Introduction

Vegetation plays an indispensable role in maintaining the ecological balance of the Earth. It is not only a crucial component of ecosystems but also a significant factor influencing the global climate, water cycle, and human society1,2,3. These factors interweave to form a complex ecosystem. The Normalized Difference Vegetation Index (NDVI), an index derived from satellite remote sensing data, is widely used to monitor and evaluate dynamic changes in vegetation cover4,5,6. However, as the impacts of precipitation, temperature, and human activities have become increasingly significant, we face the challenge of developing a deeper understanding of how these changes affect vegetation cover7,8,9,10,11,12,13. This issue not only represents a scientific inquiry but also involves revealing the extent to which human factors shape the Earth’s vegetation cover.

A wealth of studies have examined the effects of natural climate change on vegetation14,15,16,17. Jiang et al.18 investigated the impact of climate change on vegetation dynamics in Central Asia, finding that precipitation is the primary factor affecting vegetation growth in Kazakhstan, while temperature primarily influences the seasonal changes in vegetation greenness in mountainous areas and the Ruhr Sea Basin. Huang et al.19 explored climate change and vegetation dynamics in the Tibetan Plateau, showing that rising temperatures and precipitation generally promote vegetation growth but with significant spatial heterogeneity. From 1986 to 2000, rising temperatures inhibited vegetation growth in arid steppe areas during periods of slight increases in precipitation. In contrast, from 2000 to 2011, decreased precipitation in meadow areas made precipitation the dominant factor influencing vegetation growth. During drought periods, temperature became the primary limiting factor for vegetation growth in Tibet. However, with the continuous increase in human activities, new challenges have emerged4,20,21.

Human activities, like climate change, significantly affect vegetation dynamics22,23,24. These activities have both positive and negative effects on vegetation cover. On the positive side, human efforts such as agricultural cultivation, landscaping, forest protection projects, and other initiatives can promote vegetation growth and restoration, thereby improving vegetation coverage5,25,26. These efforts not only protect biodiversity but also provide essential ecological services, including oxygen production, air quality improvement, and suitable habitats for various wildlife. Conversely, human activities such as urbanization, land development, overgrazing, and illegal logging have led to widespread destruction and degradation of vegetation, directly reducing vegetation coverage27,28,29. Furthermore, industrialization releases significant amounts of pollutants and greenhouse gases, accelerating the pace and extent of climate change, which in turn affects vegetation growth and distribution. The resulting land degradation, soil erosion, and desertification have become critical issues that urgently need to be addressed in the fields of environmental protection and sustainable development.

Although many studies have explored the dynamics of vegetation cover in relation to climatic factors and human activities, there is a notable lack of research focused on the vegetation dynamics across the entire Xinjiang region, particularly by geotectonic units. Furthermore, analyses of the seasonal correlation between vegetation dynamics and climate change in this region are even rarer. Additionally, previous studies have provided limited insight into the potential effects of human activities and natural disturbances on vegetation growth. This study aims to investigate the effects of precipitation, temperature, and human activities on the NDVI in Xinjiang, with particular attention to their complex spatial and temporal distribution across different geographical regions and seasons. Using remote sensing technology and geographic information systems (GIS), we will employ MK trend testing, correlation analysis, and spatial residual analysis methods, combined with detailed meteorological data and topographic information, to reveal the intricate relationships between these impacts in various geographic subregions and seasons. This comprehensive research approach will deepen our understanding of vegetation cover dynamics in Northwest China and provide a robust scientific foundation for addressing the challenges posed by climate change and ecological protection. The results of this research will offer new insights and innovative supports for policy-making and natural resource management, fostering a more harmonious relationship between human society and the natural environment.

Study area and research methods

Study area

The Xinjiang region is located in the northwest of China (73°40′ ~ 96°23′ E, 34°25′ ~ 49°10′ N), at the heart of Eurasia, and forms one of the core areas of Western China’s development (Fig. 1). The climate in Xinjiang is diverse, encompassing arid, semi-arid, and nearly rainless conditions. The region experiences significant temperature variations throughout the year, with hot and dry summers followed by cold winters. Precipitation is predominantly concentrated in the summer months, with the southern Xinjiang basin receiving more rainfall compared to the northern Xinjiang region, which includes the Tianshan Mountains and the Altai region30. The topography of Xinjiang is complex and varied, featuring mountains, plateaus, hills, and basins. The Tianshan Mountain range runs through both the northern and southern parts of the region, serving as a major topographic dividing line, with the North Foothills Basin of the Tianshan Mountains in the east and the Tarim Basin in the west31. Additionally, Xinjiang encompasses significant geographical and geomorphological elements such as the Kunlun Mountains, Altai Mountains, and the Pamir Plateau. These topographic features play a crucial role in regulating the region’s ecological environment and influencing human activities. Based on the characteristics of landforms and climate in Xinjiang, the regional geomorphology is divided into six geomorphic units: (I) Altai-Beita Mountains, (II) Western Junggar Mountains, (III) Junggar Basin, (IV) Tianshan Mountains, (V) Tarim Basin, and (VI) Kunlun—Altun Mountains. The figure was drawn by ArcGIS 10.8, the same below.

