[Qgis-user] A problem (shift of 50 m) with some lcc projections and "on the fly" reprojection in QGIS 1.7.4 for Debian Squeeze amd64 bits

Thu Mar 29 00:17:02 PDT 2012

Hi, I do not have the time now tor read carefully allt he mail, but ti
seems to me something similar to this

http://hub.qgis.org/issues/5142

that happens only in Lunux and qgis-master.

cheers

-- Giovanni --

On Thu, 2012-03-29 at 04:09 +0200, blavet wrote:
> Dear developpers and users of QGIS 
> 
> First of all i want to thank the developers of Qgis for this excellent
> software that i follow since the version 0.9 Now the number of my
> colleagues using QGIS is increasing. Itr is very usefull for our
> experimental work in the field (with soil mapping and localisation and
> agronomical trials ) and we also start now to use Qgis with a
> differential gps, in order to localize very precisely some new
> samplings and field trials and indicate them in previous maps.
> 
> Therefore for our work, the QGIS version 1.7.4 with "on the fly"
> projection of vectors and rasters is very interesting. 
> 
> Unfortunately I have a now a quite tricky problem with Qgis 1.7.4 for
> Debian Squeeze (obtained from the Qgis repository
> http://qgis.org/debian/) . And, following my tests this problem does
> not occur with some other OS and QGIS 1.7.4 (namely Windows and
> Ubuntu) 
> 
> In Debian Squeeze with the version 1.7.4 of QGIS, whatever i do, when
> i try"on the fly projection" (of course with on "the fly reprojection"
> activated ) i have systematically a shift of about 50 meters between
> my layers in "lambert II etendu" and the layers two other projections
> (Lambert 93 and RGFCC43) .... 
> 
> I give some details of my trials in the Annex  below
> 
> 
> 
> So .. this is a bit a headache for me … i sincerely think that the
> problem comes from the Debian squeeze version .. probably my
> configuration, but may be a more general problem with Debian squeeze
> and QGIS 1.7.4 
> 
> I would want to use QGIS 1.7.4 with Debian stable (Squeeze) and to
> have good "on the fly projections" and good reprojections for my lcc
> layers … but for the moment is seems difficult … 
> 
> So thank you in advance if you could give me some help
> 
> Kind regards
> 
> 
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> 
> ANNEX about the tests that i did
> 
> 
> I first tried to verify that i did not made mistake with my layers … 
> some of my layers in lambert II etendu came from reprojections of
> "old" layers in lambert zone III and i wondererd if i made good
> reprojections . 
> But such a mistake  would not be logical because these reprojected
> layers are very well overlaid in QGIS with orthophotos in lambert II
> etendu, and also i can find very precisely the localisation of my
> observations in geospatial websites if i convert the localisation
> provided by my layers in lambert II etendu in geographical
> coordinates. 
> 
> i also tried in Debian squeeze to find a way to obtain a good
> overlaying several combinations, Then i also observed that when i make
> a real reprojection (not on the fly) with my vector layers in RGFCC43
> to lambert II etendu (with softwares such as IGNMAP) all the layers
> including the layers in CC43 were perfecty overlaid in a project in
> lambert II etendu !! This indicated to me that my layers should be
> definitively ok and should overlay. .. 
> 
> So i finally was convinced that all my layers were correct and should
> overlay perfecly !! .. and that this problem could come from somewhere
> else. 
> 
> Then, "just by chance" i had to try QGIS 1.7.4 for Windows and i tried
> with exactly the same layers that i have : several layers in lambert
> II étendu, one vector (dgps datat) in CC43 and two rasters in lambert
> 93 (just a copy from the files udes in Debian, of course same
> projections , and same parameters in Qgis with "on the fly
> reprojection activated) ) And .. Bingo !! this shift of 50 m
> disappeared immediately with Qgis 1.7.4 for Windows. 
> 
> Well, i then tested QGIS 1.7.4 for Gnu/Linux Ubuntu distribution LTS
> Lucid Lynx and second Bingo !! . In this two cases all my layers were
> immediately and perfectly overlaid together !!.
> 
> So .... this 50 m shift seems to be connected with the Debian version
> … But i still want to use Debian, that i use daily !! (:-) 
> 
> … Thereafter i tried to compare the versions of QGIS 1.7.4 in the
