基于高分六号的南昌市植被地上生物量遥感估算

Remote Sensing Estimation of Vegetation Above-Ground Biomass in Nachang Based on GF-6 Image

结合高分六号(GF-6)遥感数据对植被的光谱特征和纹理特征等参数进行提取,对于实地测量的树高和胸径等数据,使用异速生长方程将其转化为地上生物量的观测值。分别采用随机森林(random forest,RF)、支持向量机(sup⁃port vector machine,SVM)和K近邻算法(K-nearest neigh⁃bor,K-NN),结合实地采样数据构建植被地上生物量遥感反演模型,并对模型进行拟合效果检验和精度验证。结果表明,在3种机器模型中,随机森林模型的数据结果最优,其决定系数为0.6638,均方根误差为28.13。最终选择随机森林模型对南昌市植被地上生物量进行估算,得到南昌市的植被生物量分布图,为后续研究城市生物量遥感估算和城市生态研究供科学依据。

This paper firstly extracts spectral characteristics and texture characteristics of vegetation with GF-6 remote sensing data,then convert the tree height and DBH measured in the field into the observed values of aboveground biomass through the allometric growth equation.Secondly,the re⁃mote sensing inversion model of vegetation aboveground bio⁃mass is constructed through random forest,support vector ma⁃chine and K-nearest neighbor algorithm and incorporation of sampling data,.Thirdly,the fitting effect and accuracy of the model are tested.The results show that among the three ma⁃chine models,the random forest model has the best data re⁃sult,with its determination coefficient 66.38 and the root mean square error 28.13.Finally,the random forest model is selected to estimate the aboveground biomass of vegetation in Nanchang city,and the vegetation biomass distribution map of Nanchang city is obtained,which provides scientific basis for the follow-up research of urban biomass remote sensing es⁃timation and urban ecological research.