Spatiotemporal Dynamics and Climate–Human Drivers of Vegetation NPP in Northern Xinjiang, China, from 2001 to 2022

Abstract

Net Primary Productivity (NPP) stands as a crucial metric for evaluating the condition and performance of terrestrial ecosystems. This study focuses on northern Xinjiang, China, as the research site. By employing the Carnegie Ames Stanford Approach (CASA) model alongside meteorological data, we examined the spatiotemporal variations in vegetation NPP from 2001 to 2022. The model utilized monthly NDVI, climate drivers, and vegetation type raster data as inputs, while the Mann–Kendall test, We utilized Theil–Sen trend analysis and residual analysis to investigate how climatic factors and human activities drove NPP changes. Results show that from 2001 to 2022, vegetation NPP in northern Xinjiang generally rose with fluctuations, averaging 127.96 gC·m⁻²·a⁻¹ annually and growing linearly at 0.58 gC·m⁻²·a⁻¹. Spatially, NPP displayed a pattern of “high in the west and low in the east, high in mountainous areas and low in deserts.” High NPP areas are mainly clustered in the Ili River Valley and adjacent mountainous regions, encompassing eastern and southwestern Ili Prefecture, northern Tianshan slopes, Balq Mountains, and southern Borokunu foothills, where hydrothermal conditions are relatively advantageous. In the last 22 years, the mean temperature in northern Xinjiang showed a fluctuating upward trend, precipitation exhibited a fluctuating downward trend, and solar radiation demonstrated a significant declining trend. Partial correlation analysis revealed that, compared with temperature and solar radiation, precipitation had a stronger positive correlation with NPP. Residual analysis showed that in areas where vegetation NPP exhibited recovery, human activities were the dominant driving factor, accounting for 23.58% of the total area, whereas the influence of climate change was relatively minor. Conversely, in regions where vegetation NPP degraded, climate change exerted a greater impact than human activities. This research clarifies the combined impacts of climate and human actions on ecosystem productivity in arid areas, offering a scientific foundation and reference for ecological protection and regional carbon control in such regions. This provides a scientific basis for formulating rational response strategies to restore vegetation and enhance the quality of ecosystems in arid regions.

1. Introduction

Net Primary Productivity (NPP) denotes the organic matter yield generated by green vegetation per unit area over a specific time period. It is generally regarded as the remaining organic material after autotrophic respiration is subtracted from the total organic matter generated through photosynthesis, and can intuitively reflect the natural growth status of vegetation [1,2,3]. NPP is pivotal in studies of carbon sources/sinks and climate change, acting as a vital metric for assessing ecosystem health and sustainability [4,5,6]. Vegetation dynamics, shaped by both climate and human impacts, indicate vegetation restoration or degradation levels and significantly influence global change and carbon balance. Over the past few centuries, intensified and spatially expanding human activities have profoundly altered the natural and cultural landscapes of the world, exerting significant impacts on ecosystem processes and their societal functions [7]. Against the backdrop of global climate change, analyzing the spatiotemporal variations and driving mechanisms of regional vegetation NPP not only helps to reveal the responses of ecosystems to environmental changes, but also provides essential scientific support for ecological management, climate adaptation, and sustainable development strategies [8,9,10,11].

The major climatic drivers influencing vegetation productivity include temperature, precipitation, and solar radiation [12]. NPP estimation models primarily encompass climate productivity models, ecological process models, and light use efficiency models. Notably, climate productivity models, which focus solely on climatic factors, overlook numerous other influences on plant dry matter accumulation and vegetation–environment feedbacks, thus leading to relatively large uncertainties. Ecological process models, while mechanistically rich, are often limited by the difficulty of parameter acquisition, making large-scale NPP estimation challenging. In contrast, light use efficiency models offer higher accuracy, require relatively simple data acquisition, and can be effectively integrated with remote sensing approaches to simplify complex field measurements. In recent years, driven by advances in remote sensing and computing technologies, remote sensing–based NPP estimation methods have become increasingly mature [13]. Among them, the CASA model estimates NPP using remote sensing and meteorological data, avoiding the need for extensive parameterization while accurately simulating NPP over large areas [14]. The CASA model has been well validated in several regions and proven effective for accurately computing NPP. More importantly, it has been demonstrated that the CASA model is well-suited for estimating NPP in Xinjiang, China [15,16].

Currently, research on the distribution patterns of vegetation NPP in Northern Xinjiang spans from national-scale evaluations to more localized studies focusing on arid regions in Northwest China and specific topographic units. Liu et al. conducted an analysis of the spatiotemporal patterns, stability, and persistence of grassland NPP in China from 2000 to 2015, revealing that areas experiencing significant declines in grassland NPP were predominantly situated in northern Xinjian [17]. Li and Pan employed the CASA model for examining the spatiotemporal variations in NPP across arid regions of Northwest China from 2001–2012 [18]. Jiao et al. utilized a modified CASA model to estimate vegetation NPP in the Ili River Valley over the period of 2000–2014, further analyzing the spatiotemporal patterns and associated driving factors. Their results indicated that precipitation serves as the predominant climatic factor influencing interannual NPP variations in this region [19]. Cui et al. conducted an analysis on the spatiotemporal changes in grassland NPP within the Tarim River Basin from 2006–2016 by integrating a modified CASA model with MODIS data [20].

Xie et al. conducted a comprehensive analysis of the spatiotemporal variations in NPP across Xinjiang from 2000–2019 using a vegetation photosynthesis model. Their findings indicated that in regions significantly influenced by human activities, anthropogenic factors were the predominant drivers of NPP changes. Nevertheless, the total area where climate change served as the primary driver of NPP fluctuations was larger than the area predominantly affected by human activities [21]. Jiang et al. proposed a novel method, grounded in partial derivatives and residuals, to measure the respective contributions of climatic factors and human activities to fluctuations in NPP. Their findings revealed that heightened human activities exerted a greater influence on NPP changes than climatic factors [22]. Yang et al. explored the distinct roles of climate change and human activities in vegetation degradation and recovery in Northern Xinjiang. Their findings indicated that human activities significantly promoted vegetation restoration in the southern parts of the region. Furthermore, ecological restoration projects successfully alleviated vegetation degradation and facilitated recovery in the region [23]. Zhang et al. utilized the Miami model for investigating grassland dynamics in Xinjiang during 2000–2014. These findings revealed that increases in NPP were primarily attributed to human activities, whereas decreases in NPP were jointly influenced by both climate change and human activities, with nearly equal contributions from each factor [16]. Hou et al. used an improved CASA model to assess yearly NPP in urban agglomerations on the northern Tianshan slopes and created a machine learning model to distinguish the effects of climate change and ozone pollution on NPP changes in the area. Furthermore, they quantified the relative significance of climatic factors and human activities on NPP [24].

