A Multi-Criteria GIS–AHP Framework for Wildfire Risk Assessment in Northern Algeria: Integrating Environmental and Anthropogenic Factors

Wildfires pose a significant threat to ecosystems, human life, and infrastructure, particularly in South America, where diverse climatic and environmental factors contribute to their occurrence. Climate change has exacerbated extreme weather conditions such as intense heat and drought, leading to a global increase in the frequency and intensity of wildfires. Countries like Brazil have experienced significant rises in wildfire damage, highlighting the urgent need for predictive models that accurately assess future wildfire risks to mitigate their impact effectively. This thesis addresses this need by developing a wildfire risk prediction system leveraging deep learning methods and remote sensing data.Using Earth Observation (EO) APIs, the system avoids downloading and storing vast amounts of satellite imagery, enabling efficient data acquisition and preprocessing. The study focuses on key variables that influence wildfire activity, including dynamic variables such as Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), radiation, Leaf Area Index (LAI), evapotranspiration (ET), wind speed, and temperature, as well as static variables like land cover, Digital Elevation Model (DEM), and population density. The system is designed to predict wildfire risk for the next day and up to eight days, offering a robust tool for proactive wildfire management.Given the stochastic and nonlinear nature of wildfire phenomena, this research employs advanced deep learning techniques, including Random Forests (RF), Long Short-Term Memory networks (LSTM), and Convolutional LSTM (ConvLSTM) models, to predict wildfire risk in near real-time. Active fire data from MODIS products, along with their burn dates, serve as the basis for training datasets. Non-fire points are generated by mapping the land cover distribution of fire points, ensuring balanced datasets for model training. Variables are extracted and classified into dynamic and static categories to capture both temporal variability and fixed geographical characteristics.The objectives of this research are threefold: (1) to investigate existing remote sensing-based wildfire management methodologies and identify enhancements through the integration of data cubes and deep learning; (2) to develop a scalable platform for efficient data acquisition, preprocessing, and risk prediction using deep learning algorithms; and (3) to evaluate the system’s accuracy, efficiency, and scalability with real-world datasets and disaster scenarios.Preliminary results highlight the effectiveness of integrating remote sensing data with deep learning models for wildfire risk prediction. Dynamic variables such as EVI, LST, and NDVI, along with human influence factors like Global Human Modification Index (gHM), emerged as key predictors, demonstrating the interplay of environmental and anthropogenic drivers in wildfire occurrences. Seasonal analysis from 2021 to 2024 revealed a strong correlation between fire activity, elevated temperatures, and declining vegetation indices from November to April. The Random Forest model achieved 83% accuracy, while the LSTM model showed promise with 75% accuracy, emphasizing the potential of both static and temporal data. These findings lay a robust foundation for enhancing wildfire risk management through advanced machine-learning approaches.

Forest fire is one of the most common natural hazards occurring in the Western Ghats region of Kerala and is one of the reasons for forest degradation. This natural disaster causes considerable damage to the biodiversity of this region during the dry fire season. The area selected for the present study, Eravikulam National Park, which is predominantly of grassland vegetation, is also prone to forest fires. This study aims to delineate the forest fire risk zones in Eravikulam National Park using remote sensing (RS) data and geographic information system (GIS) techniques. In the present study, methods such as Analytic Hierarchy Process (AHP) and Frequency Ratio (FR) were used to derive the weights, and the results were compared. We have used seven factors, i.e. land cover types, normalized difference vegetation index, normalized difference water index, slope angle, slope aspect, distance from the settlement, and distance from the road to prepare the fire risk zone map. The area of the prepared risk zone maps is divided into three zones, namely low, moderate, and high. From the study, it was found that the fire occurring in this area is due to natural as well as anthropogenic factors. The prepared forest fire risk zone maps are validated using the fire incidence data for the period from January 2003 to June 2019 collected from the records of the Forest Survey of India. The investigation revealed that 72% and 24% of the fire incidences occurred in the high risk zone of the maps prepared using the AHP and FR methods, respectively, which ascertained the superiority of the AHP method over the FR method for forest fire risk zone mapping. The receiver operating characteristic (ROC) curve analysis gives an area under the ROC curve (AUC) value of 0.767 and 0.567 for the AHP and FR methods, respectively. The risk zone maps will be useful for staff of the forest department, planners, and officials of the disaster management department to take effective preventive and mitigation measures.

