI is a remote sensing index, constructed based on the distinct spectral contrast between the strong absorption of the red-light bands by vegetation chlorophyll and the high reflectance of the near-infrared spectrum by leaf cellular structures [14]. Its calculation formula, (NIR − Red)/ (NIR + Red), effectively amplifies this difference, enabling the index to quantitatively reflect the density, health status, and photosynthetic activity of surface vegetation cover [15]. Consequently, the spatial and temporal dynamics of NDVI are widely used to monitor vegetation phenology, growth conditions, and vegetation responses to environmental stress, which can more accurately reflect the dynamic changes in surface vegetation growth [16,17,18]. Moreover, leveraging its advantages for large-scale, long-term time series monitoring, NDVI enables researchers to effectively track vegetation development, estimate ecosystem status, and explore in-depth investigations into the underlying driving mechanisms within data-scarce plateau environments [19].

The evolution of plateau ecosystems, in particular, is governed by complex interactions between climate change and human activities. Climatic factors such as regional rapid warming and shifts in precipitation patterns intertwine with multiple anthropogenic drivers, such as adjustments in grazing pressure, land use/land cover change (LUCC), the implementation of ecological projects, and urban expansion, resulting in synergistic, antagonistic, or additive effects [20,21]. Such interactions are superimposed upon the environmental heterogeneity caused by the complex topography of a plateau and exhibit significant differences at different spatial and temporal scales, which may trigger non-linear responses or even threshold shifts in the ecosystem state [22]. Therefore, accurately identifying and quantifying the relative contributions of various natural and anthropogenic drivers to vegetation changes in plateaus, while also characterizing the complexity and heterogeneity of vegetation responses, constitutes a core scientific challenge in current research [23].

To address this challenge, researchers have employed a variety of analytical techniques, including linear regression, Theil–Sen trend analysis, the Mann–Kendall test, and wavelet analysis, to examine NDVI change patterns across various temporal and geographical scales, based on datasets such as NOAA/AVHRR, MOD13Q1 NDVI, and others [24,25]. Concurrently, the residual trend method based on multiple regression has also been utilized to investigate the relationship between human activities and NDVI [26,27,28,29,30,31,32]. Vegetation growth and soil moisture variations are directly influenced by the spatial and temporal differences in key climatic parameters (temperature and precipitation) [33,34]. Seasonal temperature and precipitation are considered to be primary variables for exploring changes in vegetation cover and its climatic responses [35,36]. Human activities also strongly influence NDVI variations, which are often more geographically localized compared to the pervasive influence of climate on vegetation [37]. Human activities such as overgrazing, urbanization, and population expansion can have negatively impacted vegetation growth. However, there is limited research concerning the effects of ecological restoration and environmental construction projects [38].

The Tibetan Plateau, possessing unique topography and thermal forcing, exerts significant influence on regional and even global climate systems. Concurrently, it serves as a ‘sentinel’ for terrestrial ecosystem responses to climate change [39,40]. The Yarlung Tsangpo, Lhasa, and Nianchu River (YLN) basin, as the economic, political, religious, and cultural center of the plateau, features an extremely fragile ecological environment that is highly sensitive to climate change and human activities. Under the pressures of global warming and accelerating urbanization, the ecological and environmental problems faced by this region are becoming increasingly prominent, subsequently constraining sustainable socio-economic development of the area. Notably, the Tibetan Plateau is experiencing rapid warming at a rate significantly higher than the global average, accompanied by alterations in precipitation patterns [41,42]. These climatic changes profoundly affect vegetation growth and distribution within the basins. Meanwhile, human activities, such as traditional grazing, land use/land cover change (LUCC) driven by rapid urbanization, and large-scale ecological restoration projects implemented in recent years [43,44], are also exerting increasingly complex and intense influences on the vegetation development of the YLN Basin. Numerous NDVI-based studies have documented a greening trend of the vegetation in parts of the plateau, attempting to attribute it to climate warming or ecological conservation measures [45]. However, other studies have revealed regional vegetation degradation or complex, non-linear responses constrained by factors such as precipitation fluctuations and overgrazing [46,47].