Fig. 1
figure 1

Elevation and zoning diagram of the study area (GS(2019)3333).

Full size image

Data sources

In this study, NDVI, precipitation, and temperature datasets were utilized, with a time series spanning from 2001 to 2020. The NDVI data were downloaded from the National Tibetan Plateau Data Center (https://data.tpdc.ac.cn/). This dataset is based on 250m, 16-day NDVI and pixel reliability products from the Aqua/Terra-MODIS satellite sensor MOD13Q1, along with land use data. The data processing involved several steps: preliminary reconstruction to address noise pixels in single-phase images, S-G filtering of long-time series images, retention of high-quality pixels, synthesis of 16-day composites into monthly products, and stitching for the China region. This resulted in 250-m monthly products for China from 2000 to 20223. Subsequently, ArcGIS was used for resampling to a resolution of 1000 m.

The precipitation and temperature datasets used in this study have a spatial resolution of 1000 m and cover monthly data from 1901 to 2022. These datasets include the global 0.5° climate data released by CRU and the global high-resolution climate data provided by WorldClim, both generated through Delta space downscaling. Verification was conducted using data from 496 independent meteorological observation points in China, with the results proving reliable. Annual statistics were calculated using a natural year algorithm, which sumed the data from January to December of each year. Seasonal data are collected as follows: spring from March to May, summer from June to August, autumn from September to November, and winter from December to February of the following year32,33.

Research methods

Trend analysis of NDVI and meteorological factors

The specific process of this study is shown in Fig. 2. In this study, the Mann–Kendall (M–K) trend test was employed to analyze the changes and characteristics of climate factors and NDVI. The M–K test is a rank-based nonparametric statistical method used to detect trends in time series data. It can identify various types of trends, including linear, non-linear, and cyclical trends34. In the M–K test, two key statistics were calculated: the statistic S and the standardized test statistic Zmk. The calculations for these statistics are as follows:

$$S = \sum\limits_{i = 1}^{n - 1} {\sum\limits_{j = i + 1}^{n} {{\text{sgn}} \left( {X_{j} - X_{i} } \right)} }$$
(1)
$${\text{sgn}} \left( {X_{j} - X_{i} } \right) = \left\{ {\begin{array}{*{20}c} { + 1} \\ 0 \\ { - 1} \\ \end{array} } \right.\begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} {if} \\ {if} \\ {if} \\ \end{array} \begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} {\left( {X_{j} - X_{i} } \right)} \\ {\left( {X_{j} - X_{i} } \right)} \\ {\left( {X_{j} - X_{i} } \right)} \\ \end{array} \begin{array}{*{20}c} > \\ = \\ < \\ \end{array} \begin{array}{*{20}c} 0 \\ 0 \\ 0 \\ \end{array}$$
(2)
$$\sqrt {Var\left( S \right)} = \frac{1}{18}\left[ {n\left( {n - 1} \right)\left( {2n + 5} \right) - \sum\limits_{q = 1}^{p} {t_{q} \left( {t_{q} - 1} \right)\left( {2t_{q} + 5} \right)} } \right]$$
(3)
$$Z_{mk} = \left\{ {\begin{array}{*{20}c} {\frac{S}{{\sqrt {Var\left( S \right)} }}} \\ 0 \\ {\frac{S - 1}{{\sqrt {Var\left( S \right)} }}} \\ \end{array} } \right.\begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} {if} \\ {if} \\ {if} \\ \end{array} \begin{array}{*{20}c} {} \\ {} \\ {} \\ \end{array} \begin{array}{*{20}c} S \\ S \\ S \\ \end{array} \begin{array}{*{20}c} > \\ = \\ < \\ \end{array} \begin{array}{*{20}c} 0 \\ 0 \\ 0 \\ \end{array}$$
(4)
Fig. 2
figure 2

Framework of this study.

Full size image

In Eqs. (1), (2), (3) and (4), Xi and Xj are the values of year i and year j of the time series, n is the length of the time series data, tq is the value corresponding to the q digit, and Zmk represents the trend change in the time series data.

If Zmk > 0, it indicates that the time series data show an increasing trend over time. If Zmk < 0, it indicates that the time series data show a decreasing trend over time. If |Zmk|> Z(1−a/2), then the null hypothesis of a significant trend in the time series data is not valid. The value of Z(1−a/2) can be found in the standard normal distribution. When a = 5% is the significance level, the corresponding value of Z(1−a/2) is 1.96; when a = 1% is the significance level, the corresponding value of Z(1−a/2) is 2.58.

