Vegetation Dynamics and Their Response to Climate Change in the Yellow River Basin: Based on Climatic Dry and Wet Zoning Scales

WANG Ruifang¹, LYU Baoqi², ZHANG Wenjing¹
¹Henan College of Surveying and Mapping, Zhengzhou 450000, Henan, China
²Henan Institute of Surveying and Mapping, Zhengzhou 450000, Henan, China

Abstract

The Yellow River Basin represents a critical region for ecological protection and economic development in China. Investigating vegetation change characteristics across different dry and wet zones within the basin is essential for adjusting ecological restoration strategies to address potential threats from environmental change. Based on the kernel normalized difference vegetation index (kNDVI) and key meteorological drivers [precipitation (PRE) and temperature (TEM)] from 2000 to 2022, this study employed multivariate statistical methods to analyze spatiotemporal patterns of vegetation dynamics across different climatic dry and wet zones. The Geodetector model and constraint effect method were used to analyze driving factors of vegetation change, identifying both commonalities and regional differences in how vegetation responds to meteorological factors across dry and wet zones. Results show that: (1) Vegetation kNDVI values exhibit latitudinal distribution, with the highest average annual kNDVI (0.49) in humid zones. During 2000–2022, 84.58% of the basin showed an upward trend, with the most significant improvements in arid (68.36%) and semi-arid zones (93.08%). (2) Precipitation generally exerts stronger influence on vegetation than temperature across the Yellow River Basin, with partial correlation coefficients of 0.36 and 0.19 at the basin scale, respectively. This difference is particularly pronounced in semi-arid zones, where partial correlation coefficients reach 0.43 for precipitation and 0.22 for temperature. (3) Regarding spatial stratified heterogeneity, the q-value for precipitation (0.5338) exceeds that for temperature (0.2283) at the basin scale. Moreover, precipitation's q-value peaks in semi-arid zones (0.4519), while temperature's q-value peaks in semi-humid zones (0.2491). Meteorological factors produce distinct constraint lines characterizing vegetation dynamic responses across different dry and wet zones. These findings provide important references for adjusting and formulating basin-scale ecological protection strategies and are significant for promoting high-quality development in the Yellow River Basin.

Keywords: vegetation change; kernel normalized vegetation index (kNDVI); dry and wet zone; restraint effect; Yellow River Basin

Vegetation plays a crucial role in terrestrial ecosystems, regulating material and energy exchange between land and atmosphere. As an important indicator of environmental change, vegetation is sensitive to the combined impacts of climate change and human activities. The Yellow River Basin, as a key ecological restoration region in China, has attracted considerable attention regarding vegetation change detection and attribution. Remote sensing techniques combined with vegetation indices are the most common methods for monitoring vegetation dynamics and growth conditions. As the most widely used vegetation index, the normalized difference vegetation index (NDVI) is closely related to leaf density, photosynthetic active radiation, vegetation productivity, and accumulated biomass. However, its accuracy is affected by the abundance of thick-leaf plants and sensitivity to canopy background brightness variations. In contrast, Camps-Valls et al. proposed the kernel normalized difference vegetation index (kNDVI), an improvement based on kernel methods that addresses difficulties in scale conversion and nonlinear problems. By introducing kernel technology, kNDVI provides more robust and accurate vegetation monitoring across different scales and under nonlinear conditions. kNDVI captures more information about vegetation biomass accumulation and demonstrates stronger correlation with actual vegetation productivity than traditional NDVI. While traditional NDVI saturates when vegetation density reaches a certain threshold, kNDVI effectively reduces this saturation effect through kernel function application, maintaining sensitivity and accuracy at higher vegetation densities, reducing bias, and adapting to material cycling. This has proven its effectiveness in evaluating vegetation dynamics at large scales.