Therefore, in the crucial YLN Basin region, systematically examining the interrelationships among vegetation, climate change, and human activities is of critical importance for regional and ecological environmental protection and vegetation restoration. The primary aim of this study is to provide a comprehensive understanding of the recent (2000–2019) vegetation dynamics within this ecologically vital and sensitive area by dissecting the complex interplay between natural and anthropogenic drivers. To achieve this, the specific objectives of this study were (1) to analyze the spatial and temporal variation characteristics of NDVI to reveal the patterns of vegetation change; (2) to determine whether the growing season NDVI and the summer NDVI respond in similar or comparable ways to human activities and climate change, thereby highlighting the crucial seasonal dimensions of the ecosystem response; and (3) to quantify the respective or distinct impacts of human activities and climatic factors on vegetation dynamics, allowing for the differentiation of key driving forces. By integrating these analyses, this research contributes a nuanced assessment of vegetation response to climate variability (particularly water limitations) and spatially heterogeneous human activities (including degradation pressures and restoration efforts) in the YLN Basin. This work underscores the necessity of seasonal-scale analysis and provides critical insights for developing targeted and adaptive management and effective ecological conservation strategies for sustainable development in this fragile plateau environment.

2. Materials and Methods

2.1. Study Region

The YLN Basin of Tibet (87°40′–92°37′ E, 28°60′–30°30′ N) is situated on the Tibetan Plateau’s southern region (Figure 1), with the Yarlung Tsangpo River as the axis, starting from Sangri County in the Shannan Region in the east and Lhazhi County in the Shigatse Region in the west. It reaches the Tsangnam River Basin in the south. The total length is about 245.5 km, involving 18 administrative regions (counties), and is a wide valley area of the southern Tibetan plain, consisting of extra-high mountains, high mountains, and river valley basins. The narrow river valley area, which covers 66,700 km², has a high altitude in the west and a low altitude in the east on average. There is an obvious vertical distribution of vegetation along the altitude. The mountain tops consist mostly of the sparse vegetation of flowing rocky beaches, affected by alpine freezing and thawing, with low vegetation coverage. The mountainsides are dominated by alpine meadows with high coverage and the foothills are dominated by alpine grasslands with some drought-tolerant shrubs and grasslands, and most of the river valley basins are distributed with agricultural land and artificial forests. The study region’s climate is classified as a temperate monsoon semi-arid zone of the plateau, with average annual temperatures of 5 to 10 °C, relative humidity averages of 40 to 60%, and precipitation of 400 to 800 mm, which mainly occurs in summer. Dramatic seasonal variations in precipitation, distinct dry and wet seasons, and an abundance of thermal and water resources in the valley make the study region a typical ecologically fragile region [48].

2.2. Data Sources and Preprocessing

2.2.1. NDVI Data

The National Tibetan Plateau Data Center’s monthly normalized difference vegetation index (NDVI) product (http://data.tpdc.ac.cn) contains the NDVI information for the years 2000–2023. This product is derived from the Aqua/Terra-MODIS satellite sensors’ MOD13Q1 data and land use data with a 250 m spatial resolution. The data synthesis method employed the monthly maximum value composite (MVC) approach [49]. For this study, data from 2000 to 2019 were selected. The downloaded MOD13Q1 data were preprocessed using the MODIS Reprojection Tool (MRT), which included tasks such as data format conversion and projection transformation.

The maximum value composite (MVC) was utilized to gather the seasonal variation NDVI statistics by separating them into summer and the growing season [50]. To lessen the effects of the soil background and boost the signal from the green vegetation, areas with NDVI 0.1 (non-vegetated areas) were removed from this study [51,52].

2.2.2. Meteorological Data

Utilizing regular rainfall and temperature statistics from the China Meteorological Science Data Sharing Service [53], along with ASTER GDEM V2 digital elevation data from the geospatial data cloud platform [54], the Anusplin method was utilized to spatially interpolate the information from 23 weather observation stations in and around the geographical research area for the purpose of calculating the seasonal temperature and rainfall [55].

