RNFdev Geocoding with PAGC Primer (long)

[Dan Putler]

Hi Dave and Frank,

Dave, excellent summary, which should give Frank the background he  
needs. It turns out that we get more than an FSA centroid file as a  
supplemental product, we get an FSA polygon file, which is a _lot_  
more useful.

Frank, the polygon creation recipe is my current thinking about how  
to handle urban area FSAs, but it is not the only approach. Others we  
have considered:

1. Directly scanning and digitalizing the Canada Post FSA pdf.  
However, this seems problematic for several reasons.

2. Using the FSA boundary roads as guides for hand digitalizing the  
FSA polygons (as opposed to using the road segments directly in  
creating the polygons). This has problems of lining polygon edges  
with road boundaries and potential problems of overlapping polygons.  
Avoiding this problem is very labour intensive.

3. Geocoding a large number of street addresses using every mailing  
list we can get our hands on, and then creating a polygon out of the  
convex hull of points for each FSA. The drawbacks are (1) getting a  
large enough address database and (2) geocoding that may addresses is  
labour intensive.

Another alternative, and one we haven't discussed, is taking an  
OpenStreetMap project approach and have people drive or bicycle the  
boundaries of an FSA while using a gps receiver to record their trip,  
after which they could upload the gps tracks of their trip around the  
boundary of an FSA. The plus to this one over the other approaches  
considered is that it has a much higher "cool factor" since it allows  
someone to use their gps toys for things other than geocaching ;-).

The real upshot is that any approach for obtaining FSA boundaries for  
Urban areas is going to be labour intensive.

Dan