ABSTRACTSpatial variation of Urban Land Surface Temperature (ULST) is a complex function of environmental, climatic, and anthropogenic factors. It thus requires specific techniques to quantify this phenomenon and its influencing factors. In this study, four models, Random Forest (RF), Generalized Additive Model (GAM), Boosted Regression Tree (BRT), and Support Vector Machine (SVM), are calibrated to simulate the ULST based on independent factors, i.e., land use/land cover (LULC), solar radiation, altitude, aspect, distance to major roads, and Normalized Difference Vegetation Index (NDVI). Additionally, the spatial influence and the main interactions among the influential factors of the ULST are explored. Landsat-8 is the main source for data extraction and Tehran metropolitan area in Iran is selected as the study area. Results show that NDVI, LULC, and altitude explained 86% of the ULST °C variation. Unexpectedly, lower LST is observed near the major roads, which was due to the presence of vegetation along the streets and highways in Tehran. The results also revealed that variation in the ULST was influenced by the interaction between altitude – NDVI, altitude – road, and LULC – altitude. This indicates that the individual examination of the underlying factors of the ULST variation might be unilluminating. Performance evaluation of the four models reveals a close performance in which their R2 and Root Mean Square Error (RMSE) fall between 60.6–62.1% and 2.56–2.60 °C, respectively. However, the difference between the models is not statistically significant. This study evaluated the predictive performance of several models for ULST simulation and enhanced our understanding of the spatial influence and interactions among the underlying driving forces of the ULST variations.

Forest are essential natural resources having the role of supporting economic activity, which plays a significant role in regulating the climate and the carbon cycle. Forest ecosystems increasingly threatened by fires caused by a range of natural and anthropogenic factors. Hence, spatial assessment of fire risk is critical to reducing the impacts of wildland fires. In the current research, the evaluation of forest fire risk (FFR) assessment performed by geospatial data of Melgaht Tiger Reserve Forest (MTRS), Maharashtra, India. We have used eleven natural and anthropogenic parameters (slope, altitude, topographic position index (TPI), aspect, rainfall, land surface temperature (LST), air temperature, wind speed, normalized differential vegetation index (NDVI), distance to road and distance to settlement) for FFR assessment based on the Analytic hierarchy process (AHP) and Frequency ratio (FR) models in a GIS framework. The results from AHP and FR models shown similar trends. The AHP model was significantly higher accuracy than the FR model. AHP and FR models based FHR maps were classified into five classes (very low, low, moderate, high, and very high). According to the generated FFR maps, the very high-risk class was found at some forest blocks (Mangtya, Kund, Gudfata, Katharmal, Amyar). The sensitivity analysis showed that some parameters (wind speed, air temperature, LST, slope, altitude, distance to settlement, and distance to the road) were more sensitive to forest fire risk. The FFR results were justified by the forest fire sample points (Forest Survey of India) and burn images (2010–2018). This work will provide a basic guideline for effective geo-environmental planning and management of Melgaht Tiger Reserve Forest.