Correlation analysis

In this study, Pearson correlation was employed to assess the relationship between NDVI and temperature and precipitation35,36. The correlation is quantified by the R value of the Pearson correlation coefficient, where a higher R value indicates a stronger correlation between the variables, and a lower R value indicates a weaker correlation. To determine the significance of the Pearson correlation coefficient, a t-test was conducted. The Pearson correlation coefficient is calculated as follows:

$$R = \frac{{\sum\limits_{i = 1}^{n} {(X_{i} - \overline{X})(Y_{i} - \overline{Y})} }}{{\sqrt {\sum\limits_{i = 1}^{n} {(X_{i} - \overline{X})^{2} } } \sqrt {\sum\limits_{i = 1}^{n} {(Y_{i} - \overline{Y})^{2} } } }}$$
(5)

In formula (5), R is the correlation coefficient; n is the calculation year (20 years); x and y represent NDVI and meteorological factors in different years.

Spatial residual analysis

Spatial residual analysis is a common method used to separate the effects of climate and human impacts. The basic process is as follows: multiple regression is carried out using climate factors and NDVI, and then the measured values are compared with the regression results37,38. The regression results are taken as climate effects, and the difference is considered as human effects. The calculation formula is:

$$NDVI_{PV} = a \times T + b \times P + c$$
(6)
$$NDVI_{HA} = NDVI_{MV} - NDVI_{PV}$$
(7)

In Eqs. (6) and (7), T represents temperature, P represents precipitation, a and b are regression coefficients, and c is the intercept. NDVIPV represents the predicted value of NDVI based on the regression model, which reflects the impact of climate change on vegetation NDVI. NDVIMV represents the measured value of NDVI, and NDVIHV represents the residual, which reflects the impact of human activities on vegetation NDVI.

Relative contribution calculation

Univariate linear regression was used to calculate the linear trend rate of NDVIPV and NDVIHA in Xinjiang from 2001 to 2020. The trend rate of NDVI change in the vegetation growing season under climate change and human influence was represented by Slope(NDVIPV) and Slope(NDVIHA), respectively. At the same time, the trend rate, Slope(NDVIMV), of NDVI was calculated, and the relative contribution rates of climate factors and human activities to vegetation NDVI change were calculated according to Table 1.

Table 1 Methods for calculating relative contribution rates of climate factors and human activities.
Full size table

Results and analysis

Analysis of temporal and spatial change of NDVI

Annual and seasonal NDVI spatial distribution and variation trends in Xinjiang are shown in Fig. 3a–j and Table 2. From 2001 to 2020, inter-annual and seasonal NDVI were affected by latitude, climate, and human activities, and they exhibit similar distribution patterns. Regions with NDVI less than 0.1 are mainly distributed in the Kunlun—Altun Mountains, Tarim Basin, and the eastern Tianshan Mountains. Areas with NDVI between 0.1 and 0.4 are primarily found in the Altai—Beita Mountains, the western mountains of Junggar, the Junggar Basin, and the Kunlun—Altun Mountains, while regions with NDVI greater than 0.4 are mainly distributed in the western Tianshan Mountains. The proportion of each NDVI range varies across different periods. The multi-year average NDVI (Fig. 3a) was less than 0.1 in 67.66% of the regions, between 0.1 and 0.4 in 30.80% of the regions, and greater than 0.4 in 1.54% of the regions. During the study period, the number of pixels with improved NDVI exceeded those with degraded NDVI. In the study area, the area of NDVI pixels increased by 91.82% and decreased by 8.18% (Fig. 3f). The annual average NDVI in spring (Fig. 3b) was less than 0.1 in 70.05% of the regions, between 0.1 and 0.4 in 28.18%, and greater than 0.4 in 1.77%. Among these, NDVI increased in 89.54% and decreased in 10.46% (Fig. 3g). The annual average NDVI in summer (Fig. 3c) was less than 0.1 in 57.50% of the regions, between 0.1 and 0.4 in 29.96%, and greater than 0.4 in 12.54%. Among these, the area of NDVI increased by 85.47% and decreased by 14.53% (Fig. 3h). In autumn, the multi-year average NDVI (Fig. 3d) was less than 0.1 in 61.80% of the regions, between 0.1 and 0.4 in 34.81%, and greater than 0.4 in 3.39%. Among these, the area of NDVI increased by 83.77% and decreased by 16.23% (Fig. 3i). The winter multi-year average NDVI (Fig. 3e) was less than 0.1 in 91.27% of the regions, between 0.1 and 0.4 in 8.71% of the regions, and greater than 0.4 in 0.02% of the regions. Among these, NDVI increased in 83.80% and decreased in 16.20% (Fig. 3j).

Fig. 3
figure 3

Multi-year mean of NDVI from 2001 to 2020 and Z-value of Mann–Kendall trend test. (a–e) annual, spring, summer, autumn and winter mean NDVI spatial distribution; (f–j) Z-values of annual, spring, summer, autumn and winter NDVI MK trend test.

