[postgis-users] ST_Buffer + grid problem

Ed Linde edolinde at gmail.com
Thu Mar 22 08:00:44 PDT 2012


I tried a few roads at random and it doesn't reproduce that error, though I
am a bit suspicious about the actual
query. This part in particular

(SELECT ST_AsRaster(ST_Extent(way_geom)::geometry, 0.000036,
> 0.000036) rast
>                          FROM buffers
>                         ) foo1

Are we saying that we want to compute the extent of each and every geometry
in my table? Or should we not
be getting the maximum extents for ALL the geometries?
So when I say something like "FROM buffers where osm_id in (... list of 10
IDs)... " this query runs forever!
Eventhough checking them one by one returns with results immediately.
So I am not entirely sure how to do this or how I can reproduce this bug to
help you guys.
Any suggestions?


On Thu, Mar 22, 2012 at 3:23 PM, Ed Linde <edolinde at gmail.com> wrote:

> Yeah only problem being that I don't know which geometry is the one that
> fails... its a table with 300K geometries.
> Have never written any PgSQL, so this I guess would require a loop to
> check which OSM ID its failing on. Unless
> you got a good idea to pin-point this error?
>
>
> On Thu, Mar 22, 2012 at 3:16 PM, Pierre Racine <
> Pierre.Racine at sbf.ulaval.ca> wrote:
>
>> What I mean is to replace:
>>
>> SELECT ST_AsRaster(ST_Extent(way_geom)::geometry, 0.000036, 0.000036)
>> rast FROM buffers  where osm_id = 94695311
>>
>> by something like:
>>
>> SELECT ST_AsRaster(ST_GeomFromText('POLYGON((11.1539754732244
>> 55.5772503545236,11.1540114732244 55.5772503545236,11.1540114732244
>> 55.5772863545236,11.1539754732244 55.5772863545236,11.1539754732244
>> 55.5772503545236))'), 0.000036, 0.000036)
>>
>> with one of the geometry reproducing the bug so we don't need your
>> dataset to test and reproduce.
>>
>> Pierre
>>
>> > -----Original Message-----
>> > From: manohar.kaul at gmail.com [mailto:manohar.kaul at gmail.com] On Behalf
>> > Of Ed Linde
>> > Sent: Thursday, March 22, 2012 9:59 AM
>> > To: Pierre Racine
>> > Cc: PostGIS Users Discussion; Bborie Park (bkpark at ucdavis.edu)
>> > Subject: Re: [postgis-users] ST_Buffer + grid problem
>> >
>> > sure. Did you mean ST_GeomfromText or show it as text?
>> >
>> > SELECT ST_GeomfromText ((gvxy).geom), ((gvxy).x - 1) * rwidth +
>> (gvxy).y gridid
>> > FROM (SELECT ST_PixelAsPolygons(rast) gvxy, ST_Width(rast) rwidth
>> >             FROM (SELECT ST_AsRaster(ST_Extent(way_geom)::geometry,
>> 0.000036,
>> > 0.000036) rast
>> >                          FROM buffers
>> >                          where osm_id = 94695311
>> >                         ) foo1
>> >            ) foo2;
>> >
>> > ERROR:  parse error - invalid geometry
>> > HINT:  "010300000001000000050000005" <-- parse error at position 27
>> within
>> > geometry
>> >
>> > ********** Error **********
>> >
>> > ERROR: parse error - invalid geometry
>> > SQL state: XX000
>> > Hint: "010300000001000000050000005" <-- parse error at position 27
>> within
>> > geometry
>> >
>> >
>> >
>> > On Thu, Mar 22, 2012 at 2:53 PM, Pierre Racine <
>> Pierre.Racine at sbf.ulaval.ca>
>> > wrote:
>> >
>> >
>> >       I guess this is related to a GDAL problem where you cannot create
>> a
>> > raster using ST_AsRaster with pixel size smaller than 1. Was this fixed
>> Bborie?
>> >
>> >       Ed could you reduce the list of geometries passed to ST_Extent to
>> one
>> > and write it as ST_GeomfromText so we can debug this more easily?
>> >
>> >
>> >       Pierre
>> >
>> >       > -----Original Message-----