Strength of association between vegetation greenness and its drivers across China between 1982 and 2015: Regional differences and temporal variations

Soil erosion is one of the main and most widespread types of soil degradation, determined by environmental and anthropogenic factors, which significantly impact the physical, chemical and biological properties of the soil, favoring its degradation. Consequently, through this research we sought to evaluate the risk of soil erosion in the canton of Mocache through the application of multicriteria analysis techniques and the use of geographic information systems (GIS). The CORINE model included several criteria for the zoning of potential and actual erosion risk, including erosivity, defined by the Modified Fournier Index (MFI) and the Bagnouls-Gaussen Aridity Index (BGI); erodibility, composed of soil texture, stoniness and depth; soil occupation; slope-orientation; and vegetation cover, through the estimation of the Normalized Difference Vegetation Index (NDVI). An expert consultation supported by Analytic Hierarchy Process (AHP) was applied to define the contribution of each criterion to erosion risk. The resulting layers were combined using a weighted relationship (map algebra) within the GIS environment. It was obtained that slope-orientation and erosivity explain a large part of the erosion risk, with a weighting of 42.46% and 27.48% in order, according to the expert evaluation and the AHP. The high erosivity was contributed by the predominant climatic factors: intense precipitation and high temperatures. In addition, it was found that the potentialand current erosion risk represent 13.1% (7,244 ha) and 10.7% (5,905 ha) of the territorial extension. It is concluded that physical-natural, meteorological factors and anthropogenic activities contribute to the risk of potential and current soil erosion in Mocache canton

Flood susceptibility mapping is a vital tool for effective disaster management and urban planning, particularly in regions experiencing recurrent flooding.This study assesses flood susceptibility in the Kolkata Municipal Corporation (KMC) area, covering 206.08 km 2 with a population of approximately 4.6 million (2011 census), using an integrated multi-criteria decision-making approach.The analysis combines the Analytical Hierarchy Process (AHP) with a Weighted Overlay Technique (WOT) within a Geographic Information System (GIS) framework.Kolkata's location is in the lowermost riparian zone of the Ganga Basin, along with its exposure to Himalayan River inflows, cyclonic storms, drainage congestion, embankment failures, and persistent waterlogging, makes it highly vulnerable to flooding.The methodology involves three major stages which includes data collection, preparation of thematic layers, and parameter weighting using AHP.Thematic maps were generated for key flood-influencing factors, including elevation, slope, aspect, drainage density, flow accumulation, flow direction, topographic wetness index (TWI), proximity to rivers, annual rainfall, land use/land cover (LULC), normalized difference vegetation index (NDVI), surface roughness, and contour characteristics.Flood susceptibility was classified into five zones such as very low, low, moderate, high, and very high.AHP-derived weights highlight precipitation, proximity to rivers, and LULC as dominant contributors, followed by elevation, TWI, NDVI, slope, proximity to roads, drainage density, and aspect.Model performance was quantitatively validated using Receiver Operating Characteristic (ROC) curve analysis based on historical flood occurrence data.The resulting Area Under the Curve (AUC) value of 0.89 indicates very good predictive accuracy and strong agreement between predicted susceptibility zones and observed flood locations.The inclusion of ROC-based validation, along with acceptable consistency ratio values from AHP, enhances the reliability and transparency of the analysis.The findings provide a robust decision-support framework for urban planners and disaster management authorities to prioritize mitigation strategies and improve flood resilience in Kolkata.

Research on wildfire risk can quantitatively assess the risk of wildfire damage to the population, economy, and natural ecology. However, existing research has primarily assessed the spatial risk of wildfires across an entire region, neglecting the impact of different land-use types on the assessment outcomes. The purpose of the study is to construct a framework for assessing wildfire risk in different land-use types, aiming to comprehensively assess the risk of wildfire disasters in a region. We conducted a case study in Central China, collecting and classifying historical wildfire samples according to land-use types. The Light Gradient Boosting Machine (LGBM) was employed to construct wildfire susceptibility models for both overall and individual land-use types. Additionally, a subjective and objective combined weighting method using the Analytic Hierarchy Process (AHP) and Entropy Weight Method (EWM) was utilized to build the wildfire vulnerability model. By integrating susceptibility and vulnerability information, we comprehensively assessed the combined risk of wildfire disasters across land-use types. The results demonstrate the following: (1) Assessing wildfire susceptibility based on different land-use types compensated for limitations in analyzing overall wildfire susceptibility, with a higher prediction performance and more detailed susceptibility information. (2) Significant variations in wildfire susceptibility distribution existed among different land-use types, with varying contributions of factors. (3) Using the AHP-EWM combined weighting method effectively addressed limitations of a single method in determining vulnerability. (4) Land-use types exerted a significant impact on wildfire risk assessment in Central China. Assessing wildfire risk for both overall and individual land-use types enhances understanding of spatial risk distribution and specific land use risk. The experimental results validate the feasibility and effectiveness of the proposed evaluation framework, providing guidance for wildfire prevention and control.