Full size image
Table 2 Regression coefficients of NDVI, precipitation, and temperature meteorological factors.
Full size table

In the four seasons, the area of increased NDVI was the largest in spring and decreased the most in winter. On the scale of geographical regions (Table 2), NDVI in all subregions showed an increasing trend from 2001 to 2020. On the annual scale, region III showed the largest trend of significant (p < 0.001) increase, which was 0.0015/yr. On the seasonal scale, the trend value of region II in spring was the largest (p < 0.01), which was 0.0025/yr; the trend value of region III in summer and autumn was the largest (p < 0.001), which was 0.0032/yr and 0.0014/yr respectively; the trend value of region IV in winter was the largest (p < 0.001), at 0.0009/yr.

Analysis of temporal and spatial changes of temperature and precipitation factors

The spatial distribution and variation trend of annual and seasonal temperature and precipitation factors in Xinjiang are shown in Fig. 4a–t and Table 2. During 2001–2020, the temperature exhibited a consistent distribution pattern across different periods. Areas with temperatures below −15 ℃ were primarily located in the Kunlun—Altun Mountains. Regions with temperatures between −15 and 0 ℃ were mainly found in Altai—Beita Mountains, western Tianshan and Kunlun—Altun Mountains. Areas with temperatures above 0 ℃ were mainly distributed in the western mountains of Junggar and Tarim Basin, while regions with temperatures exceeding 15 ℃ were mainly distributed in the eastern Tianshan Mountains and western Tarim Basin. Throughout the study period, the distribution of average temperature varied by period. The annual mean temperature (Fig. 4a) was below −15 °C in 0.79% of the region, between -15 °C and 0 °C in 24.47%, and above 0 °C in 74.75%. During the study period, the temperature increased were more common than decreases. The area with an increase in temperature accounted for 58.79%, while the area with a decrease accounted for 41.21% (Fig. 4f). In spring, the annual mean temperature (Fig. 4b) was lower than -15 ℃ in 1.39% of the regions, between −15 and 0 ℃ in 22.35%, and between 0 and 15 ℃ in 67.66%. Of these, 92.14% of the regions experienced a temperature increase, while 7.86% saw a decrease (Fig. 4g). In summer, the annual mean temperature (Fig. 4c) was below 0 ℃ in 2.79% of regions, between 0 ℃ and 15 ℃ in 26.57%, between 15 and 30 ℃ in 70.38%, and above 30 ℃ in 0.26%. Among these, 53.15% of the regions experienced a temperature increase, while 46.85% experienced a decrease (Fig. 4h). In autumn, the annual mean temperature (Fig. 4d) was below -15 ℃ in 0.30% of the regions, between −15 ℃ and 0 ℃ in 23.69%, and above 0 ℃ in 76.02%. Of these, 10.07% experienced a temperature increase, and 89.93% experienced a decrease (Fig. 4i). In winter, the annual mean temperature (Fig. 4e) was below -30 ℃ in 0.18% of the regions, between −30 and −15 ℃ in 26.88%, and between −15 and 0 ℃ in 72.94%. Among these, 22.17% of the regions experienced a temperature increase, while 77.83% experienced a decrease (Fig. 4j). Among the four seasons, the largest increase in temperature occurred in spring, while the greatest proportion of temperature decreases was observed in autumn. On the scale of geographical regions (Table 2), from 2001 to 2020, the region with the highest annual temperature increase was Region I, with a trend value of 0.0255 °C/yr. On the seasonal scale, the largest temperature increase in Region I occurred in spring, with a trend value of 0.0986 °C/yr. In Region II, the largest temperature increase was observed in summer and winter, with trend values of 0.0235 °C/yr and 0.0289 °C/yr, respectively. The largest temperature decrease in Region II occurred in autumn, with a trend value of -0.0613 °C/yr.

Fig. 4
figure 4

Multi-annual mean of temperature and precipitation from 2001 to 2020 and the Z-value of Mann–Kendall trend test. (a–e) annual, spring, summer, autumn and winter mean temperature spatial distribution; (f–j) Z-values of annual, spring, summer, autumn and winter temperature MK trend test; (k–o) throughout the year, spring, summer, autumn and winter years average rainfall spatial distribution; (p–t) Z-values of annual, spring, summer, autumn and winter precipitation MK trend test.

