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：2011
卷号：XXXVIII - 4/W19
页码：159-164
DOI：10.5194/isprsarchives-XXXVIII-4-W19-159-2011
出版社：Copernicus Publications
摘要：Understory trees in multi-layer stands are often ignored in forest inventories. Information about them would benefit silviculture, wood procurement and biodiversity management. Cost-efficient inventory methods for the assessment of the presence, density, species- and size-distributions are called for. LiDAR remote sensing is a promising addition to field work. Unlike in passive image data, in which the signals from multiple layers mix, the 3D position of each hot-spot reflection is known in LiDAR data. The overstory however prevents from obtaining a wall-to-wall sample of understory, and measurements are subject to transmission losses. Discriminating between the crowns of dominant and suppressed trees can also be challenging. We examined the potential of LiDAR for the mapping of the understory trees in Scots pine stands (62°N, 24°E), using carefully georeferenced reference data and several LiDAR data sets. We present results that highlight differences in echo-triggering between sensors that affect the near-ground height data. A conceptual model for the transmission losses in the overstory was created and formulated into simple compensation models that reduced the intensity variation in second- and third return data. The task is highly ill-posed in discrete-return LiDAR data, and our models employed the geometry of the overstory as well as the intensity of previous returns. We showed that even first-return data in the understory is subject to losses in the overstory that did not trigger an echo. Even with compensation of the losses, the intensity data was deemed of low value in species discrimination. Area-based LiDAR height metrics that were derived from the data belonging to the crown volume of the understory showed reasonable correlation with the density and mean height of the understory trees. Assessment of the species seems out of reach in discrete-return LiDAR data, which is a drastic drawback
