The differences in spatial-temporal distribution patterns and dominant driving factors of vegetation evolution among sub-regions during different historical periods were not clear. Therefore, this study introduced the gravity center model and geodetector to analyze the spatial-temporal change characteristics and dominant driving factors of normalized difference vegetation index (NDVI) in China during 1981–2019 from the perspective of geographic divisions. Results showed that (1) during 1981–2019, the average vegetation coverage in China showed an increasing trend and zones with obviously increasing vegetation NDVI were mainly distributed in the middle reaches of the Yellow River basin and the upper reaches of the Yangtze River. (2) During 1981–2019, the gravity center of national vegetation NDVI was mainly concentrated in Yan’an City and Tongchuan City, showing a southward migration trend as a whole, which indicates that the increment and growth rates of the southern part were greater than those of the northern part. (3) The growth of vegetation in southern China was primarily affected by the temperature, while that of northern China was largely influenced by precipitation. (4) During 1981–2019, the dominant interactive factors of vegetation change for all subregions changed greatly: natural factor (climate or terrain) ∩ other factors → landuse ∩ other factors.
As an important part of the terrestrial ecosystem, vegetation plays an important role in the material cycle, energy flow, and information transmission of the terrestrial ecosystem [1,2,3,4]. Vegetation is the natural link among soils, atmosphere, and water, which is most sensitive to climate change [5,6,7]. During the past decades, the spatial-temporal evolution patterns of vegetation have been comprehensively influenced by climate change, land use, soil and water conservation, and human activities [8,9,10,11]. Previous studies have shown that the climate change led to a significant increase in vegetation biomass in the northern hemisphere, and the same trend of vegetation change has also been detected in China [12,13,14]. Vegetation changes will trigger the fluctuation of the whole terrestrial ecosystem to a certain extent. Quantitative analysis of spatial-temporal change patterns of vegetation change and exploring its driving factors are of great significance for the improvement of the ecological environment.
Vegetation index derived from satellite remote sensing images has been adopted to analyze the hydrological response of vegetation at different scales, which has provided a convenient approach to reveal the characteristics of the regional terrestrial ecological environment. As an important indicator of the vegetation growth, the normalized difference vegetation index (NDVI) can accurately reflect the growth rate and the state of surface vegetation, which has been widely applied in the investigation of spatial-temporal changes in vegetation [15,16,17,18,19]. Zhong et al.  analyzed the spatial-temporal pattern of vegetation NDVI on the Qinghai-Tibet Plateau based on SPOT-VGT NDVI from 1998 to 2006 and found that the spatial distribution of vegetation NDVI was consistent with the climatic pattern of the Qinghai-Tibet Plateau. Wu et al.  investigated the changes in global vegetation coverage over the past 30 years and found that the annual variation of global average vegetation coverage fluctuated between 0.2 and 0.6, and most of the global continental vegetation showed an increasing trend (except for Oceania). Piao et al.  analyzed the changes in vegetation coverage in the cool temperate zone of Eurasia from 1982 to 2006 based on GIMMS NDVI data and found that the change of vegetation NDVI in Eurasia had obvious stages. Chen et al.  analyzed the influencing factors of vegetation in different climatic regions of the Yellow River Basin from 1982 to 2015, indicating that precipitation exerted the greatest impact on vegetation NDVI in semi-arid areas, whereas temperature had the greatest impact on vegetation NDVI in semi-humid areas. By exploring the response of global vegetation NDVI to extreme weather events, Liu et al.  found that vegetation in semi-humid and semi-arid areas, especially temperate grasslands and hardwood forests, were very sensitive to extreme precipitation events. Li et al.  utilized the datasets of GIMMS NDVI of the vegetation growing season from 1982 to 2012 to investigate the spatial-temporal response characteristics of NDVI, temperature, and precipitation of the eight typical vegetation coverage types in China. Zhang et al.  analyzed the characteristics of vegetation change in Inner Mongolia based on SPOT NDVI data from 2000 to 2015 and found that annual precipitation, soil type, vegetation type, and annual average temperature played a dominant role in the spatial-temporal change process of vegetation NDVI. The aforementioned previous studies showed that the average NDVI of the growing season was significantly correlated with the average temperature and precipitation with obviously great differences. In addition, human activities had played more important roles in the evolution process of surface vegetation rather than natural factors. However, the differences in dominant factors among subregions in China during different periods were urgent to clarify, which was critical for regional protection of the vegetation ecosystem [27,28]. Moreover, how to distinguish the effects of natural factors and human activities was still a difficult problem and a hot issue in the field of ecology [29,30,31].
Therefore, based on the datasets of GIMMS NDVI, MODIS NDVI, and meteorological data, this article introduced the gravity center model and geodetector to analyze the spatial-temporal evolution patterns of vegetation NDVI in China from 1981 to 2019 and then quantitatively determined the dominant factors of subregions in China during different periods.
2 Materials and methods
2.1 Study area and the geographical division
China is located in the eastern part of Asia and, along the west coast of the Pacific Ocean (Figure 1). China covers a vast territory, spanning nearly 50 degrees latitude from north to south and more than 60 degrees longitude from east to west, with an area of about 9.6 million km2 for China mainland. The eastern part of China has a monsoon climate, while the northwest region and Qinghai-Tibet plateau are dominated by temperate continental climate and alpine climate, respectively. The topography of China is complex and diverse, including plateaus (19%), mountains (26%), hills (10%), plains (12%), and basins (19%). The vegetation types in China are greatly affected by climatic factors, which can be categorized into eight types: cold temperate coniferous forest, temperate coniferous and deciduous broad-leaved mixed forest, warm temperate deciduous broad-leaved forest, subtropical evergreen broad-leaved forest, tropical monsoon rain forest, temperate steppe, temperate desert, and alpine vegetation on the Qinghai-Tibet Plateau. To explore and distinguish the differences in spatial and temporal evolution patterns of vegetation among different regions of China, the geographical division shown in Figure 1 has been conducted according to the similarity or consistency in climate, vegetation type, and topographic features.
2.2 Data sources and preprocessing
The meteorological data were obtained from the China Meteorological Data Network (http://data.cma.cn/), including daily precipitation (0.1 mm), daily average temperature (0.1°C), sunshine hour (0.1 h), and other factors such as wind speed (0.1 m/s) and surface temperature (0.1°C,0 cm) from 824 ground meteorological stations in China. Then, the grid datasets, including temperature, precipitation, sunshine, and accumulated temperature, with a spatial resolution of 8 km, were obtained by interpolation with the Kriging method of ArcGIS 10.2 and C#. The NDVI datasets including GIMMS NDVI (1981–2006a, at a 15d temporal resolution and 8 km spatial resolution) and MODIS NDVI data (2000–2019a, at a 16d temporal resolution and 1 km spatial resolution) were downloaded from the Resource and Environment Data Cloud Platform (https://www.resdc.cn/) and Geospatial Data Cloud (http://www.gscloud.cn/) [32,33]. To confirm the consistency of the spatial resolution, the MODIS NDVI datasets were resampled into images with 8 km, and the annual NDVI was obtained utilizing the maximum value calculation method. Considering the different sources of the two datasets, this study constructed the correction model by regression analysis based on the overlapping year’s data from 2000 to 2006: Y MODISNDVI = 1.0777X GIMMSNDVI + 0.0046 (R 2 = 0.925, P < 0.001) . Then, the aforementioned model was utilized to correct the MODIS NDVI from 2007 to 2019. The land use data for 1980 and 2020 with the spatial resolution of 1 km were obtained from the Resource and Environment Data Cloud Platform (https://www.resdc.cn/), which was used to analyze the influences of anthropogenic interference on vegetation. Digital elevation model (DEM) data derived from STRM3 were collected from Geospatial Data Cloud with a spatial resolution of 90 m. All these aforementioned datasets with different spatial resolutions were interpolated to grids with the spatial resolution of 8 km utilizing the tool of Resample of ArcGIS 10.2.
