Vector length of common boundaries

Q: Any way to get the length of common boundaries shared by neighbouring areas in a map?

A:

Sample data set: sudden infant deaths data from North Carolina, data imported from SHAPE file with v.in.ogr.

Let's have a look at the data:

d.mon x0
d.vect sids_nc

We add a second layer to the map which references the boundaries of polygons. In the vector geometry we generate an ID (category) for each boundary:

v.category sids_nc out=sids_nc2 layer=2 type=boundary option=add

Underlying idea: we'll fetch the IDs (categories) of the polygons left and right from each boundary and store it into the attribute table linked to layer 2. In general:

# cat_of_boundary | cat_of_left_polygon | cat_of_right_polygon | 
length_of_boundary

We want only one category per boundary, that's why the sides check is needed (a boundary may consist of several pieces) So we create a new attribute table and link it to the new layer 2 of the vector map:

v.db.addtable sids_nc2 layer=2 col="left integer,right integer,length integer"

Now we query the polygon/boundary relationships and store it into the attribute table linked to layer 2:

v.to.db map=sids_nc2 option=sides col=left,right layer=2 

Now we have unique categories for the boundaries and can calculate the lengths:

v.to.db map=sids_nc2 option=length col=length layer=2 

Done.

See the new attribute table containing the boundary lengths:

v.db.select sids_nc2 layer=2

Verification (let's check boundary #193):

d.vect sids_nc2 cat=193 layer=2 col=red type=boundary
d.zoom
d.measure
# LEN:     12756.00 meters

What does the attribute table say:

v.db.select sids_nc2 layer=2 | grep '^193'
190|65|68|12814

This is reasonably close since on screen digitization in d.measure isn't always that precise ...