[postgis-users] [postgis] Optimal tile size for Raster

Heng Zhi Feng (zheng@hsr.ch) zheng at hsr.ch
Wed Dec 4 23:25:48 PST 2013


Hi Pierre,

Thanks for the explanation. I will look at this direction to understand things better.

Regards,
Zhi Feng 

-----Original Message-----
From: postgis-users-bounces at lists.osgeo.org [mailto:postgis-users-bounces at lists.osgeo.org] On Behalf Of Pierre Racine
Sent: Mittwoch, 4. Dezember 2013 18:31
To: PostGIS Users Discussion
Subject: Re: [postgis-users] [postgis] Optimal tile size for Raster

I guess maybe you are only looking at PostGIS for its capacity to deal with raster in the database. You have to know that rasters and geometries are located in space (georeferenced) and might very well intersects if they overlap in space. Geometries are generally loaded from shapefiles. To create some polygons that overlap your raster just open the latter in a GIS, create a new shapefile and add it some geometries.

> -----Original Message-----
> From: postgis-users-bounces at lists.osgeo.org [mailto:postgis-users- 
> bounces at lists.osgeo.org] On Behalf Of Heng Zhi Feng (zheng at hsr.ch)
> Sent: Wednesday, December 04, 2013 12:07 PM
> To: PostGIS Users Discussion
> Subject: Re: [postgis-users] [postgis] Optimal tile size for Raster
> 
> Hi Pierre,
> 
> Thanks for the suggestion.
> 
> But how could I get a geometry table in Postgres that is sure to 
> intersect the raster image?
> The reason why I created another raster table is so that I am able to 
> vary the query size. How can I do it with your suggestions (using 
> polygons) since the polygon may not intersect with the raster.
> 
> Please correct me if I am wrong, as I am still new to this concept.
> 
> Regards,
> Zhi Feng
> 
> -----Original Message-----
> From: postgis-users-bounces at lists.osgeo.org [mailto:postgis-users- 
> bounces at lists.osgeo.org] On Behalf Of Pierre Racine
> Sent: Mittwoch, 4. Dezember 2013 17:43
> To: PostGIS Users Discussion
> Subject: Re: [postgis-users] [postgis] Optimal tile size for Raster
> 
> You are getting fast result because you are not operating at the pixel level.
> Only at the raster extent level (converted to geometry). A more usual 
> query would be to intersect a raster with some polygons:
> 
> SELECT (ST_Intersection(r.rast, q.geom)).* FROM ilatlon32x32 as r, 
> geometrytable as q WHERE ST_Intersects(r.rast, q.geom)
> 
> Why do you want to intersect two rasters? Generally it is to mask one 
> raster with another.
> 
> Pierre
> 
> > -----Original Message-----
> > From: postgis-users-bounces at lists.osgeo.org [mailto:postgis-users- 
> > bounces at lists.osgeo.org] On Behalf Of Heng Zhi Feng (zheng at hsr.ch)
> > Sent: Wednesday, December 04, 2013 10:56 AM
> > To: postgis-users at lists.osgeo.org
> > Subject: [postgis-users] [postgis] Optimal tile size for Raster
> >
> > Hello,
> >
> >
> >
> > I am having to import Raster (.tiff) into Postgres using the 
> > raster2pgsql
> tool.
> > The raster used is 12MB in size and the Postgres version is 9.1 on 
> > Ubuntu 13.10.
> >
> >
> >
> > The virtual machine has the following hardware specs.
> >
> > -          17408 Memory
> >
> > -          4 Processors
> >
> > -          16GB HDD
> >
> > The command used to import (with the parameters) is: raster2pgsql -s
> > 4326 -I -C -M -R -d -l 4 U:\Desktop\ilatlon_float.tif -F -t 8x8
> > ilatlon8x8|psql - d Raster
> >
> > And the tile size varies from 8x8 all the way up till 2048 (power of 2s).
> >
> >
> >
> > My results from the query ran showed that with larger tile size, the 
> > query has a faster returned timing.
> >
> > This trend is however opposite to some of the similar tests 
> > conducted by others (from the internet), where their conclusion was 
> > smaller tile size will have better performance.
> >
> >
> >
> > This is my one of the query I ran:
> >
> > EXPLAIN (ANALYZE,BUFFERS)
> >
> > SELECT DISTINCT ST_Intersection(r.rast::geometry,q.rast::geometry)
> >
> > FROM ilatlon32x32 as r, boundedilatlon32x32 as q;
> >
> >
> >
> > Is someone able to give advice on this, if this result that I 
> > obtained was legit? Or suggest where could the problem be that gave this result?
> >
> >
> >
> > Regards,
> >
> > Zhi Feng
> 
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users
_______________________________________________
postgis-users mailing list
postgis-users at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-users


More information about the postgis-users mailing list