In summary, existing research on vegetation NPP has predominantly focused on large-scale geographic regions such as China as a whole or the arid regions of northwestern China, as well as on smaller geomorphological units such as the Tianshan Mountains, Tarim River Basin, and Ili River Valley. Northern Xinjiang, located in China’s arid and semi-arid northwest, encompasses the Junggar Basin and its surrounding mountain ranges including the Altai and Tianshan Mountains. It serves as a crucial ecological security barrier for China but faces a series of environmental challenges, including water scarcity, grassland degradation, desertification, and biodiversity loss. Although previous studies have examined NPP dynamics in Xinjiang, few have provided continuous mesoscale quantitative data for northern Xinjiang spanning 2001–2022 while explicitly distinguishing the combined effects of climatic and anthropogenic factors. Research has typically focused on single vegetation types (such as grasslands) or shorter time series. Considering the pronounced temporal and spatial heterogeneity of vegetation responses to climate change, focusing solely on grassland ecosystems is insufficient for a comprehensive understanding of the regional carbon cycle. Therefore, an in-depth understanding of the spatiotemporal variations and driving mechanisms of NPP in northern Xinjiang is of great significance for maintaining regional ecological stability and achieving sustainable resource management.

The main aims of this research can be summarized as: (1) to estimate vegetation NPP from 2001–2022 using an enhanced CASA model and to analyze its spatiotemporal distribution characteristics; (2) to examine the response of vegetation NPP in Northern Xinjiang to climate change; and (3) to quantitatively differentiate the impacts of climate change and human activities on the variations in vegetation NPP in Northern Xinjiang.

2. Material and Methods

2.1. Overview of the Study Area

The Northern Xinjiang region is located in the northern part of the Tianshan Mountains in Xinjiang., with geographical coordinates of (40°52′17″~49°10′59″ N, 79°53′10″~96°23′5″ E), with a total area of 5.98 × 107 hm² (as shown in Figure 1). This region includes major administrative divisions such as Urumqi City, Karamay City, Shihezi City, Beitun City, Changji Hui Autonomous Prefecture, Ili Kazakh Autonomous Prefecture, Kekedala City, Altay Region, Tacheng District, Bortala Mongolian Autonomous Prefecture, Turpan City, Hami City, Baiyang City, and others. Geographically, the scope of this area centers on the Junggar Basin, bordered by the Altai Mountains to the north and the Tianshan Mountains to the south, characterized by a terrain that slopes from west high to east low. The region displays a diverse geographical environment, encompassing mountains, basins, deserts, and grasslands, mountain vegetation is mainly composed of mixed forests and grasslands, while the basins are dominated by deserts and farmlands, with elevations spanning from −165 m to 5460 m. Northern Xinjiang is subject to a typical temperate continental climate, marked by pronounced arid and semi-arid characteristics. Precipitation exhibits substantial variation depending on topography and seasonality, averaging approximately 174 mm annually. Summers tend to be relatively warm, whereas winters are cold and protracted, resulting in a significant annual temperature range. The annual mean temperature is approximately 5 °C. The annual sunshine duration ranges from 2860 to 3400 h, while solar radiation varies between 5585.19–5943.73 MJ·m⁻², with an annual mean value of 5782.50 MJ·m⁻². Owing to the barriers created by the Tianshan and Altai Mountains, certain localized areas exhibit distinctive microclimatic conditions.

2.2. Data and Preprocessing

The data used in this study includes remote sensing and meteorological data from 2001 to 2022, with detailed information provided in

Table 1. Summary of the dataset.

Product	Type	Unit	Temporal Resolution	Spatial Resolution	Data Resources
MOD13Q1	NDVI	/	16 d	250 m	https://data.tpdc.ac.cn/, (accessed on 3 December 2024)
MCD12Q1	Land Cover	Class	year	500 m	https://search.earthdata.nasa.gov/, (accessed on 3 December 2024)
Temperature	TEM	°C	month	1 km	http://data.tpdc.ac.cn/, (accessed on 29 November 2024 )
Precipitation	PRE	mm	month	1 km	http://data.tpdc.ac.cn/, (accessed on 29 November 2024 )
Sunshine Duration	SD	h	day	1 km	https://data.cma.cn/, (accessed on 3 December 2024)
SRTM	DEM	m	/	90 m	https://www.resdc.cn/, (accessed on 3 December 2024)
MOD17A3HGF	NPP	gC·m−2·a−1	year	500 m	https://ladsweb.modaps.eosdis.nasa.gov/, (accessed on 3 December 2024)

. Vegetation type data were sourced from the “1:1,000,000 Vegetation Atlas of China,” made available by the Resource and Environment Science Data Center of the Chinese Academy of Sciences. Based on the research requirements, vegetation types were reclassified into categories such as cropland, urban areas, unused land, water bodies, grasslands, and forestlands. Meteorological station data were obtained from the China Meteorological Data Service Center (https://data.cma.cn/, (accessed on 29 November 2024)), including sunshine duration records. The monthly mean sunshine duration was first calculated from the daily observations, and then the Kriging spatial interpolation method was applied to generate solar radiation data. After format conversion and spatial clipping, the radiation dataset covering the Northern Xinjiang region was obtained. This dataset served as one of the key input variables in the CASA model for estimating the NPP of vegetation in Northern Xinjiang. Prior to CASA modeling calculations, all input data underwent standardized pre-processing using ArcGIS. Spatial resolution was unified by projecting all datasets into WGS 1984 UTM Zone 45 North. Land use, monthly NDVI, air temperature, precipitation, and solar radiation data were resampled to 1 km resolution to ensure consistent spatial resolution.