2.3. Data Analysis

2.3.1. Trend Analysis and Mann–Kendall Test

The Theil–Sen median slope estimation provides a robust, non-parametric approach for quantifying the magnitude of trends in time series data. It works by calculating slopes between all possible pairs of data points in a series and uses the median of these slopes as the overall trend estimate [56]. A key advantage over simple linear regression is its robustness to outliers and does not require the data to follow a normal distribution [57,58]. In this study, the Theil–Sen slope was employed to compute the rate of change in NDVI for every pixel from 2000 to 2019. Thus, the specific rate and direction of the change, such as increase or decrease in the vegetation cover, was revealed. The calculation formula is as follows:

$${\mathrm{s}}_{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}=\mathrm{M}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{n}({\displaystyle \frac{{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{j}}-{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{i}}}{\mathrm{j}-\mathrm{i}}})\hspace{1em}0<\mathrm{I}<\mathrm{j}<\mathrm{n}$$

(1)

where NDVI_j and NDVI_i denote the NDVI values in year j and year i of the image element, respectively, and n denotes the length of the time series. When S_NDVI > 0, it shows an upward trend in the NDVI; when S_NDVI < 0, it shows a downward trend in the NDVI.

The Mann–Kendall (MK) test is also a non-parametric statistical procedure employed to determine monotonic trends within time series data [59]. It assesses the likelihood of the existence of a trend by comparing the relative magnitude of the values at all time points in the series to construct a statistic. The key advantages of the MK test include its independence from the assumptions of data normality and its considerable robustness to both missing data and outliers [60,61]. In this study, the MK test was applied in conjunction with the Theil–Sen slope method with the aim of assessing the statistical significance (e.g., p < 0.05) of the computed NDVI trends for individual pixels, thereby facilitating the differentiation of true vegetation change patterns from stochastic variability. The formula for calculating the value of the statistic Z in the Mann–Kendall method is as follows.

$$\mathrm{Z}=\left\{\begin{array}{c} {\displaystyle \frac{\mathrm{S}-1}{\sqrt{\mathrm{v}\mathrm{a}\mathrm{r}(\mathrm{S})}}} (S>0)\\ 0 (S=0)\\ {\displaystyle \frac{\mathrm{S}+1}{\sqrt{\mathrm{v}\mathrm{a}\mathrm{r}(\mathrm{S})}}} (S<0)\end{array}\right.$$

(2)

Among them,

$$\mathrm{S}=\sum _{\mathrm{i}=1}^{\mathrm{n}-1}\sum _{\mathrm{j}=\mathrm{i}+1}^{\mathrm{n}}\mathrm{s}\mathrm{g}\mathrm{n}({\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{j}}-{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{i}})$$

(3)

$$\mathrm{s}\mathrm{g}\mathrm{n}({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}})=\left\{\begin{array}{c}1 {\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{j}}-{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{i}}>0\\ 0 {\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{j}}-{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{i}}=0\\ -1 {\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{j}}-{\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}}_{\mathrm{i}}<0\end{array}\right.$$

(4)

$$\mathrm{v}\mathrm{a}\mathrm{r}(\mathrm{S})={\displaystyle \frac{\mathrm{n}(\mathrm{n}-1)(2\mathrm{n}+5)}{18}}$$

(5)

where the trend of NDVI is separated into significant increase (β > 0.0005 |Z| > 1.96), significant decrease (β < −0.0005, |Z| > 1.96), non-significant increase (β > 0.0005, |Z| < 1.96) at a 0.05 confidence level, non-significant decrease (β < −0.0005, |Z| < 1.96) at a 0.05 confidence level, and stable (−0.0005 < β < 0.0005).

2.3.2. Partial Correlation Analysis

From 2000 to 2019, the correlations between NDVI, temperature, and precipitation were investigated using partial correlation analysis. With a confidence level of 0.05, the partial correlation analysis t-test was calculated. The calculation formula is as follows:

$${\mathrm{r}}_{\mathrm{a}\mathrm{b}}={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})({\mathrm{y}}_{\mathrm{i}}-\overline{\mathrm{y}})}{\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({{\mathrm{x}}_{\mathrm{i}}-\overline{\mathrm{x}})}^2{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({{\mathrm{y}}_{\mathrm{i}}-\overline{\mathrm{y}})}^2}}}$$

(6)

where n is the number of years, and x_i and y_i are the value of the independent variables x and y in the ith year, respectively. $\overline{\mathrm{x}}$ and $\overline{\mathrm{y}}$ are the means of the values of the independent variables x and y for the study period.

