[postgis-users] gis noob question

[Iulian Margarintescu]

Paragon Corporation wrote:
> Iulian,
> 
> Hmm not sure how much of a noob you are since the below looks kind of
>  sophisticated for a newbie with some minor syntactical errors Should
> be and my parethesis may be off

My first contact with GIS was a few days a go, but i've done lots of
reading and tests these few days ... still i feel like a noob. ( and i 
went from think geo to NetTopologySuite to PostGIS

> select (st_dump( (select  ST_SimplifyPreserveTopology (
> st_collect(the_geom),0.01) from zipcodes ))).geom
> 
> 1) Question 1 - This assumes you are a total noob. Did you add a
> spatial index to your geometry? 
 > 2) Question 2 - Might help to cluster
> on that index as well since this is a pretty static table I assume.

I already had the index added, clustering did non made noticeable 
improvements. The data is around 450Mb so i guess no meter how is 
indexed it will take time to move it from the db to a bitmap.

> 3) Where did you get these zipcodes from?  I know for example US
> zipcodes are not that large as far as complexity goes (though they
> are large in area), so I am questioning why they need to be
> simplified in the first place unless you are trying to display all of
> them at once on the screen.

They wore already used in this project i'm working on, but i'll try to 
get more details and maybe a "simplified" version can be obtained.

> 
> 4) Regarding the below whether it is sane to do that - I would say
> probably not, but I could be wrong.
> 
> The reason I say no is that that if you collect the geometries first
> and then simplify, I think even simplifypreserve will loop thru each
> geometry anyway and simplify.  Though I'm not absolutely positive of
> that.  So the whole exercise of collect simplify dump is a bit
> fruitless if that was your intention.  Not a bad thought and maybe
> you are right so maybe someone can correct me on that.

I don't see how it could preserve topology ( witch i've understood it 
means it will keep the same number of "holes" ) by working independently 
on every zipcode... but it's very possible that i'm wrong.

> 5) If you are getting too many gaps, you may just want to lower your
> 0.01 to 0.001 or 0.005.  The basic idea being the more you simplify
> the more gaps you will get, but the less noticeable they are the
> further away the observer is - so for deep in zoom you want no
> simplification and for far out you want lots of simplification.

 > Oops I forgot one important point.  DO NOT simplify In longlat 
projection.
 > You get somewhat bizarre results and that may be perhaps what you are
 > experiencing.
 >
 > I think you will want to reproject to a planar project and then 
project back
 >
 > So
 >
 > SELECT
 > ST_Transform(ST_SimpilfyPreserverTopology(ST_Transform(the_geom,2163),
 > 100),4326) As newsimpgeom
 > FROM zipcode;
 >
 > (2163 is US National Atlas meters so you may need to change that, and 
4326
 > is WGS 84 longlat.  If you got zips from census you may need to 
change that
 > to 4269 NAD 83 long lat).
 >
 > You'll also need to play around with the simplification metric.
 >

Now this is something i've got to try.

Thanks for your ideas,
Iulian