期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2009
卷号：XXXVIII-3/W8
页码：135-140
出版社：Copernicus Publications
摘要：In the present study, the objective was to compare the accuracy of low-pulse airborne laser scanning (ALS), high-resolution noninterferometric TerraSAR-X (TSX) radar data and their combined feature set in the estimation of forest variables at the plot level. The variables studied included mean volume, basal area, mean height and mean diameter. Feature selection was based on a genetic algorithm (GA). The nonparametric k-nearest neighbour (k-NN) algorithm was applied to derive the estimates. The research material consisted of 125 tree level measured circular plots located in the vicinity of Espoo, Finland. The relative RMSEs for ALS were 30.6%, 29.4%, 12.1% and 17.5% for mean volume, basal area, mean height and mean diameter, respectively. For TSX thes e were 47.4%, 39.3%, 20.3% and 22.4%, and for the combined feature set 29.5%, 29.0%, 12.6% and 17.0%. The accuracies of ALS-based estimations were higher in all forest variables. The best performing combined feature set obtained by GA contained 15 features, 10 of them originating from the ALS data. The combined feature set outperformed the ALS feature set only slightly. However, due to its favourable temporal resolution, satellite-borne radar imaging is a promising data source for updating large-area forest inventories performed by low-pulse ALS inventory