> three OS  
> 
> As indicated in the message "about" i have this specifications 
> 
> 
> Qgis 1.7.4 for Debian squeeze is compiled from the version "exported"
> of the code
> and is using  :
> GDAL/OGR version 1.6.3
> Version de QWT : 5.2.0.
> is compiled with Qt 4.6.3 and run with Qt 4.6.3 
> 
> Qgis 1.74 for lucid Lynx LTD (which does not have also this 50 m
> shift) is compiled from the version "exported" of the code
> and is using  :
> GDAL/OGR version 1.8 .0
> Version de QWT : 5.2.0.
> is compiled with Qt 4.6.2 and run with Qt 4.6.2 
> 
> Qgis 1.7.4 for windows (which does not have this 50 m shift") is
> compiled from version 411aff6 of the code
> and is using  :
> GDAL/OGR version 1.8.1
> Version de QWT : 5.2.1.
> is compiled with Qt 4.7.1 and run with Qt 4.7.1
> 
> From this comparison I supposed that my problem could be linked with
> the version of GDAL/OGR which is a bit older in the Debian Squeeze.
> Then i uninstalled Qgis 1.7.4 in Debian (with purge), also in
> uninstalled all the libgdal and I reinstalled QGIS 1.7.4 after having
> added the gfoss repository to obtain libgdal1-1.7.0. Unfortunately,
> even with libgdal1-1.7.0 installed in Debian, QGIS 1.7.4 still says
> that it is using GDAL/OGR 1.6.3.. Also if i try to just uninstall
> libgdal1-1.6.0 (which is named version 1.6.3-4+b1 (stable)) the
> synaptic package manager says that it would uninstall Qgis.... 
> 
> ------------------------------------------------------------------------------------------------
> 
> By the way i also observed a  strange  behaviour in the metadata for
> the Debian version
>  (i'm not sure that this is connected with the previous one, but it
> would not be impossible)
> 
> 
> In my three Operating systems, the proj4 definition of the projection
> Lambert II etendu IGNF:LAMBE ID : 10090 indicates the use of a nadgrid
> ntf_rgf93 like this :
> +proj=lcc +nadgrids=ntf_r93.gsb,null +a=6378249.2000
> +rf=293.4660210000000 +pm=2.337229167 +lat_0=46.800000000
> +lon_0=0.000000000 +k_0=0.99987742 +lat_1=46.800000000 +x_0=600000.000
> +y_0=2200000.000 +units=m +no_defs 
> 
> 
> However in Debian Squeeze version of Qgis 1.7.4 when i "right click"
> on a layer in "Lambert II etendu" to obtain the properties and the
> medatata about the SCR the +nadgrids=ntf_r93.gsb,null simply
> disappears in the medata , and i just obtain this :
> +proj=lcc +lat_1=46.8 +lat_0=46.8 +lon_0=0 +k_0=0.99987742 +x_0=600000
> +y_0=2200000 +a=6378249.2 +b=6356514.999978254 +pm=2.337229167
> +units=m +no_defs 
> 
> (by the way i think that there is no problem with the b instead of rf,
> it is ok, as b = a*(1 - 1/rf)) The only  problem seems to be with
> +nadgrids=ntf_r93.gsb,null which have disappeared in the metadata
> provided 
> 
> 
> Moreover, in the other OS (Ubuntu Lucid lynx and Windows) i still have
> the +nadgrids=ntf_r93.gsb,null in the metadata. I obtain this by a
> right click on the layer and going the metadatya  :
> +proj=lcc +lat_1=46.8 +lat_0=46.8 +lon_0=0 +k_0=0.99987742 +x_0=600000
> +y_0=2200000 +a=6378249.2 +b=6356514.999978254
> +nadgrids=ntf_r93.gsb,null +pm=2.337229167 +units=m +no_defs 
> 
> (by the way I also verified that i have the same file ntf_rg83.gsb in
> usr/share/proj in Ubuntu and in Debian versions .. it is ok i have the
> same file (same size, same md5sum)... 
> 
> 
> finally for the proj, proj-bin and proj-data package i have 4.7.0-1
> in, Debian squeeze and 4.7.0-2 in Ubuntu Lucid Lynx (i'm note sure
> also if this is important or not) 
> 
> 
> 
> 
> 
> -- 
> 
> 
> _______________________________________________________________________________________
> 
> Didier Blavet 
> 
> Institut de Recherche pour le Développement - IRD (http://www.ird.fr)
> 
> UMR Eco&Sols - Ecologie Fonctionnelle & Biogéochimie des Sols &
> Agroécosystèmes 
> 
> (Montpellier SupAgro-CIRAD-INRA-IRD)
> (http://www.montpellier.inra.fr/ecosols)
> 
> Bâtiment 12, 2 place Viala, F-34060 Montpellier Cedex 2 , France 
> 
> Tel bureau : +33 (0)4 99 61 21 33 
> 
> Secrétariat : +33 (0)4 99 61 21 01 / Fax : +33 (0)4 99 61 21 19 
> 
> Courriel : didier.blavet at ird.fr 
> 
> _______________________________________________________________________________________
> 
> 
> _______________________________________________
> Qgis-user mailing list
> Qgis-user at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/qgis-user