2.3.3. Hurst Exponent Method

In this paper, future trends in vegetation change were predicted using the Hurst exponent method and TS trend analysis [62]. The Hurst index (H) is a measure of the long-term dependence of a time series and is usually computed by rescaled range (R/S) analysis [63,64]. Future trends were expected to be the inverse of historical trends when the Hurst exponent H was between 0 and 0.5. The estimated future trend coincided with the historical pattern when H was in the range of 0.5 and 1. Stronger continuity was suggested by H closer to 1, while stronger discontinuity was indicated by H closer to 0 [57]. The long-term dependency of the time series when H = 0.5 may be reflected by the rescaled range (R/S) analysis, which was established by Hurst in the middle of the 20th century. The time series data were a mutually independent random series with uncertain future trends.

2.3.4. Residual Analysis

Residual analysis provides a standard method for separating the effects of different drivers on a target variable. The basic idea began with building a statistical model that predicted NDVI using primary natural drivers including temperature and precipitation. This model was then used to predict the NDVI values attributable solely to these climate variables. The difference between the observed NDVI and the predicted NDVI constituted the residual series, representing NDVI changes not accounted for by the climate model. A trend analysis, employing Theil–Sen/MK methods, was then applied to these residuals. In this research, residual analysis served to quantitatively isolate the relative contributions of climate change versus human activities to NDVI dynamics. Assuming that temperature and precipitation were the main natural drivers, the residual trend was interpreted as the net effect of human activities on vegetation. Positive residual trends implied human activities enhanced vegetation growth relative to climatic expectations, whereas negative trends suggested the human-induced inhibition of vegetation growth [65].

Seven levels of individual activity-induced plant NDVI impacts were identified according to the NDVI residuals: Significant inhibition, Moderate inhibition, Slight inhibition, Basically no impact, Slight promotion, Moderate promotion, and Significant promotion [66] (

Table 1. Classification of the impacts of climatic change and human activities on vegetation NDVI (10⁻³ a⁻¹).

Slope	Influence Level
<−2.0	Significant inhibition
−2.0~−1.0	Moderate inhibition
−1.0~−0.2	Slight inhibition
−0.2~0.2	Basically no impact
0.2~1.0	Slight promotion
1.0~2.0	Moderate promotion
≥2.0	Significant promotion

).

3. Results

3.1. Spatiotemporal Changes in Vegetation Dynamics

Spatially, the multi-year average vegetation NDVI showed a clear pattern of higher values in the east decreasing towards the west for both summer (Figure 2a) and the growing season (Figure 2b). This spatial distribution highly coincided with the spatial distribution patterns of precipitation, confirming that moisture was the key factor limiting the growth of vegetation in the region. The NDVI in the majority of land areas ranged from 0.2 and 0.6 (73.93% for summer and 71% for the growing season).

Temporally, the basin-average summer NDVI showed a significant decreasing trend (p = 0.001) from 2000 to 2019 with an average NDVI of 0.41 (Figure 2c), while the growing season NDVI exhibited no significant overall trend (p = 0.6198) with an average NDVI of 0.53 (Figure 2d). This variability in summer and growing season NDVI trends highlighted the complexity of analyzing only a single time period, such as the annual average and summer, which may not fully capture vegetation dynamics, and that trends during the growing season, a critical period for vegetation activity, may be affected by a combination of changes in different months or phenological periods, resulting in an overall trend that is not significant. In particular, the growing season NDVI declined until 2010 and then remained almost constant except for in 2015, suggesting that there may be drivers like climatic fluctuations or anthropogenic intervention policies that changed over the study period.

There were strong spatial patterns of change trends in the vegetation NDVI for both summer and the growing season (Figure 3). Quantitively, the summer NDVI, in more than 80% of land areas, experienced a decreasing trend, with 31.66% reaching a significant level (

Table 2. NDVI change trend statistics.

β	Z	NDVI Trend Types	Area Ratio (%)
			Summer	Growing Season
β < −0.0005	|Z| > 1.96	Significant decrease	31.65	3.18
β < −0.0005	|Z| < 1.96	Non-significant decrease	50.42	31.18
β > 0.0005	|Z| < 1.96	Non-significant increase	7.93	39.56
β > 0.0005	|Z| > 1.96	Significant increase	1.85	7.63
−0.0005 < β < 0.0005	Z	Stable	8.15	18.45

) which widely distributed in the study areas. This suggested that the decline of vegetation during the summer months is a generalized phenomenon. However, the NDVI increased in about 50% of the land area during the growing season and remained stable in 18.45% of the land area. These areas were mainly located along the Yarlung Tsangpo, Lhasa, and Nianchu rivers.

