Jin Aibo, Li Hui, Wang Xiangrong, Wang Ziyao
School of Landscape Architecture, Beijing Forestry University, Beijing, 100083, China.
Department of Landscape Architecture, School of Architecture, Tsinghua University, Beijing, 100084, China.
Sci Rep. 2025 Apr 16;15(1):13136. doi: 10.1038/s41598-025-97156-3.
Climate change and human expansion are primary drivers of ecological degradation in metropolitan areas, underscoring the necessity of examining the complex interplay between environmental factors and ecological quality. This study investigates the spatial-temporal evolution of ecological quality within the Beijing Metropolitan Area (BMA) from 2000 to 2020 and proposes a comprehensive assessment framework integrating machine learning techniques and spatial heterogeneity analyses. Ecological quality is quantitatively evaluated using the Remote Sensing Ecological Index (RSEI), leveraging MODIS imagery, climate data, land use patterns, and soil characteristics. Spatial clustering patterns of ecological quality are identified through RSEI calculations and spatial autocorrelation analyses, while future trends are projected utilizing the coefficient of variation, Sen and Mann-Kendall methods, and the Hurst index. The XGBoost algorithm elucidates the multifaceted driving mechanisms, and geographically weighted regression (GWR) quantifies the spatial variability of these drivers. The application of XGBoost reveals nonlinear relationships among ecological drivers, and GWR enhances spatially explicit interpretations of these relationships. Results indicate an overall improvement in ecological quality, with the RSEI rising from 0.428 in 2000 to 0.480 in 2020, corresponding to an annual average increase of approximately 0.55%. Notable spatial variability exists, with ecological quality consistently higher in the Taihang Mountains relative to lower-altitude plains. Current ecological protection policies have effectively mitigated ecological degradation in approximately 32.35% of the study area; however, significant environmental pressures persist in urban-rural transition zones and plain regions. Topography and soil properties emerge as dominant influencing factors, while climate indirectly influences ecological quality by shaping vegetation coverage. Human activities predominantly exert negative impacts within urban expansion zones. This research offers a robust quantitative framework for regional ecological conservation, providing critical insights to inform sustainable development and environmental policy-making.
气候变化和人类扩张是大都市地区生态退化的主要驱动因素，这凸显了研究环境因素与生态质量之间复杂相互作用的必要性。本研究调查了2000年至2020年北京大都市地区（BMA）生态质量的时空演变，并提出了一个整合机器学习技术和空间异质性分析的综合评估框架。利用中分辨率成像光谱仪（MODIS）图像、气候数据、土地利用模式和土壤特征，通过遥感生态指数（RSEI）对生态质量进行定量评估。通过RSEI计算和空间自相关分析确定生态质量的空间聚类模式，同时利用变异系数、森和曼-肯德尔方法以及赫斯特指数预测未来趋势。XGBoost算法阐明了多方面的驱动机制，地理加权回归（GWR）量化了这些驱动因素的空间变异性。XGBoost的应用揭示了生态驱动因素之间的非线性关系，而GWR增强了这些关系的空间明确解释。结果表明生态质量总体有所改善，RSEI从2000年的0.428上升到2020年的0.480，年均增长约0.55%。存在显著的空间变异性，太行山区的生态质量始终高于低海拔平原地区。当前的生态保护政策已有效缓解了研究区域约32.35%的生态退化；然而，城乡过渡区和平原地区仍存在巨大的环境压力。地形和土壤性质是主要影响因素，而气候通过影响植被覆盖间接影响生态质量。人类活动主要在城市扩展区产生负面影响。本研究为区域生态保护提供了一个强大的定量框架，为可持续发展和环境政策制定提供了关键见解。