>> >       > From: manohar.kaul at gmail.com [mailto:manohar.kaul at gmail.com]
>> > On Behalf
>> >       > Of Ed Linde
>> >
>> >       > Sent: Thursday, March 22, 2012 9:45 AM
>> >       > To: Pierre Racine
>> >       > Cc: PostGIS Users Discussion
>> >       > Subject: Re: [postgis-users] ST_Buffer + grid problem
>> >       >
>> >       > Hi Pierre,
>> >
>> >       > So I am trying to have a 4meter by 4meter cell and I converted
>> > 4meters to
>> >       > degrees, because I think ST_Extent will compute it in degrees
>> as my
>> > geometry is
>> >       > in srid 4326. But I get this GDAL error which I have no clue
>> what it
>> > means. Any
>> >       > ideas how I can fix it? Also I noticed that if I set it to 1.0,
>> 1.0 I get
>> >       > 21 rows but the gridid isn't continuous either (which is weird).
>> >       >
>> >       > CREATE TABLE vectorgrid AS
>> >       > SELECT (gvxy).geom, ((gvxy).x - 1) * rwidth + (gvxy).y gridid
>> FROM
>> > (SELECT
>> >       > ST_PixelAsPolygons(rast) gvxy, ST_Width(rast) rwidth
>> >       >             FROM (SELECT
>> ST_AsRaster(ST_Extent(way_geom)::geometry,
>> > 0.000036,
>> >       > 0.000036) rast
>> >       >                          FROM buffers
>> >       >                         ) foo1
>> >       >            ) foo2;
>> >       >
>> >       > ERROR:  rt_raster_gdal_rasterize: Unable to add band to
>> GDALDataset
>> >       >
>> >       > ********** Error **********
>> >       >
>> >       > ERROR: rt_raster_gdal_rasterize: Unable to add band to
>> GDALDataset
>> > SQL state:
>> >       > XX000
>> >       >
>> >       >
>> >       >
>> >       >
>> >       > On Thu, Mar 22, 2012 at 2:13 PM, Pierre Racine
>> > <Pierre.Racine at sbf.ulaval.ca>
>> >       > wrote:
>> >       >
>> >       >
>> >       >       You can control how the grid is aligned by using more
>> ST_AsRaster
>> >       > parameters. See:
>> >       >
>> >       >       http://postgis.refractions.net/documentation/manual-
>> >       > svn/RT_ST_AsRaster.html
>> >       >
>> >       >       If you want it to align on your point, just align it on
>> your points...
>> >       >
>> >       >       Pierre
>> >       >
>> >       >
>> >       >       > -----Original Message-----
>> >       >       > From: manohar.kaul at gmail.com
>> > [mailto:manohar.kaul at gmail.com]
>> >       > On Behalf
>> >       >       > Of Ed Linde
>> >       >       > Sent: Thursday, March 22, 2012 9:05 AM
>> >       >       > To: PostGIS Users Discussion
>> >       >       > Cc: Pierre Racine
>> >       >       > Subject: Re: [postgis-users] ST_Buffer + grid problem
>> >       >       >
>> >       >       > Hi Pierre,
>> >       >       > Thanks for the grid idea last night. I have a problem
>> though, I
>> > want to
>> >       > have a 4m
>> >       >       > by 4m cell sized grid over my buffer geometries which
>> are in SRID
>> > =
>> >       > 4326. And
>> >       >       > then I have another data set of 2D points which are
>> also in SRID
>> >       > 4326...
>> >       >       > wondering how I can figure out from just a given
>> (lat,long) which
>> > cell
>> >       > ID in the
>> >       >       > buffer grid it would belong to? I am thinking that I
>> might run into
>> >       > "alignment
>> >       >       > issues" because shouldn't the extent of the grid be
>> exactly the
>> > same
>> >       > on the 2D
>> >       >       > point cloud as well, so that the grid cell IDs will
>> match? Unless I
>> > am
>> >       > missing
>> >       >       > something here?
>> >       >       >
>> >       >       > Cheers,
>> >       >       > Ed
