Individual tree mapping and classification using high density airborne lidar
Files
(Published version)
Date
2011
Authors
Bruce, D.A.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Conference paper
Citation
Sensors and Models in Photogrammetry and Remote Sensing, SMPR 2011, 2011, pp.1-9
Statement of Responsibility
Conference Name
SMPR 2011 (18 May 2011 - 19 May 2011 : Iran)
Abstract
Modern airborne lidar scanners provide high density three dimensional points with high accuracy and precision. Lidar points can be used for precise measurement and mapping of land cover and features. This paper examines the potential of lidar data in tree classification and mapping at the individual tree level. The data was acquired in May 2010 by Airborne Research Australia, using a Regel LMS-Q560 and covered most of the Adelaide City Council area in Adelaide, South Australia. The test area was the Adelaide southern park lands, Park 17 in particular. Two filtering methods were used to extract the required lidar data and classify the laser returns (1 to 4) into classes. Lidar returns were used to create terrain and surface digital elevation models (DEM), from which a Crown Height Model (CHM) was created. The spacing of the canopy lidar returns was used to create a Density of High Points model (DHP). CHM and DHP were used to delineate tree canopy boundaries using two methods, inverse watershed delineation (IWD) and focal statistics. The accuracy of IWD was limited and was largely affected by the size of the trees in the study area. Several parameters are required for running IWD with each of them affecting the output size of the tree canopies. Focal statistic applies to the density of lidar point showed better positioning of tree centres. The accuracy of separating adjacent trees varied depending on their properties.