Introduction

According to the IPCC (Intergovernmental Panel on Climate Change) report 《Climate Change 2021: Public Summary》, the average temperature of the earth’s surface between 2011 and 2020 has increased 1.1℃, compared to the late 19th century, it was higher than at any time in the past 125,000 years1. The IPCC also has released that: over the past hundred years, human excessive reclaiming wasteland and large-scale activities, resulting in the emissions of greenhouse gases have increased, which was the most important cause of global warming2. Human activities not only lead to global warming, but also induce ecological security problems such as extreme weather, vegetation degradation, soil erosion and so on. Ecological environment is the basic condition of people’s survival and the guarantee of human sustainable development. Vegetation as the main component of ecosystem, has an important indication of regional ecosystem environmental change and plays a very important role in global carbon cycle and the climate regulation.

Fractional Vegetation Cover(FVC) refers to the percentage of the vertical projected area of vegetation on the ground in the total area of the statistical area, which is an important indicator to reflect the growth trend of vegetation community on the surface and an important basic data to describe the ecosystem3,4,5,6,7,8. The five countries of the LMR are located in Southeast Asia, it is an important link between East Asia and South Asia, the Indian Ocean and the Pacific Ocean.The ecological environment of this region is of great significance to maintaining regional stability and promoting economic development. However, little attention has been paid to this region, and few studies have been conducted on the vegetation cover in this area. The analysis of temporal and spatial variation trend of vegetation cover and its influencing factors can provide important guidance for ecological environment protection, social and economic development.

There are many methods to estimate FVC, such as the ground survey, the empirical model, the vegetation index, the decision tree classification method and so on. Some methods have disadvantages, such as strong dependence on measured data, complicated calculation and large amount of sample data. It is especially difficult to achieve in remote areas. The rapid development of remote sensing technology and the arrival of the era of remote sensing big data makes it possible to accurately monitor the vegetation in remote areas on a large spatial and temporal scale9. Among them, MODIS NDVI products have been used by a large number of studies for the calculation of vegetation coverage and the quantitative analysis of its dynamic changes, and the dimidiate pixel model is a commonly used FVC estimation method, which is relatively simple to operate and can be directly calculated using NDVI data10,11,12,13,14.

Current studies on vegetation coverage variation trend analysis method include linear regression analysis15,16, correlation analysis17, trend analysis method combining Sen’s slope estimator and Mann-Kendall significance test18. The spatial and temporal distribution and change of vegetation cover are affected by many factors, which makes its trend analysis complicated. The method of trend analysis method combining Sen’s slope estimator and Mann-Kendall significance test is not affected by data distribution and outliers, and has a good application effect on the quantitative description of trends in long time series. Some studies have applied it to climate change, vegetation variation, environmental monitoring and so on19,20. For the LMR, which have large northern and southern spans, the characteristics and causes of vegetation variation in different regions are also different, so it is necessary to use this method to more accurately explore the characteristics of vegetation variation and its future development trend.

Human factors and natural factors influence vegetation variation. In the past, most studies used partial correlation analysis and regression analysis to study the influencing factors of vegetation coverage variation21, these methods assume a linear relationship between vegetation and various environmental factors. However, the factors affecting the growth of vegetation in the environment are complex, and there may not be a strict linear relationship. Geographical detector (for short Geodetector), not only can be used to detect spatial differentiation and identify its main influencing factors, but also to detect the interaction between different factors22,23,24. This method has no linear assumption, can avoid the multicollinearity problem, and the result is more accurate in calculating the contribution rate of influencing factors. The improved OPGD (Optimal Parameters-Based Geographical Detector) model is added with a parameter optimization module, which can discretize multiple ways and multiple breakpoints according to the characteristics of independent variables, select the optimal discretization way and the optimal breakpoints number, and improve the significant difference between various categories25,26. Geographical detector is used to detect the influencing factors of vegetation of LMR.

Zhong carried out research on the spatial-temporal trends and driving forces of vegetation coverage in the Nujiang-Salween River basin based on BFAST and Hurst1. Han studied the temporal and spatial patterns of fluctuations of vegetation in the Great Mekong Subregion, further explored the relationship between climatic factors and anthropogenic factors27. Zhao analysed the state of forest health in the Lancang-Mekong Region in 2020 by forest cover index、leaf area index、FVC and gross primary productivity28. Potapov took the LMR as the study area and proposed a semi-automatic algorithm to monitor the canopy cover and height of woody vegetation on a regional scale using freely available Landsat time series data29. At present, relevant studies on vegetation in the Mekong River basin are either about the Nujiang-Salween River basin or the Great Mekong Subregion1,27. Vegetation studies carried out in the LMR are mainly about forest ecosystem variations or about woody vegetation structure28,29, there are few studies on vegetation coverage in the LMR. In the study, vegetation refers to tree cover、shrubland、grassland、cropland、bare or sparse vegetation、herbaceous wetland、mangroves、moss and lichen, rather than just tree cover.

Based on the above analysis, it can be seen that although many studies on the spatial and temporal variation trend analysis and influencing factors of vegetation have been carried out, there are few related studies on vegetation coverage in the five countries of LMR. Therefore, this study is based on MODIS NDVI data from 2000 to 2022, combined with meteorological, topographic and humanistic data, through Sen’s slope estimator, Mann-Kendall trend test and geographic detector to explore the spatial and temporal variation characteristics and trends of vegetation cover in LMR in recent 23 years, analyzed the spatial differentiation characteristics of each factor of vegetation coverage, to provide a scientific basis for the sustainable development of the study area and guarantee the ecological environment quality.

