| 基于XGBoost-SHAP模型的北京市生态系统服务空间格局及驱动因素分析 |
| 摘要点击 530 全文点击 45 投稿时间：2025-01-15 修订日期：2025-04-06 |
| 查看HTML全文
查看全文 查看/发表评论 下载PDF阅读器 |
| 中文关键词 生态系统服务（ESs） 空间格局 权衡与协同 驱动因素 XGBoost算法 |
| 英文关键词 ecosystem services（ESs） spatial pattern trade-offs and synergies driving factors XGBoost algorithm |
| DOI 10.13227/j.hjkx.20260230 |
|
| 中文摘要 |
| 研究生态系统服务之间的空间格局及其背后的驱动因素，对于强化生态管理及促进环境的可持续发展至关重要. 以北京市为研究区域，应用InVEST模型对2000~2020年生境质量、碳储量、产水量以及土壤保持采用空间自相关、冷点/热点分析及双变量空间自相关分析方法研究生态系统服务的空间相关性、权衡与协同关系，并通过XGBoost-SHAP模型剖析影响生态系统服务的关键因素. 结果表明：①生境质量的高值区域主要集中在地势较高且人类活动干扰较小的地区；碳储量呈现出西北高、东南低的空间分布态势；产水量的高值区集中在城镇区域；土壤保持的高值区域主要分布在西南部，而在北部则呈现零散分布状态. ②全局空间自相关分析显示，4种生态系统服务的全局Moran's I 指数均通过显著性检验，且均表现出显著的高值聚集特征. ③生境质量、碳储量和土壤保持之间存在显著的协同关系，而产水量与生境质量、碳储量、土壤保持之间则表现出一定的权衡关系. ④XGBoost回归模型在训练集与测试集上均展现出良好的预测性能，且训练集的预测效果优于测试集. SHAP模型解析表明，高程是影响4种生态系统服务的关键驱动因子，坡度显著影响生境质量、碳储量和土壤保持，人口密度主要作用于生境质量和产水量，而年降水量则对产水量和土壤保持具有重要影响. 研究结果可为北京市生态系统服务空间格局优化及生态保护策略的制定提供科学支撑. |
| 英文摘要 |
| Studying the spatial patterns of ecosystem services and their driving factors is crucial for strengthening ecological management and promoting sustainable environmental development. This research focuses on Beijing as the study area. The InVEST model was applied to analyze the spatial correlation， trade-offs， and synergies of habitat quality， carbon storage， water yield， and soil retention from 2000 to 2020. The analysis utilized methods such as spatial autocorrelation， cold/hot spot analysis， and bivariate spatial autocorrelation analysis. Additionally， the XGBoost-SHAP model was employed to identify the key factors affecting ecosystem services. The results showed that： ① The high-value areas of habitat quality were mainly concentrated in regions with higher terrain and less interference from human activities. Carbon storage exhibited a spatial distribution trend that was high in the northwest and low in the southeast. The high-value areas of water yield were concentrated in urban areas， while the high-value areas of soil conservation were primarily distributed in the southwest and were more scattered in the north. ② Global spatial autocorrelation analysis indicated that the global Moran's I indices for the four ecosystem services all passed the significance test and demonstrated significant high-value aggregation characteristics. ③ There was a significant synergistic relationship between habitat quality， carbon storage， and soil conservation. However， there was a trade-off between water yield and these factors. ④ The XGBoost regression model showed good prediction performance on both the training set and the test set， with the predictive performance on the training set being better than that on the test set. The SHAP model analysis indicated that elevation was the key driving factor affecting the four ecosystem services. Slope significantly affected habitat quality， carbon storage， and soil conservation. Population density mainly affected habitat quality and water yield， while annual precipitation had an important influence on water yield and soil conservation. The research results can provide scientific support for optimizing the spatial patterns of ecosystem services and formulating ecological protection strategies in Beijing. |
