Spatiotemporal Variations in MODIS EVI and MODIS LAI and the Responses to Meteorological Drought across Different Slope Conditions in Karst Mountain Regions

Abstract

Based on monthly MODIS EVI and LAI data from 2001 to 2020, combined with the Standardized Precipitation Evapotranspiration Index (SPEI), this study employs Theil–Sen trend analysis, Mann–Kendall (MK) test, Hurst index analysis, and correlation analysis to comparatively analyze the overall vegetation trends, spatial distribution characteristics, and future trends of different vegetation types in Guizhou Province under varying slope conditions. The study also explores the response of vegetation to SPEI at different time scales across different slopes. The results indicate the following: (1) From 2001 to 2020, the average values of EVI (0.34%/a) and LAI (1.4%/a) during the growing season exhibited an increasing trend, with the improved vegetation areas primarily concentrated in the western region of Guizhou, while the degradation areas were mainly located in the central and eastern regions. (2) Under different slope conditions, EVI generally showed slight improvement, while LAI exhibited significant improvement, with dry-lands experiencing the largest changes. Future trends indicate continuous improvement, but the proportion of vegetation improvement area decreases with increasing slope. When the slope is less than 5°, the proportion of vegetation improvement area is the highest. (3) The positive correlation between EVI, LAI, and SPEI at different time scales is stronger than the negative correlation, with the strongest correlations observed when the slope is less than 5°. When the slope exceeds 35°, the relationship between vegetation and drought response is almost unaffected by the slope. These findings provide a scientific basis for vegetation growth monitoring and the study of climate change and vegetation interactions in Guizhou Province.

1. Introduction

Drought is one of the most common and destructive natural disasters worldwide, characterized by its slow onset, far-reaching impacts, and widespread effects [1]. The IPCC AR6 report clearly states [2,3] that with further global warming, the frequency and intensity of extreme heat events will continue to rise, leading to changes in global precipitation patterns. This will further exacerbate water scarcity and uneven distribution, thereby affecting the spatial and temporal distribution, intensity, and frequency of droughts, with profound implications for the sustainable development of socio-economic systems and ecological environments. Vegetation, as a crucial component of terrestrial ecosystems, plays a key role in material storage and energy cycling [4]. It is not only a vital indicator of the structural and functional stability of terrestrial ecosystems [5], but also serves as an important feedback medium for drought severity. As a critical driver affecting vegetation growth [6,7,8,9,10], the intensity, frequency, and spatial extent of drought directly determine the condition of vegetation growth.

The extent and manner in which drought affects vegetation vary depending on the region, vegetation type, and the duration and intensity of the drought itself [11,12,13]. Differences in regional climate types and underlying surface conditions lead to variations in precipitation and surface water retention capacity, resulting in differing frequencies and intensities of drought, as well as varied responses of vegetation to drought within different regions [14,15]. Therefore, timely and accurate monitoring and assessment of drought, along with an in-depth exploration of the relationship between drought and vegetation response, are crucial for formulating effective drought management strategies. In recent years, remote sensing data and vegetation indices have been widely used in drought monitoring. Vegetation indices are indicators that can rapidly and effectively reflect the state of vegetation growth, with the Normalized Difference Vegetation Index (NDVI) being one of the most commonly used indices in vegetation studies [16,17,18]. Although there is a wealth of research on NDVI, it has certain limitations, such as the presence of atmospheric noise and the tendency to saturate in areas with moderate-to-high vegetation cover [19]. To overcome these limitations, researchers have proposed new vegetation indices such as the Enhanced Vegetation Index (EVI) and the Leaf Area Index (LAI), and extensive studies have been conducted on these indices [20,21,22]. Studies have shown that EVI improves upon some of NDVI’s shortcomings while retaining its unique advantages, whereas LAI provides a more accurate reflection of dynamic vegetation changes and better reveals the interactions between vegetation and climate [23,24].

As a typical karst region characterized by ecological fragility and climate sensitivity, Guizhou Province faces critical challenges in vegetation restoration and reconstruction, which are pivotal for regional ecological management. Against the backdrop of global climate warming, extreme climate events, such as droughts, pose severe threats to regional vegetation [25,26]. In this context, investigating the differential responses of vegetation to climate change across the region, based on various vegetation indices, is of significant practical importance. However, current research predominantly focuses on the relationship between drought and vegetation response, with limited consideration given to the influence of topographic factors. As a representative karst mountainous area, Guizhou’s topographic factors, such as slope, play a crucial role in influencing the relationship between vegetation and drought response. Yet, research on the response of vegetation to meteorological drought under different slope conditions remains limited, and the underlying response mechanisms are not yet fully understood. Therefore, this study focuses on Guizhou Province, utilizing MODIS EVI and LAI images, and site-based climate data from 2001 to 2020. Employing methods such as Theil–Sen slope analysis, Mann–Kendall trend test, and Hurst index analysis, combined with spatial and statistical analysis tools in ArcGIS v10.3, this study conducts a comprehensive analysis of the spatiotemporal changes in vegetation in Guizhou over the past 20 years. It delves into the future trends of different vegetation types under varying slope conditions in the karst mountainous region and their responses to meteorological droughts at different temporal scales. The primary objectives of this study are (1) to analyze the vegetation change trends and spatial distribution characteristics in the karst mountainous region, and thereby predict future trends in vegetation types; (2) to reveal the response relationship between slope and vegetation in the karst region; and (3) to elucidate the response relationship between vegetation and the Standardized Precipitation Evapotranspiration Index (SPEI) at different temporal scales across various slope conditions. To achieve these objectives, the study is structured around the following four key areas: (1) using the Theil–Sen median slope analysis and Mann–Kendall trend test to analyze the spatiotemporal changes in vegetation during the growing season (Section 3.1); (2) based on the findings in Section 3.1, utilizing ArcGIS for spatial analysis of vegetation changes across different slopes (Section 3.2); (3) applying Theil–Sen median slope analysis and Mann–Kendall trend test to assess the spatiotemporal changes in meteorological drought at different temporal scales (Section 3.3); and (4) investigating the response of vegetation to SPEI across different slopes and temporal scales using correlation analysis (Section 3.4). The results of this study aim to provide scientific evidence for selecting appropriate vegetation indices and predicting future vegetation trends in Guizhou Province. Additionally, they offer valuable insights for enhancing the research and monitoring of vegetation and climate change in karst regions, and have important implications for assessing regional vegetation restoration and reconstruction.