Vegetation greening exhibits confirmed spatiotemporal heterogeneity across seasons, regions, and land cover types. While climate change directly determines vegetation growth physiological activities, the main meteorological drivers of vegetation development vary by region. Separating the relationship between different dry/wet zone types and vegetation change is challenging. Considering increasing evaporation demand and decreasing soil moisture availability, negative effects of warming and water stress (such as drought) on vegetation greening have been observed in many regions, including the Amazon, temperate and boreal Eurasia, and the Congo Basin. Additionally, drought trend magnitude fluctuates with temporal scale expansion, and vegetation growth responses to meteorological drought show significant differences across multi-scales and vegetation categories. Evidence indicates that vegetation responses to temperature in north temperate ecosystems have weakened over the past 30–40 years. In summary, vegetation greening is a complex dynamic process influenced not only by climatic factors but also by different environmental conditions. To better manage vegetation resources across different climate zones, investigating the relationship between climate and vegetation under different dry/wet zone types is essential.

1.1 Study Area Overview

The Yellow River Basin is located in northwestern China (95°53′–119°05′E, 32°10′–41°50′N), spanning nine provinces with a total length of 5,464 km. Originating from the Bayan Har Mountains and flowing into the Bohai Sea, the basin extends from the Yinshan Mountains in the north to the Qinling Mountains, with terrain gradually descending from west to east. The western source region consists of high mountains averaging 4,000 m in elevation, while the central Loess Plateau sits at 1,300–2,200 m with severe soil erosion. The downstream area comprises the flat Huang-Huai-Hai Plain. The basin features typical arid, semi-arid, semi-humid, and humid climate zones, with vegetation showing significant spatial differentiation. The upper source region is dominated by alpine vegetation, including alpine meadows, steppes, and marsh/aquatic vegetation. The middle Loess Plateau features grasslands, shrublands, and mixed forest-steppe, while the downstream alluvial plain is dominated by cultivated vegetation with local coniferous-broadleaf mixed forests.

1.2 Data Sources

1.2.1 kNDVI Data

kNDVI is a vegetation index based on kernel (machine learning) functions, representing an improvement on NDVI that addresses difficulties in scale conversion and nonlinear problems. Through kernel technology, kNDVI provides more robust and accurate vegetation monitoring across different scales and under nonlinear conditions. In this study, MODIS data products (MOD13Q1) with 1,000 m spatial resolution were obtained from the US Earth Resources Observation System Data Center (https://ladsweb.modaps.eosdis.nasa.gov) for 2000–2022. Monthly kNDVI values were calculated based on formulas proposed by Camps-Valls et al. The kernel function uses a hyperbolic tangent function, where NIR and Red represent near-infrared and red spectral reflectance, τ is a scaling parameter, and σ is a length scale parameter linearly proportional to the mean values of near-infrared and red reflectance. When σ = 0.5(μ_NIR + μ_Red), kNDVI maintains accuracy while ensuring simplicity.

1.2.2 Driving Factor Data

Annual average temperature and precipitation datasets for the Yellow River Basin were obtained from the Resource and Environmental Science Data Platform (http://www.resdc.cn/). Terrain data came from the Shuttle Radar Topography Mission digital elevation model. Land use data spanning 2000–2020 were obtained from the National Earth System Science Data Center, classifying land use into six major types: cropland, forest, grassland, shrubland, water bodies, ice/snow, barren land, construction land, and wetland. All data were resampled to match kNDVI resolution before analysis. Detailed dataset descriptions are provided in Table 1.

1.3 Trend Analysis

To analyze trends in Yellow River Basin vegetation, this study employed linear regression on a per-pixel basis. The slope of the linear regression equation indicates the trend magnitude, calculated as: Slope = (nΣ(i·kNDVI_i) - ΣiΣkNDVI_i) / (nΣi² - (Σi)²), where n is the number of years (23 in this study), i is the data point index, and kNDVI_i is the kNDVI value for year i. When Slope > 0, vegetation shows a greening trend; when Slope < 0, vegetation shows a browning trend.

