基于全球数据库的森林土壤呼吸模型研究  

Study on Forest Soil Respiration Model Based on Global Databases

作　　者：田秋燕 赵正勇[1] 杨旗 朱航勇 TIAN Qiuyan;ZHAO Zhengyong;YANG Qi;ZHU Hangyong(Forestry College of Guangxi University,Nanning,Guangxi 530004,China;Guangdong Provincial Key Laboratory ofSilviculture,Protection and Utilization/Guangdong Academy of Forestry,Guangzhou,Guangdong 510520,China)

机构地区：[1]广西大学林学院,广西南宁530004 [2]广东省森林培育与保护利用重点实验室/广东省林业科学研究院,广东广州510520

出　　处：《林业与环境科学》2019年第6期1-6,共6页Forestry and Environmental Science

基　　金：广西自然科学基金项目(2018GXNSFAA050135);广东省林业科技计划项目(2019-07)

摘　　要：森林土壤呼吸在陆地生态系统的碳平衡中发挥了重要作用,准确估算森林土壤呼吸量对于了解陆地碳平衡的变化至关重要。这项研究以全球气候数据、全球森林土壤呼吸数据库为基础数据,通过开发人工神经网络(ANN)模型建立由年平均气温(MAT)、年平均降水(MAP)、森林类型驱动的土壤呼吸模型,预测全球森林土壤呼吸变化。模型估算的结果表明,从1960年到2017年,全球森林平均年土壤呼吸量为40.10±0.48 PgCyr",全球森林土壤对全球土壤呼吸的贡献在40.9%-49.8%之间。人工神经网络模型预测的准确度达到0.63,进一步改善了全球森林土壤呼吸模型预测的精度。Forest soil respiration is a critical process in the carbon cycling of terrestrial ecosystems.It is very important to estimate the rate of annual forest soil respiration for understanding the change of carbon balance.Based on global climate data and global forest soil respiration database,this study developed an artificial neural network(ANN)model driven by mean annual temperature(MAT),mean annual precipitation(MAP)and forest types to predict the changes of global forest soil respiration.From 1960 to 2017,the average annual soil respiration of global forest was 40.10±0.48 Pg C yr'1,which contributed between 40.9%and 49.8%to the global soil respiration.The prediction accuracy of artificial neural network model reached 0.63 in this study,which further improved the performance of the global forest soil respiration model.

关 键 词：全球气候数据 全球森林土壤呼吸数据库 森林生态系统 人工神经网络 

分 类 号：S718[农业科学—林学]