Brad Weigle, Photo Science, bweigle@photoscience.com (Presenter)
Dan Casey, Photo Science, dcasey@photoscience.com


Three main components in a comprehensive vegetation map are vegetative units (species and species complexes), vegetation size, and canopy cover. Using remote sensing techniques, vegetative units can be reliably derived from spectral, topographic, and environmental data layers. While LiDAR is very effective for measuring tree height and canopy cover, it is not yet affordable for many organizations. Until the advent of LiDAR, vegetation size and canopy cover classes were predicted from data mining and Classification and Regression Tree (CART) analysis of data layers using software such as See5 and Random Forest instead of utilizing direct measurements. As a lower cost alternative to LiDAR, we tested commercial automated terrain extraction (ATE) software to determine whether derived vegetation heights are comparable to LiDAR measurements. For our forested study area in eastern Kentucky, three DEMs were available: 3m NED, 10m NED and 30m SRTM. Our investigation showed that ATE software could determine the canopy height within an average of 3m of LiDAR height data when an existing high resolution DEM is available since creation of a usable DEM from ATE software still requires too much manual editing to be cost effective in forested areas. Test results and processing recommendations will be discussed along with additional tests conducted in two Idaho forests where field-monitoring sites were available.