Study area and data resources

Study area

The Lancang-Mekong River is the sixth longest river in the world and the third longest river in Asia. It is an important trans-boundary international river in Asia. It originates from Zatqu at an altitude of 5,000 m in the northern foot of the Tanggula Mountains in the central part of the Qinghai-Tibet Plateau of China. The basin flows from north to the south through six countries, including China, Laos, Myanmar, Thailand, Cambodia and Vietnam, and empties into the South China Sea at Ho Chi Minh City, Vietnam. It is located in the center of the Asian tropical monsoon region, and is affected by three climate types: plateau mountain climate, subtropical monsoon climate and tropical monsoon climate, vegetation is affected by altitude, water and heat conditions, and its distribution shows latitudinal and vertical zonality30.The basin is called the Lancang River within China and the Mekong River outside China31.The Lancang-Mekong River is generally divided into two sub-basins, the upper basin is called China’s Lancang River basin, and the lower basin is composed of five countries: Myanmar, Laos, Thailand, Cambodia and Vietnam32.The study area of this study is five countries in the LMR, including Myanmar, Laos, Thailand, Cambodia, and Vietnam (Fig. 1).The region is geographically important, with a greater concentration of human activity than upper basin, and the forest coverage is high, but it is facing the threat of deforestation28, it is urgent to carry out study on temporal and spatial variation characteristics of vegetation cover in the LMR and the influencing factors to provide a scientific basis for vegetation protection and ecological environment restoration.

Fig. 1
figure 1

The administrative map of study area.

Full size image

Data sources and preprocessing

The data used in this study mainly include MODIS NDVI, DEM (Digital Elevation Model), annual mean temperature, annual precipitation, climate types, population density, GDP, and LULC data. The data of administrative boundary was obtained from the Data Platform of Resources and Environmental Science (https://www.resdc.cn/). All raster data were resamped to a uniform spatial resolution (250 m) using the nearest neighbor interpolation function (Table 1; Fig. 2). Data is described as follows.

MODIS NDVI

Although there are remote sensing images with higher resolution that can be used for NDVI data acquisition, considering the range of study area, long time series data acquisition and data processing efficiency, MODIS NDVI data is selected for FVC calculation. NDVI is the normalized vegetation index, which is the most commonly used index to characterize the growth characteristics of vegetation at present, and can reflect the vegetation coverage status to a certain extent. MODIS is the Moderate-Resolution Imaging Spectroradiometer, aboard the Terra and Aqua satellites. It has a band range of 0.4–14 μm, a total of 36 bands, and a spatial resolution of 250 m, 500 m, and 1000 m. Its product range covers four topics: atmosphere, land, ice and snow, and ocean, with a total of 44 data types. This study uses MODIS tertiary product data, named MOD13Q1, which is the land vegetation product of MODIS. The MOD13Q1 data can be downloaded free from the USGS (United States Geological Survey ) website (http://glovis.usgs.gov/), as well as freely available on the GEE (Google Earth Engine) platform.

MODIS NDVI data used in this study were invoked, preprocessed and downloaded through GEE remote sensing big data cloud platform, its temporal resolution is 16d and spatial resolution is 250 m, it is preprocessed by mosaic, projection, transformation and clipping in GEE platform(Table 1).In order to eliminate the influence of cloud, atmospheric effect, scanning angle and solar zenith angle on image quality, Maximum Value Composite (MVC) was used to synthesize NDVI data.

Table 1 Source and description of research data.
Full size table

Influencing factors

Natural factors mainly affect the growth and distribution of vegetation through natural processes such as climate change, soil characteristics, topography and natural disasters. Human factors have a significant impact on vegetation coverage through land use change, agricultural management activities and policy planning. The variation of vegetation coverage is influenced by natural factors and human activities. Although human activities can change the status of surface vegetation in a short period of time, the growth of vegetation is mainly restricted by environmental conditions in a longer period of time. Consider the actual situation and data availability, in this study, natural factors such as elevation, slope, aspect, annual mean temperature, annual precipitation and climate type, human factors such as GDP, population density and LULC were selected as the influencing factors of vegetation coverage9,17,33,34.

Topographic data

DEM used in the study comes from the Shuttle Radar Topography Mission(SRTM)Version 4 of the Consultative Group on International Agricultural Research Consortium for Spatial Information(CGIAR-CSI). The data download site is http://srtm.csi.cgiar.org/. The spatial resolution is 90 m and the format is Geo-Tiff. Through the GEE platform for mosaic, clipping, download and extracting such as elevation, slope, aspect as factors for vegetation coverage influencing analysis.

