h) Displacement of features: 
When decreasing the presentation scale of data, 
features often conflict with each other and must 
be displaced for cartographic representation. 
An example of a feature being displaced in this 
data set was a rock symbol that had to be 
interactively moved because it was too close to 
the shoreline at the 1:250 000 scale. 
3.1.3 Building Topology 
The requirement for clean structured data for 
this type of an application was apparent. Since 
the data set was not cleaned or structured, it was 
decided that sample cleaning would be done in 
order to utilize some of the automated 
generalization functionality. The two features 
that were selected for the sample cleaning were 
roads and gravel pits. The following is a brief 
description of the cleaning performed and the 
generalization applied to the clean data: 
* The roads were cleaned by building their 
topological network, which deleted 
overshoots, snapped undershoots, and 
removed pseudo nodes and duplicated lines. 
This allowed the file's arcs (roads) to be 
categorized by length, the different 
categories to be highlighted on the display, 
and the deletion of a category of lines 
interactively or in batch. 
* Since the gravel pit file was relatively small, 
its full polygon topology was built. This 
enabled: all the polygons below minimum 
size to be highlighted on the display, the 
grouping of the ones that belonged together, 
and the use of the "dissolve polygon" 
CARIS functionality to automatically delete 
the remaining polygons smaller than 
minimum Size. 
3.1.4 Edge Matching of Map Data Sets 
Edge matching is a term used to describe the 
process of making two adjacent data sets 
graphically continuous so that the features on 
both sheets continue smoothly from one data set 
into the other without breaks or disjoints. This 
functionality would be required for joining the 
sixteen 1:50 000 digital files that make up a 
1:250 000 data set to create a seamless file. 
There were many options available in the 
CARIS edge joining routine and, depending on 
the application, the operator can customize the 
routine with the different option qualifiers. 
Some of the options were as follows: edge 
matching within one file or between two files; 
matching lines within a specified maximum 
distance (map or ground distance); restricting 
the processing to a given layer number(s) or 
feature code(s); and joining lines with the same 
contour values. Also, the operator can specify a 
method of joining the lines such as straight 
extend (straight line between end points) or 
modify the lines so the join is made smoothly. 
The results of the execution of the routine were 
displayed and found that approximately 60% of 
the lines had joined. Since the tolerance must 
be kept low to avoid incorrect joins, there will 
always be the need for interactive editing. 
3.1.5 Metrication and Editing of Contours 
Since the contours from the 1:50 000 data set 
were metric and those from the 1:250 000 data 
set were imperial, the decision to use the metric 
contours was consistent with CCM's policy 
with respect to the requirement for all new 
products to be in metric. A 20m interval would 
be sufficient, thus every second contour line 
was selected from the 1:50 000 data set. 
Editing of contours was required for the re- 
entrance with single line rivers that had been 
generalized from double line rivers and to 
ensure that they did not enter the newly derived 
gravel pits. 
3.1.6 Quality Control 
Quality control is more a procedure that must be 
set, than a functionality that can be specified. 
The initial step in the quality control procedure 
was to run a statistical listing for the derived 
data set and to check that the feature codes were 
correct for the 1:250 000 product. Plots of each 
derived layer were then produced at the 
1:250 000 scale and inspected for content, 
minimum size and cartographic presentation. 
These plots were also overlaid on each other 
and checked for fit. The results of the 
comparison of the manually revised map sheet 
and the derived data plots were very favorable 
with slight variations due to generalization 
choices, within the accuracy tolerances. 
3.2 Revision Using Landsat TM Imagery 
The 1991 Landsat TM imagery that was 
acquired consisted of seven spectral bands and 
covered approximately the area of a 1:250 000 
map sheet. Landsat TM has a resolution of 
30m. The image was subdivided and 
reformatted to PIP (PCI-BSQ intermediate file 
290 
  
  
format 
on the 
Systen 
the PI 
was re 
32.1. 
Refere 
In pre 
referer 
the La 
image 
stretch 
the rec 
the de 
identif 
collect 
must f 
systen 
proper 
mappi 
unava 
displa 
image 
vector 
transf 
Colvo 
Thirty: 
interac 
raster 
refere 
especi 
of sirr 
road | 
fields. 
and re 
functi 
graph: 
contrc 
adjust 
points 
large 1 
adjust 
anoth: 
correc 
perfor 
transfc 
Were : 
image 
The s 
residu 
(410. 
neight 
and th 
obtair