Understanding spatiotemporal shifts in vegetation and their climatic and anthropogenic regulatory factors can offer a crucial theoretical basis for environmental conservation and restoration. In this article, the normalized difference vegetation index (NDVI) of the Miaoling area from 2000 to 2020 is studied using a trend analysis and the Mann–Kendall mutation test (MK test) to review the vegetation’s dynamic changes. Our study uses the Hurst index, a partial correlation analysis, and a geographic detector to investigate the contributions of climate change and human activities to regional vegetation changes and their drivers. We found that Miaoling’s annual average NDVI was between 0.66 and 0.83 in 2000–2020, with a mean of 0.766. The overall trend was slow upward (0.0009/year), and 53.82% of the region continued to grow and gradually increased from west to east in the spatial domain, among which the karst regional NDVI distribution area and its growth rate were higher than those of non-karst sites. Based on correlations between climatic factors and NDVI, precipitation seasonality (coefficient of variation, CV) had the strongest correlation (positive correlation) with NDVI, while vapor pressure deficit (VPD) had a negative correlation with NDVI. In the interaction, human activities played a dominant role in the influence of NDVI on the vegetation of Miaoling. The night light index had the most explanatory power on the NDVI (q = 0.422), and the interaction between anthropogenic factors and other factors dominated its explanatory power. This study has academic and practical importance for the management, protection, and sustainable development of karst basins.