Climate data

The influence of meteorological factors on vegetation variation is beyond doubt, and the change of precipitation and temperature plays a very important role in the dynamic variation of vegetation, which are important factors determining the growth and development of vegetation. The meteorological data used in the study mainly include annual mean temperature, annual precipitation and climate type data. Among them, the temperature data were the fifth generation ECMWF global climate and atmospheric reanalysis data provided by ERA5 (The Fifth Generation ECMWF Atmospheric Reanalysis of the Global Climate), which were preprocessed and downloaded by GEE platform and selected as mean_2m_air_temperature (unit: K). Precipitation data from Climate Hazards Group InfraRed Precipitation with Station, download from https://chc.ucsb.edu/data/chirps, the unit is mm, the resolution is 0.05°.The temperature and precipitation data from 2000 to 2022 were collected and preprocessed in ArcGIS. Then, the average temperature synthesis method was used to synthesize the temperature data from January to December each year to obtain the annual mean temperature data. The precipitation data from January to December were accumulated to obtain the annual precipitation data of the study area. The climate type data is the Koppen Global Spatial Distribution dataset of Climate types with a spatial resolution of 0.1°, which is downloaded from the National Earth System Science Data Center (https://www.geodata.cn/) as one of the analysis factors for driving forces of vegetation coverage.

Human activity factor data

Human activity factor data mainly include GDP data and population density data. Population density data from Worldpop, the download website is https://hub.worldpop.org/geodata/listing?id=64, the unit is people/km2, the spatial resolution is 1 km. Global GDP data from Github, the download website is https://github.com/Gowosad/global_population_and_gdp, the resolution is 0.5°*0.5°.

LULC data

There is a close relationship between LULC and FVC, and the change of land use type directly affects the vegetation growth environment. The LULC data in this study is ESA WorldCover 10 m v100 from ESA(https://esa-worldcover.org/en) of 2020, which is also preprocessed and downloaded through GEE platform. The data divides land use types into 11 categories, including tree cover、shrubland、grassland、cropland、built-up、bare or sparse vegetation、snow and ice、permanent water bodies、herbaceous wetland、mangroves、moss and lichen. Different land cover types are described in detail, which is considered as one of the influencing factors of human activities in vegetation coverage.

Fig. 2
figure 2

Spatial distribution of FVC driver factors.

Full size image

Methodology

The dimidiate pixel model

The main methods to obtain FVC by remote sensing techniques include empirical model, vegetation index and pixel decomposition model. The dimidiate pixel model of the pixel decomposition method has the advantages of simple, reliable calculation, easy to obtain data parameters and high inversion accuracy. So the dimidiate pixel model is used to obtain the FVC of the study area. According to the research needs, in order to reduce the impact of non-vegetated areas, this study uses GEE platform and NDVI to calculate the FVC from 2000 to 2022 through the dimidiate pixel model. The formula is as follows:

$$\:\text{F}\text{V}\text{C}\:=\:\frac{\text{N}\text{D}\text{V}\text{I}-{\text{N}\text{D}\text{V}\text{I}}_{\text{s}\text{o}\text{i}\text{l}}}{{\text{N}\text{D}\text{V}\text{I}}_{\text{v}\text{e}\text{g}}-{\text{N}\text{D}\text{V}\text{I}}_{\text{s}\text{o}\text{i}\text{l}}}$$
(1)

Where NDVIsoil and NDVIveg are the NDVI values of pixels not covered by vegetation and completely covered by vegetation, respectively. Combined with the current vegetation coverage in the study area, NDVI values with cumulative frequency of 2% and 98% were selected as NDVIsoil and NDVIveg, respectively.

Sen’s slope and Manna-Kendall

Sen’s slope can calculate the median value of a long time series to determine the amplitude of trend increase or decrease in the long time series data. This method can eliminate the interference of outliers, and has a strong ability to avoid data errors. It is widely used in long-term trend analysis of vegetation, meteorology, hydrology, and ecology. Manna-Kendall significance test is a non-parametric statistical test used to determine the significance of the trend. It has a solid statistical theoretical basis for the significance test of the variation trend. Combined with Sen’s slope and Manna-Kendall significance test, it has become an important method to judge the trend of long time series data. The advantage of this method is that data does not need to follow a certain distribution and is less affected by data outliers. It has a strong ability to resist data errors, which improves the accuracy to a certain extent35,36. It is suitable for analyzing the spatial and temporal variation of NDVI in long time series15,37. Sen’s slope calculates the median slope of n(n-1)/2 data combinations, its formula is as follows:

$${S}_{FVC} = median \left(\frac{{FVC}_{j}-{FVC}_{i}}{j-i}\right) \quad (2000 \leq {i} < {j} \leq 2022)$$
(2)

Where SFVC is used to represent the variation trend of FVC, FVCi and FVCj is the FVC value in the i and j years respectively. SFVC>0 indicates that the FVC value shows an increasing trend during the study period. On the contrary, if SFVC<0, it indicates that the FVC value has a decreasing trend during the study period. Refer to relevant study results35,38, took − 0.0005 and 0.0005 as the boundary, divided the regions with SFVC between − 0.0005 and 0.0005 as stable and unchanged, and the regions with SFVC value greater than or equal to 0.0005 as increased regions, regions with SFVC values less than − 0.0005 are divided as decreased regions. The significance of the above trends will be further tested by Mann-Kendall, and its calculation formula is as follows:

Set {FVCi}, i = 2000,2001, …, 2022.

