Exploring the impact of natural and human activities on vegetation changes: An integrated analysis framework based on trend analysis and machine learning.
Journal:
Journal of environmental management
PMID:
39826360
Abstract
Climate, human activities and terrain are crucial factors influencing vegetation changes. Despite their crucial role, there is a notable lack of research exploring the nonlinear relationships between them and vegetation changes, especially over extended time series. This research integrates trend analysis with machine learning and SHAP technology, proposing a methodological analysis framework named Theil-Sen - Mann-Kendall - XGBoost - SHAP (TMXS), aiming to explore the nonlinear relationships between vegetation changes and their influencing factors. Taking vegetation changes in the Chengdu-Chongqing urban agglomeration from 2003 to 2022 as an example. The results indicate that the TMXS analytical framework can effectively quantify the nonlinear impact of assessment factors on vegetation changes. During the specified period, there was a general increase in LAI across the study area, with an annual growth rate of 0.0245/a. Notably, a significant decrease in LAI was observed in urban cores and areas undergoing rapid urbanization. The combined contribution of climatic and human activity factors to vegetation changes exceeded 65% in all regions, with temperature distribution being more critical than precipitation. Human activities accounted for 45.73% of the contribution to vegetation degradation in the study area, and vegetation degradation was more prone to occur in densely populated regions. Among the topographic factors, alterations in slope gradient have a relatively significant effect on changes in vegetation cover. The research findings demonstrate that the TMXS method is reliable in elucidating the nonlinear relationships between vegetation changes and influencing factors, which can aid in guiding regional vegetation restoration projects and ecological environment policies.