The workflow of this study is shown in Figure 2.
2.3.1 Linear trend and change rate
where slope refers to the gradient of the trend line, i represents the year i, n refers to the serial number of year (n = 39), and M NDVI,i denotes the maximum value of the year (i). If slope >0, it indicates that the vegetation condition has been improved; otherwise, it means that the vegetation condition has been degraded.
The change rate of vegetation is used to indicate the change degree of vegetation NDVI in study periods and is expressed as follows:
where NDVIcr refers to the change rate of NDVI during the study period, NDVImean refers to the average value of NDVI during the study period, and n refers to the length of the study period.
2.3.2 Correlation analysis
Correlation analysis mainly reflects the degree of correlation between vegetation NDVI and natural factors, which is measured by Pearson correlation coefficient pixel-by-pixel in this study. Then, partial correlation analysis is unitized to investigate the relationship between the two elements. The corresponding calculations formulas are as follows [38,39,40]:
where r x,y represents the correlation coefficients between the independent variable (NDVI) and the dependent variable (climate elements), refer to the average value of the two variables, and x k , y k denote the value of the year (k) of the two variables, refers to the partial correlation value between x and y 1 after fixed y 2, and represents the same meaning of . refers to the correlation coefficient between x and y 1 and , and represents the same meaning as . refers to the multiple correlation coefficient between x and y 1, y 2. Considering the confidence levels of P < 0.05 and P < 0.01, the significant relationships between NDVI, temperature, and precipitation have been discussed. T test (equation (5)) is used to analyze the significance of the partial correlation coefficient, and the F-test (formula (6)) is utilized to analyze the significance of the multiple correlation coefficient.
2.3.3 Gravity center model and standard deviation ellipse
Gravity center can better reflect the unbalance of spatial-temporal distribution of geographical elements, which has been widely utilized in economy, population studies, ecology, and other fields. In addition, the migration trajectory of the gravity center of vegetation NDVI can also directly indicate the imbalance of growth rate and increments in different parts. The gravity center of a region is defined as follows: if a region is composed of n subregions, where the coordinate of gravity center of the subregion i is , M i stands for the attribute value of the subregion, and then, the geographical coordinate of gravity center of the region is expressed as follows [41,42]:
The standard deviation ellipse is mainly composed of the rotation angle, the standard deviation along the principal axis (major axis), and the auxiliary axis (minor axis). The long axis represents the direction of the largest spatial distribution of gravity center, while the short axis denotes the direction of the least spatial distribution of gravity centers . The standard deviation ellipse can be applied to identify the change and migration trend of vegetation NDVI gravity centers in the region because it can more directly reflect the spatial evolution trend of vegetation coverage.
The geodetector model is utilized to detect the spatial differentiation of vegetation NDVI and then reveal the explanatory power of a factor to NDVI [44,45]. The correlation degree is measured by the q value that ranges from 0 to 1. The higher the value, the larger the explanatory power of the influencing factor (X) to vegetation NDVI (Y), and vice versa, which can be expressed as follows:
where h = 1, …., L represents the classifications of variable Y or X, N h denotes the number of units of class h, and N represents the number of units of the whole region. refers to the variance of Y value of class h and means the variance of Y value of the whole region. SSW refers to the sum of the variance in layer, and SST represents the total variance of the whole region.
3.1 Change rate of vegetation NDVI in China during 1981–2019
Based on the change rate of vegetation NDVI in China during 1981–2019 (Figure 3), it could be found that in the past 40 years, zones of change rate between −10 and 10% that belonged to stable areas had the largest area that accounted for 35.57%. Zones of the change rate between 10 and 30% that belonged to slightly increased areas accounted for 23.42%. Zones of the change rate of less than −30% that belonged to significantly reduced areas had the smallest area, accounting for 6.93%. The aforementioned analysis results showed that the average vegetation coverage in China has shown an overall increasing trend in the past 40 years with the stable areas (−10 to 10%) and increased areas (>10%) amounting to 78.22% of the total study area.
As shown in Figure 4, zones of the change rate between −10 and 10% were the most widely distributed, mainly concentrated in southern Xinjiang, western Inner Mongolia, and the northern Qinghai-Tibet Plateau. Zones of the NDVI change rate >30% were mainly distributed in the middle and upper reaches of the Yellow River, the middle and upper reaches of the Yangtze River, and the Liaohe River basin. The slightly reduction zones of NDVI (−30% < change rate < −10%) were mainly distributed in northern China, including northern Heilongjiang and northeastern Inner Mongolia.
3.2 Inter-annual variation of vegetation coverage in China
From the inter-annual variations of vegetation NDVI (Figure 5), the overall fluctuation range of vegetation NDVI in China was relatively larger in the past 40 years. Among them, the annual average vegetation NDVI in the northeast China was the largest, which amounted to 0.830, followed by that of the central-south part (0.752). The northwest part dominated by desert and grassland was restricted by precipitation, which had the minimum annual NDV of 0.294. During the study period, the minimum value of NDVI in most subregions appeared around 2000, which was a consequence of meteorological disasters and human activities. Since 2000, the projects of “returning farmland to forest and grassland” and afforestation have been implemented in various regions of China with remarkable results, and the average NDVI of the national scale and different subregions showed a fluctuating increasing trend. The results were consistent with the results of Zhao et al.  and Jin et al. . Before 2000, the NDVIs in most areas of China decreased in varying degrees, among which the northeast region was the most serious, and its NDVI value decreased from 0.834 in 1981 to 0.757 in 2000, with a decrease rate of 0.077/20a. After 2000, the NDVI in China has increased in varying degrees, basically consistent with the previous research results .
3.3 Migration trajectory of gravity centers
3.3.1 Distributions of gravity center of vegetation coverage at national and regional scale
To explore the imbalance and bias of the spatial distribution of vegetation NDVI in the study area, the gravity center model was utilized to analyze the dynamic changes of national and regional NDVI in the recent 40 years. As shown in Figure 6, the standard deviation ellipse was adopted to identify the change and migration trend of gravity centers. Utilizing a polar coordinate system and taking the average gravity center of the NDVI during 1981–2019 as the origin, the distance (polar radius) and offset angle (polar angle) to the origin were calculated to directly reflect the migration direction and the distance of the gravity center.