1.4 Partial Correlation Analysis

Partial correlation analysis examines the degree of association between two variables while controlling for other factors. In models or systems with multiple factors, partial correlation coefficients represent relationships while keeping other factors constant. The formula is: r_xy,z = (r_xy - r_xz·r_yz) / √[(1 - r_xz²)(1 - r_yz²)], where r_xy,z is the partial correlation between variables x and y controlling for z; r_xy, r_xz, and r_yz are correlation coefficients between respective variable pairs. This study calculated partial correlations between kNDVI and meteorological factors at the pixel scale.

1.5 Geodetector Model

The Geodetector model detects spatial differences and reveals underlying driving forces. This study used factor detection to quantitatively analyze the influence of precipitation and temperature on kNDVI changes across different climatic dry/wet zones. The q-statistic measures explanatory power, calculated as: q = 1 - (ΣN_hσ_h²) / (Nσ²), where q ∈ [0,1] indicates stronger explanatory power with larger values; h is the stratification variable; N_h and N are sample sizes for stratum h and the entire region; σ_h² and σ² are variances of Y values for stratum h and the entire region. Interaction detection identifies interaction effects between different extreme climate indices by comparing individual and joint explanatory powers q(X1), q(X2), and q(X1∩X2).

1.6 Constraint Effect

The constraint line extraction method provides new insights for exploring relationships and mechanisms between two variables. The process involves: (1) dividing the constraint factor's value range into intervals to generate columns on the x-axis; (2) selecting the 95th percentile of each column as boundary points to reduce outlier effects; and (3) determining constraint line type based on scatterplot shape and goodness-of-fit (R²). This study constructed two-dimensional coordinate systems with meteorological factors as the x-axis and kNDVI as the y-axis, using quantile partitioning to plot constraint lines between variable pairs.

2.1 Spatiotemporal Patterns of Vegetation Dynamics and Meteorological Variables

2.1.1 Intra-annual Variation Characteristics

Across different dry/wet zones in the Yellow River Basin, monthly kNDVI shows significant peak fluctuations. Humid and semi-humid zones exhibit the highest kNDVI values, reaching peaks of 0.68 and 0.61, respectively. Arid and semi-arid zones show lower annual kNDVI fluctuations, with peaks of 0.31 and 0.41. This difference primarily results from insufficient precipitation and water supply limiting vegetation growth in arid/semi-arid zones, while adequate precipitation in humid/semi-humid zones provides favorable moisture conditions, enabling higher kNDVI values during growing seasons.

Intra-annual precipitation shows clear seasonality, concentrated in June–September and accounting for over 70% of annual precipitation. In semi-humid zones, precipitation shows a single peak in July (149.62 mm). Precipitation trends in humid, semi-arid, and arid zones show consistent fluctuating increases, with maximum values appearing near July–August (138.11 mm, 98.68 mm, and 44.74 mm, respectively). Temperature trends across all zones show similar unimodal curves, with semi-humid, semi-arid, and arid zones having comparable temperatures significantly higher than humid zones. Peak temperatures occur in July, reaching 22.78°C in arid zones, 19.53°C in semi-humid zones, 18.33°C in semi-arid zones, and 9.15°C in humid zones.

2.1.2 Spatial Distribution Patterns

The long-term average kNDVI during 2000–2022 shows significant spatial heterogeneity, decreasing from southeast to northwest in a latitudinal distribution. Average annual kNDVI values are 0.49 in humid zones, 0.39 in semi-humid zones, 0.29 in semi-arid zones, and 0.17 in arid zones. Precipitation shows a northwest-to-southeast increasing gradient, with average annual precipitation of 698.39 mm in humid zones, 617.51 mm in semi-humid zones, 435.37 mm in semi-arid zones, and 205.35 mm in arid zones. Mean annual temperature increases stepwise from west to east, with the highest temperature in eastern semi-humid zones (11.93°C) and the lowest in western semi-humid zones (-3.18°C). Temperatures in arid, semi-arid, and humid zones fall between these extremes at 8.35°C, 5.82°C, and 1.25°C, respectively.

