[postgis-users] extracting zipcode bundaries from tiger2005sebasedata

[Stephen Woodbridge]

Hi Dan,

This is why I suggested creating the block polygons and coalescing them 
into zcta polygons. The will not work for you in Canada because you do 
not have the additional topology that is available in the Tiger/Line data.

-Steve

Dan Putler wrote:
> Hi All,
> 
> There are a couple of interesting points on this matter from the Census 
> Bureau's FAQ. The relevant FAQs are questions 17 and 18. The link is:
> 
> http://www.census.gov/geo/www/tiger/faq-index.html
> 
> I am part of a small group working on creating an open Canadian Forward 
> Sortation Area (FSA) polygon layer. An FSA is Canada Post's equivalent 
> of a US five digit zip code. In urban areas we are attempting to use the 
> boundary road segments for an FSA from the recently released Statistics 
> Canada 2006 Road Network File (the Canadian counterpart to the 
> TIGER/Line road layer). However, these produce boundaries with gaps. So 
> far we haven't found a solution that doesn't involve hand digitalizing 
> some of the boundary segments, and my reading of the TIGER documentation 
> suggests that the same problem will occur when TIGER/Line data is used 
> to create zip code polygons. One important difference between Canadian 
> FSAs and US zip codes is that FSAs are assigned based on spatial 
> polygons, while this is not the case for US zip codes. The problem with 
> Canadian FSAs is that things other than roads are used as boundaries in 
> a number of instances.
> 
> Dan
> 
> On 18-Oct-06, at 9:50 AM, Stephen Woodbridge wrote:
> 
>> Dylan Beaudette wrote:
>>> Hi everyone,
>>> On the GeoServer website there are excellent directions on how to 
>>> extract specific information from the tiger2005 base data. However 
>>> there are no instructions covering the extraction of zipcode 
>>> boundaries. Does anyone know of a good way to go about doing this in 
>>> postgis ? I already have the tiger2005se database setup.
>>
>> Dylan,
>>
>> Here are a couple of possible solutions.
>>
>> A. this is how the census does it
>>
>> 1. create all the block polygons
>> 2. assign each block polygon a zipcode based on analysis of the 
>> zipcodes on the inter edge of the block polygon
>> 3. union all the block polygons for each unique zipcode
>> 4. look for small islands where a block is surrounded by other other 
>> zipcodes and deal with it.
>>
>> B. probably a bad option
>>
>> 1. collect all edges that belong to a zipcode excluding '00000'
>> 2. pass this to buildarea() function.
>> 3. this will probably fail because of gaps in the edges where zipcodes 
>> are not assigned.
>>
>> C. download the zcta polygons for Census, inspite of the fact they are 
>> based on Tiger2000 files.
>>
>> -Steve
>> _______________________________________________
>> postgis-users mailing list
>> postgis-users at postgis.refractions.net
>> http://postgis.refractions.net/mailman/listinfo/postgis-users
> 
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users