The gravity centers of vegetation NDVI in northeast China were mainly concentrated in Harbin, Heilongjiang Province. The area of the standard deviation ellipse was 92.62 km2 (Table 1), which was the smallest among the six subregions, indicating that the distribution of NDVI gravity centers was relatively concentrated. In the polar coordinate system, the proportion of the gravity centers in the north half (northeast quadrant and northwest quadrant) was 84.62%, whereas the proportion of the gravity centers in the south half (southwest quadrant and southeast quadrant) was merely 15.38%, which was much smaller than that in the north half, demonstrating that the grown rate and increment of the vegetation NDVI in the south half was smaller than that in the north half. The reason was that lots of biological restoration projects had been applied in the northern part to enhance the vegetation restoration combined with the global warming. The gravity centers of NDVI in north China were mainly distributed in Xilingol League and Chifeng City, Inner Mongolia. The area of standard deviation ellipse in north China was significantly larger than that of northeast China, which showed that the change of vegetation NDVI in north China was more salient than that in northeast China. The gravity centers of vegetation NDVI in northwest China were mainly concentrated in the vicinity of the polar angle of the line from 150° to 330°, which showed a northwest-southeast distribution pattern as a whole. The shape of the standard deviation ellipse was flatter, which had the largest area of 1549.87 km2 among the six subregions, showing that the spatial distribution of the gravity centers was obviously uneven. The reason was that the vegetation was mainly distributed in the Ili river valley affected by ice melt water and southeastern parts affected by southeast warm and humid monsoon.
|Standard deviation ellipse parameters||Northeast||North||Northwest||East||Central South||Southwest||National|
|Rotation angle (°)||104.86||74.80||151.03||90.75||71.50||157.62||92.61|
|Major axis (km)||7.30||23.89||54.79||10.91||10.36||20.21||28.93|
|Minor axis (km)||4.04||10.94||9.01||3.55||4.53||5.35||15.96|
The distribution of gravity centers of vegetation NDVI in east China was concentrated mainly in Huangshan City, Anhui Province. The rotation angle of the standard deviation ellipse was 90.75°, and the gravity centers were distributed in the north-south direction. During 1981–2019, the gravity centers of vegetation NDVI were distributed in the northern half, accounting for 64.10%, while the gravity centers of vegetation NDVI in the southern half accounted for 35.90%, indicating that the increment and the growth rate of vegetation NDVI in the northern parts was higher than that in the southern parts. The reason was that the increasing precipitation and temperature were both conductive to the vegetation grown and restoration combined with the implementation of Three North Shelterbelt ecological projects. The gravity centers of vegetation NDVI in the central south were mainly distributed in Zhaoyang City, Hunan Province, and the angle of the standard deviation ellipse was 71.50°. The gravity centers were distributed from northeast to southwest as a whole. The gravity centers of NDVI in southwest China were distributed in the northwest-southeast direction, and the standard deviation ellipse major axis was in the northwest-southeast direction with a length of 20.21 km. The minor axis was in the northeast-southwest direction with a length of 5.35 km, demonstrating that the discreteness of the gravity centers of NDVI in the northwest-southeast direction was greater than that in the northeast-southwest direction. As shown in Figure 6, the deviation distances of the gravity centers of NDVI in 1981 and 1988 were larger, indicating that the increment of NDVI in the northwest was more significant in these two years. The reason was that during the past decades, global warming greatly promoted the growth and restoration of vegetation in alpine areas, especially in Qinghai-Tibet Plateau.
The national gravity centers of NDVI were mainly distributed in Yan’an City and Tongchuan City, Shaanxi Province. The standard deviation ellipse area was 1450.26 km2, and its spatial distribution of the gravity centers was relatively discrete. The distribution of the gravity centers in the four quadrants under the polar coordinate system was relatively uniform, in which the gravity centers in the northeast quadrant were the largest with 28.20%, followed by that of the northwest quadrant (25.64%). The southwest quadrant and the southeast quadrant had the same proportion of 23.08%. The proportion of the gravity centers in the north half (53.84%) was greater than that in the south half (46.16%), indicating that the increment of the NDVI gravity center in the northern half was higher than that in the southern half. Before 2000, the gravity center of NDVI was mostly concentrated in the northern part, and after 2000, the gravity center was mostly concentrated in the southern part, indicating that the gravity centers showed a trend of southward movement.
3.3.2 Migration of gravity centers of vegetation NDVI at national and regional scales
To further analyze the spatial variation of vegetation NDVI in the study area, the gravity center migration of vegetation NDVI at 5-year scale during 1981–2019 was calculated. As shown in Figure 7, during 1981–2000, the gravity centers of vegetation NDVI in northeast China showed a migration trend of southwest-northeast-southwest. The gravity centers of 2006–2010 moved to the southwest compared with that of 2001–2005, indicating that the growth rate and increments of vegetation NDVI in the southwest were larger than that in the northeast during this period. However, the gravity centers of 2006–2010 showed a northward swinging trend compared with those of 2011–2015 and 2016–2019. On the whole, in the past 40 years, the gravity center of vegetation NDVI in northeast China showed a moving trend of south-north. The gravity centers of vegetation NDVI in north China exhibited obvious regularity. The gravity centers of 2006–2015 shifted northward, and in other periods, the gravity centers gradually shifted to the south, suggesting that the increment of the gravity center in the south was larger than that in the north. Among them, the migration distance of the gravity center of 1996–2000 → 2001–2005 was the largest, which amounted to 15.07 km, while that of 1991–1995 → 1996–2000 was the smallest with 2.04 km. The gravity centers of NDVI in northwest China migrated northwestward in three periods, namely, 1981–1985 → 1986–1990, 1986–1990 → 1991–1995, and 1991–1995 → 1996–2000, respectively. The gravity centers all moved to the southeast in the last four periods, namely, 1996–2000 → 2001–2005, 2001–2005 → 2006–2010, 2006–2010 → 2011–2015, and 2011–2015 → 2016–2019. Among them, the migration distance from 1996–2000 to 2001–2005 was the maximum with 50.53 km. From 2001–2005 to 2006–2010, the migration distance was the minimum with 4.73 km.
The gravity centers of vegetation NDVI of 1986–1990 in east China shifted to the northwest compared with that of 1981–1985. During the five periods from 1986 to 2015, the migration trajectory of the gravity centers of NDVI exhibited a circular trend of “southwest to northeast to northeast to southwest to south,” whereas the gravity center of 2016–2019 continued to migrate southward compared with that of 2011–2015. The migration trajectory of the gravity center of NDVI in central south China showed a “Z” shape. The gravity center of 1981–1985 → 1986–1990 migrated northeastward, while that of 1986–1990 → 1991–1995, 2006–2010 → 2011–2015, and 2011–2015 → 2016–2019 moved to the southwest. Meanwhile, the gravity centers migrated to the northwest in the three periods of 1991–1995 → 1996–2000, 1996–2000 → 2001–2005, and 2001–2005 → 2006–2010. The gravity center of vegetation NDVI in the central south China showed a moving trend of the southwest as a whole. The gravity centers of NDVI in southwest China in 1986–1990 and 1991–1995 shifted to the northwest compared with that in 1981–1985 and the migration distance of 1981–1985 → 1986–1990 was the largest. The gravity centers of 1991–1995, 1996–2000, 2001–2005, and 2006–2010 shifted to the southeast, while the gravity center of 1996–2000 migrated to the southwest compared to that of 2001–2005. Gravity centers of 2011–2015 and 2016–2019 moved northwest compared with that of 2006–2010.