Full size image

During 2001–2020, precipitation exhibited a consistent distribution pattern across different periods. Areas with precipitation below 200 mm were mainly distributed in Altai-Beita Mountains, Junggar Basin, eastern Tianshan, Tarim Basin, and Kunlun—Altun Mountains. Regions with precipitation between 200 and 400 mm were mainly distributed in the western mountains of Junggar and western Tianshan Mountains. Areas with more than 400 mm of precipitation were mainly distributed in the western Tianshan Mountains. Throughout the study period, the distribution of precipitation varied across different periods. The annual average precipitation (Fig. 4k) was less than 200 mm in 86.73% of regions, between 200 and 400 mm in 12.68%, and greater than 400 mm in 0.59%. During the study period, the area with increased precipitation accounted for 52.12%, while the area with decreased precipitation accounted for 47.88% (Fig. 4p). In spring, the annual average precipitation (Fig. 4l) was less than 200 mm in 86.62% of the regions, between 200 and 400 mm in 12.62% and greater than 400 mm in 0.76%. During this period, the area with increased precipitation accounted for 44.01%, while the area with decreased precipitation accounted for 55.99% (Fig. 4q). In summer, the annual average precipitation (Fig. 4m) was lower than 200 mm in 66.44% of the regions, between 200 and 400 mm in 22.12%, and above 400 mm in 11.44%. Among these, 53.07% of the regions saw an increase in precipitation, while 46.93% experienced a decrease (Fig. 4r). In autumn, the annual average precipitation (Fig. 4n) was below 200 mm in 91.17% of the regions, between 200 and 400 mm in 8.60%, and above 400 mm in 0.23%. Among these, the area with increased precipitation accounted for 45.84%, and the area with decreased precipitation accounted for 54.16% (Fig. 4s). In winter, the average annual precipitation (Fig. 4o) was below 200 mm in 99.67% of the regions, with 0.33% of the regions receiving more than 400 mm. Among these, the area with increased precipitation accounted for 23.37%, while the area with decreased precipitation accounted for 76.63% (Fig. 4t). Of the four seasons, the largest increase in precipitation occurred in summer, while the largest decrease was observed in winter. On the scale of geographical regions (Table 2), from 2001 to 2020, the region with the highest annual precipitation increase was Region IV, with a trend value of 0.6056 mm/yr. On the seasonal scale, the largest precipitation increase in spring was observed in Region I, with a trend value of 0.986 mm/yr; the largest precipitation decrease in summer occurred in Region II, with a trend value of -1.0476 mm/yr; the largest precipitation increase in autumn was in Region II, with a trend value of 3.6344 mm/yr; and the largest precipitation decrease in winter occurred in Region III, with a trend value of -0.9236 mm/yr.

Correlation analysis

The relationship between vegetation dynamics and climate factors was analyzed by calculating the correlation coefficient R between average NDVI and seasonal precipitation and temperature in Xinjiang from 2001 to 2020. The average NDVI responded differently to annual and seasonal changes in temperature and precipitation, as shown in Figs. 5, 6. The results show that the correlation coefficients between NDVI and temperature and precipitation exhibit strong spatial and temporal heterogeneity. At the annual scale (Figs. 5a,f, 6), NDVI in regions I–IV showed a positive correlation with temperature, with correlation coefficients of 0.436, 0.405, 0.362, and 0.353, respectively. In contrast, the correlation coefficients in regions V and VI were negative, at -0.063 and -0.067, respectively, which were not statistically significant (p > 0.05). NDVI was positively correlated with precipitation in regions I–V, with correlation coefficients of 0.180, 0.311, 0.011, 0.400, and 0.235, respectively. In region VI, the correlation coefficient was negative (-0.023), which also did not reach statistical significance (p > 0.05). On the seasonal scale, NDVI in spring (Figs. 5b, 6) was positively correlated with temperature in all regions, with correlation coefficients of 0.728, 0.675, 0.550, 0.787, 0.232, and 0.450, respectively, all of which were statistically significant (p < 0.05). In contrast, NDVI showed a negative correlation with precipitation in regions I, V, and VI, with correlation coefficients of -0.191, -0.092, and -0.278, respectively. In regions II–IV, the correlation coefficients were positive, at 0.023, 0.210, and 0.136, respectively, but these were not statistically significant (p > 0.05). In summer (Figs. 5c, 6), NDVI in regions I, III, and IV was positively correlated with temperature, and the correlation coefficients of 0.051, 0.281, and 0.205, respectively. In regions II, V, and VI, the correlation coefficients were negative, at -0.066, -0.043, and -0.039, respectively, and did not reach statistical significance (p > 0.05). Regarding precipitation in summer (Figs. 5h, 6), NDVI showed a positive correlation across all regions, with correlation coefficients of 0.350, 0.724, 0.168, 0.506, 0.297, and 0.163, respectively. Significant correlations (p < 0.05) were observed in regions II and IV. In autumn (Figs. 5d, 6), NDVI in regions I–III was positively correlated with temperature, and the correlation coefficients were 0.636, 0.033, and 0.007, respectively, with region I showing a significant correlation (p < 0.05). In regions IV–VI, NDVI was negatively correlated with temperature, with correlation coefficients of -0.345, -0.463, and -0.049, respectively, none of which were statistically significant (p > 0.05). Regarding precipitation in autumn (Figs. 5i, 6), a positive correlation was observed in region III, with a correlation coefficient of 0.028, while NDVI was negatively correlated with precipitation in regions I, II, IV–VI, with correlation coefficients of -0.189, -0.204, -0.079, -0.038, and -0.133, respectively, none of which were statistically significant (p > 0.05). In winter (Figs. 5e, 6), NDVI generally showed a positive correlation with temperature, although it did not reach statistical significance (p > 0.05). As for precipitation in winter (Figs. 5j, 6), there was a negative correlation between NDVI and precipitation in the all regions, and the correlation coefficients were -0.668, -0.549, -0.640, -0.556, -0.325, and -0.062, respectively. These correlations were statistically significant (p < 0.05) in regions I–IV.