$$\:\text{Z}=\left\{\begin{array}{c}\frac{\text{S}-1}{\sqrt{\text{v}\text{a}\text{r}\left(\text{S}\right)}}(S>0)\\\:0\:\:\:\:\:\:\:(S=0)\\\:\frac{\text{S}+1}{\sqrt{\text{v}\text{a}\text{r}\left(\text{S}\right)}}(S<0)\end{array}\right.$$
(3)
$$\:\text{S}=\sum_{\text{i}=1}^{\text{n}-1}\sum_{\text{j}=\text{j}+1}^{\text{n}}\text{s}\text{g}\text{n}({\text{F}\text{V}\text{C}}_{\text{i}}-{\text{F}\text{V}\text{C}}_{\text{j}})$$
(4)
$$\:\text{s}\text{g}\text{n}({\text{F}\text{V}\text{C}}_{\text{i}}-{\text{F}\text{V}\text{C}}_{\text{j}})=\left\{\begin{array}{c}1\:\:({\text{F}\text{V}\text{C}}_{\text{i}}-{\text{F}\text{V}\text{C}}_{\text{j}})>0\\\:0\:\:({\text{F}\text{V}\text{C}}_{\text{i}}-{\text{F}\text{V}\text{C}}_{\text{j}})=0\\\:-1\:\:({\text{F}\text{V}\text{C}}_{\text{i}}-{\text{F}\text{V}\text{C}}_{\text{j}})<0\end{array}\right.$$
(5)
$$\:\text{v}\text{a}\text{r}\left(\text{S}\right)=\frac{\text{n}(\text{n}-1)(2\text{n}+5)}{18}$$
(6)

The meanings of FVCi and FVCj are the same as above, n is the length of time series; sgn is a symbolic function. The statistic Z, which has the range of (+∞, -∞), is used to test the significance of FVC. Given the significance level α, if\(\:\left|Z\right|>{Z}_{1-\raisebox{1ex}{$\alpha\:$}\!\left/\:\!\raisebox{-1ex}{$2$}\right.}\), indicates that the study sequence has a significant change at the α level. In this study, take α = 0.05, the corresponding value was 1.960, to determine the significance of FVC time series change trend at the 0.05 confidence level.With significant change defined as (|Z|≧1.96), insignificant change as (|Z|<1.96). The Sen_MK trend was obtained by coupling Sen’s slope estimator with Mann-Kendall significance test (Table 2).

Table 2 Sen_MK trend type division.
Full size table

Geographic detector

Geographic detector is a spatial hierarchical heterogeneity analysis method. The main idea is to divide the study space into subregions by variables and compare the spatial variances within each subregion and between different subregions in order to assess the importance of potential explanatory variables. It mainly includes factor detection, interaction detection, risk detection and ecological detection. Factor detection, as the core part of the geographic detector, is mainly used to quantify the relative importance of different geographic variables. In this study, factor detection and interactive detection were used to analyze the geographical factors affecting vegetation in the LMR. The factor detector is mainly used to analyze the influence of different driving factors on the spatial stratification heterogeneity of FVC. The degree of influence is measured by q value, and the formula is as follows:

$$\:\text{q}=1-\frac{\sum\:_{\text{h}=1}^{\text{L}}{\text{N}}_{\text{h}}{{\upsigma\:}}_{\text{h}}^{2}}{\text{N}{{\upsigma\:}}^{2}}$$
(7)

Where q is the explanatory power of geographical factor X; L is the classification of the dependent variable FVC or geographical factor X; h is the partition variable h = 1,…, L), Nh and \(\:{\sigma\:}_{h}^{2}\)are the sample number and variance of subregions h respectively; N and are the number and variance of the whole study area respectively. The value range of q is [0,1], the larger the value, the stronger the influence of X on the spatial differentiation of FVC.Interaction factors are used to identify whether the interaction between two geographic factors enhances or weakens the effect on the dependent variable FVC, or whether the effects of the two factors are independent of each other.

The interaction detector is used to determine the influence degree of the interaction of different drivers X1 and X2 on FVC. By comparing the q value of individual driver and the q value of interaction according to the evaluation interval in Table 3, the types of interaction between different drivers can be analyzed.

Table 3 Driver interaction type.
Full size table

In this study, natural factors such as elevation, slope, aspect, annual mean temperature, annual precipitation and climate type, human factors such as GDP, population density and LULC were selected as the influencing factors of vegetation coverage, the factors were X1-X9 respectively, the vegetation coverage FVC was Y. X and Y were used as the data input of the geographic detector, the spatial resolution was uniformly sampled at 250 m. Because of these factors were continuous variables, and the input data of the geographic detector was required to be classified data, according to the data discretization method proposed by Wang39, OPGD is used to optimize parameters in the R language environment. In order to make the results more realistic, the study also carried out artificial debugging of the discretization, and finally got a relatively ideal result. The specific discretization methods and categories are shown in Table 4. At the same time, Arcgis is used to create a fishing net, through the sampling tool of spatial analysis, the grid is input for all the reclassified X and Y. The sampling point is the fishing network in the study area, the values of independent variables and dependent variables are extracted to the point, the data is input as the geographic detector.

Table 4 Factor discretization methods and categories of geographic detectors.
Full size table

There are uncertainties in the influencing factor analysis of geographical detector, such as the uncertainty of data quality and the uncertainty of the discretization method. These uncertainties should be fully considered to improve the accuracy and reliability of analysis results.

