communications, hydrology, vegetation, 
structures). It was decided to store data on the 
basis of individual 1:250,000 sheet MAPS’. 
Level-II ‘logic blocks’ are represented by the 
categories’, such as (in the case of the class 
'communications'), ‘private’ or ‘official’. 
Level-III is represented by the ‘sub-categories’, 
such as ‘Highway Class A’, ‘Highway Class B’, etc. 
Individual database objects consist of road, rail, 
river, canal, powerline, etc. segments; symbolised 
point features (e.g. trees, wells, important 
buildings); symbolised area features (e.g. builtup 
areas); and, alphanumeric character strings as 
represented on the 1:250,000 series. 
Individual database objects are identified at the 
sub-category’ level, and the spatial attributes 
of each object are described in its ‘geometric 
block’. A particular ‘geometric block’ may be 
linked to several (usually) Level-III (i.e. 
sub-category’) ‘logic blocks’. 
The ‘geometric block’ contains a definition of the 
database object as a graphic element (e.g. line, 
arc, text, symbol, point, area), sufficient 
coordinates (eventually, for India’s National 
Topographic Database these will be Latitude and 
Longitude and not Cartesian coordinates) and other 
parameters (e.g. linetype, pen number, display 
colour, display text font, plot text font, angle) 
to cartographically display the database object. 
It can be seen that the data model is strongly 
influenced by the main perceived user of the 
National Topographic Database, namely the SOI. 
4.2 The Test Documents 
From a consideration of the database objects 
described above it was concluded that five (namely 
1,3,5,6, and 8 ) of the nine production separates 
given in Section 2 and the trig. point coordinate 
lists could be the sources from which the National 
Topographic Database could be built. 
A common procedure in performing ’benchmarks’ or 
‘time trials’ of conversion (or map data capture) 
systems is to use test documents which are 
representative of all the task(s) to be performed. 
For example sheets showing low, medium and high 
planimetric and contour detail may be used. 
Furthermore digitizing/scanning and any necessary 
subsequent processing should be performed by 
experienced operators. À problem we encountered 
was that the team vas not actually experienced in 
all four production lines! For reasons of the 
predicted professional future of one of the team 
members, it was decided that this single team 
member should become experienced in all four 
production lines and perform the ‘time trials’. 
Because of the time needed for this 
familiarisation, it was not possible for the test 
to be very extensive. Nevertheless a testing 
methodology was developed which could be repeated 
with low, medium, and high detail density map 
fragments. A single test document (represented by 
its 5 separates) was used. This document was a 400 
sq km. fragment of the SOI 1:250,000 sheet No. 
56A, containing medium detail density, with the 
exception of the contours. In the chosen area the 
contour density was low, so an identically sized 
medium contour density separate fragment was 
selected instead. 
Map conversion procedures generally use a set of 
grid intersection points as the control points 
whose Cartesian coordinates are known, to obtain 
200 
The SOI 1:250,000 
series had no projection grid, but only the 
graticule on it, trig. points (whose Cartesian 
coordinates are known) could be used instead. But, 
in the test area there were insufficient trig. 
transformation constants. As 
points for obtaining transformation constants, SO 
it was decided to register a  photo-optically 
plotted 4 cm grid with the separates, and 
photographically produce a new set of separates. 
The resulting nine grid intersection points (in 
the test area) were given appropriate fictional 
Cartesian coordinates in the range OmE,OmN to 
20000mE, 20000mN. 
The FIGURE 1 shows these test fragments. 
4.3 The Time Trials 
In the following four sub-sections the procedure 
followed for the investigation of each production 
line is outlined. The results are given in section 
5. 
4.3.1 Manual Digitizing Production Line 
Set-up involved creating the empty file, setting 
the mapping units, partitioning the digitizer, 
selecting the design options, and establishing the 
control points. This took about 5 minutes per 
separate. All features in each separate were 
digitized and the contours were assigned 
attributes (by assigning them to appropriate 
layers); times were recorded. 
On completion of the digitizing, check plots vere 
made. Missing details were identified, counted, 
and interactively corrected. There were four 
missing entities (1 black, 1 red, 2 brown). Time 
spent on interactive correction was recorded. 
Logical consistency was checked on the black and 
red separates only. No inconsistencies were found. 
The TABLE 2 below summarises the times for manual 
digitizing. 
  
  
  
  
  
  
  
Operation Times in minutes/separate 
Black Red Blue Green Brown 
Digitizing 90.00 105.00 10.00 10.00 110.00 
Editing 20.00 20.00 0.00 0.00 30.00 
Total 110.00 125.00 10.00 10.00 140.00 
Table 2 - Summary of elapsed times for manual 
digitizing 
As the grid was not digitized an estimate of RMSE 
values and thence MAXIMUM ERROR was obtained from 
the error percentages provided at digitizer set up 
(control pointing). This approach provides an 
estimate of static rather than dynamic digitizing 
accuracy. Results are shown in TABLE 3.