Fig. 5
figure 5

Spatial distribution of correlation coefficients between annual and seasonal NDVI and temperature and precipitation. (a–e) spatial distribution of correlation coefficients between NDVI and temperature in year, spring, summer, autumn and winter; (f–j) Spatial distribution of correlation coefficients between NDVI and precipitation in year, spring, summer, autumn and winter.

Full size image
Fig. 6
figure 6

Annual and seasonal correlation coefficients of NDVI with temperature (RNDVI-MAT) and precipitation (RNDVI-MAP) in Xinjiang from 2001 to 2020. I: Altai-Beita Mountains; II: Western Junggar Mountains; III: Junggar Basin; IV: Tianshan Mountains; V: Tarim Basin; VI: Kunlun—Altun Mountains.

Full size image

Effect of human activities on NDVI

Human activities play an important role in the ecosystem of Xinjiang, and their impact on vegetation growth cannot be ignored. As shown in Fig. 7, from 2001 to 2020, the population of Xinjiang increased steadily, with the rural population declining and the urban population rising. This shift indicates an intensification of urbanization, which has been detrimental to the positive development of vegetation cover. However, since 2000, the implementation of the Three-North Shelter Forest Program and Grain for Green Project has resulted in the annual establishment of large areas of planted forests, significantly promoting the recovery and growth of vegetation cover. At the same time, the number of livestock has decreased slightly since 2005, alleviating the pressure caused by grazing to some extent. Since 2000, Xinjiang’s GDP has shown consistent annual growth, with all industries experiencing expansion. Notably, the secondary and the tertiary industry have seen substantial growth, contributing to increased industrialization. This trend, however, has had a negative impact on vegetation development to some extent.

Fig. 7
figure 7

Indicators related to human activity. Population Total population of Xinjiang, Urban population Urban population of Xinjiang, Rural population Rural population of Xinjiang, FA Forest area, GGP Afforestation area under the Grain for Green project, TNSFP Afforestation area under the Three-North Shelter Forest Program, LN Livestock number, Sheep Sheep number, GDP Gross domestic product of Xinjiang, PS Gross product of Xinjiang’s primary industry, SS Gross product of Xinjiang’s Secondary sector, TS Gross product of Xinjiang’s Tertiary industry.

Full size image

To quantify the distinction between anthropogenic influence and climate control, the annual, spring, summer, autumn, and winter NDVI residual trends were calculated using climate variables including temperature and precipitation, from 2001 to 2020 (Fig. 8a–e). On the annual scale (Fig. 8a), the area of residual reduction was 8.42%, mainly distributed in regions II, III, and IV, while the area of increase accounted for 91.58%. On the seasonal scale, the residual reduction area in spring (Fig. 8b) was 12.13%, distributed across the whole study area, with a concentration in the southeastern part of region VI, while the area of increase was 87.87%. In summer (Fig. 8c), the area of residual reduction was 14.05%, primarily in regions I–IV, with an increase area of 85.07%, and a significant (p < 0.05) increase in 0.88% of the area, mostly in regions IV-V. In autumn (Fig. 8d), the area of residual reduction was 14.18%, concentrated in regions I–IV, with the area of increase at 85.82%. In winter (Fig. 8e), the area of residual reduction was 20.20%, predominantly in regions I-V, while the area of increase was 79.80%.

Fig. 8
figure 8

Trend of annual and seasonal average NDVI residuals in Xinjiang from 2001 to 2020. Annual, spring, summer, autumn and winter trend division of spatial residual analysis. I: Altai-Beita Mountains; II: Western Junggar Mountains; III: Junggar Basin; IV: Tianshan Mountains; V: Tarim Basin; VI: Kunlun—Altun Mountains.

Full size image

The analysis of contribution degree shows (Fig. 9) that human activities have a significant impact on NDVI changes across different regions and time periods. On the annual scale, the contribution of human activities to region V was the largest (92.17%), while the contribution to region I was the smallest (77.83%). On the seasonal scale, the contribution of human activities in spring to region V was the largest (86.92%), and the contribution to region II was the lowest (68.90%). In summer, the contribution of human activities to region V remained the highest (88.65%), while the contribution to region VI was the lowest (80.35%). In autumn, the contribution to region I was the largest (80.04%), and the contribution to region II was the smallest (67.92%). In winter, the contribution to region VI was the largest (89.09%), while the contribution to region III was the smallest (59.43%). Based on the above analysis, the contribution of human activities to the change in vegetation NDVI is generally exceeds 60%, indicating a dominant influence.

