fferent 
ons is 
ients at 
1ild up 
iracy of 
e this 
] (i.e. 
1e photo 
7 become 
ion and 
ealistic 
ained in 
e, for a 
h is 
for all 
cale 
0 cm 
5 cm 
10 cm 
20 cm 
50 cm 
can be 
ammetric 
inits m) 
  
|.[50 cm 
(units m) 
  
map scale 
1:10,000|1:5,000 |1:2,500 |1:1,000 |1:500 
photo sca En 
  
  
1:30,000 1.62 0.96 0.71 0.62 0.60 
1:20,000 1.55 0.85 0.55 0.43 0.41 
1:10,000 1.51 0.78 0.42 0.25 0.21 
1: 5,000 1.50 0.76 0.39 0.18 0.12 
1: 2,500 1.50 0.75 0.39 0.16 0.09 
  
  
  
  
  
  
  
  
1.12 
[159.78 
| |] 0.54 
Ef 0.51 
0.50 
  
  
  
lection 
le gives 
ital data 
possible 
til the 
GIS, the 
r initial 
as that 
rement to 
al data. 
15 um at 
owing the 
rammetric 
er, that 
different 
The table 
. with no 
Table 4: Graphical plot of photogrammetric digital 
data (with no identification error) 
3.1.3 Digitising existing maps Digitising 
existing maps is an attractive alternative to 
photogrammetric digital data collection in order 
to set up a digital data base. In the feasibility 
study, however, careful consideration must be 
given to the following two issues: 
- the age of the maps and the rate of development 
of the area concerned, which determine the 
number of changes and hence the revision effort 
required. In this context, changes also imply 
changes in the map content requirements, since 
the original maps were produced 
- in view of the loss of accuracy through the 
digitising process, one will in general want to 
digitise at a larger map scale than has to be 
finally plotted and this, of course, reduces the 
economic attractiveness. 
  
Phase I: data collection 
This initial data collection phase covers the 
whole digitising process. The input errors are the 
positional errors in the map original. In 
evaluating these, account will have to be taken of 
the original production method and parameters used 
such as photo scale, control distribution in the 
AT... etc. etc.; the original accuracy 
specifications (which imply a standard deviation 
of 0.3 mm at map scale with the standard 
specifications of "90% within 0.5 mm", but only 
applicable to well-defined features); the 
identification errors to be added to all other non 
well-defined feature classes and finally the 
generalisation errors introduced in the form of 
feature displacements, during the cartographic 
phase. 
  
The measuring errors in the digitising process 
vary with the digitising method used: point 
digitising and the sampling density in relation to 
the complexity of the feature; stream digitising 
and the digitising speed, again in relation to the 
nature of the feature; automated scanning in the 
raster mode and the subsequent raster to vector 
conversion; automated scanning in the vector mode; 
the precision of the digitiser or scanner; the map 
quality (thickness of lines, errors in the 
original, etc.) and the operator acuity. 
Assuming an accuracy in the map original of 0.3 mm 
(excluding identification and generalisation 
errors) and a digitising accuracy of 0.15 mm, the 
digitised data will have an accuracy of 0.35 mm at 
map scale. This is illustrated for different map 
scales in table 5. 
597 
  
  
(units m) 
map scale| c 
1:50,000 |17.50 
1:25,000 | 8.75 
1:10,000 3.50 
1: 5,000-| 1.75 
1: 2,500 | 0.88 
1: 1,000 0.35 
  
  
  
  
Table 5: Digitised map data (excl. identification 
and generalisation errors) 
Phase II: data storage and presentation 
Since no further processing takes place until this 
data is incorporated into a GIS, the accuracy of 
the digitised map data after initial storage and 
processing will be in the same as that after data 
collection, shown in table 5. 
Phase III: data presentation 
A graphical plot is quite a normal requirement to 
be produced from digitised map data. The accuracy 
of this graphical plot is determined from the 
accuracy of the digitised map data and the 
plotting accuracy, which can be taken to be 0.15 
mm at plotting scale. Note, however, that 
different tables are needed for the different 
values of the identification and generalisation 
error. The table given below refers to the 
intrinsic data accuracy i.e. accuracy 
excluding identification and generalisation error. 
(units m) 
  
plotting 
digitising scale|1:50,000|1:25,000/1:10,000|1: 5,000|1: 2,500|1: 1,000 
scale 
  
  
  
  
  
  
  
  
1:50,000 19.04 - - - - - 
1:25,000 11.52 9.52 - - - - 
1:10,000 8.28 5.13 3.81 - - - 
1: 5,000 7.70 4.14 2.30 1.90 - - 
1: 2,500 7.55 3.85 1.74 1.15 0.95 - 
1: 1,000 7.51 3.77 1.54 0.82 0.51 0.38 
  
  
Table 6: Graphical plot of digitised map data 
(excl. identification and generalisation 
errors) 
3.2 Attribute accuracy standards 
Attribute accuracy expresses the correctness and 
completeness of the digital data stored in a data 
base and is built-up from an evaluation of the 
following different characteristics defining the 
quality of the digital data: 
- data classification: have errors been made in 
not going to a sufficant degree of detail in 
data collection i.e. streams not further 
classified as perennial, intermittent, dry; was 
the feature definition clear enough so that the 
correct boundary has been picked up i.e. not 
clear if boundary of a highway area to be taken 
from cadastral boundaries as depicted by the 
fences or the limits of the hardened highway 
surface; has account been taken of temporal 
changes in the classification system e.g. roads 
department changes its road classification