International Archives of Photogrammetry and Hemote Sensing, Vol. 32, Part 3W14, La Jolla, CA, 9-11 Nov. 1999
LASER-SCANNING FOR THE DERIVATION OF FOREST STAND PARAMETERS
W. Rieger!, O. Eckmiillner?, H. Müllner!, T. Reiter
|: Institute for Surveying, Remote Sensing, and Land Information
University of Agricultural Sciences, Vienna
2: Institute for Forest Growth and Yield
University of Agricultural Sciences, Vienna
3: Institute for Photogrammetry and Remote Sensing
Vienna University of Technology
Austria
Correspondence: rieger@edv1.boku.ac.at
Commission III, WG 5 and WG 2
KEY WORDS: Laser scanner, DEM, Forest inventory, Forest stand.
ABSTRACT
Forest stand parameters are derived from airborne laser scanner data. Data from three laser scanner flights by the company "TopoSys" are
used, a winter flight with last pulse recorded and two summer flights, with last and first pulse recorded respectively. The laser data are
reduced to relative (ground) heights with a high quality ground elevation model (DEM) that was interpolated from the winter data. The
proportion of crown coverage of coniferous trees are estimated from data for 50 stands. Stand heights and basal area proportion of
coniferous versus deciduous trees are estimated through regression analysis against terrestrial reference data from 110 angle count points,
each with approximatly 10 trees. The results are very promising with regression coefficients of 0.8 to 0.95 for all measures. The usual
systematic under-estimation of tree heights cannot be proved due to the high sample point density. The mean errors of the tree heights are
in the range of £1.8 — +2.7 m. Basal and projected crown area proportions can be estimated with similar precision. Summer last and first
pulse elevations differ insignificantly, suggesting there is no need to have both flights. Only one winter flight is necessary.
KURZFASSUNG
Forstbestandesparameter werden aus den Daten eines flugzeugbasierten Laser-Scanners abgeleitet. Insgesamt stehen drei Fliige mit dem
System der Firma "TopoSys" zur Verfügung: Ein Winterflug mit letztem empfangenen Signal und zwei Sommerflüge mit erstem bzw.
letztem Signal. Die Laserhóhen werden um ein aus den Winterdaten abgeleitetes Bodenmodell reduziert. Aus diesen Daten werden
Schirmflüchenanteile für Nadelbiume im Gegensatz zu Laubbüumen für 50 Referenzbestünde abgeleitet. Anhand von 110 Winkelzähl-
proben mit je ca. 10 aufgenommenen Báumen werden Bestandeshóhen sowie Grundflüchenanteile über eine Regressionsanalyse ge-
schützt. Für die Baumhóhen ergeben sich Regressionskoeffizienten von 0,8-0,95. Die üblicherweise beobachtete systematische Unter-
schiitzung der Baumhóhen kann auf Grund der hohen Datendichte nicht verifiziert werden; die mittleren Fehler der Höhen liegen bei
+1,8 - £2,7 m. Grund- und Schirmflüchenanteile ergeben sich mit ähnlicher Qualität. Zwischen den Datensätzen mit den ersten und
letzten aufgezeichneten Pulsen der Sommerflüge besteht kein signifikanter Unterschied, sodaß auf einen der beiden Flüge verzichtet
werden kann. Der Winterflug muß nur einmal erfolgen.
1 INTRODUCTION until today. In some cases Laser scanning is an serious
competition with DEM due to some benefits that overcome major
Data of airborne laser scanners include much information about problems of stereo-photogrammetry, which are;
the ground coverage, i. e. vegetation, off-terrain objects, and the
similar. This information is provided through the distribution e the small laser dots (ground diameter approx. 20-25 cm)
pattern of the reflected laser dots, mainly through their elevation may penetrate vegetation even through small holes.
distribution, but also through their plane positions. e the usage of an active scanning system allows data even in
the shadowed ground areas of dense forests.
In some way laser scanning is similar to photogrammetry. It has e only one ray is needed in order to collect an object point
been mainly used for the derivation of digital elevation models so compared to the two rays needed for stereo-
far, a field which has been one of the domains of photogrammetry