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
FULL WAVEFORM LIDAR EXPLOITATION TECHNIQUE AND ITS EVALUATION IN
THE MIXED FOREST HILLY REGION
S. Chhatkuli ^ *, K. Mano?, T. Kogure ®, K. Tachibana ® H. Shimamura ®
* PASCO CORPORATION, R & D Center, 2-8-11 Higashiyama, Meguro Ku, Tokyo, Japan (cshahb7460, teorsu8478,
kainka9209, hairdu9189)@pasco.co.jp
® PASCO CORPORATION, MMS Division, 1-2-3 Minatomachi, Naniwa-ku, Osaka, Japan, koonua2121@pasco.co.jp
Commission VII, Theme Session VII/7, III/2, V/3
KEY WORDS: LIDAR, Analysis, DEM/DTM, Point Cloud, Accuracy
ABSTRACT:
In this paper a full waveform exploitation technique and its evaluation in the mixed forest hilly region is presented. The increment in
ground penetration by using the full waveform exploitation technique compared to the discrete LiDAR pulses during autumn and
winter season is evaluated. The results showed that the technique used for the full waveform exploitation has effectively increased
the ground penetration by 50 % and 20 %, respectively, during autumn and winter in the mixed forest hilly region compared to the
discrete return pulses. The accuracy test of the LiDAR derived terrain model constructed from the discrete LiDAR pulses and full
waveform LiDAR pulses obtained during autumn and winter has also been performed. The RMSE of the LiDAR derived DTM with
Im grid size constructed from the discrete LIDAR pulses obtained during autumn and winter were 0.73 m and 0.22 m respectively.
Likewise, the RMSE of the LiDAR derived DTM constructed from the full waveform LiDAR pulses obtained during autumn and
winter were 0.59 m and 0.21 m respectively. The results also showed that by using full waveform return pulses, DTM constructed
for both seasons were improved compared to the DTM generated from discrete LiDAR pulses.
1. INTRODUCTION This paper presents a robust full waveform exploitation
technique which can effectively increase the ground penetration
Recently, LiDAR systems that digitize and record entire echo —— and vegetation detection even during leaf on condition. For the
waveform are increasingly available. By suitably exploiting the practical application of the full waveform exploitation
full waveform return pulses, very weak or buried signals that ^ technique, objective analysis of total increment in ground
were not recorded as discrete return pulses could also be detection and the validation of its accuracy are very important.
detected. Hence, the accuracy evaluation of the ground penetration from
the proposed full waveform exploitation technique and discrete
The data collected as discrete return pulses during leaf-on ^ return pulses is also carried out. The total increment in ground
condition may be ideal for determining canopy elevation but ^ penetration by using the proposed full waveform exploitation
not ground elevation. Conversely, discrete laser pulses during technique against the discrete return pulses has been analyzed
leaf-off condition may not provide good estimate of canopy for the mixed forest hilly region. The analysis is performed for
height (Hodgson et.al, 2003). autumn (September) and winter (December) to investigate the
changes in ground penetration due to the seasonal variation in
Chauve et.al, (2008) performed the processing of full the forest condition as well.
waveform LiDAR data to extract forest parameters and DTM.
They found that detecting weak echoes, when processing full
waveform data, enabled to better describe the canopy shape and 2. MATERIALS AND METHOD
penetrate deeper into the forest cover compared to discrete
return pulses. However, DTM was not significantly improved. LiDAR survey was performed in the Ogawa region Japan,
Reitberger et.al., (2008) found that the tree shape is better which is located about 150 km north of Tokyo, on September
detected by the full waveform data compared to discrete return 29 and December 5, 2010. Figure 1 shows the orthophoto of the
pulses and it could also detect small trees in the lower forest area that has been demarcated for the analysis purpose. The
layer. Pirotti (2011) presented the review of existing methods demarcated area is about 2km in length and 0.3km in width.
for processing full waveform LiDAR data for the estimation of The study area is a mixed forest hilly region containing patches
forest parameters. Depending on the full waveform exploitation of evergreen trees (E) like Japanese red pine trees, Japanese
technique, certain models were applicable at either the plot or cedar trees, Japanese cypress trees and deciduous trees (D) like
regional level, and some methods were influenced by leaf- Japanese beech trees and Japanese chestnut trees among other
off/leaf-on condition. varieties.
* Corresponding author.
505