[Qgis-developer] [Geoserver-users] WCS 1.1.1 response rotated

[Radim Blazek]

On Wed, Aug 8, 2012 at 10:39 PM, Andrea Aime
<andrea.aime at geo-solutions.it> wrote:
> On Wed, Aug 8, 2012 at 9:47 PM, Radim Blazek <radim.blazek at gmail.com> wrote:
>>
>> Hi,
>> I cannot get correct response from WCS 1.1.1, the result is rotated.
>> The server is here:
>>
>>  http://www.sogeo.ch/geoserver/wcs
>>
>> request:
>>
>>
>> http://www.sogeo.ch:80/geoserver/wcs?SERVICE=WCS&VERSION=1.1.1&REQUEST=GetCoverage&FORMAT=image/tiff&IDENTIFIER=Img_Sample&BOUNDINGBOX=21.5843815789473652,-129.8681617142856908,53.2349184210526332,-62.9889182857142913,urn:ogc:def:crs:EPSG::4326&GRIDBASECRS=urn:ogc:def:crs:EPSG::4326&GRIDORIGIN=53.2349184210526332,-129.8681617142856908&GRIDCS=urn:ogc:def:cs:OGC:0.0:Grid2dSquareCS&GRIDTYPE=urn:ogc:def:method:WCS:1.1:2dSimpleGrid&GRIDOFFSETS=1.7583631578947370,1.9670365714285709
>>
>> returns rotated raster. I have checked all params and I believe that
>> with GRIDTYPE=urn:ogc:def:method:WCS:1.1:2dSimpleGrid it should never
>> be rotated.
>
>
> Hi Radim, long time no see.
> The WCS spec uses the urn: form, urn:ogc:def:crs:EPSG::4326,  that mandates
> the data to be expressed in lat/lon order, GeoServer respects that to the
> letter.

Yes, the request above has lat/lon order in BOUNDINGBOX, GRIDORIGIN
and GRIDOFFSETS (North America), that is not a problem.

> However you are right, the spec says no rotation allowed:
>
> GridType: default is “urn:ogc:def:method:WCS:1.1:2dSimpleGrid”
> (This GridType disallows rotation or skew relative to the GridBaseCRS –
> therefore
> GridOffsets has only two numbers.)
>
> Now... is it possible to maintain lat/lon axis order and still use
> 2dSimpleGrid?

Why not? Anyway, can you find in GeoServer code the place where the
output is rotated with params given above?

>> Does somebody has any idea? Are the params wrong? Which free WCS
>> client is known to work with GeoServer WCS 1.1.1?
>
> The was no released WCS client to use at the time, GeoServer is the
> reference implementation,
> it was tested as such, nobody from OGC complained.
> However I believe Frank tried out GDAL against GeoServer too in that time,
> can't remember outstanding issues there either.

GDAL does not work with WCS 1.1 in geographic CRS because it does not
switch axis order.

>> The result with 1.1.1 is also larger by 1 pixel in each direction
>> (1.0.0 works OK), is it taken into account that 1.1.1 grid is using
>> cell centers, not edges?
>
> OGC mandates cell centers as the one and only way to deal with raster data
> as far as I know, I believe we should be using it on both WCS versions but
> I'm not sure, they were implemented by two different people and the code
> paths are quite different.

AFAIK, the WCS version 1.0 should use cell edges (bbox) like WMS 1.x does.

Radim

> Cheers
> Andrea
>
> --
> ==
> Our support, Your Success! Visit http://opensdi.geo-solutions.it for more
> information.
> ==
>
> Ing. Andrea Aime
> @geowolf
> Technical Lead
>
> GeoSolutions S.A.S.
> Via Poggio alle Viti 1187
> 55054  Massarosa (LU)
> Italy
> phone: +39 0584 962313
> fax:   +39 0584 962313
> mob:   +39  339 8844549
>
> http://www.geo-solutions.it
> http://twitter.com/geosolutions_it
>
> -------------------------------------------------------
>