PDF HTML阅读 XML下载 导出引用 引用提醒 盐池县2000-2012年植被变化及其驱动力 DOI: 10.5846/stxb201403050374 作者: 作者单位: 1. 西北土地退化与生态恢复国家重点实验室培育基地宁夏大学,1. 西北土地退化与生态恢复国家重点实验室培育基地宁夏大学 作者简介: 通讯作者: 中图分类号: 基金项目: 国家重点基础研究计划(973)前期专项(2012CB723206);国家自然科学基金项目(41201438);宁夏大学211建设项目 Vegetation dynamics over 2000-2012 and its driving factors in Yanchi County, Ningxia Province Author: Affiliation: 1.Cultivation Base for State Key Lab. of Restoration and Reconstruction of Degraded Ecosystem in Northwestern of China,Yinchuan,1.Cultivation Base for State Key Lab. of Restoration and Reconstruction of Degraded Ecosystem in Northwestern of China,Yinchuan Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:荒漠草原区的植被对防治荒漠化、维护生态屏障具有决定性作用,宁夏盐池县作为其典型代表,近13年的植被变化深受气候变化和人类活动的综合影响.基于MODIS NDVI等数据,运用趋势分析、经验模态分解和空间叠置分析等方法,对盐池县2000-2012年的植被动态变化进行研究,结果表明:(1)2000-2012年盐池县NDVI在0.2-0.4之间呈波动上升趋势,上升幅度为0.078/10 a,上升趋势显著;总体来说,植被稳定性低,年际间波动或转换频繁、幅度大;(2)NDVI的波动分量与残余分量方差贡献率各占50%,且NDVI波动呈减弱趋势.促使NDVI波动的主控因子是年降水量,但其影响在减弱;(3)推动NDVI趋势性上升的主要因素是土地利用方式改善和类型变化,但土地利用方式改善对NDVI的贡献远远大于土地利用类型变化对NDVI的贡献.因此,荒漠草原区的生态改善应以保护为主,辅之以必要的生态重建,走以适度开发带动整体保护的道路. Abstract:The vegetation of the desert steppe plays an important role in preventing desertification, maintaining ecosystem stability, and constructing "Ecological Barrier" in Northwest China. However, most desert steppe ecosystems are very fragile and constantly face the risk of degradation. Yanchi County, located in eastern Ningxia province, is a typical desert steppe, and its vegetation dynamics in the past 13 years have been affected by both climate change and human activities. In order to explore the process and driving factors based on the Normalized Difference Vegetation Index (NDVI) derived from the Moderate-resolution Imaging Spectroradiometer (MODIS), we studied the vegetation dynamics in Yanchi County from 2000 to 2012. We used 296 scenes of MODIS NDVI data, all of these were converted to an Albers conical projection system and GeoTIF format by using the MODIS Reprojection Tool, and smoothed using the Savitzky-Golay filter to reconstruct a high-quality NDVI time-series data set. Annual and quarterly NDVI were synthesized using the Maximum Value Composite (MVC) method. In addition to MODIS NDVI data, the land use data of Yanchi in 2000 and 2011, meteorological data, and social statistical data were also used in this study. Multiple methods were used to analyze the vegetation dynamics in Yanchi County. A linear regression with an F test was used to analyze the trend of NDVI and its significance. The non-parametric Mann-Kendall test was used to detect the abrupt change in the long-term NDVI from 2000 to 2012. A non-linear and non-stationary signal analysis method, Empirical Mode Decomposition (EMD), was used to isolate the amplitude-frequency determining the temporally varying trend of NDVI, and spatial overlay analysis was used to analyze the influence of land use and land cover change on vegetation dynamics. The results showed: (1) the average NDVI values of all pixels in Yanchi County, which were composited annually by the MVC method, ranged from 0.2 to 0.4 in the period 2000-2012. The vegetation index in this area was very low because there is a typical desert steppe, but significantly (P < 0.05) increased by 0.078 per 10 a, which was faster than that in the Three-North Shelter Forest Program region. Overall, the vegetation in Yanchi County lacks stability and has frequent, large-amplitude inter-annual fluctuations. (2) The Empirical Mode Decomposition found that the NDVI time-series data included two Intrinsic Mode Function (IMF) components with 4 and 9 year quasi-periodic fluctuations. The variance contribution of the first IMF component was about 50%, almost the same as that of the NDVI residue component, which increased significantly over the 13 years. However, the intensity of NDVI fluctuation decreased because the fluctuation in precipitation, one of its main driving factors, declined. (3) Improvements in land use and land cover change were the main drivers for NDVI increase. The former made a larger contribution than the latter. Therefore, protection is the first option for improving the ecological environment, and proper reconstruction can be used as a supplement in desert steppe. A proper strategy for vegetation restoration and maintenance should be adopted, and overall protection can be implemented through scientific and harmonized development. 参考文献 相似文献 引证文献

Land Surface Temperature (LST) affects exchange of energy between earth surface and atmosphere which is important for studying environmental changes. However, research on the relationship between LST, Land Use Land Cover (LULC), and Normalized Difference Vegetation Index (NDVI) with topographic elements in the lower Himalayan region has not been done. Therefore, the present study explored the relationship between LST and NDVI, and LULC types with topographic elements in the lower Himalayan region of Pakistan. The study area was divided into North-South, West-East, North-West to South-East and North-East to South-East directions using ArcMap 3D analysis. The current study used Landsat 8 (OLI/TIRS) data from May 2021 for LULC and LST analysis in the study area. The LST data was obtained from the thermal band of Landsat 8 (TIRS), while the LULC of the study areas was classified using the Maximum Likelihood Classification (MLC) method utilizing Landsat 8 (OLI) data. TIRS collects data for two narrow spectral bands (B10 and B11) with spectral wavelength of 10.6 μm–12.51 μm in the thermal region formerly covered by one wide spectral band (B6) on Landsat 4–7. With 12-bit data products, TIRS data is available in radiometric, geometric, and terrain-corrected file format. The effect of elevation on LST was assessed using LST and elevation data obtained from the USGS website. The LST across LULC types with sunny and shady slopes was analyzed to assess the influence of slope directions. The relationship of LST with elevation and NDVI was examined using correlation analysis. The results indicated that LST decreased from North-South and South-East, while increasing from North-East and South-West directions. The correlation coefficient between LST and elevation was negative, with an R-value of −0.51. The NDVI findings with elevation showed that NDVI increases with an increase in elevation. Zonal analysis of LST for different LULC types showed that built-up and bare soil had the highest mean LST, which was 35.76 °C and 28.08 °C, respectively, followed by agriculture, vegetation, and water bodies. The mean LST difference between sunny and shady slopes was 1.02 °C. The correlation between NDVI and LST was negative for all LULC types except the water body. This study findings can be used to ensure sustainable urban development and minimize urban heat island effects by providing effective guidelines for urban planners, policymakers, and respective authorities in the Lower Himalayan region. The current thermal remote sensing findings can be used to model energy fluxes and surface processes in the study area.

