In order to evaluate the biomass of mangrove based on remote sensing, the biomass model of mangrove is estimated with the multiple regression analysis, by extracting spectral information and textural features from TM images,and combining the field survey biomass data. The results show that the accuracy order of biomass estimation model is Rhizophora stylosa> Bruguiera gymnoihiza>Vicennia mariana>Aegiceras corniculatum>mixed species,and thatKandelia candel cannot be modeled. During the process of multiple regression analysis, the effect point is removed by ellipsoidal multivariate trimming and the multicollinearity is e-liminated by cluster analysis and factor analysis. All the biomass estimation models got through relevant test and can be applied on the biomass estimation of mangrove.%采用广西海南两地2004年红树林TM影像和红树林群落样地调查数据,应用多元逐步回归分析方法,选取影像光谱特征及纹理特征等作为自变量,对红海榄(Rhizophora stylosa)、木榄(Bruguiera gymnoihiza)、白骨壤(Vicenna mariana)、桐花树(Aegiceras corniculatum)、秋茄(Kandelia candel)以及混合(不分树种)红树林生物量遥感估算建模,研究红树林生物量的遥感估算方法.结果,生物量估算模型拟合效果红海榄>木榄>白骨壤>桐花树>混合,秋茄模型无法拟合.回归过程中采用稳健诊断方法去除影响点,并应用聚类分析和因子分析的方法排除多重共线性.得到的生物量估算模型通过相关检验,可以高效、快速地进行红树林生物量的估测.
展开▼