Results

Temporal and spatial variation characteristics of FVC

LMR has rich forest resources and vast agricultural areas. The spatial distribution map of average FVC in the study area during 2000–2022 was obtained by using the spatial analysis visualization tool (Fig. 3a). According to the analysis, the proportions of low vegetation coverage (0-0.2), relatively low vegetation coverage (0.2–0.4), medium vegetation coverage (0.4–0.6), relatively high vegetation coverage (0.6–0.8) and high vegetation coverage (0.8-1) in the LMR are as follows: 3.37%, 5.28%, 16.31%, 30.94%, 44.10%, FVC distribution has significant spatial heterogeneity. The overall vegetation coverage is higher in the north and lower in the south. The average FVC value of the region is 0.70, and the areas with high vegetation coverage and relatively high vegetation coverage account for 75.04%, indicating that the vegetation coverage in the LMR is good. This is consistent with the study of Han on the interannual variation trend of vegetation in the Greater Mekong Subregion27.

Fig. 3
figure 3

Average FVC distribution and time variation of LMR.

Full size image

Combined with LULC in the study area, the low vegetation cover areas are mainly distributed in central Myanmar, central and eastern Thailand, the main land use types in these areas are cropland and grassland. The areas with high vegetation cover are mainly in northern and western Myanmar and northern Laos, and the land use type is mainly tree cover.It can be seen from the interannual variation of FVC in Fig. 3b that the vegetation cover in the study area showed an overall fluctuating upward trend during 2000–2022, with the maximum FVC appearing in 2020 at 0.720 and the minimum FVC appearing in 2002 at 0.691, an obvious low inflection point appearing in 2002 and 2006. Many studies have shown that extreme climatic events and the natural disasters they cause are the main drivers of vegetation cover change. These events have seriously threatened the growth of vegetation, which may be the reason for the low inflection point of FVC. Taking El Niño and La Niña as examples, the appearance of El Niño has the effect of heat waves on the environment and humans, putting plants under stress due to water shortage and reducing the length of the growing season of vegetation. The biggest problem with La Niña is that it affects society and the economy, especially agriculture34.

By looking up the major events that occurred in the study area during the study period, it found that the drought events that occurred in 2004–2005, as shown in the Annual Mekong Flood Report 2017, may have contributed to the decline in FVC in 2006. In 2011, the vegetation coverage decreased significantly, which may be due to the La Niña, which caused large areas of crops to suffer floods and reduce production, and the vegetation coverage decreased in LMR. Vegetation growth also showed an obvious downward trend in 2016, the reason may be that at the end of 2015, due to the influence of El Niño, all suffered different degrees of drought in LMR, the precipitation was lower than the average level of previous years, the summer temperature rose sharply, and there was a large area of high temperature, which accelerated the evaporation of soil water and the transpiration of vegetation. With the decrease of soil moisture and air moisture, the photosynthetic capacity of vegetation is weakened, which inhibits the growth of vegetation. In 2019, Thailand suffered the worst drought in nearly 10 years. These successive extreme events may be responsible for the relatively large downward trend of FVC in the study area in 2018–201940,41. But at the same time, we also found that after a short decline, the FVC in the study area was soon improved again, which should be due to the corresponding measures taken by human beings, such as the afforestation program. Other studies have shown that strengthening agricultural activities and management level can also improve the level of FVC27, while countries in the LMR are mainly based on agricultural production, and human farming activities have a significant impact, so the vegetation improvement is relatively obvious.

Variation trend of FVC

The combination of Sen’s slope and Manna-Kendall can quantitatively and qualitatively analyze the variation trend of vegetation coverage in the LMR during 2000–2022 (Fig. 4). Figure 4a is the result of Sen’s slope, Fig. 4b is the result of Manna-Kendall, Fig. 4c is the SEN_MK trend type, including five levels of FVC variation degree: significant increase, weak increase, no change, weak decrease, and significant decrease. The results show that from 2000 to 2022, 56.33% of the areas with improved vegetation coverage, 37.55% of the areas with degraded vegetation coverage. The degraded areas are mainly distributed in northern and eastern Cambodia, southern and western Laos.The improved area is much larger than the degraded area, the overall variation trend is good, indicating that the vegetation in the LMR area has been protected to a certain extent.

The results of significance test showed that 58.01% of the regions were significant. The area with significant increase of vegetation coverage accounted for 38.97% of the whole study area, which was the largest proportion of trend types, mainly distributed in the eastern region of Thailand.The area with weak increase of vegetation coverage accounted for 17.36% of the total area of the study area, the area with weak increase was evenly distributed in the LMR. The proportion of no changed area is 6.12%, which is the smallest trend type, and the distribution is relatively uniform.The area with weak decrease accounted for 21.90%, mainly distributed in Laos, Cambodia, Vietnam and Myanmar. The area of significant decrease accounted for 15.65%, mainly distributed in Cambodia. The evolution of vegetation coverage in LMR in the past 23 years is mainly increased.

Fig. 4
figure 4

Evolution trend of FVC in the LMR from 2000 to 2022.

Full size image

In addition, we found that the FVC distribution is also closely related to LULC42. According to the distribution of FVC and LULC (Fig. 5), it found that the high vegetation coverage area are mainly distributed in the tree cover, the medium vegetation coverage area are mainly cropland, and the low vegetation coverage area are mainly built-up, permanent water bodies and so on. The analysis of Sen_MK trend change types combined with LULC showed that vegetation in cropland was improved, which proved that human farming activities promoted greening. However, it is also found that in some urban areas, vegetation coverage shows a trend of degradation to varying degrees, indicating that human activities such as economic development and urban expansion may also cause negative interference to vegetation coverage43,44.