2. Materials and Methods

2.1. Study Area

Guizhou Province is located in the eastern part of the Yunnan–Guizhou Plateau in Southwest China (103°36′–109°35′ E, 24°37′–29°13′ N) (Figure 1). The terrain in the province is high in the west and low in the east, sloping from the center to the north, east, and south, with an average altitude of approximately 1100 m. In addition, carbonate rocks are widely distributed in the province, the karst geomorphology is strongly developed, the surface is deeply cut, the terrain is broken, and it is the center of the southwestern eco-logically fragile karst landscape. The province has a humid subtropical monsoon climate, with abundant annual precipitation of 800–1600 mm but with uneven spatiotemporal distribution. It has a multi-year average temperature of 15.5 °C, little sunshine with the number of cloudy days around the territory generally exceeding 150 days, and a multi-year average relative humidity of 78.5%. The study area has diverse vegetation types, with evergreen broad-leaved forests being the main vegetation cover. Since the implementation of the policies for reforestation and grassland restoration, the forest coverage in the province has significantly increased. Guizhou Province is predominantly characterized by plateaus and mountains, with hills and basins as secondary features. The slopes vary significantly across different areas. Studying the response of vegetation to climate under different slope conditions helps monitor the suitability of vegetation growth.

2.2. Data Sources and Processing

The remote sensing image data used in this study are sourced from the Data Release Center of the National Aeronautics and Space Administration (NASA). The first type of remote sensing imagery used is the composite product MODIS EVI from the MOD13Q1 series, with a spatial resolution of 250 m × 250 m and a temporal resolution of 16 days. The second type of remote sensing image used is the synthetic product MODIS LAI from the MOD15A2H series data, with a spatial resolution of 500 m × 500 m and a temporal resolution of 8 days. Both types of remote sensing images cover a time span from 2001 to 2020, totaling 20 growing seasons (April to October) with monthly vegetation index data. Preprocessing of the raster images, including format conversion, re-projection, and mosaic cropping, was conducted using ArcGIS software.

Meteorological data were sourced from the China Meteorological Data Sharing Network (http://data.cma.cn/, accessed on 2 September 2021), which include measured data on daily precipitation, temperature, sunshine duration, wind speed, and other factors from 31 meteorological stations in Guizhou Province from 2001 to 2020. Vegetation coverage data were derived from the 2010 Remote Sensing Monitoring Data of China’s Land-Use Status provided by the Chinese Academy of Sciences (http://www.globallandcover.com/, accessed on 15 July 2021), the spatial resolution was 250 m. And different vegetation types in Guizhou Province were classified and divided based on research requirements. Ultimately, six different vegetation types were selected for analysis and research: dry-land, grassland, paddy field, evergreen broad-leaved forest, evergreen needle-leaved forest, and deciduous broad-leaved forest. Using ArcGIS software and the Digital Elevation Model (DEM) provided by the Geospatial Data Cloud platform of the Chinese Academy of Sciences (http://www.gscloud.cn/, accessed on 10 May 2021), slope data were extracted.

2.3. Methods

2.3.1. SPEI

Vicente-Serrano et al. proposed the Standardized Precipitation Evapotranspiration Index (SPEI) based on the water balance model in 2010 [27]. SPEI integrates the effects of both precipitation and evapotranspiration on drought development, and it is more sensitive to temperature variations. It combines the advantages of convenient calculation similar to the SPI (Standardized Precipitation Index) and the ability to analyze multiple time scales, making it suitable for comparisons at different scales and spatial levels. It can better reflect the new characteristics of drought under the background of climate warming. Please refer to reference [28] for the specific calculation method.

2.3.2. Pixel-Based EVI and LAI Trend Analysis

This study adopts Theil–Sen median trend analysis and utilizes the Mann–Kendall test for significance testing. The Theil–Sen median trend analysis method is a non-parametric slope estimation method [29,30], which has the advantage of being less susceptible to external interference and having strong noise resistance. The specific calculation steps are as follows:

$$\beta =Mean\left({\displaystyle \frac{x_i-x_j}{i-j}}\right)$$

(1)

In Equation (1), $x_i$, $x_j$ represent time series data. When β > 0, it indicates an increasing trend in the time series; when β < 0, it indicates a decreasing trend. The significance of the time series trend is determined using the Z-value of the Mann–Kendall test statistic.

