Modelling Biodiversity and Forest Structure Using Hyperspectral and LIDAR Remote Sensing

Abstract

"Understanding patterns of biodiversity and the linkage to forest structure is a key issue in ecological research and conservation in complex forest ecosystems. Spatially comprehensive assessments of biodiversity and forest structure require the use of remote sensing techniques. Airborne hyperspectral and LIDAR imagery promises to provide the required accuracy and information content to address these issues. We investigate plant biodiversity and forest structure in different layers of a mixed temperate forest in the Bavarian Forest National Park, Germany, using airborne HyMap and LIDAR data. The data was acquired in 2009; ground information was collected in 102 inventory plots. Data analysis was conducted by means of the Random Forest regression tree algorithm. We list a collection of biodiversity and forest structure metrics that could be derived successfully. Furthermore, we present the driving factors that govern the predictability of these metrics. The latter are subsequently used to predict the spatial patterns across the whole study area."