基于随机森林的杉木林林下草本可燃物载量估算——以安化县为例  

Estimation of Herbaceous Fuel Load in Chinese Fir Forests Using Parameter-Optimized Random Forest Modelling:A Case Study in Anhua County

作　　者：程霞 陈振雄 杨旸谷 阳胜男 丁山 樊勇明 CHENG Xia;CHEN Zhenxiong;YANG Yanggu;YANG Shengnan;DING Shan;FAN Yongming(Central South Academy of Inventory and Planning of NFGA,Changsha 410014,Hunan,China;Changsha Central South Forestry Investigation,Planning and Design Co.,Ltd.,Changsha 410014,Hunan,China)

机构地区：[1]国家林业和草原局中南调查规划院,湖南长沙410014 [2]长沙中南林业调查规划设计有限公司,湖南长沙410014

出　　处：《中南林业调查规划》2025年第1期38-44,76,共8页Central South Forest Inventory and Planning

基　　金：湖南省科技厅重点研发项目“基于多源数据集成的森林火灾监测预警及快速应急关键技术研究和应用示范”(2022NK2045)。

摘　　要：为研究杉木林林下草本可燃物载量的估算方法及空间分布,在湖南省安化县设60个杉木林样地,利用随机森林模型估算样地的草本可燃物载量,并分析各因子在估算过程中的重要程度。结果表明:1)随机森林模型对草本可燃物载量的估算精度较高,训练集的OOB error为0.1225,测试集的R^(2)为0.9159,RMSE为0.0332 t/hm^(2),均通过了0.01水平的置信度检验;2)估算过程中各因子的重要性存在显著差异,其重要性指数排名从大到小依次为林分因子(88.54%)>遥感光谱特征指数因子(6.06%)>地形因子(5.40%),且郁闭度和草本盖度在估算中起主要作用;3)安化县杉木林林下草本可燃物载量范围为0~1.93 t/hm^(2),高可燃物载量的区域主要集中于冷市镇、梅城镇、清塘铺镇、高明乡。综上,随机森林模型可较好地应用于林下草本可燃物载量估算,为森林火灾的预防、管理提供科学依据。To study the estimation methods and spatial distribution of herbaceous fuel load in Chinese fir(Cunninghamia lanceolata)forests,this study utilized remote sensing data and field plot survey results.The random forest model was employed to estimate the herbaceous fuel load across 60 standard plots of Chinese fir forests in Anhua County,Hunan Province,and to analyze the importance of various factors in the estimation process.The results indicated that:1)The random forest model exhibited high accuracy in estimating herbaceous fuel load,with an out-of-bag(OOB)error of 0.1225 for the training set,a coefficient of determination(R^(2))of 0.9159 for the test set,and a root mean squared error(RMSE)of 0.0332 t/hm^(2).All metrics passed the confidence test at the 0.01 significance level.2)There were significant differences in the importance of the factors used in the estimation process.Stand factors ranked highest in importance(88.54%),followed by spectral characteristic index factors(6.06%)and terrain factors(5.40%).Canopy density and herbaceous cover played crucial roles in the estimation process.3)The herbaceous fuel load in Chinese fir forests in Anhua County ranged from 0 to 1.93 t/hm^(2),with areas of high fuel load concentrated primarily in Lengshi Town,Meicheng Town,Qingtangpu Town,and Gaoming Township.In conclusion,the random forest model was effectively applied to estimate herbaceous fuel load,providing a scientific basis for forest fire prevention and management.

关 键 词：随机森林 草本可燃物载量 林分因子 Sentinel-2A 森林火灾 

分 类 号：S762.1[农业科学—森林保护学] TP181[农业科学—林学]