There is an increasing demand for urban vegetation mapping, and airborne laser\udscanning (ALS) has the unique ability to provide geo-referenced three-dimensional\uddata useful for mapping of surface features. This thesis examines the ability of\udfull-waveform and discrete return ALS point data to distinguish urban surface\udfeatures, and represent the three-dimensional attributes of vegetation at\uddifferent scales in a vector-based GIS environment. Two full-waveform datasets,\udat a wavelength of 1550 nm, and a discrete return dataset, at 1064 nm, are used.\udPoints extracted from the first full-waveform dataset are classified with k-means\udclustering and decision tree into vegetation, buildings and roads, based on the\udattributes of individual points and the relationships between neighbouring points.\udA decision tree is shown to perform significantly better (74.62%) than k-means\udclustering (51.59%) based on the overall accuracies. Grass and paved areas could\udbe distinguished better using intensity from discrete return data than amplitude\udfrom full-waveform data, both values proportional to the strength of the return\udsignal. The differences in the signatures of surfaces could be related to the\udwavelengths of the lasers, and need to be explored further. Calibration of\udintensity is currently possible only with full-waveform data. When the decision\udtree is applied on the second full-waveform dataset, the backscatter coefficient\udproves to be a more useful attribute than amplitude, pointing to the need for\udcalibration if a classification method using intensity is to be applied on datasets\udwith different scanning geometries. A vector-based approach for delineating tree\udcrowns is developed and implemented at three scales. The first scale provides a\udgood estimation of the tree crown area and structure, suitable for estimating\udbiomass and canopy gaps. The third scale identifies the number of trees and their\udlocations and can be used for modelling individual trees.
展开▼