Gelephu, located in the Himalayan region, has undergone significant development activities due to its suitable topography and geographic location. This has led to rapid urbanization in recent years. Assessing land use land cover (LULC) dynamics and Normalized Difference Vegetation Index (NDVI) can provide important information about urbanization trends and changes in vegetation health, respectively. The use of Geographic Information Systems (GIS) and Remote Sensing (RS) techniques based on various satellite products offers a unique opportunity to analyze these changes at a local scale. Exploring Bhutan’s mandate to maintain 60% forest cover and analyzing LULC transitions and vegetation changes using Sentinel-2 satellite imagery at 10 m resolution can provide important insights into potential future impacts. To examine these, we first performed LULC mapping for Gelephu for 2016 and 2023 using a Random Forest (RF) classifier and identified LULC changes. Second, the study assessed the dynamics of vegetation change within the study area by analysing the NDVI for the same period. Furthermore, the study also characterized the resulting LULC change for Gelephu Thromde, a sub-administrative municipal entity, as a result of the notable intensity of the infrastructure development activities. The current study used a framework to collect Sentinel-2 satellite data, which was then used for pre-and post-processing to create LULC and NDVI maps. The classification model achieved high accuracy, with an area under the curve (AUC) of up to 0.89. The corresponding LULC and NDVI statistics were analysed to determine the current status of the LULC and vegetation indices, respectively. The LULC change analysis reveals urban growth of 5.65% and 15.05% for Gelephu and Gelephu Thromde, respectively. The NDVI assessment shows significant deterioration in vegetation health with a 75.11% loss of healthy vegetation in Gelephu between 2016 and 2023. The results serve as a basis for strategy adaption required to examine the environmental protection and sustainable development management, and the policy interventions to minimize and balance the ecosystem, taking into account urban landscape.

Geo-environmental hazards are one of the most common natural disasters, posing significant dangers due to their highly unpredictable nature. The Upper Indus Basin is one of the most vulnerable regions for geo-hazards in Pakistan. The purpose of this research was to develop a technique using a geographical information system (GIS) to evaluate the Susceptibility to Multiple Geo-Environmental Hazards in the Upper Indus Basin, Pakistan. For this purpose, ten main parameters, including slope, aspect, profile curvature, land use land cover (LULC), normalized difference vegetation index (NDVI), normalized difference moisture index (NDMI), buffer zone along river, earth quake magnitude, annual mean temperature, and rainfall, were used and reclassified with specific percentage of every layer according to its ability to affect geo-environmental hazards. Reclassified LULC and elevation layers were combined using GIS spatial analysis and analytical hierarchy process (AHP) tools. According to our results, the weightage of different parameters like NDVI, NDMI, temperature, and rainfall were noted as 0.17, 0.1, 0.12, and 0.14, respectively, in the Upper Indus Basin, Pakistan. The minimum and maximum overall accuracy values were calculated in the range of 0.82 (grassland) and 0.93(built-up land), respectively. Almost 41,556 km 2 (25.21%) of the studied area was noted as a low-risk area, 72,153 km 2 (43.78%) a moderately risk area, and 51,105 km 2 (31.01%) a high-risk area for Geo-Environmental Hazards Susceptibility in Upper Indus Basin. Our outcomes showed that the area is susceptible to geo-environmental hazards that are moderately risky and high-risk area. Our research will be useful for stakeholders and policymakers who are formulating strategies for disaster mitigation and preparedness in the Upper Indus Basin, Pakistan.