The steps for calculating the Z-statistic are as follows:

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

(2)

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

(3)

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

(4)

In Equations (2)–(4), Z represents the standardized test statistic, S represents the statistical test quantity, $n$ represents the length of the time series, $x_i$, $x_j$ denote the values for the respective years, and sgn denotes the sign function. The calculation formula for the sgn function is as follows:

$$\mathrm{sgn}\left(x_j-x_i\right)=\left\{\begin{array}{ll}1,\hfill & x_j-x_i>0\hfill \\ 0,\hfill & x_j-x_i=0\hfill \\ -1,\hfill & x_j-x_i<0\hfill \end{array}\right.$$

(5)

(2001 ≤ i ≤ j ≤ 2020)

In this study, vegetation change trends were categorized into five types based on the slope $\beta$ derived from the Theil–Sen median trend analysis and the Z-value obtained from the Mann–Kendall trend test: (1) $\beta$ > 0 and Z < 1.96, indicating an insignificant improvement; (2) $\beta$ > 0 and Z > 1.96, indicating a significant improvement; (3) $\beta$ = 0 and Z = 0, indicating no change; (4) $\beta$ < 0 and Z < 1.96, indicating an insignificant degradation; and (5) Slope < 0 and Z > 1.96, indicating significant degradation.

2.3.3. Pixel-Based Hurst Index Analysis

The Hurst exponent is an efficient method used to predict whether a time series has long-term dependence. It was proposed by a renowned British hydrologist and is based on the R/S rescaled range analysis method; it has since been applied in various research fields [31]. Taking the vegetation index EVI as an example, the Hurst exponent for EVIi, where i = 1, 2, 3, 4, 5, ..., n, and m is any positive integer, can be defined. Therefore, the following can be concluded:

Differential sequence:

$$\Delta {\mathrm{EVI}}_i={\mathrm{EVI}}_i-{\mathrm{EVI}}_{i-1}$$

(6)

Mean sequence:

$$\overline{\Delta \mathrm{EVI}(m)}={\displaystyle \frac{1}{m}}{\displaystyle \sum _{i=1}^m\Delta }{\mathrm{EVI}}_i(m=1,2,\cdots ,n)$$

(7)

Cumulative deviation:

$$X(t)={\displaystyle \sum _{i=1}^m\left(\Delta {\mathrm{EVI}}_i-\overline{\Delta \mathrm{EVI}(m)}\right)}(1⩽t⩽m)$$

(8)

Range and standard deviation:

$$R(m)={\max }_{1⩽m⩽n}X(t)-{\min }_{1⩽m⩽n}X(t)(m=1,2,\cdots ,n)$$

(9)

$$S(m)=\&{\left[{\displaystyle \frac{1}{m}}{\displaystyle \sum _{i=1}^m{\left(\Delta {\mathrm{EVI}}_i-\overline{\Delta \mathrm{EVI}(m)}\right)}^2}\right]}^{{\displaystyle \frac{1}{2}}}(m=1,2,\cdots ,n)$$

(10)

The final calculation result of the Hurst exponent is

$$\raisebox{1ex}{${{\displaystyle R}}_{(m)}$}\!\left/ \!\raisebox{-1ex}{${{\displaystyle S}}_{(\mathrm{m})}$}\right.≌{\displaystyle \raisebox{1ex}{$R$}\!\left/ \!\raisebox{-1ex}{$S$}\right.}$$

(11)

If R/S ∝ m^H, it indicates the presence of the Hurst changing trend phenomenon in the observed time series. Based on the magnitude of the H value, it can be classified into the following categories: (1) When 0 < H < 0.5, it suggests that the current time series exhibits long-term dependence, but the future trend is opposite to the present, indicating anti-persistence. A smaller H value corresponds to a stronger anti-persistence. (2) When H = 0.5, it implies that past values have no influence on future development, indicating random switching. (3) When 0.5 < H < 1, it indicates that the current time series exhibits long-term persistence, and the future trend is roughly similar to the present, indicating persistence. A larger H value corresponds to a stronger persistence.

2.3.4. Correlation Analysis Based on Pixel

In this study, the Pearson correlation analysis method was used to calculate the correlation coefficients between different vegetation indices EVI and LAI and meteorological drought at different time scales under different slopes. The formula is as follows:

$$r={\displaystyle \frac{{\displaystyle \sum _{i=1}^n\left(X_i-\overline{X}\right)}\left(Y_i-\overline{Y}\right)}{\sqrt{{\displaystyle \sum _{i=1}^n{\left(X_i-\overline{X}\right)}^2}}\sqrt{{\displaystyle \sum _{i=1}^n{\left(Y_i-\overline{Y}\right)}^2}}}}$$

(12)

In Equation (12), X and Y represent the two variables; $\overline{X}$ and $\overline{Y}$ represent the mean of the two variables, respectively. The r is the correlation coefficient between the two variables, indicating the degree of linear correlation between the two variables. In this study, the F-test was used to test the significance of the correlation coefficients. p < 0.01 denotes a highly significant correlation, 0.01 < p ≤ 0.05 denotes a significant correlation, and p > 0.05 denotes a non-significant correlation.

