International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
3. RESULTS 
3.3 Image Pre-processing 
The satellite images were corrected for atmospheric effects. 
First, we carried out ortho-rectification by DEM, because the 
study area was uneven. 
Next, we reduced haze effects in the images. Equation (1) is the 
haze intensity of Tasseled Cap. 
Haze — 0.8460 TMI - 0.0730 TM2 - 0.46.1 TM3 (1) 
- 0.0032:TM4 - 0.049: 1TMS *- 0.01 19 TM7 * 0.7279 
where T Mi ^ digital numbers for band i in the original image 
The haze intensity, calculated by the equation (1), multiplied by 
1.88, 0.89, 1.02, 0.85, 1.40, and 0.71 were subtracted from the 
original image each band except the band 6. Equation (2) shows 
these calculations. 
TMI 1.88 
TM2 0.89 
TM3 1.02 
Dehazedlmages = | TM4 |—| 0.85 |x Hazelntensity (2) 
TMS 1.40 
TM6 0 
TM7 0.71 
where Haze Intensity = Equation (1) 
Next, we normalized topological effects. We did not use the 
existing Lambertian Reflection model or Minnaert method 
because we could not normalize them in this study area through 
the season, and then used methods judging shades by spectral 
characteristics. The shade intensity was judged by Equation (3), 
which is liner combination as well as Tasseled Cap. 
Shade — 0.085 TM * 0.078: TM2 * 0.119: 1TM3 (3) 
+ 0.5250TM4 + 0.6120 TM 5 0.2260 TMT7- 39.0 
This equation was obtained by reducing effects of tree species 
and soils. The shade effects was normalized by Equations (3) 
and (4). Figure 3 compares the original and processed images. 
TM!) (0.085 
TM2 0.078 
TM3 0.119 
Processed Image =| TM4 |—| 0.525 |x Shade Intensity (4) 
TMS 0.612 
TM6 0 
TM7j 10.236 
3.4 Estimating Forest Volume 
We used a timber volume, the whole volume, and its biomass. 
The timber volume data were estimated by Tokyo Metropolitan 
Government. The whole volume and its biomass were 
calculated with expansion ratios and volume density determined 
by each vegetation types. 
Next, we carried out mixed pixel decomposition. We used 
Equations (5), (6), and (7), which are a liner mixed-pixel model. 
  
^ Water 
  
o | NIR (TM4) reflectance 
  
VIS (TM3) reflectance 
Figure 2. Feature of VIS and NIR 
Figure 2 displays a distribution of VIS and NIR retlectance. 
VIS = à Il yd 1235 + 130 W (5) 
NIR = A 3; V+ a 22 S d 23 W (6) 
V+S+HW=1 (7) 
where d ;,and d »,- reflectance of vegetation 
à ,;and d ;;- reflectance of soil 
a ;;and a ,;= reflectance of water 
The land cover ratio for vegetation, soil, and water in a pixel 
was calculated by combination of these equations respectively. 
  
Figure 3. Original image (left) and processed image (right) 
852 
nh 
3 
Pixel Value 
Fig 
We 
rati 
dig 
bec 
regi 
sate 
the 
. nun 
coe 
Tab 
voli 
dete 
ban 
rati 
arez 
Wwe: 
and 
dete 
coe! 
area 
for 
E 
Tab