Aims As a typical region of soil erosion on the Loess Plateau, the Jinghe River watershed has had long-term land exploitation and soil erosion. Our objective was to study trends in the change of vegetation cover and to explore driving factors, including both climatic and anthropogenic aspects. Methods We calculated normalized difference vegetation index (NDVI) trends using GIMMS NDVI data from 1982 to 2005 in the Jinghe River watershed. Its trends were compared with precipitation and air temperature trends calculated from climate data from the 19 meteorological stations in the watershed. A 3 × 3 pixel buffer area centered on each station was used to analyze relationships between climate and vegetation. Anthropogenic factors were represented by land use data obtained from the Resource-Environment Database of the Chinese Academy of Sciences. We analyzed the proportion of each land type in areas of different NDVI trends to illustrate the effects of human activities. Important findings NDVI had no significant trends in 79.64% of the Jinghe River watershed in the 24-year period. NDVI had significant positive trends in 16.33% of the area, located in the middle and southern parts of the watershed. NDVI had significant negative trends in 4.03% of the area, located in the northern part of the watershed. Precipitation had no significant trends, and temperature had significant positive trends forall of the 19 weather stations. The spatial differences of NDVI trends could not explained by changes in precipitation and air temperature. The anthropogenic factors seemed more important. Land use analysis indicated that the percentages of land use types in areas of different NDVI trends changed little. Plantation was dominant in the area where NDVI had significant positive trends, and grassland was dominant in the area where NDVI had significant negative trends. Results suggest that the changes in plantations resulted in the significant positive trends of NDVI, and woodland loss and grassland degeneration resulted in the significant negative trends of NDVI.

Vegetation, being a core component of ecosystems, is known to be influenced by natural and anthropogenic factors. This study used the annual mean Normalized Difference Vegetation Index (NDVI) as the vegetation greenness indicator. The variation in NDVI on Hainan Island was analyzed using the Theil–Sen median trend analysis and Mann–Kendall test during 2000–2019. The influence of natural and anthropogenic factors on the driving mechanism of the spatial pattern of NDVI was explored by the Multiscale Weighted Regression (MGWR) model. Additionally, we employed the Boosted Regression Tree (BRT) model to explore their contribution to NDVI. Then, the MGWR model was utilized to predict future greenness patterns based on precipitation and temperature data from different Shared Socioeconomic Pathway (SSP) scenarios for the period 2021–2100. The results showed that: (1) the NDVI of Hainan Island forests significantly increased from 2000 to 2019, with an average increase rate of 0.0026/year. (2) the R2 of the MGWR model was 0.93, which is more effective than the OLS model (R2 = 0.42) in explaining the spatial relationship. The spatial regression coefficients of the NDVI with temperature ranged from −10.05 to 0.8 (p &lt; 0.05). Similarly, the coefficients of Gross Domestic Product (GDP) with the NDVI varied between −5.98 and 3.28 (p &lt; 0.05); (3) The natural factors played the most dominant role in influencing vegetation activities as a result of the relative contributions of 83.2% of forest NDVI changes (16.8% contributed by anthropogenic activities). (4) under SSP119, SSP245, and SSP585 from 2021 to 2100, the NDVI is projected to have an overall decreasing pattern under all scenarios. This study reveals the trend of greenness change and the spatial relationship with natural and anthropogenic factors, which can guide the medium and long-term dynamic monitoring and evaluation of tropical forests on Hainan Island.