2.3.5. Spatial Statistical Analysis

This study is based on DEM raster data, using ArcGIS software to extract slope information for Guizhou Province. The slope classification of Guizhou Province was determined according to the slope classification standards proposed by Zhang Shanshan et al. [32], dividing the province’s slopes into six categories: 1 (<5°), 2 (5–8°), 3 (8–15°), 4 (15–25°), 5 (25–35°), and 6 (>35°). Using ArcGIS, we conducted an overlay analysis of the slope classification raster and vegetation change trend raster base maps. We then performed zonal statistics to calculate the proportion of different vegetation index change trend raster cells within each slope category relative to the total number of raster cells in the study area. This allowed us to create a table (

Table 1. Proportional changes in EVI and LAI for different vegetation types under varying slope conditions.

Vegetation Type	Vegetation Index	Slope Gradient Classification
		1 (<5°)	2 (5–8°)	3 (8–15°)	4 (15–25°)	5 (25–35°)	6 (>30°)
grass	EVI LAI	Slight improvement (5.48%) Significant improvement (5.28%)	Slight improvement (3.41%) Significant improvement (3.53%)	Slight improvement (4.05%) Significant improvement (4.39%)	Slight improvement (1.12%) Significant improvement (1.28%)	Slight improvement (0.10%) Significant improvement (0.11%)	— —
evergreen broad-leaf forest	EVI LAI	Slight improvement (4.18%) Significant improvement (4.22%)	Slight improvement (2.73%) Significant improvement (2.85%)	Slight improvement (3.29%) Significant improvement (3.56%)	Slight improvement (1.14%) Significant improvement (1.22%)	Slight improvement (0.12%) Significant improvement (0.12%)	Slight improvement (0.01%) Significant improvement (0.01%)
evergreen needle-leaved forest	EVI LAI	Slight improvement (6.58%) Significant improvement (5.92%)	Slight improvement (4.06%) Significant improvement (3.81%)	Slight improvement (4.46%) Significant improvement (4.37%)	Slight improvement (0.95%) Significant improvement (0.95%)	Slight improvement (0.03%) Significant improvement (0.03%)	— —
broad-leaved deciduous forest	EVI LAI	Slight improvement (3.48%) Significant improvement (3.31%)	Slight improvement (2.43%) Significant improvement (2.43%)	Slight improvement (3.20%) Significant improvement (3.26%)	Slight improvement (1.02%) Significant improvement (1.08%)	Slight improvement (0.09%) Significant improvement (0.10%)	— —
arid lands	EVI LAI	Slight improvement (7.49%) Significant improvement (6.97%)	Slight improvement (3.41%) Significant improvement (3.61%)	Slight improvement (3.64%) Significant improvement (4.05%)	Slight improvement (0.80%) Significant improvement (0.93%)	Slight improvement (0.04%) Significant improvement (0.05%)	— —
paddy-field	EVI LAI	Slight improvement (2.43%) Significant improvement (1.99%)	Slight improvement (0.80%) Significant improvement (0.76%)	Slight improvement (0.59%) Significant improvement (0.61%)	Slight improvement (0.10%) Significant improvement (0.10%)	Slight improvement (0.01%) Significant improvement (0.01%)	— —

) that shows the different vegetation change trends across various slope categories. Although multiple vegetation change trends were observed across different slope classes, the number of raster cells for each trend varied. To ensure the significance of the findings, only the trends with relatively large cell counts were included in Table 1.

3. Results

3.1. Analysis of Spatiotemporal Variations in Vegetation during the Growing Season

3.1.1. Temporal Variation Characteristics of EVI and LAI

Figure 2 displays the inter-annual variations of EVI and LAI during the growing season in Guizhou Province from 2001 to 2020. Both EVI and LAI exhibited significant increases over this period, with linear growth rates of 0.3% per year (p < 0.01) and 1.5% per year (p < 0.05), respectively. This indicates an overall upward trend in vegetation growth. Both indices displayed a fluctuating pattern with alternating increases and decreases. The annual maximum and minimum values of EVI and LAI occurred at slightly different times. EVI reached its maximum of 0.41 in 2016 and minimum of 0.34 in 2001, while LAI peaked at 1.99 in 2019 and dipped to 1.51 in 2010.

3.1.2. Spatial Variation Characteristics of EVI and LAI

To better explore the spatial dynamics of vegetation in Guizhou Province, the spatial trends and significance levels of EVI and LAI were analyzed using the Theil–Sen median slope analysis and the Mann–Kendall trend test on an image-by-image meta-check. As shown in Figure 3, regions with positive slopes in the annual mean inter-annual variation in EVI account for 85.24% of the total area, while regions with positive slopes in LAI account for 82.02% of the total area. This further indicates an overall improvement in vegetation within the study area, primarily distributed throughout Guizhou Province, excluding Guiyang City and the Qiandongnan region centered around Kaili. The area with slight improvement in EVI accounts for the largest proportion, approximately 70.84%, while the significantly improved area accounts for 8.41%. The LAI shows a higher proportion of significantly improved area, accounting for about 70.39%, with slight improvement accounting for only 10.67%. However, the regions of improvement for both vegetation indices are largely similar, mainly distributed in the northern, western, southwestern, and southern parts of Guizhou Province. The most significant vegetation improvement is observed in western Bijie and eastern Qianxinan Prefecture, including Zhenfeng, Anlong, and Wangmo. The degraded areas of EVI and LAI are mainly concentrated in the central and eastern parts of Guizhou Province, accounting for approximately 10.20% and 16.62% of the total study area, respectively. Slight EVI degradation covers about 9.56%, while significant LAI degradation covers about 10.14%. Despite differences in degradation levels, both indices show that the primary degraded areas are in most of Qiandongnan and the border region with Qiannan.

