uniquely 
map frame 
reference 
ale of 
every 200 
> number of 
These 
rect 
>, and, 
'ontrol 
> controls 
‘the map. 
ecome the 
An accurate 
o edge 
so to 
of 
or regional 
 geodetic 
ental area 
, another 
k. The 
| type of 
t cover the 
repared for 
mber 
of base 
yrmation is 
8 
ital 
fferent 
Table 2: 
Geodetic Controls Required for Regional GIS 
  
  
  
  
  
  
  
  
  
  
  
  
  
Continent Areal Density of Total # of # of controls | New controls 
Coverage control pts. Controls available* needed 
square km | per 25 needed 2-D 3-D 
square km 
Africa 22,612,063 1 904,482 7,800 896,682 
Asia 17,601,806 1 704,072 596,000 108,072 
Australia 7,715,294 1 308,611 44,300 264,311 
Europe & 25,179,431 1 1,007,177 2,385,000** 696,088 ** 
Former 
USSR 
North 1 861,436 490,000 371,436 
America 21,535,902 
South 5,917,631 1 236,705 74,800 161,905 
America 
Total 100562127 4,022,483 3,597,900 2,498,494 
Sq. km 
  
  
  
  
* Source: World Cartography, UN Publication. 
** Former USSR has only 200,000 control points available and needs an extra 696,088 controls. 
Europe has more controls than required. 
types of thematic applications. 
The information derived from data can be of three types: 
€ spatial : locational data 
® attribute : non locational data 
® temporal : informational changes (i.e. changes in data 
due to time changes). 
Data collection or digitization can be done in either raster or 
vector mode. Most remote sensing data are acquired in 
raster mode and are stored in this format. It is less expensive 
to process and store data in compressed raster format unless 
data are specifically needed in vector format for a particular 
application. 
The conversion of map data can be done using two main 
methods: 
e manual digitization - the map is manually traced with 
a digitizing cursor. This approach is very time 
consuming and labor intensive but it works for almost 
all maps. 
e electronic scanning - the map or drawing is scanned 
electronically and stored as a raster image with no 
intelligence. Images of scanned maps are then 
generally converted to a vector map and information 
from the data base is tied to them, as for example, 
soil type or land value. Scanning can be faster than 
manual digitization, but works best for clean, 
uncluttered maps which do not contain overwritten 
data. 
After scanning a document one of three main 
approaches to raster-vector conversion can be 
followed: heads-up digitizing, semi-automatic 
vectorization and automatic vectorization. In heads-up 
digitizing an operator scans in a map and brings it up 
on the screen. S-he then traces the lines off the 
screen by using a mouse to follow them. This works 
well for a multitude of map types and complexities. 
In semi-automatic mode, the computer attempts to 
follow lines automatically but stops and waits for 
operator intervention when it encounters a problem 
area. The operator then examines the problem, makes 
a decision and the computer continues. This works 
well for relatively clean maps without too much 
overwritten text and lines. In fully automatic 
vectorization the computer traces all the lines by 
itself, often in overnight batch mode. Editing is then 
done to clear up any remaining problem areas. This 
works only for very clean maps, such as contour 
maps, or large scale cadastral plans, but it is rarely 
applicable to most local government maps. 
The problems of digital data collection in a GIS system are 
mainly: 
€ The quality of digital data for topographical mapping 
€ Interchange of digital information between different 
computer aided mapping system (CAMS). 
The Quality of Digital Data for Topographical Mapping 
Conventional topographical maps are produced within the 
accuracy required for a specific scale. Digital map data are 
collected for a specific scale, but, due to the fact that these 
data are expressed in ground coordinate values without any 
permanent accuracy statement attached to them, there is no 
way to recognize this fact from the digital map.