Realising the potential of airborne LiDAR data in high quality DEM generation : tests in applications to a catchment management region

Digital elevation models (DEMs) are becoming increasingly important components in national and regional spatial data infrastructure. High-quality DEMs can now be derived directly from airborne light detection and ranging (LiDAR) point-cloud data of high spatial density if the derivation process can be verified. However, LiDAR is relatively new compared with other technologies for terrain data collection, and, although offering the potential for providing better spatial resolution than those that have been routinely available before, will not diffuse among DEM users until the results of meeting the verification challenge are favourable enough to inspire re-organisation of spatial data in decision support for catchment management and other third-tier-of-government authorities. By way of exemplification, the research presented in this thesis concerned ways of improving the processing of the airborne LiDAR data for high-quality DEM generation in terms of both accuracy and efficiency, and explored the applications of LiDAR-derived DEMs in the region of the Corangamite Catchment Management Authority, Victoria, Australia. This thesis begins with a review of the traditional technologies for terrain data collection and DEM generation and compares them with the LiDAR technology. Accordingly, a review of the recently-reported advances in LiDAR data deployment for DEM generation is followed by reports of experiments designed to improve selection and deployment of LiDAR data filtering, modelling methods and data reduction, and the achievement of vertical accuracy for different land covers. Also reported are results of deployment of LiDAR data for ground truthing, and application of LiDAR data for the extraction of drainage networks on an area of deranged drainage: the Victorian Volcanic Plain. The show that: (a) the issues of filtering, modelling techniques, interpolation methods, DEM resolution, and data reduction are critical and must be considered carefully when using LiDAR data for a high-quality DEM generation; (b) it is efficient to use survey marks for the accuracy assessment of LiDAR data. Normal distribution must be tested in order to select a suitable measure for the accuracy assessment of LiDAR data over different land covers; (c) LiDAR data reduction can improve the terrain production efficiency without compromising the product quality. The deployment of breaklines made a significant contribution to improving the accuracy of terrain models while allowing for data reduction; (d) it demonstrated the practical feasibility of applying ground control points from LiDAR intensity image and LiDAR-derived DEM in image orthorectification. The resultant orthoimage accuracy was shown to be superior to that achieved by using (lower accuracy) data sources such as those from Vicmap data; and (e) the LiDAR-derived DEM offers the capability of extracting and delineating the drainage networks in much more detail in low¬relief terrain, including areas in which drainage is barely coherent; The advantages of using LiDAR-derived DEM over the lower-accuracy DEM emerge in terms of stream order, stream number and stream length.