On 11/10/2011 7:37 PM, MarkW wrote:
This is kind of a simplistic approach maybe - because a simple grid
doesn't have to intersect with municipal features. But could you
simply decide on a rounding level that would give you what you wanted?
Round the lat/lons to a certain decimal level, then count by / group
on those rounded lat/lons? Each one could represent centroid or a
corner of a grid.
Giving a very simplistic overview... Our current application creates "a
grid" by using
ST_SnapToGrid to collapse multiple points to one "standard" point. It
then uses a
polygon drawn on a map to create a "boundary". Given a list of points
and a polygon, using
point && polygon [with a gist index on each] for speed
and ST_Intersects(point,polygon) for accuracy you can collect
whatever points are "in" the boundary ...
Given a pre-defined grid size...
If you don't want to hassle with boundaries right now, you can artificially
create all the grid points within your lat/lon range, assign them a
neighborhood, then look for "identical"
snapped to grid points in your data with a straight equality test for
the join
to collect a count "in" a neighborhood.
Make sure you are working with the "right" (or at least same) SRIDs.
Roxanne
_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users