Fig. 9
figure 9

Contribution rates of climate change and human activities to NDVI change in Xinjiang from 2001 to 2020. PV represents the contribution of climate change to the dynamic change of NDVI; HA represents the contribution of the dynamic change of NDVI in human activities. I: Altai-Beita Mountains; II: Western Junggar Mountains; III: Junggar Basin; IV: Tianshan Mountains; V: Tarim Basin; VI: Kunlun—Altun Mountains.

Full size image

Discussion

Temporal and spatial changes of NDVI

Xinjiang encompasses vast grasslands, deserts, and mountains. Its ecological balance is related to the survival and development of the Silk Road Economic Belt, and protecting Xinjiang’s ecological environment is crucial for maintaining global ecological balance and promoting regional sustainable development. Therefore, studying the spatial and temporal changes in vegetation cover in this region, along with its main driving forces, is of great significance. As the most commonly used vegetation coverage index, NDVI can monitor and analyze the dynamic changes in vegetation across different spatial and temporal scales, providing essential support for sustainable development and natural resource management39. The results showed that NDVI in Xinjiang has strong spatial heterogeneity, with high values mainly distributed in regions I, II, and IV, and low values mainly distributed in regions III, V, and VI. This pattern is mainly due to the presence of the Gurbantünggüt Desert and Taklamakan Desert in regions III and V, leading to low NDVI values, while the average altitude of region VI, ranging between 5500 and 6000 m, coupled with its annual snow cover, limits vegetation growth, resulting in persistently low NDVI throughout the year. In this study, from 2001 to 2020, the growth rate of NDVI in Xinjiang exhibited significant variation across different periods. The growth rate in spring and summer was larger, while that in autumn and winter was lower. However, the annual growth rate of NDVI across all regions showed a significant upward trend. Despite this, the growth rate remains lower than the national average NDVI growth rate of 0.0022/yr in China40. Relevant studies show that due to the continuous increase of human activities and the destruction of vegetation caused by the implementation of the Western Development Strategy, NDVI reached its lowest level in 2001. Since then, China’s Three-North Shelter Forest Program, natural forest protection project, and Grain for Green Project have gradually restored the ecological balance and improved the vegetation coverage10,12, leading to an increase in NDVI.

Temporal and spatial changes of temperature and precipitation factors

Temperature is a key environmental factor influencing plant growth, directly affecting the physiological activities, growth rate, and life cycle of plants41,42. The results show that the temperature distribution has strong spatial heterogeneity, with high values mainly distributed in the eastern region of III, V, and IV, and low values mainly distributed in the western region of I, II, VI, and IV. This is mainly because as altitude increases, the temperature generally decreases. This phenomenon, known as the temperature lapse rate, occurs because higher altitudes are associated with lower air pressure, which causes air molecules to spread out, reducing kinetic energy and, consequently, lowering temperature. In this study, from 2001 to 2020, average temperature trends varied significantly across different periods in Xinjiang, with the average temperature in spring and summer showed an upward trend and the average temperature in autumn and winter showed a downward trend. Furthermore, the annual average temperature changes differed across regions: regions I-IV experienced an upward trend, while regions V and VI showed a downward trend, resulting in an overall upward trend across the region. Relevant studies have shown that the average temperature in Northwest China have been rising over the past few decades, with winter temperatures remaining relatively stable and the most significant increase occurring in spring43,44. The findings of this study are largely consistent with these conclusions.

Precipitation is a key limiting factor in regulating ecosystem dynamics in arid and semi-arid regions45. The results reveal that the precipitation distribution also shows strong spatial heterogeneity, with higher values primarily concentrated in the western regions of II and IV, and lower values in other areas. This is mainly because mountains such as the Tianshan Mountains often experience more frequent and intense rainfall events, resulting in higher precipitation. In addition, precipitation patterns in Xinjiang are influenced by factors such as the Siberian High, the westerlies, and the topographic characteristics of the region46,47. The topographic uplift effect leads to local changes in precipitation, resulting in higher precipitation in the northwest mountainous areas. In this study, from 2001 to 2020, the average precipitation in Xinjiang varied considerably across different periods, with an increasing trend in summer and a decreasing trend in other seasons. Precipitation changes also varied across regions: I, IV, and V showed an increasing trend, while regions II, III, and VI showed a decreasing trend, with the whole region showed an overall increasing trend. Relevant studies show that in the past few decades, the average annual precipitation in arid and semi-arid areas of northwest China has shown an increasing trend, with the most significant increase occurring in summer43. The results of this study are consistent with this conclusion.

Based on the analysis results, the entire Xinjiang region presents a trend of “warming and wetting”. Reference48,49 have shown that, over the past 60 years, both temperature and precipitation in the arid region of Northwest China have increased significantly, indicating that this region as a whole is experiencing climate warming and humidification.