3.2. Analysis of the Trend of Vegetation Types under Different Slopes

3.2.1. Analysis of EVI and LAI Variations in Different Vegetation Types

Guizhou Province is a typical karst landform region. Based on the characteristics of vegetation growth in the area, EVI and LAI of different vegetation types were extracted to analyze the changing trends from 2001 to 2020. According to Figure 4, in terms of EVI, Arid lands, evergreen broad-leaved forests, grasslands, and deciduous broad-leaved forests exhibited identical trends, each showing a highly significant increase at a rate of 0.03/10a (p < 0.01). In contrast, both evergreen coniferous forests and paddy fields demonstrated a significant increase, albeit at a slower rate of 0.02/10a (p < 0.05). Compared to EVI, grasslands had the fastest LAI (Figure 5) growth rate at 0.17/10a (p < 0.01), followed by evergreen broad-leaved forests, growing significantly at a rate of 0.17/10a (p < 0.05). Arid lands and evergreen coniferous forests had slightly lower LAI growth rates. paddy field had the smallest LAI growth rate at 0.1/10a, indicating that EVI and LAI have slightly different monitoring effects on different vegetation types.

To further investigate the changes in different vegetation types across various slope gradients, this study employed spatial statistical analysis using ArcGIS to examine the variations in EVI and LAI for different vegetation types under different slope conditions. The results are presented in Table 1. The results (Table 1) show that under different slope gradient levels, the EVI and LAI of six different vegetation types mainly exhibited slight improvement or significant improvement. The proportion of area with changes in EVI and LAI for each vegetation type, from highest to lowest, were 1 (<5°), 3 (8–15°), 2 (5–8°), 4 (15–25°), 5 (25–35°), and 6 (>35°). When the slope gradient was less than 5°, the sum of the proportions of area with changes in EVI and LAI for all vegetation types was 57.33%. Among them, the proportion of area with slight improvement in dry-land EVI and evergreen coniferous forest EVI accounted for 7.49% and 6.58% of the total area with changes, respectively. The proportion of area with significant improvement in dry-land LAI and evergreen coniferous forest LAI accounted for 6.97% and 5.92% of the total area with changes, respectively. Grassland followed, with EVI and LAI changes accounting for 5.48% and 5.28% of the total area with changes, respectively. When the slope gradient was between 5° and 8°, evergreen coniferous forest had the highest proportion of area with changes in EVI and LAI, accounting for 4.06% and 3.81%, respectively. When the slope gradient was between 8° and 15°, evergreen coniferous forest had the largest proportion of area with changes, with EVI and LAI changes accounting for 4.46% and 4.37% of all vegetation changes, respectively. When the slope gradient exceeded 35°, except for evergreen broad-leaf forest, other vegetation types did not show significant trends of change.

3.2.2. Analysis of Future Trends in Different Vegetation Types

To further analyze and study the overall and future trends of EVI and LAI within the study area, this paper conducted a Hurst index analysis based on rescaled range analysis for EVI and LAI raster images. The results were combined with the Sen trend analysis of the two vegetation indices for comprehensive analysis. The calculated results showed that the Hurst range values for EVI ranged from 0.055 to 0.997, while for LAI, the Hurst range values ranged from 0.081 to 0.928. From Figure 6, the spatial distribution of future vegetation changes in Guizhou Province can be observed. When the Hurst index range is 0 < H < 0.5 and the trend shows improvement, degradation, or remains relatively stable, the vegetation cover changes can be classified into five categories: anti-persistent slight improvement or degradation, anti-persistent significant improvement or degradation, and anti-persistent stability. On the other hand, when the Hurst index range is 0.5 < H < 1 and the trend shows improvement, degradation, or remains relatively stable, the vegetation cover changes can be classified into five categories: persistent slight improvement or degradation, persistent significant improvement or degradation, and persistent stability.

Continuous improvement or degradation refers to the future development trend remaining consistent with the past, indicating that the vegetation in Guizhou Province has been predominantly increasing or decreasing in the past and will continue to do so in the future. The areas with continuous improvement in EVI and LAI account for 15.27% and 11.94%, respectively, with a relatively scattered spatial distribution, mainly in the central and western regions. The areas with continuous degradation only account for 2.43% and 3.20% of the total region, indicating that the vegetation cover in Guizhou Province will continue to improve in the future.

On the other hand, the concept of anti-continuous improvement or degradation refers to a future trend that is opposite to the past, lacking consistency. For example, if the vegetation cover in Guizhou Province has predominantly increased in the past, it might show a decreasing trend in the future, or vice versa. The areas with anti-continuous improvement in EVI and LAI are the largest among all types of change, accounting for 63.98% and 69.48%, respectively. Among them, EVI has an area with a proportion of approximately 57.73% showing slight anti-continuous improvement, while LAI has a significantly improved area with a proportion of about 60.32%. Although the degree of change varies, both are distributed throughout Guizhou Province, except for the southeastern region, indicating that the vegetation growth in Guizhou Province is gradually reaching an optimal state or may experience intermittent stagnation. The proportions of EVI and LAI showing anti-continuous degradation trends differ, with proportions of 7.79% and 13.50%, respectively. They also have slightly different spatial distributions, with EVI mainly concentrated in the southeastern region, specifically around Kaili and the border with Duyun in the south, while LAI is mainly distributed in the central and eastern regions of Guizhou Province, with a broader monitoring range.