2.3. Method

2.3.1. CASA Model

This study utilized the CASA model improved by Zhu Wenquan to estimate vegetation NPP in Northern Xinjiang from 2001–2022 [7]. The model uses monthly Normalized Difference Vegetation Index (NDVI), climate driving factors, and annual vegetation type raster data as inputs, and is implemented using software such as ENVI 5.6, ArcGIS 10.8, and Python 2.7.0. In the CASA model, NPP estimation is determined by two variables: absorbed photosynthetically active radiation (APAR) and light use efficiency (ε) [25]. Due to its minimal parameter requirements, the model has been widely used to study and monitor global vegetation growth.

It is calculated as follows:

$$\mathrm{NPP}(\mathrm{x},\mathrm{t})=\mathrm{APAR}(\mathrm{x},\mathrm{t})\times \epsilon (\mathrm{x},\mathrm{t})$$

(1)

where NPP(x,t) (gC·m⁻²·a⁻¹) represents the vegetation NPP of pixel x at time t, and APAR(x,t) is the absorbed photosynthetically active radiation (MJ·m⁻²·d⁻¹); ε(x,t) represents the actual light energy utilization rate (gC/MJ⁻¹).

Estimation of photosynthetically active radiation absorption. APAR(x,t), which is determined by the total solar radiation (SOL) and the proportion of photosynthetically active radiation (FPAR), is calculated as follows:

$$\mathrm{APAR}(\mathrm{x},\mathrm{t})=\mathrm{SOL}(\mathrm{x},\mathrm{t})\times \mathrm{FPAR}(\mathrm{x},\mathrm{t})\times 0.5$$

(2)

where SOL(x, t) is the total solar radiation of pixel x in month t (unit: MJ·m⁻²); FPAR(x, t) is the photosynthetically active radiation absorption ratio, unitless. Among them, 0.5 is the ratio of effective solar radiation to total radiation of vegetation utilization.

Estimation of light energy utilization. The light energy utilization rate refers to the ability of the vegetation layer to absorb the incident photosynthetically active radiation per unit area and convert it into organic matter within a certain period of time. Ideally, the maximum light energy utilization rate can be reached, but under actual conditions, this maximum is affected by environmental factors, such as air temperature and moisture conditions [7], and is calculated as follows:

$$\epsilon (\mathrm{x},\mathrm{t})={\mathrm{T}}_{\epsilon 1}(\mathrm{x},\mathrm{t})\times {\mathrm{T}}_{\epsilon 2}(\mathrm{x},\mathrm{t})\times {\mathrm{W}}_{\epsilon }(\mathrm{x},\mathrm{t})\times {\epsilon }_{\max }$$

(3)

In the equation, ε_max represents the maximum light use efficiency of vegetation under ideal conditions (gC/MJ). The ε_max values for different vegetation types are based on previous studies [22,26], and the specific values are listed in

Table 2. Maximum light use efficiency values for different vegetation types.

Serial Number	Vegetation Type	εmax
1	Cropland	0.542
2	Forest land	0.475
3	Grassland	0.542
4	Water bodies	0.542
5	Urban areas	0.542
6	Unused land	0.542

. T_ε1(x,t) and T_ε1(x,t) denote the temperature stress factors affecting vegetation light use efficiency, while W_ε(x,t) is the moisture stress coefficient, representing the influence of water availability on light use efficiency.

2.3.2. Trend Analysis and Significance Tests

This study employs the Theil-Sen trend analysis and the Mann–Kendall statistical test to analyze the trends and significance of changes in NPP within the study area. The Theil-Sen median trend analysis is a robust statistical method for trend estimation, particularly effective for small sample sequences, as it is based on non-parametric statistics [27]. The formula for calculating the Theil-Sen trend is as follows:

$$\beta =\mathrm{median}\left({\displaystyle \frac{{\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}}{\mathrm{j}-\mathrm{i}}}\right),\forall \mathrm{j}>\mathrm{i}$$

(4)

In the formula, median stands for median; x_j and x_i are time series data, and 1 < i < j < n; n is the series length. β > 0 indicates that the time series is trending upward; when β < 0, it indicates that the time series is trending downward. Since β is a non-normalized parameter, it can only reflect the magnitude of the change trend of the time series itself, and the significance of the trend change cannot be judged by itself. Therefore, the significance test of the trend must be combined with the Mann–Kendall method.

The Mann–Kendall trend test can reveal the trends of specific time-series characteristic variables and detect abrupt changes in these characteristics. Moreover, it does not require the data to follow a normal distribution or assume the trend to be linear [28,29]. A significant result from the Mann–Kendall test indicates a notable trend in vegetation pixel changes within the time-series data. Conversely, a non-significant result suggests trend stability (i.e., no obvious trend). The specific calculation formula is as follows:

$$\mathrm{S}={\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}-1}{\displaystyle \sum _{\mathrm{j}=\mathrm{i}+1}^{\mathrm{n}}\mathrm{sgn}}}\left({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}\right)$$

(5)

where sgn() is a symbolic function, and the calculation formula is:

$$\mathrm{sgn}\left({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}\right)=\left\{\begin{array}{r}\hfill 1\hspace{1em} \mathrm{if} \left({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}>0\right)\\ \hfill 0\hspace{1em} \mathrm{if} \left({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}=0\right)\\ \hfill -1\hspace{1em} \mathrm{if} \left({\mathrm{x}}_{\mathrm{j}}-{\mathrm{x}}_{\mathrm{i}}<0\right)\end{array}\right.$$

(6)

The trend test is performed using the test statistic Z, and the Z-value is calculated as follows:

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

(7)

where Var is calculated as follows:

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

(8)

where n is the number of data in the sequence; m is the number of knots (recurring datasets) in the sequence; The number of duplicate data in the ith group of duplicate data.

This study uses a two-sided trend test method for a given significance level (p < 0.05 or p < 0.01); |Z| > 1.96 and |Z| > 2.58 indicate that the trend passes the significance test at the confidence levels of 0.05 and 0.01, respectively. The methods for judging the significance of the trend are shown in

Table 3. Trend categories of Mann-Kendal test.

