International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
WAVEFORM ANALYSIS FOR THE EXTRACTION OF POST-FIRE VEGETATION 
CHARACTERISTICS 
F. Pirotti® *, A. Guarnieri“, A. Vettore® 
* CIRGEO, Interdepartmental Center for Research in Geomatics, 35020 Legnaro (PD) - Italy - (francesco.pirotti, 
alberto.guarnieri, antonio. vettore)@unipd.it 
Commission VII, WG VI/4 
KEY WORDS: Full Waveform laser scanner, vegetation height distribution, airborne lidar survey 
ABSTRACT: 
Full-waveform is becoming increasingly available in today's LiDAR systems and the analysis of the full return signal can provide 
additional information on the reflecting surfaces. In this paper we present the results of an assessment on full-waveform analysis, as 
opposed to the more classic discrete return analysis, for discerning vegetation cover classes related to post-fire renovation. In the 
spring of 2011 an OPTECH ALTM sensor was used to survey an Alpine area of almost 20 km? in the north of Italy. A forest fire 
event several years ago burned large patches of vegetation for a total of about 1.5 km? . The renovation process in the area is varied 
because of the different interventions ranging from no intervention to the application of re-forestation techniques to accelerate the 
process of re-establishing protection forest. 
The LiDAR data was used to divide the study site into areas with different conditions in terms of re-establishment of the natural 
vegetation condition. The LiDAR survey provided both the full-waveform data in Optech’s CSD+DGT (corrected sensor data) and 
NDF+IDX (digitizer data with index file) formats, and the discrete return in the LAS format. The method applied to the full-waveform 
uses canopy volume profiles obtained by modelling, whereas the method applied to discrete return uses point geometry and density 
indexes. The results of these two methods are assessed by ground truth obtained from sampling and comparison shows that the 
added information from the full-waveform does give a significant better discrimination of the vegetation cover classes. 
1. INTRODUCTION Ranges from multiple returns are recorded depending on the 
sensor; two echoes (first and last) or multiple echoes (up to 
Airborne LiDAR (or Airborne Laser Scanning - ALS) in the past eight) can be recorded. Each sensor has a range resolution which 
ten years has seen rapid growth in both sensor technology and describes the minimum distance required between objects, for 
fields of application. Research on laser profilers started in the separating two return echoes. This measure depends on the 
late nineties and a lot of interest was given to the forestry outgoing pulse duration and the group velocity, some ALS 
environment because of the ability of the laser pulse to penetrate systems have a dead zone of up to 3 m (Wagner et al., 2008). 
canopy, returning ground hits, which are precious for digital Discrete-return sensors process the return echo waveform using 
terrain models (DTMs) extraction (Carson et al., 2004). Studies fast yet simple algorithms like the Constant Fraction 
over the rate of penetration show that typical coniferous and Discriminator criteria for defining a threshold for the 
deciduous forests allow 20-40% of the laser pulses to return identification of significant energy peaks (Thiel and Wehr, 
ground hits in leaf-on conditions and as much as 70% in leaf-off 2006). 
conditions (Ackermann, 1999). This can occur because the size 
of the diffraction cone (Mallet and Bretar, 2009) can vary froma Off-line processing of the waveform requires recording the full 
few centimeters up to one/two meters; thus the canopy area that waveform of the return echo by means of a digitizer which 
is illuminated is large enough to have a significant amount of samples the return energy at certain time intervals usually from 
gaps which allow part of the laser energy to pass without getting ^ one to several nanoseconds. To this day a significant amount of 
reflected by leaves of branches, all the way to the ground, which tests have been conducted on waveform data from spaceborne 
is the element which causes the last reflection. and airborne sensors. NASA's GLAS sensor, mounted on 
Discrete-return ALS systems provide data as 3D coordinates IceSAT satellite provided freely downloadable full waveform 
and intensity, commonly referred to as “point cloud”, by on-line data for several years before recent mission dismissal. The main 
processing of the retum laser echo. If one laser pulse gets objective was the estimation of the thickness and dynamics of 
reflected by different surfaces along its path, interactions of the the polar ice-sheet (Zwally et al, 2002), but its large-footprint 
laser with the objects cause a return signal whose characteristics (^65 m) waveform information was also used also for extracting 
are a mixture derived from the different optical properties of the vegetation structure (Harding and Carabajal, 2005), ground 
objects, the range and the incidence angle (Wagner et al., 2006). ^ cover classification (Pirotti, 2010) as well as direct canopy 
  
* Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author. 
523