rangeland, agriculture with a minor 
woodland component. 
Figure 1. Location of Alberta Project Area 
(1-Peace River Zone; 2-Central 
Zone; and 3-Foothills Zone) 
METHODOLOGY 
Maps 
The method of land use change detection 
involved the use of the Canada Land 
Inventory (CLI) "Present Land Use" series 
(McClellan et al., 1968) . This map series 
covers most of the settled part of 
southern Canada at a scales of 1:50 000 or 
1:63 360 and represents the most complete 
attempt at detailed, large area land 
use/land cover mapping in Canada. The 
maps date from the 1960's and have never 
been updated. 
Imagery 
Colour composites of TM bands 3, 5 and 4 
respectively assigned to blue, green and 
red were enhanced with the boreal 
reflectance enhancement. The enhancement 
is a standard product which emphasizes the 
woodland cover and provides good scene-to- 
scene consistency (Ahern and Sirois, 
1988). Boreal enhancements proved most 
useful in the Peace River and Central 
zones but less effective in the rangeland 
dominated environment of the Foothills 
zone. 
In an effort to improve the accuracy in 
the identifying and mapping losses in 
forest cover which dominated the land use 
changes, single TM band 3 black and white 
winter images for each scene centre were 
also used. In winter, open areas are 
blanketed with snow and appear white, 
while densely forested areas appear black 
or grey because of a lack of snow on the 
forest canopy. This high contrast enabled 
a rapid and straight-forward delineation 
of cleared forest areas, which were then 
be classified using the corresponding 
summer image. 
A total of 40 images with a common year of 
1986 were interpreted for the three zones. 
East-west overlap between the satellite 
paths enabled several adjacent images to 
be used. This helped to overcome 
unexpected cloud cover, discontinuous snow 
cover or poor guality of some images. 
Interpretation Technique 
Land use changes were located by comparing 
the CLI "Present Land Use" maps with 
visually interpreted 1986 Landsat Thematic 
Mapper imagery. Visual image 
interpretation was a suitable means of 
gathering the required information over 
the large area because of low cost of 
visual imagery compared with digital 
imagery and classification. Additionally, 
the ability of a visual interpreter to 
make use of textural and contextual 
information not readily available to 
digital classifications was another reason 
for employing visual interpretation. 
The image transparencies were integrated 
with the land use maps using the Procom-2 
optical transfer instrument. The 
instrument was equipped with the optional 
Stereo-Change Detection (SCD) module, a 
second projector assembly that allowed the 
winter and summer images to be co 
registered and interpreted simultaneously. 
The first step was to define a land cover 
classification. The 14 Canada Land 
Inventory "Present Land Use" classes were 
collapsed into 8 general classes capable 
of being discriminated with Thematic 
Mapper imagery (Table 1). One class 
(clearcuts) was added. It was formerly 
included as part of "unproductive 
woodland" category in the CLI 
classification. It was decided to include 
this additional class to document the 
extent of this type of change in the 
wooded environment. 
Using 1:20 000, 1:30 000 and 1:40 000 
scale aerial photography as reference data 
of the land use types in Alberta, a 
general interpretation key was developed 
based on the spectral, textural and 
contextual aspects of classes as they 
appeared on the Thematic Mapper imagery 
(Table 2). 
With the imagery registered to the map, 
the change detection procedure consisted 
of systematically comparing a land use 
polygon on the map with its class on the 
imagery as determined from the 
classification key. If whole or part of a 
polygon appeared different from the 
expected appearance for a particular 
class, the change was delineated and 
classified using a two-letter code 
representing the pre to post change land 
use combination. 
RESULTS 
There were a total of 49 possible pre to 
post change land use combinations plus an 
additional "no change" class. Thirty of 
these change combinations were actually 
found and mapped. The amount of land use 
change observed by the project are shown 
in Table 3. Over 4 6 000 polygons of 
change were mapped corresponding to some 
17 000 km 2 of changes or about 15% of the 
area surveyed. Significant regional 
variations were observed. The Peace River 
region had the greatest amount of change 
per unit area and the Foothills zone the 
least amount of change. Most of the 
change in two zones was from Woodland (T) 
359