[postgis-users] PostGIS Capabilities

[Jong S. Lee]

Tyler Mitchell wrote:

>
> Jong, I was also just thinking that if you use LRS to locate your 
> point in the first place, can't you take the attribute (i.e. parent 
> road name) and store that with the point data.  Then if you looking 
> for all the points on that road you can use regular tabular queries 
> instead of spatial queries.   Does that make sense?
>
Yes, that's possible. However, since the network for matching the 
bridges on the highway is too detail for our research purpose, we 
simplify the network.  During this process the LRS route ids are lost or 
wrongly placed since we did many merging and modifying operation to the 
original network.  The following is our procedure:

1. Matching Bridges on Network A
2 .Modifying the Nework A to Network B
3. Finding Bridges on the Network B.
( if you have any other suggestion, let me know....)

Currently, via using SQL you told me, it is working great.  I am using 
0.01 meter as the tolerance (or distace b/w features).  I pick some 
lines and bridges and verified those.  It looks good.

> postgis-users-bounces at postgis.refractions.net wrote on 02/20/2004 
> 09:00:51 AM:
>
> > > I agree with you opinion. However, the bridge data (point shape 
> file) is
> > > created based on the highway data.  The highway data has LRS (linear
> > > referencing system) to locate the point (or event) on the highway. 
>  The
> > > bridge data is created via LRS.  Therefore, it should be exactly 
> on the
> > > highway.  But, I am not sure. I can't trust ESRI product 100 %.  If I
> > > assume that ESRI product makes the right bridge data on the highway,
> > > the precision of the coordinates might be changed when I converted
> > > shapefile to PostGIS db table.  Still I suspect the ESRI product, too.
> >
> > Just because you used an LRS to generate the points doesn't mean
> > that they will
> > necessarily be located exactly on the line, in a strict 
> mathematicalsense. It
> > may not be possible to describe the location on the line exactly
> > using a given  
> > number system (whether it's binary or decimal doesn't matter). For 
> example,
> > what if you want to describe the position that is 1/3 of the way
> > along the line
> > from 0,0 to 1,2 - you cannot describe this exactly with decimal
> > numbers, in any
> > given precision. So the point that you get will be at, say 0.333333,
> > 0.666667 -
> > which is not, in a strict mathematical sense, on the line. It is 
> very close,
> > and I suspect that Arcview is realizing that it is very close (as 
> close as
> > possible with the numerical precision available) and accepting it as
> > exactly on
> > the line.  Whereas, I would expect GEOS to up and tell you that no, 
> it is not
> > on the line.
> >
> > I could be completely wrong here... but this is my suspicion.
> > Anyways, AFAIK it
> > is best to ensure that there is an actual node in your linestring 
> (highway)
> > located at the exact same location as the point feature (bridge) in 
> order to
> > ensure that they are and will reamin touching.
> >
> > Chris
> >
> > _______________________________________________
> > 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
>  
>


-- 
Jong Sung Lee, Graduate Research Assistant
Department of Urban and Regional Planning
University of Illinois at Urbana-Champaign
111 Temple Hoyne Buell Hall
611 East Loredo Taft Dr.
Champaign, IL  61820

Tel:(217)244-6564 
E-mail:jonglee1 at uiuc.edu