β	Z	Trend Characteristics
β > 0	2.58 < Z	Extremely significant increase
	1.96 < Z ≤ 2.58	Significant increase
	|Z| ≤ 1.96	There were no significant changes
β < 0	1.96 < |Z| ≤ 2.58	Significantly reduced
	2.58 < |Z|	Dramatically reduced

:

2.3.3. Correlation Analysis

The second-order partial correlation was used to calculate the relationship between NPP and climatic factors (temperature, precipitation, and solar radiation). This method reflects the degree and direction of correlation between two factors while excluding the influence of the other two factors [30]. The formula for calculating the correlation coefficient is as follows:

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

(9)

where R_xy denotes the correlation coefficient; x and y represent NPP values and meteorological factors, respectively; $\overline{\mathrm{x}}$ and $\overline{\mathrm{y}}$ denotes the average of x and y; n is the length of the time series, which is the number of years 22 years.

The formula for calculating the first-order partial correlation coefficient is as follows:

$${\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_1,{\mathrm{y}}_2}={\displaystyle \frac{{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_1}-{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_2}{\mathrm{r}}_{{\mathrm{y}}_1{\mathrm{y}}_2}}{\sqrt{(1-{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_2}^2)(1-{\mathrm{r}}_{\mathrm{y}1\mathrm{y}2}^2)}}}$$

(10)

The formula for calculating the second-order partial correlation coefficient is as follows:

$${\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_1,{\mathrm{y}}_2{\mathrm{y}}_3}={\displaystyle \frac{{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_1,{\mathrm{y}}_2}-{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_3,{\mathrm{y}}_2}{\mathrm{r}}_{{\mathrm{y}}_1{\mathrm{y}}_3,{\mathrm{y}}_2}}{\sqrt{(1-{\mathrm{r}}_{\mathrm{x}{\mathrm{y}}_3,{\mathrm{y}}_2}^2)(1-{\mathrm{r}}_{{\mathrm{y}}_1{\mathrm{y}}_3,{\mathrm{y}}_2}^2)}}}$$

(11)

where r denotes the partial correlation coefficient; x and y₁ are the elements for calculating the partial correlation coefficient; y₂, y₃ are control variables. The significance test of the partial correlation coefficient was carried out by the T-test method, and the correlation was significant when p < 0.05.

2.3.4. Multiple Regression Residual Analysis

Multiple regression residual analysis separates the effect of climate change on vegetation dynamics. This method helps extract signals of NPP variation caused by human activities [31]. It uses vegetation NPP as the dependent variable and air temperature, precipitation, and solar radiation as independent variables. A multiple linear regression model is built to predict the impact of climate change on vegetation NPP (NPP_CC). Residuals (NPP_HA) are calculated by finding the difference between predicted values and remotely sensed observations (NPP_obs). These residuals represent the influence of human activities. Multiple linear regression was used to calculate the NPP_CC, and the formula is as follows:

$$\mathrm{NPPcc}=\mathrm{A}\times \mathrm{T}+\mathrm{B}\times \mathrm{P}+\mathrm{C}\times \mathrm{S}+\mathrm{D}$$

(12)

In the formula, NPP_CC is the predicted value of NPP obtained by regression analysis, T, P and S are mean temperature, precipitation and solar radiation, A is the regression coefficient between NPP and temperature, B is the regression coefficient between NPP and precipitation, C is the regression coefficient between NPP and solar radiation, and D is the regression constant term.

The residual analysis is used to separate the impacts of climate change and human activities on NPP_obs, and the formula is as follows:

$$\mathrm{N}\mathrm{P}{\mathrm{P}}_{\mathrm{HA}}=\mathrm{N}\mathrm{P}{\mathrm{P}}_{\mathrm{obs}}-\mathrm{N}\mathrm{P}{\mathrm{P}}_{\mathrm{CC}}$$

(13)

In the formula, NPP_obs is the observed value in the remote sensing image, NPP_CC is the predicted value of NPP obtained by regression analysis, and NPP_HA is the residual value between the actual value of NPP and the predicted value of NPP, that is, the impact of human activities on vegetation NPP.

2.3.5. Evaluation Methodology

The relative contributions of climate change and human activities to vegetation NPP variations were analyzed using the slope of the trend line. When Slope (NPP_HA)^c > 0, it indicates that human activities have a positive impact on NPP changes. Conversely, when Slope (NPP_HA)^c < 0, it suggests that human activities have a negative impact on NPP changes, as shown in

Table 4. Calculation of the relative contributions of climate change and human activities to vegetation NPP.

			Contribute (%)
Slope (NPPobs) a	Slope (NPPCC) b	Slope (NPPHA) c	Climatic Change	Human Activity	Dominant Factor
	>0	>0	${\displaystyle \frac{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{CC}})}{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{obs}})}}$	${\displaystyle \frac{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{HA}})}{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{obs}})}}$	Co-dominates elevated NPP
>0	>0	<0	100	0	Climate-dominated NPP increases
	<0	>0	0	100	Human activities predominate for the increase in NPP
<0	<0	<0	${\displaystyle \frac{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{CC}})}{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{obs}})}}$	${\displaystyle \frac{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{HA}})}{\mathrm{slope}({\mathrm{NPP}}_{\mathrm{obs}})}}$	Co-lead NPP reduction
	<0	>0	100	0	Climate-led NPP declines
	>0	<0	0	100	Human activities lead to a decrease in NPP

Note: Slope (NPP_obs) ^a: Trend slope of observed vegetation NPP; Slope (NPP_CC) ^b: Trend slope of NPP induced by climate change; Slope (NPP_HA) ^c: Trend slope of NPP induced by human activities.

.