The gravity centers of national vegetation NDVI were mainly concentrated in Yan’an City, Shaanxi Province in the first four periods (1981–2000), and Tongchuan City in the last four periods (2001–2019). The gravity center of vegetation NDVI shifted to the northwest during 1986–1990 compared with that of 1981–1985, indicating that the increment and the growth rate of vegetation NDVI in the northwest of the study area was larger than that in the southeast during this period. The gravity centers of vegetation NDVI in the two periods of 1991–1995 and 1996–2000 moved to the southwest as a whole compared with that of 1986–1990, indicating that the increment of vegetation NDVI in southwest China was larger than that in northeast China in the recent decade. The gravity center of 1996–2000 migrated to the southeast compared to that of 2001–2005, and its migration distance was the maximum. The gravity center of 2006–2010 migrated westward compared with 2001–2005. From 2006 to 2019, the gravity center first moved to the northwest and then to the southwest. Base on the analysis, in the past 40 years, the gravity center of vegetation NDVI shifted to the south as a whole, and the migration distance amounted to 36.25 km.
3.4 Influence of single climate factor on vegetation NDVI
The vegetation grown was significantly influenced by climate factors. In this study, we analyzed the correlation between vegetation NDVI and single climate factor, and the correlation coefficients between NDVI and temperature, precipitation, sunshine, and accumulated temperature were calculated at the pixel scale.
As shown in Figure 8(a), zones with a significantly positive correlation (P < 0.01) between NDVI and temperature accounted for 53% of the total area, which were mainly distributed in the south parts around latitude 35°N. The increasing temperature was conductive to prolong the phenological period of vegetation and contribute to the photosynthesis of vegetation. Zones with a negative correlation (P < 0.05) between NDVI and temperature were mainly concentrated in the northern part of China (near latitude 40°N), including central Xinjiang, the Inner Mongolia-Gansu border zone, northwest of Gansu and Qinghai, the Inner Mongolia Plateau, and the central Northeast Plain. High temperature could increase the evapotranspiration of surface water, which led to different degrees of drought. As shown in Figure 8(b), zones with a positive correlation (P < 0.05) between NDVI and precipitation accounted for 60% of the total area, which were mainly distributed in the inner Mongolia Plateau and the middle and upper reaches of the Yellow River basin. In arid and semi-arid regions, the increasing precipitation was conducive to the growth and restoration of vegetation. Zones with a negative correlation (P < 0.05) between NDVI and precipitation were mainly concentrated in the Tarim Basin, the Qaidam Basin in Xinjiang, and the western part of Inner Mongolia Plateau. In southern China, in areas such as Yunnan, Hubei, Anhui, and Jiangsu, the correlation between NDVI and precipitation was also negative. It was because the intensive precipitation has led to severe soil erosion and geological disasters, which would destroy the regional vegetation ecosystem. Figure 8(c) showed that 42% of the total area exhibited a positive correlation (P < 0.05) between NDVI and sunshine, and zones with obvious positive correlation mainly concentrated in central Xinjiang, Gansu and northwest Qinghai, and the western Yunnan-Guizhou Plateau. Zones with a negative correlation (P < 0.05) between NDVI and sunshine were mainly concentrated in central China (between 105°E and 115°E), such as Guizhou, Shanxi, and Henan. The increasing sunshine could contribute to the photosynthesis of vegetation in the alpine region, but it would also cause drought in arid zones. Figure 8(d) shows that zones with a negative correlation (P < 0.05) between NDVI and accumulated temperature accounted for 50% of the total area, which were mainly concentrated in northern China, including the three provinces in northeast China, Inner Mongolia, and Xinjiang. The increasing accumulated temperature was conducted to the vegetation growth in north cold areas.
3.5 Influence of double climate factors on vegetation NDVI
The vegetation NDVI was affected by a variety of climatic factors, and the climatic factors were often interactive. Considering that temperature and precipitation were the main factors affecting the change of vegetation NDVI, this study calculated the partial correlation coefficients and multiple correlation coefficients between vegetation and temperature, precipitation in China during 1981–2019 and tested its level of significance.
As shown in Figure 9(a), the partial correlation coefficients between NDVI and temperature ranged from −0.82 to 0.87. Zones with positive correlation were mainly concentrated in east China, south-central, and southwest China, accounting for 53.65% of the total area, while zones with negative correlation were mainly distributed in the central area of Xinjiang, Inner Mongolia, Qinghai and Gansu, and the eastern area of the Northeast Plain, amounting to 46.35% of the total area. As shown in Figure 8(c), zones with a positive correlation between NDVI and precipitation were mainly distributed in north China and the border area of Shaanxi-Ningxia-Gansu, accounting for 60.34% of the total area. Meanwhile, the areas with a negative correlation were mainly distributed in central Xinjiang, western Inner Mongolia, and most areas in the south, amounting to 39.66% of the total area. As shown in Figure 9(b) and (d), the zones with the extremely significant positive correlation between NDVI and temperature accounted for 18.32% (P < 0.01, R > 0), which were mainly concentrated in the Qinghai-Tibet Plateau, Loess Plateau, and southeastern coastal areas. Zones with significant positive correlation (P < 0.05, R > 0) accounted for 8.43%, while the area proportion of extremely significant negative correlation (P < 0.01, R < 0) was 15.33%, mainly concentrated in northern arid and semi-arid areas, such as central Xinjiang and western Inner Mongolia. Zones with significant negative correlation (P < 0.05, R < 0) accounted for only 5.92%. Zones with extremely significant positive correlation accounted for 8.19%, which were mainly distributed in the Inner Mongolia Plateau and the border area of Shaanxi-Ningxia-Gansu. Zones with significant positive correlation (P < 0.05) accounted for only 6.25%, while zones with extremely significant negative correlation (P < 0.01) covered the smallest area, accounted for only 0.81%. Meanwhile, zones with significant negative correlation area accounted for 2.48%.
As shown in Figure 9(e), the multiple correlation coefficients between NDVI temperature and precipitation ranged from 0 to 0.88. Zones with larger multiple correlation coefficients were mainly concentrated in central and southern Xinjiang, northern Inner Mongolia, Gansu-Ningxia-Shaanxi border zone, southern Tibet, northern Qinghai, and the middle part of the middle and lower reaches of the Yangtze River, while zones with smaller multiple correlation coefficients were mainly distributed in the Northeast Plain and southeast coastal areas.
3.6 Interactive influence of multiple factors on vegetation NDVI
Interactive detection was utilized to evaluate the impacts of the interaction of different driving factors on vegetation NDVI, which was conductive to investigate the driving mechanism of vegetation coverage change in various subregions of China. The results showed that the q values of the interaction among factors in most areas were larger than that of the single factor, indicating that the influences of driving factors on vegetation NDVI were not independent, but a process of interaction and mutual enhancement. In this article, the vegetation NDVI of 1981 (average status of 1981, 1982, and 1983) and 2019 (average status of 2017, 2018, and 2019) were selected to explore the driving mechanisms of vegetation NDVI.