The average Hurst index of the research area varied from 0.088 to 0.911 with an average of 0.45 for summer and varied from 0.098 to 0.978 with an average of 0.54 for the growing season. The Hurst index, in more than 76% of land areas, was lower than 0.5 (Figure 4a), indicating an inverse to the historical trends, while more than 23% of the land area remained consistent with historical trends (Figure 4b and

Table 3. Persistence of the change trend of the NDVI in the study region from 2000 to 2019.

Change Patterns	Area Ratio (%)
	Summer	Growing Season
Continuous significant decrease	11.22%	7.46%
Continuous non-significant decrease	8.12%	0.99%
Uncertain	76.19%	76.26%
Continuous significant increase	0.62%	2.11%
Continuous non-significant increase	1.95%	8.93%
No significant change	1.90%	4.26%

).

3.2. Changing Trends of Climatic Factors

The YLN Basin exhibits a significant diversity of climate zones that is strongly associated with elevation. These zones transition from subtropical conditions in the downstream areas to temperate and subarctic zones in the midstream reaches, culminating in alpine climates at the highest elevations [67]. This climatic gradient is a primary driver of vegetation distribution. Annual precipitation within the basin displays a pronounced spatial pattern, with downstream eastern regions receiving significantly higher amounts (>2000 mm) compared to the upstream western regions (around 200 mm) (Figure 5a) [68]. This marked spatial differentiation in precipitation directly influences regional water availability, subsequently shaping the east–west differences in vegetation types. The overall precipitation trend showed a slight decrease (Figure 5c,e), and evidence suggests an intensification of drought conditions since the early 21st century [69], indicating complex hydroclimatic variability within the basin. The majority of the basin’s rainfall was concentrated in the summer months, consistent with the influence of the Indian Summer Monsoon [70]. This seasonal precipitation pattern dictates the timing of vegetation growth and water availability. The air temperature generally decreased from the southeast to the northwest across the basin (Figure 5b). Furthermore, the temperature consistently decreased with increasing elevation. This vertical temperature gradient is a fundamental factor controlling vegetation types and their productivity. Temperature trends from 2000 to 2019 indicated no significant overall trend across the entire YLN Basin, although a significant increase was observed between 2012 and 2019 (Figure 5f).

3.3. NDVI Responses to Temperature and Precipitation

The correlation coefficients between the NDVI and temperature varied between −0.86 and 0.90 in summer and between −0.87 and 0.88 in the growing season. Spatially, positive correlations were widespread, particularly during the growing season where over 50% of the area showed a positive relationship (Figure 6d,

Table 4. Correlation significance between the NDVI and temperature/precipitation from 2000 to 2019.

Correlations	Area Ratio (%) of Temperature		Area Ratio (%) of Precipitation
	Summer	Growing Season	Summer	Growing Season
Significant negative correlation	3.68%	2.27%	1.27%	1.75%
Significant positive correlation	56.96%	44.51%	5.38%	9.77%
No significant negative correlation	36.24%	50.60%	<0.05%	26.47%
No significant positive correlation	3.12%	2.62%	93.35%	62.00%

), although significant positive correlations were less common. This suggested that, where other factors like water availability are not limiting, increased accumulated temperature can indeed promote vegetation’s photosynthetic activity and growth. However, the northern part of the study area showed the strongest negative correlation between the summer NDVI and temperature (about 40% negatively correlated, Figure 6a, Table 4), while the central and southern parts of the study area showed the largest negative correlation (about 53% negatively correlated, Figure 6c, Table 4) during the growing season. This negative correlation, especially during the growing season, may reflect the fact that increased evapotranspiration accompanying higher temperatures may exacerbate water stress over a range of temperatures, especially in the areas or periods of relatively deficient precipitation, thereby inhibiting vegetation growth.

