气候变化研究进展2026，Vol.22Issue(3)：316-330,15.DOI:10.12006/j.issn.1673-1719.2026.029
1986-2024年京津冀地区植被时空动态评估:基于XGBoost-SHAP方法的驱动机制研究
Spatiotemporal dynamics of vegetation in the Beijing-Tianjin-Hebei region during 1986-2024:a driving mechanism analysis based on XGBoost-SHAP
摘要
Abstract
Vegetation dynamics in the Beijing-Tianjin-Hebei(BTH)region,a major urban agglomeration in northern China,are critical for ecological barrier construction and regional sustainable development.Based on a cloud-free and seamless Landsat normalized difference vegetation index(NDVI)dataset with a spatial resolution of 30 m from 1986 to 2024,this study investigated the spatiotemporal dynamics of vegetation using the Theil-Sen slope estimator,Mann-Kendall test,Hurst exponent,and coefficient of variation.The driving mechanisms were further quantified within the XGBoost-SHAP framework.The results showed that the growing-season NDVI in the BTH region exhibited an overall increasing trend during the past 39 years,with a mean value of 0.5068 and an annual increasing rate of 0.0037,corresponding to an increase of 30.46％,indicating a significant improvement in regional vegetation cover.Stage-based analysis revealed that the proportion of vegetation degradation areas decreased from 33％to 10％,whereas the proportion of improvement areas increased to 79％.Approximately 86％of the region exhibited persistent vegetation improvement;however,nearly 45％of the area still showed high interannual variability.Spatially,areas with high vegetation coverage were mainly concentrated in the mountainous forest regions of northeastern BTH,whereas areas with low vegetation coverage were sporadically distributed in urban built-up areas and high-altitude regions.The contributions of driving factors exhibited distinct stage-dependent characteristics.In the first stage,natural factors,particularly slope and precipitation,dominated vegetation changes.In the second stage,the contribution of land cover type increased substantially,reflecting the influence of ecological restoration policies.In the third stage,anthropogenic factors,including population density and nighttime light intensity,became increasingly important,indicating a shift toward human-nature coupled dynamics.Significant regional heterogeneity in driving factors was also observed:vegetation dynamics in Beijing and Tianjin were more strongly influenced by human activities,whereas topographic and climatic factors dominated in Hebei province.Overall,the findings suggest that vegetation improvement in the BTH region was primarily driven by the synergistic effects of ecological policies and climatic factors,although disturbances associated with rapid urbanization still warrant attention.This study provides scientific support for regional ecological management and the achievement of China's "dual-carbon"goals.关键词
京津冀/归一化植被指数(NDVI)/植被动态/XGBoost-SHAP/驱动机制Key words
Beijing-Tianjin-Hebei(BTH)region/Normalized Difference Vegetation Index(NDVI)/Vegetation dynamics/XGBoost-SHAP/Driving mechanisms引用本文复制引用
张欣,权畅,王小飞,张方敏..1986-2024年京津冀地区植被时空动态评估:基于XGBoost-SHAP方法的驱动机制研究[J].气候变化研究进展,2026,22(3):316-330,15.基金项目
国家重点研发计划(2023YFF0805402) （2023YFF0805402）
