413
selected and were examined with the zoom transfer scope. All were
located in areas of research interest, where imagery (either "Quick-
Look" prints or positive transparencies) was available for the year
1973. The results of this examination were below expections for
the following reasons:
- A continued inability to discriminate between river banks
and shore leads, or between islands and surrounding open
water;
- an inability to identify many islands shown on the maps;
- some marked discrepancies, so far unexplained, in the
positions of islands relative to the shore, when one ERTS
image was compared with another;
- distortions, so far unexplained, of shorelines used in
image to map scaling with the Zoom Transfer Scope; and
- an inability in later break-up stages to detect differences
between fast ice and moving ice as well as between "dirty"
ice and "clean" ice.
In an attempt to heighten the discrimination between land
and water areas several images were examined using a colour
densitometer. There was no gain in the amount of interpretable
data, compared with the results gained by using the Zoom Transfer
Scope (Figures 5 and 6). The problem lies in the range of density
levels provided in the "Quick-Look" prints.
The change in position of islands, shoreline distortions
and the inconsistent shapes of islands seem to be due to a
combination of two factors. Firstly, the use of "Quick-Look" imagery
may imply acceptance of uncontrolled local distortions in the
printed image which are beyond the capability of the Zoom Transfer
Scope to eliminate. Secondly, differences in publication date,
aerial photography, interpretation and compilation exist in the
1:250,000 maps. It is possible that some changes in the shape and
position of islands represent genuine channel changes caused by
erosion and deposition over 20-25 years. This possibility is given
more weight because maximum agreement between ERTS imagery and the
maps occurred where bedrock control of channel morphology is
dominant (for example, near Wrigley and Arctic Red River), while
maximum distortions appear in areas of more easily worked materials
(for example, Birch Island and Old Fort Point). The changes in
shape between frames, however, indicate that significant
distortions are inherent in the imagery. This can exceed 0.5 KM
in positional error. This exceeds the ability of the system to
accurately detect the important shore leads.
Perhaps the best illustration of the problems encountered
during interpretation can be given by examining the test plots of