In order to explore and monitor the future trends of different vegetation types in Guizhou Province and provide an objective basis for selecting suitable vegetation types in the study area, a statistical analysis was conducted on the future trends of various vegetation types. Figure 7 shows the future trends of EVI and LAI for different vegetation types. The results indicate that the proportion of a slight anti-continuous increasing trend is the highest among all vegetation EVI trends, ranging from 53% to 60%. Among them, evergreen broad-leaved forests have the highest proportion, approximately 59.79%. The proportion of areas showing significant anti-continuous degradation trends is the lowest, ranging from 0.07% to 0.45%. The proportion of regions with a slight continuous increasing trend ranges from 11% to 16%, with dry-land being the most significant at 15.77%, followed by paddy fields at 15.14%. For LAI, approximately 50% to 66% of the total area exhibits a significant anti-continuous increasing trend, with evergreen broad-leaved forests being the most prominent at 65.87%. Shrub-lands and dry-lands, which predominantly show a significant continuous increasing trend, account for 11.60% and 12.94% of the total area, respectively. Therefore, it is important to pay close attention to the growth conditions of evergreen broad-leaved forests, dry-lands, and shrub-lands in the future and implement appropriate vegetation cultivation measures.

3.3. Analysis of Spatial and Temporal Variations in Meteorological Drought at Different Time Scales

Different time scales of the Standardized Precipitation Evapotranspiration Index (SPEI) can more comprehensively reflect the meteorological drought variations in the study area. The Sen’s slope analysis of different time scales of SPEI (Figure 8a–e) reveals that the mean drought index for the past 20 years in Guizhou Province shows a significant increasing trend overall. As the time scale increases, the cycle of drought alternation lengthens and the frequency decreases, with the most significant period observed from 2009 to 2013. The rates of change in SPEI vary slightly among different time scales, with SPEI-3 and SPEI-9 showing the highest rates at 0.17/10 years, followed by SPEI-6 (0.16/10 years) and SPEI-12 (0.15/10 years). The lowest rate of change is observed in SPEI-1 (0.1/10 years), indicating that the seasonal scale experiences the fastest rate of change while the monthly scale changes the slowest. The lowest values of SPEI and their corresponding years also differ among the five time scales. The lowest value for SPEI-1 occurred in 2009 at −0.400, SPEI-3 reached its lowest in 2013 at −0.601, both SPEI-6 and SPEI-9 had their lowest values in 2011 at −0.889 and −0.880, respectively, and SPEI-12 reached its lowest value in 2010 at −0.804. This indicates that as the time scale increases, the lowest values of drought gradually increase, suggesting an intensification of drought conditions. However, overall, the trend of drought improvement remains predominant.

When the slope of SPEI change is negative, it indicates a trend towards drought in the study area. Based on the Theil–Sen median slope analysis method, the spatial variations in SPEI at different time scales in Guizhou Province from 2001 to 2020 are explored (Figure 9). Analyzing and calculating the negative value areas of SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12 reveals that the proportions of negative value areas relative to the total area of the study area are 15.44%, 18.93%, 21.99%, 24.35%, and 25.18%, respectively. This indicates that as the time scale increases, the area exhibiting a drought trend in Guizhou Province continues to expand. However, the proportions of positive value areas for the five different time scales of SPEI are 84.56%, 81.07%, 78.01%, 75.65%, and 74.82%, respectively, with the positive value areas far exceeding the negative value areas. This suggests that the climate change in Guizhou Province from 2001 to 2020 is mainly characterized by a trend of alternating between wet and dry conditions. In terms of spatial distribution, the differences in the trend and distribution of SPEI changes among different time scales are not significant. The areas exhibiting a drought trend are mainly located in the northern, western, and southwestern parts of Guizhou Province. Specifically, the regions with a more significant drought trend are mainly found in Zunyi, Tongren, their bordering areas, Liupanshui, Xingyi, their bordering areas, and the eastern part of Bijie. Conversely, the regions exhibiting a trend towards wetter conditions are primarily located in the southern and central–eastern parts of Guizhou Province.