Fig. 5
figure 5

FVC and LULC distribution of LMR.

Full size image

Influencing factors of FVC

Factor detection and interactive detector (Tables 5 and 6) were used to obtain the detection results of the vegetation coverage driving force in the LMR. All the p value of influencing factors are 0, pass the significance test.The q value of LULC was the largest, which is 0.4729, indicating that land use type had the greatest influence on vegetation coverage. The main reason is that different land use types have different impacts on vegetation coverage, for example, tree cover usually has higher vegetation coverage. In addition, it may be related to infrastructure construction in the LMR. With the rapid economic development, a large number of infrastructure facilities such as dams and roads have been constructed, and these construction activities often require a large amount of land. As a result, the original vegetation has been cut down or destroyed, thus affecting the vegetation coverage. At the same time, in order to meet the growing demand for food and agricultural products in the Mekong Delta region, a large number of swamps have been converted into cropland, resulting in a decline in vegetation coverage.

Slope and elevation, which are 0.3416 and 0.2860 respectively, have the influence second only to LULC. Because slope and elevation, as important topographic factors, have a great impact on vegetation coverage by affecting soil stability, water conditions and climate conditions.Annual mean temperature, population density, GDP and annual precipitation followed, with q values of 0.2168, 0.1900, 0.1697 and 0.1496. Aspect and climate types have less influence, with q values of 0.0240 and 0.0135. Although there are some seasonal changes in the climate of the LMR, such as high temperature, high humidity and seasonal precipitation changes, the climate in the region is relatively stable on the whole, and the vegetation coverage can be maintained at a relatively stable level, except for extreme climate events that cause devastating blows to vegetation. And, the LMR has realized the importance of ecological environment protection and introduced a series of policies to limit the destruction of vegetation.These measures have alleviated the adverse effects of human activities on vegetation cover to a certain extent. Climate type originally had a relatively large influence on FVC, but most of the regions in the study area belonged to the same climate type as tropical monsoon climate, and only a small part of northern Myanmar and northeastern Vietnam had subtropical monsoon climate, so the influence of climate type on FVC was not reflected.

Table 5 The q values of LMR.
Full size table

The interactive detection results show that the interaction of any two factors will enhance its influence, and the interaction between each factor and LULC reaches the maximum influence value, indicating that LULC has a strong influence on vegetation coverage in LMR. In addition to human factors, the interaction of natural factors with elevation and slope also has a significant effect on vegetation coverage, it is consistent with the factor detection results, indicating that the FVC spatial differentiation of LMR is the result of the interaction between natural and human drivers.

Table 6 Influencing factor detection results of FVC.
Full size table

Discussion

Spatial and temporal characteristics

In the past 23 years, vegetation coverage in the LMR was generally good, with an average FVC of 0.70. In order to further analyze the spatial characteristics of vegetation coverage, discussions will be conducted according to different countries in the LMR (Figs. 6 and 7).

For Vietnam, the average FVC during 2000–2022 is 0.69, and the proportions of low vegetation coverage (0-0.2), relatively low vegetation coverage (0.2–0.4), medium vegetation coverage (0.4–0.6), relatively high vegetation coverage (0.6–0.8) and high vegetation coverage (0.8-1) are as follows: 4.91%, 4.32%, 13.34%, 43.47%, 33.96%, the proportion of areas with high vegetation coverage and relatively high vegetation coverage reached 77.43%, indicating that the vegetation coverage of Vietnam was good. According to the land use type and topography, the areas with high vegetation coverage are mainly distributed in mountainous and plateau areas in the north, tropical and monsoon rainforest areas in the south. It can be seen from the interannual variation of FVC in Fig. 6a, vegetation coverage in Vietnam showed an overall fluctuating upward trend during 2000–2022. The maximum value of FVC was 0.704 in 2015, and the minimum value was 0.674 in 2006. Vegetation growth showed an obvious upward trend during 2006–2015.

For Myanmar, the average FVC from 2000 to 2022 is 0.73, and the proportion of low, relatively low, medium, relatively high and high FVC is as follows: 4.02%, 5.45%, 9.70%, 24.69%, 56.14%, the proportion of areas with high vegetation coverage and relatively high vegetation coverage reached 80.83%, the north, west and east have higher vegetation coverage. It can be seen from Fig. 6b, vegetation coverage in Myanmar showed an overall fluctuating upward trend during 2000–2022. The maximum value of FVC was 0.747 in 2020, and the minimum value was 0.718 in 2002. Vegetation growth showed an obvious upward trend during 2012–2020.

For Thailand, the average FVC from 2000 to 2022 is 0.66, and the proportion of low, relatively low, medium, relatively high and high vegetation coverage are 2.34%, 5.90%, 29.54%, 31.61% and 30.61%, the area of high vegetation coverage and relatively high vegetation coverage reached 62.22%, the northern vegetation coverage was higher because of the higher altitude and dense forest in this region. The central region is a plain area with frequent human activities and relatively low vegetation coverage. It can be seen from Fig. 6c, vegetation coverage in Myanmar showed an overall fluctuating upward trend during 2000–2022, with the maximum FVC appearing in 2020 at 0.693 and the minimum FVC appearing in 2002 at 0.631, vegetation growth showed an obvious upward trend during 2006–2022.

