International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 Intei 
  
4. APPLICATION OF 3D VISUALIZATION OF | Figu 
SEDIMENT DISASTAR Figu 
| phot 
4.1 Heavy Rainfall Disaster in the Kyushu District in pasti 
Japan | view 
Between two days from 19 to 20 in July 2002, a heavy rainfall | 
attacked Kyushu district in Japan, which had caused many 
kinds of disasters in various areas (Ushiyama, 2004). Especially, | 
a sediment disaster in Minamata City, Kumamoto prefecture, The 
was largest in this season, and 19 civilians were dead due to this Mr. 
disaster. Basically, occurrences of a sediment disaster in Rese 
mountain areas depend on not only soil state, but also 3 Scie 
dimensional structures of land slopes. In this study, 3D 
visualization of landform, where the sediment disaster occurred, 
is created using 10m DEM and aerial photos. 
Ushi 
4.2 Integration of 10m DEM and Aerial Photos in th 
138 gi. 3 Soi m ; Soci 
A 1/25000 digital map by GSJ is adopted as reference data for 
Minamata City area. A precise digital map in local areas is not Hokl 
  
so popular such as ones in large cities, so a standard 1/25000 
digital map is used. 10m DEM including Minamata City is 
integrated with the reference data, which is shown in Figure 6. 
The aerial photo, taken on 22 in July. is shown in Figure 7, at 
which point the sediment disaster has occurred. 
Figure 4. IKONOS colour image 
©Space Imaging Co. 2002 
An affine transformation is performed for integration of 
IKONOS image with a digital map. There are several problems 
in combining a raster image and a vector image. Especially, it is 
difficult to select GCP in vector images correctly, because the 
images are symbolized, and elements in a digital map are not 
corresponding to a satellite image in pixel level (one-meter 
resolution). On the other hand, it is easy to select GCP in 
satellite images. Then, we have chosen central lines of roads as 
feature layer in GIS. We show a sample for selecting GCP in 
each image, using snapping effect. Thus, GCP for 
transformation are selected for 30 points over the image, 
equivalently. After several trials, the RMS error of 30 GCP is 
finally reduced to less than one pixel, which are comparatively 
good. The result of integration of IKONOS images with a 
digital map is shown as in Figure 5. 
  
  
Figure 5. 3D view of IKONOS image 
©Space Imaging Co. 2002 
  
Figure 7. Aerial photo at occurrence of the sediment disaster 
394