| 基于随机森林的广西岩溶区与非岩溶区土壤有机碳时空演变特征及驱动因素 |
| 摘要点击 1482 全文点击 21 投稿时间：2025-07-29 修订日期：2025-10-01 |
| 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 岩溶区和非岩溶区 土壤有机碳密度（SOCD） 时空分布 驱动因子 随机森林（RF） 沙普利加和解释（SHAP） |
| 英文关键词 karst and non-karst areas soil organic carbon density（SOCD） spatiotemporal distribution driving factors random forest （RF） SHapley Additive exPlanations（SHAP） |
| DOI 10.13227/j.hjkx.202507378 |
| 作者 | 单位 | E-mail | | 令狐小钰 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | lhxylhxy2021@163.com | | 魏兴萍 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | xingpingwei@126.com | | 蒲俊兵 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | | | 杨帆 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | | | 陈诗蝶 | 北京师范大学地理科学学部, 地表过程与资源生态国家重点实验室, 北京 100875 | | | 李慧 | 湖南师范大学地理科学学院, 长沙 410081 | | | 黄柔曼 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | | | 蔡云丽 | 重庆师范大学山区生态系统碳循环与碳调控重庆市重点实验室, 重庆 401331
重庆师范大学地理与旅游学院, 重庆 401331 | |
|
| 中文摘要 |
| 岩溶区与非岩溶区有机碳密度（SOCD）差异影响省域尺度土壤碳汇的精确评估，是土壤碳库研究的热点与难点. 以广西岩溶区和非岩溶区为例，基于1980年和2010年土壤剖面及环境变量数据，构建随机森林模型以剖析岩溶区与非岩溶区表层 SOCD的时空分异特征；并通过SHAP值和偏依赖图揭示两区域中不同驱动因子对表层SOCD的响应特征. 结果表明：①1980年和2010年岩溶区SOCD均值均高于同期非岩溶区. 岩溶区SOCD年均变化率的减少区与石漠化扩张区在空间上呈现出高度一致性. ②全氮含量和坡度均为影响两区域SOCD的关键驱动因子. 粉粒含量、土壤类型和归一化植被指数对岩溶区SOCD有显著影响，而年均气温、高程和年总降水量则对非岩溶区SOCD产生重要影响. ③SOCD与各驱动因子之间呈多样化特征. SOCD与全氮含量、粉粒含量、归一化植被指数和高程呈非线性正相关，与年均气温和年总降水量呈非线性负相关，而与坡度呈先负相关后正相关的非线性关系，且各类关系中均存在明显阈值. 研究结果可为系统评估土壤碳汇潜力、制定土壤碳库调控措施及碳管理分区方案提供科学依据. |
| 英文摘要 |
| The difference in soil organic carbon density （SOCD） between karst and non-karst areas affects the accurate assessment of soil carbon sinks at the provincial scale， making it a hot and difficult topic in soil carbon pool research. Taking the karst and non-karst regions of Guangxi as an example， and based on soil profile and environmental variable data from 1980 and 2010， a random forest model was constructed to analyze the spatiotemporal differentiation characteristics of topsoil SOCD in karst and non-karst region. Furthermore， SHAP values and partial dependence plots were employed to reveal the response characteristics of different driving factors to topsoil SOCD in the two regions. The results show that： ① The mean SOCD in the karst region was higher than that in the non-karst region in both 1980 and 2010. The areas with a decrease in the annual mean change rate of SOCD in the karst region showed a high spatial consistency with the expansion areas of rocky desertification. ②Total nitrogen content and slope were both key driving factors affecting SOCD in the two regions. In addition， silt content， soil types， and normalized difference vegetation index significantly influenced SOCD in the karst region， while mean annual temperature， elevation， and total precipitation exerted important impacts on SOCD in the non-karst region. ③ The relationships between SOCD and the driving factors were characterized by diversity. SOCD showed a nonlinear positive correlation with total nitrogen content， silt content， normalized vegetation index， and elevation； a nonlinear negative correlation with mean annual temperature and total annual precipitation； and a nonlinear relationship with slope that was initially negative and then positive. Moreover， there were distinct thresholds in all these relationships. These research findings provide a scientific basis for systematically assessing the potential of soil carbon sinks， formulating measures for regulating soil carbon pools， and developing carbon management zoning strategies. |
