To accomplish the project's purpose and objectives, several stages must be implemented.  The computational procedures required four main software products: ER-Mapper, Freehand 8, Idrisi32 and AdobePhotoshop.  A major task was to convert the vegetation map from an analogue to a digital format, and match it to the Ikonos imagery.  The Ikonos imagery had to be imported using ER-Mapper, and thereafter a wide range of image interpretation visually and computationally, led to a final classification stage.  Several supervised and unsupervised classification techniques were compared to identify general trends of changes in vegetation and to reveal the characteristics of each method.  Tabular output as well as a visual change analysis between 1979 and 2000 were the final stage.  Finally, the Modifiable Areal Unit Problem (MAUP) was addressed by changing the resolution of the imagery to examine the influence of this factor on our results.

Processing of the Old Map
The original map is a blue draft map in the size of 36 x 23 inches, which was created by Professor English in 1979.  Therefore, it must be converted from an analogue map into a digital map so that it can be used in remote-sensing and GIS applications for comparison with the Ikonos classification.  The map shows the distribution of different plant assemblages in the Slave River delta.

 
Map Manipulation
Colouring the map involves filling the different polygons with different colours according to the vegetation classification in the old map.  The old map had to be photocopied first, thereafter, colour pens were used to manually colour the different vegetation classes.  Then the coloured map was scanned into the computer in eight sections, which were about 8 x 13 inches each.  Finally, the map was merged using the Freehand-software to produce a complete digital map.

Enhancement of the Digital Map
The digital map was inaccurate; thus the classification resulted in a noisy image with different kinds of vegetation in the same class.  Moreover, polygons with similar colours were often combined together.  Consequently, enhancement was performed manually in Adobe Photoshop one by one polygon to ensure a better classification result.

Geo-Referencing
Our three base maps of the Slave River delta were in three different formats.  The old map was not referenced because it was drawn from aerial photographs, the topographic map uses the UTM 12n system while, the Ikonos image utilizes the US 27tm 12n reference system.  For useful comparison it was necessary to convert all the maps to a common projection with the same reference system.  The geo-referencing process encompasses two procedures: a) resample the old map to the topographic map and b) referencing the resultant maps from a) to the Ikonos image (Images).  The resampling procedure was used as rubber sheet transformation that stretches and warps the old map to fit the topographic map with its known reference system. Control points with X,Y, coordinates were selected and recorded on same features in the two maps.  In Idrisi, these control points were used to calculate the individual residual errors and the total “root mean square (RMS)”.  These residuals express how far the individual control points deviated from the best fit equation.  Values with less than 0.05 are considered favourable and were kept whereas, high values were omitted.  After resampling, the old vegetation map should be reoriented to match the topographic map with the UTM 12n reference system.  The second step was to match the resultant map's UTM 12n reference with that of the Ikonos image's US27tm12n.  The same procedure would be repeated again and the final output would have the resultant map match the Ikonos reference system.

Classification of the Old Map
The last step was to classify the map to get distinct vegetation classes, each carrying a unique identifier to be able to perform spatial analysis.  The Isoclust performed the best in retaining the original classes, and the result was used for the spatial analysis process.

 
 

Cartographic Model