Similarly, the correlation values for precipitation throughout the growing season with the summer NDVI ranged from −0.86 to 0.86 and from −0.83 to 0.88 with the growing season NDVI. Crucially, precipitation showed a predominantly positive correlation with the NDVI across the basin. The summer NDVI was positively correlated with precipitation for an overwhelming 93.35% of the study area, which was widely distributed (Figure 7a,b; Table 4). The growing season NDVI was positively correlated with precipitation for more than 62% of the land area and was widely distributed over the study area except in the northeast (Figure 7c,d; Table 4). This result strongly suggested that precipitation, representing water availability, is the main positive climatic factor driving vegetation growth and determining vegetation greenness in this predominantly semi-arid region, especially during the relatively dry summer and across most of the growing season. Improved moisture availability generally promoted vegetation greenness. The growing season NDVI was negatively correlated with precipitation in the rest of the region (about 28%, Table 4), mainly in the northeastern part of the study area (Figure 7c,d). This may have occurred in specific alpine and high humidity environments where excess precipitation leads to excessively wet soil, reduced insolation, or coupling with low temperatures that are detrimental to vegetation growth.

3.4. The Impact of Human Activities on the Vegetation NDVI

In addition to climate, vegetation in the YLN Basin was also influenced by human activities. Multiple regression residual trend analysis showed that human activities had a significant and spatially variable influence on the vegetation NDVI (Figure 8). This suggests that on top of the climatic context, human activities have become another key force shaping vegetation patterns and dynamics in the region, and that their impacts are not unidirectional, but rather facilitative and inhibitory. The vegetation NDVI was suppressed by human activities in 43% of the land areas during the summer and 39% during the growing season (Figure 8). Of this, 22.2% of the land area was significantly suppressed during the summer and 13.2% during the growing season, and these areas were primarily located in the northern portion of the study area. The significantly suppressed areas in the north may be related to the high intensity of grazing activities, especially in the summer pastures, where excessive gnawing and trampling exceeded the carrying capacity of the grassland. In addition, some infrastructure construction or resource development activities may also lead to vegetation destruction. The vegetation NDVI on about 50% of the land areas during the summer and the growing season was promoted by human activities. However, the vegetation NDVI in 27% of the summer and 18% of the growing season land areas was significantly promoted by human activities. These areas were usually located on the banks of rivers. The significant contribution on both sides of the river likely reflected the effectiveness of ecological restoration projects, such as reforestation, fallowed land and grassland, as well as active anthropogenic interventions such as irrigated agriculture and rational land management, which were more pronounced in the river valleys where hydrothermal conditions were relatively favorable.

4. Discussion

4.1. Spatiotemporal Patterns of NDVI

Spatially, the vegetation NDVI in the YLN Basin showed a decreasing pattern from east to west during both the summer and growing seasons, which was consistent with the climatic pattern of the YLN Basin. Climatic conditions, especially higher precipitation (Figure 2, Figure 3 and Figure 5) was more favorable for plant growth in the east compared to the west, confirming that moisture was one of the dominant factors influencing vegetation distribution patterns across the Tibetan Plateau [71,72]. However, one of the key findings of this study was the significant difference between the summer and the growing season NDVI trends: the summer NDVI showed a clear downward trend during the study period, while the growing season NDVI did not show a consistent temporal change [73]. This finding not only contrasted with the overall greening trend of the Tibetan Plateau reported in some studies but also highlighted the high spatial heterogeneity of vegetation response in the context of global climate change [57]. More importantly, it emphasized that relying only on annual or single seasonal NDVI values may not be able to fully capture the complexity of the plateau ecosystem’s response to environmental change, and highlighted the need for multi-seasonal or higher temporal resolution analyses, as interannual or single-seasonal averages may mask critical intra-seasonal dynamics and interannual variability [38]. We attributed this difference in seasonal trends in part to the variability of precipitation within the study area (Figure 5), i.e., fluctuating or decreasing trends in precipitation over the study period had a more direct negative impact on summer vegetation growth, which was more dependent on in-seasonal moisture inputs.