3.4. Response Analysis of Vegetation to Different Time Scales of SPEI under Different Slopes

Correlation analysis between vegetation under different slope gradients in Guizhou Province and different time scales of Standardized Precipitation Evapotranspiration Index (SPEI) reveals the following (Figure 10): (1) When the slope gradient is level 1, the Enhanced Vegetation Index (EVI), Leaf Area Index (LAI), and different time scales of SPEI show a predominantly positive correlation, with no significant differences. The strongest positive correlation for EVI is found with the annual scale SPEI-12, at 30.64%, while LAI has the strongest positive correlation with the monthly scale SPEI-1, reaching 30.52%. The proportion of negative correlations between EVI, LAI, and different time scales of SPEI ranges from approximately 11.28% to 15.07%. (2) When the slope gradient is level 2, EVI and LAI still exhibit a positive correlation with SPEI at various time scales, but the proportion decreases significantly compared to level 1. The highest positive correlations for EVI and LAI are with SPEI-12, at 17.28% and 17.45%, respectively. The largest negative correlation for EVI is with SPEI-9, with a proportion of 8.50%, while LAI shows the highest negative correlation with SPEI-6, at 7.95%. (3) When the slope gradient is level 3, the correlation between the two vegetation indices and SPEI at different time scales is slightly higher than level 2 but lower than level 1. EVI and LAI both exhibit the strongest positive response to SPEI-12, at proportions of 19.98% and 20.53%, respectively. Regarding negative correlations, EVI shows the strongest correlation with SPEI-9, while LAI has the strongest correlation with SPEI-6, with proportions of 10.46% and 9.16%, respectively. (4) When the slope gradient is level 4 or 5, although EVI and LAI still show a higher positive correlation with SPEI at different time scales compared to negative correlation, the overall proportions are relatively small. The maximum positive correlation with SPEI-12 is around 5%, while the maximum negative correlation with different time scales of SPEI is only 0.23%. (5) When the slope gradient is level 6, the response of vegetation to meteorological drought at different time scales is minimal, with low correlation. The proportions can be considered negligible. The above analysis indicates that under different slope gradients, the positive correlation between vegetation and meteorological drought at different time scales is generally higher than the negative correlation, but the differences are not significant. Moreover, the response of vegetation to meteorological drought decreases gradually with increasing slope gradient.

4. Discussion

Vegetation growth dynamics result from the combined effects of various factors, including climate change, topography, and human activities. This study, based on MODIS EVI and LAI data from the growing seasons in Guizhou Province between 2001 and 2020, analyzes the future trends and development directions of EVI and LAI for different vegetation types under varying slopes. Additionally, the study explores the correlation between these two vegetation indices and the Standardized Precipitation Evapotranspiration Index (SPEI) at different time scales. The results indicate that, firstly, both EVI and LAI show a significant upward trend in inter-annual variation over the time series, suggesting an overall increase in vegetation cover across Guizhou Province. This finding aligns with the research results of Zhang Beibei [33], Dai Renli [34], and others. However, the differences in the magnitude of change between the two indices might be attributed to their distinct threshold ranges and sensitivities to vegetation. Secondly, regarding the spatial distribution of the trends in the two indices, the areas of vegetation improvement and degradation in Guizhou Province are relatively consistent. The primary areas of improvement are concentrated in the northern and western regions, while the main degradation zones are located in the southeastern part of the province. These findings are consistent with the research of Liu Yang [35] and others. Since 2000, the implementation of various ecological restoration and management projects across the province has led to significant improvements in the regional ecological environment [36,37], promoting vegetation growth and substantially increasing vegetation cover [38]. Notably, in the western regions of Guizhou, including Anshun, Bijie, Liupanshui, Qiannan, and Qianxinan Prefectures, where rocky desertification was previously widespread and severe, significant progress has been made in mitigating rocky desertification, resulting in substantial increases in vegetation cover. This corresponds with the vegetation improvement areas identified in this study [39,40]. When using EVI and LAI to represent the spatiotemporal changes in vegetation in Guizhou Province, both indices demonstrate considerable consistency in their spatiotemporal patterns, which aligns with existing research findings. However, there are notable differences in how each index reflects specific changes in vegetation. EVI results indicate that vegetation changes from 2000 to 2020 are primarily characterized by slight improvements, whereas LAI suggests more significant changes. Both indices have unique advantages in reflecting vegetation changes in areas with high vegetation cover. However, determining which index is more suitable for karst regions remains a crucial focus for our future research.

This study analyzes the variation characteristics of EVI and LAI across different vegetation types under varying slope conditions and finds that as slope increases, the proportion of areas with vegetation improvement decreases. Specifically, areas with slopes less than 5° show the highest proportion of vegetation improvement. Overall, across different slope categories and vegetation types, the proportion of vegetation improvement decreases with increasing slope, though the extent of change varies slightly among vegetation types. When the slope is less than 5°, dry-lands exhibit the largest improvement in EVI and LAI, followed by grasslands, with paddy fields showing the least improvement. Conversely, when the slope exceeds 35°, all vegetation types, except evergreen broad-leaf forests and grasslands, exhibit a decline in the area of vegetation improvement. Previous studies have shown [41,42,43] that in the karst region of Southwest China, topographic factors play a crucial role in vegetation dynamics. Elevation is a key determinant of mountain temperature and a primary factor causing vertical vegetation zonation [44]. Slope aspect influences vegetation growth by affecting light conditions [45], while slope affects vegetation by altering surface runoff direction and velocity, which in turn impacts soil formation conditions and water retention capacity [45]. In this study, vegetation improvement is most pronounced for dry-lands, evergreen needle-leaf forests, and evergreen broad-leaf forests in low- to mid-elevation areas. In low-altitude and low-slope regions, the combination of favorable soil parent material and optimal water and thermal conditions supports better vegetation growth. These areas are primarily occupied by evergreen broad-leaf forests, needle-leaf forests, and dry-lands. Prior to the implementation of various ecological restoration projects, rapid socio-economic growth, population increase, and escalating land-use conflicts had severely impacted vegetation in this region. However, post-2000s, human activities aimed at environmental restoration, such as reforestation and grassland rehabilitation, have effectively improved vegetation. As elevation and slope increase, natural factors such as water erosion and soil degradation intensify, reducing soil thickness and fertility. The positive impact of human activities diminishes, further limiting vegetation growth and leading to a decrease in the area of vegetation improvement [40]. Consequently, with the exception of vegetation types such as evergreen broad-leaf forests and grasslands, which have strong slope adaptability and water retention capabilities, most other vegetation types do not show significant improvement. By analyzing vegetation dynamics under different slope conditions, this study further confirms the critical role of slope in vegetation growth in karst regions and provides a theoretical basis for developing region-specific strategies for sustainable development in karst areas.

