marshes; and (3) potential nesting area analysis for the 
other parts of Hokkaido using the model developed 
during the first and second stages. This paper 
summarizes the preliminary results of the stage (1). 
2. STUDY AREA AND AVAILABLE DATA SETS 
The study area of the first stage encompasses ten 
1:25,000 scale topographic map sheets covering most 
of the Kushiro Wetland (Figure 1). These topographic 
maps were updated in 1992 and 1993. In addition to the 
annual crane census conducted in winter, another 
annual survey has been conducted by the Environment 
Agency of the Japanese Government for the number 
and location of nests and breeding pairs during 
breeding seasons. Identifying cranes and their nests 
during breeding seasons is very difficult from the 
ground because of tall reeds and alders in the wetlands. 
Nests are identified from helicopters or airplanes and 
their locations plotted on 1:50,000 scale topographic 
maps. The distribution of the original nest location 
maps is very limited to protect the habitat from human 
disturbances. This project employs nest location data 
re-plotted and provided on 1:200,000 scale maps by 
the Environment Agency. The positional accuracy of the 
nest data is not accurately known, but it is roughly 
estimated not to exceed 100 m on the ground (H. 
Masatomi, personal communication, 1996). Nest 
location maps have been compiled every year. In this 
study, only those compiled for 1994 was used. A total 
number of 39 nests were identified in the Kushiro 
Wetland in 1994. The quality and the completeness of 
the nest location maps may not be uniform every year 
due to different weather and ground conditions for nest 
observations. 
Vegetation data of high spatial resolution is considered 
important to develop a nesting site selection model 
because of its presumed relation to nesting conditions. 
But at the beginning of the project, the data was not 
available, and wetland areas delineated on 1:25,000 
topographic maps were used instead.Masatomi, et al. 
(1990) employed satellite images to analyze the 
vegetation composition around each existing nest and 
found large fluctuation between the composite ratios, 
which suggests the difficulty of modeling the vegetation 
conditions for nesting. 
Protected area maps were also employed to analyze 
the relationship between existing nesting sites and 
unprotected areas. 
3. GIS DATABASE 
The maps described in the previous section were 
digitized in vector format to make a GIS database for 
this project. The features digitized from the 1:25,000 
topographic maps include road and river networks, 
buildings and wetland areas. A tablet was employed to 
digitize the nest location points on 1:200,000 scale 
topographic maps. 
4. ANALYSIS 
Since no complete model for nesting site selection has 
been developed and is difficult to develop, we 
developed a simple model based on a rule of thumb. 
Wetland areas and rivers are considered favorable for 
nesting while roads and buildings should be 
unfavorable due to possible human interference and 
intrusion of predators. Each nest must also be spatially 
separated with each other to avoid mutual interference. 
Every road and river segment and every building were 
assumed to have equal influence to cranes and/or 
nesting conditions. In this preliminary analysis, the 
wetland area was assumed uniform, even though 
different vegetation types such as reed, sedge, alder 
and sphagnum are present. The 39 existing nest 
location data of 1994 was then used to measure the 
distance between the nearest ground features and 
nesting sites, and between adjacent nests. In order to 
avoid unusual or exceptional cases of nesting sites, 
10 % of the nests with extreme values were ignored for 
each ground feature. 
The measurement was done by changing the buffer 
size of ground features or existing nests. For example, 
the river network data was buffered to the extent that 
the buffered areas include 90 % of the existing nests. 
The resulting buffer size (90 % buffer size) was 265 m. 
Figures 2 shows the distribution of existing nests and 
river networks with 265 m buffer. Figure 3-5 show the 
results of the same analysis for road networks, 
buildings and adjacent nests with the 90 % buffer size 
of 104 m, 364 m and 1320 m, respectively. It should be 
noted for the case of adjacent nests that the 90 % buffer 
size is half of the distance from an adjacent nest. The 
relationship between the Kushiro Wetland area and 
existing nests found in 1994 is shown in Figure 6. 
The model for crane nesting thus derived has the 
following conditions: 
(1) Distance from the nearest river network is less 
than 265 m; 
(2) Distance from the nearest road network and the 
nearest building is larger than 104 m and 364 m, 
respectively; 
(3) Distance from the existing nests is larger than 2640 
m; and 
(4) Nest is located within the wetland area. 
5. RESULT 
This model is then employed to find potential nesting 
sites in the Kushiro Wetland. The result is shown in 
Figure 7, which indicates the following: 
576 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
sal 
res 
dei 
dos 
cor 
The 
tot: 
int 
are 
hig 
Sin 
unk 
em| 
usit 
isn 
stut 
Hov 
of tl 
ides 
eas 
alsc 
redi 
dev 
new 
(Ma: 
expl 
Scal 
wer 
The 
nee 
map 
acci 
topc