3. Results

3.1. Accuracy of Simulated Values of Vegetation NPP

Using the modified CASA model and the NDVI data provided by the National Tibetan Plateau Data Center, the vegetation NPP of Northern Xinjiang from 2001 to 2022 was estimated. To verify the reliability of the estimation results, the CASA model-derived NPP values were contrasted with those of the MOD17A3 NPP product. (Figure 2). As shown in Figure 2a, the overall interannual variation trends of NPP from both datasets are highly consistent, indicating that the CASA-estimated NPP captures the temporal fluctuation characteristics of the MOD17A3 product well. Furthermore, a correlation analysis was conducted using 1827 randomly generated sampling points after removing outliers. As depicted in Figure 2b, the coefficient of determination between the NPP estimated within this research and the MODIS NPP product is R² = 0.64 (p < 0.001), demonstrating a good level of agreement and confirming the reliability of the model results.

3.2. Spatiotemporal Distribution and Variation in NPP

3.2.1. Temporal Variation Characteristics of NPP

To more comprehensively analyze NPP variation trends over the past 22 years, annual pixel-scale NPP values for the study area were aggregated, and the mean annual NPP was computed (Figure 3). From 2001–2022, the average NPP in Northern Xinjiang demonstrated a predominantly fluctuating upward trend, growing linearly at a rate of 0.58 gC·m⁻²·a⁻¹. The highest average NPP, 157.76 gC·m⁻²·a⁻¹, occurred in 2016, whereas the lowest value was observed at 109.27 gC·m⁻²·a⁻¹ in 2001. The multi-year average NPP was 127.96 gC·m⁻²·a⁻¹, with NPP values generally below this average before 2012 and above it after 2012 (Figure 3). Between 2001–2014, the NPP exhibited an increasing trend, while after 2017, the fluctuations became less pronounced.

Based on the multi-year mean NPP of different vegetation types in northern Xinjiang, significant differences were observed among various land cover types over the past 22 years (Figure 4). The annual mean NPP decreased in the following order: cropland (312.06 gC·m⁻²·a⁻¹), forest (311.51 gC·m⁻²·a⁻¹), urban areas (224.83 gC·m⁻²·a⁻¹), grassland (201.54 gC·m⁻²·a⁻¹), water bodies (130.90 gC·m⁻²·a⁻¹), and unused land (38.88 gC·m⁻²·a⁻¹). This pattern can be attributed to the fact that croplands in northern Xinjiang are mainly distributed across oasis plains with favorable irrigation conditions, where sufficient water and nutrient availability create human-optimized environments for vegetation growth. Consequently, the photosynthetic efficiency of croplands is generally higher than that of forests constrained by natural environmental factors, resulting in slightly higher NPP per unit area. The annual mean NPP of cropland, forest, urban areas, water bodies, and grassland showed certain interannual fluctuations, whereas that of unused land remained relatively stable. After 2010, the fluctuations in NPP for cropland, forest land, water bodies, and grassland became significantly more pronounced. In 2006 and 2008, the NPP for cropland and grassland decreased sharply due to extreme drought events during these years [32] which suppressed vegetation growth as a result of low precipitation. Notably, the NPP for forest land did not decline significantly in 2008 because the drought event that year was associated with sustained high temperatures. The increase in temperature somewhat benefited the NPP of forest land, offsetting the decline caused by drought conditions. The NPP for forest land, grassland, and water bodies reached their lowest levels in 2012, when solar radiation peaked during this year, coinciding with extreme drought events.

3.2.2. Spatial Distribution Characteristics of NPP

The vegetation NPP in Northern Xinjiang from 2001–2022 demonstrates pronounced spatial heterogeneity (Figure 5). The annual average vegetation NPP in Northern Xinjiang is 127.96 gC·m⁻²·a⁻¹. The spatial distribution pattern of vegetation NPP reveals higher values in the western region compared to the eastern region, with mountainous areas exhibiting significantly higher values than desert regions. Areas with high NPP values (>400 gC·m⁻²·a⁻¹) constitute 5.98% of the total vegetated area, predominantly located in the Tianshan Mountains, the eastern part of Ili, and the mountain fringe zones in the northern and southern parts, forming a semi-circular distribution. Regions with NPP values ranging between 200 and 300 gC·m⁻²·a⁻¹ make up 10.71% of the vegetated landscape, primarily concentrated into the Altai Mountains and Tianshan Mountains in the northern region. Low-value areas (<100 gC·m⁻²·a⁻¹) cover 63% of the total vegetated area, predominantly located in the arid central parts of the Junggar Basin, Turpan Basin, and Gaoshun Gobi.

3.2.3. Vegetation NPP Change Trend and Its Significance Test in Northern Xinjiang

The spatial distribution of interannual changes in vegetation NPP in Northern Xinjiang from 2001–2022 is illustrated in Figure 6. The interannual variation in NPP values ranges from −28.94 to 30.38 gC·m⁻²·a⁻¹, with a long-term regional average of 0.58 gC·m⁻²·a⁻¹. Spatially, regions experiencing an increase in NPP (β > 0) account for 63.57% of the total area, whereas areas with a decrease in NPP (β < 0) constitute 36.43%. The Mann–Kendall (MK) significance test was utilized for identifying trends of NPP increase and decrease. The largest proportion of this study area (76.99%) exhibited no significant changes, while 10.70% and 10.21% displayed highly significant and significant increases, respectively, mainly in the Altai Mountains, Ertix, Urumqi, and Ili River Basins, the Tianshan northern slope oasis, Bogda Mountains, Turpan, and parts of northern and western Hami. Areas with highly significant and significant decreases in NPP account for relatively small percentages, at 0.66% and 1.44%, respectively, scattered in urban areas like Urumqi, Yining, Altai, and near Tacheng, potentially because of urbanization.

This study reveals that the proportion of regions exhibiting an increasing trend in vegetation NPP from 2001–2022 is significantly higher than that of regions with a decreasing trend, suggesting an overall enhancement in vegetation growth conditions.

3.3. Analysis of the Driving Mechanism of Vegetation NPP Change in Northern Xinjiang

3.3.1. Characteristics of Interannual Variation in Climatic Factors

From 2001–2022, the annual average temperature in Northern Xinjiang rose irregularly, with a non-significant increase at a rate of 0.03 °C·a⁻¹. The multi-year average temperature was 5.33 °C, peaking at 6.02 °C in 2007 and bottoming at 4.29 °C in 2012. Meanwhile, the average precipitation was 173.82 mm, declining irregularly at a rate of −0.81 mm·a⁻¹, indicating a fluctuating downward trend. The highest annual total precipitation was 236.92 mm (in 2016), along with the lowest was 138.18 mm (in 2022). The annual average solar radiation was 5782.50 MJ·m⁻², with a significant decreasing trend at a rate of −6.88 MJ·m⁻²·a⁻¹ (p < 0.05)( Figure 7).