2.2 Spatiotemporal Variation Characteristics of Vegetation Dynamics and Meteorological Factors

2.2.1 Interannual Variation Characteristics

Using pixel-scale averages, this study obtained comprehensive indicators of vegetation condition and climate change across the Yellow River Basin. kNDVI shows a clear upward trend from 2000 to 2022, increasing at 0.0044·a⁻¹, indicating significant vegetation greening. Annual precipitation and temperature show substantial fluctuations with large interannual differences. Precipitation increases slowly across all zones at rates of 1.8565 mm·a⁻¹ (arid), 3.0710 mm·a⁻¹ (semi-arid), 3.7315 mm·a⁻¹ (semi-humid), and 5.6063 mm·a⁻¹ (humid). Temperature trends are similar across zones, with arid and semi-arid zones showing highly consistent warming at 0.0226°C·a⁻¹ and 0.0210°C·a⁻¹, respectively.

2.2.2 Spatial Distribution of Dynamic Changes

To examine spatial distribution of trends, this study calculated Sen's slope for kNDVI, precipitation, and temperature across the basin. High kNDVI slope values concentrate in semi-arid regions (mean: 0.0055·a⁻¹), followed by semi-humid (0.0030·a⁻¹), arid (0.0022·a⁻¹), and humid zones (0.0007·a⁻¹). Precipitation slope shows significant spatial heterogeneity, increasing at 3.341 mm·a⁻¹ overall. Temperature slope varies relatively little across zones, with a basin-wide increase of 0.024°C·a⁻¹ and distinct east-west distribution patterns.

Using Theil-Sen trend analysis and Mann-Kendall tests, vegetation changes were classified into five types: significant increase, slight increase, no change, slight decrease, and significant decrease. Across the entire basin, 65.36% of the area shows significant greening, primarily in semi-arid and semi-humid zones; 19.22% shows slight greening; and only 1.87% shows significant browning, mainly in densely populated cities like Xi'an, Sanmenxia, and Zhengzhou in semi-humid zones.

Different dry/wet zones show distinct vegetation change patterns. Semi-arid zones have the largest proportion of significant greening (77.86%), followed by semi-humid zones (57.18%), reflecting the impact of ecological restoration projects. Slight greening dominates in humid zones (40.84%). In arid zones, besides significant greening (48.55%), no-change areas account for 28.67%—the highest proportion among all zones—due to harsh climate conditions limiting vegetation recovery.

2.3 Partial Correlations Between Vegetation Dynamics and Meteorological Factors

The interaction between temperature and precipitation significantly influences vegetation dynamics. Partial correlation analysis at the pixel scale reveals that kNDVI shows positive correlations with both precipitation and temperature, but precipitation dominates vegetation changes across the Yellow River Basin. The partial correlation coefficient between kNDVI and precipitation (0.36) exceeds that with temperature (0.19), a pattern consistent across all dry/wet zones but most pronounced in semi-arid zones (0.43 vs. 0.22).

Spatially, high partial correlations between kNDVI and precipitation concentrate in arid and semi-arid zones, while weaker correlations appear in semi-humid and humid zones. Temperature partial correlations show less spatial variation, with coefficients of 0.20 (arid), 0.19 (semi-arid), 0.18 (semi-humid), and 0.17 (humid).

2.4 Driving Factors and Constraint Effects

2.4.1 Analysis of Driving Factors

The Geodetector model quantified meteorological factors' influence on kNDVI through q-values. At the basin scale, precipitation's q-value (0.5338) substantially exceeds temperature's (0.2283), confirming precipitation as the dominant factor. Under interactive effects, the combined q-value reaches 0.6847, surpassing individual factors and indicating synergistic meteorological controls on vegetation.

Across dry/wet zones, precipitation's influence peaks in semi-arid zones (q = 0.4519) and is weakest in arid zones (q = 0.1823). Temperature's influence peaks in semi-humid zones (q = 0.2491) and is also weakest in arid zones (q = 0.0987). The interaction effect is strongest in semi-arid zones (q = 0.6128), where water availability and temperature jointly determine vegetation growth cycles.

2.4.2 Constraint Effects on Vegetation Dynamics

In complex ecosystems, constraint lines help eliminate confounding factors to reveal maximum response values under limiting conditions. This study explored constraint mechanisms between meteorological factors and kNDVI across different dry/wet zones.

