The paper describes a methodology for tree species classificationusing features that are derived from small?footprint full waveformLight Detection and Ranging (LIDAR) data. First, 3?dimensional coordinatesof the laser beam reflections, the intensity, and the pulse widthare extracted by a waveform decomposition, which fits a series ofGaussian pulses to the waveform. Since multiple reflections are detected,and even overlapping pulse reflections are distinguished, a muchhigher point density is achieved compared to the conventional first/last?pulsetechnique. Secondly, tree crowns are delineated from the canopy heightmodel (CHM) using the watershed algorithm. The CHM posts are equallyspaced and robustly interpolated from the highest reflections inthe canopy. Thirdly, tree features computed from the 3?dimensionalcoordinates of the reflections, the intensity and the pulse widthare used to detect coniferous and deciduous trees by an unsupervisedclassification. The methodology is applied to datasets that havebeen captured with the TopEye MK II scanner and the Riegl LMS?Q560scanner in the Bavarian Forest National Park in leaf?on and leaf?offconditions for Norway spruces, European beeches and Sycamore maples.The classification, which groups the data into two clusters (coniferous,deciduous), leads in the best case to an overall accuracy of 85%in a leaf?on situation and 96% in a leaf?off situation.     «

The paper describes a methodology for tree species classificationusing features that are derived from small?footprint full waveformLight Detection and Ranging (LIDAR) data. First, 3?dimensional coordinatesof the laser beam reflections, the intensity, and the pulse widthare extracted by a waveform decomposition, which fits a series ofGaussian pulses to the waveform. Since multiple reflections are detected,and even overlapping pulse reflections are distinguished, a muchhigher point density is achi...     »