For Laos, the average FVC from 2000 to 2022 is 0.79, the proportion of low, relatively low, medium, relatively high and high vegetation coverage are 0.69%, 1.82%, 6.66%, 24.71% and 66.12%, the proportion of areas with high vegetation coverage and relatively high vegetation coverage reached 90.3%. The vegetation coverage in Laos is high in the north and low in the south, according to the analysis of geographical environment, the terrain of Laos is high in the north and low in the south. The north is plateau mountain, the south are plain and hills. The rate of forest in the plateau and mountain areas is higher, so the vegetation coverage is relatively high. The southern region is more affected by human activities, the vegetation coverage is relatively low. It can be seen from Fig. 6d, the overall vegetation cover in Laos showed an overall fluctuating decline trend during 2000–2022. The maximum value of FVC was 0.799 in 2015, and the minimum value was 0.773 in 2018, which may be caused by deforestation and land reclamation caused by agriculture, forestry and mining in Laos. It has a significant effect on vegetation coverage.

For Cambodia, the average FVC from 2000 to 2022 is 0.64, the proportion of low, relatively low, medium, relatively high and high vegetation coverage are 3.75%, 9.14%, 21.92%, 38.30% and 26.89%, and the proportion of high and relatively high FVC areas reached 65.19%.Cambodia is high on all sides and low in the middle, with plateaus and mountains on the east, north and west sides, plains in the middle and south.Such topography makes the vegetation coverage relatively high in the east and west, relatively low in the central and southern regions where human activities are more frequent. It can be seen from Fig. 6e, the overall vegetation cover in Cambodia showed an overall fluctuating decline trend during 2000–2022. The maximum value of FVC was 0.660 in 2007, and the minimum value was 0.592 in 2022, which may be caused by the ineffective implementation of forest management, the protection policies, the deforestation and reclamation of forests in Cambodia.

Fig. 6
figure 6

Average FVC distribution and temporal variation in the five countries of LMR from 2000 to 2022.

Full size image
Fig. 7
figure 7

Average FVC distribution in the five countries of LMR from 2000 to 2022: a.Vietnam; b.Myanmar; c.Thailand; d.Laos; e.Cambodia.

Full size image

The average FVC of Laos and Myanmar is relatively high (0.79 and 0.73 respectively) (Fig. 7), the proportion of areas with high vegetation is also relatively high (66.12% and 56.14% respectively), while the average FVC of Cambodia and Thailand is relatively low (0.64 and 0.66 ), and the proportion of areas with high vegetation is also relatively low (26.89% and 30.61%), indicating that Laos and Myanmar increased the vegetation coverage level of the whole region, which was related to LULC, topography and geomorphology in the LMR, and will be discussed in the influencing factors analysis.

Variation trend analysis

In order to further analyze the variation trend of FVC in the LMR, discussions will be conducted according to different countries (Fig. 8).

For Vietnam, Fig. 8a shows that from 2000 to 2022, the proportion of five grades (significant decrease, weak decrease, no change, weak increase and significant increase) are 18.50%, 19.98%, 6.23%, 13.12% and 42.17%, respectively. The proportion of vegetation coverage increase area is 55.29%. The main area of significant increase was tree cover in northeast Vietnam. The proportion of vegetation decrease area was 38.48%, and the main significant decrease area were grassland, cropland and tree cover in southern Vietnam. In general, the increase area was larger than the decrease area, the area with significant increase accounted for 42.17% of the whole study area, which was the largest proportion of trend types, indicating that the overall variation trend of vegetation coverage in Vietnam was good.

For Myanmar, Fig. 8b shows that from 2000 to 2022, the proportion of the five grades were 11.25%, 22.68%, 8.08%, 20.07% and 37.92%, respectively. The area of vegetation coverage increase accounted for 57.99%, the area of vegetation decrease accounted for 33.93%. The increase area was larger than the decrease, the area with significant increase accounted for 37.92% of the whole study area, which was the largest proportion of trend types, indicating that the overall variation trend of vegetation coverage in Myanmar was good.

For Thailand, Fig. 8c shows that from 2000 to 2022, the proportion of the five grades were 12.49%, 18.04%, 4.20%,19.14% and 46.13%, respectively. The area of vegetation coverage increase accounted for 65.27%, the area of vegetation decrease accounted for 30.53%. The increase area was larger than the decrease, the area with significant increase accounted for 46.13% of the whole study area, which was the largest proportion of trend types, indicating that the overall variation trend of vegetation coverage in Thailand was good.

In Laos, Fig. 8d shows that from 2000 to 2022, the proportion of the five grades were 18.79%, 30.02%, 5.54%, 15.09% and 30.56%, respectively. The area of increase accounted for 45.65%, the area of decrease accounted for 48.81%. The decrease area is larger than the increase. Combined with the average FVC distribution in Laos during 2000–2022, the degradation of vegetation led to the decrease of average vegetation coverage in Laos.

In Cambodia, Fig. 8e shows that from 2000 to 2022, the proportion of the five grades were 32.31%, 22.96%, 4.53%,12.59% and 27.61%, respectively. The area of increase accounted for 40.20%, the area of decrease accounted for 55.27%. The decrease area is larger than the increase, which indicates that the overall variation trend of vegetation coverage in Cambodia is not developing well.