3.3.2. Correlation Analysis Between NPP and Climate Factors

NPP serving as a critical indicator for monitoring vegetation dynamics, is directly influenced by the synergistic effects of multiple environmental factors. Among these, hydrothermal conditions are regarded as primary determinants that significantly impact vegetation NPP.

The partial correlation coefficients among vegetation NPP and mean annual temperature in Northern Xinjiang from 2001–2022 ranged from 0.991 to 0.998 (Figure 8a,b). Positive correlations were observed in 80.13% of the area, with 21.38% exhibiting a significant positive correlation (p < 0.05). These areas were predominantly located in Altay Prefecture, eastern Changji, Hami City, Turpan City, and the Tianshan Mountains. Conversely, regions with a negative correlation constituted 19.87% of the total area, with merely 0.89% demonstrating a significant negative correlation (p < 0.05), scattered across the Altay Mountains and western Changji regions.

In Northern Xinjiang, the partial correlation coefficients between vegetation NPP and annual precipitation span from −0.85 to 0.94 (Figure 8c,d). Positive correlations are dominant, encompassing 76.77% of the region, with 26.54% showing significant positive correlations (p < 0.05), mainly in the Tianshan Mountains and Tacheng areas. Negative correlations account for 23.23% of the area, including significantly negative correlations (p < 0.05) that cover 1.86% of the region, which are sporadically distributed in the northeastern and western parts of Hami City.

The spatial distribution of the partial association between vegetation NPP and annual solar radiation in Northern Xinjiang is predominantly negative, forming a spatial pattern characterized by coexisting positive and negative correlations (Figure 8e,f). The partial correlation coefficients range from −0.91 to 0.90, with an average value of −0.03. Regions exhibiting positive relationships account for 44.64% of the total area, among which significantly positive correlations (p < 0.05) cover 5.66%, primarily distributed in central Hami City, the Altay region, and other areas. Conversely, regions with negative correlations constitute 55.36% of the total area, with significantly negative correlations (p < 0.05) accounting for 8.89%. These are mainly located in the Irtysh River Basin, the Ulungur River Basin, the Altai Mountains, areas north of the Tianshan Mountains, Turpan, the western Tacheng region, the northern and eastern parts of the Bortala Mongol Autonomous Prefecture, Ili, and other regions.

Overall, the annual mean temperature and precipitation in Northern Xinjiang are predominantly positively correlated with vegetation NPP, whereas annual solar radiation exhibits a predominantly negative correlation, characterized by both positive and negative correlations. Based on the spatial distribution of vegetation NPP trends and their statistical significance as depicted in Figure 8, it is evident that in regions where vegetation NPP undergoes significant changes influenced by hydrothermal conditions, the correlations between vegetation NPP and precipitation or temperature are also highly significant. This indicates that climatic elements are pivotal in driving vegetation NPP changes in Northern Xinjiang. Among these factors, precipitation has the most pronounced impact, with increased precipitation generally promoting higher vegetation NPP. However, in most areas, no statistically significant correlations are observed, indicating that vegetation NPP changes in Northern Xinjiang cannot be solely ascribed to climate change.

3.3.3. Contribution of Climate Change and Human Activities to Vegetation NPP in Northern Xinjiang

Figure 9a shows that 49.48% of the total area in Northern Xinjiang experiences positive contributions to vegetation NPP changes due to climate change. These regions are predominantly located in the Altai Mountains, Ulungur Lake, Bogda Mountain Range, Turpan Basin, northern and western parts of Hami City, western Tacheng, western Changji, eastern Bortala Mongol Autonomous Prefecture, Ili, the Ili River Basin, and the Kashgar River Basin, among others. In contrast, areas where climate change negatively affects vegetation NPP changes account for 50.52% of the total area, mainly located in the Junggar Basin, southern Hami City, the Tianshan Mountains, western Bortala Mongol Autonomous Prefecture, and eastern Changji, among other locations.

Regions in which human activities positively influence changes in NPP comprise 58.50% of the total area (as shown in Figure 9b), with these regions predominantly located in the Altai Mountains, Tianshan Mountains, Turpan Basin, central Hami City, Bogda Mountain Range, Changji, Ili, Bortala Mongol Autonomous Prefecture, and the eastern and southern parts of the Tacheng region, among others. Conversely, areas where human activities negatively affect NPP changes account for 41.50%, primarily concentrated in the Kashgar River Basin, Karamay City, Urumqi City, the western Tacheng region, northern Hami City, and the Turpan Basin, among other locations. Clearly, human activities have a greater impact on vegetation NPP in Northern Xinjiang than climatic factors.

3.3.4. The Dominant Driving Factors Influencing the NPP Change of Vegetation in Northern Xinjiang

Based on trend and contribution analyses, the dominant driving factors for vegetation NPP changes in Northern Xinjiang from 2001–2022 were spatially categorized into distinct zones (as shown in Figure 10). Regions where both climate change and human activities synergistically contributed to NPP increases accounted for 15.89%, predominantly located in the Altai Mountains, Irtysh River Basin, Ulungur River Basin, northern central Tianshan Mountains, Bogda Mountain, and Shihezi, among other areas. Regions where NPP increases were exclusively attributed to climate change covered 22.71%, primarily distributed across Ulungur Lake, northern and western Hami City, Turpan Basin, Ebinur Lake, and Karamay, among other areas. Regions where NPP increased exclusively due to human activities accounted for 23.58%, primarily located in northern and southern Hami City, southern Altay, central Tianshan, southern Tianshan, and northern Bortala Mongol Autonomous Prefecture. Regions experiencing NPP degradation due to climate change and human activities combined made up 7.91%, mainly distributed in Altay City, Urumqi City, Changji City, Yining City, Shihezi City, and urban areas such as Tacheng City. Regions where NPP degradation was attributed solely to human activities accounted for 10.89%, primarily distributed in the Barluk Mountains, Mayile Mountains, Baiyang River Basin, and the Kashgar and Ili River Basins. Regions experiencing NPP degradation solely attributed to climate change account for 19.03%, primarily located in Altay, Changji, eastern Tacheng, and Hami City, among other areas. Overall, 23.80% of regions are jointly affected by climate change and human activities; 41.73% and 34.47% are solely driven by climate change and human activities, respectively. In NPP-recovered regions, human activities dominate, with climate change having a minor impact. By contrast, in regions where vegetation NPP has degraded, climate change exerts a significantly greater influence than human activities.