4.2. Effects of Climate on Vegetation NDVI

In the context of global climate change, the annual precipitation in the study area showed a slight decrease in fluctuation (Figure 5), while the mean annual temperature showed no significant trend (Figure 5), which was different from some macro-trends of “warming and humidification” in the Tibetan Plateau [74]. The bias correlation analysis further confirmed that precipitation was the key limiting climatic factor in regulating vegetation growth in the YLN Basin. The positive correlation between NDVI and precipitation for more than 90% of the summer season and more than 60% of the growing season (Figure 7 and Table 4) clearly indicated that water availability was a fundamental constraint in determining vegetation growth in the region [75]. In contrast, the effect of temperature on the NDVI showed a more complex seasonal pattern. More than 60% of the regional NDVI was positively correlated with temperature in the summer, while this proportion dropped to about 47% in the growing season (Figure 6 and Table 4). This suggested that in the summer, when water was relatively abundant and the hydrothermal conditions are good, the increase in temperature promoted the photosynthesis and growth of vegetation, while in the whole scale of the growing season, the limitation of water was more prominent, and the effect of temperature was relatively weaker. Comparison of the present results with other high-elevation or arid and semi-arid regions revealed both regional commonalities and differences: compared with the Qilian Mountains in northwestern China, the latter had a decrease in NDVI at lower elevations due to decreasing precipitation and increasing temperatures, and an increase at higher elevations driven mainly by increasing temperatures [76]. This contrasts with the positive “precipitation-NDVI” relationship that prevailed over much of the YLN Basin, and highlights that even within similar highland environments, the dominant control of vegetation by climatic factors may vary depending on the region-specific hydrothermal mix and altitudinal gradient. In contrast to Nepal, another Himalayan country, there was also a significant positive correlation between the NDVI and temperature in some parts of the country, which was consistent with the findings in the YLN Basin. However, the NDVI was negatively correlated with rainfall in some parts of Nepal [38], which was different from the positive “rainfall-NDVI” relationship prevailing in the YLN Basin. Such differences may suggest regional differences in rainfall patterns such as intensity, frequency, soil water-holding capacity, or the adaptation strategies of vegetation types to water stress. Considering that the southwestern Tibetan Plateau was likely to face “warmer and drier” climate conditions in the future [72,77], sustained warming accompanied by reduced or more volatile precipitation may increase the risk of drought [78]. This risk will pose a serious challenge to the vegetation in the YLN basin, where moisture was the main limiting factor, and may inhibit growth or even exacerbate desertification. Therefore, future regional development and ecological conservation strategies need to fully consider the water resource challenges posed by climate change, like by promoting water-saving agriculture practices, improving irrigation systems, or developing adaptive management measures, to mitigate the impacts of potential droughts and maintain ecosystem stability and productivity.

4.3. The Effects of Human Activities on Vegetation Growth

Our study found both favorable (~50% of the land area) and deleterious (43%) effects of anthropogenic activities on the vegetation NDVI (Figure 8), a finding that was similar to previous studies [79] and suggested that anthropogenic activities have had a profound impact on vegetation dynamics in the YLN Basin. The analysis of anthropogenic-specific indicators (Figure 9) provided a deeper perspective for understanding this dual impact.

On the one hand, positive human impacts, promoting vegetation growth, were observed in approximately half of the study area. Figure 9a showed that forested areas from ecological restoration projects increased significantly since 2000 (from 965 km² to 2019 km²) [80], while grassland restoration projects implemented around 2007 also contributed to vegetation growth [81]. This was highly consistent with our finding that the areas of significant NDVI promotion observed in Figure 8 (especially reaching 27% in the summer and 18% in the growing season) were mainly distributed in the riverine areas. This suggested that targeted artificial vegetation restoration measures, such as reforestation and grassland rehabilitation, were a key positive factor driving vegetation improvement in specific locations. Furthermore, as noted in our initial results, the stability or increase in the NDVI observed along the Yarlung Tsangpo, Lhasa, and Nianchu rivers during the growing season (Figure 4b,d) likely reflects the benefits derived from these ecological restoration efforts, potentially combined with active anthropogenic interventions such as irrigated agriculture and rational land management, which are more pronounced and effective in the river valleys where hydrothermal conditions are relatively favorable.

On the other hand, the equally significant areas of NDVI suppression in Figure 8 (22.2% in summer and 13.2% in the growing season), especially in the northern portion of the study area, reflected ongoing anthropogenic pressures. Although the total number of livestock (fitted by LOWESS) shown in Figure 9b fluctuated overall but did not show a sharp increase during the study period, considering the fragility of alpine meadow ecosystems, even the existing grazing intensity, especially intensive grazing in summer and the growing season [82], may exceed the carrying capacity of some areas, leading to vegetation degradation [83]. This was consistent with the significant suppression we observed in the northern rangelands. More importantly, Figure 9c,d clearly demonstrate the continued population growth during the study period (from about 1.87 million in 2000 to 3.1 million in 2019) which was a significant proportion of the total population of Tibet [84], as well as the rapid expansion of impervious surface area (from 62 km² in 2000 to 360 km² in 2020) [85]. Not only did population growth directly increase the demand for natural resources and intensify land-use pressures, but also rapid urbanization and infrastructure development led to the direct loss of vegetation and changes in the nature of the land surface. This is in comparison with the Lake Tahoe region of Sierra Nevada, USA, which is also an alpine environment and faces environmental pressures from rapid tourism development, population growth and settlement expansion, which are already having an impact on its vegetation cover [86]. This example provides a cautionary analogy of the possible ecological consequences of a growing population and development pressures in the YLN Basin.

