Full text: ISPRS 4 Symposium

325 
o classification of units should not rely heavily 
on spatial relationships - For instance, if it is 
necessary to know that one unit has a particular 
shape in order to reliably classify it, this 
method would probably not be appropriate to class 
ify that unit. 
RECOMMENDATIONS FOR IMPROVEMENT OF THE METHOD 
The cubic convolution type of DICS tape was chosen early 
in the project, before some important factors were known. 
It is probable that another resampling algorithm would have 
proven more useful. The cubic convolution algorithm involves 
use of several pixels in determining the spectral values 
assigned to the new pixel. Use of the "nearest neighbour" 
algorithm could have eliminated the smoothing effect of the 
cubic convolution method, which may have disguised some 
small areas of each unit. 
The digitizing of sample pixels involved, once the polygons 
had been transferred from airphotos to print of image, the 
calculation and writing down of Universal Transverse Mercator 
coordinates, assignation of UNIT value and entering of these 
numbers into a data file to be used in extracting the band 
values from the DICS tape. Had a polygon-to-raster conver 
sion program been operational at the time, it would have 
been possible to digitize more data points, thus possibly 
improving the discriminating power of the analysis. It is 
recommended that this methodology not be undertaken in future 
without capability to perform this transformation. 
RECOMMENDATIONS FOR FURTHER RESEARCH 
It is recommended that a modified version of this methodology 
be evaluated in analysis of other regions, particularly 
those which are simpler in surface expression than the 
Peace-Athabasca Delta. 
CONCLUSION 
The methodology described has relatively little potential 
for monitoring of geomorphological change in the Athabasca 
Delta. The major reason for this is the complex nature of, 
and small unit size within, the Delta. Analysis results 
indicate that, if a smaller number of fairly generalized 
units were the subject of analysis, this type of digital 
analysis could be useful in ecological management. 
ACKNOWLEDGMENTS 
This study was funded partially by a grant from the Alberta 
Remote Sensing Center, Edmonton, Alberta and partially by 
Faculty of Environmental Design Research Assistantships. 
Logistic support was provided by TES Research and Consulting 
Ltd. of Calgary, Alberta.
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.