)12 
arguments 
the data 
data and 
performed 
additional 
the whole 
r hand the 
"xpensive. 
ore effort. 
> used for 
pping, to 
(periences 
y control 
following 
f features 
se aspects 
hecks are 
ass levels. 
nsistency, 
\ software 
topology. 
etry errors 
mity and 
corrected 
the error. 
e data and 
ich means 
rrors. The 
ed to be 
he source 
e captured 
juisites. À 
; 1) if the 
features in 
jbutes are 
> captured 
nt, line or 
nsity, 5) if 
c location 
rol, that is 
ne defined 
y %5 to 10 
nately %5 
d. 
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 
C HAM; HFl | NFN 
Accurde UNK' 'N A NA ‘kurudere 
  
Show: AM i Selected Records (1 out af 752 Selected) 
    
    
j Figure 2. uncaptured river La 
The second level is the control of vector data over the 
previously produced raster or vector maps. À control stuff 
checks all the captured data and looks at; 1) if there is a feature 
which exists on the old data but not in the newly produced 
data, 2) if the captured features are compatible in the old and 
new data, 3) if the assigned attributes are compatible according 
to the old data. Here the found error conditions can be not an 
error. These are only suspicious conditions that should be 
checked over the source and ancillary data which are used for 
data capturing. In Figure 3, a condition found is shown. In that 
condition, the extent of the captured river and the extent of the 
same river over the old topographic map are different and 
should be checked. This control takes approximately 965 of the 
production time of data capture and approximately 962 to 3 
additional conditions for the captured data are detected. 
Approximately %50 of the conditions is real errors. 
  
  
  
Show: All i Selected Records (1 aut af 752 Selected) 
  
  
M 
Weis Pes ; ho E ol 
Figure 3. An uncaptured river part 
At the third level the printouts of the vector data are taken with 
the appropriate scale and controlled according to the logical 
consistency. An experienced control stuff checks out the 
printouts for logical inconsistencies such as a wet river flowing 
through a dry river, a hard paved road connected to a loose 
unpaved cart track, a building over an empty area with no road 
connection to that house etc. The examples for this control can 
be extended. As the second level control, again the found error 
conditions can be not an error and should be checked over the 
source and ancillary data which are used for data capturing. In 
25 
Figure 4 such a condition is shown. Two hard paved roads are 
connected with a loose unpaved road. It can be a real situation 
or an error, but this is a suspicious condition should be 
checked. This control takes approximately %5 of the 
production time of data capture and approximately %2 to 3 
additional conditions of the captured data are detected. 
Approximately %30 of the conditions is real errors. Generally 
by the use of computers for the vector data or map production, 
this kind of control over printouts are begun to be neglected. 
But the main importance of this control is the errors detected 
here are generally important and coarse errors that affect the 
overall quality of data. An operator who is capturing data on a 
small area can mix up a pipeline with road, but it can be easily 
detected over the printout by the very long and straight 
geometry and connection of pipeline. This kind of many 
important and coarse errors are detected and corrected as a 
result of this control. 
    
   
Figure 4. Connected roads with different qualities 
At the final level, geonames are controlled by using different 
sources such as USA (United States of America) NGA 
(National Geospatial-Intelligence Agency) Geonames Server, 
old topographic maps and plans, other gazetteers etc. A control 
stuff checks all the captured geonames over the printouts and 
digital data and looks at; 1) if the correct name is assigned, 2) 
if the extent of the name is given correctly. In Figure 5 extent 
of a river is signed with yellow that a single geoname should 
be assigned. After that yellow part, the name of the river 
changes. This control takes approximately %5 of the 
production time of data capture and approximately %5 
additional conditions of the captured data are detected. 
    
Figure 5. A geoname correction 
3. CONCLUSIONS 
Reliability of a GIS mainly depends upon its accuracy. The 
accuracy of such information can be achieved by two main 
approaches. In one approach the data is controlled and