>> >       >       >
>> >       >       >
>> >       >       >
>> >       >       > On Wed, Mar 21, 2012 at 8:46 PM, Pierre Racine
>> >       > <Pierre.Racine at sbf.ulaval.ca>
>> >       >       > wrote:
>> >       >       >
>> >       >       >
>> >       >       >       > I am not sure if this is possible, but I have
>> computed (using
>> >       > ST_Buffer)
>> >       >       > a sort of
>> >       >       >       > buffer around several LINESTRINGs. Now I would
>> like to lay
>> > some
>> >       > sort
>> >       >       > of grid of
>> >       >       >       > say 1m^2 cell size on top of this collection of
>> geometries
>> > and
>> >       > then
>> >       >       > compute the
>> >       >       >       > intersection... so in the end I would like to
>> for example,
>> > know
>> >       > that grid
>> >       >       > cell ID = 1
>> >       >       >       > intersects with buffers 1 and 10, grid cell 2
>> intersects with
>> > buffers
>> >       > 7, 10
>> >       >       > etc..
>> >       >       >       >
>> >       >       >       > Is there a simple way of doing this in postgis?
>> Maybe
>> > someone
>> >       > could
>> >       >       > point me to
>> >       >       >       > some documentation of how I can generate such a
>> grid in
>> > postgis
>> >       > and
>> >       >       > maybe
>> >       >       >       > then I can use just ST_Intersect once I have
>> these two
>> >       > geometries?
>> >       >       >
>> >       >       >
>> >       >       >       With the raster type you can now easily create a
>> vector grid
>> > like
>> >       > this:
>> >       >       >
>> >       >       >       CREATE TABLE vectorgrid AS
>> >       >       >       SELECT (gvxy).geom, ((gvxy).x - 1) * rwidth +
>> (gvxy).y gridid
>> >       >       >       FROM (SELECT ST_PixelAsPolygons(rast) gvxy,
>> ST_Width(rast)
>> >       > rwidth
>> >       >       >                   FROM (SELECT
>> > ST_AsRaster(ST_Extent(geom)::geometry,
>> >       > 1.0,
>> >       >       > 1.0) rast
>> >       >       >                                FROM yourbuffertable
>> >       >       >                               ) foo1
>> >       >       >                  ) foo2;
>> >       >       >
>> >       >       >       Make sure a spatial index exist on both tables:
>> >       >       >
>> >       >       >       CREATE INDEX yourbuffertable_geom_idx  ON
>> > yourbuffertable
>> >       > USING
>> >       >       > gist  (geom);
>> >       >       >       CREATE INDEX vectorgrid _geom_idx  ON vectorgrid
>> USING
>> > gist
>> >       >       > (geom);
>> >       >       >
>> >       >       >       You can then perform a normal intersect query:
>> >       >       >
>> >       >       >       CREATE TABLE interresult AS
>> >       >       >       SELECT b.bufferid, g.gridid,
>> ST_Intersection(g.geom, b.geom)
>> >       > geom
>> >       >       >       FROM vectorgrid g, yourbuffertable b
>> >       >       >       WHERE ST_Intersects(g.geom, b.geom);
>> >       >       >
>> >       >       >       Pierre
>> >       >       >       _______________________________________________
>> >       >       >       postgis-users mailing list
>> >       >       >       postgis-users at postgis.refractions.net
>> >       >       >
>> http://postgis.refractions.net/mailman/listinfo/postgis-users
>> >       >       >
>> >       >       >
>> >       >
>> >       >
>> >       >
>> >
>> >
>> >
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20120322/f716b52d/attachment.html>


More information about the postgis-users mailing list