4. Discussion

4.1. Temporal and Spatial Variation Characteristics and Driving Forces of Vegetation NPP

From 2001–2022, Northern Xinjiang’s vegetation NPP increased annually at a rate of 0.58 gC·m⁻²·a⁻¹. This growth stems partly from local ecological restoration initiatives and possibly from rising atmospheric CO₂ levels CO₂. Although elevated CO₂ concentrations may exert a fertilizing effect on vegetation productivity, particularly by enhancing plant growth through improved water use efficiency, this CO₂ fertilization effect may be constrained by soil water availability in arid regions subject to severe water limitations. Consequently, the impact of CO₂ on net primary production (NPP) is complex. Beyond variations in CO₂ concentration, local climatic conditions—such as precipitation, air temperature, solar radiation—and soil type also play crucial roles in influencing vegetation growth processes [33]. Furthermore, vegetation NPP in Northern Xinjiang exhibits marked spatial variation, being higher in the west than in the east and in mountainous areas compared to deserts. These findings are consistent with previous studies [34]. Zhang et al. [35] revealed that the significant spatial heterogeneity in Xinjiang is primarily attributed to variations in topography, elevation, and latitude, with precipitation contributing the most to this heterogeneity. This pattern is closely linked to the local climatic conditions and unique topographical features. Specifically, mountainous regions such as the Tianshan and Altai ranges exhibit higher Net Primary Productivity (NPP), primarily attributable to the intensification of precipitation by mountainous topography. Particularly in foothill areas and intermontane basins—such as the Ili River Valley and along the Tianshan range—relatively abundant water resources promote vegetation growth. The westerly circulation brings warm and moist air masses from the North Atlantic. When these air masses encounter topographic barriers such as the Tianshan Mountains and Altai Mountains, they rise and cool adiabatically, leading to precipitation. This process significantly enhances vegetation NPP levels along the Ili River Valley and the Tianshan Mountain ranges [36]. In contrast, the northern region of Northern Xinjiang, particularly the area centered around the Gurbantunggut Desert, exhibits relatively lower NPP due to its complex topographical characteristics. This region is enclosed by mountains on three sides and is geographically distant from oceanic moisture sources, which significantly limits water vapor transport. As a result, the NPP levels in this area remain comparatively low. In summary, the western part of Northern Xinjiang demonstrates higher NPP values compared to the eastern part [37].

This study further quantified the trends of NPP changes across different regions, revealing that areas with a notably increasing trend make up 10.70% of Northern Xinjiang’s total area. This provides a positive indication for the future health of ecosystems and vegetation restoration in the region. However, areas with a slightly decreasing trend account for 26.03%, particularly in regions with frequent human activities such as urban expansion and agricultural development, highlighting the potential threat of human activities to NPP reduction. These results align with the study conducted by Gao et al. [38], which highlighted the growing influence of human activities on ecosystems in arid and semi-arid areas.

4.2. The Influence of Climate Change on Vegetation NPP

Prior research has shown that climatic factors rank among the top influences on vegetation NPP [39]. Vegetation growth is closely associated with precipitation [40], temperature [41], and solar radiation [42]. The partial correlation analysis between vegetation NPP and climatic factors reveals that in most areas of Northern Xinjiang, vegetation NPP exhibits a positive correlation with both temperature and precipitation. This finding aligns with the results reported by Peng et al. [43], Xu et al. [44], and Yisilayili et al. [45], who also observed a stronger positive correlation between vegetation NPP and precipitation in Xinjiang, which may be attributed to the region’s arid and semi-arid climate. For instance, in high-altitude mountainous regions, rising temperatures can accelerate glacier melt, thereby benefiting vegetation growth to some extent. However, in desert areas, rising temperatures unaccompanied by increased precipitation can intensify drought, potentially impeding vegetation growth. Moreover, solar radiation’s dominant and consistent impact on the spatial distribution and variability of vegetation NPP has been validated by previous research [35].

4.3. The Contribution of Human Activities to the Changes in NPP

Beyond climatic factors, human activities constitute a significant influence on vegetation NPP, exhibiting a dual-natured impact. Negative activities such as urbanization, cultivation, and overgrazing markedly diminish vegetation NPP, whereas positive measures like ecological restoration projects promote vegetation recovery. Consequently, quantifying the effects of human activities on vegetation growth is of paramount importance. Yang et al. [23] examined the respective contributions of climate change and human activities to vegetation degradation and recovery in Northern Xinjiang from 2001 to 2010. Their findings revealed that vegetation restoration was predominantly driven by human interventions. Nevertheless, contrary to earlier studies, this investigation indicates that in recent years, climatic factors have emerged as the predominant driver of vegetation degradation in Northern Xinjiang. This is consistent with the findings of Wan et al. [46]. in their analysis of the response of net primary production (NPP) to climate change and human activities in the Xinjiang region from 2001 to 2022. This transformation can be attributed to global warming, which has resulted in elevated temperatures and diminished precipitation levels in arid and semi-arid regions. Such changes have heightened evaporation rates, exacerbating water stress on vegetation and consequently affecting plant growth and survival. Discrepancies in data sources, temporal spans, spatial scales, and analytical methodologies may account for inconsistencies in research outcomes.