As shown in Figure 10(a1), in northeast China in 1981, the interactive factors with higher explanation power on vegetation NDVI change were sequenced as follows: land use ∩ sunshine (0.657) > elevation ∩ sunshine (0.585) > soil type ∩ sunshine (0.568) > sunshine ∩ temperature (0.566) > aspect ∩ sunshine (0.528) > accumulated temperature ∩ sunshine (0.525), indicating that sunshine ∩ other factors acted as dominant factors. As shown in Figure 10(a2), in 2019, the interactive factors with high explanation power were sequenced as follows: land use ∩ aspect (0.473) > land use ∩ temperature (0.446) > land use ∩ precipitation (0.356) > soil type ∩ land use (0.347) > soil type ∩ aspect (0.312) > land use ∩ sunshine (0.305). The explanatory power of the interaction between land use and other factors was dominant, while the interaction between climatic factors was relatively weak, suggesting that the climate was not the main factor affecting the change of vegetation in this period. As shown in Figure 10(b1), the maximum explanation power of interactive factors in north China in 1981 was sunshine ∩ temperature (0.752), while that of the aspect ∩ slope (0.111) was minimum. In addition, it could be found that the explanation power of the interaction between soil type and other factors was larger in this period. As shown in Figure 10(b2), the land use ∩ precipitation had the largest explanation power of 0.770 in 2019, whereas the aspect ∩ slope had the smallest explanation power of 0.149.The interaction of precipitation, land use, soil type ∩ other factors had greater explanatory power in 2019. As shown in Figure 10(c1), the soil type ∩ precipitation (0.669) and land use ∩ precipitation (0.642) in northwest China in 1981 exhibited a larger explanation power for vegetation NDVI change,. During this period, the explanatory power of the interaction between precipitation and other factors was the largest. In 2019, the explanation powers of land use ∩ precipitation (0.699) and land use ∩ sunshine (0.681) were larger, while that of the aspect ∩ slope was the smallest.
As shown in Figure 10(d1) and (d2), the aspect ∩ precipitation (0.374) had the largest explanation power in east China in 1981, while that of accumulated temperature ∩ temperature was the smallest with 0.074. The explanatory power of interactions between climate factors in this period was relatively smaller. In 2019, land use ∩ aspect (0.412) had the largest explanation power, while that of accumulated temperature ∩ temperature (0.073) was the smallest. It could be seen from the figure that the explanatory power between land use ∩ other factors was the largest. As shown in Figure 10(e1), the interactive factors with larger explanation power for vegetation NDVI change in central south China in 1981 were ranked as follows: soil type ∩ aspect (0.303) > aspect ∩ precipitation (0.276) > soil type ∩ sunshine (0.268) > accumulated temperature ∩ sunshine (0.255) > land use ∩ precipitation (0.251) > sunshine ∩ precipitation (0.250). The explanatory power of interaction among aspect, soil type, and precipitation was larger. As shown in Figure 10(e2), the interactive factors with larger explanation power in 2019 were ranked as follows: land use ∩ aspect (0.476) > land use ∩ precipitation (0.442) > soil type ∩ land use (0.407) > land use ∩ elevation (0.3920) > land use ∩ accumulated temperature (0.3919) > land use ∩ temperature (0.391). In this period, the explanatory power of the interaction between land use and other factors acted as dominant factors, and the explanatory power between land use and other factors was greater than that in the previous period. The interaction among climatic factors was relatively weak, revealing that climate was not the main factor influencing the change of vegetation NDVI in this period. As shown in Figure 10(f1), the elevation ∩ sunshine (0.810) had the largest explanation power in southwest China in 1981, while that of aspect ∩ slope was the smallest with 0.087. As shown in Figure 10(f2), the elevation ∩ sunshine had the largest explanation power of 0.807 in 2019, while the aspect ∩ slope had the smallest value of 0.084. For the northeast parts of China in 1981, the interactions between sunshine and other factors played important roles in the change process of vegetation NDVI, while land use ∩ other factors became the dominant factors in 2019. In the northwest parts, the dominant interactive factors shifted from precipitation ∩ other factors to land use ∩ other factors.
4.1 Comparisons of spatial-temporal change patterns of vegetation NDVI in different subregions
During 1981–2019, there was an increasing trend of vegetation coverage in China, especially in the middle reaches of the Yellow River basin, the middle and upper reaches of the Yangtze River basin, and the Liaohe River basin, which was consistent with the studies of the surrounding countries [49,50]. This was due to the fact that during the past decades, national policies of “returning farmland to forest and grassland,” “Yangtze River Ecological Protection and restoration project,” “Water and soil loss control project in the Yellow River Basin,” and “Three North Shelterbelt Ecological Projects” had been implemented and remarkable ecological improvements had been achieved [51,52]. In addition, the vigorous implementation of the afforestation project in the Liaohe River basin had greatly improved the regional vegetation condition. During the past 40 years, the change trend of vegetation NDVI in southern Xinjiang, and western Inner Mongolia was not obvious, which was consistent with the studies of Sun et al.  and Fan and Liu . The reason was that the desert and gobi were widely distributed in these above regions with scarce precipitation, which was not conducive to the growth and restoration of vegetation [55,56]. In 2000, the NDVI showed a minimum value in all subregions. The reason was that human activities such as urbanization, steep slope reclamation, and overgrazing had greatly damaged the regional vegetation ecosystem. During 1999–2000, there occurred different degrees of drought in most subregions of China, which exacerbated the degradation of vegetation. During 1981–2000, the vegetation NDVI showed a slightly decreasing trend in the subregions of east, central, and southwest. In these aforementioned zones, the rapid urbanization process, population growth rate, and economic development had led to the intensification of the contradiction between man and land . Although certain protection measures had been taken, the speed of social and economic development far exceeded the pace of ecological and environmental protection, so the change in vegetation cover was not obvious . After 2000, the vegetation NDVI in most subregions showed an increasing trend, which was consistent with the previous studies . During 1981–2019, the gravity center of vegetation NDVI moved toward the south, which indicated that the grown rate and increment in the south parts were larger than that of the north parts. Although the vegetation NDVI in most regions showed an increasing trend due to the global warming (prolonged vegetation growth period, accelerated decomposition of soil organic matter, and the release of nutrient elements) and human activity (improving the agricultural management and implementing vegetation construction project), the rapidly increasing temperature in North China also led to varying degrees of drought, especially in the northwest parts with scare precipitation and water resource [60,61]. Therefore, the grown increment of the vegetation NDVI was smaller than that of the south parts. For the subregions, the gravity center of the vegetation NDVI in northeast and north moved toward the south during the past 40 years because of the implementations of “Three North Shelterbelt ecological projects” and the rising temperature that were all both conductive to the grown and restoration of vegetation . The gravity center of the northwest showed a southward trend because the climate in this area had been becoming warm and moist, especially in the northeast edge of Qinghai Tibet Plateau . In addition, the implementation of “ecological environment protection project of Three-river Source Region” had also contributed to the restoration of vegetation in the southeast parts . The gravity center of vegetation NDVI in east and central south both moved toward the south because of the implementation of “Returning farmland to forest and grassland” and other ecological protection projects . The gravity center of vegetation NDVI showed a southwestward migration trend, and it was because under the stress of global warming, the increasing temperature and increased precipitation had greatly contributed to the vegetation grown as well as the implementation of “Returning grazing to grassland” [16,30].