Fig. 8
figure 8

FVC variation trend in the five countries of LMR from 2000 to 2022: a.Vietnam; b.Myanmar; c.Thailand; d.Laos; e. Cambodia.

Full size image

Influencing factors analysis

The study selected the geographic detector to explore the factors affecting vegetation coverage in the LMR. The results showed that LULC had the greatest influence on FVC in the LMR. The second is the terrain factor and the climate factor. Population density, GDP and other factors have a low impact on vegetation in LMR.

There are some similarities and differences of influencing factors on vegetation coverage in different countries. The similarity is LULC has the greatest influence on vegetation coverage in the five countries, because different land use types determine the level and type of vegetation coverage. For example, tree cover and grassland have higher vegetation coverage, while built-up, snow and ice, and water tend to have lower vegetation coverage. The difference is that topography and climate factors are more important to vegetation coverage in some countries, while human factors have a greater impact on vegetation cover in other countries. For Vietnam, Thailand and Laos, the topographic factor is second only to LULC, because these countries have rich topographic changes from the north to the south, resulting in differences in the distribution of water, soil and climate resources, which influence the growth conditions and coverage of vegetation. For Myanmar and Cambodia, the precipitation factor is second only to LULC, because water is one of the basic requirements for vegetation growth. Cambodia has a humid climate and relatively abundant precipitation, the distribution and change of precipitation will also influence sudden water content and fertility, eventually influencing vegetation coverage. There are significant differences in the seasonal and spatial distribution of precipitation in Myanmar, which directly influence the growth environment of vegetation and the vegetation coverage.

In conclusion, there are similarities and differences in the influence of various environmental factors on vegetation coverage. Countries need to comprehensively consider the spatial distribution of FVC under different environmental conditions and the spatial correlation of each impact factor, identify environmental constraints in different regions, formulate ecological restoration measures according to local conditions, plan regional economic construction.For Cambodia and Laos, the vegetation coverage has a decrease trend, it is necessary to prevent the further decrease of vegetation coverage and strengthen ecological restoration, such as optimizing land use structure, developing modern ecological agriculture, changing farming types, implementing crop rotation, fallow management, returning farmland to forest, building nature reserves and implementing ecological migration, striving for national policy support. Pay attention to the common development of social economy and ecological environment.For Myanmar, Vietnam and Thailand, natural environmental factors have a greater impact on vegetation cover, it is important to ensure the ecological environment is not destroyed, maintain the stability of the ecosystem, develop an efficient, diversified and sustainable ecological economy. Vegetation dynamics is a complex process. It should be given full play to the positive role of human activities on ecological environment, found the balance between ecological protection and economic development, promote the coordinated and sustainable development of ecology and economy.

Conclusions

Based on MODIS NDVI data from 2000 to 2022, this study calculated FVC of five countries in the LMR, discussed the spatial and temporal characteristics, variation trends and influencing forces of vegetation coverage in recent 23 years. The main conclusions are as follows:

  1. (1)

    In recent 23 years, the overall development trend of vegetation coverage in the LMR was good, the vegetation coverage is higher in the north and lower in the south, the average FVC value is 0.70. Under the influence of weather conditions, topography and human activities, the distribution of vegetation coverage has obvious spatial heterogeneity. Vegetation coverage in Myanmar, Laos and Vietnam are significantly better than in Thailand and Cambodia. In recent years, the vegetation coverage in Vietnam, Myanmar and Thailand showed a fluctuating increasing trend, while in Laos and Cambodia showed a decreasing trend.

  2. (2)

    Sen_MK variation trend analysis showed that from 2000 to 2022, 56.33% of the area of vegetation coverage increase, 37.55% of the area decrease in the LMR. The increased area was larger than the decreased area. The overall variation trend was positive, indicating that the vegetation in the LMR had been protected to a certain extent. In Vietnam, Myanmar and Thailand, the vegetation coverage increased area is larger than the decreased area, the trend is good. In Laos and Cambodia, the decreased area is larger than the increased area, the overall vegetation coverage variation trend is not good.

  3. (3)

    The results of the geographic detector show that LULC has the greatest influence on vegetation coverage in both the whole region and each country of the LMR, which further indicates the close relationship between land use type and vegetation coverage in the study area. For Vietnam, Thailand and Laos, the influence of elevation and slope are second only to LULC. For Myanmar and Cambodia, the influence of precipitation is second only to LULC. Therefore, reasonable vegetation protection measures should be formulated according to the environmental constraints in different regions of the study area.

  4. (4)

    Human activities have two sides to vegetation development, it actively improves the ecological environment through a series of ecological projects, but also destroy the surface vegetation due to unreasonable activities. For LMR, high vegetation coverage and improvement areas, the existing resources should be effectively protected, the stability of the ecosystem should be maintained, and an efficient, diversified and sustainable ecological economy should be developed. For areas with degraded vegetation, it should give full play to the positive role of human activities in transforming the ecological environment and strengthen ecological construction. In addition, vegetation growth is a complex process, the prevention and control of extreme events should be strengthened to prevent further degradation of vegetation in the context of climate change. At the same time, five countries of LMR should strengthen cooperation, jointly formulate transnational vegetation protection and restoration plans, clarify the responsibilities and obligations of each country, and ensure policy coordination. In the future, more FVC calculation methods and trend analysis methods adapted to LMR will also be discussed, the influencing factors adapted to local conditions will be explored.