At the basin scale, constraint lines between kNDVI and meteorological factors show humpback shapes—promoting vegetation growth initially but constraining it after reaching thresholds. In arid zones, precipitation constraint lines show multi-peak fluctuations due to unstable precipitation patterns, while temperature constraint lines show stepwise patterns with rapid vegetation changes within 7–8°C thresholds but diminishing temperature benefits beyond this range. In semi-arid and semi-humid zones, both precipitation and temperature show single-hump constraint lines, promoting growth within thresholds but inhibiting beyond them. Humid zones show constraint lines consistent with the basin pattern: initial promotion followed by inhibition and synergy.

3 Discussion

This study comprehensively analyzed the spatial distribution, interannual variation, and meteorological responses of kNDVI across different dry/wet zones in the Yellow River Basin. Compared with previous research, this work emphasizes differential vegetation changes across dry/wet zones and their distinct responses to meteorological factors.

Climate change directly determines vegetation physiological activities, but primary meteorological drivers vary regionally. In the Yellow River Basin, vegetation kNDVI increased at 0.0044·a⁻¹ from 2000–2022. Significant spatial heterogeneity exists in both meteorological factors and vegetation responses. Climate change positively contributed to vegetation activity mainly in semi-arid and semi-humid zones, where warming promotes growth and increased precipitation meets photosynthetic water demands, facilitating vegetation recovery. In arid and semi-arid regions, increased precipitation promotes soil organic matter decomposition, improving nutrients and moisture to enhance vegetation activity. Overall, vegetation dynamics reflect the synergistic effects of multiple meteorological factors.

Previous studies indicate that vegetation growth in arid/semi-arid regions is primarily limited by thermal conditions (solar radiation, temperature). Under global warming, Yellow River Basin vegetation shows increasing trends, with water conditions surpassing thermal factors as the most important meteorological influence—consistent with our findings that precipitation dominates vegetation dynamics. Temperature directly affects vegetation growth and metabolism; suitable temperatures promote growth and increase coverage, while extremes limit growth. Semi-arid vegetation shows certain adaptability to water and temperature, but changes beyond adaptive ranges significantly impact growth and coverage.

4 Conclusions

1) Vegetation kNDVI in the Yellow River Basin shows latitudinal distribution, gradually increasing from northwest to southeast, indicating improving vegetation density and condition from arid to humid zones. Significant differences exist across zones: arid zones have the lowest kNDVI (0.17), which increases progressively through semi-arid (0.29), semi-humid (0.39), and humid zones (0.49). During 2000–2022, 84.58% of the basin showed increasing vegetation trends.

2) Precipitation dominates vegetation changes across the Yellow River Basin compared to temperature, with mean partial correlation coefficients of 0.36 versus 0.19. This pattern holds across all dry/wet zones, particularly in semi-arid regions where the difference is most pronounced. In arid zones, precipitation is the key factor for vegetation change.

3) Based on q-values, precipitation is the primary vegetation growth factor, with stronger influence than temperature. The interaction between precipitation and temperature explains vegetation growth better than individual factors. Precipitation's influence peaks in semi-arid zones, while temperature's influence peaks in semi-humid zones, providing important guidance for vegetation management and ecological protection strategies.

4) At the basin scale, relationships between kNDVI and meteorological factors show humpback constraint lines, indicating enhanced constraints beyond thresholds. Across dry/wet zones, constraint line morphologies vary: arid zones show fluctuating patterns, while other zones show humpback shapes, demonstrating that vegetation growth is initially promoted by meteorological factors but inhibited after exceeding thresholds.

References

[1] Ma Xiaohong, Lin Fei, Yuan Liming, et al. Vegetation coverage change and its response to ecological protection project in Fenhe River Basin[J]. Journal of Desert Research, 2023, 43(3): 86-95.

[2] Luo Min, Meng Fanhao, Wang Yunqian, et al. Mutual feedback relationship between vegetation GPP and soil moisture in China under climate change[J]. Acta Geographica Sinica, 2024, 79(1): 218-239.