4.2 Dominant factors of different subregions in historical periods
Zones with a positive relation between temperature and vegetation NDVI were mainly distributed in the southwest, east, and central parts of China, and the main reason was that the increasing temperature could prolong the growth period of vegetation and was conductive to the photosynthesis of vegetation . In addition, the higher temperature would accelerate the decomposition of soil organic matter and the release of nutrient elements, which improved the growth environment of vegetation . The temperature had negative relations with vegetation NDVI in the northwest and north parts, which was due to the fact that the increasing temperature would lead to severe moisture evaporation that resulted in various degrees of drought . Zones with positive relation between precipitation and vegetation NDVI were mainly located in the Inner Mongolia Plateau and the middle and upper reaches of the Yellow River basin. The reason was that the grasslands were widely distributed in these areas, which was more sensitive to the change in soil moisture . Zones with the obvious positive correlation between sunshine and vegetation NDVI were mainly concentrated in the junction zones of Xinjiang-Gansu-Inner Mongolia and Yunnan. The reason was that sufficient light was conductive to the photosynthesis of the plant, especially in the alpine region with lower temperature . Zones with negative correlation with sunshine were mainly located in the central parts of China, and the reason was that the prolonged sunshine hours would increase the surface heat, which intensified the surface evapotranspiration that increased the drought risk . Zones with a positive correlation of accumulated temperature were mainly located in the southern, southwestern (Qianghai-Tibet plateau), and eastern China. In these regions, the increasing accumulated temperature was conductive to prolong the grown period of vegetation .
In the northeast parts of China in 1981, the dominant interactive factors were sunshine ∩ others, while in 2019, the land use ∩ other factors had played more an important role in the evolution process of the vegetation system. This was because that the sufficient sunshine could improve the grown environment and promote the plant photosynthesis . With the implements of “Afforestation project” and urbanization, the land use ∩ other factors gradually became the dominant interactive factors . In the north part of 1981, the sunshine ∩ temperature has the largest explanatory power, while the land use ∩ precipitation was the dominant interactive factor. This was due to the fact that both sunshine and temperature were conductive to plant photosynthesis, especially in zones with low temperature . In addition, the grassland was more sensitive to the change in temperature. However, under the stress of global warming and urbanization, the precipitation and human activity gradually became the dominant factors in the evolution process of the vegetation system . In the northwest parts, the precipitation ∩ other factors had the largest explanatory power in 1981, while the land use ∩ other factors played a more important role in the change of vegetation NDVI. This was because the water resources were one of the limiting factors for the vegetation growth in arid and semi-arid areas, which were mainly derived from precipitation . With the development of regional economic and social development, human activities, such as afforestation, returning grazing to grass, and urbanization, greatly weakened the influences of natural factors on the vegetation system . In the south part, the dominant interactive factor was the soil type ∩ aspect in 1981, while the explanatory powers of land use ∩ other factors were larger in 2019. This was due to the fact that the sufficient water and heat sources were not the restrictive factors for vegetation grown, so that the soil type and terrain had played a more important role in the evolution process of the vegetation system. In addition, the intensity of human interference severer during the past 40 year, which led to drastic changes in land use . In the southwest part, the elevation and sunshine had the largest explanatory power in both 1981 and 2019. This was because the eco-environment of these regions was characterized by high elevation and large topographic relief, which greatly influenced the spatial and temporal changes of vegetation NDVI .
Based on the datasets of GIMMS NDVI and MODIS NDVI, the gravity center model and geodetector had been introduced to quantitatively analyze and discussed the spatial-temporal evolution pattern of vegetation NDVI in different subregions from the perspective of geographical division and furthermore determined the dominant driving factors of vegetation NDVI evolution in different historical periods of each subregion. The main conclusions were as follows:
The spatial distribution of vegetation NDVI in China showed a decreasing trend from the southeast to the northwest as a whole. Before 2000, the average vegetation NDVI in most subregions of China showed a decreasing trend, while there was an obvious increasing trend in all areas after 2000.
During1981–2019, the gravity center of national vegetation NDVI was mainly concentrated in Yan’an and Tongchuan in Shaanxi Province, showing a southward migration trend as a whole, which indicated that the increment and the growth rate of the vegetation NDVI in the southern part was greater than that of the northern part.
There were significant differences in the correlation between vegetation NDVI and climatic factors in different subregions. The growth of vegetation in southern China was mainly affected by temperature, whereas that in the north was mainly affected by precipitation.
During 1981–2019, the dominant interactive factors of vegetation change for all subregions changed greatly: natural factor (climate or terrain) ∩ other factors → land use ∩ other factors.
Funding information: This work was supported by the technology project of State Grid Corporation of China headquarters (grant no.5500-202140127A) and the Open Fund of Key Laboratory of Urban Land Resources Monitoring and Simulation, Ministry of Natural Resources (grant no. KF-2020-05-001).
Author contributions: ShiJun Wang, Ning Wang: conceptualization, methodology, software, data curation, writing – original draft preparation; Chang Ping, Jing Wen, Ke Zhang: investigation, revision; Kun Yuan, Jun Yang: supervision, writing – reviewing and editing, and revision.
Conflict of interest: All authors declare that they have no known conflict of interest or personal relationships that could have appeared to influence the work reported in this article.
Data availability statement: The data that support the findings of this study are available from the corresponding author, Wang N, upon reasonable request.
 Guo B, Wen Y. An optimal monitoring model of desertification in Naiman Banner based on feature space utilizing Landsat 8 OLI Image. J Eng. 2020;8:4761–8.10.1109/ACCESS.2019.2962909Search in Google Scholar
 Gomes E, Inacio M, Bogdzevič K, Kalinauskas M, Karnauskaitė D, Pereira P. Future land-use changes and its impacts on terrestrial ecosystem services: a review. Sci Total Environ. 2021;781:146716.10.1016/j.scitotenv.2021.146716Search in Google Scholar PubMed
