基于混合效应模型的西双版纳栎类林生物量遥感估测

刘子豪; 黄天宝; 欧光龙; 施凯泽

文章摘要

刘子豪,黄天宝,欧光龙,施凯泽.基于混合效应模型的西双版纳栎类林生物量遥感估测[J].林业调查规划,2026,51(2):17-26

基于混合效应模型的西双版纳栎类林生物量遥感估测

Remote Sensing Estimation of Oak Forest Biomass in XishuangbannaBased on Mixed Effects Model

DOI：

中文关键词: 栎类林森林生物量遥感估测混合效应模型森林异质性机器学习 Landsat 8 OLI 西双版纳

英文关键词: oak forests forest biomass remote sensing estimation mixed effect model forest heterogeneity machine learning Landsat 8 OLI Xishuangbanna

基金项目:

作者	单位
刘子豪	西南林业大学,云南昆明 650224
黄天宝	西南林业大学,云南昆明 650224
欧光龙	西南林业大学,云南昆明 650224
施凯泽	云南省林业调查规划院,云南昆明 650051

摘要点击次数: 347

全文下载次数: 24

中文摘要:

利用Landsat 8 OLI影像,基于小班面状数据引入林分信息、小班空间信息构建混合线性模型、多元线性回归模型、机器学习算法(支持向量机、随机森林、梯度提升树、K-近邻)对西双版纳州栎类林进行生物量估测。结果表明,引入龄组为随机效应的混合线性模型的R²为0.62,RMSE为31.52 t/hm²,相较于多元线性回归模型和机器学习算法具有较好的拟合效果和较宽的估测范围,在一定程度上减少了光学遥感在高森林生物量地区遥感估测中高值低估和低值高估带来的影响,改善了热带地区的遥感估测能力。在4种机器学习算法中,随机森林的估测精度高于其他3种模型,R²和RMSE分别为0.57、34.89 t/hm²。

英文摘要:

Landsat 8 OLI images were used to estimate the biomass of oak forest in Xishuangbanna by introducing stand information and spatial information of subcompartment based on the surface data to construct a linear mixed model, multiple linear regression model, and machine learning algorithms (support vector machine, random forest, gradient lift tree, and K-nearest neighbor). The results showed that the R² and RMSE of the linear mixed model with random age group were 0.61 and 31.52 t/hm², which had better fitting effect and wider estimation range than the multiple linear regression model and machine learning algorithm. To some extent, the influence of high value underestimation and low value overestimation in remote sensing estimation of high forest above-ground biomass area was reduced, and the estimation ability of tropical area was improved. Among the four machine learning algorithms, the estimation accuracy of random forest was higher than that of the other three models, R² and RMSE were 0.57 and 34.89 t/hm², respectively.

查看全文查看/发表评论下载PDF阅读器

关闭