International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
The data from these entities had the following characteristics. 
  
Management | Characteristics 
entity 
  
Complies with national standards and has geodetic 
Nation coordinates. 
Specific features are expressed with polygons. 
  
Has geodetic coordinates. 
Prefecture * All the features are expressed with lines. 
Some data is rasterized. 
  
  
Has geodetic coordinates. 
City * All the features are expressed with lines. 
* Some data is rasterized. 
  
  
  
  
Table 2. Characteristics of as-built drawings by road 
management entity 
Not all the as-built drawings of the prefecture and 
municipalities were available as vector data. We excluded 
rasterized data for the verification. 
Regarding the three types of as-built drawings above, we 
examined whether they had adequate location accuracy to be 
used for road data updating. The drawings of the prefectural 
and city roads did not have geodetic coordinates. We performed 
the orientation of the drawings and compared the coordinates 
on the drawings with those measured by field survey. As the 
table below shows, the drawings had location accuracy 
equivalent to or better than their scales. 
  
  
  
  
  
  
  
  
  
  
Number of 
Verified drawing verified Scale RMSE 
points 
Co-1 6 1:1000 0.422 
Nation Co-2 9 1:1000 0.566 
Co-3 6 1:1000 0.635 
P-1 4 1:1000 0.87 
Prefecture P-2 6 1:500 0.553 
P-3 4 1:500 0.231 
Ct-1 3 1:300 0.064 
Municipality Ct-2 7 1:500 0.076 
Ct-3 4 1:500 0.057 
  
  
  
  
  
  
  
Table 3. Evaluation result of location accuracy of as-built 
drawings 
Using the data above, we updated road data of a spatial data 
infrastructure and found the following. 
* There were no problems with data updating from the 
perspective of location accuracy. 
+ The procedures for drawing road shapes were not 
standardized and the original data and updated data 
sometimes differed in the positions of road boundaries. 
In suburbs in particular, some road boundaries were 
not clear. Some standards needed to be established. 
+ Data created by prefectures and municipalities did not 
have absolute positional coordinates. From the 
viewpoint of working efficiency, it would be helpful if 
reference information for identifying the area to be 
updated were available in advance. 
30 
  
Figure 7. Difference in road boundaries 
5.2 Updating of building data 
To update building data, we also conducted a field study on 
Mie Prefecture. 
We asked the fixed asset departments of three municipalities 
how they created their house ledgers. We found that they paid 
little attention to absolute location accuracy. To use the house 
ledgers, it bears keeping in mind that the ledgers vary in 
location accuracy from municipality to municipality depending 
on the updating method that is applied. In some municipalities, 
the staff draws building shapes by hand. 
House ledgers are increasingly digitized. These ledgers can be 
used to update building data without affecting efficiency. 
+ Scale: 1:500 or 1:1000 (location accuracy unknown) 
+ Scope: Flat land (area in which houses stand) 
+ Number of houses updated each year: 1% to 2% of 
the total 
¢ Updating frequency: Every year 
We verified the location accuracy of the house ledgers. We 
identified the locations of the houses on the GIS, compared 
them with those measured by field survey and calculated 
RMSE. 
The result shows that no data has location accuracy as high as 
the scales of the house ledgers. However, their accuracy is 
equivalent to a scale of 1:2500, which is good enough to be 
used for the updating. 
  
  
  
  
  
: Number of 
City verified et Scale RMSE 
City A 36 1:1000 1.074 
City B 42 1:1000 0.824 
City C 30 1:500 1.086 
  
  
  
  
  
Table 4. Verification result of location accuracy of house 
ledgers 
As we performed the updating, we discovered the following 
with regard to the method of partially updating building data. 
* Identifying the area to be updated in the old data 
required that all newly constructed, lost or altered 
houses be located by comparing the house ledgers for 
two years and then all the buildings on the old data 
that were overlapped by houses in those locations had 
to be selected to avoid omissions. 
+ When we checked the data selected in identifying the 
area to be updated, we found that excess data was 
included because of subtle differences in shape, as 
shown below. Visual checks always needed to be 
conducted to remove that excess data. 
+ When building data is integrated with road data, some 
  
Building 
  
6.2 
Fo 
que 
frei 
op: 
qu: 
nec 
sta 
sol 
anc 
dis 
dat 
Spa 
the 
pla 
be