[3] Wu Ling, Liu Xiangnan, Liu Meiling, et al. Review of the detection and attribution of multi-type forest disturbances using an ensemble of spatio-temporal spectral information from remote sensing images[J]. National Remote Sensing Bulletin, 2024, 28(3): 558-575.

[4] Zeng Y, Hao D, Huete A, et al. Optical vegetation indices for monitoring terrestrial ecosystems globally[J]. Nature Reviews Earth & Environment, 2022, 3(7): 477-493.

[5] Tan C W, Zhang P P, Zhou X X, et al. Quantitative monitoring of leaf area index in wheat of different plant types by integrating NDVI and Beer-Lambert law[J]. Scientific Reports, 2020, 10(1): 929.

[6] Han Wanqiang, Jin Guili, Yue Yonghuan, et al. Spectral characteristics and modified vegetation index of three main plants of Seriphidium transiliense desert grassland[J]. Xinjiang Agricultural Sciences, 2020, 57(5): 950-957.

[7] Gao Q, Zhu W, Schwartz M W, et al. Climatic change controls productivity variation in global grasslands[J]. Scientific Reports, 2016, 6(1): 26958.

[8] Kang Yao, Guo Enliang, Wang Yongfang, et al. Application of temperature vegetation dryness index for drought monitoring in Mongolian Plateau[J]. Chinese Journal of Applied Ecology, 2021, 32(7): 2534-2544.

[9] Li H, Cao Y, Xiao J, et al. A daily gap-free normalized difference vegetation index dataset from 1981 to 2023 in China[J]. Scientific Data, 2024, 11(1): 527.

[10] Jin H, Chen X, Wang Y, et al. Spatio-temporal distribution of NDVI and its influencing factors in China[J]. Journal of Hydrology, 2021, 603: 127129.

[11] Camps-Valls G, Campos-Taberner M, Moreno-Martínez Á, et al. A unified vegetation index for quantifying the terrestrial biosphere[J]. Science Advances, 2021, 7(9): eabc7447.

[12] Wang Q, Moreno-Martínez Á, Muñoz-Marí J, et al. Estimation of vegetation traits with kernel NDVI[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2023, 195: 408-417.

[13] Li Zhijie, Huang Guichao, Ma Ziyi, et al. Spatial and temporal variation patterns of urban habitat quality and its influencing factors based on land use change[J]. Journal of Chinese Urban Forestry, 2024, 22(2): 135-143.

[14] Song Jinxi, Qi Guizeng, She Dunxian, et al. Response of vegetation productivity to wet and dry changes in China[J]. Acta Geographica Sinica, 2023, 78(7): 1764-1778.

[15] Han Yunhuan, Ma Zhuguo, Li Mingxing, et al. Vegetation changes and their relationship with climate factors in different dry/wet areas over China[J]. Chinese Journal of Atmospheric Sciences, 2023, 47(6): 1680-1692.

[16] Tian Qian, Yang Fang, Wang Zhaohuan, et al. Response of soil CO₂ emission in terrestrial ecosystems to simulated warming and its influencing factors[J]. Acta Ecologica Sinica, 2024, 44(5): 1928-1939.

[17] Guo Fuyin, Liu Xiaohuang, Zhang Wenbo, et al. Evolution of spatiotemporal patterns of habitat quality and analysis of driving forces in Yellow River Basin (Henan section) from 2000 to 2040[J]. Geoscience, 2024, 38(3): 599-611.

[18] He B, Chen A, Jiang W, et al. The response of vegetation growth to shifts in temperature trend in China[J]. Journal of Geographical Sciences, 2017, 27(7): 801-816.

[19] Zhang J Y, Yin Y H, Li B Y. A new scheme for climate regionalization in China[J]. Acta Geographica Sinica, 2010, 65(1): 3-12.

[20] Yang J, Huang X. The 30 m annual land cover dataset and its dynamics in China from 1990 to 2019[J]. Earth System Science Data, 2021, 13(8): 3907-3925.