The projection of future vegetation trends in Guizhou Province suggests that certain areas may reach saturation, potentially leading to temporary stagnation in vegetation growth. This finding provides a scientific theoretical basis for the implementation of ecological management projects in the province; however, the specific causes of this phenomenon require further investigation. Over the past 20 years, SPEI (Standardized Precipitation Evapotranspiration Index) at various time scales indicates a declining trend in drought conditions in Guizhou, with an increasing trend towards more humid conditions, which is largely consistent with existing research findings [46,47]. The improvement in drought conditions has, to some extent, facilitated the enhancement of the vegetation growth environment. By analyzing the response of vegetation to meteorological drought under different slope conditions, it is evident that the overall positive correlation between EVI, LAI, and SPEI at various time scales in Guizhou Province is stronger than the negative correlation. Moreover, as slope increases, both positive and negative correlations gradually weaken, indicating that the influence of slope on the relationship between vegetation and drought response has a threshold range. Although Guizhou Province is located in a typical subtropical monsoon region with abundant rainfall, drought remains a significant factor affecting vegetation growth [48], closely related to the province’s unique geological and geomorphological characteristics. Guizhou is situated on the eastern slope of the Yunnan–Guizhou Plateau, with terrain that is higher in the west and lower in the east. Mountainous areas account for more than 90% of the province’s total land area, and karst landscapes are extensively developed, covering 73% of the province’s territory. The karst regions feature a unique dual structure of surface and underground systems, characterized by poor soil formation conditions, thin surface soil layers, low soil fertility, and widespread underground rivers and caves. This distinctive underlying surface combination leads to weaker surface runoff interception and water retention capacities compared to non-karst areas [49,50]. Consequently, vegetation growth conditions are inherently poor, and topographic factors such as slope further weaken regional vegetation growth by directly influencing the redistribution of surface material and energy, indirectly affecting the response relationship between vegetation and drought [51]. In the karst regions, the influence of topographic factors like slope results in variations in vegetation types and distribution, as well as differences in drought resistance [41]. In gentle slope areas, vegetation growth conditions are better than in steep slope areas, and the vegetation types mainly include rice paddies, evergreen broad-leaf forests, and needle-leaf forests, which are highly dependent on soil moisture and nutrients, making them more sensitive to drought. As the slope increases, the effects of water erosion intensify, surface runoff interception capacity decreases, and soil moisture and nutrient content decline, leading to an increase in ecological problems such as rocky desertification and soil erosion. This restricts regional vegetation growth and enhances the drought resistance of vegetation, resulting in lower drought sensitivity in steep slope areas compared to gentle slope areas [52]. This study focused on the response of vegetation to meteorological drought under different slope conditions but did not consider the influence of other factors, such as surface incision depth, slope aspect, and human activities. Additionally, the study only analyzed the overall response of vegetation to meteorological drought at different time scales under varying slopes, without considering the response of different vegetation types to drought. Therefore, future research on the correlation between vegetation and meteorological drought should pay more attention to the impact of these comprehensive factors on their relationship.

5. Conclusions

(1)

From 2001 to 2020, both the EVI and LAI of vegetation during the growing season in Guizhou Province showed significant upward trends, with linear growth rates of 0.3% per year (p < 0.01) and 1.5% per year (p < 0.05), respectively. This indicates an overall improvement in vegetation growth across the study area, with the primary regions of improvement located in western Guizhou and the main areas of degradation concentrated in the central and eastern parts of the province.

(2)

Hurst index analysis reveals that the areas of sustained improvement in EVI and LAI account for 15.27% and 11.94% of the total area, respectively, while areas of sustained degradation account for only 2.43% and 3.20%. This suggests that the future trend in vegetation cover across Guizhou Province is likely to continue improving.

(3)

Under different slope categories, EVI for various vegetation types primarily exhibits slight improvement, while LAI shows significant improvement. As slope increases, the proportion of vegetation improvement area decreases, with the largest improvement observed in regions with slopes less than 5°.

(4)

The mean values of SPEI across different time scales show a gradual increasing trend, indicating a significant shift towards more humid conditions in Guizhou Province. The humidification trend is most pronounced in the southern and central–eastern regions, while areas of increasing aridity are mainly located in the northern and southwestern parts of the province. Across different slope categories, the positive correlations between vegetation indices (EVI and LAI) and SPEI are stronger than the negative correlations, and these correlations weaken as slope increases. This suggests that slope exerts a threshold effect on the relationship between vegetation and drought response. When slopes exceed 35°, the response of vegetation to drought becomes almost independent of slope.