Correlation between NDVI and temperature and precipitation factors

Studying the correlation between NDVI and temperature and precipitation is essential for comprehensively understanding the impact model of climate change on vegetation dynamics, predicting ecosystem responses, providing a scientific basis for sustainable development under climate change, and promote harmonious coexistence between humans and the natural environment. The results indicate that NDVI is significantly correlated with temperature and precipitation annually. Temperature is the primary meteorological factor affecting NDVI changes in spring and autumn, while precipitation is the main factor influencing NDVI changes in summer and winter. Liu et al.50 studied the relationship between vegetation cover change and climate factors in Northwest China and found that temperature was the main factor affecting NDVI dynamics in spring and autumn, and precipitation was the primary factor in summer and winter. These findings are similar to the conclusions of this study. However, Liu et al. found a positive correlation between winter precipitation and NDVI changes, whereas this study found a significant negative correlation. This discrepancy may be attributed to differences in the specific study areas. Liu et al.’s study encompassed Shaanxi Province, Qinghai Province, Gansu Province, Ningxia Hui Autonomous Region, and Xinjiang Uyghur Autonomous Region, leading to more varied results. In this study, NDVI in regions I to IV was positively correlated with interannual temperature changes, while in regions V and VI it was negatively correlated. NDVI was positively correlated with the interannual variation of precipitation in regions I to V, negatively correlated in region VI, and positively correlated overall, though none of these correlations were statistically significant. Relevant studies suggest a weak positive correlation between vegetation NDVI and interannual changes in temperature and precipitation in northwest China51, which aligns with the findings of this study.

Effects of human activities on vegetation dynamics

Human activities are significant driving factors behind vegetation dynamics. The dynamic changes in vegetation across different periods may be influenced by various human disturbances, including grazing, afforestation, policy-driven land use changes, and ecological restoration efforts52. This study highlights the substantial role of human activities in promoting vegetation restoration in Xinjiang from 2001 to 2020. Research indicates that over half of the global vegetation restoration areas are closely linked to land cover changes directly caused by human activities. Particularly in regions with deteriorating climate conditions, human activities can mitigate or even offset the adverse impacts of climate change on vegetation22. As social economies develop and populations grow, the influence of human activities on regional vegetation cover has increasingly become a significant factor in vegetation change19. The landforms and underlying surfaces in northwest China are complex, and the region’s fragile ecological environment is highly susceptible to human activities. Since 1998, China has implemented several nationwide ecological restoration projects, such as the Three-North Shelter Forest Program and the Grain for Green Project. These initiatives have led to a gradual recovery of forest resources and grasslands16, and have effectively curbed the ongoing vegetation degradation in Northwest China. Large-scale ecological engineering is the human activity with the most significant impact on vegetation change in China53 and is a key driver of vegetation restoration. Consequently, since the early twenty-first century, human activities have emerged as the primary factors influencing vegetation restoration in Northwest China.

Research deficiencies

This study focused on NDVI changes in Xinjiang from 2001 to 2020, which limits the analysis to a relatively short time series. We did not investigate the vegetation status prior to the implementation of ecological engineering projects in 2000, nor did we account for the impacts of extreme climate events on vegetation. To offer more effective recommendations for future vegetation restoration and sustainable development in Xinjiang, further research is needed to assess the long-term sustainability of vegetation changes and the effects of extreme climatic events.

Conclusion

In this study, we analyzed the temporal and spatial changes in NDVI and their responses to temperature and precipitation factors across spring, summer, autumn, and winter in Xinjiang from 2001 to 2020. We also quantified the relative effects of climate change and human activities on NDVI. The main conclusions are as follows:

  1. (1)

    In the past 20 years, the climate in Xinjiang has exhibited a trend of warming and wetting on an interannual scale. Seasonally, this translates to warmer and more humid conditions in spring and summer, colder and more humid conditions in autumn, and warmer and drier conditions in winter. Correspondingly, the NDVI changes of vegetation have varied across different periods but generally show a significant growth trend. On an annual scale, the growth trend is most pronounced in Region III, with a rate of 0.0015 per year. Seasonally, Region II experienced the highest growth trend in spring, at 0.0025 per year, while Regions III saw the largest growth trends in summer (0.0032 per year) and autumn (0.0014 per year). In winter, Region IV had the highest growth trend, at 0.0009 per year. Notably, vegetation degradation has been most concentrated in northern Xinjiang.

  2. (2)

    The effects of temperature and precipitation on NDVI exhibit significant spatiotemporal heterogeneity. Temperature is the primary climate factor influencing interannual, spring, and autumn NDVI changes, whereas precipitation is the main factor affecting NDVI changes in summer and winter. This indicates that climatic factors have varying impacts on vegetation growth across different periods.

  3. (3)

    Residual analysis revealed that some areas, particularly in northern Xinjiang, experienced vegetation degradation due to human activities. However, the overall trend in vegetation change was positive, indicating overall improvement. Contribution degree analysis showed that human activities accounted for over 60% of NDVI changes across different periods and regions. This substantial impact is largely attributed to various ecological policies implemented since 2001, which have significantly promoted vegetation restoration in the study area.