i| 2004 
ESTIMATION ON TREE COVER PERCENTAGE USING TERRA/ASTER DATA 
patial WITH AIRBORNE LASER SCANNING DATA 
1gular 
tional 
ity of H. P. SATO ", R. TATEISHI " 
ble at 
ugust " Geography and Crustal Dynamics Research Center, Geographical Survey Institute, 
Tsukuba 305-0811, Japan 
Earth. * hsato@gsi.go.jp 
screte "" Center for Environmental Remote Sensing (CEReS), Chiba University, 
ilable Chiba 263-8522, Japan 
last 
and KEY WORDS: Forest, Canopy, Cover, Density, Lidar, Airborne laser scanning, ASTER 
shers 
ABSTRACT: 
iangle 
ctions It is expected that the remote sensing techniques will play a major role in establishing the carbon sink assessment system to meet the 
need of the Kyoto Protocol. Tree cover percentage is one of the useful parameters in estimating CO; absorption. Purpose of this 
space study is to investigate the correlation between tree cover percentage and 15-30 m resolution TERRA/Advanced Spaceborne Thermal 
emote Emission and Reflection radiometer (ASTER) optical sensor data. First, airborne laser scanning data — better known as airborne light 
detection and ranging (LIDAR) sensor data, are obtained in the forest of 1 km? in area, Japan. The forest as the study area is classified 
: An into five kinds of landform such as 1) crest slope, 2) head hollow, 3) slope facing east, 4) slope facing west and 5) valley. Tree cover 
  
ional percentage in cach landform is calculated by the LIDAR data. The tree cover percentage is overlapped on the TERRA/ASTER data, 
larch and the relation between them is investigated. As a result, an adequate correlation is estimated between the tree cover percentage and 
TERRA/ASTER data in the crest slope and the valley. In order to estimate global level tree cover percentage, there is the possibility to 
on of adopt this correlation to 1 km resolution SPOT/VEGETATION data. But more investigation will be needed in the future study. 
“lence 
1. INTRODUCTION 
faces 
s 3rd ; : Dn 
pore, The Kyoto Protocol was adopted in 1997 in Kyoto, at the third , 9n | 
nger, Conference of the Parties to the United Nations Framework | 
Convention on Climate Change (UNFCCC). In the Protocol Put 
eture Japan makes an effort to meet its 6% CO, emissions reduction . gat 
- 8" target. It is necessary to research and monitor carbon sinks such see 
ling, as forest to meet the target, and tree cover percentage is one of 
on. 3 the useful parameters to estimate the volume of CO, absorption. gi D 
view, et > 
Tree cover percentage is also important parameter to make N ba SME TY 1 
1992, global land cover classification map. The classification system, «p AT C 
i for example, Land Cover Classification System (LCCS) of Food ie d 
22. and Agricultural Organization (FAO), often needs tree cover SON 
based percentage. 
here. Figure 1. Study area 
| by Tree cover percentage is efficiently and objectively calculated 
through airborne laser scanning data — better known as airborne 
light detection and ranging (LIDAR) sensor data. It needs high 
costs to obtain the LIDAR data in the wide area, but the satellite 
optical sensor data is more reasonable and cover wider area than 
the LIDAR data. 
correlation between the tree cover percentage and the 15-30 m 
Purpose of this study is to investigate the 
resolution satellite sensor data. Furthermore, the possibility to 
estimate the tree cover percentage using 1 km resolution the 
Systeme pour l’Observation de la Terra (SPOT)/VEGETATION 
(VGT) data is described. 
2. STUDY AREA 
The study area that covers 1 km? in area is Hachioji, near Tokyo, 
Japan (Figure 1). The hilly area is extended in wide area around 
Tokyo, and study area is located in the typical hilly area. 
797 
Landform has an effect on the distribution ofthe trees in the hilly 
area. The study area was classified into five landforms such as 
crest slope, head hollow, slope facing east, slope facing west and 
valley (Figure 2). In the crest slope and the valley soil are dry 
and wet, respectively. The head hollow is susceptible to 
landslide. Hill slopes were classified into two types; slope 
facing east and slope facing west. This is the reason why these 
differences may appear in the satellite optical sensor data. The 
Table 1 shows the kinds of trees in each landform.