[21] Fu Kaixiang, Jia Guodong, Yu Xinxiao, et al. Evaluation of ecological environment quality and analysis of driving mechanism in Tulufan-Hami region based on improved remote sensing ecological indices[J]. Acta Ecologica Sinica, 2024, 44(9): 3911-3923.

[22] Liu Ze, Chen Jianping. Correlation between temporal-spatial changes of vegetation and climate factors in Beijing[J]. Geological Bulletin of China, 2021, 40(12): 2159-2166.

[23] Wang Chunya, Wang Jinniu, Cui Xia, et al. Spatio-temporal change in vegetation patterns and its climatic drivers in the core region of Three Parallel Rivers in southeast Tibet[J]. Geographical Research, 2021, 40(11): 3191-3207.

[24] Zhang Z, Song Y, Wu P. Robust geographical detector[J]. International Journal of Applied Earth Observation and Geoinformation, 2022, 109: 102782.

[25] An Q, Yuan X, Zhang X, et al. Spatio-temporal interaction and constraint effects between ecosystem services and human activity intensity in Shaanxi Province, China[J]. Ecological Indicators, 2024, 160: 111937.

[26] Xue Lianqing, Wang Wenzhuang, Liu Yuanhong, et al. Spatiotemporal accumulation response of vegetation water use efficiency to drought in the Yellow River Basin[J]. Water Resources Protection, 2023, 39(4): 32-41.

[27] Xue Lianqing, Xiao Ying, Liu Yuanhong, et al. Response of gross primary productivity of vegetation to persistent drought-induced water deficit in the Yellow River Basin[J]. Water Resources Protection, 2024, 40(3): 44-51.

[28] Bai X, Fan Z, Yue T. Dynamic pattern effect relationships between precipitation and vegetation in the semi-arid and semi-humid area of China[J]. Catena, 2023, 232: 107425.

[29] Li C, Fu B, Wang S, et al. Drivers and impacts of changes in China's drylands[J]. Nature Reviews Earth & Environment, 2021, 2(12): 858-873.

[30] Liu L, Gou X, Wang X, et al. Relationship between extreme climate and vegetation in arid and semi-arid mountains in China: A case study of the Qilian Mountains[J]. Agricultural and Forest Meteorology, 2024, 348: 109938.

[31] Jiang Kaisheng, Liu Yu, Liu Ying, et al. Research on the relationship between vegetation coverage and climate factor response in northwest China[J]. Science of Surveying and Mapping, 2023, 48(6): 172-180.

[32] Sun G Q, Li L, Li J, et al. Impacts of climate change on vegetation pattern: Mathematical modeling and data analysis[J]. Physics of Life Reviews, 2022, 43: 239-270.

[33] Wu Dafang, Ma Peifang, Li Long, et al. Spatio-temporal variation and factor detection of land use transition based on geodetector: A case study of Zhuhai[J]. Journal of South China Normal University (Natural Science Edition), 2023, 55(4): 50-61.

[34] Wang K, Zhou J, Tan M L, et al. Impacts of vegetation restoration on soil erosion in the Yellow River Basin, China[J]. Catena, 2024, 234: 107547.

[35] Zhu Y, Zhang Y, Zhang Z, et al. Converted vegetation type regulates the vegetation greening effects on land surface albedo in arid regions of China[J]. Agricultural and Forest Meteorology, 2022, 324: 109119.

[36] Yin Y, Deng H, Ma D. Complex effects of moisture conditions and temperature enhanced vegetation growth in the arid/humid transition zone in northern China[J]. Science of the Total Environment, 2024, 160: 111937.

[37] Bita C, Gerats T. Plant tolerance to high temperature in a changing environment: Scientific fundamentals and production of heat stress tolerant crops[J]. Frontiers in Plant Science, 2013, 4: 273.

[38] Xu H J, Wang X P, Zhao C Y. Drought sensitivity of vegetation photosynthesis along the aridity gradient in northern China[J]. International Journal of Applied Earth Observation and Geoinformation, 2021, 102: 102418.