The findings of this study suggest that, on the whole, the restorative impacts of human activities on vegetation significantly surpass their degradative effects. This phenomenon is likely to be closely linked to China’s ecological restoration initiatives and conservation policies enacted in recent years. Since the launch of the 13th Five-Year Plan, Xinjiang has successfully completed key ecological protection and restoration projects spanning 5166 square kilometers, reforested 3195.87 square kilometers of former agricultural land, converted 676.66 square kilometers of cropland to grassland, rehabilitated 4014 square kilometers of degraded grasslands, and established 61.8 square kilometers of comprehensive sand control demonstration forests [47,48]. These projects have significantly contributed to the recovery and enhancement of vegetation productivity in the region.

The findings of this study reveal that substantial vegetation degradation, as measured by NPP, is predominantly concentrated in regions such as Urumqi, Yining, Altay, and Tacheng. In recent years, the rapid advancement of industrialization and urbanization has resulted in a continuous increase in land allocated for construction purposes [49], thereby encroaching upon grassland and forest resources. Additionally, the exacerbation of the urban heat island effect and the intensification of drought conditions have contributed to a decline in vegetation NPP, further heightening the risk of ecological degradation [50].

According to prior research [51,52], afforestation has the potential to enhance vegetation coverage in arid and semi-arid regions. Nevertheless, as afforestation initiatives expand, soil moisture deficits may intensify. Consequently, when designing and implementing ecological restoration programs in Xinjiang, it is imperative to thoroughly evaluate the natural environmental conditions and adopt site-specific strategies. A comprehensive and multifaceted approach is crucial for ensuring the efficacy of these projects, rather than over-relying on afforestation as a singular solution.

5. Limitations and Future Research Directions

The driving factors considered in this study were relatively limited, as the analysis focused solely on temperature, precipitation, and solar radiation. Future research should aim to integrate additional natural factors (e.g., relative humidity, soil type) and anthropogenic factors (e.g., population density, gross domestic product, afforestation area) to provide a fuller evaluation of their effects on NPP.

As a key indicator in the terrestrial ecosystem carbon cycle, the monitoring and assessment of NPP constitute a long-term and continuous process. Due to limitations in data availability, there is a paucity of studies analyzing the temporal variation characteristics of NPP at monthly and seasonal scales. The CASA model utilizes remote sensing and meteorological data, yet discrepancies frequently arise in their spatial and temporal resolutions during practical use. For example, in this research, the MODIS NDVI (MOD13Q1 product) features a 250 m spatial resolution, while the meteorological data possess a 1 km spatial resolution. The conversion between datasets with differing resolutions introduces uncertainties into the CASA model’s estimations [53,54]. Future research should strengthen consistency testing and error correction across different data sources to reduce potential inconsistencies and enhance the reliability of data integration. Low-resolution meteorological data struggles to capture small-scale climatic variations (such as extreme precipitation) within microclimatic regions like mountains and basins, potentially introducing biases in NPP estimates. Furthermore, several simplifying assumptions within the CASA model fail to fully account for crucial ecological processes, such as soil moisture, plant species variation, and vegetation changes throughout the growing season. This may lead to overestimation or underestimation of NPP. For instance, the model assumes consistent effects of soil moisture and plant species on productivity, yet marked variations in carbon uptake capacity among plant species could reduce simulation accuracy. To enhance result reliability, subsequent studies should incorporate additional ecological process parameters or employ more complex models to more accurately characterize these factors’ influence on NPP.

6. Conclusions

Utilizing the CASA model, this research analyzed the spatiotemporal dynamics and underlying drivers of vegetation NPP in Northern Xinjiang between 2001 and 2022, leading to the following conclusions:

(1)

On the temporal scale, the annual variation in vegetation NPP in Northern Xinjiang from 2001–2022 demonstrated a fluctuating upward trend, with an average annual growth rate of 0.58 gC·m⁻²·a⁻¹. This indicates an overall enhancement in vegetation health during this period. Spatially, the distribution of vegetation NPP generally exhibited higher values in the western region compared to the eastern region and in mountainous areas relative to desert regions. High NPP values are primarily concentrated in the Ili River Valley and its surrounding mountainous areas, including eastern and southwestern Ili Prefecture, the northern slopes of the Tianshan Mountains, the Balq Mountains, and the southern foothills of the Borokunu Mountains. These regions exhibit relatively favorable hydrological and thermal conditions. Regions with increasing vegetation NPP (β > 0) constituted 63.57% of the total study area, while areas with decreasing NPP (β < 0) accounted for 36.43%. Notably, areas with a highly significant increase (10.70%) and a significant increase (10.21%) were primarily distributed across the Altai Mountains, the Irtysh River Basin, the Ulungur River Basin, the Ili River Basin, the Tianshan Mountains, Bogda Mountain, Turpan, and parts of northern and western Hami. Conversely, the percentage of areas with a highly significant decrease and a significant decrease in NPP was relatively small, comprising 0.66% and 1.44%, respectively.

(2)

In the past 22 years, Northern Xinjiang’s mean temperature has shown a fluctuating but rising trend, while precipitation has shown a fluctuating downward trend, and solar radiation has significantly declined. Vegetation NPP demonstrated a significant positive correlation with the annual average temperature and total annual precipitation, whereas its correlation with annual solar radiation was predominantly negative, though both positive and negative correlations coexisted. Partial correlation analysis revealed that precipitation exerted a stronger impact on vegetation NPP in Northern Xinjiang than temperature and solar radiation.

(3)

Vegetation NPP variations in Northern Xinjiang stem from both climatic shifts and human actions, with climate factors explaining 49.48% of the impact and human activities contributing slightly more, at 50.52%. In regions where vegetation NPP is recovering, human activities dominate as the primary driver, covering 23.58% of the area, with climatic factors playing a secondary role. Conversely, in areas experiencing degradation of vegetation NPP, the impact of climatic changes significantly surpasses that of human activities, underscoring the substantial challenges posed by climate change to ecological stability in Northern Xinjiang. To enhance vegetation productivity and ecosystem restoration capacity, it is recommended to prioritize water-saving and integrated water-resource management in low-NPP basins, to focus on ecological restoration and sustainable grazing or land use practices in areas where human activity drives recovery, and to monitor and evaluate restoration projects through remote sensing in order to scale successful interventions.