摘要
秦巴山区作为中央绿芯和中华碳库,发育了多样而独特的森林生态系统,其森林土壤有机碳(Soil organic carbon,SOC)动态评估对维系区域碳平衡具有至关重要的作用。对比6种机器学习算法,选取最优模型模拟2000—2023年秦巴山区森林SOC时空分布格局,利用沙普利叠加解释(Shapley additive explanations,SHAP)方法揭示环境因子和表层(0—20cm)SOC的非线性关系。结果表明:(1)XGBoost模型在SOC空间模拟中取得相对最优性能(R^(2)=0.73,RMSE=21.98g/kg),验证了其对复杂山地环境变量的交互解析能力。(2)环境协变量与秦巴山区森林SOC之间存在着非线性关系,生长季太阳辐射、海拔、生长季降水量和生长季均温作为关键因子在XGBoost模型中的贡献率分别为26.18%、14.50%、8.76%和5.77%,并存在阈值效应。(3)2000—2023年秦巴山区表层森林SOC呈“西高东低”空间格局,虽在不同时段内波动但总体呈上升趋势,且高海拔地区森林SOC对气候波动较为敏感。研究结果为深入理解区域碳循环机制提供科学依据,为制定精准森林管理和碳汇提升策略提供理论支持。
The Qinling-Daba Mountains,recognized as the Central Green Core and China's Carbon Reservoir,harbored a rich variety of unique forest ecosystems.Consequently,assessing the dynamics of forest soil organic carbon(SOC)in this region was pivotal for maintaining regional carbon balance.In this study,we compared six machine-learning algorithms and selected the optimal model to simulate the spatiotemporal distribution of forest SOC in the Qinling-Daba Mountains from 2000 to 2023.We subsequently applied the Shapley additive explanations(SHAP)method to elucidate the nonlinear relationships between environmental factors and surface SOC(0—20cm).The results showed that:(1)The XGBoost model demonstrated the best performance in spatial SOC simulation(R^(2)=0.73,RMSE=21.98g/kg),confirming its strength in analyzing interactions among complex mountain environmental variables;(2)Environmental covariates and forest SOC exhibited nonlinear relationships,with solar radiation during the growing season,elevation,precipitation during the growing season,and mean temperature during the growing season contributing 26.18%,14.50%,8.76%,and 5.77%,respectively,and displaying threshold effects;(3)Between 2000 and 2023,surface forest SOC presented a spatial pattern of high values in the west and low values in the east,showed a generally increasing trend despite temporal fluctuations,and proved more sensitive to climate variations at higher elevations.These findings provided a scientific basis for a deeper understanding of the regional carbon cycle and offered theoretical support for developing precise forest management and carbon-sink enhancement strategies.
作者
王晓峰
章玥
周潮伟
陈吉臻
黄志霖
刘世荣
王筱雪
周继涛
孙泽冲
白娟
吕一河
WANG Xiaofeng;ZHANG Yue;ZHOU Chaowei;CHEN Jizhen;HUANG Zhilin;LIU Shirong;WANG Xiaoxue;ZHOU Jitao;SUN Zechong;BAI Juan;LÜYihe(School of Land Engineering,Chang'an University,Xi'an 710054,China;Key Laboratory of Forest Ecology and Environment of National Forestry and Grassland Administration Ecology and Nature Conservation Institute,Chinese Academy of Forestry,Beijing 100091,China;State Key Laboratory of Urban and Regional Ecology,Research Center for Eco-Environmental Sciences,Chinese Academy of Sciences,Beijing 100085,China)
出处
《生态学报》
2026年第3期1193-1207,共15页
Acta Ecologica Sinica
基金
国家自然科学基金项目(72349002)
中国林业科学研究院基本科研业务费专项(CAFYBB2024ZA001)
长安大学中央高校基本科研业务费专项基金(chd220235240599)。
关键词
森林土壤有机碳
数字土壤制图
机器学习
秦巴山区
forest soil organic carbon
digital soil mapping
machine learning
Qinling-Daba Mountains