 Wu HW, Guo B, Fan JF, Yang F, Han BM, Wei CX, et al. A novel remote sensing ecological vulnerability index on large scale: a case study of the China-Pakistan Economic Corridor region. Ecol Indic. 2021;2021:107955.10.1016/j.ecolind.2021.107955Search in Google Scholar
 Hu CJ, Guo L. Advances in the research of ecological effects of vegetation restoration. Ecol Environ Sci. 2012;21(9):1640–6.Search in Google Scholar
 Guo B, Zang WQ, Luo W. Spatial-temporal shifts of ecological vulnerability of Karst Mountain ecosystem-impacts of global change and anthropogenic interference. Sci Total Environ. 2020;74:140256.10.1016/j.scitotenv.2020.140256Search in Google Scholar PubMed
 Lu QQ, Jiang T, Liu DL, Liu ZY. The response characteristics of NDVI with different vegetation cover types to temperature and precipitation in China. Ecol Environ Sci. 2020;29(1):23–34.Search in Google Scholar
 Li XB, Shi PJ. Sensitivity analysis of variation in NDVI, temperature and precipitation in typical vegetation types across China. Acta Phytoecol Sin. 2000;03:379–82.Search in Google Scholar
 Wu ZL, Jia WX, Zhao Z, Zhang YS, Liu YR, Chen JH. Spatial-temporal variations of vegetation and its correlation with climatic factors in Qilian Mountains from 2000 to 2012. Arid Land Geogr. 2015;38(6):1241–52.Search in Google Scholar
 Shi DD, Yang T, Hu JM. Spatial-temporal variation of NDVI-based vegetation during the growing-season and its relation with climatic factors in the Yellow River Source Region. Mt Res Dev. 2018;36(2):184–93.Search in Google Scholar
 Guo B, Zang WQ, Luo W, Wen Y, Yang F, Han BM, et al. Detection model of soil salinization information in the Yellow River Delta based on feature space models with typical surface parameters derived from Landsat8 OLI image. Geomat Nat Haz Risk. 2020;11(1):288–300.10.1080/19475705.2020.1721573Search in Google Scholar
 Fang JY, Piao SL, He JS. Vegetation activities in China have been increasing in the past 20 years. Sci China Ser C. 2003;6:554–65.Search in Google Scholar
 Liu XF, Zhu XF, Pan YZ, Li YZ, Zhao AZ. Spatiotemporal changes in vegetation coverage in China during 1982-2012. Acta Ecol Sin. 2015;5(16):5331–42.Search in Google Scholar
 Guo B, Zang WQ, Yang F, Han BM, Chen ST, Liu Y, et al. Spatial and temporal change patterns of net primary productivity and its response to climate change in the Qinghai–Tibet Plateau of China from 2000 to 2015. J Arid Land. 2020;12(1):1–17.10.1007/s40333-019-0070-1Search in Google Scholar
 Liu LL, Liu LY, Hu Y. Remote sensing monitoring and analysis of the start time of Eurasian vegetation growing season from 1982 to 2006. Prog Phys Geog. 2012;31(11):1433–42.Search in Google Scholar
 Jiang HL, Xu X, Guan MX, Wang LF, Huang YM, Jiang Y. Determining the contributions of climate change and human activities to vegetation dynamics in agro-pastural transitional zone of northern China from 2000 to 2015. Sci Total Environ. 2020;718:134871.10.1016/j.scitotenv.2019.134871Search in Google Scholar PubMed
 Zheng J, Feng WL, Niu XJ, He BW. Vegetation change and its correlation with meteorological factors in Sichuan Province. Bull Soil Water Conserv. 2016;36(2):99–104.Search in Google Scholar
 Jin K. The Spatial-temporal change of vegetation cover in China and its relationship with climate and human activities. Xi’an: Northwest A & F University; 2017.Search in Google Scholar
 Guo B, Zang WQ, Yang X, Huang XZ, Zhang R, Wu HW, et al. Improved evaluation method of the soil wind erosion intensity based on the cloud–AHP model under the stress of global climate change. Sci Total Environ. 2020;746:141271.10.1016/j.scitotenv.2020.141271Search in Google Scholar PubMed
 Zhong L, Ma YM, Salama MS, Su ZB. Assessment of vegetation dynamics and their response to variations in precipitation and temperature in the Tibetan Plateau. Clim Change. 2010;103(3–4):519–35.10.1007/s10584-009-9787-8Search in Google Scholar
 Wu D, Wu H, Zhao X, Zhou T, Tang BJ, Zhao WQ, et al. Evaluation of spatiotemporal variations of global fractional vegetation cover based on GIMMS NDVI data from 1982 to 2011. Remote Sens. 2014;6(5):4217–39.10.3390/rs6054217Search in Google Scholar
 Piao SL, Wang XH, Ciais P. Changes in satellite derived vegetation growth trend in temperate and boreal Eurasia from 1982 to 2006. GCB Bioenergy. 2011;17(10):3228–39.10.1111/j.1365-2486.2011.02419.xSearch in Google Scholar
 Chen C, Wang YM, Li YY, Zhou S. Vegetation changes and influencing factors in different climatic regions of the Yellow River Basin from 1982 to 2015. J Yangtze River Sci Res Inst. 2021;39(2):56–62.Search in Google Scholar
 Liu G, Liu H, Yin Y. Global patterns of NDVI-indicated vegetation extremes and their sensitivity to climate extremes. Environ Res Lett. 2013;8(2):25009.10.1088/1748-9326/8/2/025009Search in Google Scholar
 Li SJ, Wang JM, Zhang M, Tang Q. Characterizing and attributing the vegetation coverage changes in North Shanxi coal base of China from 1987 to 2020. Resour Policy. 2021;74:102331.10.1016/j.resourpol.2021.102331Search in Google Scholar
 Zhang SY, Nie Y, Zhang HY, Li YL, Han YD, Liu XH, et al. Spatiotemporal variation of vegetation NDVI and its driving forces in Inner Mongolia based on geodetector. Acta Agrest Sin. 2020;28(5):1460–72.Search in Google Scholar
 Liu Y, Yan L, Li SC, Safa M. Spatial and temporal patterns of global NDVI trends: correlations with climate and human factors. Remote Sens. 2015;7(10):13233–50.10.3390/rs71013233Search in Google Scholar
 Quamar MDF. Holocene vegetation and climate change from central India: an updated and a detailed pollen-based review. Holocene Clim Change Environ. 2022;9:29–36.10.1016/B978-0-323-90085-0.00013-9Search in Google Scholar
 Gao JB, Jiao KW, Wu SH. Investigating the spatially heterogeneous relationships between climate factors and NDVI in China during 1982 to 2013. Sci Press. 2019;29(10):1597–609.10.1007/s11442-019-1682-2Search in Google Scholar
 Liu HY, Zhang MY, Lin ZS, Xu XJ. Spatial heterogeneity of the relationship between vegetation dynamics and climate change and their driving forces at multiple time scales in Southwest China. Agr Forest Meteorol. 2018;256–257:10–21.10.1016/j.agrformet.2018.02.015Search in Google Scholar
 Jyoti S, Busnur RM, Keshava B, Prajithb A, Manjunatha HV, Jithin J, et al. Quantitative pollen-based reconstruction of the vegetation diversity in response to the late-Holocene climate change near Karwar, south-west coast of India. Quatern Int. 2021;599-600:95–106.10.1016/j.quaint.2021.03.026Search in Google Scholar
 Rasmus F, Simon RP. Evaluation of Earth observation based global long term vegetation trends – comparing GIMMS and MODIS global NDVI time series. Remote Sens Environ. 2012;119:131–47.10.1016/j.rse.2011.12.015Search in Google Scholar
 Zhao L, Dai AG, Dong B. Changes in global vegetation activity and its driving factors during 1982–2013. Agr Forest Meteorol. 2018;249:198–209.10.1016/j.agrformet.2017.11.013Search in Google Scholar
 Tian F, Fensholt R, Verbesselt J, Grogan K, Horion S, Wang YJ. Evaluating temporal consistency of long-term global NDVI datasets for trend analysis. Remote Sens Environ. 2015;163:326–40.10.1016/j.rse.2015.03.031Search in Google Scholar
 Zhang JH, Feng ZM, Jiang LG, Yang ZY. Analysis of the correlation between NDVI and climate factors in the Lancang River Basin. J Nat Resour. 2015;30(9):1425–35.Search in Google Scholar