Taken together, Figure 9 reveals the inherent complexity of anthropogenic impacts in the YLN Basin: active ecological restoration measures coexist with negative pressures such as continued grazing pressure, a growing population, and land use/land cover changes driven by them. The spatial pattern of the NDVI changes observed in Figure 8 was in fact a combination of these different directions and intensities of human activities interacting and superimposing on each other spatially, and were moderated by local climatic conditions, especially moisture limitation [87,88]. This remarkable heterogeneity and complexity of anthropogenic impacts highlights the need to implement more refined, spatialized, and typological management measures when formulating regional sustainable development and ecological conservation strategies in the future. The focus of future management should be on consolidating and expanding ecological restoration gains, as well as targeting the identification and mitigation of anthropogenic disturbance pressures caused by high-intensity grazing and rapid urbanization. At the same time, adaptation to climate change must be integrated into long-term planning to synergistically promote the overall stability and resilience of regional ecosystems.

4.4. Limitations and Prospects

Although this study provided valuable insights into understanding vegetation dynamics and its drivers in the YLN Basin, a few limitations still existed. The MODIS NDVI data used in this study had a spatial resolution of 250 m. Although suitable for regional-scale analysis, the data may not have been able to capture vegetation changes at smaller scales and finer details, especially in areas with complex topography or highly fragmented land use. Future studies may try to integrate higher resolution remote sensing data for nested analysis to reveal more microscopic details of vegetation dynamics. The NDVI suffered from saturation problems in areas with high vegetation cover, which may have underestimated the actual greenness of the most lushly vegetated areas. In addition, the NDVI is sensitive to soil background and atmospheric conditions. In the future, other vegetation indicators that more directly reflect the photosynthetic activity of vegetation, such as solar-induced chlorophyll fluorescence (SIF), can be introduced for comparative validation to improve the accuracy of monitoring. Finally, the residual analysis method, although commonly used, was based on the assumption that the climate impacts and anthropogenic impacts were considered relatively independent and of a linear superposition, which may be an oversimplification in reality. There were complex interactions and feedback mechanisms between climate change and human activities. The use of more advanced attribution models could be explored in the future to quantify more precisely the contributions of different drivers.

5. Conclusions

This study systematically examined the spatial and temporal dynamics of the NDVI and its drivers in the YLN Basin on the Tibetan Plateau from 2000 to 2019. It was found that vegetation changes in the basin showed significant seasonal differences, with the NDVI exhibiting a significant downward trend in summer and no overall temporal pattern in the growing season, revealing that vegetation monitoring relying only on a single season or an annual average NDVI may not accurately capture the complex response of plateau ecosystems to environmental changes and highlighting the necessity of seasonal-scale analyses. The partial correlation analysis confirmed that precipitation was the key limiting climatic factor regulating vegetation growth in the region, and its positive effect dominated most of the region, suggesting that moisture availability was the underlying limiting condition across the basin, likely exerting the strongest control in the drier western parts. At the same time, human activities had a significant dual impact on the vegetation NDVI and strong spatial heterogeneity; positive impacts from ecological restoration and potentially agriculture were evident along river valleys, while negative pressures from factors like grazing and urbanization likely influenced other areas, particularly the northern pastoral regions and around settlements. Therefore, the observed spatial and temporal patterns of the vegetation NDVI in the YLN Basin were not the result of a single factor dominating everywhere, but rather a complex outcome of the regional climatic context interacting with these spatially intertwined and superimposed impacts of diverse human activities. In view of this spatial complexity and the combined effects of natural and anthropogenic drivers, future ecological protection and sustainable development strategies must go beyond “one-fits-all” and adopt spatially refined and differentiated management measures, targeting both the consolidation of restoration gains and the mitigation of disturbance pressures in vulnerable areas, while integrating adaptation to climate change into long-term planning to promote the health and resilience of regional ecosystems.