 Li X, Yuan JG, Meng D. Spatial-temporal distribution of vegetation net primary productivity and its relationship with climate factors in Hebei province from 2005 to 2014. Res Soil Water Conserv. 2018;25(6):109–14.Search in Google Scholar
 Sun YL, Qian S, Yan H, Xu LL, Wang Y, Cao Y. Spatial-temporal dynamics of vegetation ecosystem in typical coal mining area of Yunnan province during 2000–2018: a case study in Lincang. Ecol Environ Sci. 2019;28(12):2381–9.Search in Google Scholar
 Yang X, Guo B, Han BM, Chen ST, Yang F, Fan YW, et al. Analysis of the spatial-temporal evolution patterns of NPP and its driving mechanisms in the Qinghai-Tibet Plateau. Resour Environ Yangtze Basin. 2019;28(12):3038–50.Search in Google Scholar
 Chen ST, Guo B, Yang F, Han BM, Fan YW, Yang X, et al. Spatial and temporal patterns of NPP and its response to climate change in the Qinghai-Tibet Plateau from 2000 to 2015. J Nat Res. 2020;35(10):2511–27.10.31497/zrzyxb.20201016Search in Google Scholar
 Peng KF, Jiang WG, Hou P, Sun CX, Zhao X, Xiao RL. Spatiotemporal variation of vegetation coverage and its affecting factors in the Three-river-source National Park. Chin J Ecol. 2020;39(10):3388–96.Search in Google Scholar
 Rong HF, Fang B. Measurement of the matching degree between urbanization and ecology in Anhui based on barycenter model. Chin Land Sci. 2017;31(6):34–41.Search in Google Scholar
 Guo B, Luo W, Han BM. Multiple spatial-temporal scale change patterns of rainfall erosivity in China over past 58 years based on gravity centre model. Geomat Nat Haz Risk. 2019;10(1):2200–19.10.1080/19475705.2019.1685012Search in Google Scholar
 Zhu L, Yang AM, Xia XX, Wu HQ. Spatial distribution pattern and change characteristics analysis of cultivated land in the Manas River Basin from 1975 to 2015. Chin J Eco-Agric. 2020;28(6):887–99.Search in Google Scholar
 Li JG, Yan QW, Xiong JB, Huang YY. Variation of vegetation index in coal Mining Areas in Guizhou Province and its affecting factors. J Ecol Rural Environ. 2016;32(3):374–8.Search in Google Scholar
 Peng WF, Zhang DM, Luo YM, Tao S, Xu XL. Influence of natural factors on vegetation NDVI using geographical detection in Sichuan Province. Acta Geogr Sin. 2019;74(9):1758–76.Search in Google Scholar
 Zhao J, Du ZQ, Wu ZT, Zhang H, Guo N, Ma ZT, et al. Seasonal variations of day and nighttime warming and their effects on vegetation dynamics in China’s temperate zone. Acta Geogr Sin. 2018;73(3):395–404.Search in Google Scholar
 Jin K, Wang F, Han JQ, Shi SY, Ding WB. Contribution of climatic change and human activities to vegetation NDVI change over China during 1982–2015. Sci Nat Resour Res. 2020;75(5):961–74.Search in Google Scholar
 Prăvălie R, Sîrodoev I, Nita IA, Cristian P, Monica D, Bogdan R, et al. NDVI-based ecological dynamics of forest vegetation and its relationship to climate change in Romania during 1987–2018. Ecol Indic. 2022;136:108629.10.1016/j.ecolind.2022.108629Search in Google Scholar
 Ram CD, Craig MT, Diane EA, Peter MK. A study over 33 years shows that carbon and nitrogen stocks in a subtropical soil are increasing under native vegetation in a changing climate. Sci Total Environ. 2021;772:145019.10.1016/j.scitotenv.2021.145019Search in Google Scholar PubMed
 Michal A, Radoslaw J, Christiaan VDT, Anshu R. Impact of climate change-induced alterations in peatland vegetation phenology and composition on carbon balance. Sci Total Environ. 2022;827:154294.10.1016/j.scitotenv.2022.154294Search in Google Scholar PubMed
 Andrey AG, Yulia VS, Olga VE, Svetlana YS, Elizaveta IV, Martin W. Expansion of Juniperus sibirica Burgsd. as a response to climate change and associated effect on mountain tundra vegetation in the Northern Urals. J Mater Sci. 2020;17(10):2339–53.10.1007/s11629-019-5925-6Search in Google Scholar
 Sun JK, Niu HP, Yuan M. Analysis on NPP spatial pattern change of terrestrial vegetation ecosystem in China. Trans Chin Soc Agric Mach. 2020;51(6):162–8.Search in Google Scholar
 Kerrylee R, Neil S, Colin DW. Surface elevation change and vegetation distribution dynamics in a subtropical coastal wetland: Implications for coastal wetland response to climate change. Estura Coast Shelf S. 2014;149:46–56.10.1016/j.ecss.2014.07.009Search in Google Scholar
 Ma ZC, Yu HB, Cao CM, Zhang QF, Hou LL, Liu YX. Spatiotemporal characteristics of fractional vegetation coverage and its influencing factors in China. Resour Environ Yangtze Basin. 2020;29(6):1310–21.Search in Google Scholar
 Han XJ, Yu JL, Shi LN, Zhao XC, Wang JJ. Spatiotemporal evolution of ecosystem service values in an area dominated by vegetation restoration: quantification and mechanisms. Ecol Indic. 2021;131:108191.10.1016/j.ecolind.2021.108191Search in Google Scholar
 Zhang M, Lin H, Long XR, Cai YT. Analyzing the spatiotemporal pattern and driving factors of wetland vegetation changes using 2000–2019 time-series Landsat data. Sci Total Environ. 2021;780:146615.10.1016/j.scitotenv.2021.146615Search in Google Scholar PubMed
 Azin Z, Esmaeil A, Ataollah E, Mohammad J, Arash M, Hamid MN, et al. Predictions of future grassland vegetation cover fluctuation under climate change scenarios. Ecol Indic. 2020;119:106858.10.1016/j.ecolind.2020.106858Search in Google Scholar
 Huo H, Sun CP. Spatiotemporal variation and influencing factors of vegetation dynamics based on geodetector: a case study of the northwestern Yunnan Plateau, China. Ecol Indic. 2021;130:108005.10.1016/j.ecolind.2021.108005Search in Google Scholar
 Liu HY, Jia JH, Lin ZS, Wang ZY, Gong HB. Relationship between net primary production and climate change in different vegetation zones based on EEMD detrending – a case study of Northwest China. Ecol Indic. 2021;122:107276.10.1016/j.ecolind.2020.107276Search in Google Scholar
 Xu XJ, Liu HY, Jiao FS, Gong HB, Lin ZS. Nonlinear relationship of greening and shifts from greening to browning in vegetation with nature and human factors along the Silk Road Economic Belt. Sci Total Environ. 2021;766:142553.10.1016/j.scitotenv.2020.142553Search in Google Scholar PubMed
 Ning K, Wang NA, Yang ZJ, Zhang LL, Wang YX, Li ZL, et al. Holocene vegetation history and environmental changes inferred from pollen records of a groundwater recharge lake, Badain Jaran Desert, northwestern China. Palaeogeogr Palaeocl. 2021;577:110538.10.1016/j.palaeo.2021.110538Search in Google Scholar
 Mu SJ, Li B, Yao J, Yang GS, Wan RR, Xu XB. Monitoring the spatio-temporal dynamics of the wetland vegetation in Poyang Lake by Landsat and MODIS observations. Sci Total Environ. 2020;725:138096.10.1016/j.scitotenv.2020.138096Search in Google Scholar PubMed
 Guilherme CO, Daniel MA, Elpídio IFF, Gustavo VV, Márcio RF, Carlos EG, et al. Soil predictors are crucial for modelling vegetation distribution and its responses to climate change. Sci Total Environ. 2021;780:146680.10.1016/j.scitotenv.2021.146680Search in Google Scholar PubMed
 Andrés MS, Hyeyeong C, Ryan MB, Paul RE, James HT. Climate exposure shows high risk and few climate refugia for Chilean native vegetation. Sci Total Environ. 2021;785:147399.10.1016/j.scitotenv.2021.147399Search in Google Scholar
© 2022 ShiJun Wang et al., published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.